Fatigue Analysis - Automotive Components

Download as pdf or txt
Download as pdf or txt
You are on page 1of 306

A Dissertation

entitled

Manufacturing Process Effects on Fatigue Design and


Optimization of Automotive Components –
An Analytical and Experimental Study

by
Mehrdad Zoroufi

Submitted as partial fulfillment of the requirements for


the Doctor of Philosophy in Engineering Science

Graduate School

The University of Toledo


December 2004
ii
An Abstract of

Manufacturing Process Effects on Fatigue Design and Optimization


of Automotive Components – An Analytical and Experimental Study

Mehrdad Zoroufi

Submitted as partial fulfillment of the requirements for


the Doctor of Philosophy in Engineering Science

The University of Toledo

December 2004

Numerous fatigue-critical parts could be found in ground vehicles, and time-varying

loads have always challenged automotive designers. Fatigue design and life assessment of

these components are essentially influenced by the material used and manufacturing

processes chosen. Exploring the design criteria and optimization potentials with respect to

manufacturing processes is vital to the industry.

This study was aimed at developing general procedures for fatigue analysis and

optimization of safety-critical automotive components with manufacturing considerations. A

literature survey was conducted, specimen and component tests were performed, and finite

element stress analysis and durability and optimization evaluations of similar components

produced by different manufacturing technologies were made to achieve the objectives. The

typical example component chosen was a vehicle steering knuckle made of three competing

materials and manufacturing processes including forged steel, cast aluminum and cast iron.

In the literature survey, manufacturing processes were studied and compared with

focus on mechanical behavior. The methods used in the literature for fatigue life evaluation

and prediction of automotive components, as well as for optimization studies with respect to

geometry, material and manufacturing aspects were also reviewed. Specimen strain-

iii
controlled tests were conducted to obtain material monotonic and cyclic deformation and

fatigue properties. Components’ fatigue behaviors were investigated via constant-amplitude

load-controlled fatigue tests. Comparisons of materials monotonic and fatigue properties,

and components’ fatigue behaviors were made for competing material and manufacturing

processes. In terms of structural performance and durability, based on both material testing

and component evaluation, forged steel was found superior to cast iron which in turn was

found superior to cast aluminum.

Finite element models of the components were analyzed, using linear and nonlinear

stress analyses. The nominal stress and local stress and strain approaches were employed to

assess durability of the components. Experimental and analytical stress and fatigue life

results were compared to evaluate the validity of the analytical approaches. The strength and

shortages of the applied models and alternative analyses were also investigated. It was

concluded that the local life prediction approaches in combination with either nonlinear

finite element analysis results, or linear finite element analysis results corrected for local

plasticity, yielded satisfactory predictions.

A procedure was developed to optimize forged automotive components for weight

reduction and cost savings with fatigue strength as the key performance indicator. By

considering geometry variations, alternative materials and manufacturing process parameters

as design variables, the example part was optimized. Guidelines were developed and

limitations were identified for the optimization procedure. Although the optimization results

showed limited changes for the particular example component, the approach that was

followed is applicable to other forged components. It was emphasized that geometrical

optimization of manufactured components could only be realistic and practical if other

important parameters like material, manufacturability, and cost are taken into account.

iv
Acknowledgements

Dr. Ali Fatemi, my advisor, has always impressed me with his outstanding

professional conduct, his strong conviction for science, and his belief that a doctorate

program is only a start of a “life-long learning” experience. I appreciate his consistent

support from the first day I applied to this graduate program to these concluding moments.

I am truly grateful for his progressive vision about my training in science, his tolerance of my

naïve mistakes, and his commitment to my future career.

I am grateful to my dissertation committee members: Dr. Mehdi Pourazady for our

fruitful discussions in finite element analysis, Dr. Efstratios Nikolaidis for his suggestions in

optimization, and Dr. Naser Mostaghel, Dr. Cedric Mousseau and Dr. Mohamed Samir

Hefzy for their helpful criticism and suggestions.

I would like to thank my labmates at the Fatigue and Fracture Research Laboratory,

especially those who mentored me when I first joined the laboratory. Also, the training and

assistance of Mr. John R. Jaegly, Mr. Randall S. Reihing and Mr. Tim Grivanos of the

Machining Shop of the MIME Department are greatly appreciated.

I appreciate the support of the Forging Industry Educational and Research

Foundation, especially Mr. George Mochnal (Director of Research and Education) and Ms.

Karen Lewis (Executive Director), and the American Iron and Steel Institute, especially Mr.

David Anderson (Director, Bar and Rod Programs), who facilitated this research program.

For helpful suggestions related to various aspects of this work, I would like to thank,

in no particular order, Mr. Peter Bauerle (Senior Specialist) of Daimler-Chrysler Corporation,

Mr. Michael Crews, Mr. Jay Hedges, Mr. Mohammad Saeed, Mr. Andrew R. Snow, and Mr.

Gary Studt of Metaldyne Corporation, Mr. Tom Oakwood formerly of Inland Steel

v
Corporation, Dr. Chinchan Chu of Ford Motor Company, Mr. Joseph N. Grosso (Program

Manager) of Citation Corporation, and Mr. Amir R. Shayan of The University of Toledo.

Dr. Majid Mirzaei, my advisor in the Master’s program in Iran, was the one who

introduced me to the field of fatigue and inspired me to continue my graduate studies. It was

he who recommended that I would pursue my doctoral degree under the direction of Dr.

Fatemi. I remain grateful to him for having opened this door for me.

I would also like to thank Dr. Sonia M. Najjar at the Medical College of Ohio, a

brilliant scientist and a remarkable person, for her continuous advice during this journey.

Lastly, I am grateful for the many sacrifices my family has given to help me complete

this program. I can not find the appropriate words that could properly describe my

appreciation for their devotion, support and faith in my ability to attain my goals. They have

been unforgettably exceptional for their encouragement of my career goals. My parents have

been especially inspirational with their strength to overcome hurdles. It is through their

teaching, positive attitude in life and their love that I have been able to face the challenges of

carrying out my graduate studies away from home.

vi
Table of Contents

Abstract iii
Acknowledgements v
Table of Contents vii
List of Tables x
List of Figures xii
Nomenclature xix

1 Introduction 1
1.1 Motivation 2
1.2 Objectives 4
1.3 Scope of this Study and Overview of the Dissertation 4
1.3.1 Experimental Work 5
1.3.2 Analytical Work 6

2 Background and Literature Review 9


2.1 Forging Process and the Influencing Parameters 9
2.1.1 Forging Processes 10
2.1.2 Important Factors in Forging Process 13
2.2 Vehicle Engine and Suspension 16
2.2.1 Vehicle Engine 17
2.2.2 Suspension System 17
2.2.3 Steering Knuckle 19
2.3 Comparison of Manufacturing Processes 21
2.4 Durability Assessment of Automotive Components 31
2.5 Optimization with Material, Manufacturing and Cost Considerations 40

3 Material Fatigue Behavior and Comparisons 74


3.1 Experimental Program 75
3.1.1 Material and Specimen Fabrication 75
3.1.2 Testing Equipment 77
3.2 Test Methods and Procedures 78
3.2.1 Monotonic Tension Tests 78
3.2.2 Constant Amplitude Fatigue Tests 79
3.3 Experimental Results, Observations, and Analysis 80
3.3.1 Monotonic Deformation Behavior and Comparisons 80
3.3.2 Cyclic Deformation Behavior and Comparisons 83
3.3.2.1 Transient Cyclic Response 83

vii
3.3.2.2 Steady-State Cyclic Deformation 84
3.3.3 Strain-Controlled Fatigue Behavior and Comparisons 86

4 Finite Element Analysis 118


4.1 Component Geometry Generation 119
4.2 Model Boundary Conditions 119
4.3 Meshing of the Finite Element Models 120
4.4 Material Nonlinear Behavior Models 122
4.5 Model Solutions 123
4.6 Analysis of the Results 124

5 Component Fatigue Behavior and Comparisons 139


5.1 Identifying the Components and their Service Loadings 140
5.2 Fixture Assembly Design and Preparation 141
5.3 Pre-Test Analysis and Considerations 142
5.4 Collecting Experimental Data 145
5.5 Experimental Results, Analysis, and Comparisons 146

6 Fatigue Life Predictions and Comparisons with Experimental Results 160


6.1 Nominal Stress Approach 161
6.2 Local Strain Approach Using Nominal Stresses 162
6.3 Local Stress and Strain Approaches Using FEA Results 165
6.4 Fatigue Performance Comparison of Components 169
6.5 Other Considerations in Durability Assessment of Steering Knuckle 169

7 Optimization Including Material, Manufacturing and Cost Considerations 182


7.1 Component Specifications 183
7.2 Optimization Problem Definition 187
7.2.1 Objective Function 190
7.2.2 Constraints 190
7.2.3 Design Variables 191
7.3 Optimization Analysis, Results and Discussions 192
7.3.1 Stage I – No Change in Attachment Geometry 192
7.3.1.1 Optimization Using Geometry Variables 193
7.3.1.2 Modifications to Manufacturing Processes 197
7.3.2 Stage II – Limited Change in Attachment Geometry 203
7.3.2.1 Optimization Using Geometry Variables 203
7.3.2.2 Modifications Using Alternative Materials 204
7.3.2.3 Modifications to Manufacturing Processes 207
7.3.3 Discussions 209

viii
8 Summary and Conclusions 231

References 239

Appendix A - Summary of Literature Review for Fatigue Life Prediction of


Automotive Components 248

Appendix B - Life Prediction Calculation Details 272

ix
List of Tables

Table 2-1 Classification of metals in decreasing order of forgeability


(Kalpakjian and Schmid, 2000). 51
Table 2-2 Properties of cast and forged steels swivel hub (Cristinacce et al.,
1998). 51
Table 2-3 Summary of characteristic comparison between manufacturing
processes. 52
Table 2-4 Test setup for steering knuckle and other relevant automotive
components. 53
Table 3-1 Summary of mechanical properties and their comparative ratios
(forged steel is taken as the base for ratio calculations). 90
Table 3-2 Summary of monotonic test results for forged steel 11V37, cast
aluminum A356-T6, and cast iron 65-45-12. 91
Table 3-3 Summary of constant amplitude completely reversed fatigue test
results for forged steel 11V37- Direction A. 92
Table 3-4 Summary of constant amplitude completely reversed fatigue test
results for forged steel 11V37- Direction B. 93
Table 3-5 Summary of constant amplitude completely reversed fatigue test
results for forged steel 11V37- Direction C. 94
Table 3-6 Summary of constant amplitude completely reversed fatigue test
results for cast aluminum A356-T6. 95
Table 3-7 Summary of constant amplitude completely reversed fatigue test
results for cast iron 65-45-12. 96
Table 3-8 Summary of fatigue constants for the three geometrical directions
A, B and C of forged steel 11V37. 97
Table 4-1 Components of stress and strain for the critical locations of the
forged steel (M = 1515 N.m), cast aluminum (M = 2230 N.m) and
cast iron (M = 2230 N.m) steering knuckles. 128
Table 4-2 Nominal moment ratios (Rm) and local stress ratios (Rσ) indicating
variation of Rσ resulting from different levels of notch residual
stress generated during cycling as a function of Mmax. 129
Table 5-1 Measured and predicted strain values at 680 N.m static moment.
Locations of the gages are also shown. 148
Table 5-2 Component test data of forged steel and cast aluminum steering
knuckles investigated. 149
Table 6-1 Nominal and local stresses and predicted and experimental fatigue
lives for the forged steel steering knuckle using nominal stress and
local strain approaches. 174

x
Table 6-2 Comparison of FEA results and predicted stresses and strains
obtained from Neuber’s rule, and Seeger Seeger-Heuler’s
generalized application of Neuber’s rule for the forged steel steering
knuckle. The predicted fatigue lives are calculated from strain-life
method. 175
Table 6-3 Comparison of alternating and mean stress values, SWT parameter
and fatigue life for two stress calculation methods; nonlinear FEA,
and elastic FEA plus Neuber-correction using σ-N and ε-N
methods. Fatigue lives for σ-N and ε-N approaches were both
obtained from the SWT model. Nonlinear FEA results were taken
as the base for comparison. 176
Table 7-1 Cost attributes for manufacturing the steering knuckle. 212
Table 7-2 Preliminary list of alternative materials with superior fatigue
strength, as compared to the currently used alloy steel (SAE
11V37). 213
Table 7-3 Effect of common alloying elements in automotive steels (Bayer,
2003). 214
Table 7-4 Potential alternative materials to replace the current 11V37 steel. All
the steels are in the form of bars. 215
Table 7-5 Summary of the results of optimization stages. 216

xi
List of Figures

Figure 1-1 From left forged steel, cast aluminum and cast iron steering
knuckles selected as example parts for this study. 8
Figure 2-1 Flow chart of the forging process (Kalpakjian and Schmid, 2000). 58
Figure 2-2 (a) Stages in impression-die forging of a connecting rod for an
internal combustion engine. (b) Fullering and (c) edging operations
to distribute the material when pre-shaping the blank for forging
(Kalpakjian and Schmid, 2000). 58
Figure 2-3 Typical vehicle components that are subject to competition among
manufacturing processes (Cristinacce et al., 1998). 59
Figure 2-4 Components of the engine block. 59
Figure 2-5 Front-wheel-drive car with MacPherson-strut front suspension and
strut-type independent rear suspension (Saturn Corp.). 60
Figure 2-6 Arrangement for SLA suspension. The steering knuckle/spindle
assembly in SLA suspension supports the wheels and attaches to
the control arm with ball and socket joints. The control arm
attaches to the frame of the vehicle through rubber bushings to
help isolate noise and vibration between the road and the body
(Halderman and Mitchel, 2000). 60
Figure 2-7 Multilink double-wishbone suspension uses a spring and damper
unit. The long curved steering knuckle and very angled upper
control arm allow its use in areas of limited size, Courtesy of
American Honda Motor Co., Inc. (Birch, 2000). 61
Figure 2-8 Steering knuckle in strut suspension. In this type of suspension, the
strut rod is the longitudinal support to prevent front-to-back wheel
movement, Courtesy of Ford Motor Company (Birch, 2000). 61
Figure 2-9 Spindle assembly in a solid axle suspension, Courtesy of Ford
Motor Company (Birch, 2000). 62
Figure 2-10 A twin I-beam front suspension. This particular swing axle design
uses two ball joints to connect the steering knuckle to the axle,
Courtesy of Ford Motor Company (Birch, 2000). 62
Figure 2-11 Static lateral input (left) and static longitudinal input (Haeg, 1997). 63
Figure 2-12 Steering knuckle assembly force reactions with longitudinal input
isometric view (left) and plan view (Haeg, 1997). 63
Figure 2-13 Forging process parameters, the manufacturing influenced
parameters and the effect on mechanical properties. Each
parameter and its connector lines are coded with the same color. 64
Figure 2-14 Constant amplitude fatigue crack growth behavior of Q&T vs. cast
steels (Rice, 1997). 65

xii
Figure 2-15 Superimposed plots of true stress amplitude versus reversals to
failure for forged steel and powder metal materials used in
connecting rods (Afzal and Fatemi, 2004). 65
Figure 2-16 Experimental stress amplitude vs. cycles to failure for forged steel
and powder metal connecting rods (Afzal and Fatemi, 2004). 66
Figure 2-17 Changes in fatigue performance of vehicle front suspension arm
due to surface defects from forged and cast manufacturing
processes (Blarasin and Giunti, 1997). 66
Figure 2-18 Capability of each manufacturing process of producing parts to
some characteristic tolerance and surface finish under typical
conditions (Schey, 2000). 67
Figure 2-19 Influence of forging reduction on anisotropy for a 0.35% carbon
wrought steel. Properties for a 0.35% carbon cast steel are shown in
the graph by a star (*) for purposes of comparison (Blair and
Monroe, 2002). 67
Figure 2-20 Relative unit costs of a small connecting rod made by various
forging and casting processes (Kalpakjian and Schmid, 2000). 68
Figure 2-21 Fatigue design flow chart (Stephens et al., 2000). 68
Figure 2-22 Methods of structural durability assessment (Berger et al., 2002). 69
Figure 2-23 Timing of past (top) and current product development periods
(Berger et al., 2002). 69
Figure 2-24 Wheel loads and moments on the test road indicating the existence
of overloads are part of the service history (Lee et al., 1995). 70
Figure 2-25 Design model (left) and meshes of the steering knuckle (Botkin,
1991). 71
Figure 2-26 Baseline superposed over optimized steering knuckle (Krishna,
2001). 71
Figure 2-27 Fatigue strength of microalloyed (SV40CL1) and Q&T (S40C)
forged connecting rods (Kuratomi et al., 1990). 72
Figure 2-28 Case studies to optimize manufacturing processes of automotive
components: (a) drive shaft flange, (b) output shaft, (c) flanged
cylinder, and (d) link shaft (Nägele et al., 2000). 72
Figure 2-29 Effect of shot-peening on fatigue behavior of carburized gears
(Stephens et al., 2000). 73
Figure 2-30 Original design (left), topology optimization results (middle) and
final design of an engine mount bracket (Schramm et al., 2002). 73
Figure 3-1 Specimen configuration and dimensions (all dimensions in mm). 98
Figure 3-2 From left to right: forged steel, cast aluminum, and cast iron
steering knuckles. The three perpendicular geometrical directions of

xiii
the forged steel steering knuckle are; one along spindle, one along
the arm perpendicular to the body, and one along the body. 98
Figure 3-3 True stress versus true plastic strain for (a) forged steel 11V37, (b)
cast aluminum A356-T6, and (c) cast iron 65-45-12. 99
Figure 3-4 Monotonic stress-strain curves for (a) forged steel 11V37, (b) cast
aluminum A356-T6, and (c) cast iron 65-45-12. 100
Figure 3-5 Superimposed monotonic stress-strain curves for the three
materials; forged steel 11V37, cast aluminum A356-T6, and cast
iron 65-45-12. 101
Figure 3-6 True stress amplitude versus normalized number of cycles for (a)
forged steel 11V37, (b) cast aluminum A356-T6, and (c) cast iron
65-45-12. 102
Figure 3-7 True stress amplitude versus number of cycles for (a) forged steel
11V37, (b) cast aluminum A356-T6, and (c) cast iron 65-45-12. 103
Figure 3-8 True stress amplitude versus calculated true plastic strain amplitude
for (a) forged steel 11V37 in three directions, (b) cast aluminum
A356-T6, and (c) cast iron 65-45-12. 104
Figure 3-9 True stress amplitude versus true strain amplitude for (a) forged
steel 11V37 in three directions, (b) cast aluminum A356-T6, and (c)
cast iron 65-45-12. 105
Figure 3-10 Superimposed true stress amplitude versus true strain amplitude
curves for the three materials; forged steel 11V37 in direction A,
cast aluminum A356-T6, and cast iron 65-45-12. 106
Figure 3-11 Superimposed plots of cyclic and monotonic stress-strain curves for
(a) forged steel 11V37 in direction A, (b) cast aluminum A356-T6,
and (c) cast iron 65-45-12. 107
Figure 3-12 Superimposed plot of cyclic and monotonic stress-strain curves for
the three materials; forged steel 11V37 in direction A, cast
aluminum A356-T6, and cast iron 65-45-12. 108
Figure 3-13 Composite plot of midlife hysteresis loops for (a) forged steel
11V37 in direction A, (b) cast aluminum A356-T6, and (c) cast iron
65-45-12. 109
Figure 3-14 (a) True stress amplitude versus reversals to failure, (b) calculated
true plastic strain amplitude versus reversals to failure, and (c) true
strain amplitude versus reversals to failure for forged steel 11V37 in
three directions. 110
Figure 3-15 True stress amplitude versus reversals to failure for (a) forged steel
11V37 in direction A, (b) cast aluminum A356-T6, and (c) cast iron
65-45-12. 111

xiv
Figure 3-16 Superimposed true stress amplitude versus reversals to failure for
the three materials; forged steel 11V37 in direction A, cast
aluminum A356-T6, and cast iron 65-45-12. 112
Figure 3-17 Calculated true plastic strain amplitude versus reversals to failure for
(a) forged steel 11V37 in direction A, (b) cast aluminum A356-T6,
and (c) cast iron 65-45-12. 113
Figure 3-18 Superimposed calculated true plastic strain amplitude versus
reversals to failure for the three materials; forged steel 11V37 in
direction A, cast aluminum A356-T6, and cast iron 65-45-12. 114
Figure 3-19 True strain amplitude versus reversals to failure for (a) forged steel
11V37 in direction A, (b) cast aluminum A356-T6, and (c) cast iron
65-45-12. 115
Figure 3-20 Superimposed true strain amplitude versus number of reversals to
failure for the three materials; forged steel 11V37 in direction A,
cast aluminum A356-T6, and cast iron 65-45-12. 116
Figure 3-21 Neuber fatigue life curves for forged steel 11V37, cast aluminum
A356-T6, and cast iron 65-45-12. 117
Figure 4-1 From left to right the solid models of the forges steel, cast
aluminum and cast iron steering knuckles. 130
Figure 4-2 From top to bottom boundary conditions for the forged steel, cast
aluminum and cast iron steering knuckles. 130
Figure 4-3 von Mises stress versus mesh size at critical locations of (a) forged
steel , (b) cast aluminum and (c) cast iron steering knuckles to check
mesh convergence. 131
Figure 4-4 From top to bottom, the generated mesh for the forged steel, cast
aluminum and cast iron steering knuckles. 132
Figure 4-5 Superimposed cyclic stress-strain curves and the bilinear models
used in the nonlinear FEA for the forged steel, cast iron and cast
aluminum steering knuckle materials. The data points are from the
material cyclic tests. 133
Figure 4-6 Actual material cyclic stress-strain behavior and the corresponding
local von Mises stress and strain from FEA (shown with plus signs)
for (a) forged steel, (b) cast aluminum and (c) cast iron steering
knuckles. 134
Figure 4-7 From top to bottom, contours of von Mises stress at highest
moment levels for forged steel (Mmax = 1515 N.m), cast aluminum
(Mmax = 2230 N.m) and cast iron (Mmax = 2230 N.m) steering
knuckles. The stress values of the color bar are in MPa. 135
Figure 4-8 From top to bottom, contours of von Mises stress at lowest
moment levels for forged steel (Mmax = 825 N.m), cast aluminum
(Mmax = 1195 N.m) and cast iron (Mmax = 1195 N.m) steering
knuckles. The stress values of the color bar are in MPa. 136

xv
Figure 4-9 The distribution of stress at the spindle 2nd step fillet and at a
section remote from the 2nd step fillet of the forged steel steering
knuckle for a moment of 1515 N.m. The nominal stress profile was
used to obtain the nominal values of stress. 137
Figure 4-10 Signed von Mises stress history at different locations of the forged
steel steering knuckle. 138
Figure 5-1 Nomenclature used for forged steel (top) and cast aluminum
steering knuckles in this study. 150
Figure 5-2 Forged steel steering knuckle within the rear suspension system of a
4-cylinder sedan. 151
Figure 5-3 Cast aluminum steering knuckle within the front suspension system
of a 6-cylinder minivan. 152
Figure 5-4 Simulated primary loading and restraints on the forged steel steering
knuckle for component testing. 153
Figure 5-5 Simulated primary loading and restraints on the cast aluminum
steering knuckle for component testing. 153
Figure 5-6 Exploded view of the forged steel steering knuckle test fixture: 1)
steering knuckle, 2) spindle grip, 3) moment arm, 4) side blocks, 5)
end bar, 6) test bench, 7) load actuator, 8) load cell, 9) rod-end
bearing and pin, 10) needle roller bearings. 154
Figure 5-7 Schematic drawing (left) and the fixture for forged steel steering
knuckle test arrangement. 154
Figure 5-8 Exploded view of the cast aluminum steering knuckle test fixture: 1)
steering knuckle, 2) body fixing blocks, 3) arm grip pins, 4) moment
arm adjusting pins, 5) moment arm, 6) test bench, 7) load actuator,
8) load cell, 9) rod-end bearing and pin, 10) needle roller bearings. 155
Figure 5-9 Cast aluminum steering knuckle test arrangement showing fixturing
schematic (left), and arm fixturing close up. 155
Figure 5-10 Load application through rod-end bearing and a pair of needle
roller bearings to avoid out-of-plane bending and horizontal friction
force. 156
Figure 5-11 Displacement amplitude versus normalized cycles for typical forged
steel and cast aluminum steering knuckles during tests. 156
Figure 5-12 Applied moment amplitude versus fatigue life curves for (a) forged
steel and (b) cast aluminum steering knuckles. 157
Figure 5-13 Superimposed stress amplitude versus fatigue life curves for forged
steel and cast aluminum steering knuckles. 158
Figure 5-14 Typical fractured (left) and fracture surface of the forged steel
steering knuckle. 158

xvi
Figure 5-15 Typical fractured (top) and fracture surface of the cast aluminum
steering knuckle. 159
Figure 6-1 Durability assessment methodology pursued in this study. 177
Figure 6-2 Superimposed experimental life and stress-life prediction using the
nominal stress approach for the forged steel steering knuckle. 178
Figure 6-3 Superimposed experimental life and local strain-life predictions
using nominal stresses as input for the forged steel steering knuckle. 178
Figure 6-4 Superimposed local stress amplitude versus experimental life and
predictions of stress-life model for (a) forged steel and (b) cast
aluminum steering knuckle using nonlinear FEA results. 179
Figure 6-5 Superimposed local strain amplitude versus experimental life and
predictions of strain-life model for (a) forged steel and (b) cast
aluminum steering knuckle using nonlinear FEA results. 180
Figure 6-6 Stress amplitude versus life curves based on the stress-life approach
for the forged steel, cast aluminum and cast iron steering knuckles. 181
Figure 6-7 SWT parameter versus life curves based on the strain-life approach
for the forged steel, cast aluminum and cast iron steering knuckles. 181
Figure 7-1 Manufacturing process flow chart for the steering knuckle. 217
Figure 7-2 Locating the forged steering knuckle on the V-block to turn
spindle. 218
Figure 7-3 Unmachined forging loaded on the fixtures to be milled. The
steering knuckle shown in this picture is slightly different from the
one being studied here. 218
Figure 7-4 Steering knuckle optimization stages followed in this study. 219
Figure 7-5 Steering knuckle optimization procedure. 220
Figure 7-6 Stress distribution under primary loading at the vicinity of top strut
attachment (top left), bottom strut attachment (top right), hub
mounting part (bottom left) and the upper curve of steering knuckle
body. 221
Figure 7-7 Pattern of the section to reduce the thickness in Stage I
optimization. 221
Figure 7-8 Actual design goal (solid line), the design goals upper bound
(dashed line) and the design goals lower bound (dotted line) versus
number of iterations. The mass includes the moment arms and the
unit is kg. 222
Figure 7-9 Stress limit in the optimized area versus number of iterations. The
stress unit is MPa. 222
Figure 7-10 Design variable (depth of the removed area) versus number of
iterations. The design parameter unit is mm. 223
Figure 7-11 Optimized section in Stage I optimization. 223

xvii
Figure 7-12 Dimensions of the redesigned portion of the steering knuckle’s
body in Stage I optimization. All dimensions are in mm. 224
Figure 7-13 von Mises stress distribution for an arbitrary moment of 690 N.m
before (top) and after optimizing for the removed section from the
body (first part of Section I optimization). The stress unit is MPa. 224
Figure 7-14 The original (left) and optimized dimensions of hub mounting and
lateral link in Stage I optimization. The dimensions are in mm. 225
Figure 7-15 von Mises stress distribution for an arbitrary moment of 690 N.m
before (top) and after optimizing for Stage I for the optimized area.
The stress unit is MPa. 225
Figure 7-16 The forged part showing the strut and hub mounting holes. 226
Figure 7-17 Stage I (left) and Stage II models with redesigned spindle. 226
Figure 7-18 Spherical roller bearing with tapered bore to be installed on the
tapered section of the redesigned spindle (SKF Catalog). 226
Figure 7-19 von Mises stress distribution on the spindle-redesigned steering
knuckle due to an arbitrary moment of 690 N.m. The stress unit is
MPa. 227
Figure 7-20 Dimensions of the original (top) and redesigned-spindle (bottom)
steering knuckles. The dimensions are in mm. 227
Figure 7-21 Front (left) and isometric views of Stage II optimization. 228
Figure 7-22 Cutting speed and feed rate (both indexes of machinability) versus
hardness of some steels (Trent and Wright, 2000). 228
Figure 7-23 Results of fatigue tests (R = -0.6) for induction hardened and non-
hardened truck stub axle (top) and depth of induction hardened
zone revealed by etching (bottom) (Schijve, 2001). 229
Figure 7-24 Spindle fillets as the highest stressed areas of the steering knuckle
are recommended for inducing compressive residual stress. 230
Figure 7-25 Effect of rolling of the notch root on fatigue behavior under
rotating-bending fatigue of 37CrS4 steel (Kloos et al., 1987). 230

xviii
Nomenclature

a material characteristic length


Ao , Af initial, final cross section area
b fatigue strength exponent
c fatigue ductility exponent
e engineering strain
e* modified nominal strain
E modulus of elasticity
E' midlife cyclic modulus of elasticity
%EL percent elongation
HB, HRB, HRC, HV Brinell, Rockwell B-Scale, Rockwell C-Scale, Vickers hardness number
K strength coefficient
K' cyclic strength coefficient
Kf fatigue notch factor
Kt elastic stress concentration factor
Ktq equivalent stress concentration factor
Lo , Lf initial, final specimen gage length
M moment
Ma moment amplitude
n strain hardening exponent
n’ cyclic strain hardening exponent
N50%, (Nf)10%, (Nf)50% number of cycles to midlife, to 10% load drop, to 50% load drop
Nf cycles to failure
P load
Pf load at fracture
Pp plastic limit load
Pu ultimate load
%RA percent reduction in area
r radius of the notch
S engineering stress, nominal stress
S* modified nominal stress
Sa nominal stress amplitude
Sf fatigue strength
SNf equivalent completely reversed nominal stress amplitude
Su ultimate strength
Sy yield strength
S'y cyclic yield strength

xix
to , tf initial, final specimen thickness
wo , wf initial, final specimen width
Δε strain range
Δεe elastic strain range
Δεp plastic strain range
Δσ stress range
ε total strain
ε1 , ε2 , ε3 maximum, middle, minimum principal strain
εa strain amplitude
εeq elastic-plastic notch equivalent strain
ε
e eq elastically calculated notch equivalent strain
εe elastic strain
εf true fracture ductility
ε 'f fatigue ductility coefficient
εm mean strain
εp plastic strain
εx ,εy ,εz ,εxy ,εxz ,εyz local components of strain tensor
εVM local von Mises strain
γmax local maximum shear strain
σ true stress, local stress
σ1 ,σ2 ,σ3 maximum, middle, minimum principal stress
σa local stress amplitude
σe1 , σe2 , σe3 local elastic principal stresses
σeq elastic-plastic notch equivalent stress
σ
e eq elastically calculated notch equivalent stress
σf true fracture strength
σ 'f fatigue strength coefficient
σmax local maximum stress
σm local mean stress
σNf equivalent completely reversed local stress amplitude
σVM local von Mises stress
σx ,σy ,σz ,τxy ,τxz ,τyz local components of stress tensor
τmax local maximum shear stress

xx
Chapter One

1 Introduction
Manufacturing processes face major competitions in automotive industry to produce

lighter, cheaper and more efficient components that exhibit more precise dimensions, need

less machining and require less part processing. Technology leaders follow two main routes

to improve these processes; development and integration of currently available technologies,

and invention of new technologies optimized with respect to various design or commercial

aspects. The value of the know-how depends on a process’s ability to usefully differentiate

its capabilities from those of its competitors. Material mechanical properties and

manufacturing parameters play decisive roles and the weaknesses and strengths of each

manufacturing process need to be available to designers in these respects, to enable them to

choose the optimum choice for the specific component and application.

In automotive industry, designers have a wide range of materials and processes to

select from. Steel and aluminum forgings and castings, cast irons, and powder forgings have

found broad applications in automotive safety-critical systems. The competition is

particularly acute in the chassis, and it is not unusual to find a range of different materials

and manufacturing technologies employed within modern chassis components.

Many safety-critical components in the vehicle experience static as well as time-

varying loadings, and obviously they undergo the latter during a major portion of their

service life. However, material selection for these components made by various

manufacturing techniques is often based on monotonic rather than cyclic properties. Fatigue

1
2

behavior is, therefore, a key consideration in design and performance evaluation of

automotive components, and to address the issue effectively and economically, engineers

need to model and design for mechanical fatigue early in the product design stage.

In automotive design, durability evaluation of components based on experimental

assessments is time-consuming and expensive, so analytical approaches that include limited

number of component verification tests have gained more attention. A problem that arises at

the fatigue design stage of such components is the transferability of data from smooth

specimens to the component. The component geometry, loading, and manufacturing

process parameters such as surface conditions often deviate from that of the specimen

investigated and neither a nominal stress nor a notch factor can be defined in most cases.

In automotive industry, the significant increase of the demand for lighter, more fuel

efficient vehicles, reduced design-testing iterations, and satisfactory reliability level has

promoted the adoption of optimum materials and components. The studies on optimization

have focused on three aspects; geometry optimization, alternative materials and

manufacturing process modifications. From the viewpoint of geometry and considering the

boundary conditions, an optimized component could be obtained using different methods of

geometry optimization. On the other hand, alternative materials are being put into trial and

manufacturing processes are being re-evaluated to achieve lighter, cheaper and more efficient

components.

1.1 Motivation
This research was motivated by a practical need to assess and compare fatigue

performance of components produced by competing manufacturing processes, to develop a

general durability assessment methodology for automotive chassis (and similar) components,
3

and to implement an optimization methodology that incorporates structural durability

performance, material properties, manufacturing and cost considerations for such

components.

In automotive industry, the evaluation of manufacturing processes with regards to

structural performance is often performed simply based on monotonic rather than cyclic or

fatigue characteristics. However, the stress-strain behavior obtained from a monotonic

tension or compression test can be quite different from that obtained under cyclic loading.

Although fatigue is one of the main parameters that differentiate among competing

manufacturing processes and one of the major responsible phenomena that cause

mechanical failure in components, it is often overlooked by design engineers, resulting in

inefficient design methods and many over-designed parts that lack optimization potential.

Accordingly, a comparison of competing manufacturing processes regarding fatigue

performance that also considers cost of manufacturing was one of the main motivations of

this study.

The analytical approach combined with a limited number of component testing

reduces design cycle time due to reduced testing, allows inexpensive evaluation of changes in

geometry, material, loading and manufacturing process through performance simulation, and

finally provides evaluation techniques for product optimization and failure analysis. An

advantage of the limited number of component tests is that the effects of material,

manufacturing process parameters, and geometry are inherently accounted for, even though

synergistically. This dissertation was partially motivated to implement the principles of this

analytical approach to fatigue life assessment of automotive components. In this respect,

developing an optimization methodology for these components, as an effective tool for

design modification, becomes feasible.


4

The common practice in optimization of automotive components in the automotive

industry is to focus on geometry, material or manufacturing parameters individually. An

approach that incorporates these three main aspects, even though challenging, has the

advantage of enabling the designer to evaluate a design in a comprehensive manner, to

investigate the cost parameters and to determine practical optimization potentials. Therefore,

the third motivating concept for this study was the need to investigate optimization of

safety-critical automotive components considering the combined effects of geometry,

material and manufacturing parameters and subsequently, cost.

1.2 Objectives
The overall objectives of this research program were:

1. To assess fatigue life and compare fatigue performance of competing manufacturing

processes;

2. To develop a general durability assessment methodology for safety-critical automotive

components;

3. To develop a method for efficient and reliable optimization of such components that

satisfies performance criteria and considers geometry, material, manufacturing

parameters and costs.

1.3 Scope of this Study and Overview of the


Dissertation
This study involves several main topics; these are 1) a background study on forging

and its competing manufacturing processes, and vehicle engine and chassis components that

are produced by these competing processes discussed in Chapter 2, 2) a literature review that

focuses on comparison of competing manufacturing processes, and durability assessment


5

and optimization of automotive components also presented in Chapter 2, 3) experimental

work including specimen and component testing, and 4) analytical work including durability

assessment and optimization analysis. In light of the high volume of forged steel vehicle

components, the forging process was considered as the base for comparison of competing

manufacturing processes.

Vehicle steering knuckles of three materials/processes were selected as the example

parts for this study. These included forged steel SAE Grade 11V37 steering knuckle of the

rear suspension of a 4-cylinder sedan weighing 2.4 kg, cast aluminum ASTM A356-T6

steering knuckle of front suspension of a 6-cylinder minivan weighing 2.4 kg, and cast iron

ASTM A536 Grade 65-45-12 steering knuckle of the front suspension of a 4-cylinder sedan

weighing 4.7 kg. Figure 1-1 shows the three components.

1.3.1 Experimental Work


Strain-controlled monotonic and fatigue tests of specimens made of forged steel, cast

aluminum and cast iron steering knuckles based on ASTM standard test methods and

recommended practices were conducted. From these experiments, monotonic as well as

baseline cyclic deformation and fatigue properties of the three materials were obtained. The

data obtained made it possible to compare deformation response, fatigue performance, and

failure mechanisms of the base materials, without introducing the effects and interaction of

complex design parameters. In addition, the required baseline data for life prediction analysis

to predict component fatigue life and performance under actual service loading conditions

became available. Chapter 3 discusses the specimen testing methods, results and

comparisons of monotonic and fatigue behaviors of the three materials.


6

Load-control component tests for the forged steel and cast aluminum steering

knuckles were conducted. Such data provides a direct comparison between fatigue

performances of the components made of competing processes. The comparison inherently

includes design effects such as surface finish, component size, residual stress, and stress

concentration. In addition, the component test results made it possible to verify the

analytical durability assessment. Chapter 5 explains the pre-test analysis, test configuration,

and post-test analysis related to component testing.

1.3.2 Analytical Work


The analytical work in this study consisted of finite element analysis (FEA),

durability assessment and optimization analysis. Linear and nonlinear finite element analyses

of the steering knuckles were conducted to obtain critical locations of, and stress and strain

distributions of each component. Details of geometry generation, boundary conditions,

mesh specification, nonlinear material model, model solution and post processing are

detailed in Chapter 4.

A general life prediction methodology for the subject components was developed

and is described in Chapter 6. Material monotonic and cyclic data and results of the FEA

were used in life prediction methods applicable to safety-critical automotive components.

The strengths and shortages of each method are discussed.

Based on the results of analyses and testing performed, an analytical optimization

study of the forged steel steering knuckle was performed. Such optimization sought to

minimize weight and manufacturing costs while maintaining or improving fatigue strength of

the component by targeting geometry, material and manufacturing parameters. The

optimization methodology, manufacturing considerations and alternative materials are


7

discussed in Chapter 7 and recommendations are made on the optimized geometry.

Consequently, Chapter 8 summarizes the results of this study.


8

Forged steel Cast aluminum Cast iron


SAE Grade 11V37 ASTM A356-T6 ASTM A536
steering knuckle of steering knuckle of Grade 65-45-12
rear suspension of front suspension of a steering knuckle of
a 4-cylinder sedan 6-cylinder minivan front suspension of a
weighing 2.4 kg weighing 2.4 kg 4-cylinder sedan
weighing 4.7 kg

Figure 1-1 From left forged steel, cast aluminum and cast iron steering knuckles selected
as example parts for this study.
Chapter Two

2 Background and Literature Review


The background study includes two main parts. The first part provides an

introduction to forging process and its parameters as a process that possesses high volume in

automotive components. The second part gives an overview of vehicle power train and

suspension components that are the subject of competition among manufacturing processes,

and the service conditions of steering knuckle as the example part of this study. The

literature review covers a number of studies on comparison of competing manufacturing

processes with a focus on material mechanical behavior, life prediction methods

implemented in automotive components, and studies on optimization of steering knuckle

and similar components with manufacturing and material considerations. A brief report on

the literature review is also presented in Fatemi and Zoroufi (2002).

2.1 Forging Process and the Influencing Parameters


Bulk metal forming, which could also generally be called forging, denotes a family of

processes by which plastic deformation of the work piece is carried out by compressive

forces. In automotive industry, forged components are commonly found at points of

alternating and impact stresses such as steering knuckles, spindles, kingpins, axle beams and

shafts, torsion bars, ball studs, idler arms, pitman arms and steering arms. Another common

application is in the power train, where crank shafts, camshafts, connecting rods,

transmission shafts and gears, differential gears, clutch hubs and universal joints are often

9
10

forged. Although typically forged from carbon or low-alloy steel, other materials such as

aluminum and high strength low-alloy (microalloyed) steels are being increasingly used in

forged automotive applications.

Figure 2-1 shows the sequences involved in forging process. In this process, a slug or

billet is prepared, or the workpiece is preformed by shearing, sawing, or cutting off, either

cold or hot. If necessary, the surfaces are cleaned by such means as shot blasting. For hot

forging, the workpiece is heated in a furnace or by induction and, if necessary, descaled after

heating. Descaling may also occur during the initial stages of forging when the scale, which is

usually brittle, falls off during plastic deformation of the part. For hot forging, the dies are

preheated and lubricated and for cold forging, the blank is lubricated.

The workpiece is forged in appropriate dies and in the proper sequence. The excess

material such as flash is removed by trimming, machining or grinding. The forging is

cleaned, its dimensions are checked and, if necessary, it is totally or partly machined to final

dimensions and tolerances. Additional operations such as heat treating are performed to

obtain the desired mechanical and metallurgical properties. Any finishing operations that

may be required are conducted too. Finally the forging is inspected for any internal and

external defects or imperfections. Different forging methods exercised are discussed in the

following.

2.1.1 Forging Processes


The methods contemporarily practiced for forging process are usually categorized as

open-die forging, impression-die forging, and miscellaneous forging operations including

cold forging, ring forging, etc.


11

Open-Die Forging
Open-die forging, also called upsetting, is performed when a workpiece is placed on

a lower die and its height is reduced by the downward movement of the top die. Friction

between the end faces of the workpiece prevents the free, lateral spread of the ends of the

workpiece and results in a barrel shape. Open-die forging is distinguished by the fact that the

metal is never completely confined or restrained, and that the dies used are rather simple and

universal. All types of hammers or presses may be used in open-die forging. Forgings are

made by this method if (a) the forging is too large to be produced in closed dies; under open

dies it is produced in many steps by forging only a part of it in each step, or (b) the quantity

required is too small to justify the cost of complex closed dies.

Practically all of forgeable ferrous and non-ferrous alloys can be open-die forged,

including age-hardening superalloys and corrosion-resistant refractory alloys. Open-die shape

capability is wide and this process can produce step shafts, hollows cylindrical in shape, ring-

like parts, and contour-formed metal shells like pressure vessels. Multiple open-die forging

operations can be combined to produce the required shape, or these forging methods can be

tailored to attain the proper amount of total deformation and optimum grain-flow structure,

thereby maximizing property enhancement and ultimate performance for a particular

application. Forging an integral gear blank and hub, for example, may entail multiple drawing

or solid forging operations, and then upsetting. Similarly, blanks for rings may be prepared

by upsetting an ingot, then piercing the center, prior to forging the ring.

Impression-Die Forging
In the simplest form of impression-die forging, commonly referred to as closed-die

forging, a cylindrical or rectangular workpiece is placed in the bottom die. The dies contain

no provision for controlling the flow of excess material. Figure 2-2 shows impression-forged
12

connecting rods along with the matching dies. As the two dies are brought together, the

workpiece undergoes plastic deformation until its enlarged sides touch the sidewalls of the

die impressions. At this point a small amount of material begins to flow outside of the die

impressions, forming flash. In the further course of the die approach, this flash is thinned

gradually. As a consequence, it cools rapidly and presents increased resistance to

deformation. In this sense, the flash becomes a part of the tool and helps to build up high

pressure inside the bulk of the workpiece. This pressure can aid material flow into parts of

the impression previously unfilled so that, at the end of the stroke, the die impressions are

nearly filled with the workpiece material.

Impression-die forging can produce a variety of three dimensional shapes with a

wide range in weight. Because metal flow is restricted by the die contours, this process can

yield more complex shapes and closer tolerances than open-die forging processes. Additional

flexibility in forming symmetrical and non-symmetrical shapes comes from various pre-

forming operations (sometimes bending) prior to forging in finisher dies. Part geometries

range from simple spherical shapes, block-like rectangular solids, and disc-like configurations

to components with thin and long sections that incorporate thin webs and relatively high

vertical projections like ribs and bosses. Although many parts are generally symmetrical,

others incorporate all sorts of design elements (flanges, protrusions, holes, cavities, pockets,

etc.) that combine to make the forging very non-symmetrical. In addition, parts can be bent

or curved in one or several planes, whether they are basically longitudinal, equi-dimensional

or flat.

Most engineering metals and alloys can be forged via conventional impression-die

processes, among them carbon and alloy steels, tool steels, and stainless, aluminum and

copper alloys, and certain titanium alloys. Strain-rate and temperature-sensitive materials
13

(magnesium, nickel-based alloys, refractory alloys and some titanium alloys) may require

more sophisticated forging processes or special equipment for forging in impression dies.

Miscellaneous Forging Operations

Most forging is done as hot work, at temperatures up to 1300°C; however, a

variation of impression-die forging is cold forging. The temperature of metals being cold

forged may range from room temperature to several hundred degrees. Cold forging

encompasses many processes: bending, cold drawing, cold heading, coining, extrusion,

punching, thread rolling and more, to produce a diverse range of part shapes. These include

various shaft-like components, cup-shaped geometries, hollow parts with stems and shafts,

all kinds of headed and bent configurations, as well as combinations. Material options range

form low alloy and carbon steels to stainless steels, selected aluminum alloys, brass and

bronze. There are times when warm forging practices are selected over cold forging

especially for higher carbon grades of steel or where annealing can be eliminated.

Seamless rolled ring forging is another forging process and is typically performed by

punching a hole in a thick, round piece of metal (creating a donut shape), and then rolling

and squeezing (or in some cases pounding) the donut into a thin ring. Rings forged by the

seamless ring rolling process can weigh less than 0.5 kg up to 150 tons, while outer diameters

range from a few centimeters up to 10 m in diameter. Other miscellaneous forging

operations include isothermal and hot die forging, rolling, radial forging, etc.

2.1.2 Important Factors in Forging Process


Forgeability
The forgeability of a metal can be defined as its capability to undergo deformation by

forging without cracking or defects. This definition can be expanded to include the flow
14

strength of the metal. Thus a material with good forgeability is one that can be shaped with

low forces without cracking. A commonly used test of forgeability is to upset a solid

cylindrical specimen and observe any cracking on the barreled surface. The greater the

deformation prior to cracking, the greater the forgeability of the metal (Kalpakjian and

Schmid, 2000). If notch sensitivity of the material is high, surface defects will affect the

results by causing premature cracking.

In another test of forgeability called hot-twist test, a round specimen is twisted

continuously in the same direction until it fails. The test is performed at various

temperatures, and the number of turns that each specimen undergoes before failure is

observed. The optimal forging temperature is then determined. This test is particularly useful

in determining the forgeability of steels, although upsetting tests can also be used for that

purpose. Small changes in the composition of or impurities in the metal can have a

significant effect on forgeability.

Hydrostatic pressure has a significant beneficial effect on the ductility of metals and

nonmetallic materials. Experiments have indicated that the results of room-temperature

forgeability tests are improved, i.e. cracking takes place at higher strain levels, if the tests are

conducted in an environment of high hydrostatic pressure. Based on the results of various

tests and observations, the forgeability of several metals and alloys is determined and shown

in Table 2-1. It is based on such considerations as the ductility and strength of the metal,

forging temperature required, frictional behavior and quality of the forging obtained.

Lubrication
During the deformation phase of a conventional hot forging operation in which the

die temperature is much less than the billet temperature, the lubricant provides three basic

effects (Byrer et al., 1985); lubricity, ensuring that the correct coefficient of friction is
15

maintained between the hot billet and the tooling; physical barrier, preventing physical

contact between billet and dies; and thermal insulation, retarding the rate of heat transfer

between the hot metal and the dies.

The degree of metal flow affects all of these effects. The manner in which a part is

forged can also affect the proper functioning of a lubrication system. Lubricants can be

classified into four groups, although a lubricant compound can be a blend of the first two

(or the first three) categories to obtain the desired performance or to improve performance;

Solids, meltable pigments, organic chemicals and water solubles.

Surface Finish
Surface finish of the forging depends on the effectiveness of the lubricant,

preparation of the blank, die surface finish, and die wear. As described by Rice (1997), in

addition to bulk effects, forgings subjected to cyclic or repeated service loading are

significantly influenced by surface conditions. Many forgings are used in service with some

or the entire as-forged surface remaining. It follows that the fatigue life is reduced by

features such as forging laps, folds, cracks and other imperfections that can act as crack

initiation sites. Surface metallurgical features can also be important. For example, steels may

show decarburization and other alloys exhibit changes in chemistry and phase distribution

resulting from the heating and cooling cycles associated with forging and heat treatment.

Heat Treatment
Heat treatment involves specific controlled thermal cycles of heating and cooling to

improve one or more properties of the forged part. The primary objectives of such

treatment may be to relieve internal stresses (as in tempering and stress relieving), control

distortion, optimize the depth of hardening (hardenability), or develop the final specification
16

for mechanical or physical properties, because the forged component must meet

performance engineering design requirements and ensure safety and reliability in service.

Underlying the selection of the proper treatment are cost considerations that must

factor in the variables of forging shape and size, metal composition, machining, and final

properties. For example, it would be wasteful of time and energy to reheat a forging when

cooling from forging heat would satisfy the specification, or to add a temperature cycle when

normalizing suffices.

Anisotropy
During the forging process, nonuniformities in alloy chemistry, second-phase

particles, inclusions, and the crystalline grains themselves are aligned in the directions of the

greatest metal flow. The directional pattern of the crystals following working is known as the

grain-flow pattern. Grain flow produces directional characteristics in properties such as

strength, ductility, and resistance to impact and fatigue. The grains could be oriented in the

forged part to lie in the direction that maximum strength is needed. The important

anisotropies are in ductility, notched-bar impact strength, and fatigue. The resistance to

stress corrosion cracking can be highly directional, for example, in high-strength aluminum

alloys. The die sequence and the designs of the dies may be altered in order to control the

grain-flow pattern. Forging and lubrication techniques can also be used for this purpose

(Byrer et al., 1985).

2.2 Vehicle Engine and Suspension


For a typical motor vehicle, the main competition among manufacturing processes is

in the components of power train and suspension system. Typical forged components used

in vehicles are crankshaft, connecting rod, camshaft, and suspension components such as
17

control arm, steering knuckle and wheel hub, as shown in Figure 2-3. In order to have a

better understanding of the vehicle components, the technical features of vehicle engine and

suspension system components and the steering knuckle in particular, are briefly reviewed

here.

2.2.1 Vehicle Engine


The main components of a typical engine, as shown in Figure 2-4, are cylinder block

and crankcase, pistons and rings, connecting rods and bearings, crankshaft assembly and

bearings, cylinder head and gasket, camshaft, valve train and timing drive, and engine

support mountings. Among these components, manufacturing processes compete over

power train components, mainly crankshaft, transmission shaft, camshaft and connecting

rod. Most vehicle engine components operate at certain speeds for very high number of

cycles.

2.2.2 Suspension System


Suspension system consists of the springs and related parts intermediate between the

wheels and the frame, subframe, or side rails of a unitized body. The suspension supports

the weight of the upper part of a vehicle on its axles and wheels, allows the vehicle to travel

over irregular surfaces with a minimum of up-and-down body movement, and allows the

vehicle to corner with minimum roll or loss of traction between the tires and the road. Four

types of springs used in automotive suspension are coil, leaf, torsion bar, and air spring.

In a typical suspension system for a vehicle with front-engine and front-wheel-drive

(Figure 2-5), the weight of the vehicle applies an initial compression to the coil springs.

When the tires and wheels encounter irregularities in the road, the springs further compress
18

or expand to absorb most of the shock. The suspension at the rear wheels is usually simpler

than for the front wheels, which require multiple-point attachments so the wheels can move

up and down while swinging from side to side for steering.

A telescoping hydraulic damper, known as a shock absorber, is mounted separately

or in the strut at each wheel to restrain spring movement and prevent prolonged spring

oscillations. The shock absorber contains a piston that moves in a cylinder as the wheel

moves up and down with respect to the vehicle body or frame. As the piston moves, it

forces a fluid through an orifice, imposing a restraint on the spring. Spring-loaded valves

open to permit quicker flow of the fluid if fluid pressure rises high enough, as it may when

rapid wheel movements take place. As categorized by Birch (2000), there are essentially five

types of suspensions used on cars, pickups and trucks:

Short-Long Arm (SLA) Suspension: This is the typical RWD car’s front suspension. It

consists of two control arms (a short upper arm and a longer lower arm), a steering knuckle

with spindle, and the necessary bushings and ball joints (Figure 2-6). The outer ends of both

control arms connect to the steering knuckle, which includes the spindle, through a ball

joint.

Multilink Suspension: Some SLA designs have evolved so the steering knuckle has

become taller, about to the top of the tire, and the spring is strut mounted over the shock

absorber, as shown in Figure 2-7.

MacPherson Strut Suspension: This arrangement has no upper control arm or upper ball

joint. The steering knuckle connects to a spring and shock absorber assembly, which is the

strut. The upper end of this assembly connects to the car body through a pivot-damper unit.

A lower control arm is used; it serves the same purpose as the lower arm of an SLA

suspension (Figure 2-8).


19

Solid Axles Suspension: This kind of suspension is commonly used on trucks, 4WDs, and

some pickups because of their simpler, stronger, and less expensive construction. In

addition, they have traditionally been used for rear suspension. A solid axle is simply a

strong, solid beam of steel (usually I shaped) with a kingpin at each end to connect to the

steering knuckle (Figure 2-9).

Swing Axles Suspension: Ford Motor Company has used this kind of suspension on its

pickups, 4WDs, and light trucks. Twin I-beam axles combine some of the sturdiness and

simplicity of a solid axle with some of the improved ride and handling characteristics of an

independent suspension. A twin I-beam axle is a compromise between these two suspension

types (Figure 2-10).

2.2.3 Steering Knuckle


Suspensions use various links, arms, and joints to allow the wheels to move freely up

and down; front suspensions also have to allow the front wheels to turn. All suspensions

must provide transverse (or side to side) as well as longitudinal (front-to-back) wheel

support. Steering knuckle/spindle assembly, which might be two separate parts attached

together or one complete part, is one of these links. Its geometry depends on the type of

suspension.

Figure 2-6 shows the assembly of the steering knuckle and spindle on an SLA, which

is the typical rear-wheel-drive car’s front suspension. The Multilink Suspension, in which the

long curved steering knuckle and angled upper control arm allow its use in areas of limited

size, is shown in Figure 2-7. The steering knuckle for a MacPherson Strut Suspension, which

has no upper control arm or upper ball joint, is shown in Figure 2-8. In the Solid Axle

Suspension, a beam of steel is connected to the steering knuckle, as shown in Figure 2-9.
20

The arrangement of steering knuckle/spindle in the Swing Axle Suspension could be seen in

Figure 2-10. In spite of different configurations of the steering knuckle/spindle assembly for

each type of vehicle suspension, the assembly is intended to play a common role in all type,

and that is to accommodate the service loading.

Haeg (1997) investigated the simulation of steering system dynamics and studied

steering system inputs and responses. As he states, there are many factors that influence

steer motions and forces in a vehicle suspension. These range from the kinematics of the

suspension itself to its response to severe off-center lateral impact loading such as an oblique

curb strike event. It must be remembered that even axles that are not steered, still experience

steer axis-related forces during normal vehicle operation.

Most forces are imparted to the vehicle suspension via the tire patch. Figure 2-11

shows a static lateral force input at the tire patch center. The primary lateral reaction paths

are through the upper and lower horizontal restraints. Very little force is reacted by the

steering system if the static lateral force input vector intersects the line of the steer axis.

Figure 2-11 also shows static longitudinal force applied at the tire patch center. When the tire

patch or face of the tire encounters resistance or longitudinal load, it is resolved into a force

and a moment about the axis of rotation of the tire/wheel (spindle center). The force is

resolved at the centerline of the spindle itself. This moment (torque) can be reacted by the

drive-train or brakes, or it can rotationally accelerate the tire.

In most suspensions the steer axis is well inboard (towards the center of the vehicle),

therefore longitudinal force reacted at the spindle center imparts a moment about the steer

axis. The steer moment is reacted via opposing forces (a couple) at the horizontal restraints

and the tie rod end, as illustrated in Figure 2-12. The tie rod forces are reacted back through

the steering gear to the body or frame. The distribution of these reaction forces is a function
21

of the relative stiffness of the various mounts and components, the steer angle, and loading

of the suspension on the opposite side of the vehicle.

2.3 Comparison of Manufacturing Processes


Various mechanical and metallurgical properties, environmental considerations, and

above all cost competitiveness aspects are the main driving forces shaping the future

direction of the manufacturing processes. In the ground vehicle industry conventional

forged components include crankshaft, camshaft, connecting rod, piston crown, steering

lever, suspension arm, steering knuckle, wheel hub, drive flange and axle beam. Some of

these components are also manufactured by die-casting, and more recently by powder

forging and composite technologies. Powder metallurgy (P/M) processes using sintering,

which can offer net-shaped products, have also been used, though having limitations for

large parts and parts with challenging geometries. Composites offer lightweight and

directional properties, but are comparatively expensive.

Mechanical properties of the manufactured component are influenced by its

manufacturing process. As an example, the forging process is investigated in this study to

identify the process parameters that affect mechanical properties and material behavior of

the forged component and the result is a chart shown in Figure 2-13. The chart has four

columns; the first column includes the process influential parameters. Each parameter and its

connector lines are coded with the same color. The second column entails the mechanical

and metallurgical parameters that play a bridge role between the process and mechanical

properties. Defects, surface finish, residual stresses generated in the component and the

microstructure of the workpiece material are those bridge parameters affected by the

process. The third and fourth columns cover the general mechanical properties and the
22

breakdown of monotonic, cyclic and fatigue properties, respectively, of the workpiece

material that are subsequently affected.

A number of selected examples from literature are provided in the following to

indicate the behavior comparison of competing manufacturing processes.

Casting vs. Forging


Gunnarson et al. (1987) investigated replacing conventional forged quenched and

tempered (Q&T) steel with precipitation hardened pearlitic-ferritic cast steels for connecting

rod, steering knuckle, crank shaft, control arm and other automotive components. They also

compared toughness and machinability characteristics of forging versus casting components.

They observed insufficient toughness but higher machinability for cast components. For the

case of the vehicle steering knuckle, equivalent fatigue strength was found for the cast

steering knuckle, but only with increased spindle diameter and reduced length, as compared

with the forged steel steering knuckle.

The positive trend for the application of cast components is mainly due to lower cost

incentives. However, weaker mechanical properties of cast components due to a wide variety

of flaws and their low ductility have always been a matter of concern. Houshito et al. (1989)

performed a feasibility study on the application of high strength ductile iron to automotive

chassis parts, namely steering knuckles. They state that while the shape of forged part is

limited by the manufacturing process, the shape of casting part can be optimized by

balancing the stress distribution. They intended to reduce weight and cost of the steering

knuckle by replacing the forged part by a cast part with optimized shape and comparable

strength. In this regard, stress and rigidity analyses and fatigue and impact experiments were

conducted on the steering knuckle. They concluded that the Young’s modulus of castings
23

was lower than that of forging by 20%, and fatigue strength of the cast steering knuckle was

lower than the original forged steering knuckle by 23%.

Fatigue crack growth resistance can also be an important consideration, when

evaluating fatigue performance. Fatigue crack growth behavior of Q&T steels, which are

often used to produce forged products, is compared to cast steel in Figure 2-14 (Rice, 1997).

The Q&T 4140 steel exhibits superior fatigue crack growth resistance compared to the cast

SAE 0030 steel. Machined bars and plates may be more susceptible to fatigue and stress

corrosion because machining cuts material grain pattern. When a fabricated component has

a high number of inches of weld in critical stress area, a greater chance exists that problems

in the weld itself or from micro structural defects in the neighboring heat affected zone will

cause failure in the field. High-stress welded joints are generally less capable in overloads or

cyclic fatigue than junctions formed in the forging or metal casting process.

Traditionally, forged components have been produced from heat treated carbon and

low alloy steels. Although heat-treated steels are still widely used, air cooled forging steels are

becoming increasingly popular. Through a reduction in energy consumption, fewer process

steps and lower inventories, these forgings can offer significant cost savings. Cristinacce et

al. (1998) provide some recent examples of the range of components produced from air-

cooled forging steels. An air-cooled 0.53%C steel was used in the production of hubs and

spindles, where an approximately 400% increase in hardness was achieved compared to the

heat-treated forgings. In another case, a swivel hub was redesigned as a forging in place of

steel casting that had been proposed originally. The steel casting exhibited unacceptable

distortion of the steering arm in heat treatment, surplus material leading to high machining

cost, and excessive weight affecting the unsprung mass of the suspension design. The

redesigned swivel hub was forged and control air-cooled. The results of mechanical tests on
24

both the forging and the casting are given in Table 2-2. The forging had superior strength,

ductility, and hardness, compared to the heat-treated casting. The use of the forging also

resulted in lower weight by 21%, better dimensional control, less machining, and avoidance

of heat treatment costs.

van Bennekom and Wilke (2003) compared forged and cast stainless steels for

physical, mechanical and corrosive properties. They concluded that “the forged components

exhibited better corrosion resistance, better mechanical properties, superior machinability

and surface finish and were more suitable for non-destructive testing. However, more

complex components could be obtained by castings and the cost per component produced

was lower. For the specific case of fatigue properties, steel cleanliness, uniformity of the

grain size and microstructure, being free from segregation, and being free from slag and

other metallic or non-metallic inclusions were included as factors that affected the fatigue

resistance of cast and forged steel components. The effect of surface finish was ignored,

since most components were machined to attain the final dimensions and tolerance after

casting or forging.” It should, however, be noted that machined forged surfaces have

superior fatigue resistance as compared to machined castings. This is because casting

porosities, which significantly affect fatigue behavior of castings, are still present after

machining.

van Bennekom and Wilke (2003) also added that “the cleanliness of forgings was

superior to the cleanliness of similar components that have been cast. The same applied to

the presence of slag and other metallic or non metallic inclusions. The grain size of forgings

was substantially finer than that of castings since the coarser solidification structure has been

destroyed and refined by the hot forming operation. Finer grain sizes resulted in a substantial

increase in the fatigue strength of the stainless steel. The fatigue properties were also
25

considered to depend to a large extent on grain orientation effects. During the solidification

process in castings, the grains structure was highly orientated in the transverse direction to

the surface since solidification started at the surface and then propagates inwards via the

growth of coarse columnar grains. This grain orientation was particularly detrimental from a

fatigue point of view and as such the forged components in which the grains were aligned

with the profile of the forging displayed superior fatigue properties.”

P/M Forging vs. Conventional Forging


P/M offers high precision and low weight tolerances leading to less machining

operations in comparison to classically forged or cast components. On the other hand,

fatigue resistance and toughness are not generally as good as steel forgings (Esper and

Sonsino, 1994). Jang et al. (2000) conducted a study on powder materials and production

processes by producing the clutch disk spline hub of automobile, to replace the existing

forged component. They also investigated mechanical properties and microstructure along

with the performance of torsional durability test of three types of powder materials. They

concluded that one of the produced P/M samples, which is a diffusion alloy powder and is

treated with carburizing -tempering, performed better in torsion durability tests and wear

resistance than that of existing forged steel component. They also concluded that toughness

could be improved to the same level as forged metal, if powder metal is sintered and treated

with adequate condition. There was no mention of cost and comparison with the forged

part, however.

In an investigation about P/M connecting rods, Whittaker (2001) states that powder

forged connecting rods are around 30% more expensive than the drop forged product at the

as-formed blank stage. However, the machining cost savings compared to a conventionally

split drop forging are sufficiently high to create a cost advantage of around 10% for the
26

powder forged product as a fully machined connecting rod. In the meanwhile, the recent

introduction of fracture splitting of drop forged steel connecting rods has reversed this cost

advantage, as the finished machined cost of drop forged fracture-split steel connecting rods

is 3% lower than fracture split powder forged one, but 19% higher than the single

press/sinter connecting rods. Another study by Repgen (1998) believes that the fracture-

cracking technology reduces production cost of connecting rods by 25% compared to the

conventional forging.

A Comparative study was performed by Afzal and Fatemi (2004) on fatigue

behavior of forged steel and P/M connecting rods including strain-controlled specimen tests

and component load-controlled bench tests. Specimen testing showed long-life fatigue

strength defined at 106 cycles of the forged steel to be 27% higher than that of the P/M, as

shown in Figure 2-15. This resulted in about an order of magnitude longer life for the forged

steel, as compared with the P/M. Forged steel and P/M connecting rod bench test results as

shown in Figure 2-16 indicated that the forged steel connecting rod exhibits 37% higher

fatigue strength, as compared with the powder metal connecting rod. This increased strength

resulted in about two orders of magnitude longer life for the forged steel connecting rod.

The difference in fatigue performance between the two connecting rods increased with

longer lives.

In general, the main advantages of P/M parts are reduction of waste material as well

as machining operations and low unit cost when mass-produced; while their main

disadvantages are high cost of dies, typically lower physical properties, higher cost of

materials, limitations on the design, and the limited range of materials which can be used.
27

Thixoforming vs. Forging


Thixoforming has come to play an increasing role as an alternative to forging,

especially in obtaining high-strength aluminum components for lightweight automotive

designs. In this process a semi-solid metal is injected into a closed die. Thixo-formed

aluminum components are often intended to replace steel forgings to form near-net shape

components, or nodular cast iron components to reduce the solidification shrinkage. Hirt et

al. (1997) used a pilot thixoforming system to redesign, thixoform and test an aluminum

steering knuckle, as a thin walled structural component subject to high loads, and to

compare the product with the original steel-forged steering knuckle. They combined steering

knuckle thick-walled and thin-walled areas with stiffening ribs and undercuts to gain

necessary yield strength, fracture toughness and stiffness. As a result, the weight of the new

part was 50% below that of conventional forged steel design, despite identical functional

capabilities.

There are numerous challenges for applying this technology to carbon and alloy

steels, however, according to the Steel Industry Technology Roadmap (AISI, 2002). These

include the high melting points and relatively small differences between liquidus and solidus

temperatures and the need for advanced mold/die materials to withstand the high

temperature & pressures. This makes the process more costly than conventional processes.

Process controls for steel cleanliness, chemical uniformity and microstructure consistency

throughout the section are also needed.

Comparisons of Multiple Processes


The influence of surface roughness and defects on fatigue life for various

manufacturing processes is a key consideration in durability performance. Processes such as

P/M have sometimes been more appealing, if they require less machining after production
28

(Blarasin and Giunti, 1997; Rice, 1997). Surface effects also include differences in

microstructure, chemical composition, and residual stresses. Figure 2-17 shows a comparison

of fatigue strength in various manufacturing processes for a front suspension arm. Even

though for the hot forged steel surface defects of blank surfaces reduce fatigue strength by

30% from that of the machined surface, the fatigue strength is still considerably better than

that of the nodular cast iron arm.

Surface enhancement processes like shot peening and shot blasting can affect fatigue

performance of components. In the previously-mentioned study by Afzal and Fatemi (2004)

on fatigue behavior of forged steel and P/M connecting rods, it was found that in the bench

test of the P/M connecting rods the crack originated either at the surface or subsurface,

while for the forged steel connecting rods cracks started subsurface. It was emphasized that

for the forged steel connecting rod, the S-N approach predictions are reasonable if the

predictions are based on smooth rather than forged surface finish. This was attributed to the

fact that the beneficial compressive residual stresses on the surface from the shot blasting

process nullify the detrimental effect of forged surface finish.

Strain hardening of the surface layer has a strong influence on fatigue behavior and

depends on the depth of the deformed layer. Surface decarburization can occur after hot

forging, which can cause different defects on the surface layer, reduce strain hardening of the

surface layer, and consequently reduce the fatigue strength. Excessive strain hardening

resulting from large deformations can also produce cracking and flaking of the surface and

significantly reduce the fatigue strength.

Each manufacturing process is capable of producing a part to a certain surface finish

and tolerance range without extra expenditure. Figure 2-18 compares the surface roughness

and tolerance range for a number of manufacturing processes. It shows that forging is
29

superior to sand casting and even die-casting in terms of producing parts to some

characteristic tolerance and surface finish, but inferior in comparison with machining

processes. Therefore, forgings and castings generally require additional finishing operations,

such as heat-treating, to modify properties, and then machining to obtain accurate finished

dimensions. For forged components, the cost of machining is typically 50% of the total cost

of the part (Naylor, 1998). Surface defects can be the source of fatigue failures, and they may

lead to such other problems as corrosion and wear during the service life of the component.

The forging process tends to reduce surface porosity and discontinuities (and may close up

small internal cavities). Surface porosity and discontinuities occasionally appear on steel

castings and require weld repair.

To compare directional properties, cast and powder forged parts do not exhibit grain

flow or directional strength. Because hot working refines grain pattern and imparts high

strength, ductility and resistance properties, forged products have lower possibility of

internal defects compared to castings and they are manufactured without the added costs for

tighter process controls and inspection that are required for casting. However, tensile

strength, elongation and impact properties in forged products decrease in the transverse

direction(s). Thus forgings are anisotropic. In castings, the metal is typically isotropic with

similar properties in all directions, although flow lines, porosity and several other casting

defects can show up depending on casting practice.

Figure 2-19 shows the variation in impact strength, yield strength and percent

elongation with direction. While the properties for the cast part remains constant, the impact

strength, reduction of area, and percent elongation change significantly from longitudinal to

transverse direction in the forged part, though tensile strength and yield strength remain

nearly constant. Additionally, the service conditions of the components must be carefully
30

evaluated. If the loading is uniaxial along the longitudinal axis, then the directionality of the

forging is an advantage. As the stresses become multiaxial, directionality becomes more

complex. In comparison to welding, selective heating and non-uniform cooling that occur in

welding can yield undesirable metallurgical properties like inconsistent grain structure that

could rarely be found in an appropriately forged component. In use, a welded seam may act

as a metallurgical notch that can lead to part failure.

Although forging and casting processes are capable of offering favorable shape

complexity, the level of dimensional tolerances drops as the component becomes more

complex. P/M offers good dimensional control and, in many instances, results in reduction

of machining and finishing operations; in this way it reduces scrap and waste and saves

energy. But the nature of P/M process imposes limitations on part size and shape

complexity. A problem with P/M parts is lack of proper density at sharp corners or stress

concentration areas, and for larger parts due to the limitations of the process. This is even

more prominent in more geometrically complex shapes.

Production quantity is another major factor. A cost comparison for a typical

automotive component (connecting rod) is provided in Figure 2-20. All other factors being

the same and depending on the number of pieces required, manufacturing a certain part by,

for example, expandable mold casting may well be more economical than doing so by

forging and on the other hand, for large quantities forging is more economical.

To conclude, forged products are compared to competitor cast and powder-forged

products with respect to a number of important mechanical properties and manufacturing

specifications in Table 2-3. Forged products exhibit higher strength, ductility and toughness

compared to casting and P/M parts and the fatigue crack growth resistance of the forged

parts is superior. Directional strength, that could be a favorable property if used properly,
31

only exists in forging. Internal defects exist in forged and cast products, while the oxygen

trapped in the powder during the compacting process of P/M parts results in porosity and

induces a detrimental effect on mechanical properties. From the economic perspective, the

P/M process has higher initial tooling cost. This process generates the least material waste

among the three processes, though this advantage can be overshadowed by higher cost of

raw material and the non-uniform distribution of density in P/M products. Surface finish is

another important factor where forged parts have better specifications compared to the

other two processes. Forgings have good response to surface enhancement processes; for

instance, the surface defects in forgings may be reduced and/or made harmless by shot

blasting process. Choosing the right process for manufacturing a component needs thorough

acquaintance with different aspects including part’s design criteria, service conditions, and

economic aspects.

2.4 Durability Assessment of Automotive Components


This part of the literature review is dedicated to the procedures for designing

automotive components for durability and fatigue. The results of selected studies on

determination of local stresses and strains, notch analysis, force and moment measurements,

multiaxial stress/strain, fatigue failure diagnosis and analysis guidelines, and component

fatigue test are briefly discussed. A number of methods and models implemented in

durability assessment of steering knuckle or similar fatigue-critical automotive components

are explained. A summary of the procedures of fatigue life prediction methods implemented

in literature for automotive components is provided in Appendix A.

The flow chart in Figure 2-21 shows the fatigue design process of components. In

this design process the designer gathers input data including geometry, loading history,
32

material properties, and environmental parameters. Implementing the design criteria, the

configuration, material and manufacturing processes of the component are selected. Stress

and strain analysis enables evaluating the critical locations of the designed component under

the assigned loading condition. Generally four fatigue life analysis models are commonly

used; the nominal stress-life (S-N) model, the local strain-life (ε-N) model, the fatigue crack

growth (da/dN-ΔK) model, and the two-stage model, which combines the second and third

model to incorporate both macroscopic fatigue crack formation (nucleation) and fatigue

crack growth. While nominal stress approach may fall short due to complexity of geometry

and loading in many cases, local stress or strain approach has been popularly used in the

automotive industry. When designing components subjected to occasional overloads,

particularly for notched components where cyclic plastic deformation can be significant,

cyclic ductility plays an important role. This is typical of suspension components, such as

steering knuckle, that are considered as fatigue critical parts.

The load history of an automotive component is typically variable amplitude and, for

instance in the case of suspension components, complex load spectrums exist. Choosing a

proper damage model is, therefore, the next step in fatigue design that accounts for the

cumulative effects of the cycles. The fatigue life calculated for the component is verified at

the next step by component or vehicle test and the component’s configuration, material and

manufacturing processes are modified in an iterative process to achieve the optimum design.

The methods for component durability assessment could be categorized into

experimental and numerical procedures as shown in Figure 2-22. While, historically, the

numerical pre-dimensioning was followed by experimental optimization of particular

components and experimental proof-out of a system consisting of different components, the

present industrial trend interacts these phases of product development with each other by
33

simultaneous engineering, Figure 2-23, in order to reduce time. This procedure can deliver a

reliable design only if the numerical assessment considers service experience and is

accompanied by experimental verification. The experimental methods comprise material

testing (stress or strain controlled), strain analysis and component testing (uni- or multiaxial);

numerical methods consist of stress/strain analysis by finite or boundary elements, the

simulation of dynamic system behavior with regard to service stresses, the simulation of

component strength properties and fatigue life calculations (Berger et al., 2002).

Blarasin and Farsetti (1989) adopted a model that utilizes the fatigue properties of

the material from specimen testing, nominal load history and notch factors to predict fatigue

life of similar steering knuckles made from Q&T and micro-alloyed steels. They assumed

that the fatigue behavior of the component is equivalent to that of the smooth specimen

with respect to material properties, metallurgical structure and surface conditions, subject to

a stress history corresponding to that acting in the critical zone of the component. The local

behavior of the material was described by utilizing the cyclic stress-strain curve. The damage

parameter utilized was the Smith-Watson-Topper (SWT) parameter. Component bench tests

were performed by subjecting the component to a time history of the primary load. The

results showed that predictions over- and under-estimate fatigue life for Q&T and micro-

alloyed steel steering knuckles, respectively. They attributed the discrepancy to an under-

estimate of the stress concentration factor value as well as to differences between the

microstructures of the components (where in the critical region, in a matrix of tempered

martensite quantities of ferrite were found at the austenitic grain boundaries) and the fatigue

specimens (characterized by homogeneous tempered martensite). Other sources of this

discrepancy could be the component critical location stress states and gradient, surface

condition, residual stresses, etc.


34

Conle and Mousseau (1991) used vehicle simulation and finite element (FE) results

to generate fatigue life contours for chassis components using automotive proving ground

load history results combined with computational techniques. They concluded that the

combination of vehicle dynamics modeling, finite-element analysis, and fatigue analysis is a

viable technique for the fatigue design of automotive components.

In an application of the local strain approach, Lee et al. (1995) developed a

methodology to quantitatively assess fatigue lives of automotive structures and to identify

critical and non-damaging areas for design enhancement and weight reduction. An MS-

3760A cast iron steering knuckle was the example component of this study. The

methodology combines the load-time history file with results from elastic finite element

analysis (FEA) to estimate fatigue lives. The load- and moment-time histories of the wheel

(Figure 2-24) showed the nature of loading on a typical steering knuckle, that includes

occasional overloads. The differences between the fatigue lives observed in bench tests and

predicted lives in the inelastic range were found to be factors of 3.9 and 1.4 of the fatigue life

with reliability of 50% and confidence level of 50% for fore/aft and lateral loading tests,

respectively.

Complexity of loading, plastic deformation due to overloads, and the influence of

manufacturing process parameters in fatigue assessment of the steering knuckle are

emphasized by Diboine (1996). It is indicated that the loading condition in service is non-

proportional multiaxial, variable amplitude and much more complex than for engine

components. In his work, the loading was simplified to two block loadings called normal and

accidental, whose levels were defined such that the same damage values from the track load

recordings with respect to the loading conditions were obtained. The accidental block was

found to represent only a few percent of the total number of cycles, but is important to
35

fatigue analysis as it is usually very close to plastic loading. The commercial software tools

used for this component gave less accurate predictions when the loading is a mixture of

elasto-plastic and fully elastic cycles. Modelling the residual stresses under variable amplitude

loading, the surface finish effects, the manufacturing process, and formulating fatigue criteria

for both intermediate and infinite lives with provision for overloads are named as a number

of problems not fully solved.

A practiced procedure in durability analysis is described in a review paper by Conle

and Chu (1997) and features using strain-life results from material tests, simulating three-

dimensional stress-strain models and multiaxial deformation paths to assess fatigue damage,

and searching for a critical plane for the most damaging direction. After the complex load

history is reduced to a uniaxial (elastic) stress history for each critical element, a Neuber

plasticity correction method is used to correct for plastic behavior.

Devlukia and Bargmann (1997) conducted fatigue assessment of a suspension arm

using deterministic and probabilistic approaches. The strength reduction effect due to

surface roughness was accounted for by representing the surface as a collection of notches

and making use of Neuber’s rule. The strength reduction effects due to the surface

roughness were assumed to be similar under constant and variable amplitude loading. It was

concluded that residual stress demonstrated a more pronounced effect under constant

amplitude loading as compared to variable amplitude loading. Cumulative damage under

variable amplitude loading sequence of long duration on simple specimens was non-

conservative by a factor of about two as compared to measured data. The prediction of

component lives based on specimen data and strain-life method was conservative by a factor

of about two.
36

Sonsino, Kaufmann, Foth and Jauch (1997) performed constant and variable

amplitude fatigue tests on induction-hardened automatic transmission shafts made from Ck

35 mod SAE 1038 steel, in order to evaluate safety reserves of the shafts and to apply several

selected methods to assess fatigue life, to compare these methods and to gain insights for

future component development. They concluded that the required data and criteria in order

to make a reliable prediction of fatigue life are:

S-N curve of the component on the basis of local stresses/strains including the

influences of geometry, material, surface condition and residual stresses on fatigue

behavior;

actual local material states (e.g. hardened or non) at the locations that are critical for

failure;

component and material mean stress modification model for the critical location;

cyclic stress-strain curves for the particular material states at the critical failure

locations; and

local (equivalent) strains or stresses in the critical area.

It was concluded that methods based on local stresses and strains (such as SWT parameter

and component-related Haigh diagram) predict fatigue life better than the methods based on

nominal loads.

Taylor (1997) proposed a technique for the prediction of fatigue failure in the

presence of stress concentrations. The technique, called “crack modeling”, predicts fatigue

failure in the presence of stress concentration through a modification of linear elastic

fracture mechanics (LEFM). FEA was used in conjunction with a modeling exercise in order

to extend the method to include bodies of arbitrary shape subjected to any set of loads. The

method was first tested using standard notch geometries (blunt and sharp notches in beams),
37

where accurate predictions of fatigue limit could be achieved. It was then applied to an

industrial problem, giving a reasonably close prediction of high-cycle fatigue behavior for an

automotive crankshaft. It was concluded that LEFM can be extended to predict the fatigue

behavior of bodies containing notches of standard geometry, instead of cracks. The method

is postulated to require only simple mechanical property data (the material fatigue limit and

stress-intensity threshold) and to use only linear-elastic FE modeling. Fracture mechanics

theory was used without the need to specifically model the presence of a crack and uses far-

field elastic stresses to infer behavior in the region of a stress concentration.

To verify this crack modeling method for torsional loading, Taylor, Zhou,

Ciepalowicz and Devlukia (1999) described the analysis of fatigue failure in an automotive

crankshaft as a result of loading under test conditions in bending and torsion. They

concluded that the crack-modeling method, previously used for the analysis of the

crankshaft under bending loads, could be extended to consider torsional loading. The model

was claimed to correctly predict the fatigue limit and also the crack initiation point and

growth direction. Also, the method was expected to be successful in analyzing any

combination of bending and torsion applied in-phase. Out-of-phase loading was not

considered. Moreover, predictions were not significantly affected by refinement of the FE

mesh, even though this had a strong effect on the value of the hot-spot stress.

Due to the significance of determining assembly stresses in aluminum knuckles,

Taylor et al. (1999) proposed analytical and numerical methods based on mechanics of

materials principles and FEA, respectively, for determining the stress distributions in steering

knuckle/tapered stud assemblies. The results showed that plane stress solutions for the

assembly radial and hoop stresses in a steering knuckle boss provide good correlation to the

FEA solutions for given draw distances. Furthermore, the effects of the coefficient of
38

friction do not significantly affect the hoop and radial stress distributions in the steering

knuckle.

Witter et al. (1999) converted a steering knuckle into a 6-DOF transducer to be able

to estimate the operating wheel translation force and moment inputs to a Mercury Sable

steering knuckle. To reach this goal, a 6-DOF load cell was used to provide an estimate of

the six forces and moments’ inputs at a point on the plate bolted to the steering knuckle.

They concluded that the calibration matrix did not vary significantly with suspension height,

but did vary significantly with large steering angles. Moreover, the strain gage responses were

sensitive to moments. It was suggested to place the vehicle on a 4-poster and apply forces

and moments through the 4-poster exciter to prevent the effect of un-suspended vehicle

array calibration problems. The results of their analysis was compared with two cornering

tests on the same design, showing an 11% error in forces and moments with respect to the

physical test results conducted on the prototypes.

Local strain approach was used to analyze fatigue behavior of a hardened and

tempered forged vehicle axle steering arm made of 41Cr4 low alloyed steel by Savaidis

(2001). The elastic-plastic strain-time path was evaluated using the strain-time sequence

measured at the failure critical location, taking into account the Masing and memory

behavior. The SWT damage parameter and a parameter developed by means of elastic-plastic

fracture mechanics (J-integral) based on a semi-elliptical surface micro-crack were used for

estimation of damage caused by the local stress-strain path. Introducing a factor to describe

the decrease of the endurance limit stress due to surface roughness, it was shown that the

experimentally determined fatigue lives agreed well with those calculated for the critical

location using the parameter based on J-integral.


39

Multiaxial fatigue is an important factor that should be considered in the fatigue

analysis of components. Kocabicak and Firat (2001) proposed a bi-axial load-notch strain

approximation for proportional loading to estimate the fatigue life of a passenger car wheel

during the cornering fatigue test under plane stress conditions. In their work the elasto-

plastic strain components were calculated analytically using the total deformation theory of

plasticity. In addition, the input for the load–notch strain analysis was the measured or

calculated plastic strain state at the notch together with the material stabilized cyclic stress-

strain curve evaluated from unnotched axial specimens. The damage accumulation was based

on the Palmgren-Miner rule.

One of the objectives of the present study was to conduct component testing on

steering knuckle in order to obtain component fatigue performance. The configuration of

the test machine, the control mode, fixturing of the steering knuckle, simulation of the loads

and moments applied to the component in service, and testing frequency are among the

considerations to conduct a well-simulated component test from which reliable data could

be obtained. To reach this goal, a number of studies, which include component testing on

steering knuckle and other similar automotive parts, were reviewed. These include studies by

Lee (1986), Osuzu et al. (1986), Gunnarson et al. (1987), Blarasin and Farsetti (1989),

Houshito et al. (1989) and Witter et al. (1999) on vehicle steering knuckle and the study by

Savaidis (2001) on steering arm. Table 2-4 shows a summary of the test configurations used

in these studies. In addition, brief descriptions of testing details are provided.


40

2.5 Optimization with Material, Manufacturing and


Cost Considerations
The results of literature review for the optimization of steering knuckle and similar

automotive components are discussed in this section. The focus of the review, as well as the

focus of the optimization task in this study has been on more of a global view of the

optimization problem including geometry optimization, alternative materials, and

manufacturing process modifications and costs. In this respect, apart from issues like

mathematical shape optimization, those studies were more emphasized that either include

manufacturing and alternative material considerations, or would lend to that purpose. In this

regard, the first set of the review is dedicated to shape optimization of automotive

components. The second set relates to manufacturing and cost considerations in

optimization.

Size and Shape Optimization


Botkin (1991) used a shape design modeling with three-dimensional mesh generation

to model and optimize vehicle suspension components, namely suspension arm and steering

knuckle. In the model the control arm geometry was assembled from two parts: boss and

arm, and similarly the steering knuckle consists of boss, slab and hub (see Figure 2-25). A

preliminary set of design primitives were developed which can be assembled into complete

solid models. The resulting models were associated with design variables. It was stated that

the mesh generation was capable of discretizing 3-D surfaces and solids into triangles and

tetrahedral, respectively. It was concluded that the most accurate FEA solution results were

required on the surface of a part and the curved faces, edges and vertices were necessary to

be precisely represented.
41

Lee et al. (1995) presented the design process for a proposed weight-reduced steering

knuckle. In their work, the non-damaging areas were identified by the life-contour plots of

the original steering knuckle and a new design for weight reduction was proposed. Then, the

original and proposed finite element steering knuckle models were loaded and constrained.

They found that the proposed steering knuckle, being lighter, should survive the entire

vehicle test field schedule by having fatigue damage values within acceptable limit. The

percent reduction in weight due to this optimization was not mentioned.

The application of shape optimization to a steering knuckle using thermal shape

vectors was presented by Krishna and Fetcho (1998). Since the steering knuckle was not

uniform in shape, it was cumbersome to generate shape vectors, which define the possible

variations in the shape of the component, using the traditional loads and displacement

methods. So thermal loading was used to generate shape vectors. For this purpose, the

baseline model was assumed to be at a temperature of zero degree (Kelvin) and a

temperature raise of 100 degrees was applied to twelve different regions of the steering

knuckle, independently. These were the regions where shape was allowed to be modified. A

shape optimization problem was defined; Minimize (objective): volume of the structure, Subject

to (Constraints): maximum von Mises stresses less than or equal to material ultimate tensile

strength. No safety factor was considered for the constraint. The redesigned steering knuckle

offers a weight reduction of 7.6 percent. A shortcoming of such analysis is the fact that the

real service loads applied to the component are not static, and considering ultimate tensile

strength as constraint may not be reasonable when cyclic loading exists. However, the

analysis method can be implemented by using cyclic rather than static stresses and fatigue

strength instead of ultimate tensile strength. This issue was not discussed in this study.
42

Krishna (2001) investigated the application of shape optimization techniques to

reduce the weight of a steering knuckle of a heavy truck suspension. A rigid body dynamic

analysis of the suspension axle was carried out and steering knuckle load cases were

obtained. Five different worst load cases were identified. A baseline FEA of the model

showed that some parts of the steering knuckle experienced low stresses during all the five

load cases. Shape vectors, which are the possible shape variations that the steering knuckle

could undergo to reduce weight, were generated by numerical interpolation method. To

generate the shape vectors, a group of elements were first identified and then a domain

element was created to enclose these elements. All the nodes of the domain element and the

other regular elements were grouped together as a node set. Perturbation vectors were

identified to specify how the deformation is to be affected. The shape optimization problem

was defined as follows –Minimize (Objective): Weight of the steering knuckle. Subject to

(Constraints): Maximum von Mises stresses less than or equal to limiting value for each of the

five load cases. The Sequential Linear Programming method as recommended by

MSC/NASTRAN was used to solve the optimization problem and give the directions for

changes. The shape optimization took 18 iterations for convergence. The objective (weight)

was brought down by 12.7%, which was considered to be an appreciable saving both in

material cost and final product cost. It was concluded that shape optimization techniques

can be used effectively to bring down the weight of a complicated component like a steering

knuckle, and multiple load cases can be handled by considering them separately. Figure 2-26

shows the superposed baseline and optimized steering knuckles of the study by Krishna

(2001).

Ferreira et al. (2003) investigated the basic concepts of structural optimization and

design for automotive durability. A trailing arm bracket seam welded to the reinforcement
43

bracket and bolted to vehicle body rail, was the case study. The optimization criteria were

defined in terms of size optimization: component thicknesses (sheet metal components)

were calculated in order to use a minimal amount of material and to provide maximum

durability. For this purpose, a standard structural optimization procedure using Altair

Optistruct (a general purpose finite element software and optimization package) was

conducted with volume of the bracket as objective function, maximum admissible von Mises

stress less than fatigue limit of the material (SAE 1010) as constraint, and thickness of

trailing arm bracket, reinforcement and welds as design variables. The optimization problem

converged after 7 iterations and 40 percent reduction in the volume was achieved. Then the

optimized components were submitted to fatigue calculation in a full vehicle finite element

model using FDynam (in house Ford software for durability calculation). The durability

results showed that the bracket reinforcement was overloaded in the full vehicle model with

complete load history. The solution was to increase the thickness of the components using

high strength steel (SAEJ1392050) for the bracket reinforcement. This revised thickness and

material configuration achieved the durability requirement for the components.

Lee and Lee (2003) presented optimization design methodologies in the design

stages of a cast aluminum control arm for a suspension. Using topology optimization, the

optimal layout and the reinforcement structure were obtained, and then the detail designs

were carried out using shape optimization for structural rigidity and strength. The baseline of

structural safety was considered to be yield strength and ultimate strength. In comparison

with a stamped steel control arm, the mass reduction was 50 percent and the structural

rigidity and static strength were improved up to 40 percent. Even though a control arm

undergoes cyclic load history, fatigue strength was not mentioned in this study.
44

Material and Manufacturing Considerations


Application of forged parts made of microalloyed steels to automobile parts is

becoming increasingly common due to their superior properties and cost reductions

compared to conventional Q&T carbon steels. Microalloyed steels do not require heat

treatment and, therefore, no additional machining for correcting distortions due to heat

treatment after forging is necessary. The fatigue properties and toughness of microalloyed

steel forgings have been demonstrated to fit for purpose. However, compared with heat

treated low alloy steels their fracture toughness may be somewhat lower, even though still

significantly superior to castings.

Farsetti and Blarasin (1988) investigated the possibility of replacing forged Q&T

steels by forged microalloyed steels of appropriate composition and microstructure. They

concluded that replacing Q&T steels with microalloyed steels is possible with the following

considerations; for mechanical components that can be made from 800 MPa class steels,

microalloyed steels could be used. With low-C high-Mn steels, satisfactory strength values

can be attained, keeping good toughness properties. The higher-strength class microalloyed

steel lends itself to an increase in strength and toughness by optimizing the micro structural

parameters. Small ferrite grain and presence of bainitic phase will increase strength of

microalloyed steels significantly.

Following the same goal, Kuratomi et al. (1990) developed lightweight connecting

rods based on fatigue resistance analysis of microalloyed steels. Rotating bending fatigue

tests on smooth and notched specimens as well as component buckling and load-controlled

fatigue tests were conducted. Figure 2-27 shows the fatigue test results obtained with actual

connecting rods made of SV40CL1 microalloyed steel (0.4% carbon steel) and S40C Q&T

(equivalent to SAE 1040) steel. It was concluded that connecting rods made of the forged
45

microalloyed steel exhibit 25% higher fatigue limit than the similar forged Q& T steel and

are 10% lighter in weight. In addition, in selecting high fatigue strength materials for

connecting rods, they emphasized considering notch sensitivity because actual parts have

small notches on their surface. They added that when the grain size became larger with

increasing temperature, notch sensitivity of fatigue properties was not observed. They

attributed this to the relation between the effective grain size and notch depth in a way that

when the effective grain size was the same as or larger than one-half the notch depth, notch

sensitivity decreased.

Material type and manufacturing process considerations are two key factors in

reducing component cost. To reduce component cost, Repgen (1998) modified a connecting

rod with respect to material and manufacturing process. For the material modification, a

direct comparison of steel-forged versus hot formed powder metal connecting rods with

similar cross section showed a 21 percent higher fatigue strength level for the forged part.

To reach a considerable cost reduction for connecting rods it was found necessary to analyze

the total component costs consisting of rough part and machining. It was noted that steel

forged rough parts have the advantage of low material and production costs. With the

development of fracture splitting the connecting rods a total cost reduction up to 25 % was

achieved compared to conventionally designed connecting rods. Further steps were taken by

improving the forging technology to produce an optimized rough part with regards to

tolerances that reduce the machining costs.

Nägele et al. (2000) reported the optimization of the process flow between bulk

forming of parts and the subsequent machining processes of the forgings. Four case studies

were presented in which the design of the part was modified in order to optimize the whole

manufacturing process. The production process of a drive shaft flange (Figure 2-28a) that
46

was originally designed to be machining was switched to cold forming (forging). By doing

this, cost was reduced by replacing expensive machining processing methods such as spline

slotting. In addition, mechanical properties in the root of the spline teeth were enhanced by

as much as 25-30% due to grain flow orientation, and weight and scrap reduction were

achieved by optimum material utilization and process considerations.

For the case of an output shaft of the study by Nägele et al. (2000), Figure 2-28b, the

original design required a hot forging with a cold calibration process where the excess

material is removed at room temperature. To optimize the process, the cold calibration

process was eliminated and replaced by a subsequent machining process in-house. This

machining process incorporated not only the standard turning operation, but also a deep-

hole drilling, induction hardening, cold rolling for a spline to standard and several threads.

The greatest potential for cost savings was mainly found in technically advanced processing

and forming operations. It was demonstrated that the emphasis needs to be on the

manufacturing process as a whole and not just sectors of the manufacturing requirement.

This example proved to be approximately 20% less expensive than the original intention of

manufacturing for this component. The cost savings were attributed to elimination of the

cold calibration requirements, same source forming and machining that allows a greater

material utilization, which in turn improves die life, scrap and rework requirements, and

delivering the component assemble-ready to the customer with no further processing

required.

Another case of the same study by Nägele et al. (2000) involved a flanged cylinder.

Figure 2-28c illustrates a warm-formed workpiece and the part after machining. The

production process included warm forging and turning. An improved manufacturing process

was developed where the forging process was modified and elimination of machining the
47

head was implemented. This was achieved by an additional cold calibration operation of the

head in the area of the pierced elongated holes. A chamfer was added to the part which

eliminated the formation of fins, reducing the machining time considerably, and eliminated a

deburring operation. In addition to this, the shaft diameter was reduced which also helped to

minimize the machining time. These changes did increase the cost of the forming operation,

but the total part cost was reduced by 9% compared to the original design practices.

The final case of the study by Nägele et al. (2000) was a link shaft as shown in Figure

2-28d, in which a machined component (left) was replaced by a formed part (right). The

original design started with a hot forged preform that required several subsequent machining

operations. An optimal design was proposed for this component, in which the preform was

warm formed in a three stage process with an additional cold calibration operation for the

outside diameters. A final cold forming of the external spline to print specification produced

the spline as assemble-ready. Finally the part was machined in a dual spindle lathe with two

separate setups. The material usage for this component was 24% less and the weight

reduction was 10.3% less material. The cost saving to produce this part with the new design

amounted to 51.2%. In addition to these reductions, the drive shaft was also able to be re-

designed due to the size reduction of this link shaft which was an overall reduction in weight

of the transmission in whole.

The key aspects of the study by Nägele et al. (2000) were to demonstrate the

significance of combining the forming and machining operations to be optimized by one

supplier, and how this can be more cost effective than the alternative of simply delivering a

formed part which then has to be subsequently machined. Also, it was emphasized that an

intelligent combination of individual manufacturing processes may lead to the reduction of


48

individual processes and this leads to an increased cost effectiveness and competitiveness of

the component.

Flashless forging is introduced as an efficient method to reduce manufacturing cost.

In conventional hot forging of connecting rods, the material wasted to the flash accounts for

approximately 20-40% of the original workpiece (Vazquez and Altan, 2000). In order to

reduce the cost of forged products, the forging must be performed in a closed cavity to

obtain near-net or net shape parts. In flashless forging, the volume distribution of the

preform 1 must be accurately controlled to avoid overloading the dies and to fill the cavity.

Additionally, the preform must be simple enough to be mass-produced. A study by Vazquez

and Altan (2000) deals with the preform design for flashless forging of a connecting rod and

introduces a new tooling concept for forging of complex parts with a controlled amount of

flash. A hot forging tooling was developed that would allow the forging of a connecting rod

with a controlled amount of flash. It was established that 5% material waste for connecting

rod may be reasonable under the present production conditions. The tooling was also

targeted to be simple enough to be used in mass production.

To improve fatigue performance of forged components, reducing surface defects

and/or their effects play an important role. One way to achieve this goal is to induce surface

compressive residual stresses. Applying localized inelastic deformation through processes

such as shot peening or surface rolling, are the common available methods of inducing

residual stresses. Fifty percent greater fatigue strength has been reported in rolled threads,

compared with cut or ground threads made of high strength steel (Stephens et al., 2000).

Figure 2-29 illustrates the beneficial effect of shot peening on fatigue resistance of gears.

1 Preform is the forging operation in which stock is preformed or shaped to a predetermined size and contour
prior to subsequent die forging operations; the operation may involve drawing, bending, flattening, edging,
fullering, rolling, or upsetting.
49

Thermal processes can also induce residual stresses. For example, in surface hardening of

steel by processes such as induction hardening, carburizing, or nitriding, in addition to the

hard surface produced, a beneficial compressive residual stress is also created on the surface.

This compressive residual stress can very effectively prevent the formation of cracks.

To estimate machining process quality and cost, Nicolaou et al. (2002) presented a

method for formulating a model for first estimating quality, cost and cutting fluid wastewater

treatment impacts of two machining operations (end milling and drilling), and then for

tradeoff decision making. A case study of an automotive steering knuckle was presented,

where decision variables include material choice (cast iron versus aluminum), feed rate,

cutting speed and wet versus dry machining. The results of Nicolaou et al. study was used to

specify the cost attributes of the steering knuckle of the present study and is discussed in

more detail in Section 7.1.

Manufacturing considerations in structural optimization is addressed in a study by

Schramm et al. (2002). One of the techniques described is a manufacturing constraint for

topology optimization. In this technique a draw direction is defined to open up the design

into one direction only. This would allow for better casting, forging, and machining

manufacturability. As an example, an engine mount bracket was optimized. A new layout

with a reduced mass for a given design was sought. The new design should have had equal

or better strength characteristics than the existing design. The mass of the initial design,

which was manufactured from cast aluminum, was 950 g. Figure 2-30 shows the original

part. The design space was defined by the space of the existing design. The following six

load cases were of interest: Drive off forward; drive off backwards; driving into a pot hole;

driving out of a pot hole; loads from an attached belt transmission; transportation of the

engine block for assembly. The result of the topology optimization (Figure 2-30) was
50

transformed into a new design that was further improved using shape optimization. The

final design (Figure 2-30) has a structural mass of 730 g, with deformations equal to or less

than the initial design and stresses less than the allowable stress.

Shenoy (2004) and Shenoy and Fatemi (2005) investigated weight and cost reduction

opportunities of a mid-size sedan forged steel connecting rod. Cost was reduced by replacing

the conventional forged steel material with crackable C-70 forged steel. They concluded that

fatigue strength was the most significant design driving factor in optimization of the

component. Geometry optimization was performed under maximum engine operating speed

and maximum gas load (service worst case conditions) and the optimized connecting rod

satisfied all the constraints defined. The optimized geometry, while feasible to be

manufactured, was 10% lighter and 25% less expensive than the conventional forged steel

connecting rod, despite the lower strength of C-70 steel compared to the conventional

forged steel.
51

Table 2-1 Classification of metals in decreasing order of forgeability (Kalpakjian and


Schmid, 2000).

Metal or Alloy Approximate Range of


Hot Forging Temperature (°C)

Aluminum alloys 400–550


Magnesium alloys 250–350
Copper alloys 600–900
Carbon and low–alloy steels 850–1150
Martensitic stainless steels 1100–1250
Austenitic stainless steels 1100–1250
Titanium alloys 700–950
Iron-base superalloys 1050–1180
Cobalt-base superalloys 1180–1250
Tantalum alloys 1050–1350
Molybdenum alloys 1150–1350
Nickel-base superalloys 1050–1200
Tungsten alloys 1200–1300

Table 2-2 Properties of cast and forged steels swivel hub (Cristinacce et al., 1998).

0.2% PS
Lower Su EI R/A 3mm
C Si Mn S V Hv
Sy (N/mm2) (%) (%) U (J)
(N/mm2)
Casting 0.31 0.45 1.34 0.014 - 440 655 15- 63 190-
(wide 25 205
arm)
Forging 0.39 0.26 1.28 0.075 0.099 669 969 17 46.6 9 290-
(wide 300
arm)
52

Table 2-3 Summary of characteristic comparison between manufacturing processes.

Process
Forging Casting Powder Metallurgy
Property

Strength high medium medium

Ductility high low low

Toughness high medium medium

Fatigue crack growth resistance good poor fair

Directional strength capability yes none none

Heat treatment response good requires close control good

Internal defects possible many possible

Production volume high high moderate to high

low (sand casting) to


Production rate high high
high (die casting)
low (sand casting) to
Initial tooling cost medium high
medium (die casting)

Production cost low low low

Material waste yes yes limited

Shape complexity limited limited limited

Dimensional versatility high limited limited

Dimensional accuracy medium medium high

Surface finish good to poor poor fair

Material versatility high limited limited


53

Table 2-4 Test setup for steering knuckle and other relevant automotive components.

Author(s)/ Component(s) Test Type(s) Component Test Configuration


Year
Lee/ 1986 Steering Load-controlled A steering knuckle consisted of wheel
knuckle/other component fatigue stud, carrier tube, and steering arm. The
components tests, characteristic load amplitudes were 4.0 kN
Vehicle field tests for wheel stud and carrier tube (test A) and
4.5 kN for steering arm (test B).

Acceptable criteria were no crack initiation


and no permanent deformation during the
5
test until 2×10 cycles. Different grips were
developed for tests A and B.
54

Author(s)/ Component(s) Test Type(s) Component Test Configuration


Year
Osuzu, et Axle spindle, Specimen tensile Impact tests were conducted on complete
Knuckle arm and impact tests, axle spindle forgings using a drop weight
al. / 1986
Load-controlled test procedure.
component fatigue Fatigue tests were conducted on forged
tests axle spindles.

Axle spindle

Drop weight impact test of axle spindle

Fatigue test of axle spindle


Gunnarson Steering Load-control Schematic of fatigue testing is shown
knuckle/other component fatigue below:
et al. /1987
components test

Fatigue test setup of knuckle.


55

Author(s)/ Component(s) Test Type(s) Component Test Configuration


Year
Blarasin Steering Load-control Fatigue bench tests were carried out on
knuckle component fatigue components mounted as they would be on
and
tests, a vehicle, the articulated joints being
Farsetti/ Specimen strain- substituted by appropriate flexible strip
controlled fatigue joints.
1989
tests,
Component road
tests

Locations of strain gages

mm
WHEEL

LOAD

Layout of the steering knuckle test bench.


56

Author(s)/ Component(s) Test Type(s) Component Test Configuration


Year
Houshito Steering Tensile tests, Vertical, lateral and longitudinal load tests
knuckle Load-control were conducted as shown below:
et al. /1989
component fatigue
tests,
Impact tests
57

Author(s)/ Component(s) Test Type(s) Component Test Configuration


Year
Witter et Steering Force/moment A small triangular plate was constructed
knuckle simulation on a and bolted to the steering knuckle at the
al. /1999
vehicle locations shown. The strain-gaged steering
knuckle and plate were reattached to the
vehicle.

Steering knuckle instrumented with 18 ICP


strain gages.

Savaidis/ Steering arm Specimen strain- The load was induced in one side of the
controlled fatigue component via hydraulic actuator while its
2001
tests, other side was born by a block gear shaft
Load control positioned in the lug of steering arm.
component fatigue
bench tests

Steering arm and load configuration.


58

Figure 2-1 Flow chart of the forging process (Kalpakjian and Schmid, 2000).

Figure 2-2 (a) Stages in impression-die forging of a connecting rod for an internal
combustion engine. (b) Fullering and (c) edging operations to distribute the
material when pre-shaping the blank for forging (Kalpakjian and Schmid,
2000).
59

Figure 2-3 Typical vehicle components that are subject to competition among
manufacturing processes (Cristinacce et al., 1998).

Figure 2-4 Components of the engine block.


60

Figure 2-5 Front-wheel-drive car with MacPherson-strut front suspension and strut-type
independent rear suspension (Saturn Corp.).

Figure 2-6 Arrangement for SLA suspension. The steering knuckle/spindle assembly in
SLA suspension supports the wheels and attaches to the control arm with
ball and socket joints. The control arm attaches to the frame of the vehicle
through rubber bushings to help isolate noise and vibration between the road
and the body (Halderman and Mitchel, 2000).
61

Figure 2-7 Multilink double-wishbone suspension uses a spring and damper unit. The
long curved steering knuckle and very angled upper control arm allow its use
in areas of limited size, Courtesy of American Honda Motor Co., Inc. (Birch,
2000).

Figure 2-8 Steering knuckle in strut suspension. In this type of suspension, the strut rod
is the longitudinal support to prevent front-to-back wheel movement,
Courtesy of Ford Motor Company (Birch, 2000).
62

Figure 2-9 Spindle assembly in a solid axle suspension, Courtesy of Ford Motor
Company (Birch, 2000).

Figure 2-10 A twin I-beam front suspension. This particular swing axle design uses two
ball joints to connect the steering knuckle to the axle, Courtesy of Ford
Motor Company (Birch, 2000).
63

Figure 2-11 Static lateral input (left) and static longitudinal input (Haeg, 1997).

Figure 2-12 Steering knuckle assembly force reactions with longitudinal input isometric
view (left) and plan view (Haeg, 1997).
64

Figure 2-13 Forging process parameters, the manufacturing influenced parameters and
the effect on mechanical properties. Each parameter and its connector lines
are coded with the same color.
65

Figure 2-14 Constant amplitude fatigue crack growth behavior of Q&T vs. cast steels
(Rice, 1997).

Figure 2-15 Superimposed plots of true stress amplitude versus reversals to failure for
forged steel and powder metal materials used in connecting rods (Afzal and
Fatemi, 2004).
66

Figure 2-16 Experimental stress amplitude vs. cycles to failure for forged steel and
powder metal connecting rods (Afzal and Fatemi, 2004).

Figure 2-17 Changes in fatigue performance of vehicle front suspension arm due to
surface defects from forged and cast manufacturing processes (Blarasin and
Giunti, 1997).
67

Figure 2-18 Capability of each manufacturing process of producing parts to some


characteristic tolerance and surface finish under typical conditions (Schey,
2000).

Figure 2-19 Influence of forging reduction on anisotropy for a 0.35% carbon wrought
steel. Properties for a 0.35% carbon cast steel are shown in the graph by a
star (*) for purposes of comparison (Blair and Monroe, 2002).
68

Figure 2-20 Relative unit costs of a small connecting rod made by various forging and
casting processes (Kalpakjian and Schmid, 2000).

Figure 2-21 Fatigue design flow chart (Stephens et al., 2000).


69

Figure 2-22 Methods of structural durability assessment (Berger et al., 2002).

Figure 2-23 Timing of past (top) and current product development periods (Berger et al.,
2002).
70

Fx

Fy

Fz

Mx

Mz

Figure 2-24 Wheel loads and moments on the test road indicating the existence of
overloads are part of the service history (Lee et al., 1995).
71

Figure 2-25 Design model (left) and meshes of the steering knuckle (Botkin, 1991).

Figure 2-26 Baseline superposed over optimized steering knuckle (Krishna, 2001).
72

Figure 2-27 Fatigue strength of microalloyed (SV40CL1) and Q&T (S40C) forged
connecting rods (Kuratomi et al., 1990).

(a) (b)

(c) (d)

Figure 2-28 Case studies to optimize manufacturing processes of automotive


components: (a) drive shaft flange, (b) output shaft, (c) flanged cylinder, and
(d) link shaft (Nägele et al., 2000).
73

Figure 2-29 Effect of shot-peening on fatigue behavior of carburized gears (Stephens et


al., 2000).

Figure 2-30 Original design (left), topology optimization results (middle) and final design
of an engine mount bracket (Schramm et al., 2002).
Chapter Three

3 Material Fatigue Behavior and Comparisons


Material mechanical properties can be used to compare material and/or processes.

Such properties are also often necessary as inputs for different analyses like life prediction

and optimization. Modulus of elasticity, yield and tensile strengths, and ductility and strain

hardening properties are obtained from tensile test. These properties are used to evaluate

material’s tensile behavior under static or quasi-static loading conditions during elastic and

plastic deformations.

In design situations where cyclic loading is involved, considering monotonic strength

and ductility as the design variables is not sufficient to obtain a safe and reliable design, due

to introduction of cyclic softening or hardening. In such cases, to which many automotive

components are subject to them, considering cyclic deformation of the material is also

essential.

From strain-controlled fatigue tests, cyclic deformation and fatigue properties are

extracted. Cyclic modulus of elasticity represents cyclic stiffness prior to cyclic yielding.

Cyclic yield strength shows the yield point of the material in cyclic loading. Cyclic strength

coefficient and cyclic strain hardening exponent are the properties that help the designer to

investigate cyclic deformation after yielding, as well as to compare the cyclic and monotonic

deformation characteristics while the material plastically deforms.

74
75

Fatigue strength coefficient and exponent represent S-N behavior of the material

often used to evaluate long-life fatigue performance, while fatigue ductility coefficient and

exponent are properties used to describe ε-N behavior to evaluate low-cycle fatigue

performance.

In the specimen testing program of this study as discussed in this chapter, it was

intended to obtain monotonic, cyclic and fatigue properties of forged steel 11V37, cast

aluminum A356-T6, and cast iron 65-45-12, and consequently compare them with respect to

those properties. The monotonic properties included are stiffness, yield and ultimate

strengths, percent elongation, percent reduction in area, strength coefficient, strain

hardening exponent, and true fracture strength and ductility. The cyclic deformation

parameters consist of cyclic stiffness, cyclic yield strength, cyclic strength coefficient, and

cyclic strain hardening exponent. In addition, cyclic deformation characteristics such as

hardening and softening were also investigated. The fatigue properties include fatigue limit,

fatigue strength coefficient, fatigue strength exponent, fatigue ductility coefficient and fatigue

ductility exponent. These properties were used to characterize and compare stress-life and

strain-life material behaviors.

3.1 Experimental Program

3.1.1 Material and Specimen Fabrication


The materials were in the form of vehicle steering knuckles made of forged steel

11V37, cast aluminum A356-T6, and cast iron 65-45-12. Although utilized in three different

vehicles, the three steering knuckles were selected from relatively similar-engine-size vehicles

to make the comparisons.


76

Identical flat plate specimens with rectangular cross section were prepared out of the

received steering knuckles for the monotonic and fatigue tests. The specimen configuration

and dimensions are shown in Figure 3-1. This configuration was chosen such that the gage

section length could be minimized to prevent buckling. A detailed finite element analysis

(FEA) was performed to evaluate stress concentration at the radius. The FEA results

indicate the stress concentration factor of about 1.05 for the specimen.

All specimens were machined in the Mechanical, Industrial, and Manufacturing

Engineering Machine Shop at the University of Toledo. Figure 3-2 shows the steering

knuckles of the three investigated materials, from which the specimens were machined.

Specimens in three geometrical orientations were made from the forged steel steering

knuckle to investigate the effect of directionality (see Figure 3-2). For cast aluminum and

cast iron steering knuckles, since the material properties are independent of geometrical

orientation, the specimens were machined from the hub and one of the arms, respectively.

The specimens were initially rough cut to a rectangular strip out of the vehicle

steering knuckles using a milling machine and then inserted into a fixture for cutting the

required geometry in another CNC milling machine. Using the CNC milling machine, final

machining was performed to achieve the tolerable dimensions specified on the specimen

drawings.

After machining, specimens were checked carefully for flatness on a machinist’s

stone-flat surface. Any specimen with camber exceeding 0.1 mm (0.004 in) from end to end

was rejected. The specimen gage section edges in the thickness and width directions were

polished, the polishing marks coinciding with the specimen’s longitudinal direction. The

polished surfaces were carefully examined to ensure complete removal of machining marks

in the test section.


77

3.1.2 Testing Equipment


An INSTRON 8801 closed-loop servo-hydraulic axial load frame in conjunction

with a Fast-Track 8800 digital servo-controller was used to conduct the tests. The calibration

of this system was verified prior to beginning the test program. The load cell used had a

capacity of 50 kN (11 klb). In order to achieve the best alignment of the specimens, two

stops were designed and mounted on the hydraulically operated universal wedge grips with

flat faces, during all tests. These stops helped to align the specimen’s ends in series with the

load train. Any twisting of the specimens was avoided by using precisely machined blocks to

prevent any relative rotation of the grips.

Total strain was controlled for all tests using an extensometer rated as ASTM class

B1 (ASTM E83-02, 2002). The calibration of the extensometer was verified using

displacement apparatus containing a micrometer barrel in divisions of 0.0001 in. The

extensometer had a gage length of 0.2362 in (6 mm) and was capable of measuring strains up

to 10%. In order to protect the specimen’s surface from the knife-edges of the

extensometer, M-coat D mixture was used to “cushion” the attachment. The extensometer

was carefully positioned at the center section of the specimen uniform gage section.

All tests were conducted at room temperature and were monitored using a digital

thermometer. In order to minimize temperature effects upon the extensometer and load cell

calibrations, fluctuations were maintained within ± 2 oC (± 3.6 oF) as required by ASTM

Standard E606 (1998). Also, the relative humidity of the air was monitored using a precision

hydrometer.

Significant effort was put forth to align the load train (load cell, grips, specimen, and

actuator). Misalignment can result from both tilt and offset between the central lines of the

load train components. According to ASTM Standard E606 (1998), the maximum bending
78

strains should not exceed 5% of the minimum axial strain range imposed during any test

program. For this study, the minimum axial strain range was 0.0025 in/in. Therefore, the

maximum allowable bending strain was 125 microstrains. ASTM Standard E1012 (1999),

Type A, Method 1 was followed to verify specimen alignment. A 0.25 in×0.25 in square

cross-section bar with eight strain gages was used and the maximum bending strain was

much smaller than that allowed by the ASTM standard.

3.2 Test Methods and Procedures

3.2.1 Monotonic Tension Tests


All monotonic tests in this study were performed using test methods specified by

ASTM Standard E8-02 (2002). Two specimens for forged steel and one for each of cast iron

and cast aluminum were used to obtain the monotonic properties. Due to the limitations of

the extensometer, strain-control was used only up to 10% strain. Unless the specimen failed

prior to this level, displacement-control was used after this point until fracture. A stress

versus strain plot was obtained automatically for each test.

For the elastic and initial yield region (0% to 0.5% strain), a strain rate of 0.003125

mm/mm/min was chosen. This strain rate was three-quarters of the maximum allowable

rate specified by ASTM Standard E8-02 (2002) for the initial yield region. After yielding

(0.5% to 10% strain), the strain rate was increased by a factor of three (i.e., 0.0094

mm/mm/min). After the extensometer was removed, a displacement rate of 0.215 mm/min

was used. This displacement rate provided approximately the same strain rate as that used

prior to switching control modes.

After the tension tests were concluded, the broken specimens were carefully

reassembled. The final gage lengths of the fractured specimens were measured with a
79

Vernier caliper having divisions of 0.001 in. Using an optical comparator with 10x

magnification and divisions of 0.001 in, the final cross section dimensions were measured. It

should be noted that prior to the test, the initial cross section was measured with this same

instrument.

3.2.2 Constant Amplitude Fatigue Tests


All constant amplitude fatigue tests in this study were performed according to ASTM

Standard E606 (1998). It is recommended by this standard that at least 10 specimens be used

to generate the fatigue properties. For this study, three sets of tests for forged steel (in three

geometrical orientations) and one set for each of cast aluminum and cast iron were

conducted.

For forged steel, a total of 29 tests were conducted, in order to obtain the fatigue

data as well as to investigate directionality effects. Specimens of three geometrical directions

were tested at 6 different strain amplitudes ranging from 0.15% to 0.7%, with a total of 11, 8

and 10 specimens in directions A, B and C respectively. Figure 3-2 shows a schematic of the

forged steering knuckle and the above-mentioned directions. For cast aluminum, a total of

14 fatigue tests at 8 strain amplitude levels ranging from 0.125% to 0.7% and for cast iron, a

total of 10 fatigue tests at 5 strain levels ranging from 0.13% to 0.5% were conducted.

There were two control modes used for these tests; strain-control and load-control.

Strain-control was used in all tests, except for: 1) a number of tests that were switched to

load-control due to stabilized cyclic behavior during strain-control mode. For these tests,

strain-control was used initially to determine the stabilized load. Then load-control was used

for the remainder of the test. The reason for the change in control mode was due to

frequency limitations of the extensometer. 2) A number of tests conducted in load-control


80

mode due to negligible predicted amount of plastic deformation. And 3) run-out tests that

were conducted in load-control mode. For the strain-control tests, the applied frequencies

ranged from 0.1 Hz to 2 Hz. For the load-control tests including run-out tests, the frequency

was increased to up to 30 Hz in order to shorten the overall test duration. All tests were

conducted using a triangular waveform.

Strain amplitudes larger than 0.7% were not possible due to specimen buckling

limitation. Instron LCF and SAX software tools were used for strain-control and load-

control tests, respectively. During each strain-control test, the total strain was recorded using

the extensometer output. Test data were automatically recorded, periodically, throughout

each test.

3.3 Experimental Results, Observations, and Analysis

3.3.1 Monotonic Deformation Behavior and Comparisons


The properties determined from monotonic tests were the following: modulus of

elasticity (E), yield strength (Sy), ultimate tensile strength (Su), percent elongation (%EL),

percent reduction in area (%RA), true fracture strength (σf), true fracture ductility (εf),

strength coefficient (K), and strain hardening exponent (n).

True stress (σ), true strain (ε), and true plastic strain (εp) were calculated from

engineering stress (S) and engineering strain (e), according to the following relationships,

which are based on constant volume assumption:

σ = S (1 + e ) (3-1a)

ε = ln (1 + e ) (3-1b)

σ
ε p = ε − εe = ε − (3-1c)
E
81

The true stress-true strain (σ−ε) plot is often represented by the Ramberg-Osgood equation:

1
σ
⎛σ ⎞ n
ε = εe + ε p = + ⎜ ⎟ (3-2)
E ⎝K⎠

The strength coefficient, K, and strain hardening exponent, n, are the intercept and slope of

the best line fit to true stress (σ) versus true plastic strain (εp) data in log-log scale:

σ = K (ε p ) n (3-3)

In accordance with ASTM Standard E739 (1998), when performing the least squares

fit, the true plastic strain (εp) is the independent variable and the stress (σ) is the dependent

variable. These plots for the two tests of forged steel and one test for each of cast aluminum

and cast iron are shown in Figure 3-3(a), (b) and (c), respectively. To generate the K and n

values, the range of data used in this figure was chosen according to the definition of

discontinuous yielding specified in ASTM Standard E646 (2000). The data range used was

between 1% and 9% total strain.

The true fracture strength, σf , was calculated from:

Pf Pf
σf = = (3-4)
Af wf t f

where Pf is the load at fracture, and wf and tf are the width and thickness at fracture,

respectively. The true fracture ductility, εf , was calculated from the relationship based on

constant volume:

⎛ A ⎞ ⎛ ⎞ 1
ε f = ln ⎜⎜ o ⎟⎟ = ln ⎜ ⎟ (3-5)
⎝ Af ⎠ ⎝ 1 − RA ⎠

where Af is the cross-sectional area at fracture, Ao is the original cross-sectional area, and RA

is the reduction in area.


82

A summary of the monotonic properties for the three materials is provided in Table

3-1, including the ratios of each property with respect to that of forged steel. As discussed

later in this report, since direction A is the primary loading direction of the component, it

was selected to obtain monotonic data of forged steel. Table 3-2 summarizes the monotonic

test results. Separate monotonic stress-strain curves are shown in Figure 3-4 for each

material, and Figure 3-5 presents a direct comparison of the three materials’ monotonic

stress-strain behavior.

Comparing E values from the monotonic tests, the stiffness of cast aluminum and

cast iron are 39% and 96% of forged steel’s, respectively. The E value of cast iron in the

literature was found to be about 170 MPa, which is somewhat less than what was found in

this study. This could be due to the characteristics of cast products, and parameters such as

porosity and inclusions. To investigate the difference, further metallurgical and

microstructural analyses are required that are beyond the scope of this investigation.

Ultimate tensile strength of forged steel is far above the other two materials, cast

aluminum and cast iron being 37% and 57% of forged steel, respectively. The yield strength

of cast aluminum and cast iron is also lower, 42% and 54% of forged steel, respectively. As a

criterion to measure ductility, the percent elongation of cast aluminum and cast iron were

found to be 24% and 48% of forged steel, respectively. This shows that cast aluminum and

cast iron behave less ductile compared to the behavior of forged steel. This could also be

observed from the fracture ductility results where true fracture ductility of cast aluminum

and cast iron is about 23% and 59% of forged steel, respectively. Thus, with respect to

strength and ductility, which are two major mechanical properties measured from monotonic

loading, forged steel shows superiority to the two cast materials. As can be seen in Figure
83

3-5, strain hardening is more prominent for forged steel and cast iron, while cast aluminum

behaves toward elastic-perfectly plastic behavior.

Although comparison of monotonic strength and ductility data as two major design

parameters show superiority of forged steel, it should be noted that considering only

monotonic properties for design purposes involving cyclic loading may lead to false

conclusions due to introduction of cyclic softening and/or hardening.

3.3.2 Cyclic Deformation Behavior and Comparisons

3.3.2.1 Transient Cyclic Response


Transient cyclic response describes the process of cyclic-induced change in

deformation resistance of a material. Data obtained from constant amplitude strain-

controlled fatigue tests were used to determine this response. Plots of stress amplitude

variation versus applied number of cycles can indicate the degree of transient cyclic

softening/hardening. Also, these plots show when cyclic stabilization occurs.

A composite plot of the transient cyclic response for the three materials studied is

shown in Figure 3-6 and Figure 3-7. The transient response is normalized on the rectangular

plot in Figure 3-6, and a semi-log plot is shown in Figure 3-7. While the first set of

normalized plots present the response over the entire life, the second set magnifies the stress

response over the early cycles, in order to facilitate observing the variation in stress. Even

though multiple tests were conducted at each one of the strain amplitudes in most cases,

data from one test at each one of the strain amplitudes tested are shown in these plots. For

forged steel, the data plotted are for direction A only. These figures show some transient

softening for forged steel, while cast iron and cast aluminum behave in a cyclically hardening

manner.
84

3.3.2.2 Steady-State Cyclic Deformation


Another cyclic behavior of interest is the steady state or stable response. Data

obtained from constant amplitude strain-controlled fatigue tests were also used to determine

this response. The properties determined from the steady-state hysteresis loops are the

following: cyclic strength coefficient (K'), cyclic strain hardening exponent (n'), and cyclic

yield strength (Sy'). Half-life (midlife) hysteresis loops and data were used to obtain the stable

cyclic properties.

Similar to monotonic behavior, the cyclic true stress-strain behavior can be

characterized by the Ramberg-Osgood type equation:

Δε Δε e Δε p Δσ ⎛ Δσ ⎞
1
n′
= + = +⎜ ⎟ (3-6)
2 2 2 2E ⎝ 2K ′ ⎠

It should be noted that in Equation (3-6) and the other equations that follow, E is

the monotonic modulus of elasticity, obtained from monotonic tests. However, since the

value of E obtained from monotonic test of ductile cast iron was different from that

published in the literature of ductile cast iron as discussed previously, the average E value

from the first cycle of strain-control fatigue tests was used to calculate plastic strains for this

material.

The cyclic strength coefficient, K', and cyclic strain hardening exponent, n', are the

intercept and slope of the best line fit to true stress amplitude (Δσ/2) versus true plastic

strain amplitude (Δεp/2) data in log-log scale:

n′
Δσ ⎛ Δε p ⎞
= K ′⎜⎜ ⎟⎟ (3-7)
2 ⎝ 2 ⎠

In accordance with ASTM Standard E739 (1998), when performing the least squares

fit, the true plastic strain amplitude (Δεp/2) is the independent variable and the stress
85

amplitude (Δσ/2) is the dependent variable. The true plastic strain amplitude was calculated

by the following equation:

Δε Δε Δσ
p
= − (3-8)
2 2 2E

These plots are shown in Figure 3-8. To generate the K' and n' values, the range of

date used in the figure was chosen for 0.003 ≤ εa ≤ 0.007 for forged steel, 0.00375 ≤ εa ≤

0.007 for cast aluminum, and 0.002 ≤ εa ≤ 0.005 for cast iron, based on the plastic strain

range being considerable at these strain range domains.

The cyclic stress-strain curve reflects the resistance of a material to cyclic

deformation and can be vastly different from the monotonic stress-strain curve. The cyclic

stress-strain curve of each process is shown separately in Figure 3-9. It could be observed

that the Ramberg-Osgood equation, Equation (3-6), provides good representation of cyclic

deformation behavior. In addition, the forged steel curves for the three geometrical

directions (Figure 3-9a) show that cyclic deformation curve is independent of direction. The

results of two previous studies on fatigue properties of cast aluminum A356-T6 (Wigant and

Stephens, 1987) and cast iron 65-45-12 (Meritor Automotive, 1997) are also illustrated in

Figure 3-9b and Figure 3-9c, respectively. The curves from these two studies are similar to

those found in this study. In Figure 3-10 the cyclic curves of the three materials are

compared. The cyclic yield strength (Sy') of cast aluminum and cast iron is 54% and 75% of

forged steel, and the cyclic strain hardening exponent (n') is 46% and 55% of forged steel,

respectively. This shows the higher strength of forged steel against cyclic yielding, and its

higher resistance to cyclic plastic deformation.

Superimposed plots of monotonic and cyclic curves for each material are shown in

Figure 3-11. Figure 3-12 represents the same superimposed plot for all three materials. As
86

can be seen in Figure 3-12, forged steel has mixed mode cyclic behavior. Initially, this

material cyclically softens, but then hardens at strain amplitudes larger than 0.54%. Cast

aluminum and cast iron show cyclic hardening behavior by about 25% and 30%,

respectively. Figure 3-13 shows a composite plot of the steady-state (midlife) hysteresis

loops. The stable loops from only one test at each of the strain amplitudes are shown in this

plot, even though at some strain levels more than one test was conducted.

3.3.3 Strain-Controlled Fatigue Behavior and Comparisons


Constant amplitude strain-controlled fatigue tests were performed to determine the

strain-life curve. The following equation relates the true strain amplitude to the fatigue life:

Δε Δε e Δε p σ ′f
= + = (2 N f )b + ε ′f (2 N f )c (3-9)
2 2 2 E

where σf' is the fatigue strength coefficient, b is the fatigue strength exponent, εf' is the

fatigue ductility coefficient, c is the fatigue ductility exponent, E is the monotonic modulus

of elasticity, and 2Nf is the number of reversals to failure (which was defined at 50% load

drop, as recommended by ASTM Standard E606 (1998). A summary of the cyclic properties

for the three materials is provided in Table 3-1. Table 3-3 to Table 3-5 provide the summary

of the fatigue test results for directions A, B, and C of forged steel, and Table 3-6 and Table

3-7 provide these results for cast aluminum and cast iron, respectively.

The fatigue strength coefficient, σf', and fatigue strength exponent, b, are the

intercept and slope of the best line fit to true stress amplitude (Δσ/2) versus reversals to

failure (2Nf) data in log-log scale:

Δσ
= σ ′f (2 N f )
b
(3-10)
2
87

In accordance with ASTM Standard E739 (1998), when performing the least squares

fit, the stress amplitude (Δσ/2) is the independent variable and the reversals to failure (2Nf)

is the dependent variable. This plot for the three directions of forged steel is shown in Figure

3-14a. To generate the σf' and b values, the range of data used in this figure for directions A,

B and C was chosen for 0.00175 ≤ εa ≤ 0.007. Table 3-8 compares fatigue constants for these

three directions. It could be seen that direction A provides a longer fatigue life in the high-

cycle region. It should be noted that this direction is the primary direction of stressing in the

forged steel steering knuckle. Therefore properties generated in this direction are used as the

basis for comparison with cast aluminum and cast iron.

Figure 3-15 presents the three materials S-N behavior and Figure 3-16 shows direct

comparison of the three materials with respect to S-N behavior. Direction A of forged steel

was used for this comparison. To generate the σf' and b values for cast aluminum and cast

iron, the range of data used in these figures was chosen for 0.0015 ≤ εa ≤ 0.007 and 0.0013 ≤

εa ≤ 0.005, respectively. Comparison of long-life fatigue strength, Sf , which is defined as the

fatigue strength at 106 cycles, shows that the fatigue limit of cast aluminum and cast iron are

35% and 72% of forged steel, respectively. In addition, while the fatigue strength of forged

steel at 106 cycles is expected to remain about constant at longer lives, fatigue strength of the

two cast materials continuously drops with longer lives (see Figure 3-16). Figure 3-16,

therefore, indicates significantly better S-N fatigue resistance of forged steel, as compared

with the two cast materials.

The fatigue ductility coefficient, εf', and fatigue ductility exponent, c, are the intercept

and slope of the best line fit to calculated true plastic strain amplitude (Δεp/2) versus

reversals to failure (2Nf) data in log-log scale:


88

⎛ Δε p ⎞
⎜⎜ ⎟⎟ = ε ′f (2 N f )
c
(3-11)
⎝ 2 ⎠ calculated

In accordance with ASTM Standard E739 (1998), when performing the least squares

fit, the calculated true plastic strain amplitude (Δεp/2) is the independent variable and the

reversals to failure (2Nf) is the dependent variable. Calculated true plastic strain amplitude

was determined from Equation (3-8). This plot for the three directions of forged steel is

shown in Figure 3-14b. To generate the εf' and c values, the range of data used in this figure

was chosen for 0.003 ≤ εa ≤ 0.007. Table 3-8 compares fatigue ductility constants for these

three directions. Fatigue ductility parameters determine the short-life or low-cycle fatigue

behavior of the material and its resistance against plastic deformation in that domain. Figure

3-14b shows that direction A provides better fatigue ductility than the other two directions.

Figure 3-17 presents the three materials plastic strain-life behavior and Figure 3-18

shows a direct comparison of them with this respect. The range of data used in these figures

for cast aluminum and cast iron was chosen for 0.00375 ≤ εa ≤ 0.007 and 0.002 ≤ εa ≤ 0.005,

respectively. Figure 3-18 shows the superiority of forged steel at short lives and to cyclic

plastic deformation. In automotive design, cyclic ductility is a major concern when designing

components subjected to occasional overloads, particularly for notched components, where

significant plastic deformation can occur.

The true total strain amplitude versus reversals to failure plot, based on Equation (3-

9), for the three directions of forged steel is shown in Figure 3-14c. The effect of anisotropy

and directionality of forged steel could be observed, where it is shown that direction A offers

about twice more life than the other two directions. This superiority of direction A remains

nearly the same over the whole life spectrum. As indicated earlier, the loading direction of

the forged steering knuckle induces the primary stresses in this direction.
89

Figure 3-19 displays the strain-life curves (Equation 3-9), the elastic strain portions

(Equation 3-10), the plastic strain portions (Equation 3-11), and superimposed fatigue data

for each material. Wigant and Stephens (1987) also investigated the low-cycle fatigue

behavior of A356-T6 cast aluminum. The results of their study are also shown in Figure

3-19b. Fatigue properties of 65-45-12 cast iron were also available from Meritor Automotive

(1997) and superimposed in Figure 3-19c. These Figures show that the current study resulted

in somewhat better fatigue behavior of cast aluminum and cast iron compared to the other

two studies, in the low-cycle fatigue region. This could be due to a number of reasons such

as microstructural variations, as well as degree of alignment of the test machines used.

Comparisons of strain-life fatigue behavior of the three materials, as shown in Figure

3-20, demonstrate the superiority of forged steel over cast aluminum and cast iron. Forged

steel provides about a factor of 5 longer life in short-life regime compared to cast aluminum

and cast iron. In high-cycle regime, forged steel results in more than an order of magnitude

longer life than cast iron, and about a factor of 3 longer life than cast aluminum.

The so called Neuber stress (σ aε a E ) versus life, known as Neuber plot, is shown
in Figure 3-21. This plot is useful when analyzing component geometries with stress

concentrations, where the notch root fatigue behavior is a function of both local stress and

strain. Therefore, rather than considering the individual effects of stress amplitude (Figure

3-16) or strain amplitude (Figure 3-20), a Neuber plot considers the combined effects of

both stress and strain amplitudes. This plot shows forged steel to have about two orders of

magnitude longer life than cast iron and about four orders of magnitude longer life than cast

aluminum.
90

Table 3-1 Summary of mechanical properties and their comparative ratios (forged steel is
taken as the base for ratio calculations).

Forged Steel Cast Aluminum Cast Iron


11V37 A356-T6 65-45-12
Monotonic Properties
Modulus of elasticity, E, GPa (ksi) 201.5(29,231) 78.1(11,327) 193.0(27,991)
Ratio 1 0.39 0.96
Yield strength (0.2% offset), Sy, MPa (ksi) 556.2(80.7) 232.4(33.7) 300.0(43.5)
Ratio 1 0.42 0.54
Ultimate tensile strength, Su, MPa (ksi) 821.2(119.1) 302.7(43.9) 471.2(68.3)
Ratio 1 0.37 0.57
Percent elongation, %EL (%) 21 5 10
Ratio 1 0.24 0.48
Percent reduction in area, %RA (%) 37 10 25
Ratio 1 0.27 0.68
Strength coefficient, K, MPa (ksi) 1,347.3(195.4) 417.8(60.6) 796.5(115.5)
Ratio 1 0.31 0.59
Strain hardening exponent, n 0.157 0.095 0.187
Ratio 1 0.6 1.19
True fracture strength, σf, MPa (ksi) 496(71.9) 301(43.7) 219.2(31.8)
Ratio 1 0.6 0.44
True fracture ductility, εf (%) 47 10 28
Ratio 1 0.23 0.59
Cyclic and Fatigue Properties
Cyclic modulus of elasticity, E', GPa (ksi) 194.9(28,267) 73.3(10,636) 169.4(24,568)
Ratio 1 0.38 0.87
Fatigue strength coefficient, σf', MPa (ksi) 1,156.8(167.8) 665.9(96.6) 760.8(110.3)
Ratio 1 0.58 0.66
Fatigue strength exponent, b -0.082 -0.117 -0.076
Ratio 1 1.42 0.92
Fatigue ductility coefficient, εf' 3.032 0.094 0.864
Ratio 1 0.03 0.28
Fatigue ductility exponent, c -0.791 -0.610 -0.771
Ratio 1 0.77 0.97
Fatigue strength, Sf @ 106 cycles, MPa 352(51.0) 122(17.6) 253(36.6)
(ksi) 1 0.35 0.72
Cyclic yield strength, Sy', MPa (ksi) 541.2(78.5) 290.7(42.2) 407.3(59.1)
Ratio 1 0.54 0.75
Cyclic strength coefficient, K', MPa (ksi) 1,269.5(184.1) 430.3(62.4) 649.1(94.1)
Ratio 1 0.34 0.51
Cyclic strain hardening exponent, n' 0.137 0.063 0.075
Ratio 1 0.46 0.55
Table 3-2 Summary of monotonic test results for forged steel 11V37, cast aluminum A356-T6, and cast iron 65-45-12.

Sy
wo , (offset=
SPECIMEN to , mm tf , mm wf , mm Lo , mm Lf , mm E, GPa Su , MPa K , MPa
mm 0.2%), n %EL %RA εf σf
ID (in) (in) (in) (in) (in) (ksi) (ksi) (ksi)
(in) MPa
(ksi)
Forged Steel
FS2A3 2.57 2.53 2.24 1.92 6.00 7.23 193.5 566.2 830.2 1,359.7 0.157 20% 34% 41% 469

(0.101) (0.100) (0.088) (0.076) (0.24) (0.28) (28058) (82.1) (120.4) (197.2) (68.0)

FS2A4 2.53 2.53 1.98 1.91 6.00 7.24 209.6 546.2 812.3 1,334.8 0.158 21% 41% 53% 523

(0.100) (0.100) (0.078) (0.075) (0.24) (0.29) (30404) (79.2) (117.8) (193.6) (75.9)

Average values 201.5 556.2 821.2 1,347.3 0.157 21% 37% 47% 496

(29231) (80.7) (119.1) (195.4) (71.9)

Cast Aluminum

CA1-1-3 2.58 2.59 2.44 2.46 6.00 6.32 78.1 232.4 302.7 417.8 0.095 5% 10% 10% 301

(0.102) (0.102) (0.096) (0.097) (0.24) (0.25) (11327) (33.7) (43.9) (60.6) (43.7)

Cast Iron

CI1-12 2.50 2.54 2.16 2.21 6.00 6.60 193.0 300.0 471.2 796.5 0.187 10% 25% 28% 219.2

(0.098) (0.100) (0.085) (0.087) (0.236) (0.260) (27991) (43.5) (68.3) (115.5) (31.8)

91
92

Table 3-3 Summary of constant amplitude completely reversed fatigue test results for
forged steel 11V37- Direction A.

At Midlife(N50%)
1st Cycle
Test Test Δεp/2 [a] Δεp/2 Δσ/2 σm 2N50% 2(Nf)10% 2(Nf)50%
Specimen E Value Δε/2 Failure
(calculated) (measured) MPa
[b] [c] [d]
control freq. MPa
ID GPa % location[e]
(ksi)
mode Hz % % (ksi) (ksi) reversals reversals reversals

FS1A1 196.9 strain 0.10 0.698% 0.403% 0.377% 592.8 -4.1 2,048 3,798 4,044 IGL
(28,550) (86.0) (-0.6)

FS2A1 189.9 strain 0.50 0.499% 0.231% 0.200% 539.5 -3.9 4,096 9,008 10,716 IGL
(27,542) (78.2) (-0.6)

FS3A3 208.6 strain 0.50 0.499% 0.218% 0.204% 565.5 -3.5 4,096 7,752 8,354 IGL
(30,252) (82.0) (-0.5)

FS1A5 189.7 strain 1.20 0.300% 0.067% 0.048% 468.9 45.2 16,384 41,418 46,392 IGL
(27,515) (68.0) (6.6)

FS3A6 193.6 strain 1.20 0.300% 0.070% 0.070% 463.6 4.8 16,384 33,810 34,308 IGL
(28,073) (67.2) (0.7)

FS1A3 195.6 strain 2.00 0.200% 0.009% 0.007% 385.4 14.6 4,096 661,904 661,914 IGL
(28,362) load 15.00 (55.9) (2.1)

FS3A4 206.2 strain 1.80 0.200% 0.004% 0.006% 395.5 13.0 4,096 893,294 893,316 IGL
(29,901) load 15.00 (57.4) (1.9)
FS2A2 220.5 strain 2.00 0.175% 0.002% 0.000% 347.1 30.9 33,992 471,934 471,944 IGL

(31,976) load 15.00 (50.3) (4.5) [f]


FS3A8 216.6 load 30.00 0.175% 0.000% 0.000% 352.7 0.0 - - 1,922,906 OGIT

(31,417) (51.2) (0.0)


7
FS3A7 187.5 load 30.00 0.175% 0.000% 0.000% 352.7 0.0 - - >10 not failed

[g] (27,194) (51.2) (0.0)


7
FS3A1 199.7 load 30.00 0.150% 0.000% 0.000% 302.3 0.0 - - >10 not failed

[g] (28,970) (43.8) (0.0)

[a] Δεp/2(calculated)=Δε/2-Δσ/2E.
[b] N50% is defined as the midlife cycle.
[c] (Nf)10% is defined as 10% load drop.
[d] (Nf)50% is defined as 50% load drop.
[e] IGL= inside gage length; OGIT = outside gage length but inside test section.
[f] Invalid test due to inclusion detected at failure surface. Data from this test is not included in data fittings.
[g] Data from these tests are not included in data fittings.
93

Table 3-4 Summary of constant amplitude completely reversed fatigue test results for
forged steel 11V37- Direction B.

At Midlife(N50%)
1st Cycle
Test Test Δεp/2 [a] Δεp/2 Δσ/2 σm 2N50% 2(Nf)10% 2(Nf)50%
E Value Δε/2 Failure
(calculated) (measured) MPa MPa
[b] [c] [d]
Specimen ID control freq.
GPa % location[e]
(ksi)
mode Hz % % (ksi) (ksi) reversals reversals reversals

FS1B1 193.3 strain 0.10 0.699% 0.410% 0.379% 580.6 1.4 1,024 2,404 2,702 IGL
(28,040) (84.2) (0.2)

FS2B2 196.9 strain 0.50 0.500% 0.222% 0.208% 559.5 7.8 4,096 10,238 12,476 IGL
(28,550) (81.1) (1.1)

FS1B3 189.9 strain 1.20 0.300% 0.063% 0.067% 477.0 12.5 8,192 12,110 12,484 IGL
(27,542) (69.2) (1.8)

FS2B4 208.6 strain 1.20 0.300% 0.061% 0.056% 481.6 11.2 16,384 32,768 36,326 IGL
(30,252) (69.9) (1.6)

FS2B6 189.7 strain 1.80 0.199% 0.005% 0.008% 391.6 -1.8 8,192 1,089,354 1,089,370 IGL
(27,515) load 15.00 (56.8) (-0.3)

FS1B4 193.6 strain 2.00 0.175% 0.007% 0.003% 337.7 -1.1 92,942 702,898 702,904 IGL
(28,073) load 15.00 (49.0) (-0.2)
7
FS2B1 195.6 load 30.00 0.175% 0.000% 0.000% 352.7 0.0 - - >10 not failed

[f] (28,362) (51.2) (0.0)


7
FS1B2 206.2 load 30.00 0.150% 0.000% 0.000% 302.3 0.0 - - >10 not failed

[f] (29,901) (43.8) (0.0)

[a] Δεp/2(calculated)=Δε/2-Δσ/2E.
[b] N50% is defined as the midlife cycle.
[c] (Nf)10% is defined as 10% load drop.
[d] (Nf)50% is defined as 50% load drop.
[e] IGL= inside gage length; OGIT = outside gage length but inside test section.
[f] Data from these tests are not included in data fittings.
94

Table 3-5 Summary of constant amplitude completely reversed fatigue test results for
forged steel 11V37- Direction C.

At Midlife(N50%)
1st Cycle
Test Test Δεp/2 [a] Δεp/2 Δσ/2 σm 2N50% 2(Nf)10% 2(Nf)50%
Specimen E Value Δε/2 Failure
(calculated) (measured) MPa
[b] [c] [d]
control freq. MPa
ID GPa % location[e]
(ksi)
mode Hz % % (ksi) (ksi) reversals reversals reversals

FS2C1 193.3 strain 0.10 0.697% 0.415% 0.377% 567.4 1.2 1,024 2,338 2,354 IGL
(28,040) (82.3) (0.2)

FS2C3 196.9 strain 0.10 0.698% 0.411% 0.389% 579.0 -2.6 1,024 2,080 2,166 IGL
(28,550) (84.0) (-0.4)

FS1C5 189.9 strain 0.50 0.499% 0.237% 0.198% 528.2 -2.8 2,048 5,454 6,252 IGL
(27,542) (76.6) (-0.4)

FS1C6 208.6 strain 0.50 0.499% 0.236% 0.205% 529.2 -3.5 4,096 7,276 8,080 IGL
(30,252) (76.8) (-0.5)

FS2C2 189.7 strain 1.20 0.299% 0.057% 0.066% 488.6 -23.1 8,192 24,544 24,870 IGL
(27,515) (70.9) (-3.4)

FS1C4 193.6 strain 1.20 0.300% 0.070% 0.054% 462.0 7.2 16,384 31,560 35,416 IGL
(28,073) (67.0) (1.0)

FS1C3 195.6 strain 2.00 0.199% 0.004% 0.010% 392.8 -44.5 98,678 191,256 191,260 IGL
(28,362) load 15.00 (57.0) (-6.5)

FS1C2 206.2 strain 1.80 0.200% 0.016% 0.008% 370.1 81.3 127,166 171,280 171,290 IGL
(29,901) load 15.00 (53.7) (11.8)

FS1C1 220.5 strain 2.00 0.174% 0.011% 0.004% 329.1 -3.2 80,198 267,976 267,980 IGL
(31,976) load 15.00 (47.7) (-0.5)
FS2C4 216.6 load 30.00 0.175% 0.000% 0.000% 352.7 0.0 - - 758,248 OGIT

(31,417) (51.2) (0.0)

[a] Δεp/2(calculated)=Δε/2-Δσ/2E.
[b] N50% is defined as the midlife cycle.
[c] (Nf)10% is defined as 10% load drop.
[d] (Nf)50% is defined as 50% load drop.
[e] IGL= inside gage length; OGIT = outside gage length but inside test section.
95

Table 3-6 Summary of constant amplitude completely reversed fatigue test results for
cast aluminum A356-T6.

At Midlife(N50%)
1st Cycle
Test Test Δεp/2 [a] Δεp/2 Δσ/2 σm 2N50% 2(Nf)10% 2(Nf)50%
E Value Δε/2 Failure
(calculated) (measured) MPa MPa
[b] [c] [d]
Specimen ID control freq.
GPa % location[e]
(ksi)
mode Hz % % (ksi) (ksi) reversals reversals reversals

CA1-1-4 76.9 strain 0.10 0.699% 0.322% 0.272% 294.4 -2.2 64 152 168 IGL
(11,158) (42.7) (-0.3)

CA1-1-17 73.9 strain 0.10 0.699% 0.317% 0.273% 298.2 -5.3 256 464 488 IGL
(10,715) (43.3) (-0.8)

CA1-1-1 76.1 strain 0.50 0.499% 0.125% 0.112% 291.3 -7.4 256 512 650 OGIT
(11,042) (42.3) (-1.1)

CA1-1-12 70.7 strain 0.50 0.500% 0.131% 0.087% 287.4 -1.1 512 1,408 1,416 IGL
(10,259) (41.7) (-0.2)

CA1-1-20 72.9 strain 1.10 0.374% 0.048% 0.022% 254.6 0.3 2,048 6,214 6,272 OGIT
(10,574) (36.9) (0.0)

CA1-2-9 75.8 strain 1.10 0.375% 0.035% 0.026% 265.1 8.8 4,096 7,138 10,584 IGL
(10,997) (38.4) (1.3)

CA1-1-2 76.7 strain 1.20 0.300% 0.009% 0.005% 227.4 -6.4 8,192 16,384 18,208 OGIT
(11,117) (33.0) (-0.9)

CA1-1-11 72.5 strain 1.20 0.299% 0.018% 0.005% 219.6 33.9 8,192 16,166 19,958 IGL
(10,510) (31.8) (4.9)

CA1-1-22 74.7 strain 1.50 0.225% 0.008% 0.002% 169.3 -1.9 8,192 463,482 463,486 IGL
(10,828) load 20.00 (24.6) (-0.3)

CA1-1-14 72.8 strain 1.80 0.199% 0.014% 0.002% 144.6 -7.1 16,384 862,032 862,044 OGIT
(10,558) load 20.00 (21.0) (-1.0)

CA1-1-16 78.3 load 15.00 0.200% 0.000% 0.000% 156.6 0.0 - - 308,330 IGL
(11,348) (22.7) (0.0)

CA1-1-7 77.0 load 20.00 0.149% 0.002% 0.000% 115.0 -0.1 - - 846,826 OGIT
(11,166) (16.7) (-0.0)

CA1-2-5 77.0 load 20.00 0.150% 0.002% 0.000% 115.5 -9.5 - - 1,993,158 IGL
(11,166) (16.7) (-1.4)
7
CA1-1-15 77.0 load 30.00 0.125% 0.002% 0.000% 96.3 0.0 - - >10 not failed
[f] (11,166) (14.0) (0.0)

[a] Δεp/2(calculated)=Δε/2-Δσ/2E.
[b] N50% is defined as the midlife cycle.
[c] (Nf)10% is defined as 10% load drop.
[d] (Nf)50% is defined as 50% load drop.
[e] IGL= inside gage length; OGIT = outside gage length but inside test section.
[f] Data from these tests are not used in data fittings.
96

Table 3-7 Summary of constant amplitude completely reversed fatigue test results for
cast iron 65-45-12.
At Midlife(N50%)
1st Cycle
Test Test Δεp/2 [a], [f] Δεp/2 Δσ/2 σm 2N50% 2(Nf)10% 2(Nf)50%
Specimen E Value Δε/2 Failure
(calculated) (measured) MPa
[b] [c] [d]
control freq. MPa
ID GPa % location[e]
(ksi)
mode Hz % % (ksi) (ksi) reversals reversals reversals

CI1-1 176.0 strain 0.50 0.500% 0.269% 0.247% 412.5 -10.9 512 1,032 1,122 IGL
(25,524) (59.8) (-1.6)
CI1-8 179.5 strain 0.50 0.499% 0.267% 0.245% 413.4 -7.6 1,024 2,100 2,122 IGL
(26,029) (60.0) (-1.1)
CI1-3 179.4 strain 1.20 0.299% 0.080% 0.068% 392.2 -20.0 8,192 12,348 13,816 IGL
(26,018) (56.9) (-2.9)
CI1-9 166.4 strain 1.20 0.299% 0.086% 0.065% 381.5 -5.7 8,192 13,680 13,788 IGL
(24,140) (55.3) (-0.8)
CI1-5 185.3 strain 1.80 0.197% 0.012% 0.016% 330.9 -40.3 8,192 87,888 87,894 IGL
(26,880) load 15.00 (48.0) (-5.8)
CI1-10 184.2 strain 1.80 0.199% 0.015% 0.019% 328.9 19.5 8,192 93,352 93,366 IGL
(26,710) load 15.00 (47.7) (2.8)
CI1-6 180.6 load 30.00 0.150% 0.000% 0.000% 270.9 0.0 - - 611,254 IGL
(26,192) (39.3) (0.0)
CI1-11 180.6 load 10.00 0.150% 0.000% 0.000% 270.9 0.0 - - 794,946 IGL
(26,192) (39.3) (0.0)
CI1-13 180.6 load 15.00 0.130% 0.000% 0.000% 234.8 0.0 - - 5,674,804 IGL
(26,192) (34.0) (0.0)
CI1-14 180.6 load 30.00 0.130% 0.000% 0.000% 234.8 0.0 - - 4,491,966 IGL
(26,192) (34.0) (0.0)

[a] Δεp/2(calculated)=Δε/2-Δσ/2E.
[b] N50% is defined as the midlife cycle.
[c] (Nf)10% is defined as 10% load drop.
[d] (Nf)50% is defined as 50% load drop.
[e] IGL= inside gage length; OGIT = outside gage length but inside test section.
[f] For plastic strain values, the calculated value is the average of E values from the first cycle of strain-controlled
fatigue tests. (Range: 24140-26880 ksi, average value: 25884 ksi).
97

Table 3-8 Summary of fatigue constants for the three geometrical directions A, B and C
of forged steel 11V37.

Fatigue Properties Direction A Direction B Direction C

Fatigue strength coefficient, σ´f, MPa (ksi) 1,156.8 (167.8) 1265.2 (183.5) 1316.3 (190.9)

Fatigue strength exponent, b -0.082 -0.093 -0.103

Fatigue ductility coefficient, ε'f 3.032 26.893 1.814

Fatigue ductility exponent, c -0.791 -1.062 -0.770

Fatigue strength, Sf @ 106 cycles, MPa


352 (51.0) 326 (47.2) 295 (42.7)
(ksi)

Cyclic strength coefficient, K', MPa (ksi) 1,269.5 (184.1) 1049.1 (152.2) 948.6 (137.6)

Cyclic strain hardening exponent, n' 0.137 0.106 0.094


98

Figure 3-1 Specimen configuration and dimensions (all dimensions in mm).

A
C

Figure 3-2 From left to right: forged steel, cast aluminum, and cast iron steering
knuckles. The three perpendicular geometrical directions of the forged steel
steering knuckle are; one along spindle, one along the arm perpendicular to
the body, and one along the body.
99

(a) 1000

True Stress, σ (MPa)


FS2A3 (top) FS2A4 (bottom)
0.1566 0.1582
σ =1359.7(εp) σ =1334.8(εp)
K =1359.7 MPa K = 1334.8MPa
n =0.1566 n = 0.1582
R2 = 0.9737 R2 = 0.981
100
1.0% 10.0%
True Plastic Strain, ε p (% )

(b) 1000
CA1-1-3
0.0946
σ = 417.8 (εp)
True Stress, σ (MPa)

K =417.8 MPa
n = 0.0946
R2 = 0.9966

100
1.0% 10.0%
True Plastic Strain, ε p (% )

(c) 1000
True Stress, σ (MPa)

CI1-12
0.1871
σ = 796.5 (εp)
K =796.5 MPa
n = 0.1871
R2 = 0.9972
100
1.0% 10.0%
True Plastic Strain, ε p (% )

Figure 3-3 True stress versus true plastic strain for (a) forged steel 11V37, (b) cast
aluminum A356-T6, and (c) cast iron 65-45-12.
100

(a) 900

800

Engineering Stress, S (MPa)


700

600

500

400
Specimen ID:
300 (top to bottom)
200
FS2A3
FS2A4
100

0
0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10%
Engineering Strain, e (%)

(b) 350
Engineering Stress, S (MPa)

300

250

200

150

100
Specimen ID:
50 CA1-1-3

0
0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10%
Engineering Strain, e (%)

(c) 500
Engineering Stress, S (MPa)

400

300

200

Specimen ID:
100
CI1-12

0
0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10%

Engineering Strain, e (%)

Figure 3-4 Monotonic stress-strain curves for (a) forged steel 11V37, (b) cast
aluminum A356-T6, and (c) cast iron 65-45-12.
900

800

Engineering Stress, S (MPa) 700

600

500

400

300

200
Forged Steel 11V37
100 Cast Iron 65-45-12
Cast Aluminum A356-T6
0
0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10%

Engineering Strain, e (% )

Figure 3-5 Superimposed monotonic stress-strain curves for the three materials; forged steel 11V37, cast aluminum A356-T6, and cast
iron 65-45-12.

101
102

(a) 700

True Stress Amplitude, Δσ /2 (MPa)


Strain Amplitudes:
650 (top to bottom)
εa=0.7%
600 εa=0.5%
εa=0.3%
550 εa=0.2%
εa=0.175%
500

450

400

350

300
0.0 0.2 0.4 0.6 0.8 1.0
Cycle Ratio, (N/N f )

(b) 300
True Stress Amplitude, Δσ /2 (MPa)

250

Strain Amplitudes:
(top to bottom)
200 εa=0.7%
εa=0.5%
εa=0.375%
εa=0.3%
150
εa=0.225%
εa=0.2%

100
0.0 0.2 0.4 0.6 0.8 1.0
Cycle Ratio, (N/N f )

(c) 500
True Stress Amplitude, Δσ /2 (MPa)

400

300
Strain Amplitudes:
(top to bottom)
εa=0.5%
εa=0.3%
200
εa=0.2%
ε a=0.15%
ε a=0.13%

100
0.0 0.2 0.4 0.6 0.8 1.0
Cycle Ratio, (N/N f )

Figure 3-6 True stress amplitude versus normalized number of cycles for (a) forged steel
11V37, (b) cast aluminum A356-T6, and (c) cast iron 65-45-12.
103

(a) 800
Strain

True Stress Amplitude, Δσ /2 (MPa)


700 Amplitudes:
(top to bottom)
εa=0.7%
600
εa=0.5%
εa=0.3%
500 εa=0.2%
εa=0.175%
400

300

200

100
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7
Cycles, N

(b) 300
Strain
True Stress Amplitude, Δσ /2 (MPa)

Amplitudes:
(top to bottom)
250 εa=0.7%
εa=0.5%
εa=0.375%
εa=0.3%
200 εa=0.225%
εa=0.2%

150

100
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7
Cycles, N

(c) 500
True Stress Amplitude, Δσ /2 (MPa)

Strain
Amplitudes:
(top to bottom)
400 εa=0.5%
εa=0.3%
εa=0.2%
εa=0.15%
300
εa=0.13%

200

100
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7

Cycles, N

Figure 3-7 True stress amplitude versus number of cycles for (a) forged steel 11V37, (b)
cast aluminum A356-T6, and (c) cast iron 65-45-12.
104

(a) 1000

True Stress Amplitude, Δσ /2 (MPa)

Direction A Direction B Direction C Data - Direction A


Δ σ/2 =1269.5 ( Δ εp /2)0.1372 Δ σ/2 =1049.1 (Δ εp /2)0.1056 Δ σ/2 =948.6 ( Δ εp /2)0.0942 Trendline - Direction A
K ' = 1269.5 MPa K ' = 1049.1 MPa K ' = 948.6 MPa Data - Direction B
n ' = 0.1372 n ' = 0.1056 n ' = 0.0942 Trendline - Direction B
R2 = 0.9285 R2 = 0.9140 R2 = 0.8830 Data - Direction C
Trendline - Direction C
100
0.01% 0.10% 1.00%
True Plastic Strain Amplitude, Δε p /2 (% )

(b) 1000
True Stress Amplitude, Δσ /2 (MPa)

0.0631
Δσ/2 =430.3(Δεp /2)
K ' = 430.3 MPa
n ' = 0.0631
2
R = 0.8230

100
0.01% 0.10% 1.00%

True Plastic Strain Amplitude, Δε p /2 (% )

(c) 1000
True Stress Amplitude, Δσ /2 (MPa)

0.0750
Δσ/2 =649.1 (Δεp /2)
K ' = 649.1 MPa
n ' = 0.0750
2
R = 0.8920
100
0.01% 0.10% 1.00%

True Plastic Strain Amplitude, Δε p /2 (% )

Figure 3-8 True stress amplitude versus calculated true plastic strain amplitude for (a)
forged steel 11V37 in three directions, (b) cast aluminum A356-T6, and (c)
cast iron 65-45-12.
105

(a) 900

True Stress Amplitude, Δσ /2 (MPa)


800

700

600

500

400
Data - Direction A
300 Trendline - Direction A
Data - Direction B
200 Trendline - Direction B
Data - Direction C
100
Trendline - Direction C
0
0.0% 0.1% 0.2% 0.3% 0.4% 0.5% 0.6% 0.7% 0.8% 0.9% 1.0%
True Strain Amplitude, Δε /2 (%)

(b) 400
True Stress Amplitude, Δσ /2 (MPa)

350

300

250

200

150

100
Data
50 Cyclic Stress-Strain Equation
Wigant & Stephens (1987)
0
0.0% 0.1% 0.2% 0.3% 0.4% 0.5% 0.6% 0.7% 0.8% 0.9% 1.0%

True Strain Amplitude, Δε /2 (% )

(c) 600
True Stress Amplitude, Δσ /2 (MPa)

500

400

300

200

Data
100
Cyclic Stress-Strain Equation
Meritor Automotive Inc. (1997)
0
0.0% 0.1% 0.2% 0.3% 0.4% 0.5% 0.6% 0.7% 0.8% 0.9% 1.0%
True Strain Amplitude, Δε /2 (% )

Figure 3-9 True stress amplitude versus true strain amplitude for (a) forged steel 11V37
in three directions, (b) cast aluminum A356-T6, and (c) cast iron 65-45-12.
700

600

True Stress Amplitude, Δσ /2 (MPa)

500

400

300

200

100 Forged Steel 11V37-Direction (A)


Cast Iron 65-45-12
Cast Aluminum A356-T6
0
0.0% 0.1% 0.2% 0.3% 0.4% 0.5% 0.6% 0.7% 0.8% 0.9% 1.0%

True Strain Amplitude, Δε /2 (% )

Figure 3-10 Superimposed true stress amplitude versus true strain amplitude curves for the three materials; forged steel 11V37 in
direction A, cast aluminum A356-T6, and cast iron 65-45-12.

106
107

(a) 900

800

700

True Stress (MPa)


600 Monotonic Curve
500
Cyclic Curve
400

300

200

100

0
0.0% 0.5% 1.0% 1.5% 2.0%
True Strain (% )

(b) 400

350
True Stress (MPa)

300 Cyclic Curve

250
Monotonic Curve

200

150

100

50

0
0.0% 1.0% 2.0%
True Strain (% )
(c) 500

450

400 Cyclic Curve


True Stress (MPa)

350

300 Monotonic Curve

250

200

150

100

50

0
0.0% 1.0% 2.0%

True Strain (% )

Figure 3-11 Superimposed plots of cyclic and monotonic stress-strain curves for (a)
forged steel 11V37 in direction A, (b) cast aluminum A356-T6, and (c) cast
iron 65-45-12.
700

600

Monotonic
500
Cyclic
True Stress (MPa)

Cyclic

400

Monotonic
300
Cyclic

Monotonic
200

100 Forged Steel 11V37


Cast Iron 65-45-12
Cast Aluminum A356-T6
0
0.0% 0.1% 0.2% 0.3% 0.4% 0.5% 0.6% 0.7% 0.8% 0.9% 1.0%

True Strain (% )

Figure 3-12 Superimposed plot of cyclic and monotonic stress-strain curves for the three materials; forged steel 11V37 in direction A,
cast aluminum A356-T6, and cast iron 65-45-12.

108
109

(a) 800

600

400

True Stress, σ (MPa) 200

0
-0.8% -0.6% -0.4% -0.2% 0.0% 0.2% 0.4% 0.6% 0.8%

-200
Strain Amplitudes:
-400 (starting on outside)
ε a=0.7%
ε a=0.5%
-600
ε a=0.3%
ε a=0.2%
-800
True Strain, ε (%)

(b) 400
Strain Amplitudes:
(starting on outside)
ε a=0.7% 300

ε a=0.5%
ε a=0.375% 200
True Stress, σ (MPa)

ε a=0.3%
ε a=0.225% 100
ε a=0.2%

0
-0.8% -0.6% -0.4% -0.2% 0.0% 0.2% 0.4% 0.6% 0.8%
-100

-200

-300

-400
True Strain, ε (%)

(c) 500

400

300
True Stress, σ (MPa)

200

100

0
-0.6% -0.4% -0.2% 0.0% 0.2% 0.4% 0.6%
-100
Strain
-200 Amplitudes:
(starting on
-300 outside)
ε a=0.5%
-400 ε a=0.3%
ε a=0.2%
-500
True Strain, ε (%)

Figure 3-13 Composite plot of midlife hysteresis loops for (a) forged steel 11V37 in
direction A, (b) cast aluminum A356-T6, and (c) cast iron 65-45-12.
110

(a) 1000
Data - Direction A
Trendline - Direction A

True Stress Amplitude, Δσ /2 (MPa)


Data - Direction B
Trendline - Direction B
Data - Direction C
Trendline - Direction C

Direction (A) Direction B Direction C


-0.0820 -0.0933 -0.1029
Δσ/2 =1156.8 (2Nf) Δσ/2 =1265.2 (2Nf) Δσ/2 =1316.3 (2Nf)
σf ' =1156.8 MPa σf ' =1265.2 MPa σf ' =1316.3 MPa
b = -0.0820 b = -0.0933 b = -0.1029
2 2 2
R = 0.9606 R = 0.8620 R = 0.9644
100
1E+3 1E+4 1E+5 1E+6 1E+7
Reversals to Failure, 2Nf

(b) 1.00%
True Plastic Strain Amplitude, Δε p /2 (%)

Direction A
-0.7912
Δεp /2 = 3.0315 (2Nf)
εf ' = 3.0315
c = -0.7912
2
R = 0.9653
Direction B
-1.0615
0.10% Δεp /2 = 26.8929 (2Nf)
εf ' = 26.8929
c = -1.0615
2
R = 0.9475 Direction C
-0.7695
Δεp /2 = 1.8145 (2Nf)
Direction A εf ' = 1.8145
c = -0.7695
Direction B 2
R = 0.9291
Direction C
0.01%
1E+3 1E+4 1E+5 1E+6 1E+7

Reversals to Failure, 2N f

(c) 1.00%
Data - Direction A
Trendline - Direction A
Data - Direction B
True Strain Amplitude, Δε /2, %

Trendline - Direction B
Data - Direction C
Trendline - Direction C

0.10%
1E+3 1E+4 1E+5 1E+6 1E+7 1E+8
Reversals to Failure, 2N f

Figure 3-14 (a) True stress amplitude versus reversals to failure, (b) calculated true plastic
strain amplitude versus reversals to failure, and (c) true strain amplitude
versus reversals to failure for forged steel 11V37 in three directions.
111

(a) 1000

True Stress Amplitude, Δσ /2 (MPa)

-0.0820
Δσ/2 =1156.8 (2Nf)
σf ' =1156.8 MPa
b = -0.0820
2
R = 0.9606

100
1E+3 1E+4 1E+5 1E+6 1E+7
Reversals to Failure, 2N f

(b) 1000
-0.1167
Δσ/2 =665.9 (2Nf)
True Stress Amplitude, Δσ /2 (MPa)

σf ' =665.9 MPa


b = -0.1167
2
R = 0.9275

100
1E+2 1E+3 1E+4 1E+5 1E+6 1E+7
Reversals to Failure, 2N f

(c) 1000
-0.0758
Δσ/2 =760.8 (2Nf)
True Stress Amplitude, Δσ /2 (MPa)

σf ' =760.8 MPa


b = -0.0758
2
R = 0.9692

100
1E+2 1E+3 1E+4 1E+5 1E+6 1E+7
Reversals to Failure, 2N f

Figure 3-15 True stress amplitude versus reversals to failure for (a) forged steel 11V37 in
direction A, (b) cast aluminum A356-T6, and (c) cast iron 65-45-12.
Figure 3-16 Superimposed true stress amplitude versus reversals to failure for the three materials; forged steel 11V37 in direction A, cast
aluminum A356-T6, and cast iron 65-45-12.

112
113

(a)

True Plastic Strain Amplitude, Δε p/2 (%)


1.00%
-0.7912
Δεp /2 = 3.0315 (2Nf)
εf ' = 3.0315
c = -0.7912
2
R = 0.9868

0.10%

0.01%
1E+3 1E+4 1E+5 1E+6 1E+7

Reversals to Failure, 2N f

(b) 1.00%
True Plastic Strain Amplitude, Δε p/2 (%)

-0.6095
Δεp /2 = 0.0944 (2Nf)
εf ' = 0.0944
c = -0.6095
2
R = 0.9193

0.10%

0.01%
1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8

Reversals to Failure, 2N f

(c) 1.00%
-0.7609
Δεp /2 = 0.8645 (2Nf)
True Plastic Strain Amplitude, Δε p/2 (%)

εf ' = 0.8645
c = -0.7609
2
R = 0.9590

0.10%

0.01%
1E+2 1E+3 1E+4 1E+5 1E+6

Reversals to Failure, 2N f

Figure 3-17 Calculated true plastic strain amplitude versus reversals to failure for (a)
forged steel 11V37 in direction A, (b) cast aluminum A356-T6, and (c) cast
iron 65-45-12.
1.000%

True Plastic Strain Amplitude (%)

0.100%

0.010%

Forged Steel 11V37


Cast Iron 65-45-12
Cast Aluminum A356-T6
0.001%
1E+2 1E+3 1E+4 1E+5 1E+6

Reversals to Failure, 2N f

Figure 3-18 Superimposed calculated true plastic strain amplitude versus reversals to failure for the three materials; forged steel 11V37
in direction A, cast aluminum A356-T6, and cast iron 65-45-12.

114
115

(a) 10.00%
Strain-Life Equation

True Strain Amplitude, Δε /2, %


Plastic Strain Data
Elastic Strain Data
Total Strain Data
1.00%
Δε / 2
Δ ε p /2

Δε e / 2

0.10%

0.01%
1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8
Reversals to Failure, 2N f

(b) 10.00%
Strain-Life Equation
True Strain Amplitude, Δε /2, %

Plastic Strain Data


Elastic Strain Data
Total Strain Data
1.00% Δε / 2
Wigant & Stephens (1987)

Δε e / 2

0.10%

Δ ε p /2

0.01%
1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8

Reversals to Failure, 2N f

(c) 10.00%
Strain-Life Equation
True Strain Amplitude, Δε /2, %

Plastic Strain Data


Elastic Strain Data
Total Strain Data
1.00% Meritor Automotive Inc. (1997)
Δε / 2

Δε e / 2

0.10%

Δ ε p /2

0.01%
1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8

Reversals to Failure, 2N f

Figure 3-19 True strain amplitude versus reversals to failure for (a) forged steel 11V37 in
direction A, (b) cast aluminum A356-T6, and (c) cast iron 65-45-12.
1.00%
Forged Steel 11V37

Cast Aluminum A356

Cast Iron 65-45-12

True Strain Amplitude (%)

0.10%
1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8

Reversals to Failure, 2Nf

Figure 3-20 Superimposed true strain amplitude versus number of reversals to failure for the three materials; forged steel 11V37 in
direction A, cast aluminum A356-T6, and cast iron 65-45-12.

116
Figure 3-21 Neuber fatigue life curves for forged steel 11V37, cast aluminum A356-T6, and cast iron 65-45-12.

117
Chapter Four

4 Finite Element Analysis


The objective of the stress analysis was to obtain the complete three dimensional

stress and strain distributions at a potential failure site, facilitating fatigue life predictions.

Depending on the method of fatigue life prediction, stress analysis can be linear or

nonlinear. Linear elastic analysis is the most common type of stress analysis pursued in

automotive design and analysis. The majority of automobile engine components like

camshaft, crankshaft and connecting rod operate within the material elastic region, and

therefore linear elastic analysis is feasible to model their behavior. On the other hand, most

components in chassis and suspension undergo occasional overloads in service and cyclic

plasticity becomes a major factor to define their stress-strain response, particularly at stress

concentrations. In these cases linear analysis is not sufficient to analyze component’s

behavior and nonlinear elastic-plastic analysis should be performed.

Although in this study the components underwent unidirectional loading condition,

the state of stress at many locations was found to be multiaxial. In addition, gross yielding

existed for a number of experimental applied moment cases. Therefore, both linear finite

element analysis (FEA) to be used along with a stress/strain correction method, and

nonlinear FEA to directly estimate elastic-plastic distributions of stress and strain were

performed. IDEAS FEA program and its help library bookshelf were employed for the

analysis. Details of the FEA are discussed in this chapter.

118
119

4.1 Component Geometry Generation


The geometries of components were generated as solid parts using a coordinate

measuring machine and the modeler tools of the software. The resulting solid models are

presented in Figure 4-1. The weights of the generated models were 2.35 kg, 2.38 kg and 4.62

kg versus the actual components weight of 2.4 kg, 2.4 kg and 4.7 kg for the forged steel, cast

aluminum and cast iron steering knuckles, respectively. These show differences of 2%, 1 %

and 2 %, respectively, and verify the accuracy of the geometries generated for analysis. The

fillets and chamfers of the components were removed (except at the critical locations) in the

models used for the analysis in order to reduce the complexity of the models and the run-

time.

4.2 Model Boundary Conditions


The boundary conditions and loading were selected to represent the assumed

component testing conditions as shown in Figure 4-2. To verify the model with the specified

boundary conditions other alternatives were analyzed by switching the loading and boundary

conditions and also by releasing any one of the fixed points to ensure the critical locations

remained the same.

For the cast aluminum steering knuckle in service while the loading is applied to the

strut joints through struts, the four hub bolt holes are connected to the wheel assembly.

Several trials for boundary conditions were analyzed including fixing the whole area of the

four hub bolt holes, fixing the centerline of the hub bolt holes, only fixing the pair of bolt

holes away from the load application point, and fixing two points at the middle area of the

hub. It was found that except for the case of fixing the whole area of the bolt holes, all the

other three cases provided approximately similar results at the critical location. This critical
120

location was the node at the area that the crack initiated during component testing (Section

5.5). Moreover, the strain readings from strain gages were reasonably close to the strain

values for these three options as discussed in Section 5.3. The option to fix the whole area of

the bolt hole resulted in lower value of stress at the node of this critical location and the

highest stresses occurred at a different location. In addition, for this option the strain value

at the location where strain gage was installed was different from the strain gage reading.

Based on these two observations, the choice of fixing the hub bolt hole-centerlines was

selected as the base boundary condition case for the cast aluminum steering knuckle. For the

cast iron steering knuckle, where the geometry and service conditions were close to the cast

aluminum steering knuckle, similar loading and boundary conditions were applied.

4.3 Meshing of the Finite Element Models


To generate the mesh for the components, free meshing feature of the software was

employed since it has no geometry restrictions and it could be defined on complicated

volumes. The free mesh generator uses an algorithm that minimizes element distortion.

Three-dimensional linear tetrahedron solid elements with three translational degrees of

freedom were used. Although parabolic element offers more accurate results, the analysis

time, especially for nonlinear analysis, is immensely more than that of linear elements.

Nonlinear analysis at the highest moment level with parabolic elements showed negligible

difference in results with those using linear tetrahedral element and, therefore, the linear

element was selected for modeling. For the complex three-dimensional geometry of the

steering knuckles, solid element offers more accurate results and IDEAS-10 software

employs tetrahedron solid elements. In spite of a global mesh size for each component, free
121

local meshing feature was used to increase the number of elements at the vicinity of the

critical points.

Convergence of stress and strain energy was considered as the criteria to select the

mesh size. Too much refinement at the critical points would result in extremely lengthy

analysis time and was, therefore, avoided. Figure 4-3 shows the variation of stress at the

critical points of each steering knuckle. Global mesh size of 5.1 mm for the forged steel and

cast aluminum steering knuckles and 3.8 mm for the cast iron steering knuckle and local

mesh size of 0.1 mm for the forged steel and 0.64 mm for the cast aluminum and cast iron

steering knuckles were specified. The critically-stressed locations were spindle fillets for the

forged steel, and hub bolt-holes for the cast aluminum and cast iron steering knuckles. Due

to higher stress concentration at the spindle fillet of the forged steel steering knuckle, large

stress gradient existed at this location, and therefore, more refined mesh size was

implemented. As another verification of convergence of the mesh, the averaged and

unaveraged stress values at the critical nodes were confirmed to be approximately equal.

Figure 4-4 shows the three meshed components with the darker areas representing those

with higher mesh density.

Adaptive meshing module as a tool to improve a mesh by moving nodes, splitting

elements, or remeshing the model, was examined. The software derives the new mesh by

analyzing the data variation along the boundaries and within the interior regions of the faces.

Adaptive meshing could be used to reduce elemental distortion or refine a mesh in areas

where error estimates are highest. Energy error norm is used as basis for the adaptive

meshing method. The energy error norm (EEN) is based on the strain energy error. It is

computed from the elemental strain energy errors and the model’s total strain energy from

Zienkiewicz-Zhu parameter (IDEAS-10 Help Library Bookshelf, 2002):


122

⎡ ∑ elemental strain energy errors ⎤


Energy error norm = ⎢ ⎥ × 100 (4-1)
⎣ total strain energy ⎦

where the summation is over the entire volume of the component.

The Convergence is determined by comparing EEN in the model to the EEN

specified. For the forged steel steering knuckle with an initial mesh size of 5.08 mm, EEN

was calculated to be 10.7%. Remeshing option of adaptive meshing that increases the

number of elements at the areas with high strain energy error was applied and the first

adaptive trial resulted in EEN equal to 7.9%. No more adaptive meshing was possible due to

high distortion of the newly generated elements. This is attributed to the geometric

complexity of the component. The component’s appropriate loading and boundary

conditions were applied to the refined model and the results were compared to the results

from the converged mesh using free local meshing, where they were found not to be equal.

In addition, the averaged and unaveraged stress and strain results for the adaptive-meshed

model were compared and they were not equal either. These show that the mesh did not

converge. As a result, the final meshed models used for solution were those with free local

meshing and the adaptive meshing tool was found not to be suitable for the components.

4.4 Material Nonlinear Behavior Models


To define the limit of elastic behavior in nonlinear analysis, von Mises yield criterion

was assumed, using an associated flow rule in which the plastic potential function is the same

as the yield function and the components of the plastic strain increment are given by a

Prandtl-Reuss type equation. Ziegler-Prager kinematic hardening rule was selected, for which

a bilinear stress-strain curve of the material is assumed. The appropriate portion of the

material’s cyclic stress-strain curve was used to define material properties for each cyclic
123

moment level; i.e. for each applied cyclic moment level, a slope for the second line of the

bilinear model was assumed and the stress and strain results were obtained from FEA. This

slope was found by trail and error in an iterative process until the resultant stress-strain point

matched the experimental material cyclic stress-strain curve. Figure 4-5 shows the material

model for the two higher moment levels.

Figure 4-6 illustrates the actual cyclic stress-strain curves at different cyclic moment

levels including the values of stress and strain at the critical points of failure at each level

obtained from FEA also superimposed. It was observed that the assumed model provides

stress/strain values in the component with a reasonable error of less than 10% with respect

to the actual material stress-strain curve. Note that even at the lower moment level, which

can be considered as an indication of long-life service of the components, the material

undergoes local plastic deformation. This is evidence that mere use of linear elastic FEA is

not sufficient for reliable fatigue life predictions.

4.5 Model Solutions


The model solution equation solver for linear or nonlinear analysis creates a set of

simultaneous linear equations that must be solved. For instance, the number of equations

created for the highest load level of the forged steel steering knuckle model with 4245 nodes

and 19098 elements was 12570 equations. The direct method is represented by Gaussian

elimination along with Choleski decomposition or factorization method. With Choleski

decomposition, the matrix of equations is factorized into a lower triangular, diagonal, and

upper triangular matrix; the upper triangular matrix is the transpose of the lower triangular

matrix. After factorization, the equations are solved for the unknown values by performing a
124

forward and backward substitution upon the load vector. Following decomposition of the

stiffness matrix, the displacements are calculated and strain and stresses are obtained.

A nonlinear solution proceeds incrementally with equilibrium iterations performed at

every solution point. The stiffness matrix in the nonlinear solution doesn’t participate in the

equilibrium statement as is the case in linear analysis. The stiffness relates incremental

displacements to incremental forces. The software computes tangent stiffness based on

current geometry, stress, and plastic strain. Then it computes applied loads based on current

configuration, and computes internal force based on current displacements and plastic

strains. In the next step, it computes force residual, computes displacement increment,

updates displacements, computes plastic strain increments at element integration points, and

checks for convergence. Energy convergence ratio is the convergence criterion that uses

both displacements and forces. The amount of work done by the residual forces in the ith

iteration is compared with the first iteration (IDEAS-10 Help Library Bookshelf, 2002). As

an example, the nonlinear solution for the highest load level of the forged steel steering

knuckle took four iterations to converge.

4.6 Analysis of the Results


The finite element models as shown in Figure 4-2 were solved for loading and

unloading under the assumed component testing conditions with the assumed moment

levels as listed in Table 5-2. The procedure used to obtain maximum and minimum stresses

and strains in a loading cycle is as follows; first, loading was applied to the FEA model at the

maximum moment assumed to be used in the test and von Mises stress at the critical

location was obtained (σVM,max). Then the moment range assumed to be used in the

experiment was applied to the FEA model and von Mises stress range at the same critical
125

location was obtained (ΔσVM). It should be noted that when using the maximum moment,

the stress-strain relation used was the material cyclic curve based on amplitudes, whereas

when using the moment range, the cyclic stress-strain curve was based on ranges (i.e.

equation of the hysteresis loop). Having σVM,max and ΔσVM , the range and mean values of

stresses and strains were then calculated.

Figure 4-7 shows contours of von Mises stress for the maximum moment level of

1515 N.m for forged steel and 2230 N.m for cast aluminum and cast iron steering knuckles.

For the forged steel and cast aluminum steering knuckles at the higher moment levels,

yielding occurred both gross (at the spindle and the hub, respectively) and locally (at the fillet

and hub bolt-holes, respectively), whereas for the cast iron steering knuckle only local

yielding occurred at the critical points (hub bolt-hole). The contours of stress and strain at

the lower moment levels were different, showing only local plastic deformation, as shown in

Figure 4-8. In this case, the spindle and the hub area for the forged steel and cast aluminum

steering knuckles undergo elastic deformation.

Table 4-1 lists the x, y and z components, principal and equivalent von Mises values

of stress and strain for the three components at the critical locations for the highest moment

level applied in the analysis. The z direction aligns with the spindle axis for the forged steel

steering knuckle, and the y and x directions align with the strut arm for the cast aluminum

and cast iron steering knuckles, respectively. The spindle 1st step fillet area for the forged

steel and hub bolt hole for the cast aluminum and cast iron steering knuckles were found to

be the high-stressed locations with high stress gradient. The effect of stress concentration

can be seen in Figure 4-9 that shows the distribution of local stress at the spindle 2nd step

fillet and nominal stress at a cross section remote from the fillet for the maximum moment

level of 1515 N.m. For spindle radius smaller than 10 mm the material behavior is elastic,
126

while for larger radii it becomes inelastic. Figure 4-10 shows the variation of stress at

different locations of the forged steel steering knuckle with respect to moment steps to

provide a comparison between the level of stress at the critical location and other locations.

The stress concentration produced by a notch depends on the mode of loading applied to

the component. Not only the stress concentration at the fillet of the forged steel steering

knuckle increases the stress level, but also the stress gradient is higher than, for instance,

tension loading due to the loading mode (bending).

Examining the data in Table 4-1 clearly demonstrates the multiaxial nature of stress

and strain at the critical locations, although the primary loading on the components are

unidirectional. The ratio of σy/σz in the case of the forged steel steering knuckle is 0.36,

σx/σy for the cast aluminum steering knuckle is 0.69, and τyz/τxy is -0.36 for the case of the

cast iron steering knuckle. Therefore, equivalent values of stress and strain should be

calculated to account for multiaxiality. Since the stresses and strains obtained form these

analyses were used to predict fatigue life, the proportionality of stresses is also an important

issue for selection of an appropriate fatigue life prediction model. Only one source of load

exists for the primary loading that these components undergo, therefore the stresses are

proportional throughout the components, i.e. the stresses increase and decrease in-phase as

the primary loads increase and decrease, respectively. For proportional stressing, von Mises

stress and strain have been found effective in calculating the equivalent values as a result of

multiaxiality (Stephens et al., 2000), and were used for subsequent fatigue life analyses.

Another observation that could be made from these data is the stress state at the

critical locations. The values of σ2/σ1 for the critical locations are 0.36, 0.12 and 0.06 while

the values of ε2/ε1 are -0.09, -0.35 and -0.13 for the forged steel, cast aluminum and cast iron

steering knuckles. This indicates that at the critical location, the state of nearly plane strain
127

prevails for the forged steel steering knuckle, while the state of stress at the critical locations

of the cast aluminum and cast iron steering knuckles is closer to plane stress. Slightly

different results were obtained for lower moment levels due to limited plastic deformation,

but the same conclusion about the state of stress could be drawn for them too.

An important point for the fatigue analysis to be mentioned here is the effect of the

notch on local stresses and strains due to cycling. Cyclic loading generates residual stress at

the notch and therefore, makes the local deformation behavior different from the nominal

behavior. Table 4-2 lists the nominal moment ratios and local stress ratios as results of

loading and unloading simulations of the components. As the nominal R-ratio remains

almost constant (close to zero), significant negative R-ratio is observed for most of the

simulations as a result of the residual stress generated at the stress concentrations (i.e. critical

locations). This phenomenon is more pronounced at the higher moment levels, as expected,

because of increased plastic deformation and nonlinear material behavior.


128

Table 4-1 Components of stress and strain for the critical locations of the forged steel
(M = 1515 N.m), cast aluminum (M = 2230 N.m) and cast iron (M = 2230
N.m) steering knuckles.

Forged Steel Cast Aluminum Cast Iron


(2nd step fillet) (hub bolt hole) (hub bolt hole)

Stress
σx (MPa) 44 145 -8
σy (MPa) 258 210 200
σz (MPa) 710 40 -6
τxy (MPa) 3 146 207
τxz (MPa) 43 -16 36
τyz (MPa) -3 -32 -74
σ1 (MPa) 713 332 334
σ2 (MPa) 259 41 19
σ3 (MPa) 42 23 -167
τmax (MPa) 336 154 250
σVM (MPa) 603 296 431

Strain
εx -0.00430 0.00110 -0.00160
εy -0.00060 0.00340 0.00360
εx 0.00700 -0.00270 -0.00160
εxy 0.00010 0.01000 0.01100
εxz 0.00140 -0.00110 0.00180
εyz -0.00009 -0.00230 -0.00380
ε1 0.00710 0.00770 0.00710
ε2 -0.00064 -0.00270 -0.00096
ε3 -0.00430 -0.00330 -0.00570
γmax 0.01100 0.01100 0.01300
εVM 0.00670 0.00710 0.00740
129

Table 4-2 Nominal moment ratios (Rm) and local stress ratios (Rσ) indicating variation of
Rσ resulting from different levels of notch residual stress generated during
cycling as a function of Mmax.

Nominal Local
Mmax Mmin Moment σVM, max σVM, min Stress
(N.m) (N.m) Ratio (MPa) (MPa) Ratio

Forged Steel
1515 75 0.05 603 -264 -0.44
1240 75 0.06 565 -214 -0.38
965 75 0.08 514 -123 -0.24
825 75 0.09 478 -61 -0.13

Cast Aluminum
2230 75 0.03 296 -134 -0.45
1595 75 0.05 279 -24 -0.09
1305 75 0.06 251 5 0.02
1195 75 0.06 237 14 0.06

Cast Iron
2230 75 0.03 431 -276 -0.64
1595 75 0.05 412 -229 -0.56
1305 75 0.06 405 -114 -0.28
1195 75 0.06 388 -85 -0.22
130

Figure 4-1 From left to right the solid models of the forges steel, cast aluminum and
cast iron steering knuckles.

Figure 4-2 From top to bottom boundary conditions for the forged steel, cast aluminum
and cast iron steering knuckles.
131

(a) Local Mesh Size (mm)


1.0 0.8 0.6 0.4 0.2 0.0
300

Equivalent von Mises Stress (MPa)


325

350

375

400

425

450

475

500

525

(b) Local Mesh Size (mm)


1.0 0.8 0.6 0.4 0.2 0.0
120
Hub bolt hole 1

Equivalent von Mises Stress (MPa)


Hub bolt hole 2 130

140

150

160

170

180

190

(c) Local Mesh Size (mm)


1.0 0.8 0.6 0.4 0.2 0.0
0
Equivalent von Mises Stress (MPa)

Hub front bolt hole


Arm root
100

200

300

400

Figure 4-3 von Mises stress versus mesh size at critical locations of (a) forged steel , (b)
cast aluminum and (c) cast iron steering knuckles to check mesh
convergence.
132

Figure 4-4 From top to bottom, the generated mesh for the forged steel, cast aluminum
and cast iron steering knuckles.
133

700
Forged Steel THE MATERIAL RAMBERG-OSGOOD TYPE BEHAVIOR
Cast Iron
600
Cast Aluminum
Stress Amplitude (MPa)

500

400

300

200

100
THE BILINEAR MODEL DEFINED FOR THE SOFTWARE

0
0 0.2 0.4 0.6 0.8 1
Strain Amplitude (%)

Figure 4-5 Superimposed cyclic stress-strain curves and the bilinear models used in the
nonlinear FEA for the forged steel, cast iron and cast aluminum steering
knuckle materials. The data points are from the material cyclic tests.
134

700
350

600 300

500 250

400 200

150
300
Local Stress (MPa)

Local Stress (MPa)


100
200
50
100
0
0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 -50
-100
-100
-200
-150

-300 -200

-400 -250

-500 -300
Local Strain (%) Local Strain (%)

(a) (b)

500

400

300

200
Local Stress (MPa)

100

0
0 0.2 0.4 0.6 0.8

-100

-200

-300

-400
Local Strain (%)

(c)

Figure 4-6 Actual material cyclic stress-strain behavior and the corresponding local von
Mises stress and strain from FEA (shown with plus signs) for (a) forged steel,
(b) cast aluminum and (c) cast iron steering knuckles.
135

Figure 4-7 From top to bottom, contours of von Mises stress at highest moment levels
for forged steel (Mmax = 1515 N.m), cast aluminum (Mmax = 2230 N.m) and
cast iron (Mmax = 2230 N.m) steering knuckles. The stress values of the color
bar are in MPa.
136

Figure 4-8 From top to bottom, contours of von Mises stress at lowest moment levels
for forged steel (Mmax = 825 N.m), cast aluminum (Mmax = 1195 N.m) and
cast iron (Mmax = 1195 N.m) steering knuckles. The stress values of the color
bar are in MPa.
137

700
Notch Behavior
600 Nominal Behavior
Maximum Stress (MPa)

500

400

300
NOMINAL
NOTCH
200

100

0
0 2 4 6 8 10 12 14 16
CENTER Radius (mm) SURFACE

Figure 4-9 The distribution of stress at the spindle 2nd step fillet and at a section remote
from the 2nd step fillet of the forged steel steering knuckle for a moment of
1515 N.m. The nominal stress profile was used to obtain the nominal values
of stress.
138

2 1

4 3

10

400
Signed von Mises Stress (MPa)

300
Location 1
200 Location 2
Location 3
100
Location 4
Location 5
0
Location 6

-100 Location 7
Location 8
-200 Location 9
Location 10
-300

-400
0 200 400 600 800
Moment Steps (N.m)

Figure 4-10 Signed von Mises † stress history at different locations of the forged steel
steering knuckle.

†Signed von Mises uses the sign of the absolute maximum principal stress to give either a positive or negative
von Mises stress (This procedure is used in FEA/fatigue software codes such as nCodeTM and MSC.FatigueTM).
Chapter Five

5 Component Fatigue Behavior and Comparisons


In fatigue design problems, material properties, loading, geometry, manufacturing

aspects, and service environment are considered as five major aspects to be dealt with, and

the effect of geometry has always been one of the challenging issues. The methodologies to

transfer material properties obtained from specimen fatigue tests to fatigue behavior of real

components where neither a nominal stress nor a notch factor could be defined, have been

questionable. In addition, simulating manufacturing parameters like surface finish and

residual stress, if not impossible, is complicated. Therefore direct component testing, though

time consuming and expensive, has become a necessity in fatigue design. Contrary to

specimen testing, for which numerous standards have been developed, component testing is

more a matter of designer’s practice. Various parameters such as simulation of the real-life

condition, collecting useful data, and relating test results to analytical predictions determine

the applicability of the conducted tests.

To be able to compare the fatigue behavior of the steering knuckles, and to make it

possible to verify analytical life predictions, constant-amplitude load-control fatigue tests

were performed for forged steel and cast aluminum steering knuckles. The component

testing was only conducted for forged steel and cast aluminum steering knuckles. In this

chapter, the procedure to conduct the component tests is detailed, including considerations

prior to and during the tests, and test results are presented and discussed.

139
140

5.1 Identifying the Components and their Service


Loadings
The suspension systems of the vehicle, which the steering knuckles belong to, were

identified and the loading and attachment conditions were investigated. The forged steel

steering knuckle belongs to the rear suspension system of a four-cylinder sedan vehicle. It is

a symmetric component with one plane of symmetry. The cast aluminum steering knuckle

belongs to the front suspension of a six-cylinder minivan. The nomenclature for these

components is shown in Figure 5-1.

Figure 5-2 and Figure 5-3 show the forged steel and cast aluminum steering knuckles

as installed in the suspension system of the vehicles, respectively. For the case of the forged

steel steering knuckle (Figure 5-2), the strut mounting holes are connected to the strut joints,

the front and rear lateral links connect to the chassis and the tension strut joint is attached to

the tension strut that is fixed to the chassis bracket. The hub and bearing assembly mount on

the spindle. The hub bearing sits on the spindle middle step. The inner part of the hub

attaches to the mounting holes, while the outer part connects to the wheel and is free to

rotate. For the case of the cast aluminum steering knuckle (Figure 5-3), the strut mounts on

the steering knuckle vertical arm, and the control arm and the stabilizer bar attach to the

horizontal arm and lower body. The body also attaches to the caliper from its two outer bolt

holes and to the wheel hub from the four mounting holes.

The primary loading condition for the forged steel and cast aluminum steering

knuckles was simulated as shown in Figure 5-4 and Figure 5-5, respectively. For the forged

steel steering knuckle (Figure 5-4), because of the symmetrical round geometry of the

spindle, the forces and moment in the y or z directions result in a uniaxial stress along the

spindle direction and a failure location at the second step of the spindle. Therefore, the
141

loading could be simplified to a single moment applied to the spindle in the y or z direction.

The torsional moment and axial force in the x direction are minor due to the presence of

bearings on the spindle. In addition, attachment of the hub assembly to the steering knuckle

mounting holes prevents the transfer of these loads to the body. For the cast aluminum

steering knuckle (Figure 5-5) the primary loading is in the form of a moment applied in the y

direction to the arm, while the body is fixed through its four bolt hole.

5.2 Fixture Assembly Design and Preparation


Prior to preparing the fixture parts, finite element models of the simulated loading

conditions were analyzed in order to determine and verify the critical locations and the

loading and restraint arrangements. The critical points of highest stress in the components

for the primary loading conditions discussed in Sections 5.1 were detected from the stress

analysis, as described in Chapter 4. Accordingly, specific test fixtures for each one of the two

steering knuckles were designed and machined.

The exploded view of the forged steel steering knuckle fixturing is shown in Figure

5-6 and the assembled model and the real fixture assembly is shown in Figure 5-7. In this

arrangement the spindle was fixed by a 2-piece block where threaded rods tightened the

block to the spindle. A pair of L-shaped moment arms transferred the load from the testing

machine loading actuator to the spindle blocks in the form of bending load. The strut and

suspension connections on the steering knuckle body were fixed to the bench using round

and square blocks.

The exploded view of the cast aluminum steering knuckle fixturing is shown in

Figure 5-8. For this component a two-strut-attachment test was conducted, as shown in the

assembled model and the real fixture assembly of Figure 5-9. In this arrangement, the strut
142

attachment of the arm was connected from both sides to a pair of moment arms. The

moment arms transferred the bending load from the loading actuator to the steering knuckle.

The four hub bolt holes were fixed to the bench.

5.3 Pre-Test Analysis and Considerations


A number of considerations were made to ensure correctness of the tests. Strain

gages were positioned on the components and the strain readings were compared to

analytical values and finite element analysis (FEA) results. To validate the test setups, values

of strains as measured by strain gages in component testing and as predicted using FEA

were compared, and are listed in Table 5-1. The strain gages for the forged steel steering

knuckle were positioned at the vicinity of the spindle root and the first step fillets, and for

the cast aluminum steering knuckle two gages were positioned at the goose neck of the strut

arm and two at the hub bolt holes where crack initiation was observed during component

testing. These locations are identified in Table 5-1. Depending on the location of the gage,

the proper component of the strain obtained from the FEA was selected for comparison.

The differences between measured and predicted strains obtained for the two

steering knuckles were less than 18% and were considered reasonable for the complex

steering knuckle geometries. The error can be attributed to: 1) difficulty in matching the

exact corresponding node on the FEA model to the strain gage location, 2) difference in

boundary conditions between FEA and actual test setup, 3) the unwanted friction force that

exists between the point of load application and the moment arms in test (although this was

reduced significantly by using roller bearings), and 4) measurement errors related to strain

gage averaging particularly for cast aluminum steering knuckle due to high strain gradient at
143

the strain gage locations. The measured strains also confirmed the symmetry of and linear

variations with elastic loading.

For the forged steel steering knuckle, which has a relatively simpler geometry, results

of strain calculations from analytical equations of mechanics of materials are also listed in

Table 5-1. These results are mostly in between the measured and FEA-predicted strains. It

should be noted that the position of the strain gages and the magnitudes of the applied loads

were such that all measured strains were in the elastic range. In addition, it was confirmed

that change of the gripped spindle length for the forged steel steering knuckle does not

affect the strain reading at the locations close to the spindle step.

For the cast aluminum steering knuckle, the bolt pretension and the accompanied

compressive stress on the engaged area of the component was estimated using theory of

machine design (Shigley and Mischke, 1989) for a bolt torque of 13.6 N.m (applied to the

bolt during the component tests) and a bolt having major thread diameter of 12.7 mm to be

5.4 kN and 14 MPa (2 ksi), respectively. The stress value was negligible compared to the

stresses generated due to the experiment loading and were not considered in subsequent

stress analysis and predictions.

A closed-loop servo-controlled hydraulic 100 kN axial load frame was used to

conduct the tests. The calibration of the system was verified prior to the beginning of the

tests. As shown in Figure 5-6 to Figure 5-9, a rod end bearing joint was used to apply the

load from the actuator to the moment arms, in order to avoid any out of plane bending. Due

to relative rigidity of the fixtures, the effect of horizontal friction force was found to be

significant at the joint-fixture contact point. Therefore, a needle roller bearing was installed

on each side of the pin of the bearing, allowing the moment arm to roll horizontally to

minimize friction force. Figure 5-10 shows the details of attachment of the rod-end and
144

needle roller bearings to the load cell and moment arms for the forged steel steering knuckle

testing. Care was taken to ensure symmetry of the bending load transferred from the two

moment arms. All fixture bolts and nuts were tightened with identical torque values to

maintain consistency.

Table 5-2 lists the maximum moment, moment amplitude, ratio of maximum to

minimum moments, and test frequencies. The moment levels were determined based on

stress analysis results, the true stress-true strain curve of the materials, and standard load

cases. The actual load cases for these specific components were not available, but two

standard load cases, Lotus load condition that is used in Daewoo Motors (Lee and Lee,

2003) and ULSAS standard load cases (AISI, 2001) were examined. The primary load of 3g

acting on the wheel is the vertical bump for both of the load cases, where g is the

acceleration of gravity. Based on component weights of 1195 kg (2635 lb) and 1983 kg (4372

lb), and the wheel hub moment arms (from the center of the wheel bearings to the critical

location of the components) of 21.9 mm and 25.4 mm, maximum moments of 1000 N.m

and 2000 N.m were obtained for the forged steel and cast aluminum steering knuckles,

respectively. The highest maximum moment levels in the component test were specified to

be 1515 N.m and 2230 N.m for the forged steel and cast aluminum steering knuckles,

respectively.

A minimum moment of 75 N.m was used in all tests, corresponding to an R-ratio

(Mmin/Mmax) of less than 0.08. A total of seven component tests at four moment levels with

amplitudes between 380 N.m and 720 N.m for the forged steel steering knuckle, and a total

of six tests at four moment levels with amplitudes between 560 N.m and 1075 N.m for the

cast aluminum steering knuckle were conducted. The frequency of the tests ranged from 0.5
145

Hz for higher moment levels, to 5 Hz for lower moment levels. The moment levels chosen

resulted in fatigue lives between 104 and 2x106 cycles.

5.4 Collecting Experimental Data


Displacement amplitude versus cycle data of the components during each test were

monitored in order to record macro-crack nucleation (i.e. a crack on the order of several

millimeters), growth, and fracture stages. Due to the nature of the loading and restraints on

both steering knuckles, the locations of crack initiation could not be reached to enable

detecting crack nucleation. Therefore, a marked displacement amplitude increase during the

test was considered as the crack nucleation point, and a sudden increase as the final fracture.

Variations of displacement amplitude versus cycles for two typical tests of the forged

steel and cast aluminum steering knuckles are shown in Figure 5-11. As can be observed

from this figure, for the forged steel steering knuckle the displacement amplitude was nearly

constant until about the end of the test. This indicates that the time lag between macro-crack

nucleation and fracture was a small fraction of the total life. On the other hand, for the cast

aluminum steering knuckle the crack growth portion of the life was significant. The crack

lengths of the cast aluminum steering knuckles were also visually observed and recorded. For

the typical cast aluminum steering knuckle data in Figure 5-11, the crack lengths were 8 mm,

13 mm, 20 mm and 27 mm at N/Nf equal to 0.3, 0.5, 0.7 and 0.9, respectively, where crack

grew with an approximately linear trend versus number of cycles. The lives to failure used in

latter comparisons for the cast aluminum steering knuckle were considered to be those of

macro-crack nucleation.
146

5.5 Experimental Results, Analysis, and Comparisons


Table 5-2 presents the component test data and Figure 5-12 shows the applied

moment amplitude versus fatigue life curves for the two components. Although a plateau

was observed for the forged steel steering knuckle as could be seen in Figure 5-12, the cast

aluminum steering knuckle did not exhibit this behavior at the selected lower moment

amplitude levels. The stress amplitude versus fatigue life curves of the two steering knuckles

are superimposed in Figure 5-13. The stresses in this chart are the local von Mises stresses at

the critical locations of the components obtained from nonlinear FEA (as discussed in

Chapter 4). For the cast aluminum steering knuckle S-N lines based on failure defined as

either macro-crack nucleation or fracture are shown. On the average, about 50% of the cast

aluminum steering knuckle life is spent on macro-crack growth. This figure also shows that

the forged steel steering knuckle results in about two orders of magnitude longer life than

the cast aluminum steering knuckle, for the same stress amplitude level. This occurs at both

short as well as long lives. Note that the difference can be even larger at long lives, due to

the run-out data points for the forged steel steering knuckle. It could also be seen from this

figure that the highest load levels provided life in the range of 104 to 5x104 cycles. Moment

levels corresponding to this life range could be considered as representative of overload

conditions for suspension components, such as a steering knuckle, in service.

Different fracture surface characteristics were observed for the forged steel and cast

aluminum steering knuckles. As could be seen in Figure 5-14 for a typical steering knuckle,

the failed forged steel steering knuckles had a typical ductile material fatigue failure surface

including crack initiation, smooth crack growth and rough fracture sections. The failed cast

aluminum in Figure 5-15 could be seen with a relatively longer crack growth portion (as
147

observed in the displacement monitoring curve), as compared the crack growth portion of

the forged steel steering knuckle.


148

Table 5-1 Measured and predicted strain values at 680 N.m static moment. Locations of
the gages are also shown.

Gage Measured P Mc
+
Predicted Diff.
Number Strain A I Strain from FEA Meas. and FEA
(μstrain) (μstrain) (μstrain) (%)

Forged Steel
1 542 575 583 8
2 -527 -557 -546 4
3 1561 1571 1716 10
4 -1489 -1536 -1590 7

Cast Aluminum
1 455 - 434 5
2 534 - 470 12
3 228 - 268 18
4 289 - 320 11
149

Table 5-2 Component test data of forged steel and cast aluminum steering knuckles
investigated.

Nf Nf Test
Mmax Ma
R-Ratio Crack Nucleation Fracture Freq. Remarks
(Nm) (Nm)
(cycle) (cycle) (Hz)

Forged Steel
1515 720 0.05 57,200 2 1

1515 720 0.05 48,800 2 1

1240 585 0.06 105,100 3 1

965 445 0.07 >323,800 3 1,2

965 445 0.07 >1.5E6 3 1,2

965 445 0.07 >560,700 3 1,2

825 380 0.08 >1.4E6 5 1,2

Cast Aluminum
2230 1075 0.04 15,000 29,500 1
2230 1075 0.04 35,000 58,500 1
1595 755 0.05 292,000 409,000 3
1595 755 0.05 190,000 472,300 3
1305 615 0.06 393,000 831,100 3
1195 560 0.07 278,000 869,900 4

1. There was no marked difference between crack nucleation and fracture lives of the
forged steel steering knuckle.
2. These parts broke at a location other than the critical location due to fixture part breaking,
therefore they were considered as run-out tests.
150

BODY

HUB MOUNTING
ATTACHMENTS

STRUT MOUNTING
HOLES

SPINDLE

FORNT AND REAR


LINKS ATTACHMENT

TENSION STRUT
ATTACHMENT

STRUT MOUNTING
HOLES VERTICAL ARM

BRAKE CALIPER
HORIZONTAL MOUNTING BOSS
ARM

HUB MOUNTING
HOLES
STEERING ARM AND
LOWER BALL JOINT
ATTACHMENTS

BODY

Figure 5-1 Nomenclature used for forged steel (top) and cast aluminum steering
knuckles in this study.
151

1) Strut mount 2) Coil spring 3) Spring insulator 4) Strut


5) Hub and bearing assembly 6) Knuckle 7) Rear lateral link 8)Front lateral link
9) Tension strut 10) Bracket 11) Stabilizer bar 12) Bushings
13) Bracket 14) Link

Figure 5-2 Forged steel steering knuckle within the rear suspension system of a 4-
cylinder sedan.
152

1) Front hub 2) Knuckle 3) Lower control arm 4) Bushing


5) Lower ball joints 6) Strut 7) Coil spring 8) Strut dust shield
9) Strut bumper 10) Strut mount 11) Upper bearings 12) Spring seat
13) Stabilizer bar 14) Bushings 15) Link

Figure 5-3 Cast aluminum steering knuckle within the front suspension system of a 6-
cylinder minivan.
153

FIXED TO STRUT

MY

FIXED TO CONTROL ARMS

FIXED TO TENSION STRUT X Y

Figure 5-4 Simulated primary loading and restraints on the forged steel steering knuckle
for component testing.

STRUT MOUNT

MY

FIXED TO HUB

X Y

Figure 5-5 Simulated primary loading and restraints on the cast aluminum steering
knuckle for component testing.
154

10

8
4

5
6
7

Figure 5-6 Exploded view of the forged steel steering knuckle test fixture: 1) steering
knuckle, 2) spindle grip, 3) moment arm, 4) side blocks, 5) end bar, 6) test
bench, 7) load actuator, 8) load cell, 9) rod-end bearing and pin, 10) needle
roller bearings.

Figure 5-7 Schematic drawing (left) and the fixture for forged steel steering knuckle test
arrangement.
155

10
9

1 7

2 6

Figure 5-8 Exploded view of the cast aluminum steering knuckle test fixture: 1) steering
knuckle, 2) body fixing blocks, 3) arm grip pins, 4) moment arm adjusting
pins, 5) moment arm, 6) test bench, 7) load actuator, 8) load cell, 9) rod-end
bearing and pin, 10) needle roller bearings.

Figure 5-9 Cast aluminum steering knuckle test arrangement showing fixturing
schematic (left), and arm fixturing close up.
156

Figure 5-10 Load application through rod-end bearing and a pair of needle roller bearings
to avoid out-of-plane bending and horizontal friction force.

1.5
Forged Steel Knuckle
Displacement Amplitude (mm)

Cast Aluminum Knuckle

crack nucleates

0.5
0 0.2 0.4 0.6 0.8 1

Normalized Number of Cycles, N/Nf

Figure 5-11 Displacement amplitude versus normalized cycles for typical forged steel and
cast aluminum steering knuckles during tests.
157

(a)

(b)

Figure 5-12 Applied moment amplitude versus fatigue life curves for (a) forged steel and
(b) cast aluminum steering knuckles.
158

Figure 5-13 Superimposed stress amplitude versus fatigue life curves for forged steel and
cast aluminum steering knuckles.

FRACTURE CRACK
NUCLEATION

CRACK
GROWTH

Figure 5-14 Typical fractured (left) and fracture surface of the forged steel steering
knuckle.
159

FRACTURE

CRACK
GROWTH

CRACK
NUCLEATION

Figure 5-15 Typical fractured (top) and fracture surface of the cast aluminum steering
knuckle.
Chapter Six

6 Fatigue Life Predictions and Comparisons with


Experimental Results
In general, four fatigue life prediction models are commonly used. These include the

stress-life model, the strain-life model, the fatigue crack growth model, and a combination of

fatigue crack growth and strain-life models. The first two models consider the macro-crack

nucleation phase as failure. The stress-based approach is typically used in terms of nominal

stresses. Therefore, it does not directly account for the plastic strain at the notch root, which

can have significant effect on fatigue behavior of components under cyclic plasticity

conditions. The predictions of this approach for notched members, such as the fillet at the

spindle step of the forged steel steering knuckle and the hub bolt holes of the cast aluminum

and cast iron steering knuckles, are usually conservative. Local values of stress could also be

used in stress-life predictions (here called local stress-life or σ-N method).

Strain-based approaches are very common in life prediction of notched components.

In these cases, the behavior of the material at the root of the notch is best described in terms

of strain. The approaches pursued here are nominal stress approach, local strain approach

using nominal stresses, and local stress and strain approaches using finite element analysis

(FEA) results. The fatigue crack growth model was not pursued in this study, since damage

tolerance design, which assumes preexisting cracks, is not yet as practical for ground vehicle

parts due to costly and, in most cases, infeasible inspection of cracks. The overall

methodology followed for life prediction in this study is shown in Figure 6-1.

160
161

In the following sections the life prediction paths pursued are implemented and

discussed. The details of fatigue life calculations for each method at the highest moment

level for the forged steel and cast aluminum steering knuckles are provided in Appendix B.

6.1 Nominal Stress Approach


One of the approaches used for the forged steel steering knuckle is the nominal

stress approach (path 1 in Figure 6-1). This method could not be used for the cast aluminum

steering knuckle since nominal stress at the vicinity of the notch could not be explicitly

defined for this component. In nominal S-N approach, the nominal values of maximum,

mean and alternating stresses were calculated from analytical equations (i.e. for nominal

elastic behavior, S = P A + Mc I ). Nominal yielding occurred at higher maximum moment

levels of 1515 N.m and 1240 N.m. The maximum and minimum nominal stress distributions

at a section remote from the fillet, i.e. where the stress distribution becomes uniform, were

obtained by applying maximum and minimum moments to the FEA model of Figure 4-2 for

the forged steel steering knuckle and using a nominal stress distribution on a cross section

remote from the spindle 2nd step fillet as in Figure 4-9. Two mean stress correction methods

were used to account for the effect of mean stress. These include the Gerber parabola:
2
Sa ⎛ Sm ⎞
+⎜ ⎟ =1 (6-1)
S Nf ⎜⎝ S u ⎟⎠

and the modified Goodman equation:

Sa Sm
+ =1 (6-2)
S Nf S u

The fatigue life was then obtained from a Basquin-type equation:

S Nf = σ ′f (2 N f ) (6-3)
B
162

To find the exponent B, the S-N line was drawn considering the effect of the notch.

The stress concentration factor, Kt = 2.2, at the fillet was obtained from stepped shaft stress

concentration chart. Peterson’s equation, K f = 1 + K t − 1 , was used to calculate the fatigue


1+ a r

notch factor, Kf = 2.1, where a is a material characteristic length obtained from an empirical
1.8

relationship between Su and a , i.e. a = 0.0254 ⎛⎜⎜ 2070 ⎞⎟⎟ with Su in MPa and a in mm
⎝ Su ⎠

(Stephens et al., 2000). Then the fatigue limit (at 106 cycles), Sf , was reduced to Sf /Kf , and B

was found to be -0.133. No surface finish effect was considered due to highly polished

surface of the spindle at the fillet. Table 6-1 lists nominal stress amplitude and mean stress,

equivalent nominal stress amplitude for R = -1 (SNf), and predicted and experimental fatigue

lives. Figure 6-2 shows stress amplitude versus predicted and experimental lives using the

nominal S-N approach. For the Gerber model, a life difference of less than a factor of 6

compared to the experimental results could be observed. The modified Goodman

predictions were more conservative, with predicted lives shorter by about a factor of 50.

6.2 Local Strain Approach Using Nominal Stresses


The nominal values of stress were also used as input to the local strain approach

(path 2 in Figure 6-1). To use this method, a notch rule such as Neuber’s rule or strain

energy density rule is required to obtain the local stresses and strains. Neuber’s rule relates

nominal stress and strain with local stress and strain behavior in such a way that the

geometric mean value of the stress and strain concentration factors is equal to the Hookian

(theoretical) stress concentration factor (Neuber, 1961). The strain energy density rule is

based on the assumption that the strain energy density at the notch root is nearly the same

for linear elastic notch behavior and elastic-plastic notch behavior, as long as the plastic
163

deformation zone at the notch is surrounded by an elastic stress field (Molski and Glinka,

1981). Because of this requirement and considering the fact that general yielding exists at the

two higher moment levels, the latter method is not best suited for the case of forged steel

steering knuckle.

Neuber’s assumption has shown good predictions for plane stress conditions and is

usually conservative by over-predicting strain in most other cases (Stephens et al., 2000),

even though this is not always the case (Härkegård and Mann, 2003; Zeng and Fatemi,

2001). In order to investigate the applicability of the notch formula used, the state of stress

and strain at the critical sites of this steering knuckle was investigated. As discussed in

Section 4.6, the results of stress analysis for the forged steel steering knuckle at the transition

point between the fillet and the spindle second step (the highest stressed point of the

component) showed that the state of plane strain could be considered for the critical

location and Neuber’s rule is expected to provide conservative results.

To deal with nominal yielding at the two higher moment levels (1515 N.m and 1240

N.m) in local strain approach, two methods were followed. The first method is based on

considering the fact that Neuber’s rule in its original form is an approximation method that

also applies to the general case of large scale yielding (Neuber, 1961; Härkegård and Mann,

2003), the local stresses and strains were obtained from this general form:

σ ε = S e K t2 (6-4)

where the relation between nominal stress S and nominal strain e is based on Ramberg-

Osgood equation, rather than the linear relation S = e E. The state of stress at the root of the

notch is multiaxial (as discussed in Section 4.6) and to account for this, a method proposed

by Hoffmann and Seeger (1985) was implemented and the equivalent stress concentration

factor, Ktq, was used in Equation (6-4) instead of Kt, where:


164

1
K tq = K t [(1 − a e ) 2 + (1 − be ) 2 + (a e − be ) 2 ] (6-5)
2

In this equation ae and be denote the elastic stress ratios, ae = σe2/σe1 and be = σe3/σe1. The

second method applied is based on a generalized application of Neuber’s rule for nonlinear

net cross section behavior (i.e. general yielding) proposed by Seeger and Heuler (1980). In

this method, by defining modified nominal stress and strain incorporating the plastic limit

load, Neuber’s rule is generalized for the case of inelastic net section behavior. They

proposed this generalization for Neuber’s rule as:

1
K tq2 S 2 Ee∗ σ2 ⎛σ ⎞ n′
=εσ = +σ⎜ ⎟ (6-6)
⎝ K′ ⎠
*
E S E

The modified nominal stress S* is calculated from:

P
S* = Sy (6-7)
Pp

P is the maximum load of the test and Pp is the plastic limit load defined for elastic-perfectly

plastic material and characterizes the onset of general yielding of net section area. Here, the

loads were replaced by moments due to specific loading condition for the components. The

modified nominal strain, e* , is obtained from the material stress-strain law.

Once the local amplitudes of stress, σa , and strain, εa , were calculated, fatigue life

was obtained from the Smith-Watson-Topper (SWT) life equation that incorporates the

effect of mean stress:

σ max ε a E = (σ ′f ) 2 (2 N f ) 2b + σ ′f ε ′f E (2 N f ) b+c (6-8)

Table 6-2 lists the stress results from FEA and predicted stresses based on Neuber’s

rule, and Seeger-Heuler’s generalized application of Neuber’s rule. When the estimated

stresses of these two stress estimation approaches are compared, it can be found that for the
165

maximum applied moment levels where large-scale yielding existed, the latter approach

estimates 14% and 5% higher local stress amplitudes, but 30% and 25% lower mean stress

compared to nonlinear FEA results than Neuber’s rule without generalization. In this regard,

the stress and strain predictions by direct application of Neuber’s rule, as compared with the

Seeger-Heuler’s generalized application of Neuber’s rule, were less conservative and closer to

nonlinear FEA results. In this case the results of Neuber’s rule (without generalization) were

considered as the base in life predictions.

It should be emphasized that these two methods compared here (i.e. Neuber’s rule,

and Seeger-Heuler’s generalized application of Neuber’s rule) are notch deformation rules

and are used to estimate stress and strain at the notch. Fatigue life prediction for the two

higher moment levels that follows the stress and strain estimations resulted in 2 and 1.3

times more conservative lives for Neuber’s rule with Seeger-Heuler’s generalized application

of Neuber’s rule.

Table 6-1 lists the local alternating and mean stresses and predicted and experimental

lives, and Figure 6-3 illustrates SWT parameter (σ max εaE ) versus predicted and

experimental lives, where it could be seen that the predictions are conservative by more than

a factor of 7. This could be explained by the suggestion that Neuber’s rule is shown to be

more applicable to plane stress states.

6.3 Local Stress and Strain Approaches Using FEA


Results
The crack nucleation models could also be applied using local stress and strain values

obtained directly from FEA at the failure location (paths 3 and 4 in Figure 6-1). The

complex geometry of the cast aluminum steering knuckle is an example for which no notch
166

factor could be defined and according to Sonsino, Kaufmann and Grubišić (1997), for such

conditions transferability of material test data could be performed only through local

equivalent stresses or strains in the critical failure areas. Here, the local elastic-plastic stresses

and strains were obtained from FEA using two approaches; first, by full-scale elastic-plastic

FEA of the steering knuckles (path 3 in Figure 6-1), and second, a combination of elastic

FEA and a Neuber-type stress correction (path 4 in Figure 6-1). The stress-life and strain-life

methods were then implemented to predict fatigue life. To account for the multiaxial stress

state, equivalent von Mises values were used for both components. Since the maximum

principal stress theory is more commonly used for brittle materials such as cast aluminum,

life predictions using maximum principal stresses and strains were also performed for the

cast aluminum steering knuckle.

Using the nonlinear FEA results (path 3 in Figure 6-1), the local equivalent stresses

and strains corresponding to the experimental loading conditions were obtained by applying

equal moments to the simulated finite element models and performing nonlinear FEA. The

local R-ratios for stress are different from the applied moment R-ratio due to local plasticity,

as indicated in Section 4.6. In stress-life approach, the effect of mean stress was accounted

for using three mean stress models. These include Gerber’s equation (Equation (6-1), but for

local stresses), the stress-life version of the SWT model:

σ maxσ a= σ ′f 2 (2 N f ) 2b (6-9)

and the commonly used modified Goodman equation (Equation (6-2), but for local stresses).

The Basquin equation (Equation (6-3), but for local stress) was then used to obtain the

fatigue life, where B = b . For the stress-life version of SWT model, fatigue life was found

directly from Equation (6-9).


167

Superimposed local stress amplitude versus predicted and experimental lives for

different mean stress correction models based on stress-life approach is presented in Figure

6-4. Comparing the predictions with test results for the forged steel steering knuckle in

Figure 6-4a, the Gerber’s model could be seen to offer close predictions. The modified

Goodman’s prediction is the most conservative, by an order of magnitude in fatigue life. For

the cast aluminum steering knuckle in Figure 6-4b, Gerber’s line is relatively close to the

experimental life line by factors of 5 in shorter lives and a factor of 2 in longer lives, and

offers relatively conservative predictions. The SWT and modified Goodman predictions are

very conservative. In addition, comparing Figure 6-3 and Figure 6-4a, i.e. the nominal and

local approaches for the forged steel steering knuckle, it could be seen that predictions of the

local approach are closer to the experimental lives. This is partly due to the fact that the local

approach directly accounts for residual stresses generated because of local plasticity, whereas

this effect is not considered in the nominal stress approach.

In the strain-life method using the nonlinear FEA results, to take the effect of mean

stress into account in fatigue life predictions two models were considered; the Morrow’s

mean stress model:

σ ′f − σ m
εa = (2 N )
f
b
+ ε ′f (2 N f ) c
(6-10)
E

and the strain-life version of SWT model, Equation (6-8). The predictions for the cast

aluminum steering knuckle were also performed using maximum principal stress, in addition

to von Mises stress. Superimposed local strain amplitude versus predicted and experimental

lives for different mean stress correction models based on strain-life approach are presented

in Figure 6-5. Morrow’s mean stress correction model provides the closest prediction, within

a factor 2 in life for the forged steel, and within a factor of 3 in life for the cast aluminum
168

steering knuckle. The predictions for the cast aluminum steering knuckle based on the

maximum principal stress were about twice more conservative than the von Mises

predictions, and are not shown.

In the second category of local stress and strain approach using FEA stresses and

strains, elastic FEA results were used in conjunction with a Neuber-type stress correction

(path 4 in Figure 6-1) to account for inelastic deformation (Chu, 1997). To estimate the local

elastic-plastic stresses and strains, Neuber’s rule was applied in terms of equivalent quantities

of multiaxial stresses, and the equivalent stresses were computed from von Mises flow

criterion (Socie and Marquis, 2000), which is a reasonable assumption for multiaxial

proportional stresses. If eσeq and eεeq are defined as the elastically calculated notch stress and

strain, and σeq and εeq are the elastic-plastic notch stress and strain, Neuber’s rule becomes:

σeq εeq = eσeq eεeq (6-11)

The second equation to obtain elastic-plastic stress and strain is the cyclic stress-

strain curve of the material:

1
σ eq ⎛ σ eq ⎞ n′
ε eq
= + ⎜⎜ ⎟ (6-12)
E ⎝ K ′ ⎟⎠

To better observe the differences between the nonlinear FEA and linear FEA

combined with Neuber-correction, Table 6-3 compares the stress amplitude, mean stress and

fatigue life of these two approaches. In this table, the fatigue lives using SWT mean stress

correction models for σ-N and ε-N approaches are listed. It could be seen that the lives

obtained using the linear FEA and Neuber-correction are close to the nonlinear FEA

predictions. In particular, for the case of cast aluminum steering knuckle, the nonlinear FEA

results in conservative lives, while the linear FEA with Neuber-correction predicts closer

lives to experimental results. It could be concluded that the linear FEA plus Neuber-
169

correction method is a capable approach to life prediction of components, especially

considering the fact that it does not require the more complicated and time-consuming

nonlinear FEA.

6.4 Fatigue Performance Comparison of Components


Superimposed stress amplitude versus life curves based on stress-life approach, and

SWT parameter versus life based on the strain-life approach for the three steering knuckles

are presented in Figure 6-6 and Figure 6-7, respectively. To obtain stress unit in Figure 6-7,

square root of the left side of Equation (6-8) is plotted as the SWT parameter. Comparison

of the forged steel, cast aluminum and cast iron steering knuckle prediction curves in Figure

6-7 demonstrates that the forged steel steering knuckle offers more than an order of

magnitude longer life than the cast iron steering knuckle, at both short as well as long lives.

As compared with the cast aluminum steering knuckle, the predicted lives for the forged

steel steering knuckle are longer by about three orders of magnitude. The fatigue

performance comparison results of the components of this study are also presented in

Zoroufi and Fatemi (2004).

6.5 Other Considerations in Durability Assessment of


Steering Knuckle
Vehicle steering knuckle is a critical mechanical component in terms of its fatigue

behavior. Forging, casting, and machining induce residual stresses to the component.

Residual stresses can be determined analytically or computationally having detailed

knowledge of the local mechanical response during the induction of residual stress (Savaidis

et al., 2002), or experimentally using, for instance, X-ray diffraction measurements (Stephens

et al., 2000). Residual stresses at the critical locations of the component generated during the
170

manufacturing process could be a significant source of strengthening (if compressive) or

weakening (if tensile), in terms of fatigue life. For the case of the forged steel steering

knuckle under investigation, the main location suitable for inducing local compressive

residual stress is the lower section of the spindle 2nd step fillets. This could be achieved by a

number of ways including shot peening. Due to one-sided nature of loading that the

component undergoes, this is a very effective method to increase fatigue strength.

Automotive chassis components are often used without corrosion protection.

Therefore, fatigue performance under corrosive environments has to be assessed as part of

the design process. Corrosion in conjunction with the effect of time, frequency and variable

amplitude loading may play a prominent role in fatigue design, especially for aluminum alloys

that have lower fatigue limits than steels. For instance, a study on durability assessment of

forged and cast aluminum steering knuckles showed that fatigue resistance drastically drops

under cyclic loading in corrosive environments (Heuler and Birk, 2002).

The current study was based on the simplified constant amplitude experiments and

analysis, while the component in service undergoes the more complex variable amplitude

loading including overloads and underloads. For a steering knuckle, this type of loading can

be generated from panic brake, severe cornering and pot holes. As stated in (Bonnen and

Topper, 1999), experience with long cracks and low-stress-levels have shown that tensile

overloads can increase fatigue life while compressive overloads can decrease it. A tensile

overload can cause a substantial crack tip plastic zone and, as a result of elastic constraint

around the plastic zone, it leaves compressive residual stresses which result in crack growth

retardation. The converse can occur for compressive overloads which decrease compressive

residual stresses and, consequently, fatigue life.


171

As an example of the effects of variable amplitude loading on fatigue behavior of

automotive parts, Bonnen and Topper (1999) conducted variable amplitude bending-torsion

fatigue experiments on normalized SAE 1045 steel axle-shafts to determine the effects of

overloads on fatigue life. Either periodic bending overloads or static bending loads were

applied to these shafts to determine their effect on torsional fatigue. It was determined that

these yield-stress-level bending excursions both decrease the torsional fatigue limit and

shorten the torsional fatigue life at medium and long lifetimes. In the case of multiaxial

loading, the overloads resulted in substantial reductions in fatigue strength at all lives such

that at the fatigue limit this reduction was nearly a factor of three. Therefore, apart from the

overload itself, other parameters such as stress level, sequence effect, cumulative damage and

the type of loading may have substantial influence on fatigue behavior.

From another viewpoint, constant amplitude tests, as compared with realistic

variable amplitude tests, may give insufficient data about the relative fatigue strength of

components. Nevertheless, constant amplitude loading is specified in all the acceptance

fatigue tests for components by taking commercial considerations into account. From a

commercial viewpoint, an acceptance fatigue test needs to be simple, quick, and cheap. The

risk of a fatigue failure in service may be kept low by making a simple test relatively severe.

In many situations this will minimize the overall product cost, despite possible over-design

(Pook, 1997). In situations where minimum weight design is important, such as the aircraft

industry and more recently the automotive industry with the trend toward lighter and more

fuel efficient vehicles, more expensive and time consuming variable amplitude tests are

economically justified.

A number of references in the literature conducted real-time vehicle wheel load and

moment measurements to obtain the applied loads and moments on steering knuckles (for
172

instance see Lee et al., 1995). It was found that loads and moments in all three directions are

applied to a typical steering knuckle installed in a vehicle. Experiments and analyses based on

multiaxial loading, rather than unidirectional, express the real loading condition. But due to

complexity of such kind of testing for this component, unidirectional tests are typically

conducted. For instance Lee et al. (1995), conducted separate unidirectional tests in two

primary loading directions on the steering knuckle. Although unidirectional testing in the

primary loading direction, as performed in this work, does not reflect the precise effect of

real life history, it could be regarded as a simplified and practical means of estimating fatigue

tolerance of the component.

Fretting fatigue is another mode of failure. Fretting, which is a surface wear

phenomenon occurring between two contacting surfaces having oscillating relative motion

of small amplitude, may exist between the steering knuckle and hub assembly, strut joints,

suspension links, and chassis connections in service.

Surface finish effect could be very influential in fatigue evaluation and, normally, a

surface finish reduction factor is applied to the fatigue strength of a component. However,

in this study the fillet of the forged steel steering knuckle was machined and polished and,

therefore, no surface finish factor was applied to life predictions. For the cast aluminum

steering knuckle, due to the nature of the casting materials and the fact that the defects of a

casting material are uniform internally and externally, no surface finish factor was

implemented either.

In forged components, grain flow produces directional characteristics in properties

such as strength, ductility, and resistance to impact and fatigue. The effect of anisotropy and

directionality of the forged steel steering knuckle material fatigue behavior was investigated

in this study, where it was found that the primary stressing direction offers about twice more
173

life than the other two directions. In castings, flow lines, porosity and several other casting

defects can show up, depending on casting practice. Solidification is not always uniform in

castings. Directional solidification is another process that might bring variability of behavior

in different directions.

It should be noted that the influencing parameters on fatigue behavior of a steering

knuckle or similar components are not limited to those mentioned, nor are the approaches

to address such problem only confined to those implemented. One of the main purposes of

this chapter was to provide a feasible methodology to approach the durability assessment of

a fatigue-critical automotive component.


Table 6-1 Nominal and local stresses and predicted and experimental fatigue lives for the forged steel steering knuckle using nominal
stress and local strain approaches.

S-N ε-N
Applied SNf (MPa) Fatigue Life (cycles) Predicted
Moment Fatigue Experiment
Amplitude Sa Sm
Modified
Gerber
Modified
Gerber σmax εa σ maxε a E Life Fatigue Life
(N.m) Goodman Goodman
(MPa) (MPa) (MPa) (%) (MPa) (cycles) (cycles)
720 268 301 422 309 1,000 10,400 701 0.29 634 8,800 57,280
48,879

580 253 286 388 288 1,900 17,800 674 0.26 594 12,900 105,197

450 219 252 317 242 8,700 65,100 620 0.22 524 27,500 323,849*
1,565,630*
560,773*

380 185 219 253 200 47,100 278,600 570 0.18 455 81,800 1,399,370*
* The component broke at a location different from the predicted failure location due to fixture part breaking, therefore considered as run-out test.

174
Table 6-2 Comparison of FEA results and predicted stresses and strains obtained from Neuber’s rule, and Seeger Seeger-Heuler’s
generalized application of Neuber’s rule for the forged steel steering knuckle. The predicted fatigue lives are calculated from
strain-life method.

Neuber's Rule Seeger-Heuler's Generalized Application of Neuber's Rule


Applied
Moment Predicted Fatigue Predicted Fatigue
Amplitude Sa Sm σa σm εa Life σa σm εa Life
(N.m) (MPa) (MPa) (MPa) (MPa) (%) (cycles) (MPa) (MPa) (%) (cycles)
720 268 301 456 245 0.29 8,800 517 172 0.40 4,500

580 253 286 440 235 0.26 12,900 463 177 0.30 10,100

450 219 252 397 223 0.22 27,500 397 223 0.22 27,500

380 185 219 345 225 0.18 81,800 345 225 0.18 81,800

175
Table 6-3 Comparison of alternating and mean stress values, SWT parameter and fatigue life for two stress calculation methods;
nonlinear FEA, and elastic FEA plus Neuber-correction using σ-N and ε-N methods. Fatigue lives for σ-N and ε-N
approaches were both obtained from the SWT model. Nonlinear FEA results were taken as the base for comparison.

S-N ε-N

σa σm Fatigue Life σ maxε a E Fatigue Life


(MPa) (MPa) (cycles) (MPa) (cycles)

Applied
Moment Non- Linear FEA Non- Linear FEA Non- Linear FEA & Non- Linear FEA Linear FEA &
Amplitude linear & Neuber % linear & Neuber % linear Neuber linear & Neuber % Non-linear Neuber
(N.m) FEA Correct. Diff FEA Correct. Diff FEA Correct. Ratio FEA Correct. Diff FEA Correct. Ratio

Forged Steel Steering knuckle


720 434 437 -1 170 177 -5 10,400 8,900 0.9 547 567 -4 20,900 16,700 0.8
580 390 374 4 176 198 -13 29,900 36,100 1.2 454 480 -6 82,600 52,200 0.6
450 319 294 8 195 220 -13 180,900 300,700 1.7 376 394 -5 538,800 320,500 0.6
380 270 250 8 209 225 -8 776,500 1.3E+06 1.7 333 346 -4 2.1E+06 1.4E+06 0.6

Cast Aluminum Steering knuckle


1080 215 203 6 81 99 -23 2,014 2,409 1.2 238 248 -4 5,740 4,070 0.7
760 152 142 6 128 131 -2 11,711 17,255 1.5 194 196 -1 25,850 22,700 0.9
610 123 116 6 128 127 1 45,855 67,431 1.5 165 169 -2 85,090 73,420 0.9
560 112 104 6 125 119 5 87,458 150,403 1.7 153 154 0 144,800 158,300 1.1

176
177

DURABILITY ASSESSMENT
OF STEERING KNUCKLE

EXPERIMENTAL ANALYTICAL

1
2

4 3
LINEAR FEA WITH NONLINEAR
NEUBER’S CORRECTION FEA

4
NOMINAL 2 NEUBER’S 2 LOCAL STRESS 3
STRESS RULE AND STRAIN

LOAD-LIFE S-N σ-N ε-N

Figure 6-1 Durability assessment methodology pursued in this study.


178

Figure 6-2 Superimposed experimental life and stress-life prediction using the nominal
stress approach for the forged steel steering knuckle.

Figure 6-3 Superimposed experimental life and local strain-life predictions using
nominal stresses as input for the forged steel steering knuckle.
179

(a)

(b)

Figure 6-4 Superimposed local stress amplitude versus experimental life and predictions
of stress-life model for (a) forged steel and (b) cast aluminum steering
knuckle using nonlinear FEA results.
180

(a)

(b)

Figure 6-5 Superimposed local strain amplitude versus experimental life and predictions
of strain-life model for (a) forged steel and (b) cast aluminum steering
knuckle using nonlinear FEA results.
181

1000
Forged Steel Knuckle
Cast Iron Knuckle
Cast Aluminum Knuckle

Stress Amplitude (MPa)

100
1E+3 1E+4 1E+5 1E+6 1E+7
Cycles to Failure, Nf

Figure 6-6 Stress amplitude versus life curves based on the stress-life approach for the
forged steel, cast aluminum and cast iron steering knuckles.

Figure 6-7 SWT parameter versus life curves based on the strain-life approach for the
forged steel, cast aluminum and cast iron steering knuckles.
Chapter Seven

7 Optimization Including Material,


Manufacturing and Cost Considerations
The objective of the optimization study was to reduce weight and manufacturing

cost of the forged steel steering knuckle while maintaining or improving its fatigue strength.

It was attempted to have a more general look at the optimization. Therefore reducing the

mass of the component, reducing manufacturing costs, and improving fatigue performance

were focused simultaneously.

Manufacturing costs and fatigue strength of a steering knuckle depend on service

conditions, geometry, material and manufacturing processes. Service conditions are typically

dictated to the designer. Therefore, geometry, material and manufacturing parameters were

attempted in this study as design variables. The modifications were approached in two

stages; first without changing the component’s attachment geometry and focusing on

steering knuckle’s body; and second, with limited change in attachment geometry and

focusing on the spindle as well as the body. The material alternatives search in this study

considered replacing the current material with materials of superior fatigue performance, and

subsequently, reducing dimensions and weight. Manufacturing parameter modifications to

improve fatigue performance or reduce manufacturing costs included precision forging

instead of conventional forging, warm forging instead of hot forging, reducing

manufacturing steps, and surface enhancement.

182
183

In this chapter, first component specifications including material, manufacturing

process details and production cost parameters are discussed. This follows with the

definition of optimization problem that introduces the objective function, constraints and

design variables. Then, the details of optimization analysis and results are discussed.

7.1 Component Specifications


Considering the optimization task globally, makes it necessary to have knowledge

about the component, its service conditions, material of construction, manufacturing

processes and parameters that affect its cost. The nomenclature used in the optimization

study is shown in Figure 5-1, and Figure 5-2 shows the component as installed in the

suspension system of the vehicle. The exact service loading conditions for this specific

component were not available, but a general case of Figure 5-4 was considered as a valid

simulation of the primary loading condition for the optimization study.

The forged steel steering knuckle’s material, AISI/SAE 11V37, is a resulphurized

(free machining) high-strength low-alloy (or microalloyed) steel that contains 0.37% carbon.

High strength low-alloy steels, as defined in ASM Machining Handbook (ASM, 1989), are

classified as a separate steel category than alloy steels, and are similar to as-rolled mild-

carbon steels but with enhanced mechanical properties obtained by the addition of small

amounts of alloys. In some cases, special processing techniques such as controlled rolling

and accelerated cooling methods are also used. Due to these additions, low alloy steels have

mechanical properties that are superior to those of the carbon steels for designated

applications. Superior properties usually mean higher strength, hardness, wear resistance,

toughness, and more desirable combinations of these properties. Microalloyed forging steels

reduce manufacturing costs by means of a simplified thermo-mechanical treatment (i.e., a


184

controlled cooling following hot forging) that achieves the desired properties without

additional heat treatments (i.e. quenching and tempering) required by conventional carbon

and alloy steels. Addition of vanadium enhances strength and toughness of steel. It also

forms carbides that increase wear resistance. The increase in fatigue strength from vanadium

averages between 5 and 15 MPa per 0.01 wt% vanadium, depending on carbon content and

rate of cooling from hot rolling (Nakamura et al., 1993). Vanadium can have negative impact

on machinability (Bayer, 2003).

The steering knuckle’s main manufacturing processes are hot forging and machining.

Figure 7-1 shows the manufacturing process flow chart. The description of each step is as

follows, where the information about the forging and the machining processes were

obtained from the OEM and the part’s machining company, respectively.

1. SAE 11V37 is in the form of hot rolled cold drawn bar. The raw material samples are

sent to the inspection laboratory to be chemically checked.

2. In the cutting department, the bars are cut or sheared to the dimensions of the steering

knuckle.

3. The bars are induction-heated or heated in furnace between 1100°C to 1260°C.

4. The forging sequence begins with bar stock. When the stock has reached the proper

forging temperature, it is delivered to the impression die from the furnace.

5. The fullering and edging operations improve grain structure, reduce the cross-sectional

area of the stock where needed, and gather the metal for other sections ‡.

‡ Fuller is a portion of the die that is used in hammer-forging primarily to reduce the cross section and lengthen

a portion of the forging stock. The fullering impression is often used in conjunction with an edger, which is the
portion of the die impression that distributed metal, during forging, into areas where it is most needed to
facilitate filling the cavities of subsequent impressions to be used in the forging sequence.
185

6. The blocking operation forms the steering knuckle into its first definite shape. This

involves hot working of the metal in several successive blows of the hammer or press,

thus compelling the workpiece to flow into and fill the blocking impression in the dies.

7. Flash is produced and appears as flat, unformed metal around the edge of the part.

8. The exact shape of each steering knuckle is obtained by the impact of several additional

blows in the hammer that force the stock to completely fill every part of the finishing

impression.

9. The flash is removed with trimming in a mechanical press. A set of trim dies consists of

a sharp shearing edge, produced to the exact contour of the forging at the flash line,

and a punch, also contoured to fit the forging.

10. The trimmed steering knuckles are now ready for shot cleaning where the scales and

surface impurities remained form the process are removed.

11. The machining process is categorized into four steps: (1) turning the spindle, (2)

conformity checking, (3) milling the holes and making the threads, and (4) inspection

and packing.

12. In the first machining step (turning the spindle) the steering knuckle is placed and

located on the V-block area of the fixture (Figure 7-2). The turning process is

conducted in two steps by rougher and finisher tools. The turning process takes 80

seconds.

13. In the 1st inspection step, the spindle portion of the part is verified to have the required

quality by visual and eddy current inspections.

14. In the milling step, the unmachined forged steering knuckle is loaded on the fixtures

(Figure 7-3). This step takes 160 seconds. The following operations are performed:

a. The strut joint holes are milled.


186

b. The strut joint chamfers are made.

c. The mounting hole chamfers are made.

d. The tension strut steps are milled.

e. The hub mounting holes are drilled.

f. The spindle threads are cut.

g. The hub mounting hole threads are cut.

h. The lateral link hole is made.

15. In the 2nd inspection step the part is gaged and visually checked to make sure the holes

are tapped, the threads are on the spindle, and the chamfers are in the mounting holes

and flange holes.

16. The finished steering knuckles are cleaned and dispatched.

A realistic optimization process necessitates a good understanding of the production

cost attributes and engineering parameters that influence them in manufacturing of the

component. These are shown in Table 7-1 for the component of this study. In

manufacturing of the steering knuckle, the cost attributes of a unit steering knuckle consist

of the raw material, forging process and machining process costs. The cost of each process

consists of variable and fixed costs.

The engineering parameters related to manufacturing were defined in four main

categories in Table 7-1; the general production parameters, the forging process parameters,

the machining process parameters and the machining lubrication parameters. The parameters

that affect the costs were identified. For instance, a forging parameter like temperature

affects tools and initial investment costs of the forging process, while a machining parameter

like feed rate influences coolant and tooling costs of the machining process. The overhead

costs include energy consumption during the processes. The inspection costs depend on the
187

design and customer requirements, but are not affected by the process parameters. It is

known through the machining company that the price of the forged part (including the raw

material) is 50% of the finished part price and the remaining 50% is the machining process’s

share.

7.2 Optimization Problem Definition


Mathematically, a general constrained optimization problem could be defined by an

objective function, design variables and a set of constraints (Vanderplaats, 1999):

Minimize F(X) objective function

Subject to:

gj(X) ≤ 0 j = 1, m inequality constraints

hk(X) = 0 k = 1, l equality constraints

X il ≤ X i ≤ X iu i = 1, n side constraints

⎧ X1 ⎫
⎪X ⎪
⎪ 2⎪
where X = ⎪ . ⎪ design variables
⎨ ⎬
⎪ . ⎪
⎪ . ⎪
⎪X ⎪
⎩ n⎭

In terms of a structural design like the current case of steering knuckle redesign,

F(X) represents a quality to be minimized, such as the mass of the component, maximum

stress at critical locations or cost. gj(X) is a bounded constraint like allowable stress at certain

geometry locations, and hk(X) is an equality constraint like fixed dimensions and geometry

limitations. The design variables such as size and shape of the component are restricted by

the upper and lower limits. Material and manufacturing parameters are also considered as

side variables in order to obtain an improved design.


188

There are several types of optimization approaches applied to automotive structural

design (Ferreira et al., 2003). Size Optimization defines the design variables in terms of discrete

parameters of the system in study. Generally these parameters do not change the overall

shape of the component and only the size is modified. Parameters such as geometrical

properties such as thickness, diameter, and area are used as design variables in size

optimization.

On the other hand, Shape Optimization and Topological Optimization methods should

change appearance of the geometrical domain in the study. In shape optimization the

boundaries of the domain are mapped by a set of control variables that defines the

coordinates of the domain borders. These coordinates are changed in order to minimize the

objective function. Therefore the final shape of the system is changed to find the optimal

design to satisfy the requirements. Topological optimization generally changes the shape of

the system and is more applicable to the conceptual design stage. The initial geometrical

domain is roughly defined and the optimization algorithms create several voids in the whole

system in order to achieve the best material distribution to obtain the optimum design.

The overall optimization approach in this work was a combination of quantitative

and qualitative methods (i.e. mass reduction, cost reduction, and improving fatigue

performance using alternative materials and considering manufacturing aspects). The

optimization work was performed in two stages. Figure 7-4 shows a schematic of the

alternatives considered; in Stage I with no change in attachment geometry, the geometry

variables were considered to be the thicknesses at different locations of the body considering

manufacturing limitations. Additional manufacturing modifications to reduce cost were also

investigated. In Stage II with limited changes in the attachment geometry, spindle was
189

redesigned and alternative materials and additional manufacturing operations were

investigated to reduce weight and cost.

For each optimization stage of Figure 7-4, a localized shape optimization procedure

was conducted on the steering knuckle considering manufacturing limitations. The material

and manufacturing processes as design variables were more used as means of design

modification rather than optimization. Figure 7-5 shows the algorithm of the optimization

process. The general flow of the solution for geometric optimization started with identifying

pre-process design data such as design variables and constraints. Then, finite element

analysis (FEA) on the original component enabled investigating the distribution of

parameters such as stress, stress gradient, and displacement. Design constraints were

evaluated and the important ones were retained. A parametric study was performed to

observe how a defined change in the model affected structural response. Following that, the

optimization problem was created and solved, where the analysis data were updated after

each one of the iterations. Referring to the primary load calculations in Chapter 5, the

nominal stresses generated from the primary loading on the component are essentially

elastic, so linear elastic analysis was found sufficient for the optimization study. The

iterations continued as necessary till the convergence criteria were met (Vanderplaats and

Miura, 1986; Lee and Lee, 2003). For the material alternatives and manufacturing processes a

trial and error approach combined with qualitative investigation was followed by selecting

replacement materials, proposing alternative manufacturing processes, and adding or

eliminating some of the manufacturing steps. The constituents of the analysis are detailed

here after.
190

7.2.1 Objective Function


The objective function is the function or parameter that is optimized. In this

optimization of the steering knuckle, weight and manufacturing cost of the component were

considered as the objective functions.

7.2.2 Constraints
Stress and geometry constraints were defined based on the selected optimization

stage. In order to maintain fatigue performance of the component, equivalent local von

Mises stress amplitude at the original model’s critical location (spindle 2nd step fillet) under

primary loading conditions (fixed steering knuckle body and moment applied to the spindle)

was obtained at a certain moment amplitude. This moment amplitude could change during

the optimization stage, but the optimization was carried out in such a way that the equivalent

local stress amplitude at any location of the optimized model did not exceed the equivalent

stress amplitude at the critical location of the original model under equal moment amplitude:

(Δσequivalent @ any geometry location) – (Δσequivalent @ spindle second step fillet) ≤ 0

This criterion is reasonable considering the stress history at different locations of the

component (Figure 4-10). According to this figure, the stresses generated in the component

at different locations vary proportionally. It could also be noticed that for the root of the 2nd

step of the spindle (the critical location), stress is higher than at other locations.

Referring to the component’s nomenclature of Figure 5-1, the following constraints

were considered for each optimization stage.

Stage I: The optimized steering knuckle was expected to be interchangeable with the

existing one. Therefore, the following dimensions were kept unchanged:

strut joint center-points, and inner diameters;


191

front and rear lateral links center-points and inner diameters;

tension strut joint center-point and inner diameters;

center-points and size of the bolt holes of the hub mounting plate;

spindle geometry.

Stage II: The first four constraints of Stage I remained the same but the spindle geometry

could be changed.

7.2.3 Design Variables


Design variables specify what could change in the optimization process. Depending

on each specific optimization stage of Figure 7-4, the geometry variables were selected as

follows.

Stage I:

body thickness and the geometry of the portion to be removed;

strut joints outer diameters and thicknesses;

front and rear lateral links outer diameter and thickness;

tension strut joint outer diameter and thickness;

hub mounting plate geometry and thickness.

Stage II: The variables of Stage I plus spindle shape and diameter; spindle was redesigned to

a two step shaft with conic transition to reduce the stress concentration and subsequently

weight.

Considering the material as a design variable, alternative steels that have better

fatigue behavior could reduce component’s weight. In this regard, a number of high strength

steels, high strength low-alloy steels (microalloyed steels), and advanced high strength steels

were considered as possible substitutes. Taking the results of the fatigue performance
192

comparison of Chapter 6 into account, substitution of a lighter metal like aluminum or

cheaper metal like cast iron was not considered as an option since one of the goals of the

optimization of this component was to maintain the overall shape, strength, stiffness and

fatigue performance of the component. A lighter or less strong material like aluminum or

cast iron lends to an inevitably bigger component to acquire the required strength and

stiffness, and a completely altered geometry.

Manufacturing processes as design variables were investigated from two points of

view; first to reduce manufacturing costs, and second to improve fatigue strength of the

component. The manufacturing processes and process steps of Figure 7-1 were reevaluated

to investigate the possibility of replacing the current process with an improved one, and

limiting or eliminating any post-forging process without affecting the fatigue strength, in

order to reduce manufacturing costs. On the other hand, the options considered to improve

fatigue strength of the component were inducing compressive residual stress by surface

rolling at the spindle fillet, carburizing, nitriding, and surface hardening.

7.3 Optimization Analysis, Results and Discussions


The optimization analysis was based on the guideline of Figure 7-4, where the 2nd

optimization stage compliments the 1st stage. Depending on the requirements of the

designer, the proper stage could be selected as detailed here after.

7.3.1 Stage I – No Change in Attachment Geometry


Stage I of the optimization guideline (Figure 7-4) included geometry optimization

without changing the attachment geometry, and cost saving in the manufacturing process.

Since the objective of this stage was to maintain the current geometry with no attachment
193

changes, improving fatigue performance by substituting superior materials or additional

manufacturing processes, which would allow having smaller spindle dimensions, was not

considered for this stage.

7.3.1.1 Optimization Using Geometry Variables


A baseline FEA of the original model showed that some parts of the steering

knuckle experience lower stresses for the primary load case considered. The procedure of

Figure 7-5 was implemented. The shape optimization techniques were applied to lower the

weight of the component. The I-DEAS Optimization software was used to perform

automatic redesign of the component. Redesign allows varying the design parameters

(Section 7.2.3), set limits on the design (stress), and set goals for the design (minimize stress

by varying body thickness). The software then tries to find the optimal structure that satisfies

these criteria.

Following the optimization, redesign histories (design changes at each optimization

step) were examined including design goal, design limit, and design parameter for all

iterations to find whether the design is feasible and whether it is the best feasible design

possible. In a feasible design, the component falls within its stress limit while its economical

manufacturing feasibility is verified. The upper bound of the design goal is the actual value

of the design goal scaled up by the maximum stress limit violation. The lower bound is based

on limit violation and on the predicted change in design goal value obtained from the

optimization algorithm. It is always less than or equal to the actual value.

In this optimization stage the attachment geometry of the steering knuckle was kept

unchanged and the geometry variable was considered to be the thickness at different

locations of the body considering manufacturing limitations. Reducing the thickness of the
194

body close to the tension strut connection hole will likely result in distortion of the

component during the forging process and was avoided. The thickness in the area

surrounded by lateral link joint and the lower strut joint was not reduced due to presence of

stress gradient and manufacturing limitations. Therefore, the area surrounded by strut

mounting holes and hub mounting section was targeted to reduce the thickness.

To find the appropriate perimeter of the section to be removed, the rate of change

of stress at that portion of the component was determined while the component was under

primary loading condition. The distribution of stress at that portion for the original model is

shown in Figure 7-6. The red areas in this figure that are basically the stress concentration

points were found to have the highest stress and stress gradient sensitivity to thickness, thus

a pattern was sought to keep the reduced thickness area away enough from these sensitive

points. This pattern could be seen in Figure 7-7.

The depth of the removed section was selected as the design variable, while the

constraints as mentioned in Section 7.2.2 were applied to the model. The design goal was to

minimize the mass of the component. Figure 7-8 shows the variation of actual component

mass, mass upper bound and mass lower bound with the optimization iterations. It should

be noted that the mass value in this figure reflects the whole assembly including the moment

arms §. Figure 7-9 illustrates the stress limit variation in the optimized area versus number of

iterations. After 15 optimization iterations, the stress limit in the optimized area was still only

40% of the stress at the critical location of the component (spindle 2nd step fillet). In this

case, the possible detrimental effect of the forged surface in the optimization region is

accounted for and the design is feasible, i.e. the stresses generated in the optimized area do

§ This was a software limitation. Moment could not be applied directly to the spindle but should be in the form

of load since in IDEAS a moment can not be applied to a geometrical point and it should be applied to a node.
But the math programming redesign optimization task only accepts geometry based FEA models (that can not
include any boundary condition on nodes or elements).
195

not exceed the limit. Figure 7-10 shows the depth of the removed area versus number of

iterations.

The highest weight saving happens by removing the above mentioned section of the

body from the component. Investigating the forging process applicable to this modification,

it was found that the deeper a hole on a surface, the harder the forging process becomes.

The web can also be punched out resulting in additional weight saving, however forging cost

may increase since punching out would require additional process **. Therefore, a depth of 4

mm was selected following the Forger’s recommendation. The final optimized thickness of

the body in the mass-reduced area is 8 mm, versus the original thickness of 16 mm. Figure

7-11 shows the filleted schematic of the optimized portion and Figure 7-12 provides the

dimensions of the redesigned portion. Figure 7-13 shows the distribution of von Mises stress

in the component before and after optimization for this part of Stage I for an arbitrary

bending moment of 690 N.m. The stress distribution in the material-removed area becomes

more uniform for the optimized component, indicating reduction of stress gradient in the

section. It could also be seen that the level of stress in the optimized section increased after

optimization, while the maximum stress and the critical locations of the component

remained the same, showing the fact that in the optimized model the load is distributed

more evenly in the component.

For other optimization potentials at this stage, FEA showed that the strut joints

outer diameters and thicknesses were subjected to lower stress levels, and therefore could be

reduced. However, this reduction’s contribution to weight loss was seen to be minor;

therefore these dimensions were kept unchanged. The tension strut joint outer diameter and

thickness were kept unchanged due to the likelihood of distortion of that portion of the

** Private communication with Joe Grosso of Citation Corporation.


196

component in manufacturing in addition to the attachment requirements of the tension strut

to the steering knuckle.

The analysis showed that reduction of lateral link joint’s outer diameter and

thickness, and the hub mounting plate thickness was possible. In addition, such changes

would not affect the integrity of the component and are manufacturable. Following the

mathematical redesign optimization of IDEAS, the outer diameter of the joint and the

thickness of the mounting plate were selected as design parameters, separately, and the

maximum stress at the critical location of the component was chosen as the stress limit. The

optimized lateral link joint outer diameter and the thickness of the mounting plate were

found to be 20.8 mm and 11.4 mm, versus the original 24 mm and 13 mm, respectively.

Figure 7-14 shows the modified dimensions of the optimized part. In this redesign, the

component’s weight became 2.13 kg earning 9.4% weight saving compared to the original

model. Figure 7-15 shows the distribution of von Mises stress in the component before and

after optimization for the Stage I optimization (including body, hub mounting and lateral

link joint optimization) for an arbitrary bending moment of 690 N.m. The stress level,

though higher in some areas within the material-removed sections, is still lower than that of

the critical point. This optimization option does not result in additional forging cost ††.

Another factor to be considered following the optimization is the stiffness of the

optimized model compared to the original model. Comparing the displacement of the model

before and after optimization, it was found that a 15% overall displacement increase

(measured at the point of application of the load) occurs after optimization. But an

important factor to be considered along with this comparison is the absolute value of the

displacement. Taking standard load cases into account (e.g. ULSAS standard load case, AISI,

†† Private communication with Joe Grosso of Citation Corporation.


197

2001), and assuming a primary load to be the vertical bump (see Section 5.3 for details), the

maximum amount of displacement of the spindle as a result of 1 kN.m moment in the

primary loading direction (applied in the form of concentrated force to the moment arms)

equals to less than 0.1 mm. Therefore, the stiffness does not affect the functionality of the

optimized component in service.

7.3.1.2 Modifications to Manufacturing Processes


In the framework of this study and for optimization Stage I of Figure 7-4 with no

change in attachment geometry, the approaches considered to reduce manufacturing costs

were precision forging instead of conventional forging, warm forging instead of hot forging,

and reducing machining steps.

Precision Forging versus Conventional Forging


The current practice in the forging of steering knuckle results in a considerable

percentage of the material being wasted as flash, in addition to generating non-precise parts

that require machining to get to the required tolerances at the interfaces with other

suspension parts. The weight of the as-forged steering knuckle is 3.1 kg, while the weight of

the finished steering knuckle is 2.4 kg. Therefore 0.7 kg (about 22%) of the material is

removed during post forging processes, mainly machining. Reducing the amount of

machining will also cut the component’s machining cost significantly.

Despite these advantages of precision forging, it should be noted that a major

advantage of closed-die forging with flash is that the volume of the preform can vary within

a wider range than for flashless forging. This makes it possible to continuously manufacture

products with the same quality. Precision forging requires a higher accuracy and a

significantly improved process monitoring and control. Moreover, the tools needed for
198

precision forging and the process conditions have to be monitored carefully because of the

stricter product tolerances that need to be achieved. Thus a couple of process parameters

such as forces, temperature, and process-times need to be monitored (Reinsch et al., 2003).

Considering these facts, a research program could be incited for the steering knuckle

based on the studies on precision-flashless forging by Reinsch et al. (2003) and Vazquez and

Altan (2000). Although these studies were conducted on forged connecting rods, due to

similarities with steering knuckle from manufacturing point of view, they could be modified

and extended for the steering knuckle of this study with the following objectives:

1. Design a tooling concept that can save material by allowing the formation of only a small

amount of flash;

2. The volume of the initial preform and the volume of the cavity at the end of the process

must be the same. The mass distribution and positioning of the preform must be very

exact;

3. Optimize the tooling design by simulations. Find a geometry that requires the smallest

load to fill the cavity without causing defects;

4. Establish guidelines and procedures to design blockers and preforms in order to

accelerate the development of the production process of the forged part.

In order to fill the cavity and produce components, Vazquez and Altan (2000)

concluded that the formation of a reasonable amount of flash should be allowed. A guideline

to achieve this goal could be planned as:

1. Define a preform for the flashless forging of the steering knuckle by physical modeling

experiments. An option would be plasticine billets and aluminum tooling;

2. Obtain the volume distribution in the steering knuckle;


199

3. Perform FE simulations of the component production to obtain the material flow and

tool-billet contact conditions;

4. Determine the shape parameters to optimize the preform geometry;

5. Perform FE simulation with the optimized geometry;

6. Design the tool for forging of the component with controlled amount of flash by

performing several simulations with different geometrical parameters (flash location and

thickness) to optimize the geometry of the tooling.

It should be noted that, in spite of the advantage of reduced machining, precision

forging requires more complex dies that need to be replaced more often, and higher forging

loads that may increase the production cost of the component.

Warm Forging versus Hot Forging


The current forging process of the steering knuckle of this study is performed hot.

The idea of replacing hot forging with warm forging is based on the study by Aloi et al.

(1997), where an experimental study of warm and hot forged microalloyed-steel-transmission

hubs showed that the optimal combination of strength and toughness is achieved in a warm

forged, fan-cooled condition.

Warm forming is an energy efficient process that allows for a part to be

manufactured to a near net shape, in fewer operations, and with less material waste than in

hot forming. Warm forging, as used in the study of Aloi et al. (1997), includes heating a

blank to a controlled temperature between 760°C (1400°F) and 1040°C (1900°F), low

temperatures in the austenite single-phase region, and forging into shape. The technology

allows for near net shape forging of parts with intricate detail and much closer tolerances, as

compared to hot forging. By nature, warm forging also results in a significant refinement of

the austenite grain size.


200

Warm forging processes result in a more efficient utilization of material by reducing

the amounts of both flash and scale which are formed. In one type of warm forging process,

closed-die flashless forging, the tooling cavity completely surrounds the material, thereby

eliminating all flash. The blank weight must be precisely controlled so that, in combination

with warm forming, no flash trim is formed on the exterior of the forging envelope. Since no

flash is generated, a trim press is not required in processing.

Associated reduced scale formation, inherent after forging at lower temperatures, not

only reduces material losses, it also contributes to the near net shape of the final part.

Additionally, lightly scaled parts would not require scale removal in the forge press or final

shot cleaning operation. The lower forming temperature requires lower electrical

consumption not only reducing the operational cost but also the capital cost of the induction

heating power supply. The smaller power supply requires less floor space and a more

compact forging line. Die life can actually be improved in lower temperature forging

operations due to decreased thermal loading and reduced amounts of abrasive forging scale.

In warm forging of microalloyed steels, improved toughness can be accomplished through

refinement of the austenite grain size.

Warm forging does however, have limitations and requires specialized equipment

due to the fact that the resultant forming pressures in this process are extremely high. For

example, the forging pressure increases when decreasing the forging temperature. Since the

press load capacities are higher than that required for hot forging a similar size part, the

tooling must be able to withstand the higher stress levels imparted during forging. This

places an overall limitation on the size of forging that can be formed at lower temperatures.

According to the study by Aloi et al. (1997), in forging of 1037 and 1040

microalloyed steels the room temperature toughness values were 32 to 42 Joules (24 to 31 ft-
201

lbs) higher for the two microalloyed steels after forging at 1010°C (1850°F) versus 1230°C

(2250°F), while 0.2% offset yield strength was reduced by only 4 to 23 MPa (1 to 3 ksi). The

hubs passed a fatigue test of the induction hardened splines, a push-out weld test, and a

wide-open throttle durability test of the hub parts. Successful implementation of both hubs

into production was also achieved, and included acceptable weldability, induction hardening

response, and good productivity through comparable machinability.

The results obtained in the study of Aloi et al. (1997) demonstrate the advantages

with regard to mechanical properties, processability, and overall cost of warm forged

microalloyed steels for any number of applications. These benefits can be fully recognized in

applications requiring a good combination of strength and toughness, including the steering

knuckle of this study. On the other hand, since the material is forged at higher yield stress in

warm forging, the process requires higher loads that reduce die life. Therefore, while

substituting warm forging is considered as one of the options in manufacturing process

modification of the steering knuckle, it is necessary to evaluate the cost impact of the option

in a more detailed study.

Eliminating or Limiting Some of the Manufacturing Steps


Figure 7-16 shows the forged steering knuckle prior to shipping to the machining

plant. Here as a modification, the strut and hub mounting holes are proposed to be

extrusion-pierced during the forging process. In the modified process, starting with a slug or

billet from a steel bar, the steering knuckle is produced in the first stage by pressing or

hammering. Then at the second stage the hub and strut mounting holes are extrusion-

pierced. The final stage for this part finish pierces the holes. Therefore, the forging process

is modified and the machining work of the mounting holes could be reduced.
202

The changes in cost due to this process modification could be evaluated both

qualitatively and quantitatively. Referring to the cost attributes (Table 7-1), by implementing

these changes, the forging process costs attributed to tools and dies, and initial investment

and setup increase. According to Jung (2002), the machining cost is mainly based on

machining time that is composed of set-up time, operation time, and non-operation time.

The current modification has minor influence on the tool and machine set-up time. The

operation time is the duration laps from feed engagement to feed disengagement and is

composed of the rough cutting time, the finish cutting time, and tool approach time. The

main contribution of the current modification will be on the rough cutting time. The non-

operation time, which is the lapse time after a workpiece is mounted except the operation

time, has also minor effect within the current modification.

The rough cutting time is proportional to the machined volume while the finish

cutting time is proportional to the area being machined. For the majority of machining

operations, the main portion of machining time is consumed by rough machining (e.g. for

the example part investigated by Jung (2002) the finish machining was estimated to be 15%

of the rough machining time). Based on this concept, the reduction in machining time and

subsequently machining cost could be related to the reduction in removed volume. In this

modification of the steering knuckle, 8.9 cm3 less volume will be removed during the

machining process (about 10% of the total material currently being removed) saving about

10% of machining and 5% of total component production time and, subsequently, cost. It

should be noted that the changes in forging process need to be implemented only once,

imposing a one-time investment, while the saving in machining costs is attributed to every

component being manufactured. Therefore the modification proposed is justified.


203

7.3.2 Stage II – Limited Change in Attachment Geometry


Stage II of the optimization guideline (Figure 7-4) included geometry optimization

with limited change in attachment geometry (i.e. modifying spindle geometry), substituting

alternative materials with superior performance, and improving fatigue performance by

additional manufacturing operations. This stage intended to obtain a smaller and, therefore,

lighter spindle that reduces material and manufacturing costs.

7.3.2.1 Optimization Using Geometry Variables


For this optimization stage, the spindle of the optimized component from Stage I

was redesigned to obtain higher fatigue performance and reduced weight. A schematic of

this redesign compared to the original design is shown in Figure 7-17. In this model, the

diameters of the first and third steps of the spindle were maintained and the second step, on

which the bearing is installed in the current design, was reshaped to a tapered design. Instead

of the current cylindrical-bore roller bearing, a spherical roller bearing with tapered bore

(Figure 7-18) was selected. Due to removal of the second step fillet, which was the failure

location in component tests, the stress concentration at this location decreased. In the

original design, the maximum stress generated by the primary loading condition on the

component occurred at the 2nd step fillet. The diameter of the spindle’s first step in this

redesigned model was optimized in a way that the maximum stress generated by the primary

loading in the component did not exceed the maximum stress generated in the original

model with the same magnitude of load. In addition, the influence of spindle 1st step fillet

radius was investigated.

Figure 7-19 shows the stress distribution caused by the primary loading and a typical

moment magnitude of 690 N.m. In the redesigns with the fillet radii of 1.3 mm (the original
204

1st step radius) and 2.6 mm, the component’s weight became 2.07 kg and 2.03 kg, earning

2.8% and 4.7% weight savings compared to Stage I optimization, and 11.9% and 13.6%

weight savings compared to the original model, respectively. Due to the reduction of the

effect of stress concentration with increasing the fillet radius, the weight of the spindle is

reduced. But this increase of the radius depends on the design requirements of the mounting

hub and should be made compatible with the suspension system design. The dimensions of

the spindle portion after the redesign with 1.3 mm spindle 1st step fillet radius (the original

size) are shown in Figure 7-20. The forging cost difference between a tapered and a straight

stem is minimal. There would be added cost in machining a tapered stem and in induction

hardening the tapered stem after machining. Typically a tapered stem would be rough

machined, hardened, then finish machined. Even though this could result in significant cost

impact to the knuckle, it can still yield a net wheel end system cost save ‡‡. All these factors

should be taken into account at the detailed design stage of this plan. Figure 7-21 shows the

model of Stage II optimization.

7.3.2.2 Modifications Using Alternative Materials


Alternative materials with superior fatigue properties were considered as another

group of design variables within the Stage II optimization (Figure 7-4) and general

optimization procedure of Figure 7-5. In this respect, the monotonic and cyclic properties of

a number of steels were obtained (AISI bar steel fatigue database, undated; Johnson et al.,

2000). In addition, material selection guidelines recommended by Bayer (2003) for

automotive applications were considered. Since the objective of this material search was to

find a material with superior fatigue properties, fatigue strengths of alternative materials were

‡‡ Private communication with Joe Grosso of Citation Corporation.


205

compared to that of the current SAE 11V37 used to manufacture the steering knuckle.

Those materials superior in fatigue strength (Table 7-2 lists their mechanical properties) were

investigated individually.

The effect of alloying elements on mechanical properties, service conditions and

manufacturability (considering Table 7-3), strength, ductility, static strength at overloads, and

excess deformation were evaluated and the proposed alternatives are listed in Table 7-4. It

should be noted that microalloyed steels had higher priority since no heat treatment is

required to achieve the desired properties such as hardness and strength. The non-

microalloyed steels bring additional cost for heat treatment and unwanted distortion during

this process may occur, resulting in the need for additional machining and are, therefore, not

recommended. Materials with lower machinability are also not desirable due to cost impact

on machining. For the particular case of 1117 steel, a microalloyed version of this material

(i.e. 11V17), if produced and tested for mechanical properties, can be a potential alternative.

A key manufacturing parameter of interest is machinability of the selected materials.

The machinability of carbon or alloy steels is influenced by their carbon or alloy content,

microstructure, amount of soft or free machining particles, oxide inclusions, work-hardening

rate and hardness. The AISI Bar Machinability Sub-committee is in the process of generating

a database of automotive bar steel machinability under single point carbide turning

conditions, including more than 30 steel grades (Joseph and Stout, 2004). Since this database

and other machinability rating sources (e.g. ASM Handbook, Vol. 16) do not include most of

the potential materials being studied in this work, hardness of the materials was considered

as the criterion to evaluate and compare machinability of the alternative materials. It is

shown that hardness has adverse effect on cutting speed and feed rate in machining (Figure

7-22); hence it could be considered as a reasonable criterion for machinability comparison.


206

The following microstructural and manufacturability considerations should also be

noted. All of the candidate steels rely on two fundamental mechanisms to develop a proper

microstructure and the desired properties. The first is to achieve a uniform fine grained

austenite microstructure during forging, which transforms to a uniform fine grained room

temperature microstructure during cooling after forging. The second is to precipitate a fine

dispersion of vanadium particles (vanadium nitride, vanadium carbide, vanadium

carbonitride) during the transformation of austenite to the room temperature microstructure.

The forging and cooling practices required for each are functions of the composition. While

all of the grades are forgeable in that parts can be formed, the required reheating

temperatures, forging temperatures and cooling practices will be composition dependent.

The MoVTi group of steels offers some advantages in that molybdenum suppresses

unwanted high temperature precipitation of vanadium particles, and lowers austenite

transformation temperatures. The net effect is a finer dispersion of vanadium particles in a

finer room temperature microstructure resulting in better properties.

The properties of two of the steels, 1522MoVTi and 1522MoVTiS, are as hot-rolled

properties. Caution should be exercised in comparing these properties to the other grades

where the properties are given after a forging and cooling practice. The hot deformation and

cooling conditions achieved during hot rolling in a steel mill may not be the same as those

used during forging. As a result the properties could change. Also, the properties of the

15R30V grade are given after hot rolling followed by a high temperature heat treatment. If

this heat treatment is required for a forged part, the cost of the part will increase significantly

because of the additional heat treating step. §§

§§ Private communications with Dr. Tom Oakwood, formerly of Inland Steel Co.
207

Based on the results of the investigation presented in Table 7-4 and the

aforementioned discussion, the candidate materials, i.e. SAE 15V24, 15R30V, 1522 MoVTi,

1522 MoVTiS, 1534 MoVTi and 1534 MoVTiSi, are proposed as alternatives. Limited weight

saving can be achieved by replacing the potential alternative materials (0.4% to 2.1%), mainly

due to geometrical constraints. If comprehensive changes to the geometry are allowed or for

other components with fewer constraints, the weight saving will be more significant. Among

these alternatives, 15V24 is the lowest cost material in terms of raw material cost,

comparable in cost to the current 11V37. The group of MoVTi steels has significantly higher

raw material cost. In particular, molybdenum is the high cost alloy element. More specific

selection requires availability of cost data for the specific metal and performing

manufacturability tests of the metals, both of which were out of the scope of this study.

7.3.2.3 Modifications to Manufacturing Processes


Within the framework of Stage II optimization of Figure 7-4 and optimization

procedure of Figure 7-5, fatigue strength of the spindle could be improved, allowing reduced

spindle size, and therefore weight and cost. Additional manufacturing operations such as

surface hardening and shot peening to induce compressive residual stress are considered to

achieve this purpose as discussed here.

Surface carburizing, nitriding and hardening could be applied to the fatigue-critical

area (i.e. spindle fillet) of the subject steering knuckle for improvement of fatigue

performance, and also improved wear and fretting resistance, since the steering knuckle is in

contact with other chassis components. Carburizing process could be applied to steel to

increase the carbon content of the surface layer. It leads to a higher fatigue resistance and a

much improved wear resistance. Nitriding of low-alloy steels can significantly increase the
208

hardness of the surface layer by some precipitation phenomena which also improve the

fatigue resistance. Nitriding also gives a volume increase and as a result residual compressive

stresses in the surface layer.

Hardening of a surface layer is also done by induction hardening and thus without

introducing extra carbon or nitrogen. Results in Figure 7-23 show improvements obtained

for the spindle of a truck, which in service caused some fatigue problems at the smaller

radius (3 mm). After induction hardening the S-N curve was raised considerably, and fatigue

problems did not occur anymore. The etched cross section of the axle in Figure 7-23 reveals

the depth of the induction hardened layer, which in this case is on the order of a few

millimeters.

Compressive residual stresses improve fatigue resistance by acting as negative mean

stress. For the component of this study, no shot peening is performed, but there is a

possibility of compressive residual stress generation due to shot cleaning that is performed

after forging. In addition, residual stresses may exist due to the forging process. Since

measuring the residual stress generated due to processing on the component was not within

the scope of this study, no residual stress effect due to processing is considered.

To achieve higher fatigue performance, the major location that deserves inducing

local compressive residual stress is the spindle fillets, as shown in Figure 7-24. Residual

stresses can be induced by mechanical methods, thermal methods, and machining. Among

the most widely used mechanical processes for producing beneficial compressive surface

residual stresses for enhancing long and intermediate fatigue lives are surface rolling, shot

peening and shot blasting. Shot peening and shot blasting are not applicable to the case of

forged steel steering knuckle since the spindle surface is machined after forging. On the

other hand, surface rolling could be applied to the spindle steps. In Figure 7-25 a hardened
209

rolling wheel is pressed in the notch of a rotating-bending specimen. As a result, the plastic

deformation occurs in the surface layer of the notch root. As shown by the graph, the fatigue

limit under rotating-bending was significantly increased, up to more than twice the original

value.

The modifications proposed in this section should be looked into at the detailed

design stage, considering the issues of manufacturability and cost. The weight saving due to

each modification should be compared with the manufacturing limitations and cost impact

of the technique. In addition, the dimensions of the wheel end system should be taken into

consideration, where design constraints may limit dimensional changes.

7.3.3 Discussions
To summarize the results from the optimization analysis, Table 7-5 provides a

comparative list of component’s original weight and the weight reductions of each

optimization stage. Overall weight and cost reductions of at least 12% and 5%, respectively,

are estimated for the manufacturing process. The cost of the saved material is additional

reduction, though not very considerable due to small portion of material cost within the total

production cost.

The optimization results showed somewhat limited changes for this particular

component. In this regard, a few points should be clarified. First, this component is

relatively small, compared to steering knuckles with similar or relatively similar service

conditions. Second, it has many attachment compatibility constraints. A more

comprehensive change on constraints can result in drastic and unwanted design alterations in

suspension system. For instance, an optimization plan could be proposed as an interference-

fit assembly to make the component in two parts namely body and spindle, in which the
210

spindle fits in the body. This plan requires a number of major changes in suspension system

design including the geometry of strut, the location of lateral link joint, the position of hub

mounting holes, and in general the overall geometry of the component. Furthermore, some

manufacturing limitations make the optimization options limited. For instance, reducing the

thickness of the body close to the tension strut connection hole results in distortion of the

component during the forging process. Also, the forging process becomes more complicated

as the depth of the removed part from a surface increases. The web can ultimately be

punched out resulting in additional weight saving, however forging cost may increase since

punching out would require additional process. In addition, material can not be removed

from the area surrounded by lateral link joint and the lower strut joint, again due to

limitations in the manufacturing process. All these confine the optimization options to what

was discussed for Stage I (Section 7.3.1). Modifications to spindle, as considered in Stage II

(Section 7.3.2), are also restricted because of numerous changes that will be imposed on the

wheel end system. The techniques proposed in this stage to improve fatigue strength are also

pending a detailed design study to investigate weight saving versus cost impact of the

particular technique and the limitations of the wheel end system design.

It should be emphasized that, following the general goal of the present study as

mentioned in Chapter 1, this steering knuckle was used as an example part in the

optimization study to investigate a methodology for optimizing a forged automotive

component with durability constraints for weight and cost reductions. Therefore, the

emphasis in this study was more on the optimization process, rather than optimization of

the particular component used. In spite of the limited optimization achieved for this

particular component, the approach that was followed is applicable to other forged
211

components. Components with fewer geometrical restrictions have much higher potential

for weight reduction and cost savings.


Table 7-1 Cost attributes for manufacturing the steering knuckle.

General Production Machining Process Lubrication


Forging Process Parameters

Engineering Parameters
Parameters Parameters Parameters

Tool Material
Loading Rate
Program Life

Contact Time

Feed Rate
Workpiece Temperature

Area Required
Material Choice

Speed

Type of Coolant
Forging Sequence
Lubrication
Forging Temperature

Machine Selection

Coolant Concentration

Disposal Method
Annual Prod. Volume

Physical Volume
Forging Method
Die and tool Design

Number of Stations
Cost Attributes

Raw Material
- - - - - - - - - - - - - - - - - - -
Costs
Variable
Labor - - - - - - - - - - - - - - - - -
Tools and Dies - - - - - - - - - - - - -
Forging Initial Investment
Process - - - - - - - - - - -
Equipment
Fixed Setup - - - - - - - - - - - - - - - - - -
Overhead - - - - - - - - - - - - - - -
Inspection* - - - - - - - - - - - - - - - - - - - - -
Labor - - - - - - - - - - - - - -
Variable Coolant - - - - - - - - - - - - - - -
Tools - - - - - - - - - - - - - - -
Machining Initial Investment
- - - - - - - - - - - - - -
Process Equipment
Fixed Setup - - - - - - - - - - - - - - - - -
Overhead - - - - - - - - - - - - - -
Inspection* - - - - - - - - - - - - - - - - - - - - -

: Engineering parameters affecting cost attributes.

212
* The inspection costs depend on customer requirements and component design criteria, not the production parameters.
213

Table 7-2 Preliminary list of alternative materials with superior fatigue strength, as
compared to the currently used alloy steel (SAE 11V37).

SAE Process Sf @106 Sy Su %R/A %EL BHN Ref.


Grade Cycles (MPa) (MPa)
(MPa)
Specimen from forged
11V37 352 556 821 37 21 240 This study
steering knuckle
AISI
10B21 Quenched and Tempered 537 1062 1105 70.5 33.4 322
Database
Quenched and Tempered
AISI
1117 Simulated Carburized 386 655 777 55.0 37.0 193
Database
Core
Quenched and Tempered
AISI
15B35 Simulated Carburized 468 866 940 64.5 37.3 286
Database
Core
Hot Formed, Cont.
AISI
15V24 Cooled Specimens From 379 646 878 61.1 36.0 243
Database
Control Arm
Hot rolled at 1260°C,
Johnson et
15R30V Austenitized at 1150°C 404 593 861 - 17.7 255
al. (2000)
for 1 hr, Air cool
Johnson et
1522MoVTi As hot rolled 428 600 852 - 21.5 267
al. (2000)
Johnson et
1522MoVTiS As hot rolled 438 598 950 - 23 248
al. (2000)
Forced air cooled after Johnson et
1534 MoVTi 443 624 904 - 17.6 272
forging al. (2000)
Forced air cooled after Johnson et
1534MoVTiSi 483 558 1021 - 24.6 272
forging al. (2000)
AISI
4130 Al Quenched and Tempered 528 1284 1483 44.3 28.0 442
Database
AISI
4140 Quenched and Tempered 437 929 1043 61.3 30.8 325
Database
Quenched and Tempered
AISI
4320 Simulated Carburized 547 920 994 63.0 43.0 188
Database
Core
Thru Carburized AISI
4620 764 1317 2227 4.0 12.5
Simulated Case Database
AISI
5120 Quenched and Tempered 472 780 1008 58.0 46.0 252
Database
AISI
5140 Quenched and Tempered 497 957 1039 52.7 28.3 305
Database
AISI
51B60 Quenched and Tempered 771 1830 1970 21.6 36.2 450
Database
Thru Carburized AISI
8620 458 920 1202 1.0 1.0 583
Simulated Case Database
8695 LAB Lab. Melted Simulated AISI
401 780 1496 1.3 1.3 539
HEAT Carburized Case w/o IGO Database
Quenched and Tempered
AISI
8822 Simulated Carburized 442 918 1131 56.0 40.0 252
Database
Core
AISI
9254 Al Quenched and Tempered 845 2270 2950 4.0 3.9 584
Database
AISI
9254V (MA) Quenched and Tempered 710 1870 2050 35.1 16.2 536
Database
Quenched and Tempered
AISI
9310 Simulated Carburized 479 804 902 70.8 31.0 240
Database
Core
214

Table 7-3 Effect of common alloying elements in automotive steels (Bayer, 2003).

Element Effects

Carbon The principal strengthening element in steel. It can have a great effect on
numerous metal1urgical properties. Increased carbon levels can provide
increased hardness and strength. Lower carbon (less than 0.25%) improves
weldability, ductility, and toughness at the expense of strength.
Manganese Increases strength and toughness. Manganese has one of the strongest
elemental effects on steel’s hardenability (the ability of steel to harden at a
depth from the surface through quenching). Higher levels have a negative
effect on weldability.
Sulfur Considered an impurity, except when intentionally added to improve
machinability. It combines with manganese to produce manganese sulfide
(MnS) inclusions, which assist as “chip breakers” in machining steels. Higher
sulfur levels have a detrimental effect on impact resistance.
Silicon A deoxidizer, it is added to steel to tie up free oxygen. The term “killed steel” is
used when it is deoxidized, thus providing improved internal soundness and
surface quality. Higher levels slightly increase hardenability; however, silicon
can have a negative impact on machinability.
Nickel When combined with other alloying elements, it produces steels with excellent
strength and low-temperature toughness in the quenched and tempered
condition.
Chromium Provides wear resistance, hardenability, and low temperature toughness. At
high levels, it provides corrosion and oxidation resistance, and assists in
maintaining strength levels at elevated temperatures.
Molybdenum Has a strong effect on hardenability (similar to manganese). Molybdenum also
increases strength at elevated temperatures.
Aluminum Acts as a deoxidizer and helps control grain size. It can have a negative impact
on machinability.
Niobium Helps produce fine grain steel, and improves the strength of microalloyed
steels.
Vanadium Also helps produce fine grain steel. Additionally, it can be used to increase
strength, impact toughness, and hardenability. Vanadium can have a negative
effect on machinability.
Titanium Primarily a deoxidizer and nitrogen scavenger in the making of boron steels.
Also acts as a grain refiner. Titanium can have a detrimental effect on
machinability.
Boron Increases hardenability in steel with lea than 0.8% carbon, replacing other
alloying elements.
Table 7-4 Potential alternative materials to replace the current 11V37 steel. All the steels are in the form of bars.

Material Process Microstructure Application Advantage(s) Disadvantage(s)


15V24 hot forged and ferrite-pearlite control arms It has superior strength, ductility and fatigue properties
control-cooled Its machinability is comparable to the current material
vanadium- It is a microalloyed steel and the produced part will
strengthened carbon not need heat treatment
steel It is at the same cost level as 11V37
15R30V hot forged and coarse ferrite-pearlite automotive/ It has superior strength and fatigue properties Its ductility is slightly lower
control-cooled aerospace Its hardness is close to the current material
vanadium- It is a microalloyed steel so the produced part will not
strengthened steel with need heat treatment
hardening
characteristics.
1522 hot forged and acicular ferrite and automotive/ It has superior strength and fatigue properties Its hardness is slightly higher (less
MoVTi control-cooled bainitic along with aerospace It is a microalloyed steel so the produced part will not machinable)
titanium grain-refined conventional ferrite need heat treatment Titanium generates forgeability issues
nontraditional steel and pearlite Titanium particles aid in refining austenite grain size It has higher raw material cost
with softening during forging
characteristic
1522 hot forged and acicular ferrite and automotive/ It has superior strength, ductility and fatigue properties Titanium generates forgeability issues
MoVTiS control-cooled high bainitic along with aerospace Its hardness is almost equal to the current material It has higher raw material cost
sulfur (0.05 weight conventional ferrite (equivalent machinability)
percent) steel with and pearlite It is a microalloyed steel so the produced part will not
softening characteristic need heat treatment
Titanium particles aid in refining austenite grain size
during forging
1534 hot forged and acicular ferrite and automotive/ It has superior strength and fatigue properties Its hardness is slightly higher (less
MoVTi control-cooled low bainitic along with aerospace It is a microalloyed steel so the produced part will not machinable) and ductility slightly lower
silicon/0.35 weight conventional ferrite need heat treatment Titanium generates forgeability issues
percent carbon steel and pearlite Titanium particles aid in refining austenite grain size It has higher raw material cost
with softening during forging
characteristic
1534 hot forged and acicular ferrite and automotive/ It has superior strength, ductility and fatigue properties It has softening characteristics
MoVTiSi control-cooled high bainitic along with aerospace It is a microalloyed steel so the produced part will not Its hardness is slightly higher (less
silicon/0.35 weight conventional ferrite need heat treatment machinable)
percent carbon steel and pearlite Titanium particles aid in refining austenite grain size Titanium generates forgeability issues
during forging It has higher raw material cost

215
Table 7-5 Summary of the results of optimization stages.

Geometry Change Process Change Component Weight Cost Reduction Remarks


Weight Reduction (%)
(kg) (%)
Original - 2.35 -

Stage I Material removed from Strut joint holes pierced 2.13 9.4 5% + material 1

body Hub mounting bolt saving


Hub mounting thickness holes pierced
optimized
Lateral link joint thickness
optimized

Stage II Spindle redesigned 2.07 or less *** 11.9 and up 5% + material 2

saving

1. Precision forging and warm forging are other proposed process changes for Stage I. These options should be more extensively evaluated in a
detailed design stage for their cost impact.
2. Surface enhancement (carburizing, nitriding, hardening) and inducing compressive residual stress at surface by surface rolling are proposed
additional processes that will reduce component weight by increasing fatigue strength in Stage II. These options should be more extensively
evaluated in a detailed design stage for their cost impact.

216
*** Depending on spindle fillet radius.
217

RAW MATERIAL INPUT TURN SPINDLE

DIMENSIONAL CHECK

RAW MATERIAL INSPECTION


MILL STRUT JOINT HOLES

MAKE STRUT JOINT CHAMFER


CUT TO SIZE OF PART
DRILL HUB MOUNTING HOLES

MAKE HUB MOUNTING HOLE CHAMFER


INDUCTION OR FURNACE HEATING

MILL TENSION STRUT STEPS

IMPRESSION-DIE FORGING CUT SPINDLE THREADS

CUT HUB MOUNTING HOLE THREADS

TRIMMING
MAKE LATERAL LINK HOLE

SPINDLE DIAMETER AND


SHOT CLEANING THREADS INSPECION

SAMPLE EDDY CURRENT INSPECTION


INSPECTION &
DIMENSIONAL CHECK
CLEANING, FINAL INSPECTION, SHIPPING

FORGING MACHINING

Figure 7-1 Manufacturing process flow chart for the steering knuckle.
218

Figure 7-2 Locating the forged steering knuckle on the V-block to turn spindle.

Figure 7-3 Unmachined forging loaded on the fixtures to be milled. The steering knuckle shown
in this picture is slightly different from the one being studied here.
219

FORGRED STEEL STEERING KNUCKLE


OPTIMIZATION ALTERNATIVES

NO CHANGE
IN
ATTACHMENT
GEOMETRY

FOCUS: STAGE I
Minimize mass while
maintaining the overall
shape and attachment
dimensions
Modify manufacturing
process

LIMITED
CHANGE IN
ATTACHMENT
GEOMETRY

FOCUS: STAGE II
Minimize spindle’s
mass with limited
attachment changes
Modify manufacturing
process

Figure 7-4 Steering knuckle optimization stages followed in this study.


220

OBJECTIVE DESIGN CONSTRAINTS


FUNCTION VARIABLES

MATERIAL GEOMETRY MANUFACTURING


REDESIGNED PROCESSES
MODEL

INITIAL DESIGN MODEL


(GEOMETRY, MATERIAL, PROCESS)
ORIGINAL
MODEL

REDESIGN MATERIAL FEA SHAPE PROCESS


ALTERNATIVES OPTIMIZATION MODIFICATION

NO
OPTIMIZED?

YES

STOP

FINAL DESIGN
MODEL

Figure 7-5 Steering knuckle optimization procedure.


221

Figure 7-6 Stress distribution under primary loading at the vicinity of top strut attachment
(top left), bottom strut attachment (top right), hub mounting part (bottom left)
and the upper curve of steering knuckle body.

Figure 7-7 Pattern of the section to reduce the thickness in Stage I optimization.
222

Figure 7-8 Actual design goal (solid line), the design goals upper bound (dashed line) and
the design goals lower bound (dotted line) versus number of iterations. The
mass includes the moment arms and the unit is kg.

Figure 7-9 Stress limit in the optimized area versus number of iterations. The stress unit
is MPa.
223

Figure 7-10 Design variable (depth of the removed area) versus number of iterations. The
design parameter unit is mm.

Figure 7-11 Optimized section in Stage I optimization.


224

Figure 7-12 Dimensions of the redesigned portion of the steering knuckle’s body in Stage I
optimization. All dimensions are in mm.

Figure 7-13 von Mises stress distribution for an arbitrary moment of 690 N.m before (top)
and after optimizing for the removed section from the body (first part of
Section I optimization). The stress unit is MPa.
225

Figure 7-14 The original (left) and optimized dimensions of hub mounting and lateral link
in Stage I optimization. The dimensions are in mm.

Figure 7-15 von Mises stress distribution for an arbitrary moment of 690 N.m before (top)
and after optimizing for Stage I for the optimized area. The stress unit is MPa.
226

HUB MOUNTING HOLES


(4 OF THEM)

STRUT MOUNTING HOLES

Figure 7-16 The forged part showing the strut and hub mounting holes.

Figure 7-17 Stage I (left) and Stage II models with redesigned spindle.

Figure 7-18 Spherical roller bearing with tapered bore to be installed on the tapered section
of the redesigned spindle (SKF Catalog).
227

Figure 7-19 von Mises stress distribution on the spindle-redesigned steering knuckle due to
an arbitrary moment of 690 N.m. The stress unit is MPa.

Figure 7-20 Dimensions of the original (top) and redesigned-spindle (bottom) steering
knuckles. The dimensions are in mm.
228

Figure 7-21 Front (left) and isometric views of Stage II optimization.

Figure 7-22 Cutting speed and feed rate (both indexes of machinability) versus hardness
of some steels (Trent and Wright, 2000).
229

Figure 7-23 Results of fatigue tests (R = -0.6) for induction hardened and non-hardened
truck stub axle (top) and depth of induction hardened zone revealed by
etching (bottom) (Schijve, 2001).
230

FROM LEFT, SPINDLE


RD ND ST
3 , 2 AND 1 STEP
FILLETS

Figure 7-24 Spindle fillets as the highest stressed areas of the steering knuckle are
recommended for inducing compressive residual stress.

Figure 7-25 Effect of rolling of the notch root on fatigue behavior under rotating-bending
fatigue of 37CrS4 steel (Kloos et al., 1987).
Chapter Eight

8 Summary and Conclusions


The effects of manufacturing process on fatigue design and optimization of

automotive components using experimental, numerical and analytical tools were investigated.

Even though the methodologies developed apply to a wide range of automotive and other

components, vehicle steering knuckles made of forged steel, cast aluminum, and cast iron

were selected as example parts for this study.

An extensive literature review was performed to compare manufacturing processes

of automotive components focusing on mechanical behavior, identify methods used in

durability assessment and optimization aspects. Material monotonic and fatigue properties

were obtained and compared through strain-controlled tests on forged steel, cast aluminum,

and cast iron specimens. Linear and nonlinear finite element analyses were employed to

obtain three dimensional stress and strain distributions in the components to facilitate

fatigue life predictions and optimization. Component tests were conducted to compare

fatigue behavior of the components made from different manufacturing processes and to

verify analytical evaluations and life predictions.

A general procedure for durability design of automotive components was developed

and pursued for the example parts. The strengths and shortages of various approaches

including nominal stress, local stress and local strain approaches were investigated. An

optimization procedure for a forged automotive component was also developed considering

manufacturing, material, and cost aspects, using the forged steel steering knuckle as the

231
232

example part. The manufacturing process details and production cost parameters for the

example part were discussed, the optimization problem was defined and the results were

analyzed. The findings of this study are summarized below.

Material Fatigue Behavior and Comparisons


1. From tensile tests and monotonic deformation curves it is concluded that forged steel is

considerably stronger and more ductile than cast aluminum and cast iron. Cast

aluminum and cast iron reached 37% and 57% of forged steel ultimate tensile strength,

respectively. The yield strength of cast aluminum and cast iron is also lower, 42% and

54% of the forged steel, respectively. The percent elongation, as a measure of ductility,

of cast aluminum and cast iron were found to be 24% and 48% of the forged steel,

respectively.

2. From strain-controlled cyclic tests it is concluded that the cyclic deformation curve of

the forged steel is independent of the geometrical direction (i.e. isotropic behavior). For

the fatigue behavior, however, some degree of anisotropy was observed. Both the long-

life as well as the short-life fatigue of forged steel were observed to be longer (by about

a factor of two) in the direction coinciding with the primary stressing direction of the

forged steering knuckle.

3. The cyclic yield strength of cast aluminum and cast iron were found to be 54% and

75% of forged steel, respectively. The cyclic strain hardening exponent of cast

aluminum and cast iron was 46% and 55% of the forged steel, respectively. These

indicate the higher cyclic strength of forged steel against yielding, and its higher

resistance to plastic deformation.

4. Significantly better S-N fatigue resistance of the forged steel was observed, as compared

with the two cast materials. Comparison of long-life fatigue strength (defined as the
233

fatigue strength at 106 cycles) shows that the fatigue limit of cast aluminum and cast

iron are only 35% and 72% of the forged steel, respectively. In addition, while the

fatigue strength of forged steel at 106 cycles is expected to remain about constant at

longer lives, fatigue strength of the two cast materials is expected to continuously drop

with longer lives.

5. Forged steel was found to be superior to cast aluminum and cast iron with respect to

low cyclic fatigue (i.e. cyclic ductility). In automotive design, cyclic ductility can be a

major concern when designing components subjected to occasional overloads,

particularly for notched components, where significant local plastic deformation can

occur.

6. Comparisons of strain-life fatigue behavior of the three materials demonstrate the

superiority of the forged steel over cast aluminum and cast iron. The forged steel

provides about a factor of 5 longer lives in the short-life regime, compared to the cast

aluminum and cast iron. In the high-cycle regime, forged steel results in about an order

of magnitude longer life than the cast iron, and about a factor of 3 longer life,

compared to the cast aluminum.

7. Neuber stress versus life plot, which considers the combined effects of stress and strain

amplitudes, shows forged steel to have about two orders of magnitude longer life than

cast iron and four orders of magnitude longer life than cast aluminum.

Finite Element Analysis


8. In order to avoid a complex meshed model that increases the FEA run-time, a relatively

coarse global mesh size, and a finer mesh at the vicinity of the critical points using free

local meshing feature was selected for each component. This procedure increased the
234

computational efficiency of the model significantly, particularly for nonlinear models

where material deformation was elastic-plastic.

9. Even at the lower loading level, which can be considered as an indication of long-life

service of the components, the material undergoes local plastic deformation. This is

evidence that mere use of linear elastic FEA is not sufficient for reliable fatigue life

predictions.

10. The spindle 1st step fillet area for the forged steel and hub bolt hole for the cast

aluminum and cast iron steering knuckles were found to be high-stressed locations with

high stress gradient. Both stress concentration as well as stress gradient due to the mode

of loading applied (i.e. bending in this case) are major factors in making an area fatigue-

critical location.

11. Although the primary loading on the components is unidirectional, it is shown that the

stress and strain at the critical locations are multiaxial. The type of primary loading that

the components undergo generates proportional stresses throughout the components.

For proportional stressing, von Mises stress and strain have been found effective in

calculating the equivalent values as a result of multiaxiality, and were used for fatigue life

analyses.

12. At the critical location the state of plane strain prevails for the forged steel steering

knuckle, while the state of stress at the critical location of the cast aluminum and cast

iron steering knuckles is closer to plane stress. Knowledge of the state of stress and

strain at the critical location of the components helps in choosing the appropriate

deformation model, leading to more accurate fatigue life predictions.

13. FEA simulation for cyclic loading is important for fatigue analysis since cyclic

deformation material response can be vastly different from monotonic deformation


235

response. In addition, the local and nominal behaviors are generally different under

various loading conditions. For example, as the nominal stress R-ratio remains almost

constant (close to zero), significant negative stress R-ratio is observed for most of the

simulations as a result of the residual stress generated at the stress concentrations due to

local plastic deformation.

Component Fatigue Behavior and Comparisons


14. Strain gages were used to validate the readings with analytical calculations. The

differences between experimentally measured and FEA-predicted strains obtained for

the forged steel and cast aluminum steering knuckles were found to be reasonable for

the complex geometries considered.

15. Based on the component testing observations, crack growth life was found to be a

significant portion of the cast aluminum steering knuckle fatigue life (on the average,

about 50% of the cast aluminum steering knuckle life is spent on macro-crack growth),

while crack growth life was an insignificant portion of the forged steel steering knuckle

fatigue life.

16. Component testing results showed the forged steel steering knuckle to have about two

orders of magnitude longer life than the cast aluminum steering knuckle, for the same

stress amplitude level. This occurred at both short as well as long lives. Comparison of

the strain-life prediction curves of the components demonstrated that the forged steel

steering knuckle offers more than an order of magnitude longer life than the cast iron

steering knuckle.

17. The failed forged steel steering knuckle had a typical ductile material fatigue failure

surface including crack initiation, smooth crack growth and rough fracture sections. The

failed cast aluminum had a relatively longer crack growth portion as compared to the
236

crack growth portion of the forged steel steering knuckle. The failure locations in the

component tests agreed with FEA predictions.

Fatigue Life Predictions


18. The nominal stress approach cannot be used for complex component geometries, such

as the cast aluminum steering knuckle in this study due to the fact that for complex

geometries, nominal stress can not be defined explicitly. For the forged steel steering

knuckle, the predictions of the nominal S-N approach were conservative, by about a

factor of seven on fatigue life, as compared to the experimental results.

19. The local stress or strain approaches in conjunction with the FEA results were found to

provide better life predictions, as compared with the commonly used nominal S-N

approach. This is partly due to the fact that the local approaches directly account for the

residual stresses from local plastic deformation.

20. The local strain approach using nominal stresses for the forged steel knuckle in

conjunction with Neuber’s rule predicted conservative lives, by about an order of

magnitude, as compared with experimental results. This confirms the suggestion that

Neuber’s rule is more applicable to plane stress states, since plane strain state existed at

the fatigue-critical location of the forged steel knuckle.

21. Life predictions based on local approaches using linear elastic FEA results in

conjunction with Neuber-corrected stresses were found to be close to those obtained

based on nonlinear elastic-plastic FEA results. Therefore, the simpler and less time

consuming linear elastic FEA, when modified to correct for plastic deformation, is an

effective and capable approach for life prediction of components with complex

geometries and/or loadings.


237

22. For the local stress approach, Gerber’s mean stress parameter provides better predicted

fatigue lives, as compared with the experimental lives, than the commonly used

modified Goodman equation. For the local strain approach, Morrow’s mean stress

parameter provides better predicted fatigue lives than the Smith-Watson-Topper mean

stress parameter.

Optimization
23. Manufacturing process considerations, material and cost parameters are major

constituents of a general optimization procedure with durability constraints for

automotive component. A geometrical optimization without these considerations is not

a practical approach for such high volume components.

24. The proposed material alternatives provide higher fatigue strength for the component.

Manufacturability and cost are two other main issues that are critical to the final

selection of the replacing material(s). Limited weight saving is achieved by replacing the

potential alternative materials, mainly due to geometrical constraints. If comprehensive

changes to the geometry are allowed or for other components with fewer constraints,

the weight saving will be more significant.

25. Additional manufacturing operations such as surface hardening and surface rolling to

induce compressive residual stress can be considered to improve fatigue strength of the

forged steel steering knuckle at the spindle fillet area.

26. Overall weight and cost reductions of at least 12% and 5%, respectively, are estimated

for the example part following the optimization task. The cost of the saved material is

additional reduction, though not very considerable due to small portion of material cost

within the total production cost. Due to the small size of the forged steel steering

knuckle and many attachment compatibility constraints, limited changes could be


238

implemented during the optimization process. More comprehensive changes require a

more detailed design of the component and the suspension system.

27. The approach that was followed is applicable to other forged components. Components

with fewer geometrical restrictions than the steering knuckle considered have much

higher potential for weight reduction and cost savings.


239

References
AISI, Bar Steel Fatigue Database, Version 1.0, www.barsteelfatigue.autosteel.org , American Iron and
Steel Institute.

AISI, 2002, Steel Industry Technology Roadmap, Technical Report, American Iron and Steel
Institute.

AISI, 2001, UltraLight Steel Auto Suspension (ULSAS) Engineering Report, American Iron and
Steel Institute.

ASM, 1989, ASM Handbook, Vol. 16: Machining, American Society of Metals.

ASME, 2001, ASME Boiler and Pressure Vessel Code: Code Cases: Boilers and Pressure Vessels,
American Society of Mechanical Engineers.

ASTM, 1998, Standard Practice for Strain-Controlled Fatigue Testing, ASTM Standard E606-92,
Vol. 03.01, American Society of Testing and Materials.

ASTM, 2000, Standard Test Method for Tensile Strain-Hardening Exponents (n-Values) of Metallic
Sheet Materials, ASTM Standard E646-00, Vol. 03.01, American Society of Testing and
Materials.

ASTM, 1998, Standard Practice for Statistical Analysis of Linear or Linearized Stress-Life (S-N) and
Strain-Life (e-N) Fatigue Data, ASTM Standard E739-91, Vol. 03.01, American Society of
Testing and Materials.

ASTM, 2001, Standard Test Methods for Tension Testing of Metallic Materials, ASTM Standard E8-
01, Vol. 03.01, American Society of Testing and Materials.

ASTM, 2002, Standard Practice for Verification and Classification of Extensometer, ASTM Standard
E83-02, Vol. 03.01, American Society of Testing and Materials.

ASTM, 1999, Standard Practice for Verification of Specimen Alignment under Tensile Loading, ASTM
Standard E1012-99, Vol. 03.01, American Society of Testing and Materials.

Afzal A. and Fatemi, A., 2004, “A Comparative Study of Fatigue Behavior and Life
Predictions of Forged Steel and PM Connecting Rods,” SAE Technical Paper No. 2004-01-
1529, Society of Automotive Engineers.

Aloi, Jr., N. E., Burnett, M. E., and Kendrick, R., 1997, “Optimizing Forgings for
Automotive Transmission Hubs by Warm Forming Vanadium-Microalloyed Steels,” SAE
Technical Paper No. 970517, Society of Automotive Engineers.

Bayer, J., 2003, “Steel Bars for Automotive Applications,” Advanced Materials and Processes,
Vol. 161, No. 8, pp. 46-49.
240

Béranger, A. S., Bérard, J. Y., and Vittori, J. F., 1997, “A Fatigue Life Assessment
Methodology for Automotive Components,” In Fatigue Design of Components, ESIS
Publication 22, Proceedings of the 2nd International Symposium on Fatigue Design, FD’95,
5-8 September, 1995, Helsinki, Finland, Marquis, G. and Solin, J., eds., pp. 17-25.

Beretta, S., Blarasin, A., Endo, M., Giunti, T., and Murakami, Y., 1997, “Defect Tolerant
Design of Automotive Components,” International Journal of Fatigue, Vol. 19, No. 4, pp. 319-
333.

Berger, C., Eulitz, K. G., Heuler, P., Kotte, K. L., Naundorf, H., Schuetz, W., Sonsino, C.
M., Wimmer A., and Zenner, H., 2002, “Betriebsfestigkeit in Germany - An Overview,”
International Journal of Fatigue, Vol. 24, pp. 603-625.

Birch, T. W., 2000, Automotive Chassis Systems, 1st edition, Thomson Learning.

Blair, M. and Monroe, R. W., 2002, “Castings or Forgings? A Realistic Evaluation,”


www.sfsa.org, Steel Founders’ Society of America, Barrington, Illinois.

Blarasin, A. and Farsetti, P., 1989, “A Procedure for the Rational Choice of Microalloyed
Steels for Automotive Hot-Forged Components Subjected to Fatigue Loads,” International
Journal of Fatigue, Vol. 11, No. 1, pp. 13-18.

Blarasin, A. and Giunti, T., 1997, “Manufacturing Processes Influence Fatigue Life,”
Automotive Engineering, Vol. 105, pp. 93-95.

Bonnen, J. J. F. and Topper, T. H., 1999, “The Effect of Bending Overloads on Torsional
Fatigue in Normalized 1045 Steel,” International Journal of Fatigue, Vol. 21, pp. 23–33.

Botkin, M. E., 1991, “Shape Design Modeling Using Fully Automatic Three-Dimensional
Mesh Generation,” Finite Elements in Analysis and Design, Vol. 10, No. 2, pp. 165-181.

Bruder, T. and Schön, M., 2001, “Durability Analysis of Carburized Components Using a
Local Approach Based on Elastic Stresses,” Material wissenschaft und Werkstoff technik, Vol. 32,
pp. 377-387.

Byrer, T. G., ed., Semiatin, S. L. and Vollmer, D. C., associated eds., 1985, Forging Handbook,
Forging Industry Association (FIA), Cleveland, OH.

Chu, C. C. 1997, “Multiaxial Fatigue Life Prediction Method in the Ground Vehicle
Industry,” International Journal of Fatigue, Vol. 19 (Supp. No. 1), pp. S325-S330.

Conle, F. A., 1987, “Durability Analysis under Complex Multiaxial Loading,” SAE Technical
Paper No. 871969, Society of Automotive Engineers.

Conle, F. A. and Chu, C. C., 1997, “Fatigue Analysis and the Local Stress-Strain Approach in
Complex Vehicular Structures,” International Journal of Fatigue, Vol. 19, No. 1, pp. S317-S323.
241

Conle, F. A. and Mousseau, C. W., 1991, “Using Vehicle Dynamics Simulation and Finite-
Element Results to Generate Fatigue Life Contours for Chassis Components,” International
Journal of Fatigue, Vol. 13, No. 3, pp. 195-205.

Cristinacce, M., James, D. E., and Milbourn, D. J., 1998, “The Future Competitiveness of
Automotive Forging the Steelmakers’ Contribution,” In Forging and Related Technology-
ICFT’98, IMechE Conference Transactions, pp. 37-50.

Devlukia, J. and Bargmann, H., 1997, “Fatigue Assessment of an Automotive Suspension


Component Using Deterministic and Probabilistic Approaches,” In Fatigue Design of
Components, ESIS Publication 22, Proceedings of the 2nd International Symposium on Fatigue
Design, FD’95, 5-8 September, 1995, Helsinki, Finland, Marquis, G. and Solin, J., eds., pp.
436-445.

Diboine, A., 1996, “Fatigue assessment of components under complex loadings,” In


Biaxial/Multiaxial Fatigue: Multiaxial Fatigue and Design, Proceedings of the 4th International
Conference, Saint-Germain en Laye, France, Pineau, A., Cailletaud, G., and Lindley, T. C.,
eds., pp. 425-444.

Esper, F. J. and Sonsino, C. M., 1994, Fatigue Design for PM Components, European Powder
Metallurgy Association, England.

Farsetti, P. and Blarasin, A., 1988, “Fatigue Behavior of Microalloyed Steels for Hot-Forged
Mechanical Components,” International Journal of Fatigue, Vol. 10, No. 3, pp. 153-161.

Fatemi, A. and Zoroufi, M., 2002, “Fatigue Performance Evaluation of Forged versus
Competing Process Technologies: A Comparative Study,” In Proceedings, 24th Forging
Industry Technical Conference, Forging Industry Association, Cleveland, OH.

Ferreira, W. G., Martins, F., Kameoka, S., Salloum, A. S., and Kaeya, J. T., 2003, “Structural
Optimization of Automotive Components Applied to Durability Problems,” SAE Technical
Paper No. 2003-01-3547, Society of Automotive Engineers.

Firat, M. and Kocabicak, U., 2004, “Analytical Durability Modeling and Evaluation-
Complementary Techniques for Physical Testing of Automotive Components,” Engineering
Failure Analysis, Vol. 11, pp. 655–674.

Forcellese, A., Gabrielli, F., and Ruffini, R., 1996, “Application of a Decision Making
Method in the Forging Condition Optimization for Manufacturing Automotive
Components,” Journal of Materials Processing Technology, Vol. 60, pp. 125-132.

Forging Industry Association, Forging Facts, www.forging.org.

Groover, M. P., 2002, Fundamentals of Modern Manufacturing: Materials, Processes and Systems,
2nd edition, John Wiley and Sons, Inc.

Grunwald, J. and Schnack, E., 1997, “Shape Optimization for Fatigue Problems in
Mechanical Engineering,” Engineering Optimization, Vol. 29, No. 1-4, pp. 235-258.
242

Gunnarson, S., Ravenshorst, H., and Bergstorm, C. M., 1987, “Experience with Forged
Automotive Components in Precipitation Hardened Pearlitic-Ferritic Steels,” In Fundamentals
of Microalloying Forging Steels, Proceedings of an international symposium sponsored by the
Ferrous Metallurgy Committee of the Metallurgical Society, Golden, Colorado, July 8-10,
1986, Krauss, G. and Banerji, S. K. eds., pp. 325-338.

Haeg, S. R., 1997, “Steer Dynamics in road Simulation,” In SAE SP-1223, Steering and
Suspension Technology Buckley, T. J., et al., Session Organizers, Society of Automotive
Engineers.

Haiba, M., Barton, D. C., Brooks P. C., and Levesley, M. C., 2002, “Review of Life
Assessment Techniques Applied to Dynamically Loaded Automotive Components,”
Computers and Structures, Vol. 80, pp. 481–494.

Halderman, J. D. and Mitchel, Jr., C. D., 2000, Automotive Steering, Suspension, and Alignment,
2nd edition, Prentice -Hall, Inc.

Härkegård, G. and Mann T., 2003, “Neuber Prediction of Elastic-Plastic Strain


Concentration in Notched Tensile Specimens under Large-Scale Yielding,” Journal of Strain
Analysis, Vol. 38, No. 1, pp. 79-94.

Heuler, P. and Birk, O., 2002, “Durability Assessment of Automotive Aluminum Parts,”
Fatigue & Fracture of Engineering Materials & Structures, Vol. 25, pp. 1135-1148.

Heyes, P. J., Milsted, M. G., and Dakin, J., 1996, “Multiaxial Fatigue Assessment of
Automotive Chassis Components on the Basis of Finite Element Models,” In
Biaxial/Multiaxial Fatigue: Multiaxial Fatigue and Design, Proceedings of the 4th International
Conference, Saint-Germain en Laye, France, Pineau, A., Cailletaud, G., and Lindley, T. C.,
eds., pp. 317-330.

Hirt, G., Cremer, R., Witulski, T., and Tinius, H. C., 1997, “Lightweight Near Net Shape
Components Produced by Thixoforming,” Materials and Design, Vol. 18, No. 4, pp. 315-321.

Hoffmann, M. and Seeger, T., 1985, “A Generalized Method for Estimating Multiaxial
Elastic-Plastic Notch Stresses and Strain, Part I: Theory,” Transactions of ASME, Journal of
Engineering Materials and Technology, Vol. 107, pp. 250-254.

Houshito, S., Watanabe, Y., Goka, M., and Ishihara, Y., 1989, “Feasibility Study on the
Application of High Strength Ductile Iron to Automotive Chassis Parts, International Journal of
Materials and Product Technology, Vol. 4, No. 3, pp. 285-299.

IDEAS, 2002, IDEAS-10 Help Library Bookshelf, Unigraphics Solutions Inc.

Jang, G. B., Hur, M. D., and Kang, S. S., 2000, “A Study on the Development of a
Substitution Process by Powder Metallurgy in Automobile Parts,” Journal of Materials Processing
Technology, Vol. 100, pp. 110-115.

Johnson, J. A., Krauss, G., and Matlock, D. K., 2000, “Fatigue of Microalloyed Bar Steels,”
SAE Technical Paper No. 2000-01-0615, Society of Automotive Engineers.
243

Joseph, R. A. and Stout, R., 2004, “AISI Bar Machinability Database of Steels Using Sintered
Carbide Tools in Single Point Bending,” SAE Technical Paper No. 2004-01-1525, Society of
Automotive Engineers.

Jung, J. Y., 2002, “Manufacturing Cost Estimation for Machined Parts Based on
Manufacturing Features,” Journal of Intelligent Manufacturing, Vol. 13, pp. 227-238.

Kalpakjian, S. and Schmid, S. R., 2000, Manufacturing Engineering and Technology, 4th edition,
Prentice Hall, Inc., Upper Saddle River, New Jersey 07458.

Kawashima, Y., Nishikawa, J., and Paxton, G. B., 1998, “Developments in Automation for
the Manufacture of Automotive Forgings,” In Forging and Related Technology-ICFT’98, IMechE
Conference Transactions, pp. 435-444.

Kloos, Kh., Fuchsbauer B., and Adelmann J., 1987, “Fatigue Properties of Specimens Similar
to Components Deep Rolled under Optimized Conditions,” International Journal of Fatigue,
Vol. 9, No. 1, pp. 35-42.

Kocabicak, U. and Firat, M., 2001, “Numerical Analysis of Wheel Cornering Fatigue Tests,”
Engineering Failure Analysis, Vol. 8, pp. 339-354.

Krishna, M. M. R., 2001, “Finite Element Shape Optimization of a Steering Knuckle for a
Heavy Truck – A Case Study,” SAE Technical Paper No. 2001-01-0634, Society of Automotive
Engineers.

Krishna, M. M. R. and Fetcho, M. R., 1998, “Thermal Shape Vectors in Finite Element
Shape Optimization of a Steering Knuckle,” In Recent Advances in Solid and Structures, ASME
PVP, Chung, H. H., ed., Vol. 381, pp. 165-170.

Kuratomi, H., Uchino, M., Kurebayashi, Y., Namiki, K., and Sugiura, S., 1990,
“Development of Lightweight Connecting Rod Based on Fatigue Resistance Analysis of
Microalloyed Steel,” SAE Technical Paper No. 900454, Society of Automotive Engineers.

Lee, S. B., 1986, “Structural Fatigue Tests of Automobile Components under Constant
Amplitude Loadings,” In Fatigue Life Analysis and Prediction, Proceedings, International
Conference and Exposition on Fatigue, Goel, V. S., ed., American Society of Metals, pp.
177-186.

Lee, D. C. and Lee, J. I., 2003, “Structural Optimization Concept for the Design of an
Aluminum Control Arm,” Proceedings of the IMechE, Part D: Journal of Automobile Engineering,
Vol. 217, pp. 647-656.

Lee, Y. L., Raymond, M. N., and Villaire, M. A., 1995, “Durability Design Process of a
Vehicle Suspension Component,” Journal of Testing and Evaluation, Vol. 23, pp. 354-363.

Meritor Automotive Inc., 1997, Ductile Iron Cyclic and Fatigue Properties, Ductile Iron Data,
Section 10, Table 10.2, Ductile Iron Society.
244

Molski, K. and Glinka, G., 1981, “A Method of Elastic-Plastic Stress and Strain Calculation
at a Notch Root,” Materials Science and Engineering, Vol. 50, pp. 93-100.

Nägele, H., Wörner, H., and Hirschvogel, M., 2000, “Automotive Parts Produced by
Optimizing the Process Flow Forming – Machining,” Journal of Materials Processing Technology,
Vol. 98, pp. 171-175.

Nakamura, S., Mizuno, K., Matsubara, T., and Sato, Y., 1993, “Development of High
Fatigue Strength Free machining Microalloyed Steel for Connecting Rods,” SAE Technical
Paper No. 930619, Society of Automotive Engineers.

Naylor, D. J., 1998, “Microalloyed Forging Steels,” In Microalloying in Steels, Proceedings of


the international conference, Materials Series Forum, Rodriguez-Ibabe, J. M., Gutierrez, I.,
and Lopez, B., eds., Vol. 284-286, pp. 83-94.

Neuber, H., 1961, “Theory of Stress Concentration for Shear-Strained Prismatical Bodies
with Arbitrary Nonlinear Stress-Strain Law,” Transactions of ASME, Journal of Applied
Mechanics, Vol. 28, pp. 544-550.

Nicolaou, P., Thurston, D. L., and Carnahan, J. V., 2002, “Machining Quality and Cost:
Estimation and Tradeoffs,” Transactions of ASME, Journal of Manufacturing Science and
Engineering, Vol. 124, pp. 840-851.

Osuzu, H., Shiraga, T., Shiroi, Y., Taniguchi, Y., Tsujimura, K., and Kido, H., 1986,
“Application of Microalloyed Steels to Achieve High Toughness in Hot Forged Components
without Further Heat Treatments,” SAE Technical Paper No. 860131, Society of Automotive
Engineers.

Pook, L. P., 1997, “Standard Fatigue Tests for Components,” In Fatigue Design of Components,
ESIS Publication 22, Proceedings of the 2nd International Symposium on Fatigue Design,
FD’95, 5-8 September, 1995, Helsinki, Finland, Marquis, G. and Solin, J., eds., pp. 203-212.

Reinsch, S., Müssig, B., Schmidt, B., and Tracht, K., 2003, “Advanced Manufacturing System
for Forging Products,” Journal of Materials Processing Technology, Vol. 138, pp. 16-21.

Repgen, B., 1998, “Optimized Connecting Rods to Enable Higher Engine Performance and
Cost Reduction,” SAE Technical Paper No. 980882, Society of Automotive Engineers.

Rice, R. C., ed., 1997, SAE Fatigue Design Handbook, prepared under the auspices of the SAE
Fatigue Design and Evaluation Committee, Society of Automotive Engineers.

Savaidis, G., 2001, “Analysis of Fatigue Behavior of a Vehicle Axle Steering Arm Based on
Local Stresses and Strains,” Material wissenschaft und Werkstoff technik, Vol. 32, No. 4, pp. 362-
368.

Savaidis, G., Savaidis, A., Tsamasphyros, G., and Zhang Ch., 2002, “On Size and
Technological Effects in Fatigue Analysis and Prediction of Engineering Materials and
Components,” International Journal of Mechanical Sciences, Vol. 44, pp. 521–543.
245

Schey, J. A., 2000, Introduction to Manufacturing Processes, 3rd edition, McGraw-Hill, Inc.

Schijve, J., 2001, Fatigue of Structures and Materials, 1st edition, Kluwer Academic.

Schramm, U., Thomas, H., and Zhou, M., 2002, “Manufacturing Considerations and
Structural Optimization for Automotive Components,” SAE Technical Paper No. 2002-01-
1242, Society of Automotive Engineers.

Seeger, T. and Heuler, P., 1980, “Generalized application of Neuber’s Rule,” Journal of Testing
and Evaluation, Vol. 8, No. 4, pp. 199-204.

Shenoy, P. S., 2004, “Dynamic Load Analysis and Optimization of Connecting Rod,”
Master’s Thesis, The University of Toledo, Toledo, OH.

Shenoy, P. S. and Fatemi, A., 2005, “Connecting Rod Optimization for Weight and Cost
Reduction,” SAE Technical Paper, Society of Automotive Engineers (to be published).

Shigley, J. E. and Mischke, C. R., 1989, Mechanical Engineering Design, 5th edition, McGraw-Hill,
Inc.

Siljander, A., Kahonen, A., and Lehtonen, M., 1997, “Durability Assessments of Straddle
Carrier Wheel Supports,” In Fatigue Design of Components, ESIS Publication 22, Proceedings of
the 2nd International Symposium on Fatigue Design, FD’95, 5-8 September, 1995, Helsinki,
Finland, Marquis, G. and Solin, J., eds., pp. 59-71.

Socie, D. F. and Marquis, G. B., 2000, Multiaxial Fatigue, 1st edition, Society of Automotive
Engineers, Warrendale, Pa.

Sonsino, C. M., Kaufmann, H., and Grubišić, V., 1997, “Transferability of Material Data for
the Example of a Randomly Loaded Forged Truck Stub Axle,” SAE Technical Paper No.
970708 in SAE PT-67, Recent Developments in Fatigue Technology, Chernenkoff, R. A. and
Bonnen, J. J., eds., Society of Automotive Engineers.

Sonsino, C. M., Kaufmann, H., Foth, J., and Jauch, F., 1997, “Fatigue Strength of Driving
Shafts of Automatic Transmission Gearboxes under Operational Torques,” SAE Technical
Paper No. 970706 in SAE PT-67, Recent Developments in Fatigue Technology, Chernenkoff, R. A.
and Bonnen, J. J. eds., Society of Automotive Engineers.

Stephens, R. I., Fatemi, A., Stephens, R. R., and Fuchs, H. O., 2000, Metal Fatigue in
Engineering, 2nd edition, John Wiley and Sons.

Susmel, L. and Taylor, D., 2003, “Fatigue Design in the Presence of Stress Concentration,”
Journal of Strain Analysis, Vol. 38, No. 5, pp. 443-452.

Tang, J., Ogarevic, V., and Tsai, C.-S., 2001, “An Integrated CAE Environment for
Simulation-Based Durability and Reliability Design,” Advances in Engineering Software, Vol. 32,
pp. 1-14.
246

Taylor, D., 1996, “Crack Modelling: A Technique for the Fatigue Design of Components,”
Engineering Failure Analysis, Vol. 3, No. 2, pp. 129-136.

Taylor, D., 1997, “Crack Modeling: A Novel Technique for the Prediction of Fatigue Failure
in the Presence of Stress Concentrations,” Computational Mechanics, Vol. 20, pp. 176-180.

Taylor, D., Bologna, P., and Bel Knani, K., 2000, “Prediction of Fatigue Failure Location on
a Component Using a Critical Distance Method,” International Journal of Fatigue, Vol. 22, pp.
735–742.

Taylor, D., Zhou, W., Ciepalowicz, A. J., and Devlukia, J., 1999, “Mixed-Mode Fatigue from
Stress Concentrations: an Approach Based on Equivalent Stress Intensity,” International
Journal of Fatigue, Vol. 21, pp. 173-178.

Tlusty, J., 2000, Manufacturing Processes and Equipment, 1st edition, Prentice Hall, Upper Saddle
River, NJ.

Todd, R. H., Allen, D. K., and Alting, L., 1994, Manufacturing Processes Reference Guide, 1st
edition, Industrial Press Inc.

Trent, E. M. and Wright, P. K., 2000, Metal Cutting, 4th edition, Butterworth-Heinemann.

van Bennekom, A. and Wilke, F., 2003, “Comparison between the Physical, mechanical and
corrosive properties of forged and cast stainless steels,” Krupp Edelstahlprofile, Report 2003-1.

Vanderplaats, G. N., 1999, Numerical Optimization Techniques for Engineering Design, 1st edition,
Vanderplaats Research & Development, Inc., Colorado Springs, CO.

Vanderplaats, G., N. and Miura, H., 1986, “Trends in Structural Optimization: Some
Considerations in Using Standard Finite Element Software,” SAE Technical Paper No. 860801,
Society of Automotive Engineers.

Vazquez, V. and Altan, T., 2000, “Die Design for Flashless Forging of Complex Parts,”
Journal of Materials Processing Technology, Vol. 98, pp. 81-89.

Weber, M., 1993, “Comparison of Advanced Procedures and Economics for Production of
Connecting Rods,” Powder Metallurgy International, Vol. 25, No. 3, pp. 125-129.

Whittaker, D., 2001, “The Competition for Automotive Connecting Rod Market,” Metal
Powder Report, Vol. 56, Issue 5, pp. 32-37.

Wigant, C. C. and Stephens, R. I., 1987, “Low Cycle Fatigue of A356-T6 Cast Aluminum
Alloy,” SAE Technical Paper No. 870096, Society of Automotive Engineers.

Witter, M. C., Dumbacher, S. M., and Brown, D. L., 1999, “Converting a Steering Knuckle
into a 6-DOF Force Transducer,” In Proceedings, The 17th International Modal Analysis
Conference, Wicks, A. L. and DeMichele, D. J., Technical Program Chair-persons, pp. 1622-
1632.
247

Zeng, Z. and Fatemi, A., 2001, “Elasto-Plastic Stress and Strain Behavior at Notch Roots
under Monotonic and Cyclic Loadings,” Journal of Strain Analysis, Vol. 36, No. 3, pp. 287-300.

Zoroufi, M. and Fatemi, A., 2004, “Fatigue Life Comparisons of Competing Manufacturing
Processes: A Study of Steering Knuckle,” SAE Technical Paper No. 2004-01-0628, Society of
Automotive Engineers.
248

Appendix A - Summary of Literature Review for


Fatigue Life Prediction of
Automotive Components
This appendix summarizes a number of relatively recent studies related to fatigue life

prediction of automotive components. The procedure followed in each study is briefly

discussed and significant points are mentioned. The articles are arranged in the order of

years published in order to have a sense of the progress in the field.


Author(s): Blarasin and Farsetti (1989) Prediction Procedure

Article Title: A procedure for the rational choice of microalloyed steels for automotive
hot-forged components subjected to fatigue loads

Component(s): Steering knuckle of the front suspension of a commercial vehicle

Stress-Strain
Nominal stress history (corrugated proving ground) + Neuber’s rule
Calculation:

Life Prediction
Strain-life
Method(s):

Experiment
Variable amplitude uniaxial component test
Verification:

Author’s 1. Overestimated predictions for Q&T-steel components; attributed to


Remarks: underestimate of the stress concentration as well as to differences
between the microstructure of the component and the fatigue
specimens.
2. Good agreement between calculated and experimental lives for
microalloyed-steel components.

249
Author(s): Conle and Mousseau (1991) Prediction Procedure

Article Title: Using vehicle dynamics simulations and finite-element results to


generate fatigue life contours for chassis components 1. Forces generated on the wheels are recorded on proving ground.

Component(s): Upper control arm of an automotive front suspension 2. DADS 10 model computes the upper control arm forces.

Stress-Strain 3. Elastic FEA calculates loads on each element.


Elastic FEA + Neuber plasticity correction
Calculation:

Life Prediction 4. Critical fatigue activity plane is determined by an elastic multiaxial


Critical plane
Method(s): superposition routine.

Experiment 5. Stresses normal to this maximum principal stress plane are


None
Verification: rainflow counted, converted to local stress and strain with
Neuber’s rule, and summed for fatigue damage.
Author’s 1. The method finds the critical angle for fatigue calculations of
Remarks: each element for all load cases, not just the critical ones.
2. The vehicle subsystem models (e.g. bushings) need to be
correlated with improved laboratory tests and real service
load behavior.
3. Other methods for multiaxial to uniaxial equivalencing
should be investigated.
4. Verification with actual experimental results at each stage of
analysis needs to be done.

250
10 Dynamic Analysis and Design System
Author(s): Heyes, Milsted and Dakin (1994) Prediction Procedure

Article Title: Multiaxial fatigue assessment of automotive


chassis components on the basis of finite
element models

Component(s): General – steering knuckle as example

Stress-Strain Elastic FEA (+ Neuber plasticity correction


Calculation: for strain-life predictions)

Life Prediction Strain-life: abs max principal – normal strain


Method(s): abs max principal – SWT
abs max principal – Morrow
signed Tresca criterion
signed von Mises criterion
Critical plane: normal strain, εn
shear strain, γ
Bannantine normal, εn σn,max
Fatemi and Socie, γ(1+n σn,max/σy)

Experiment
None
Verification:

Author’s A two stage assessment procedure is


Remarks: proposed. First, critical events and locations
are identified through preliminary life
estimates and examination of strain ranges,
and summary statistics are obtained
concerning the behavior of the elastic stress
tensor. A more detailed analysis of the
critical events and locations is then carried
out, with the simplest credible method
appropriate to the prevailing stress-strain
state being used to assess life.

251
Author(s): Lee, Raymond and Villaire (1995) Prediction Procedure
Computer Aided Engineering in Fatigue (CEA-FATIGUE) methodology
Article Title: Durability design process of a vehicle suspensions component was applied:
1. Reference Elastic Stresses: The reference stresses corresponding to each
Component(s): Steering knuckle load source was first determined by performing static-elastic finite
element analyses, and was then multiplied by the load factor histories
Stress-Strain for the entire elastic stress histories.
Elastic + Glinka plasticity correction
Calculation: 2. Elastic Stress History Synthesis: The elastic stress components due to
numerous static loads were combined to derive the time history for
Life Prediction each stress components. That is σ ( t ) = N σ ⋅ L ( t )
Method(s):
Critical plane (defined as maximum principal stress plane) k ∑
i =1
ki i

3. Failure Mode Computation: The critical plane approach was adopted. The
Experiment Component strain gage measurements to verify the FEM. plane where the fatigue crack initiates, the critical plane, was defined as
Verification: Two-strut attachment constant amplitude fatigue tests in fore/aft the plane on which major elastic principal stresses were maximized
and lateral directions for CAE-FATIGUE life validation. during the entire multiaxial cyclic loading history. The search for the
critical plane was found necessary, because the principal stress/strain
Author’s 1. Average errors of 23% and 27% for lateral and fore/aft loads axes varies with time for non-proportional loading problems. The
Remarks: were considered reasonable for the complex knuckle’s strain maximum normal stress and normal strain amplitude on the critical
correlation. plane were the damage parameters that characterized the fatigue life
2. The life differences between the measured and predicted were through the SWT criterion. It was noted that the SWT parameters are
found by factors of 3.9 and l.4, for fore/aft and lateral tests at applicable to the failure modes characterized by crack initiation on
the R50C50 life, respectively. planes of maximum principal stress, and are especially satisfactory for
3. It is depicted that the accuracy of the predictions depends automotive suspension and engine components with cast materials.
upon (1) material properties used, (2) finite element codes, (3) 4. Rainflow Cycle Counting: The objective of cycle counting is to reduce all
finite element modeling techniques, (4) finite element irregular stress history to a table of number of cycles versus
boundary conditions, (5) measured loading history, (6) service stress/strain levels (including mean and amplitude) so that a
environment, (7) fatigue damage parameters, (8) cycle stress/strain life curve obtained with fully reversed loading cycle may
counting algorithm, (9) notch stress and strain estimates, and be used for estimating the fatigue life of the component. A cycle is
(10) definition of cracks. defined as a closed loop on a cyclic stress-strain curve. The one-pass
4. It is emphasized that CAE-FATIGUE may reduce design rainflow algorithm was adopted for performing cycle counting.
iterations, but can never replace final test validations. 5. Mean Stress Effect and Plasticity Justification: The SWT equation was used
5. It is stated that fatigue is only one of the design criteria; for mean stress effect. When the elastic stresses obtained from the
yield/ultimate strength as well as stiffness cannot be ignored. elastic finite element analysis exceeded the cyclic yield stress range, the
stresses were adjusted. Glinka’s energy density method was adopted
for the plasticity justification.
6. Damage Accumulation: The fatigue life was calculated by accumulating
the damage due to the stress/strain cycles using Miner’s (linear

252
damage) rule.
Author(s): Taylor (1996 and 1997) Prediction Procedure
Taylor, Zhou, Ciepalowicz and Devlukia (1999)

Article Title: 1. Crack modeling: A novel technique for the


prediction of fatigue failure in the presence of
stress concentrations
2. Mixed-mode fatigue from stress concentrations:
an approach based on equivalent stress intensity

Component(s): General - Crankshaft

Stress-Strain
Elastic FEA
Calculation:

Life Prediction
Crack modeling method (CMM)
Method(s):

Experiment Constant amplitude uniaxial (bending and torsion)


Verification: component test

Author’s 1. The method can be successfully applied to


Remarks: components which are analyzed using FEA.
Predictions for the fatigue limit of an
automotive crankshaft were correct to within
3% of the experimental value.
2. The method is simple to use, requiring only
elastic FE data and basic material properties. It
is an improvement over existing methods for
the analysis of sharp notches, which require
elastic/plastic analyses combined with 1. The elastic stresses along a line moving away from the point of maximum stress
empirical corrections for the effect of notch are examined and compared to the stress distribution for a crack of standard
size and stressed volume. geometry. The geometry chosen was that of a through-crack in an infinite plate
3. Predictions were not significantly affected by under uniform tension for which the variation of stress with distance from the
refinement of the FE mesh, even though this crack tip was shown by Westergaard.
had a strong effect on the value of the hot-spot 2. Results from the FE analysis of the component, in the form of a stress/distance
stress. This is very advantageous for the curve, are compared with a curve derived from the infinite plate crack equation.

253
practical use of the technique. The constants of the equation are varied until a best possible fit is obtained
Author(s): Taylor (1996 and 1997) Prediction Procedure
Taylor, Zhou, Ciepalowicz and Devlukia (1999)

4. The CMM suffers from problems when faced between the two curves.
with a notch or crack which is physically short. 3. The value of K, and thus ΔK is then known because for the Westergaard geometry
It is expected that the CMM prediction will the solution is given by ΔK = Δσ (πD )1 2 where Δσ and D are the constants of the
give a fatigue limit that is too high in these
Westergaard equation.
cases.
4. The fatigue limit for the component is predicted by finding the applied loads on
the FE model which are just sufficient to make ΔK=ΔKth. Because the FE model is
linear-elastic this is usually just a matter of scaling.
5. In line with the approach of Smith and Miller, a second prediction of fatigue limit
for the component must be made by finding the loads for which the local stress at
the stress-concentration is equal to the unnotched fatigue limit.
6. The second prediction will be the correct one if we are in the blunt-notch regime.
In any event the higher of the two predictions of fatigue limit is used.

254
Author(s): Chu (1997) Prediction Procedure

Article Title: Multiaxial fatigue life prediction method in the ground vehicle
industry

Component(s): General

Stress-Strain Strain-gaged local history


Calculation: Elastic FEA + Neuber’s plasticity correction
Elastic-plastic FEA

Life Prediction Critical plane:


Method(s): SWT (σmax εa) and biaxial version of it (σmax εa + |τ|max γa)

Experiment
Not applicable
Verification:

Author’s 1. The multiaxial fatigue analysis program is separated into two


Remarks: parts: the stress analysis part and the fatigue damage analysis
part.
2. The objective of the stress analysis is to obtain the complete 3D
stress and strain history of a potential failure site.
3. The fatigue damage analysis involves rainflow counting of the
fatigue events and the summing up of their damage. The
method is similar to that used for uniaxial problems, with the
additional complexity of the biaxial damage criterion and the
critical plane approach: using Brown and Miller’s biaxial damage
criterion as an example, for every potential failure plane,
damage events are determined by rainflowing the shear strain
component (γ). During each shear strain defined damage event,
the variation of normal strain ( ε) is also recorded. The damage
caused by each event is then assessed by the biaxial damage
parameter (γa + Kεa). The (shortest) fatigue life (on the most
critical plane) can be determined only after the fatigue analysis
is performed on each and every potential failure plane.

255
Author(s): Sonsino, Kaufmann, Foth and Jauch (1997) Prediction Procedure

Determination of the reference S-N curve for damage calculation:


Article Title: Fatigue strength of driving shafts of automatic transmission
gearboxes under operational torques

Component(s): Transmission shaft (induction-hardened surface)

Stress-Strain Elastic FEA (the measured strains showed minor plastic strains)
Calculation: Normal stress hypothesis was used for the hardened surface
because of its brittleness and distortion energy hypothesis was used
for the tougher non-hardened core.

Life Prediction
Nominal and local stress-life
Method(s):

Experiment
Constant and variable amplitude component tests
Verification:

Author’s 1. The nominal stress concept (case A) is not recommended


Remarks: because the nominal stresses do not always correlate with the
local stresses. Five fatigue life calculation procedures were followed:
2. Concepts based on local stresses and strains (Cases B and C) Case A, The nominal stress concept: the calculation starts from the external
describe the material behavior better than the methods based load that corresponds to the nominal shear stress, as well as to the local
on nominal loads (Cases A and D). strain, because of the approximate linearity observed between load and local
3. The PJ method sometimes results in overestimation of fatigue strain.
life. Concepts based on local stresses and strains:
4. The following information must be available and the following Case B: SWT (PSWT-N curve derived from the S-N curve of the shaft is
procedures must be followed in order to make a reliable used instead of the S-N curve of the unnotched specimen).
prediction of fatigue life: Case C: Component related Haigh diagram (the local rainflow matrix
S-N curve of the component on the basis of the local of strains (εa , εm , n) is initially calculated from the rainflow
stresses/strains, including the influences of the geometry, matrix of loads (MT,a , MT,m , n), the cyclic stress-strain curve
material, surface condition and residual stresses on fatigue determined using non-hardened, unnotched specimens
behavior. subjected to axial loading was used to find the local rainflow
Actual local material states at the locations that are critical matrix of stresses (σa , σm , n), and the influence of mean stress
for failure. is incorporated using a Haigh diagram).
Component and material related Haigh diagram for the Case D: Modification of S-N curve using Liu-Zenner procedure.

256
Case E: Using PJ parameter based on fracture mechanical considerations
Author(s): Sonsino, Kaufmann, Foth and Jauch (1997) Prediction Procedure

critical location. that assume a micro-crack from the very beginning of the fatigue life. The
Cyclic stress-strain curves for the particular material states local strain at the crack tip is calculated as the amplitude of the effective
at the critical failure locations. cyclic J-integral.
Local (equivalent) strains or stresses in the critical area.
Loads and local parameters in the form of a rainflow
matrix.
Real damage sums applicable for the type of spectrum,
geometry, stress and material state (surface, residual
stresses) at the locations that are critical for failures. The
real damage sum should refer to the method of fatigue
life prediction that was used. Also, a comparison must be
made between the service spectrum that was expected
and the spectrum for which the real damage sums were
derived.

257
Author(s): Sonsino, Kaufmann, and Grubišić (1997) Prediction Procedure

CASE A: Method based on rain-flow matrices; Amplitude


Article Title: Transferability of material data for the example of a randomly loaded forged transformation via the damage parameter PSWT and
truck stub axle a component and material related compensation of
the mean values using a Haigh diagram are
Component(s): Truck stub axle performed for the strain-time histories on the basis
of their respective rain-flow matrices.
Stress-Strain A1: Amplitude transformation with the damage
linear elastic FEA
Calculation: parameter PSWT;
A2: Amplitude transformation with component
Life Prediction related Haigh diagram.
Local stress-life
Method(s): CASE B: Method based on peak value sequences; Besides
introducing the rain-flow matrices, the peak value
Experiment sequences are introduced so that the sequence of
Constant and variable amplitude component tests
Verification: the loading can numerically be taken into account
cycle by cycle.
Author’s 1. The problem that arises before the fatigue life calculation is the B1: Amplitude transformation with the damage
Remarks: transferability of data predominantly determined for specimens on the parameter PSWT;
basis of nominal stress to components where neither a nominal stress nor B2: Amplitude transformation with the damage
a notch factor can be defined in most cases and where the dimensions parameter PJ.
often deviate from those of the specimens investigated. CASE C: Method based on level crossing and range pair
2. The preconditions for the transferability of material data are fulfilled only counting;
after consideration of: C1: Mean value formation between the counts
the same failure criterion (first detectable crack), of level crossings and range pairs;
the local equivalent stresses or strains and, C2: The original Miner damage calculation is
the maximum stressed/strained material volume. retained, but the reference S-N curve has the
3. In addition, differences between the materials, states of surfaces, surface mean value of the slopes of the S-N curve
layers and residual stress must also be accounted for. for the failure criteria total rupture and the
4. The local concept in its extended form is applicable, namely after taking S-N curve for crack propagation.
into account local stresses/strains, stress/strain gradients and the
maximum stressed/strained material volume.
5. If the maximum stressed/strained material volume cannot be calculated
in a relatively simple manner, it can be derived by approximation via FE
calculations using strain/stress gradients, and its influence on fatigue
strength can be evaluated.
6. The real damage sum, which is a function of material, component, load

258
Author(s): Sonsino, Kaufmann, and Grubišić (1997) Prediction Procedure

type and spectrum, is not known in most cases. This is a decisive


limitation to the transferability of real damage sums determined on
specimens. Available experience with real damage sums can therefore
only be utilized for a preliminary dimensioning within the framework of a
defined method (for particular methods, different empirical damage sums
are used).
7. When over-dimensioning is not acceptable, verification tests prior to
release are indispensable, in particular for vital components like stub
axles, which must never fail.

259
Author(s): Béranger, Bérard, and Vittori (1997) Prediction Procedure

Article Title: A fatigue life assessment methodology for automotive 1. The result of the static linear elastic FE calculation is used to generate the
components load cycle.
2. The Dang Van fatigue model ( τ (t ) + αp(t ) − τ o > 0 )was used with a material
Component(s): Renault Safrane suspension arm damage line, whose coefficients are τo=217 MPa and α=-1.38. They
correspond to the shot-peened material, for a life of 107 cycles, with a
Stress-Strain probability of failure of 50%. Several stress ratios are used to identify these
Elastic FEA
Calculation: coefficients. Therefore any loading cycle, with various stress ratios, can be
predicted with these coefficients using the Dang Van model.
Life Prediction 3. The result of the calculation is a value of a so called safety factor (Sf). This
Critical plane and hyper-sphere procedures of Dang Van
Method(s): factor can be defined as the normalized smallest distance of the loading path
in the (τ; p) Dang Van diagram to the damage line. When Sf is greater than 1
Experiment
Constant-amplitude load-control component tests the component is supposed to be safe. On the contrary when Sf of some
Verification:
elements becomes less than unity, failure may occur.
4. Several damage lines representative of the various surface finishes have been
Author’s 1. The minimum value of the safety factor on the part is
tested for the prediction of the safety factors associated to the fatigue life.
Remarks: located in the area where the failure occurred on the
test rig. Its value is 1.02. This result exhibits a very
good correlation between the experiment and the
fatigue life prediction since the theoretical value of 1
and the actual safety factor are very close.
2. It was shown that the safety factors obtained are
highly dependent on the material data. The best results
were derived from material data representative of the
exact surface finish of the part, namely shot-peened.

260
Author(s): Devlukia and Bargmann (1997) Prediction Procedure
Deterministic:
Article Title: Fatigue assessment of an automotive suspension component using The surface roughness parameters for the as-forged surface
deterministic and probabilistic approaches were found. These values, together with the tensile strength
of the material were used to predict the surface finish factor,
Component(s): Suspension arm Ksf .
This factor was used in Neuber’s rule to predict the
Stress-Strain reduction in fatigue lives. There, in place of the fatigue
Strain measurement during tests
Calculation: notch factor Kf, the term Kt/Ksf was used, Kt being the
geometric stress concentration factor.
Life Prediction The surface factor and the residual stress measured value
Strain-life (deterministic and probabilistic)
Method(s): were used in Morrow’s equation to predict their effects on
the fatigue performance.
Experiment Miner’s rule was used for damage summation.
Constant and variable amplitude component tests
Verification: Probabilistic:
In this analysis, both the surface finish factor and the
Author’s 1. Cumulative damage under variable amplitude loading sequence of residual stress were assumed to be independent random
Remarks: long duration on simple specimens is non-conservative by a factor variables. Experimental measurements carried out on a
of about 2 as compared to measured data. Possible reasons include batch of components established the normal density, mean
(a) the effect of small cycles which, in principle, requires non-linear value and standard deviation of the surface finish factor and
summation, (b) the mode of relaxation of residual stresses. the residual stress.
2. The prediction of component lives based on specimen data is The above probabilistic inputs were used to calculate the
conservative by a factor of about 2. A possible reason for this effect probability of failure based on the Coffin-Manson equation,
may be due to an approximate assumption for the stress state. the hysteresis loop equation and Neuber’s rule.
The residual stress effects were predicted using the
Goodman correction, and the surface roughness effects
were predicted substituting Kt/Ksf for Kf in Neuber’s rule.
The “CPFI-Complete Probability Fast Integration” method
was used to predict the probability of failure.
In the CPFI approach, the fatigue-life distribution of a
component, and hence the reliability, i.e. the probability that
the random lifetime N does not fall below a given value, is
always expressed in a multiple-integral closed form.

261
Author(s): Conle and Chu (1997) Prediction Procedure

Article Title: Fatigue analysis and the local stress–strain approach in complex vehicular Local stress-strain analysis options:
structures

Component(s): General

Stress-Strain
Analytical, elastic FEA + Neuber plasticity correction, elastic-plastic FEA
Calculation:

Life Prediction Total strain vs life;


Method(s): SWT parameter σmax εa vs life;
Morrow-Landgraf: replace σ'f by (σ'f-σo);
Von Mises’ equivalent stress vs life;
Browm-Miller γa + Kεa vs life;
Fatemi-Socie-Kurath γ(1+n σn,max/σy) vs life;
Chu’s modified SWT σmax εa + |τ|max γa vs life.

Experiment
Not applicable
Verification:

Author’s The following features are emphasized:


Remarks: 1. Material properties derived from standard strain life test results.
2. A three-dimensional stress–strain model, adapted from that of Mroz.
The model must be able to simulate reversed multiaxial stress/strain
paths.
3. Damage assessment must be flexible enough to handle several multiaxial
fatigue damage criteria. A critical plane search for the most damaging
direction is necessary.
4. A multiaxial Neuber type of plasticity correction method must be used
to translate the elastic local stress estimates into approximations that are
corrected for plastic stress–strain behavior.
5. Large finite element models of vehicular components tend to have many
load input points, and thus must utilize an elastic unit load analysis
combined with a superposition procedure of each load points service

262
history, to produce each element’s stress history. It is critical that the
Author(s): Conle and Chu (1997) Prediction Procedure

time base of the loading events be preserved.


Problem areas that require further research:
1. The simulation of the non-linear stress–strain or load deflection
behavior of elements such as tires and bushings.
2. Accounting for ‘small cycle’ effects caused by periodic overload
conditions in the loading.

263
Author(s): Taylor, Bologna and Bel Knani (2000) Prediction Procedure
Susmel and Taylor (2003)

Article Title: 1. Prediction of fatigue failure location on a component


using a critical distance method
2. Fatigue design in the presence of stress concentration

Component(s): Fiat Punto Wishbone suspension arm

Stress-Strain
Elastic FEA
Calculation:

Life Prediction Strain-life with elastic FEA results + Neuber


Method(s): plasticity correction
Critical distance method
1. It is proposed that in order for fatigue failure to occur, the stress level must
Experiment be high enough not only at the maximum stress point (the hot spot) but also
Variable amplitude uniaxial load-controlled component test
Verification: for some distance around the hot spot (the stress at one point, a given
distance from the hot spot or to the stress averaged over a line of given
Author’s 1. It was found that good predictions (within 20% of the length). These distances are material-dependent.
Remarks: experimental fatigue limit) could be obtained for a 2. Consider a central crack in an infinite plate subjected to a remote uniaxial
wide range of materials, notch geometries and loading
load Δσ. The stress ahead of the crack tip can be expressed, when r<<a, as:
conditions.
2. Some inaccuracies occurred in components if their
section thickness was small or if the FE mesh density
used for stress analysis was not sufficiently refined. 3. If this crack is loaded at its fatigue limit, then ΔK=ΔKth; therefore the stress
3. An interesting effect which arises from notch analysis range at the point r=L/2, θ=0 on the stress-distance curve is equal to the
is that for a component which contains more than one plain fatigue limit:
stress-concentration feature, the feature from which
fatigue failure occurs may not be the one which
creates the highest stress. For example in the case of 4. When ΔK=ΔKth, the average stress along a straight line of length 2L is equal
two notches, one of which is sharp but relatively to the plain fatigue limit:
shallow, and another which is blunt but deeper, the
highest stress may occur at the tip of the sharp notch,
but failure may occur preferentially from the blunt
notch. The critical distance method can predict this 5. Also the stress averaged over a semicircle of radius r=L, θˆ=+π/2 to -π/2,

264
behavior: since the sharp notch has a higher stress is approximately equal to the plain fatigue limit (in fact it is larger by about 10
Author(s): Taylor, Bologna and Bel Knani (2000) Prediction Procedure
Susmel and Taylor (2003)

gradient, the stress at the critical distance may be lower per cent):
than that for the blunt notch.
4. The approach was able to predict correctly the
location of fatigue failure, and also predicted the 6. The last 3 equations provide the basis for three critical distance methods for
fatigue life with an error of a factor of 1.85, equivalent estimating the fatigue limit of the cracked body: using the stress at a single
to an error of less than 6% on stress, which is within point, averaged along a line or averaged over an area.
acceptable tolerances.

265
Author(s): Tang, Ogarevic and Tsai (2001) Prediction Procedure
Software component structure:
Article Title: An integrated CAE environment for simulation-based durability and
reliability design

Component(s): General – lower control arm

Stress-Strain
quasi-static FEA
Calculation:

Life Prediction von Mises equivalent strain approach


Method(s): ASME Boiler and Pressure Vessel Code strain approach
tensile critical plane method
shear critical plane method

Experiment
None
Verification: The program runstream:
Author’s 1. The development of a simulation-based environment for
Remarks: durability and reliability analysis of a mechanical component is
implemented.
2. This environment carries out the life prediction and reliability
assessment of mechanical components based on duty cycle loads
or dynamic stress time histories. It involves multidiscipline, such
as geometry generation, structural analysis, structural dynamics,
fatigue life prediction, and reliability analysis.
3. The software structure of this environment consists of three
subworkspaces, computer servers, database and data server, and The dataflow and dependency of life prediction subworkspace:
a communication channel. Each subworkspace collects several
tools to perform an engineering activity, namely, dynamic stress
computation, fatigue life prediction, and reliability analysis.

266
Author(s): Kocabicak and Firat (2001) and Firat and Kocabicak (2004) Prediction Procedure

Article Title: 1. Numerical analysis of wheel cornering fatigue test


2. Analytical durability modeling and evaluation-complementary techniques for
physical testing of automotive components

Component(s): Passenger car wheel

Stress-Strain Elastic and elasto-plastic FEA using a linear or non-linear kinematic hardening
Calculation: material law with isotropic von Mises yield criterion

Life Prediction Strain-life:


Method(s): Strain amplitudes based on effective strain and stress measures
Critical plane concepts based on SWT and Fatemi-Socie parameters

Experiment
Constant and variable amplitude bi-axial load-controlled cornering fatigue test
Verification:

Author’s 1. Best correlations are obtained with critical plane parameters, while effective
Remarks: parameters performed relatively poor.
2. Fatigue test cycles predicted using Fatemi–Socie damage parameter based on
critical plane concepts are considerably close to test cycles for all wheel loads,
and falls in to a band within a factor of 3 to 5.
3. In order to investigate the trends predicted with increasing number of cycles,
the numerical simulation for the wheel failure location is conducted using
Fatemi–Socie parameter, and the variation of damage per cycle is determined
to be monotonically decreasing function, indicating that the estimation of
fatigue life using the damage predicted with the first cycle after monotonic
loading is an appropriate strategy resulting in an conservative results for this
particular case. The reduction in damage per cycle is determined to be
approximately 13%.
4. The critical plane parameters involving mean stress terms perform equally
similar predictions under non-proportional unbalanced loading case
considered here, and Fatemi–Socie and Smith–Watson–Topper models based
on critical plane concept constitute a pair of damage parameters applicable in
both cases within the margin of acceptable accuracy from engineering point
of view in the design of wheels with conformance to anti-fatigue

267
requirements in biaxial cornering tests.
Author(s): Savaidis (2001) Prediction Procedure

Article Title: Analysis of fatigue behavior of a vehicle axle Damage Parameters


steering arm based on local stresses and strains The SWT damage parameter: PSWT = (σ a + σ m )ε a E
and the one developed by means of elastic-plastic fracture mechanics on a semi-elliptical
Component(s): Vehicle steering arm
microcrack (PJ): P = ΔJ eff = 1.24 ⋅ Δσ eff + 1.02 ⋅ Δσ ⋅ ⎡Δε − Δσ eff ⎤
2

eff ⎢
n′ E ⎥⎦
J eff
Stress-Strain Strain-time sequence measured at the failure a E ⎣
Calculation: critical location have been used.
Since the PJ-N curve is closely related to the crack growth law, its analytical description occurs
Life Prediction local stress-strain: according to a power-law function: PJm ⋅ N = Q for PJ>Pj,E, where the constant Q denotes the
Method(s): PSWT damage parameter fatigue life at PJ=1 N/mm2. The slope m is evaluated using regression analysis throughout the
PJ damage parameter experimental results. The subscript E indicated the endurance limit.
Experiment Variable amplitude uniaxial (bending) Technological Effects
Verification: component test
Surface roughness: Factor κ describes the decrease of the stress endurance limit due to surface
roughness
Author’s The Local Strain Approach allows an analysis
Remarks: of fatigue behavior of engineering components
under operational loading from the point of
view of material mechanics. In conjunction
with the two parameters, PSWT and PJ, it is able
to display and estimate influencing factors such the influence of the residual stress-strain state on the fatigue life prediction is considered by
as mean stresses, load sequence effects above adding the value of the residual strain εR to the local strain-time sequence εcyclic measured:
and below the endurance limit and
technological effects such as the roughness and
ε (t ) = ε cyclic (t ) + ε R
residual stresses explicitly.

268
Author(s): Bruder and Schön (2001) Prediction Procedure

Local Stress-Life Approach


Article Title: Durability analysis of carburized
components using a local approach
based on elastic stresses

Component(s): Gear wheel, Transmission shaft

Stress-Strain
Elastic FEA
Calculation:

Life Prediction Local stress-life


Method(s): Local strain-life

Experiment Constant and variable amplitude load-


Verification: controlled component tests

Author’s 1. The fatigue resistance of case-


Remarks: hardened components is strongly
influenced by the heat treatment and
technological factors. Also, the
statistical size effect on the
endurance limit and static limit load
needs to be taken into account.
2. The fatigue resistance can be
estimated well with both local stress
and local strain concepts.
3. Compared with the local stress
approach, the local strain approach
which is based on cyclic material data
needs additional input. Data, such as
residual stresses and surface factors
are required.
1. The influence of heat treatment on fatigue resistance of case-hardened components is taken into
account by using S-N curves obtained from “similar” components as reference. The reference S-
N curves are transformed to a σe,a-N curve for the critical location using the stress concentration

269
Author(s): Bruder and Schön (2001) Prediction Procedure

factor Kt (or a load influence factor c). To the σe,a-N curve a highly stressed volume IV and a
Weibull exponent kW are assigned.
2. Failure criterion is crack initiation of technical size in the case.
3. “Similarity” means that the components are equal in stress state at the critical locations for crack
initiation, base material, case hardening (type, case depth, etc.), residual stresses, and surface
condition.
4. The weakest link model is recommended to consider the size effect on the crack initiation life.
5. The parameter PSWT, e proved to be suitable to take the mean stress influence into account. Cyclic
loading may be limited by the static strength of a component, e.g. with fully tensile loading (load
ratio 0 < RS < 1). If the Weibull exponent for monotonic loading kW,m is known, the size effect
on static strength can also be estimated with the weakest link model.

270
Author(s): Haiba, Barton, Brooks and Levesley (2002) Prediction Procedure

Article Title: Review of life assessment techniques applied to dynamically loaded automotive components

Component(s): Lower control arm of a double wishbone suspension system

Stress-Strain The standard time domain approach that involves the application of the quasi-static stress
Calculation: analysis method.
The more sophisticated time domain approach that involves the application of the
transient dynamics analysis method.
The frequency domain approach that involves the application of the harmonic stress
analysis method.

Life Prediction The standard time domain approach that involves stress or strain cycle counting, damage
Method(s): prediction, and finally life estimation. The stress histories required for the above processes
can be obtained from either a quasi-static or transient stress analysis.
The frequency domain approach that involves using the power spectral density of the
component stress/ strain histories, estimating the expected number of cycles at several pre-
specified stress ranges, and finally predicting the life. The stress histories required prior to
the application of this approach can be obtained from a harmonic stress analysis.
Experiment
None
Verification:

Author’s 1. The reference life assessment strategy that uses transient stress analysis and time domain life
Remarks: assessment is very expensive. Consequently, it cannot be used as a part of an optimization
algorithm based on fatigue life.
2. The strategy that involves quasi-static stress analysis and time domain life assessment is very
efficient. The minimum life predicted using this strategy is very accurate when compared with
that obtained using the reference strategy. Additionally, this strategy has the capability to
predict nodal life distribution that is identical to the distribution obtained using the reference
strategy when considering just the design regions of the component under consideration.
Consequently, this strategy can be used as a part of an optimization algorithm based on fatigue
life.
3. The life assessment strategy that uses harmonic stress analysis and frequency domain life
assessment cannot be used to predict reliable values of minimum life and the results obtained
using this strategy cannot be used within an optimization algorithm based on fatigue life.

271
272

Appendix B - Life Prediction Calculation Details


The details of life prediction calculations for the forged steel (FS) and cast aluminum

(CA) steering knuckles for each life prediction model used and for the highest load level of

the component testing are provided here.

I Nominal S-N Approach for FS Steering knuckle


Applied Loads and moments are:

Pmax = 1100 lbf = 4893 N Mmax = 1100 * 12.18 = 4893 *0.309 lbf.in= 1513 N.m

Pa = 525 lbf = 2335 N Ma = 525 * 12.18 = 2335 *0.309 lbf.in= 722 N.m

Pmin = 50 lbf = 222 N Mmin = 50 * 12.18 = 222 *0.309 lbf.in= 69 N.m

Pm = 575 lbf = 2558 N Mm = 575 * 12.18 = 2558 *0.309 lbf.in= 791 N.m

Perform nonlinear finite element analysis (FEA) at a cross section 0.5 in above spindle second step

fillet:

Apply Pmax → find Smax = 568.1 MPa & apply Pmin → find Smin = 33.0 MPa

Then Sa = 267.5 MPa & Sm = 300.6 MPa

Kt = 2.2 from stress concentration charts (e.g. see Stephens et al., 2000) for step shaft under bending.

Find Kf :

1.8
Kt −1 ⎛ 2070 ⎞
⎜ S ⎟⎟
K f = 1+ where a = 0.0254 ⎜ in mm with Su in MPa and r = 0.05 in = 1.27 mm,
1+ a r ⎝ u ⎠

→ Kf = 2.1

Find S-N line including Kf effect at 106 cycles:

2Nf SNf

1 σ'f = 1156.8 MPa

2.00E+06 Sf/Kf = 352/2.1 = 168.4 MPa


273

→ B = - 0.133 → S Nf = 1156.8 (2 N f ) -0.133

Use modified Goodman equation to account for mean stress and find life:

Sa Sm 267.5 300.6
+ =1 → + = 1 → SNf = 422.0 MPa
S Nf S u S Nf 821.1

S Nf = σ ′f (2 N f ) → S Nf = 1156.8 (2 N f ) -0.133 → Nf = 1000 cycles


B

Use Gerber equation to account for mean stress and find life:

2
Sa ⎛ Sm ⎞
2
267.5 ⎛ 300.6 ⎞
+ ⎜⎜ ⎟⎟ = 1 → +⎜ ⎟ = 1 → SNf = 308.9 MPa
S Nf ⎝ S u ⎠ S Nf ⎝ 821.1 ⎠

S Nf = σ ′f (2 N f ) → S Nf = 1156.8 (2 N f ) -0.133 → Nf = 10400 cycles


B

II ε-N Approach using Nominal Stresses for FS Steering


knuckle
Using Neuber’s Rule without Generalization

Neuber’s Rule in its original form applies to small-scale as well as large-scale yielding:

σε = SeK tq2

where S & e are nonlinear nominal stress and strain and Ktq is based on Hoffmann-Seeger’s multiaxial

version of Neuber’s Rule for local multiaxial state of stress where elastic FEA results are available. To

find Ktq:

1
K tq = K t [(1 − a e ) 2 + (1 − be ) 2 + (a e − be ) 2 ]
2

σ e2 and be = σ e3
ae =
σ e1 = 0.24 σ e1 = 0.06

σe1 , σe2 and σe3 are the principal stresses from elastic FEA solution.

→ Ktq = 1.9

The 2nd equation to be solved with the Neuber’s rule is the material law:
274

1
σ⎛σ ⎞ n′
ε = +⎜ ⎟
E ⎝ K′ ⎠

For maximum load:

1 1
S ⎛S ⎞ n′
568.1 ⎛ 568.1 ⎞ 0.137
Smax = 568.1 MPa → emax = max + ⎜ max ⎟ = +⎜ ⎟ = 0.00565
E ⎝ K′ ⎠ 201500 ⎝ 1269 ⎠

Then the equations are:

σ maxε max = S max emax K tq2 = 568.1× 0.00565 × 1.9 2 = 11.6

1
σ max ⎛σ ⎞ 0.137
ε max = + ⎜ max ⎟
201500 ⎝ 1269 ⎠

Solving the above two equations simultaneously with a numerical method (MATLAB here):

σ max = 700.6 MPa

ε max = 0.0166

For unloading:

Sa = 267.5 MPa → ΔS = 535 MPa →

1 1
ΔS ⎛ ΔS ⎞ n′
535.0 ⎛ 535.0 ⎞ 0.137
Δe = + 2⎜ ⎟ = + 2⎜ ⎟ = 0.00268
E ⎝ 2K ′ ⎠ 201500 ⎝ 2 × 1269 ⎠

Then the equations are:

ΔσΔε = ΔSΔeK tq2 = 535.0 × 0.00268 × 1.9 2 = 5.2

1
Δσ ⎛ Δσ ⎞ 0.137
Δε = + 2⎜ ⎟
201500 ⎝ 2 × 1269 ⎠

Solving the above two equations simultaneously with a numerical method (MATLAB here):

Δσ = 912.0 MPa

Δε = 0.0057

The strain amplitude is equal to ε a = 0.00285

Use Smith-Watson-Topper (SWT) equation to calculate life:


275

σ max ε a E = (σ ′f ) 2 (2 N f ) 2b + σ ′f ε ′f E (2 N f ) b+ c

700.6 × 0.00285 × 201500 = 1156.8 2 (2 N f ) 2×−0.082 + 1156.8 × 3.0315 × 201500 (2 N f ) −0.082−0.7912

→ Nf = 8800 cycles

Using Seeger-Heuler’s Generalized Application of Neuber’s Rule


1
Ktq2 S 2 Ee∗ σ 2 ⎛σ ⎞ n′
= +σ⎜ ⎟
⎝ K′ ⎠

E S E

Here S is the elastically calculated nominal stress:

28
2335 × 309.4 ×
P M c 2335 2 = 340.3 MPa
Sa = a + a = +
A I π 2 π
28 284
4 64

Similarly for Pm = 2558 N → Sm = 372.7 MPa

Therefore: Smax = Sm + Sa = 713.0 MPa

S* is the modified nominal stress:

P
S* = Sy
Pp

where Pp is the plastic limit load for elastic-perfectly plastic material at the cross section remote from

the notch. Here, the appropriate version would be:

M
S* = Sy
Mp

and Mp = 2028 N.m (equivalent to 1470 lbf load at the actuator).

Thus:

M 1515
S* = Sy = ⋅ 556.2 = 415.5 MPa
Mp 2028

1 1
S∗ ⎛ S∗ ⎞ n′
415.5 ⎛ 415.5 ⎞ 0.137

e = +⎜ ⎟ = +⎜ ⎟ = 0.00235
E ⎜⎝ K ′ ⎟⎠ 201500 ⎝ 1269 ⎠

Ee∗
→ = 1.14 > 1
S∗
276

So the Seeger-Heuler model recommends the generalized application.

For maximum load:

Smax = 713.0 MPa

Then the equations are:

1
σ max ⎛σ
1
K tq2 S max Ee∗ σ max ⎛σ ⎞
2

n′ 2
2
1.9 2 713.0 2 0.137
= + σ max ⎜ max ⎟ → 1.14 = + σ max ⎜ max ⎟ →
E S∗ E ⎝ K′ ⎠ 201500 201500 ⎝ 1269 ⎠

σ max = 689.2 MPa

1
σ max
⎛σ ⎞ 0.137
ε max = + ⎜ max ⎟ = 0.0151
201500 ⎝ 1269 ⎠

For unloading the plastic limit moment was found to be 2755 N.m (obtained from FEA for EPP

unloading material at a cross section remote from the notch). Thus:

ΔM 1445
ΔS * = Sy = ⋅ 556.2 = 396.3 MPa
ΔM p 2028

1 1
ΔS ∗ ⎛ ΔS ∗ ⎞ 0.137
396.2 ⎛ 396.2 ⎞ 0.137
Δε =∗
+ 2⎜⎜ ⎟ = + 2⎜ ⎟ = 0.00197
201500 ⎝ 2 × 1269 ⎟⎠ 201500 ⎝ 2 × 1269 ⎠

EΔe ∗
→ =1
ΔS ∗

Therefore Seeger-Heuler’s model is not needed.

ΔS = 680.6 MPa

Then the equations are:

1
K tq2 ΔS 2
1
Δσ 2 ⎛ Δσ ⎞ n′ 1.9 2 680.6 2 Δσ 2 ⎛ Δσ ⎞ 0.137
= + 2Δσ ⎜ ⎟ → = + 2Δσ ⎜ ⎟
E E ⎝ 2K ′ ⎠ 201500 201500 ⎝ 2 × 1269 ⎠

→ Δσ = 1034.9 MPa

1
Δσ ⎛ Δσ ⎞ 0.137
Δε = + 2⎜ ⎟ = 0.008
201500 ⎝ 2 × 1269 ⎠

The strain amplitude is equal to: ε a = 0.004


277

Use SWT equation to calculate life:

σ max ε a E = (σ ′f ) 2 (2 N f ) 2b + σ ′f ε ′f E (2 N f ) b+ c

689.2 × 0.004 × 201500 = 1156.8 2 (2 N f ) 2×−0.082 + 1156.8 × 3.0315 × 201500 (2 N f ) −0.082−0.7912

→ Nf = 4500 cycles

III Local σ-N Approach using Nonlinear FEA Results for FS


Steering knuckle
Perform nonlinear FEA and obtain results (von Mises) at the node located at the transition of fillet

and spindle second step shaft:

Apply Pmax = 4893 N→ find σmax = 603.3 MPa & εmax = 0.0073

Apply ΔP = 4670 N→ find Δσ = 867.4 MPa & Δε = 0.0049

Therefore:

σa = 433.7 MPa εa = 0.00246

σm = 169.6 MPa εm = 0.0049

σmin = -264.1 MPa εmin = 0.00241

Use modified Goodman equation to account for mean stress and calculate life:

σa σm 433.7 169.6
+ =1 → + = 1 → σNf = 546.6 MPa
σ Nf Su σ Nf 821.1

σ Nf = σ ′f (2 N f ) b → 546.6 = 1156.8 (2 N f ) → Nf = 4600 cycles


- 0.082

Use Gerber equation to account for mean stress and calculate life:

2
σ a ⎛σ m ⎞
2
433.7 ⎛ 169.6 ⎞
+ ⎜⎜ ⎟⎟ = 1 → +⎜ ⎟ = 1 → σNf = 453.0 MPa
σ Nf ⎝ S u ⎠ σ Nf ⎝ 821.1 ⎠

σ Nf = σ ′f (2 N f ) b → 453.0 = 1156.8 (2 N f ) → Nf = 46100 cycles


-0.082

Use the stress-life version of SWT model to account for mean stress and calculate life:

σ maxσ a= σ ′f 2 (2 N f ) 2b
→ 603.3 × 433.7 = 1156.8 2 (2 N f ) 2× −0.082
→ Nf = 10400 cycles
278

IV Local σ-N Approach using Linear FEA Results and


Neuber-Correction for FS Steering knuckle
Perform linear FEA and obtain result (von Mises) at the node located at the transition of fillet and

spindle second step shaft:

Apply Pmax = 4893 N → find eσmax = 1079.8 MPa & eεmax = 0.0046

Solve Neuber’s correction to obtain elastic-plastic stress and strain along with the material law:

σeq εeq = eσeq eεeq → σeq εeq = 4.97

1 1
σ eq
⎛ σ eq ⎞ n′
→ε
σ eq ⎛ σ eq ⎞ 0.137
ε eq
= + ⎜⎜ ⎟ eq
= + ⎜⎜ ⎟
E ⎝ K ′ ⎟⎠ 201500 ⎝ 1269.5 ⎟⎠

to find:

σmax = 614.5 MPa & εmax = 0.0081

Apply ΔP = 4670 N→ find eΔσ = 1030.8 MPa & eΔε = 0.0044

Solve Neuber’s correction to obtain elastic-plastic stress and strain along with the material law for

unloading:

Δσeq Δεeq = eΔσeq eΔεeq → Δσeq Δεeq = 4.54

1 1
Δσ eq ⎛ Δσ eq ⎞ n′
Δσ eq ⎛ Δσ eq ⎞ 0.137
Δε eq
= + 2⎜⎜ ⎟⎟ → Δε eq
= + 2⎜⎜ ⎟
E ⎝ 2K ′ ⎠ 201500 ⎝ 2 × 1269.5 ⎟⎠

to find:

Δσ = 874.5 MPa & Δε = 0.0052

Therefore:

σa = 437.3 MPa εa = 0.0026

σm = 177.3 MPa εm = 0.0055

σmin = -260.0 MPa εmin = 0.0029

Use modified Goodman equation to account for mean stress and find life:
279

σa σm 437.3 177.3
+ =1 → + = 1 → σNf = 557.6 MPa
σ Nf Su σ Nf 821.1

σ Nf = σ ′f (2 N f ) b → 557.6 = 1156.8 (2 N f ) → Nf = 3600 cycles


-0.082

Use Gerber equation to account for mean stress and find life:

2
σ a ⎛σ m ⎞
2
437.3 ⎛ 177.3 ⎞
+ ⎜⎜ ⎟⎟ = 1 → +⎜ ⎟ = 1 → σNf = 458.6 MPa
σ Nf ⎝ S u ⎠ σ Nf ⎝ 821.1 ⎠

σ Nf = σ ′f (2 N f ) b → 458.6 = 1156.8 (2 N f ) -0.082 → Nf = 39700 cycles

Use the stress-life version of SWT model to account for mean stress and calculate life:

σ maxσ a= σ ′f 2 (2 N f ) 2b
→ 614.5 × 437.3 = 1156.82 (2 N f ) 2×−0.082 → Nf = 8900 cycles

V Local σ-N Approach using Nonlinear FEA Results for CA


Steering knuckle
With von Mises Stress and Strain

Perform nonlinear FEA and obtain results (von Mises) at the node located at 45º angle of hub bolt

hole (the failure location):

Apply Pmax = 6230 N→ find σmax = 296.5 MPa & εmax = 0.0067

Apply ΔP = 6005 N→ find Δσ = 430.9 MPa & Δε = 0.00489

Therefore:

σa = 215.5 MPa εa = 0.0024

σm = 81.0 MPa εm = 0.0043

σmin = -134.5 MPa εmin = 0.0018

Use modified Goodman equation to account for mean stress and find life:

σa σm 215.5 81.0
+ =1 → + = 1 → σNf = 294.2 MPa
σ Nf Su σ Nf 302

σ Nf = σ ′f (2 N f ) b → 294.2 = 666 (2 N f ) -0.117 → Nf = 550 cycles


280

Use Gerber equation to account for mean stress and find life:

2
σ a ⎛σ m ⎞
2
215.5 ⎛ 81.0 ⎞
+ ⎜⎜ ⎟⎟ = 1 → +⎜ ⎟ = 1 → σNf = 232.1 MPa
σ Nf ⎝ S u ⎠ σ Nf ⎝ 302 ⎠

σ Nf = σ ′f (2 N f ) b → 232.1 = 666 (2 N f ) -0.117 → Nf = 4200 cycles

Use the stress-life version of SWT model to account for mean stress and calculate life:

σ maxσ a= σ ′f 2 (2 N f ) 2b
→ 296.5 × 215.5 = 6662 (2 N f ) 2×−0.117 → Nf = 2000 cycles

With Maximum Principal Stress and Strain

Perform nonlinear FEA and obtain results (maximum principal) at the node located at 45º angle of

hub bolt hole (the failure location):

Apply Pmax = 6230 N→ find σmax = 317.2 MPa & εmax = 0.0072

Apply ΔP = 6005 N→ find Δσ = 450.9 MPa & Δε = 0.0056

Therefore:

σa = 225.5 MPa εa = 0.0028

σm = 91.7 MPa εm = 0.0044

σmin = -133.8 MPa εmin = 0.0016

Use modified Goodman equation to account for mean stress and find life:

σa σm 225.5 91.7
+ =1 → + = 1 → σNf = 323.5 MPa
σ Nf Su σ Nf 302

σ Nf = σ ′f (2 N f ) b → 323.5 = 666 (2 N f ) → Nf = 240 cycles


-0.117

Use Gerber equation to account for mean stress and find life:

2
σ a ⎛σ m ⎞
2
225.5 ⎛ 91.7 ⎞
+ ⎜⎜ ⎟⎟ = 1 → +⎜ ⎟ = 1 → σNf = 248.3 MPa
σ Nf ⎝ S u ⎠ σ Nf ⎝ 302 ⎠

σ Nf = σ ′f (2 N f ) b → 248.3 = 666 (2 N f ) → Nf = 2350 cycles


-0.117

Use the stress-life version of SWT model to account for mean stress and calculate life:
281

σ max σ a= σ ′f 2 (2 N f ) 2b
→ 317.2 × 225.5 = 666 2 (2 N f ) 2×−0.117 → Nf = 1250 cycles

VI Local σ-N Approach using Linear FEA Results and


Neuber-Correction for CA Steering knuckle
Perform linear FEA and obtain results (von Mises) at the node located at 45º angle of hub bolt hole

(the failure location):

Apply Pmax = 6230 N→ find eσmax = 446.1 MPa & eεmax = 0.0051

Solve Neuber’s correction to obtain elastic-plastic stress and strain along with the material law:

σeq εeq = eσeq eεeq → σeq εeq = 2.28

1 1
σ eq
⎛ σ eq ⎞ n′
σ eq ⎛ σ eq ⎞ 0.137
ε eq
= + ⎜⎜ ⎟ → ε eq
= + ⎜⎜ ⎟
E ⎝ K ′ ⎟⎠ 201500 ⎝ 1269.5 ⎟⎠

to find:

σmax = 302.1 MPa & εmax = 0.0075

Apply ΔP = 6005 N→ find eΔσ = 430.9 MPa & eΔε = 0.0049

Solve Neuber’s correction to obtain elastic-plastic stress and strain along with the material law for

unloading:

Δσeq Δεeq = eΔσeq eΔεeq → Δσeq Δεeq = 2.11

1 1
Δσ eq ⎛ Δσ eq ⎞ n′
→ Δε eq = Δσ
eq
⎛ Δσ eq ⎞ 0.1372
Δε eq
= + 2⎜⎜ ⎟⎟ + 2⎜⎜ ⎟
E ⎝ 2K ′ ⎠ 201500 ⎝ 2 × 1269.5 ⎟⎠

to find:

Δσ = 405.6 MPa & Δε = 0.0052

Therefore:

σa = 202.8 MPa εa = 0.0026

σm = 99.3 MPa εm = 0.0049

σmin = -103.5 MPa εmin = 0.0023


282

Use modified Goodman equation to account for mean stress and find life:

σa σm 202.8 99.3
+ =1 → + = 1 → σNf = 301.8 MPa
σ Nf Su σ Nf 302

σ Nf = σ ′f (2 N f ) b → 301.8 = 666 (2 N f ) → Nf = 450 cycles


-0.117

Use Gerber equation to account for mean stress and find life:

2
σ a ⎛σ m ⎞
2
202.8 ⎛ 99.3 ⎞
+ ⎜⎜ ⎟⎟ = 1 → +⎜ ⎟ = 1 → σNf = 227.3 MPa
σ Nf ⎝ S u ⎠ σ Nf ⎝ 302 ⎠

σ Nf = σ ′f (2 N f ) b → 227.3 = 666 (2 N f ) → Nf = 5000 cycles


-0.117

Use the stress-life version of SWT model to account for mean stress and calculate life:

σ max σ a= σ ′f 2 (2 N f ) 2b
→ 302.1 × 202.8 = 666 2 (2 N f ) 2×−0.117
→ Nf = 2400 cycles

VII ε-N Approach using Nonlinear FEA Results for FS


Steering knuckle
See local stresses and strains in local σ-N approach using nonlinear FEA results for the FS steering

knuckle.

Use Morrow equation to account for mean stress and calculate life:

σ ′f − σ m
εa = (2 N ) f
b
+ ε ′f (2 N f ) c

1156.8 − 169.6
0.00246 =
201500
(2 N f ) −0.082+ 3.0315(2 N f ) −0.7912 → Nf = 31700 cycles
Use Manson and Halford equation to account for mean stress and calculate life:

c
σ ′f − σ m ⎛σ ′ −σm ⎞
(2 N ) (2 N )
b

εa = b
+ ε ′f ⎜ f ⎟ c

E
f ⎜ σ′ ⎟ f
⎝ f ⎠

−0.7912
1156.8 − 169.6 ⎛ 1156.8 − 169.6 ⎞
(2 N f ) − 0.082 + 3.0315⎜ (2 N )
− 0.082
− 0.7912
0.00246 = ⎟ f
201500 ⎝ 1156.8 ⎠

→ Nf = 9400 cycles
283

Use SWT equation to account for mean stress and calculate life:

σ max ε a E = (σ ′f ) 2 ( 2 N f ) 2b + σ ′f ε ′f E ( 2 N f ) b + c

603.3 × 0.00246 × 201500 = 1156.82 ( 2 N f )2×−0.082 + 1156.8 × 3.0315 × 201500 ( 2 N f )−0.082−0.7912

→ Nf = 20900 cycles

VIII ε-N Approach using Linear FEA Results and Neuber-


Correction for FS Steering knuckle
See local stresses and strains in local σ-N approach using linear FEA results for the FS steering

knuckle.

Use Morrow equation to account for mean stress and calculate life:

σ ′f − σ m
εa = (2 N ) f
b
+ ε ′f (2 N f ) c

1156.8 − 177.3
0.0026 =
201500
(2 N f ) − 0.082+ 3.0315(2 N f ) −0.7912 → Nf = 24300 cycles
Use Manson and Halford equation to account for mean stress and calculate life:

c
σ ′f − σ m ⎛σ′ −σm ⎞
(2 N ) (2 N )
b

εa = b
+ ε ′f ⎜ f ⎟ c

E
f ⎜ σ′ ⎟ f
⎝ f ⎠

−0.7912
1156.8 − 177.3 ⎛ 1156.8 − 177.3 ⎞
(2 N f ) − 0.082 + 3.0315⎜ (2 N )
− 0.082
− 0.7912
0.0026 = ⎟ f
201500 ⎝ 1156.8 ⎠

→ Nf = 6200 cycles

Use SWT equation to account for mean stress and calculate life:

σ max ε a E = (σ ′f ) 2 (2 N f ) 2b + σ ′f ε ′f E (2 N f ) b + c

614.5 × 0.0026 × 201500 = 1156.82 ( 2 N f )2×−0.082 + 1156.8 × 3.0315 × 201500 ( 2 N f )−0.082−0.7912

→ Nf = 16700 cycles
284

IX ε-N Approach using Nonlinear FEA Results for CA


Steering knuckle
With von Mises Stress and Strain

See local stresses and strains in local σ-N approach using nonlinear FEA with von Mises stress and

strain for the CA steering knuckle.

Use Morrow equation to account for mean stress and calculate life:

σ ′f − σ m
εa = (2 N ) f
b
+ ε ′f (2 N f ) c

666.0 − 81.0
0.0024 = (2 N f ) − 0.117 + 0.0944(2 N f ) − 0.6095 → Nf = 16100 cycles
78100

Use Manson and Halford equation to account for mean stress and calculate life:

c
σ ′f − σ m ⎛σ′ −σm ⎞
(2 N ) (2 N )
b

εa = b
+ ε ′f ⎜ f ⎟ c

E
f ⎜ σ′ ⎟ f
⎝ f ⎠
−0.6095
666.0 − 81.0 ⎛ 666.0 − 81.0 ⎞
(2 N f ) (2 N )
− 0.117
−0.117 −0.7912
0.0024 = + 0.0944⎜ ⎟ f
78100 ⎝ 666.0 ⎠

→ Nf = 12400 cycles

Use SWT equation to account for mean stress and calculate life:

σ max ε a E = (σ ′f ) 2 (2 N f ) 2b + σ ′f ε ′f E (2 N f ) b+ c

296.5 × 0.0024 × 78100 = 666.02 ( 2 N f )2×−0.117 + 666.0 × 0.0944 × 78100 ( 2 N f )−0.117−0.6095

→ Nf = 5750 cycles

With Maximum Principal Stress and Strain

See local stresses and strains in local σ-N approach using nonlinear FEA with maximum principal

stress and strain for the CA steering knuckle.

Use Morrow equation to account for mean stress and calculate life:

σ ′f − σ m
εa = (2 N ) f
b
+ ε ′f (2 N f ) c

E
285

666.0 − 91.7
0.0028 =
78100
(2 N f ) − 0.117 + 0.0944(2 N f ) − 0.6095 → Nf = 5600 cycles

Use Manson and Halford equation to account for mean stress and calculate life:

c
σ ′f − σ m ⎛σ′ −σm ⎞
(2 N ) (2 N )
b

εa = b
+ ε ′f ⎜ f ⎟ c

E
f ⎜ σ′ ⎟ f
⎝ f ⎠
−0.6095
666.0 − 91.7
(2 N f ) −0.117+ 0.0944⎛⎜ 666.0 − 91.7 ⎞⎟ (2 N )
− 0.117
− 0.7912
0.0028 = f
78100 ⎝ 666.0 ⎠

→ Nf = 3700 cycles

Use SWT equation to account for mean stress and calculate life:

σ max ε a E = (σ ′f ) 2 (2 N f ) 2b + σ ′f ε ′f E (2 N f ) b +c

317.2 × 0.0028 × 78100 = 666.02 ( 2 N f )2×−0.117 + 666.0 × 0.0944 × 78100 ( 2 N f )−0.117−0.6095

→ Nf = 2700 cycles

X ε-N Approach using Linear FEA Results and Neuber-


Correction for CA Steering knuckle
See local stresses and strains in local σ-N approach using linear FEA results with von Mises stress

and strain for the CA steering knuckle.

Use Morrow equation to account for mean stress and calculate life:

σ ′f − σ m
εa = (2 N ) f
b
+ ε ′f (2 N f ) c

666.0 − 99.3
0.0026 = (2 N f ) − 0.117 + 0.0944(2 N f ) − 0.6095 → Nf = 8100 cycles
78100

Use Manson and Halford equation to account for mean stress and calculate life:

c
σ ′f − σ m ⎛ ⎞
(2 N f ) b + ε ′f ⎜⎜ σ ′f σ−′σ m ⎟⎟ (2 N )
b

εa = f
c

E ⎝ f ⎠
286

−0.6095
666.0 − 99.3
(2 N f ) −0.117+ 0.0944⎛⎜ 666.0 − 99.3 ⎞⎟ (2 N )
− 0.117
− 0.7912
0.0026 = f
78100 ⎝ 666.0 ⎠

→ Nf = 5400 cycles

Use SWT equation to account for mean stress and calculate life:

σ max ε a E = (σ ′f ) 2 (2 N f ) 2b + σ ′f ε ′f E (2 N f ) b+ c

302.1× 0.0024 × 78100 = 666.02 ( 2 N f )2×−0.117 + 666.0 × 0.0944 × 78100 ( 2 N f )−0.117−0.6095

→ Nf = 4050 cycles

You might also like