MECHANICAL METALLURGY
Metallurgy and Metallurgical Engineering Series
Robert
F.
Mehl,
Consulting Editor
Michael
B. Bever, Associate Consulting Editor
Barrett Structure of Metals BiRCHENALL Physical Metallurgy Bridgman Studies in Large Plastic Flow and Fracture Briggs The Metallurgy of Steel Castings
•
• •
•
Butts
•
Metallurgical Problems
•
Darken and Gurry
Dieter
•
Physical Chemistry of Metals
Mechanical Metallurgy
Gaudin Flotation Hansen Constitution of Binary Alloys Kehl The Principles of Metallographic Laboratory Rhines Phase Diagrams in Metallurgy Seitz The Physics of Metals
•
•
•
Practice
•
•
Smith
•
Properties of Metals at Elevated Temperatures
•
Williams and Homerberg
Principles of Metallography
VSL
Mechanical Metallurgy
GEORGE
E.
DIETER, JR.
Professor and
Head
of
Department of Metallurgical
Engineering
Drexel Institute of Technologij Philadelphia
4,
Pa.
McGRAWHILL BOOK COMPANY
New York
Toronto
London
1961
MECHANICAL METALLURGY
Copyright 1961 by the McGrawHill Book Company, Inc. Printed United States of America. All rights reserved. This book, or parts thereof, may not be reproduced in any form without permission of the publishers. Library of Congress Catalog Number 6111385
in the
©
12 13 14 15 16
MAMM  7 5 4 3
ISBN
070168903
PREFACE
Mechanical metallurgy
cisely defined area,
it
is
the area of knowledge which deals with the Since
it is
behavior and response of metals to applied forces.
will
not a pre
mean
different things to different persons.
To
some
it
will
mean mechanical
still
properties of metals or mechanical testing,
others
may consider the field restricted to the plastic working and
shaping
of metals, while
others confine their interests to the more theoretical
field, which merge with metal physics and physical metalanother group may consider that mechanical metallurgy is In writclosely allied with applied mathematics and applied mechanics. ing this book an attempt has been made to cover, in some measure, this
aspects of the
lurgy.
Still
great diversity of interests.
The
objective has been to include the entire
scope of mechanical metallurgy in one fairly comprehensive volume.
The book has been divided into four parts. Part One, Mechanical Fundamentals, presents the mathematical framework for many of the chapters which follow. The concepts of combined stress and strain are reviewed and extended into three dimensions. Detailed consideration of the theories of yielding and an introduction to the concepts of plasticity are given. No attempt is made to carry the topics in Part One to Instead, the degree of completion required for original problem solving. the purpose is to acquaint metallurgically trained persons with the mathematical language encountered in some areas of mechanical metallurgy. Part Two, Metallurgical Fundamentals, deals with the structural aspects of plastic deformation and fracture. Emphasis is on the atomistics of flow and fracture and the way in which metallurgical structure affects these processes. The concept of the dislocation is introduced early in Part Two and is used throughout to provide qualitative explanations for such phenomena as strain hardening, the yield point, dispersed phase hardening, and fracture. A more mathematical treatment of the properThe topics covered in ties of dislocations is given in a separate chapter. Part Two stem from physical metallurgy. However, most topics are discussed in greater detail and with a different emphasis than when they
are
first
lurgy.
covered in the usual undergraduate course in physical metalCertain topics that are more physical metallurgy than mechanical
VI
Prefe rreracc
metallurgy are included to provide continuity and the necessary background for readers who have not studied modern physical metallurgy. Part Three, Applications to Materials Testing, deals with the engineering aspects of the
common testing techniques of mechanical failure of are devoted to the tension, torsion, hardness, fatigue, Chapters metals. tests. Others take up the important subjects of impact and creep, statistical analysis of mechanicalproperty data. the and stresses residual is on emphasis placed the interpretation of the tests and Three Part In metallurgical variables on mechanical behavior rather of effect on the for conducting the It is assumed that the procedures the tests. on than in of these tests will be a concurrent laboraperformance covered actual Plastic Forming of or in separate course. Part Four, a course tory common mechanical for producing usethe processes with deals Metals, emphasis is the descriptive Little given aspects of shapes. to metal ful trips and can plant illustrated this best be covered since by subject, this Instead, the main attention is given to the mechanical and lectures. metallurgical factors which control each process such as forging, rolling,
extrusion, drawing,
and sheetmetal forming.
This book
is
written for the senior or firstyear graduate student in
metallurgical or mechanical engineering, as well as for practicing engi
neers in industry.
While most universities have instituted courses
in
mechanical metallurgy or mechanical properties, there is a great diversity in the material covered and in the background of the students taking Thus, for the present there can be nothing like a standthese courses. It is hoped that the breadth ardized textbook on mechanical metallurgy.
and scope
treatment
of this
book
will
provide material for these somewhat diverse
requirements.
It is further
hoped that the existence
of a
comprehensive
of the field of
mechanical metallurgy
will stimulate the develop
which cover the total subject. is intended for college seniors, graduate students, and book Since this it is expected to become a part of their professional engineers, practicing Although there has been no attempt to make this book a handlibrary. given to providing abundant references to been thought has book, some Therefore, more references are metallurgy. on mechanical the literature
ment
of courses
included than
References have been is usual in the ordinary textbook. beyond the scope of the book, or analyses out derivations point given to or detailed controversial on information further key provide the to to
to emphasize important papers which are worthy of further In addition, a bibliography of general references will be found A collection of problems is included at the at the end of each chapter. end of the volume. This is primarily for the use of the reader who is
points,
and
study.
engaged
in industry
and who
desires
some check on
his
comprehension of
the material.
Preface
vii
facts
The task of writing this book has been mainly one of sifting and sorting and information from the literature and the many excellent texts on
specialized aspects of this subject.
To
cover the breadth of material
book would require parts of over 15 standard texts and countless review articles and individual contributions. A conscientious effort has been made throughout to give credit to original sources. For the occasional oversights that may have developed during the "boilingfound
in this
process" the author offers his apologies. He is indebted to many authors and publishers who consented to the reproduction of illustrations. Credit is given in the captions of the illustrations.
down
Finally,
the author wishes to acknowledge the
in this
many
friends
who
advised him
A.
W.
work. Special mention should be given to Professor Grosvenor, Drexel Institute of Technology, Dr. G. T. Home,
J.
Carnegie Institute of Technology, Drs. T. C. Chilton,
H. Faupel,
W.
L. Phillips,
S.
W.
I.
Pollock,
of
and Dr. A.
Nemy
and J. T. Ransom of the du Pont Company, the ThompsonRamoWooldridge Corp.
George E. Dieter, Jr.
Two Dimensions (Plane Mohr's Circle of Stress— Two Dimensions State of Stress in Three Dimensions Mohr's Circle Three Dimensions 19 — 27. Generalized StressStrain Relationships 212.. 17. 19. 32. Spherical 215. Calculation of Stresses from Elastic Strains 211. 15. 33. Strain 3.CONTENTS Preface List of v Symbols Part xvii One. 22. The Flow Curve True Strain ... Concept Concept and the Types of Stress Strain and the Types of Strain 15 2. Scope of This Book Strength of Materials Basic Assumptions Elastic and Plastic Behavior — 14. StressStrain Relations . 16.. Stress 21. and Strain Relationships for Elastic Behavior 17 1^ Introduction Description of Stress at a Point 17 Stress) 23. Mechanical Fundamentals 3 1. Description of Strain at 28. a Point Measurement of Surface Strain 29.. 13. and Deviator Components Energy and Strain . 23 24 27 31 33 35 39 41 43 46 50 ^2 Elements of the Theory of Plasticity Introduction 54 54 ^^ 57 ix 31. Stress Concentration 214. Introduction 11. • • 210. 26. Brittle Behavior 3 5 6 7 8 9 10 13 What Constitutes Failure? of Stress of 18. Theory of Elasticity of Stress 213. State of Stress in 24. 25. 12. Average Stress and Strain Tensile Deformation of Ductile Metal Ductile vs.
K
Contents
34. Yielding Criteria for Ductile 35. 36.
Metals
Combined
Stress Tests
Octahedral Shear Stress and Shear Strain
37. Invariants of Stress
38. 39.
and Strain
Basis of the Theories of Plasticity
310.
311.
312.
Flow Theories Deformation Theories Twodimensional Plastic Flow Slipfield Theory
Part
— Plane Strain
58 62 65 66 67 69 72 73 74
Two.
Metallurgical Fundamentals
4. Plastic
41.
Deformation of Single Crystals
81
81
Introduction
42.
Concepts of Crystal Geometry
43. Lattice Defects 44.
45.
Deformation by
Slip
Slip in a Perfect Lattice
46. Slip 47.
by Dislocation Movement
Resolved Shear Stress for Slip
Critical
48. Testing of Single Crystals
49.
Deformation by Twinning Deformation Bands and Kink Bands Hardening of Single Crystals
410. Stacking Faults
411.
82 85 90 95 97 99 102 104 108 110
Ill
» . . .
412. Strain
5.
Plastic
Deformation of Polycrystalline Aggregates
.118
118 119 123 128 132 135 137 145 146 149 150 153 156
.
51. 52.
Introduction
53.
54.
Grain Boundaries and Deformation Lowangle Grain Boundaries Solidsolution Hardening
55. Yieldpoint 56. Strain 57. 58. 59.
Phenomenon
Aging
510. 511.
512.
513.
Strengthening from Secondphase Particles Hardening Due to Point Defects Strain Hardening and Cold Work Bauschinger Effect Preferred Orientation Annealing of Coldworked Metal Anneahng Textures
. .
6.
Dislocation Theory
.
o
1
58
158 158 162 164 169 169
171
61. 62.
Introduction Methods of Detecting Dislocations 63. Burgers Vector and the Dislocation Loop
64. Dislocations in 65. Dislocations in the 66.
the Facecentered Cubic Lattice Hexagonal Closepacked Lattice
Dislocations in the Bodycentered Cubic Lattice
67. Stress Field of
68.
a Dislocation Forces on Dislocations 69. Forces between Dislocations
174 175
Contents
610. Dislocation
xi
Climb
177 178 179 181
183 184 186
611. Jogs in Dislocations
612. Dislocation
613. Dislocation
614. Dislocation Sources 615.
616.
7.
—Foreignatom Interaction Multiplication of Dislocations — FrankRead Source
Dislocation Pileup
and Vacancy Interaction
Fracture
190
190 190 192 194 197 199 200 204
71.
Introduction 72. Types of Fracture in Metals 73. Theoretical Cohesive Strength of Metals 74. Griffith Theory of Brittle Fracture
75. 76.
Modifications of the Griffith Theory
Fracture of Single Crystals 77. Metallographic Aspects of Brittle Fracture 78. Dislocation Theories of Fracture
79.
710. Velocity of
Delayed Yielding Crack Propagation
...
209 210
211
711. Ductile Fracture 712. 713.
714. 715.
716.
Notch Effect in Fracture Concept of the Fracture Curve Classical Theory of the DuctiletoBrittle Transition Fracture under Combined Stresses Effect of High Hydrostatic Pressure on Fracture
213 215 216 218 219
8.
Internal Friction
221
221
81. 82.
Introduction
Phenomenological Description of Internal Friction
.
83. Anelasticity 84.
Relaxation Spectrum 85. Grainboundary Relaxation
86. 87. 88. 89.
The Snoek
Effect
Thermoelastic Internal Friction
Dislocation Damping Damping Capacity
222 224 227 227 229 229 230 232
Part Three.
9.
Applications to Materials Testing
The Tension Test
Engineering StressStrain Curve TruestressTruestrain Curve Instability in Tension Stress Distribution at the Neck Strain Distribution in the Tensile Specimen Effect of Strain Rate on Tensile Properties Effect of Temperature on Tensile Properties
237
237 243 248 250 252 254 256 258 260 262 269
91. 92. 93. 94. 95. 96.
97.
98. 99.
Combined Effect of Notch Tensile Test
Strain Rate and Temperature
910. Tensile Properties of Steels
911. Anisotropy of Tensile Properties
xii
Contents
10. The Torsion Test
101. Introduction
273
273 273 276 278 279
Mechanical Properties in Torsion 103. Torsional Stresses for Large Plastic Strains 104. Types of Torsion Failures 105. Torsion Test vs. Tension Test
102.
11. The Hardness Test
111. Introduction
112. Brinell 113.
282
282 283 284 286 287 289 290
291 292 293
Hardness Meyer Hardness
114. Analysis of Indentation
115. Relationship 116.
by a Spherical Indenter
117. 118.
119.
between Hardness and the Tensileflow Curve Vickers Hardness Rockwell Hardness Test Microhardness Tests
....
Hardnessconversion Relationships 1110. Hardness at Elevated Temperatures
12. Fatigue of Metals
121. Introduction 122. Stress Cycles
296
296 297 299 301 304 307 310 314 315 320 323 326 327 329 332
°
°
123.
The
SiV
124. Statistical
Curve Nature
of Fatigue
125. Structural Features of Fatigue
126. Theories of Fatigue 127. Efifect of Stress
128. Size Effect
Concentration on Fatigue
129. Surface Effects
and Fatigue
Stress
1210. Corrosion Fatigue
1211. Effect of
Mean
on Fatigue
Stresses
1212. Fatigue under 1213. Overstressing
Combined
and Understressing
1214. Effect of Metallurgical Variables
1215. Effect of
on Fatigue Properties Temperature on Fatigue
13. Creep and Stress Rupture
131.
132. 133. 134.
335
335 336 341 342 345 347 349 354 356 359 363 367 367
The Hightemperature Materials Problem The Creep Curve The Stressrupture Test
Deformation at Elevated Temperature Elevated Temperature 136. Theories of Lowtemperature Creep 137. Theories of Hightemperature Creep 138. Presentation of Engineering Creep Data 139. Prediction of Longtime Properties 1310. Hightemperature Alloys
135. Fracture at
1311. Effect of Metallurgical Variables
1312.
Creep under Combined Stresses
1313. Stress Relaxation
Contents
14.
Brittle Failure
xiii
and Impact Testing
370
370
371
141.
142. 143.
The
Problem Notchedbar Impact Tests Slowbend Tests
Brittlefailure
144. Specialized Tests for Transition 145.
146.
Temperature Significance of the Transition Temperature Metallurgical Factors Affecting Transition Temperature
147. Effect of Section Size
148.
149.
Notch Toughness
of Heattreated Steels
Temper Embrittlement
Hydrogen Embrittlement Flow and Fracture under Very Rapid Rates
of
1410. 1411.
Loading
375 377 379 381 384 385 387 388 390
15. Residual Stresses
151. Origin of Residual Stresses
152. Effects of Residual Stresses 153.
393
393 397 398 403 407 411 415 417
Mechanical Methods of Residualstress Measurement
154. Deflection
155. 156.
Methods
of Residual Stress
Xray Determination Quenching Stresses
157. Surface Residual Stresses
158. Stress Relief
16. Statistics
161.
Applied
to Materials Testing
419
419 420 421 424 426 430 432 435 439 441 442 444 446
Why
Statistics?
162. Errors 163. 164. 165. 166.
and Samples Frequency Distribution Measures of Central Tendency and Dispersion The Normal Distribution
Extremevalue Distributions
167. Tests of Significance
168. Analysis of Variance
169. Statistical
Design of Experiments
1610. Linear Regression 1611. Control Charts 1612. Statistical Aspects of Size Effect in Brittle Fracture
1613. Statistical
Treatment
of the Fatigue Limit
Part Four.
Plastic
Forming of Metals
17. General Fundamentals of Metalworking
171. 172. 173.
453
453 455 458 459 462 466 468 470
471
174.
175.
176.
177. 178. 179.
Forming Processes Effect of Temperature on Forming Processes Effect of Speed of Deformation on Forming Processes Effect of Metallurgical Structure on Forming Processes Mechanics of Metal Forming Work of Plastic Deformation Formability Tests and Criteria Friction in Forming Operations Experimental Techniques for Forming Analysis
Classification of
xlv
Contents
18. Forging
181. Classification of Forging Processes
182. 183. 184. 185.
186.
473
473 476 479 481 483 483 484 486
187.
188.
Forging Equipment Deformation in Compression Forging in Plane Strain with Coulomb Friction Forging in Plane Strain with Sticking Friction Forging of a Cylinder in Plane Strain Forging Defects Residual Stresses in Forgings
19. Rolling of Metals
191.
488
488 489 491 492 493 494 498 501 502 503 504 508
Classification of Rolling Processes
192. 193. 194.
RoUing Equipment Hot RoUing
195. Rolling of
Cold Rolling Bars and Shapes 196. Forces and Geometrical Relationships in Rolling
197.
198. 199.
Variables in RoUing Deformation in Rolling Defects in RoUed Products
Main
1910. Residual Stresses in Rolled Products
1911. Theories of Cold Rolling
1912. Theories of 1913.
Hot RoUing Torque and Horsepower
.511
514
514 517 518 522 524 525 526 527 529
20. Extrusion
201. 202. 203. 204.
205. 206.
Classification of Extrusion Processes Extrusion Equipment Variables in Extrusion Deformation in Extrusion Extrusion Defects Extrusion under Ideal Conditions
.
207. Extrusion with Friction
and Nonhomogeneous Deformation and Tubing
....
208. Extrusion of
Tubing
209. Production of Seamless Pipe
21
.
Rod, Wire, and Tube Drawing
532
532 532 534 535 536 539
541
211. Introduction
212.
213.
Rod and Wire Drawing Defects in Rod and Wire
Wire Drawing Wire Drawing without Friction Wire Drawing with Friction Tubedrawing Processes
214. Variables in 215.
216. 217. 218.
219.
2110.
2111.
Tube Sinking Tube Drawing with a Stationary Mandrel Tube Drawing with a Moving Mandrel Residual Stresses in Rod, Wire, and Tubes
542 543 545 547
22. Sheetmetal Forming
221. Introduction
549
549
Contents
222.
xv
550 555 557 562 563 568 569
57]
223. Shearing 224. 225.
226.
227.
228.
Forming Methods and Blanking Bending Stretch Forming Deep Drawing Redrawing Operations Ironing and Sinking
229. Defects in
Formed Parts
2210. Tests for Formability
57^
Appendix.
Problems
Constants, and Conversion Factors
,
.
577
579 599
603
Answers to Selected Problems
Name
Index
Subject Index
609
LIST
OF SYMBOLS
Area
Linear distance Interatomic spacing
A
a
ao
B
6
Constant
Width
b
C
Cij
or breadth Burgers vector of a dislocation Generahzed constant
Elastic coefficients
c
Length
D E
e
of Griffith crack Diameter, grain diameter
Modulus
Base
of elasticity for axial loading
(Young's modulus)
Conventional, or engineering, strain
of natural logarithms (= 2.718) Force per unit length on a dislocation
line
exp
F
/
Coefficient of friction
G
9
Modulus
of elasticity in shear
(modulus
of rigidity)
Crackextension force
Activation energy
Distance, usually in thickness direction
Miller indices of a crystallographic plane
H
h
{h,k,l)
/
Moment
of inertia
J
Invariant of the stress deviator; polar
moment
of inertia
K
Kf
Ki
k
Strength coefficient
Fatiguenotch factor
Theoretical stressconcentration factor
Yield stress in pure shear
L
I,
Length
m, n
Direction cosines of normal to a plane
In
log
Natural logarithm Logarithm to base 10
Mb Mt
Bending moment Torsional moment, torque
xvili
List of
Symbols
m
Strainrate sensitivity
N
n
n'
Number
of cycles of stress or vibration
Sti'ainhardening exponent
Generalized constant in exponential term
P
p
q
Load
or external force
Pressure
Reduction
in area; plasticconstraint factor;
notch sensitivity
index in fatigue
R
r
*Si
Radius
of curvature; stress ratio in fatigue; gas constant
Radial distance
Total stress on a plane before resolution into normal and shear
components
Sij
Elastic compliance
s
T
Tm
t
tr
Standard deviation Temperature
Melting point
of a
sample
Time; thickness
Time
for rupture
U
Uo
u, V, IV
Elastic strain energy
Elastic strain energy per unit
volume
x,
ij,
Components
of
displacement in
and
z directions
[uvw]
Miller indices for a crystallographic direction
V
V
Volume
Velocity; coefficient of variation
W
Z
a
a,0,d,(t)
Work
ZenerHollomon parameter
Linear coefficient of thermal expansion
Generalized angles
r 7
Line tension of a dislocation Shear strain
A
8
e e e
Volume
strain or cubical dilatation; finite change Deformation or elongation; deflection; logarithmic decrement
Natural, or true, strain
Significant, or effective, true strain
Truestrain rate
im
7/
Minimum
creep rate
Efficiency; coefficient of viscosity
6
K
Dorn timetemperature parameter Bulk modulus or volumetric modulus
Interparticle spacing
of elasticity
A
X
ju
Lame's constant Lode's stress parameter
List of
p
Symbols
xix
Poisson's ratio; Lode's strain parameter
p
a
Density
Normal
stress; the
standard deviation of a population
(To
(Tq
CT
Yield stress or yield strength Yield stress in plane strain
Significant; or effective, true stress
01,02,03
0'
Principal stresses
Stress deviator
0"
(Ta
o,„
Hydrostatic component of stress
Alternating, or variable, stress
Average principal
stress;
mean
stress
ar
Range
of stress
a„
ou,
Ultimate tensile strength
Working
stress
T
<l>
Shearing stress; relaxation time Airy stress function
Specific
i/'
damping capacity
Part
One
MECHANICAL FUNDAMENTALS
.
in which case it is necessary to know something about the limiting values which can be withstood without failure. elasticity. Scope of This Book is Mechanical metallurgy the area of metallurgy which is concerned primarily with the response of metals to forces or loads. for most students of metallurgy and for practicing engineers in industry. it is worth spending the time to become familiar with the mathematics presented in Part One. The theories of strength of materials. machine design. On the one hand is the approach used in reference to strength of materials and in the theories of elasticity and plasticity. It is a combination of many disciplines and many approaches to the problem of understanding the response of materials to forces.Chapter 1 INTRODUCTION 11. 1 to 3 can be considered the mathematical framework on which much of the remainder of the book rests. and the three topics of strength of materials. where a metal is considered to be a homogeneous material whose mechanical behavior can be rather precisely described on the basis of only a very few material constants. it probaHowever. For students of engineering who have had an advanced course in strength of materials or bly will be possible to skim rapidly over these chapters. Mechanical metallurgy is not a subject which can be neatly isolated and studied by itself. and here it is necessary to know the conditions of temperature and rate of loading which minimize the forces that are needed to do the job. elasticity. the objective may be to convert a cast ingot into a more useful shape. The material covered in Chaps. such as a flat plate. The forces from the use of the metal as a member or part in a structure or machine. of their and plasticity lose much power when the structure of the metal becomes an impor3 . This approach is the basis for the rational design of structural members and machine parts. and plasticity are covered in Part One of this book from a more generalized point of view than is usually considered in a first arise may course in strength of materials. On the other hand.
Many of the alliance of the solidstate physicist with the metallurgist. Therefore. Part Four considers the metallurgical and mechanical factors involved in the forming of metals into useful shapes. which occurs in plain carbon steel. especially Chap. where the metallurgical structure may continuously change with time. Appli cations to Materials Testing. mechanical behavior them. Much of the material in Parts One and Two has been utilized in Part Three. Examples of this are in the hightemperature behavior of metals. are concerned primarily with atomistic concepts of the flow and fracture of metals. However. Metallurgiof the material in Part Two has been lurgical cal students will find that metallurgy a first covered in a previous course in physical metallurgy. Part Three. The determination of the relationship between mechanical behavior and structure (as detected chiefly with microscopic and Xray techniques) is the main responsibility of the mechanical metallurgist. fundamentals Part of the is understood in terms of metallurgical struc it is generally possible to improve the mechanical properties or at least to control Two of this book is concerned with the metalsome mechanical behavior of metals. developments in these areas have been the result of the This is an area where direct observation is practically impossible and definitive experiments of an indirect nature are difficult to conceive. certain topics which pertain more to physical metallurgy than mechanical metallurgy have course in physical metallurgy. it is an area of intense activity in which the lifetime of a concept or theory may be rather short. Moreover. The last three chapters of Part Two. these subjects are considered in greater detail than is usually the case in In addition. Attempts have been made . considers each of the common mechanical not from the usual standpoint of testing techniques. since mechanical is part of the broader field of physical metallurgy. in writing these chapters an attempt has been made to include only material which is generally valid and to minimize the controversial aspects of the subject.4 Mechanical Fundamentals [Chap. or in the ductiletobrittle transition. number of standardized mechanical tests. Basic data concerning the strength of metals and measurements for the routine control of mechanical properties are obtained from a relatively small tests. When ture. 6. been included cal students in order to provide continuity and to assist nonmetallurgi who may not have had a course in physical metallurgy. 1 tant consideration and it can no longer be considered as a homogeneous medium. It is assumed that the reader either has completed a conventional course in materials testing or will be concurrently taking a laboratory course in which familiarization with the testing techniques will be acquired. but instead from the consideration of what these tests tell about the service performance of metals and how metallurgical variables affect the results of these tests.
Sec. it is necessary to is make the internal resisting done by passing a plane through the part of the body lying on one side interest. Since the equations of equilibrium must be expressed in terms of forces in strength of materials the first step to acting external to the body. it is the integral of the stress In order to evaluate this arrived at necessary to area of the cutting plane. and external loads. deformation. and how geometrical and metallurgical the forming loads and the success of the metalworking 12. expression for the stress is then substituted into the equations of equistrain For present purposes stress is defined as the change will be given later. acting on the "free body" hold it in equilibrium. so that the internal force times the differential area over which it acts. either because of the of this book. considerable detail required or because the analysis is beyond the scope No attempt has been made to include the extensive special ized technology associated with each metalworking process. although in certain cases this has not been possible. ing know the The stress distribution of the stress over the distribution is by observ and measuring the strain distribution in the member. such as rolling or extrusion. the equations of equi librium may be applied to the problem. in length per unit length. In method of analysis used assume that the member is in equilibrium. However. The point of the body at replaced and by the forces it exerted on plane is removed cutting the of This Since the forces the cut section of the part of the body that remains. The internal resisting forces are usually expressed is by the stress^ acting over a certain area. integral. Strength of Materials — Basic Assumptions Strength of materials the general is is the body of knowledge which deals with the relation betw^een internal forces. The equations of static equilibrium are applied to the forces acting on some part of the body in order to obtain a relationship between the external forces acting on the member and the internal forces resisting the action of the external loads. emphasis has been placed on presenting a fairly simplified picture of the forces involved in each process factors affect process. the determi The nation of the strain distribution provides the stress distribution. although some effort has been made to give a general impression of the mechanical equipment required and to familiarize the Major reader with the specialized vocabulary of the metalworking field. since stress cannot be physically measured. The companion term More complete definitions . forces into external forces. 12] Introduction 5 to present mathematical analyses of the principal metalworking processes. since stress is proportional to strain for the small deformations involved in most work. * is defined as force per unit area.
rolling or forging. Howwhen metals are severely deformed in a particular direction. materials can be deformed when sub found that up to certain limiting loads a solid will recover its original dimensions when the load is removed. This relationship is known as Hooke's law. 1 and they are solved for stress in terms of the loads and dimenmember. in general. the body will experience a permanent set or deformation when the load is removed. homogeneous. cast iron. of an aggregate of crystal grains having different properties in different crystallographic directions. such that on a micro scale they are heterogeneous. Most engineering metals are made up of more than one phase. A property which varies with orientation with respect to some system of axes is said to be anisotropic. the deformation is proportional to the load. as in statistically mechanical properties may be anisotropic on a macro scale. If the elastic limit is exceeded. sions of the when that property does not vary with direction or orientation. with different mechanical properties. Hooke's law requires that the loaddeformation relationship However. Elastic and Plastic Behavior all solid Experience shows that jected to external load. Further. it is more frequently stated as stress is proportional to strain. Important assumptions in strength of materials are that the body which is being analyzed is continuous. the materials are ever. the crystal grains are so small that. even a singlephase metal will usually exhibit chemical segregation.6 Mechanical Fundamentals [Chap. A continuous body is one which does not contain voids or empty spaces of any kind. it does not necessarily follow that all mate . and isotropic. it is readily apparent when they are viewed through a microscope that they are anything but homogeneous and isotropic. For most materials. A body which is permanently deformed is said to have undergone plastic deformation. should be linear. and therefore the Metals are made up properties will not be identical from point to point. A body is considered to be isotropic with respect to some property librium. of materials describe the The reason why the equations of strength behavior of real metals is that. A body is homogeneous if it has identical properties at all points. and aluminum may appear to meet these conditions when viewed on a gross scale. for a specimen of ume. as long as the load does not exceed the elastic limit. the any macroscopic volhomogeneous and isotropic. 13. While engineering materials such as steel. The limiting load beyond which the material no longer behaves elastically is the elastic limit. The recovery of the original dimensions of a deformed body when the It is further load is removed is known as elastic behavior.
the elastic limit is decreased. consider a uniform cylindrical bar which is subjected to an axial tensile load (Fig. unstrained state and that Lo is the gage length between these marks. •(_4. so that most metals there is only a rather narrow range of loads over which Hooke's law importance.Sec. primarily of academic Hooke's law remains a quite valid relationship for engi neering design. 11. Average Stress and Strain As a starting point in the discussion of stress and strain. where a is the stress normal to the cutting plane and A is in Fig. Cylindrical bar subjected to Fig. however. 12. As the meas Rubber for uring devices become more sensitive. have shown that the elastic limits of metals are much lower than the values usually measured in engineering tests of materials. axial load. Figure 12 shows the freebody diagram for the cylindrical bar shown The external load P is balanced by the internal resisting force j(7 dA. called the deforits A load mation. This is. is an example of a material with a nonlinear stressstrain relationship that still satisfies the definition of an elastic material. 14] rials Introduction which behave elastically will have a linear stressstrain relationship. _ Strain is J^ Lo _ AL _ L Lo Lo (11) Lo 5 a dimensionless quantity since both and Lo are expressed in units of length. . P is applied to one end of the bar. Lo + S Ln^P 'a dA ^P Fig. Elastic deformations in metals are quite small and require very sensiUltrasensitive instruments tive instruments for their measurement. Freebody diagram for Fig. The average linear strain e is the ratio of the change in length to the original length. between the gage marks has increased by an amount 5. strictly applies. 11. Assume that two gage marks are put on the surface of the bar in 11). 11. and the gage length undergoes The distance a slight increase in length and decrease in diameter.
so that proportional to the average strain. 1 the crosssectional area of the bar. e = E = constant or Young's modulus.000 psi. square inch In scientific work stresses are (ksi). and the proportionality between stress and strain would have to be identical for each element. The inherent anisotropy between grains plete uniformity of stress over a metal rules out the possibility of comThe presence of macroscopic size. Tensile Deformation of Ductile Metal The basic data on the mechanical properties of a ductile metal are obtained from a tension test.) often expressed in units of kilograms per square millimeter or dynes per 10^ dynes/cml) (1 kg/mm^ = 9. in which a suitably designed specimen is The load and elongasubjected to increasing axial load until it fractures. the microscopic scale.8 Mechanical Fundamentals [Chap. 15. (13) represents an average stress. In engineering practice.81 square centimeter. the stress will not be uniform over the area A. thousands of pounds. of more than one phase also gives rise to nonuniformity of stress on a If the bar is not straight or not centrally loaded. the load is usually measured in pounds and the area in square inches. so that stress has units of pounds per square Since it is common for engineers to deal with loads in the inch (psi). in a polycrystalline body stress pattern occurs when there is an abrupt change in cross section. if the stress is distributed uniformly over the area A.000 lb. that a is con stant. 213). (12) becomes P = cr aj dA = (tA (13) = j In general. Eq. The equinbrium equation is P = jadA If (12) is. an obvious simpHfication is to work with units of The stress may be expressed in units of kips per 1. tion are measured at frequent intervals during the test and are expressed . (1 ksi = 1. called kips. (14) The constant E is the modulus of elasticity. For the stress to be absolutelyuniform. and therefore Eq. X Below the elastic limit is the average stress Hooke's law can be considered vaHd. every longitudinal element in the bar would have to experience exactly the same strain. This results in a stress raiser or stress concentration (see Sec. strains will be different for certain longitudinal elements and the stress An extreme disruption in the uniformity of the will not be uniform.
16. The yield strength is defined as the stress which will produce a small amount of permanent deformation. point B. Typical tension stressstrain routine. a of The determination limit is the elastic all Fig. 9. or offset. is based on the original area from maximum load to fracture. The maximum load divided by the original area of the specimen is the ultimate For a ductile metal the diameter of the specimen tensile strength. when the begins to decrease rapidly beyona Since the average stress it maximum off until load. Figure 13 illustrates the tension stressstrain curve of a ductile material. A is the elastic Fracture limit. For engineering purposes the limit of usable elastic behavior is described by the yield strength.2 per cent or 0. not at 13. Plastic deformation begins elastic limit is exceeded. so that the load required to continue deformation drops also decreases the specimen fractures. is OC. The proportional limit the stress at which the stressof the stressstrain strain curve deviates in this region is the from linearity. The initial linear portion of the curve OA is the elastic region within which Point Hooke's law is obeyed. Ductile vs. per. a metal such as aluminum or copas average stress section. Brittle Behavior The general behavior of materials under load can be classified as ductile or brittle depending upon whether or not the material exhibits the ability to undergo plastic deformation.002 inches per inch. of the specimen. Eventually the load reaches a maximum value. generally this a strain equal to 0. the strainmeasuring it is instrument. As the plastic deformation of the specimen increases. A completely brittle material would . 16] Introduction and strain according to the equations in the previous (More complete details on the tension test are given in Chap.Sec. the metal becomes stronger (strain hardening) so that the load required to extend the specimen increases with further straining. For these reasons often replaced is by the proportional curve point A'.) The data obtained from the tension test are generally plotted as a Figure 13 shows a typical stressstrain curve for stressstrain diagram. The slope modulus of elasticity. 13 permanent strain. limit. In Fig. Strain e quite tedious. defined as the greatest stress that the metal can withstand with out strain experiencing permanent when the load is removed. sitivity and dependent on the senof curve.
is ductile at an elevated temperature. What Constitutes Failure? fail Structural members and machine elements can intended functions in three general ways: 1. to perform their Excessive elastic deformation Yielding. such as the design because An . For different types of failure. while a brittle metal. or excessive plastic deformation 2. static strain it is possible to design for basis of situations on the stresses.10 Mechanical Fundamentals [Chap. shows (Fig. a metal which is ductile in tension at room temperature can become brittle in the presence of notches. continue to build up is when there no local yielding. high rates of loading. Furthermore. and it spreads rapidly over the section. 146). with brittle materials. a metal. 1 fracture almost at the elastic limit (Fig. different significant parameters will be important. 3. or embrittling agents such as hydrogen. (b) stressstrain curve for brittle metal with slight amount of ductility. Two general types of excessive elastic deformation may occur: (1) excessive deflection under conditions of stable equilibrium. low temperature. because it allows the mate rial to redistribute localized stresses. When localized stresses at notches and other accidental stress concentrations do not have to be considered.0). such measure of plasticity before fracture an important engineering consideration. 14. Finally. a crack forms at one or more points of stress if concentration. localized stresses Fig. important to note that brittleness is not an absolute property of A metal such as tungsten. as white cast iron. Fracture understanding of the common types of failure is important in good it is always necessary to relate the loads and dimensions of the member to some significant material parameter which limits the loadcarrying capacity of the member. some slight is Adequate ductility 14. 17. which is brittle at room temperature. fracture will still occur suddenly because the and tensile strength are practically identical. yield stress It is Even no stress concentrations are present in a brittle material. A metal which is brittle in tension may be ductile under hydrostatic compression. average com However. (a) Stressstrain curve for pletely brittle material (ideal behavior).
Excessive elastic deformation of a machine part can mean failure of the machine just as much as if the part completely fractured. Yielding produces permanent change of shape. (3) delayed fracture. 34). Yielding. (2) fatigue. A part made from a ductile metal which is loaded statically rarely fractures like a tensile specimen. plex phenomenon will be considered in greater detail in Chap. (2) sudden deflection. not critical value. occurs of the when the elastic limit metal has been exceeded. 13. which may prevent the part from functioning properly any longer. is usually The most effective way to increase the stiffness of a member by changing its shape and increasing the dimensions of its cross section.Sec. or progressive fracture. In the previous section under brittle static was shown that a brittle material fractures loads with little outward evidence of yielding. because the metal strain hardens as it deforms. or excessive plastic deformation. under conditions of unstable equilibrium. because it will first fail by excessive plastic deformation. but it must be used with a suitable failure criterion (Sec. The formation of a crack which can result in complete disruption of continuity of the member constitutes fracture. For more complex loading conditions the yield strength is still the significant parameter. metals fail by fracture in three general ways: (1) sudden brittle fracture. Generally. and an increased stress is required to produce further deformaFailure by excessive plastic deformation is controlled by the yield tion. A sudden it certain conditions. strength of the metal for a uniaxial condition of loading. The failure criterion under creep conditions is complicated by the fact that stress is not proportional to strain and the further fact that the mechanical properThis comties of the material may change appreciably during service. The sudden buckling type of failure may occur in a slender column when the axial load exceeds the Euler critical load or when the trolled external pres. metals can continuously deform at constant stress in a timedependent yielding known as creep. or the excessive deflection of closely mating parts can result in interference and damage to the parts. type of fracture can also occur in ordinarily ductile metals under Plain carbon structural steel is the most common . a shaft which is too flexible can cause rapid wear of the bearing. Instead. metallurgical control can be exercised over the elastic modulus. by the modulus little by the strength of the material. For example.sure acting against a thinwalled shell exceeds a Failures due to excessive elastic deformation are conof elasticity. or buckling. At temperatures significantly greater than room temperature metals no longer exhibit strain hardening. However. 17] Introduction 11 deflection of beam under gradually applied loads. In a ductile metal under conditions of static loading at room temperature yielding rarely results in fracture.
sary in calculating the stresses for Allowance must be made for the possibility of accidental loads of high magnitude. which in turn show a certain variability in mechanical can be influenced by changes in heat treatand approximations are usually necesall but the most simple member. A change from the ductile to the brittle type of fracture is promoted by a decrease in temperature.12 Mechanical Fundamentals of [Chap. This problem is considered in example Chap.. generally at a notch or stress concentration. and gradually spreads over the cross section until the member breaks. Further. static applications the working stress of ductile metals is usually based on the yield strength ao and for brittle metals on the ultimate tensile strength o. One common type of delayed fracture is stressrupture failure. = rj (l5j iV u . it is necessary that the allowThe able stresses be smaller than the stresses which produce failure. value of stress for a particular material used in a particular way which is For considered to be a safe stress is usually called the working stress (Tw. and therefore design for fatigue failure is based primarily on empirical relationships using nominal stresses. A minute crack starts at a localized spot. in order to provide a margin of safety and to protect against failure from unpredictable causes. 1 a material with a ductiletobrittle transition. and the presence of a complex state of stress due to a notch. Values of working stress are established by local and Federal agencies and by technical organizations such as the American Society of Mechanical Engineers (ASME). occurs at room temperature when steel is statically loaded in the presence of hydrogen. Fatigue of metals is discussed in greater detail in Chap. Fatigue failure occurs without any visible sign of yieldis ing at nominal or average stresses that are well below the tensile strength of the metal. by a number Ow "^ jTf•/V or CTu. Fatigue failure caused by a critical localized tensile stress which is very difficult to evaluate. Depending upon the stress and the temperature there may be no yielding prior to fracture. Fatigue failures occur in parts which are subjected to alternating. The working stress may be considered as either the yield strength or the tensile strength divided called the factor of safety. stresses. 14. or fluctuating. 12. Thus.. an increase in the rate of loading. ment or fabrication. which occurs when a metal has been statically loaded at an elevated temperature for a long period of time. A similar type of delayed fracture. Most fractures in machine parts are due to fatigue. uncertainties usually exist regarding the magnitude of the applied loads. in which there is no warning by yielding prior to failure. All engineering materials properties.
psi factor of safety based on yield strength factor of safety based on tensile strength to the factor of safety depends on an estimate of The value assigned all In addition. In military equipment.Sec. of safety p . the factor of safety should be increased. The the factors discussed above. which is subjected to vibration and fluctuating stresses. 18] Introduction 13 where = = (r„ = iVo = Nu = ay. the factor would be lower than in a machine. cro working stress. careful consideration should be given to the consequences which would result from failure. psi tensile strength. type of equipment will also influence the factor of safety. the factor The factor of of safety may be lower than in commercial equipment. where light weight may be a prime consideration. psi yield strength. as in a building. safety will also depend on the expected type of loading. If failure would result in loss of life. For static loading.
l5a. However. the stress will be different on any other plane passing through point 0. a normal stress a perpendicular to AA. such as the plane nn. 156. Mechanical Fundamentals [Chap. To obtain the stress at some in a plane such as mm. We then AP take an area AA If surrounding the point and note that a force acts on this area. part 1 of the body is removed and replaced point by the system of external forces on which will retain each point in section of the mm part 2 of the body in the same position as before the removal of part 1. The same stress at point in plane would be obtained if the free body were constructed by removing part 2 of the solid body. To illustrate this point. AP/AA the mm lim —r = AA cr (16) AA^O stress will be in the direction of the resultant force P and will generbe inclined at an angle to A A. consider Fig. Fig. The total stress can be resolved into two components. 1 forces due to highspeed rotation ential over the and forces due to temperature diffe)' body (thermal stress). and a shearing stress (or shear stress) r lying in the plane mm of the area. It is inconvenient to use a stress which is inclined at some arbitrary angle to the area over which it acts. 16. the limiting value of the ratio stress at the point on plane of body 2. 16. Also. This is the situation in Fig. the plane containing the normal and P intersects the plane A along a dashed line that makes an The ally mm angle <^ with the y axis. The normal a stress is given by (17) P ^ r A COS 6 . The force P makes an angle 6 with the normal z to the plane of the area A. In general the force will not be uniformly distributed over any cross body illustrated in Fig. the area AA is is continuously reduced to zero. Resolution of total stress into its components.14.
will be discussed more fully in Chap. ^ linear strain. 14 the average linear strain was defined as the ratio of the change in length to the original length of the same dimension. in general a given plane may have one normal stress and two shear stresses acting on it. or true.= \nj^ (111) The above equation defines the natural. Rather than referring the change in length to the original gage length. the strain at a point is the ratio of the deformation to the gage length as the gage length approaches zero. . For the present it should be noted that for the very small strains for which the equations of elasticity are valid the two definitions of strain give identical values.Sec. linear dimen —.^ Not only will the elastic deformation of a body result in a change in length of a linear element in the body. strain. Concept of Strain and the Types of Strain In Sec. while true strain is sometimes denoted by 5 or e. 3. 19] Introduction stress in the plane acts along the line 15 The shear OC and has the magnitude (18) T=T&\nd This shear stress may be further resolved into components parallel to the X and y directions lying in the plane. which is useful in dealing with problems in plasticity and metal forming. True strain. Linear strain is often denoted by e. it = = average linear strain often is sion divided more useful to define the strain as the change in by the instantaneous value of the dimension. but it may also result in a change Considerable variance exists in the literature regarding the notation for average and deformation. X direction y direction "^ ~ ~ P ~a ^^^ ^ ^^^^ ^ (1"9) P ~a '^ ^^^^ ^ ^^^ ^ (110) Therefore. 19. _ where e 5 ± _ AL L  Ln deformation By analogy with the definition of stress at a point. true strain.
ed. 1957." McGrawHill Book Company. The angle at A.." The Ronald Press Company. F. Frocht. Inc..: "Strength of Materials. [Chap.. ed. and Sons. York. Mechanics of Materials. the tangent angle (in of the angle and the radians) are equal. Therefore. shear strains are often expressed as angles of rotation.. Smith: "Resistance of Materials. h. 1 in the initial right angle is known is as shear strain. R.. and New J.. M. M. B. C. 1952. New York. Shear strain. The ratio a/h is also the tangent of the angle through which the element has been rotated. Inc. 1959." McGrawHill Book Company." 4th York. The angular change in a Figure 17 illustrates the strain produced by the pure shear was originally 90°. divided by the distance between the planes. 17.): 1951. Seely. which decreased by the application of a shear stress by a The shear strain y is equal to the displacement a Fig. New York." 2d John Wiley & Shanley. F. Inc. O.. H.. Crandall. John Wiley & Sons. For the small angles usually involved.: "Strength of Materials. small amount 6. 7 = 7 = tan 6 = n (112) BIBLIOGRAPHY "An Introduction to the Mechanics of Solids.16 Mechanical Fundamentals angle between any two lines. S. 1957. Dahl (eds. New . and N. of one face of a cube. ^: "Advanced New York. Inc.
however. it must act on the plane perpendicular to the x direction. Description of Stress at a Point As described at a point into in Sec. and for this reason ships for expressing the stress and strain at a point it should be given careful attention by those readers to In the space available for this subject it whom it is unfamiliar. In the general case the shear components are at arbitrary angles to the coordinate axes. The general case is shown in Fig.Chapter 2 STRESS AND STRAIN RELATIONSHIPS FOR ELASTIC BEHAVIOR 21 . 22. included in this chapter is important for an understanding of most of the phenomenological aspects of mechanical metallurgy. material covered here should. ax is the normal stress acting in the x direction. it is often convenient to resolve the stresses normal and shear components. provide a background for intelligent reading of the more mathematical literature in mechanical metallurgy. While part of the material covered in this chapter is a review of information generally covered in strength of materials. that is. ble to carry has not been possi The it to the point where extensive problem solving is possible. Stresses acting normal to the faces of the elemental cube are identified by the subscript which also identifies the direction in which the stress acts. Since it is a normal stress. 21. By convention. values less than zero 17 . the subject is extended beyond this point to a consideration of stress and strain in three dimenThe material sions and to an introduction to the theory of elasticity. 18. so that it is convenient to resolve each shear stress further into two components. values of normal stresses greater than zero denote tension. Introduction of this chapter is The purpose to present the mathematical relation and the relationships between stress and strain in a rigid body which obeys Hooke's law.
oy. normal stresses shown in Fig. shown in Fig. P. 21 are positive. direction of the z axis. Inc.18 Mechanical Fundamentals All the [Chap. tensile. The first and the second the Since a plane is most easily defined direction in which the stress acts. Goodier. If we assume that the areas of the faces of the unit cube are small enough so that the change in stress over the face is negligible. They are ax. and Tzy.. of Elasticity. by taking the sum1 S. Shearing stresses oriented in the if positive directions of the coordinate axes are positive a tensile stress on the same cube face All the shear stresses is in the positive direction of the corresponding axis. McGraw . subscript indicates the plane in which the stress acts Tyz is the shear stress on the plane perpendicular to the y axis in the Fig. "Theory Book Company. that several other systems of notation are in use. 21 are Two subscripts are needed for describing shearing stresses. 21 . Before attempting to read papers in this field it is important to become familiar with the notation which is used. N. For example. 21 that nine quantities must be defined in order to establish the state of stress at a point." 2d ed.. New York. The reader is reminded. some simplification is possible. Txz. 2 indicate compression. The notation given above is the one used by Timoshenko^ and most American workers in the field of elasticity. Howcvcr. 1951. the first subscript refers to this normal. T^x. It can be seen from Fig. Hill Timoshenko and J. t„x is the shear stress on a plane normal to the y axis in the direction of the x axis. however. Txy. by its normal. Tyx. cr^. Stresses acting on an elemental unit cube. Tyz.
Txy. The stress system will consist A stress conof two normal stresses (Xx and ay and a shear stress Txy. plane of the paper. 1953. two dimensions. problems can be simplified by considering a twodimensional This condition is frequently approached in practice when ) Fig. State of Stress in Two Dimensions (Plane Stress) Many state of stress. ^ see C. For example. To do this. there will be no stress acting perpendicular to the plate.)ec. dition in which the stresses are zero in one of the primary directions is called plane stress. Stress on oblique plane. ax. 23. New York. Inc. Tyz. T. 22. consider an oblique plane normal to the plane of the paper at an angle 6 between the For a more complete derivation McGrawHill Book Company. Txz. 23] Stress and Strain Relationships For Elastic Behavior 19 mation Tiy of the moments of the forces about the (Tyx z axis it can be shown that = Tyx. (tjcu a?/ Az) • • Ax = Ax Az) Ay (21) Txy == Tyx and in like manner TXZ ^^ TzX Tyz = Tgy Thus." pp. the state of stress at a point is completely described by six components/ three normal stresses and three shear stresses. Figure 22 illustrates a thin plate with its thickness normal to the in In order to know the state of stress at point for we need to be able to describe the stress components at any orientation of the axes through the point. ay. in a thin plate loaded in the plane of the plate. Wang. the plate. cr^. 79. .. one of the dimensions of the body is small relative to the others. "Applied Elasticity.
the areas of the sides of the element perpendicular to the x and y axes are Al and Am. it is often convenient to express Eqs. (oy — (Xx) sin 9 cos 6 (23) To aid in computation. If A is the area of the oblique it follows that I = cos 6 and m = sin 6. stress a SxA = SyA = or CFxAl f TxyA^m ay (Tx (Ty Am + TxyAl cos 6 sin 6 f Sx Sy of Sx = = Txy sin 6 Txy + cos The components and Sy Sx COS 6 in the direction of the normal 9 stress a are SxN = and SyN = Sy sin is so that the normal stress acting on the oblique plane (T cr given by = = Sx COS 6 (Tx j Sy sin 6 \ cos^ d \. From the geometry of Fig. 2 X axis and the normal A^ to the plane. It is assumed that the plane and that the eleshown in Fig.20 Mechanical Fundamentals [Chap. The direction cosines between N and the X and y axes are I and m. 22 is an infinitesimal distance from ment is so small that variations in stress over the sides of the element The stresses acting on the oblique plane are the normal and the shear stress r. we obtain can be neglected. Let Sx and Sy denote the x and y components of the total stress acting on the inclined face. By taking the summation of the forces in the x direction and the y direction. plane. . respectively.(jy sin^ d 2Txy sin 6 cos 6 (22) The shearing stress on the oblique plane T T is given by = = Sy cos 6 Txy{cos~ 6 — Sx sin 9 — sin^ 9) + 29. 22. cos 20 u + 1 . (22) and (23) in terms of the double angle This can be done with the follow ing identities: uuo .
plane through point 2. The maximum shear The an angle halfway between the maximum and minimum normal 4.000psi Fig. The maximum and minimum values of both normal stress and shear which are 90° apart. Variation of normal stress and shear stress with angle 6. 23. and shear stress on any plane through a point a body subjected to a planestress situation. These relationships are valid for any state of stress. 23] Stress and (23) Strain Relationships for Elastic Behavior 21 Equations in describe the normal stress and their equivalents. Figure 23 shows (22) and cr^ = 2. variation of normal stress and shear stress occurs in the form of a sine wave. Note the following important facts about this figure 1. . Eqs. with a period oi 6 = 180°. (24) and (25). The maximum and minimum values of normal stress on the oblique occur when the shear stress is zero. stress occurs at stresses. the variation of normal stress and shear stress with 6 for the biaxial planestress situation given at the top of the figure. stress occur at angles 3.: Sec.
for 1. 61 and 62 = 61 These roots define two mutually perpendicular planes which = tan (tt  26).)V4 + ry^^ cos 26 = + (OX  (Ty)/2 Substituting these values into Eq. (24) will give the principal stresses when values of cos 26 and 26 The values of cos 26 and sin are substituted into it from Eq. (24) results in the expression for the . relationships. The specification and their direction provides a convenient way the state of stress at a point. axes 1. and Although and 3 do not coincide with the cartesiancoordinate axes of many situations of the princiof describing that are encountered in practice the two systems of axes coincide because pal stresses symmetry of loading and deformation. the case of stress in three dimensions there apart (Fig.v(cos2 6 — sin. directions of the principal stresses are the principal in general the principal axes x. principal stresses 02. (26) has two roots. 03. principal stresses and a 2 which occur at angles that are 90° will be two ai For general 23). and the stresses normal to these For twodimensional plane stress there planes are the principal stresses. 2. Pythagorean 26 from Eq.22 Mechanical Fundamentals [Chap. and According to convention. The 3. z. Eq. its angu lar relationship with respect to the xy coordinate axes can be determined by finding the values of 6 in Eq. 2 For any state of stress it is always possible to define a new coordinate system which has axes perpendicular to the planes on which the maximum normal stresses act and on which no shearing stresses act. while 03 is the algebraically smallest stress. (Ti is the algebraically greatest principal stress. 2.6) + {uy — Ux) sin 6 cos = 1 rxy (Tx — _ (^y sin e cos d cos^ 6 — ^ >^(sin 26) ^ ^^^^ sin^ 6 cos 26 ^^ (26) 2 tan 26 = ^^^^ (Tx — Cy Since tan 26 j mr/2. by means of the sin are found (26) Equation sin 26 = ± [{ax  (r. Since by definition a principal plane contains no shear stress. (23) for which r = 0. These planes are called the principal planes. are free from shear. Ti. will be three ai. y. (26).
Sec. Mohr's Circle of Stress —Two Dimensions A graphical method for representing the state of stress at a point on any oblique plane through the point was suggested by O. Mohr. 24. The intersection of the line AB with the x axis circle. — (Jy \ 2 T' (27) 0'2 24. Figure 24 is a Mohr's circle diagram for a twodimensional state of stress. determines the center of the is At points is D and E the shear stress zero. . (7 max principal stresses for a twodimensional (biaxial) O'miD = = O"! I 0"X + (Ty \ ( a. shear stresses along the on the planes normal to the x and y axes are plotted and B. Normal y axis. 24] Stress and Strain Relationships for Elastic Behavior 23 maximum and minimum state of stress. The angle between the x axis and ax determined by angle ACD in Fig. Fig. stresses are plotted along the The stresses as points A x axis. so that these points represent the values of the principal stresses. Mohr's circle for twodimensional state of stress.
i. the sum of the normal stresses on two perpendicular planes is a constant.. It can also be seen from Fig.CE = The radius of the circle is + CTy 2 J equal to nn Thus. of the principal stresses for a threedimensional The determination coordinate system is state of stress in terms of the stresses acting on an arbitrary cartesian an extension of the method described in Sec. 24 that OF' + OG' = 20C. State of Stress in Three Di mensions The If general threedimensional state of stress consists of three unequal principal stresses acting at a point.e. 2 This angle on Mohr's axis Fig. 23 . we find that to determine the stresses on the oblique plane whose normal is at an angle 6 we must advance an angle 26 from point A in the Mohr's circle. The normal and shearing stresses on the oblique plane are given by the coordinates of point F. circle is equal to twice the angle between ai and the x From on the actual stressed body. The stresses on a plane perpendicular to mm would be obtained by advancing an additional 180° on the Mohr's circle to point G. or spherical. Therefore. This shows that the shearing stresses on two perpendicular planes are numerically equal. This is called a triaxial state of stress.24 24. Mohr's circle oblique plane mm. two of the three principal stresses are equal. Note that it acts on a plane for which 6 = x/4 (26 = 7r/2 on Mohr's circle). can also be used to determine the stresses acting on any Using the convention that 6 is positive when it is measured clockwise from the positive x axis. the plane on which r^ax acts bisects the angle between the principal stresses. <Ti "^i ~ *^2 circle is equal to the maximum shearing — (72 + r^ (28) This value is given by the maximum ordinate of the circle. while stress is three principal stresses are equal. Mechanical Fundamentals [Chap. the state of stress is if all known as cylindrical. 24 it can be determined that ^^ = OC — + CD = ^^±^^ + Ax 2 <^X <Ty\ I I 2 ' xy OC . 25. the state of said to be hydrostatic. the radius of Mohr's stress. independent of the orientation of the planes.
the forces acting on each of its faces must balance. Let I. Since the free = Area KOL = Al Sx (tI = Area J OK = Sy am S. that is. n be the direction cosines of a. 25] Stress and Strain Relationships for Elastic Behavior 85 for the twodimensional case. the cosines area A. body in Fig. an Am Area JOL An Taking the summation (tAI of the forces in the x direction results in — (TxAl — TyxAm — — Tyxtn TzxAn = (29a) which reduces to ((7 — (Tx)l — Tsxn = Summing the forces along the other two axes results in — Txyl + — Txzl — Equations (29) are three (O — Cry)m TyzVl + (cr — T^yU = — (tO = linear equations in (296) (29c) homogeneous terms of I. 25 must be in equilibrium. Fis.Sec. of the angles between <x and the x. a is the principal stress acting normal to the plane JKL. . body similar to that shown in Fig. y. m. and z axes. and S^. Sy. The components of a along each of the axes are Sx. 21 Figure 25 represents an elemental free with a diagonal plane JKL of The plane JKL is assumed to be a principal plane cutting through the unit cube. Stresses acting on elemental free body. 25.
02. I (214) . (210) are the three principal stresses and 03. m. To determine the direction. and zero. 02. y. The resulting equations must be solved simultaneously for I. Let S be the total stress before resolution into normal and shear com ponents and acting on a plane (not a principal plane). = Solution of the determinant results in a cubic equation in <J^ a. in which the principal stresses act.26 m. 2 The only solution can be obtained by setting the determinant I. and let I. — (ox + 0"^ + — O'z)"'" + {(TxCTy + (Ty(J2 + — (X x(T z {(Jx(Ty(Tz + '^iTxyTyzTxz (TxTy^ " — T . (210) The three roots of Eq. they are the maximum shearing stresses or the principal shearing stresses. and n with the help of the auxiliary relationship P ^ m^ { n"^ = 1.^y — Ty^ " T ^^ff (TyTxz ~ (^ zT xy) = ci. MechaHcal Fundamentals n. with respect to the original x. [Chap. and n be the direction cosines for the plane with respect to the three principal axes. (29). m. m. and n equal to zero. n cannot all eqtial — Txtj cr — cr„ — T. the shearing stress acting on the same plane r^ is given by = S^  <j^ = aiT + as^m^ { a^V  (cxiP cr^m^ + a^n^)^ (213) <vhich reduces to ^2 ^ (^^ _ o2)2^2m2 + (ai  asyPn^ + ((72 as)hn^n^ Values of t for the three particular sets of direction cosines listed below are of interest because they bisect the angle between principal axes. and 03 each in turn into the three equations of Eq. z axes. of the coefficients of I. S' = Sx"" + + Sy^ + + Sz' = ai'P is + a^'m' + az'n' (211) The normal stress a acting on this plane SzTl given by a = Sxl SyM = (7l/2 + 02^2 + +  (7371^ (212) Therefore. since m. two of the three Therefore. it is necessary to substitute oi.
 cTi . it is n has the largest value of shear stress and called the maximum shear stress ci — O'S (215) The maximum shear forming operations. 26] Stress and Strain Relationships for Elastic is Behavior 27 Since according to convention ai the algebraically greatest principal normal stress and 03 is the algebraically smallest principal stress.— Sec. I oi . stress is important in theories of yielding and metalFigure 26 shows the planes of the principal shear 0\ _ > •mnv ~ _ *? "~ tr.
Thus. Fig. way for a of visualizing the state of stress." 2d ed. The effectiveness of biaxial. Nadai. 28e shows that. k^2 / f Fig. New York. 27. However. if compressive stresses are applied lateral to a tensile stress. While the only physical significance of Mohr's circle is that it gives a geometrical representation of the equations that express the transformation of stress components to different sets of axes...28 Mechanical Fundamentals [Chap. 1950. Mohr's circle representation of a threedimensional state of stress. the maximum is not decreased from what it would be for uniaxial tension.and triof planes shear stress axialtension stresses in reducing the shear stresses results in a consider able decrease in the ductility of the material. Hovf ever. For the limiting case of equal triaxial tension (hydrostatic tension) Mohr's circle reduces to a point. the maximum shear stress is larger than for the case of is 1 A. the maximum shear stress is reduced appreciably. If a tensile stress is applied in the third principal direction (Fig. 2 fall It can be shown ^ that all possible stress conditions within the body within the shaded area between the circles in Fig. although if only the twodimensional Mohr's circle had been used. 28d). 9698. 28c) results in a decrease in the principal shear stress on two of the three sets on which a principal shear stress acts. pp. Note that the application of a tensile stress ai at right angles to an existing tensile stress ax (Fig. Inc. and there are no shear stresses acting on any plane in the body. McGraw . because plastic deformation produced by shear stresses. of Solids. brittle fracture is invariably associated with triaxial stresses developed at a notch or stress raiser. it is a very convenient Figure 28 shows Mohr's circle number of common states of stress. this would not have been apparent. Hill "Theory of Flow and Fracture Book Company. 27.
. U) {c] ^mox ~ ^2 ~ ^3 Fig. (d) triaxial tension (unequal). (e) uniaxial compression. biaxial tension. Mohr's (6) circles (threedimensional) for various states of stress. state of stress. c^jj^ cr. uniaxial tension plus biaxial compression. either uniaxial tension or compression. = 2 cr. 28. 33. (c) (a) Uniaxial tension. = 202 = 203 cr. of metals.2 29 (C) 7max = r2=r. Because of the high value of shear stress relative to the applied tensile stress the material has an excellent opportunity to deform plastically without fracturing under this Important use is made of this fact in the plastic working For example. 26] Stress and Strain Relationships for Elastic Behavior .Sec. greater ductility is obtained in drawing wire through a die than in simple uniaxial tension because the reaction of the metal with the die will produce lateral compressive stresses.
. (a) Pure shear (plane stress). and compression.ax = ci. Figure 29a illustrates a twodimensional state of pure shear. but occurs only on the set of planes parallel to the On the other two sets of planes the principal shear stress Note that for threedimensional pure shear two out of the three sets of shear planes have a value of shear can be obtained Tn. the axes must be rotated 45° in space or 90° on the Mohr's circle. (6) equal shear stresses.30 Mechanical Fundamentals state of stress is [Chap. An identical state of stress to pure when equal tensile applied to a unit cube (Fig. stress Txii. and compressive stresses are Once again Xmax = cri. is rxi//2. 29. biaxial tension Equivalent pureshear conditions. but to obtain complete identity with a state of twodimensional pure shear. 296). The Mohr's circle for this state of stress shows that the maximum and minimum normal stresses are equal to the shear stress and occur at 45° to the '/x im '// K ^min~ '// max" '//C '// (a) cr^ay =cr. 2 An important Mohr's stress is circle for pure shear. The maximum shear it stress is equal to the applied shear z axis. Kb) Fig. This state of readily obtained by twisting a cylindrical bar in torsion.
X. For simpUcity in illustration. 210 considers dy ^ '^Ty'^y I I M I L I dy . Fig. of the entire body be geometrically compatible. it can be seen that the x component of the displacement of to K' is the displacement of J. After strain is applied. y. given by u.JM _ [dx \ (du/dx) dx] — dx dy dx dy \ _ du " di _ dv /q ir ^ ^^ ^^^ Similarly (dv/dy) dy  JM dy dy (2166) . V. Description of Strain at a Point The brief description of linear strain and shear strain given in Sec. and w in the directions x. Fig. or planestrain. is given by K ^^ _ ~ J'K' JK J'M' — JK ~ _ . v. v. and therefore the partial derivatives w with respect to x. y.*%. of u. It is. 27] Stress and Strain Relationships for Elastic Behavior 31 27. given by (du/dx) dx. a simple matter to generalize the relationship obtained from this figure to the threedimensional case. e^. Let the coordinates of a point in an unstrained rigid body be defined by and z. 19 can be expanded into a more generaHzed description of the strain at a point in a rigid body.i. The unit linear strain in the x direction. y.Sec. occupy the same point u. plus the rate of change of u along the distance dx. ment components In order to satisfy these requirements. all the deformation is however. if This can be accomplished their gradients with respect to x. the displaceand w must vary continuously from point to point. y. Components of strain for plane strain. and z and In Fig. 210. have no discontinuities. condition where confined to the xy plane. and z enter into the analysis. and z. a twodimensional. the point will undergo displacements In order that the displacement u. 210 for the planestrain condition.__^^ dx dx . it is two particles in space or that necessary that no no voids be created within the body.
ey. > 63 are the three roots of the followcii. op. 75^^. In complete analogy with stress.32 Mechanical Fundamentals if [Chap. 23. 221227. As with linear strains at a point in the body are given by the values of the principal strains. it (du/dy) dy dy + (dv/dx) dx du dv dx ^di^di /n icj\ ^^^^^^ can be shown that du dw dw ^ ^Tz^Tx dy /o ic \ (216e) Thus.. the linear strain on any plane in a twodimensional situation can. 2 and the z direction were being considered. Methods similar to those used in Sec. 210 it ^. for the small strains involved. for the normal and shear stresses on any plane through a point in the body. from Eq. be expressed by es = ex cos^ 6 ei \ Cy sin^ 6 62 \ 72y sin 6 cos d (217) The ' three principal strains > cit. These axes are For an isotropic body the directions of the principal strains coincide with those of the principal stresses. these equations may be obtained much more easily by replacing a and r in the stress equation by e and 7/2. at a point. the largest any rotation and smallest or shearing strains. (22).LK'J'W = IBJ'M' can be seen that Since. 25. op. e^. pp. e.. and in Sec. for the derivation of the equation for the values of the three principal stresses. 7^^. can be used for the derivation^ of similar quantities in terms of strain. 7x2. . 7x^ = /KJM . For a derivation of this point see Wang. six terms are necessary completely to describe the state of strain e^. However. it is possible to define a system of coordinate axes along which there are no shear strains. from Fig.= In the same way. pp. 2 Timoshenko and Goodier. the principal strain axes. 2627. the tangent of the angle equals the angle.^ An ele ment oriented along one of these principal axes will undergo pure exten sion or contraction without principal stresses. = ^ + lAJ'K' (216c) The shearing strain 7^^ at J' is given by the change in angle of the two elements originally parallel to the x and y axes. For example.
z) = (218) following through with this analogy. can be accurately determined with a simple Wheatstonebridge circuit. the wires in the gage are strained and their electrical resistance is altered. + [e^Cy + e^e. These gages wire resistance gage. dy. (214). stability. Therefore.: Sec. + tL + ylz)]e + Mi^xll^ + e. 28] Stress and Strain Relationships For Elastic Behavior 33 ing equation [obtained from Eq.lx. and ease of application make resistance strain gages a very powerful tool for strain are determination. dz. unit of original volume. . the volume strain 61) (1 given by _ " = which for small (1 + + 62) (1 63) dx dy dz dx dy dz + — dx dy dz (1 f ei)(l +e2)(l +63)  1 strains. or cubical dilatation. strain. (210)] e^  (fx + ey \ e. + e^e. experimental measurements of stress are actually based on measured strains and are converted to stresses by means of Hooke's law and the more general relationships which are given The most universal strainmeasuring device is the bondedin Sec. gages can be oriented in these directions and the principal . 7i Tmax 73 = = = €"2 — = f's 72 ei fl  63 (219) — 62 The volume dx.)e'^ By e^eye.7L + ^ztL . made up of several loops of fine wire or foil of special composition. is the change in volume per Consider a rectangular parallelepiped with edges The volume (1 in the strained condition is ei)(l + + 62) (1 + A 63) is dx dy dz From its definition.. ei 4.nxzi. it is not possible to measure stress directly. which are bonded to the surface of the body to be studied. the equations for the principal shearing strains can be obtained from Eq..62 + 63 Measurement of Surface Strain Except in a few cases involving contact stresses. comparative ruggedness. which is proportional to strain. 210. The change in resistance. The high sensitivity. the principal directions are known. becomes (220) A = 28. frequently called the SR4 strain gage.}^i(yl. after neglecting the products of strains. When the body is deformed. For practical problems of experimental stress analysis If it is often imporstresses tant to determine the principal stresses.
Typical straingage rosettes. r 2 . (6) delta. Rectangular. 2 \a} Fig. id] (a) 211.34 Mechanical Fundamentals [Chap.
Cy. Values of the principal strains are determined by the intersection of the circle with the to the gage a is new x axis through 0. = Ee. /3. Therefore. Appl. 209. and C on the circle give the values of e and 7/2 (measured from the new x axis through 0) for the three gages. 1945. 211 will be found in Table 22. A209. A. In constructing a Mohr's circle representation of strain. Cb. Stress Analysis. While a tensile force in the » x direction produces a linear strain along that axis. Proc.. C. 3. while shear strains must be determined indirectly. Exptl. 2. and Cc are available for three gages situated at arbitrary angles a and tation of the principal strains ei to determine the magnitude and orienand ez. (217) which can be solved for e^. Construct a circle through A. 1. In the same The objective is way. This is Hooke's law simplest form. /. vol. (221) where E is the modulus of elasticity in tension or compression. Straingage read ings at three values of 6 will give three simultaneous equations similar to Eq. values of linear normal strain e are plotted along the x axis. 29] Stress and Strain Relationships for Elastic Behavior 35 readings ©f linear strain. McClintock. and Cc. and the shear strain divided by 2 is plotted along the Figure 212 shows the Mohr's circle construction^ for the genery axis. and cc corresponding to the strains Ca. 5. Straingage readings Ca. Along an arbitrary axis X'X' lay off vertical lines aa. Soc. Mech. 211. (218) can then be used to determine the principal strains. p. 12. 1951. A more convenient method of determining the principal strains from straingage readings than the solution of three simultaneous equations in three unknowns is the use of Mohr's circle. and AD. onehalf the angle AOP on the Mohr's circle {AOP = 29.Sec. 9. The center of this circle is at 0. StressStrain Relations In the first chapter it axial strain in its by means of the was shown that uniaxial stress is related to unimodulus of elasticity. From any ])oint on the line bb (middle strain gage) draw a line DA at an angle a with bb and intersecting aa at point A. bb. B. it is fixed angles in the the usual practice to use three strain gages separated at form of a "rosette. Murphy. The angular relationship of ei 26). Equations for directly converting straingage readings into principal stresses for the two rosettes shown in Fig." as in Fig. and yxy The twodimensional version of Eq. G. alized straingage rosette illustrated at the top of the figure. determined by the intersection of the perpendicular bisectors to CD 4. p. (T. Points A. lay off DC intersecting cc at point C. vol. Cb. and D. F. .
es = 0.33 has been presented by F. = .36 it Mechanical Fundamentals [Chap. [o2 — v((Ti + + 03)] (223) ^ p. R. Poisson's ratio is 0. the strain along that axis is not zero (unless ci = —02). The v. McGrawHill Book Company. (o"2 {223a> 63 = ~ E^'^^ ~^ ^^^ Note that. Shanley. effect of the principal stresses acting along the other 1 V V = and for the other ^ [ol  J'(0"2 + org)] two principal axes ^2 = = p. 62 [(1 v)<r2 wi] ' A derivation which intuitively suggests that v = 0.^ (222) Only the absolute value values^ are closer to 0. = 0. (223) reduce to ei = = ^ (oi — W2) vari) ^2 p. . even when the stress along the third axis is zero.25 for a perfectly isotropic elastic material. 2 also produces a contraction in the transverse y and z directions. of v is used in calculations. In the case of plane strain.33. and the strainstress relationships become ei = = —p^ p— —+ 1 [(1 — j/)(Ti — W2] (22^^^ . Inc. New York. 138139. denoted by the symbol Cy == e^ = PC:. "Strength of Materials. ^3 ks — 03 v((ri 02)] For a biaxial state of plane stress. but for most metals the The generalized description of Hooke's law says that for a body acted upon by a general stress system the strain along any principal axis is due to the stress acting along that axis plus the superimposed strain resulting from the Poisson two axes. 1957. and Eqs.." pp. ratio of the strain in the transverse direction to the strain in the longi tudinal direction is known as Poisson's ratio.
or the modulus of rigidity. related to shear strain by relationships similar to Eq. G. 29] Stress is and Strain Relationships for Elastic Behavior 37 Shear stress (221). Three material constants E.Sec. and v are involved in the description It will of the elastic behavior of materials. so that for an isotropic material there are only two . be shown that these con stants are related. Txy = Gy^y T„ = Gy^^ Ty^ = Gyy: (224) The proportionality constant G is the modulus of elasticity in shear.
(223) produces ^V _ _ ^X _ 1 ^V + n E V _ *~~ ^ ^ + E V Equating the two expressions for tan above equation gives (7r/4 — 7/2) and substituting the with the generalized form of Eq.0 4. Rewriting Eq.31 0.31 0.33 0. The constant k is the volumetric modulus of or the hulk modulus.0 29.5 6.8 0.5 22.0 11. 106 psi Poisson' ratio Aluminum Copper alloys Steel (plain carbon and lowalloy) Stainless steel (188) Titanium Tungsten 10.28 0. Comparing this relation the general relationship between the three elastic constants. .0 28. Shear modulus.0 58. (224) results in Typical values for a number of materials are given in Table 21. This can be _arn " _ ~ A ~ E^ 3(1  (229) 2v) elasticity. 2 Substitution into Eq. (223)] results in ei + 62 + 63 = —^— is (oi + 02 + o's) (227) The lefthand term of Eq.0 17. 38 Mechanical Fundamentals [Chap. (227) results in A where am written as is S'^™ E (1  2p) (228) the average of the three principal stresses.33 0.5 16. G = 2(1 E + 21 (226) v) Table Typical Roomtemperature Values of Elastic Constants FOR Isotropic Materials Modulus Material 10e psi of elasticity.0 9. (227) the volume strain A..27 The addition of the three equations giving the strain produced by the three principal stresses [Eq.0 6.
)(!  2u) ^ + TlTv '' + lf/. + .)(i  ^ + IT".) vE 2. E = (l + .— Sec. (232). Calculation of Stresses from Elastic Strains The general equations expressing stress in terms of strain are consider ably more complicated than the equations giving strain in terms of stress The simultaneous solution of these equations for oi. Usually the methods described in Sec. equal to the hydrostatic pressure. A shortcut procedure is to use a straingage rosette with three gages at fixed orientations for which relationships have been worked out between the strains measured by each gage and the principal stresses. (223) simultaneously for ai and I ci = = :j^^ (ei ' + + 1^62) E 1 (232) 5 02 (^2 vei) These equations can be very useful gage is for determining the values of princi pal stress from straingage measurements. 210] Stress is and Strain Relationships for Elastic Behavior 39 The bulk modulus therefore the ratio of the average stress to the vol ume strain.)a2. ai. corrections for lateral strains must be made by using Eqs.e^. Table 22 gives the relationships between the principal stresses and the .e2 \. Note that even when a it is strain oriented in the principal direction elastic not correct to multiply the strain by the modulus to get the stress in that direction. .) ^ ^ (230) E "' ~ 0"T". Because of the Poisson effect. (223)]. By using this constant the above equations can be simplified to (71 (72 (73 = XA = XA = XA = may f f 2Gei 2Ge2 2Ge3 (231) + 0) For the case of plane stress (a^ relating the stress to the strains two simple and useful equations be obtained by solving the first two (72.p)(l — 2v) is called Lame's constant. It is frequently evaluated for conditions of hydrostatic comis pression. of Eqs. 28 will have to be employed to determine the values of principal strain before these equations can be used. where o„. The term X = vE/(l \. '' where A = ei {. and [Eqs. (73 results in ''' _  vE (1 . 210.
P5 0^ tf + + + iH m t Eh ^ o > HI + + o CC (M I m S 03 H .2 "If & =^ C > ^ + + Z si + 6q IcN &q n CO .
The elastic constants 1 know symmetry from the strain in the most 6 strain components and C. *Si2. New York. necessary to 36 elastic coefficients.y are the elastic compliances. . C. Cij. • . and H. many materials that the material can be considered isotropic. The and v. A similar set of six equations relates the stress to the strain in Cl2. "The Strain Gage Primer. cTx = Ciiex + + CuCy CiiCy \ Ci^ez j Ciaxy + + Cibyyz + + Ci&yzx (234) Txy = C ilCx + Ci^Cz \ Cnyxy Cihjyz C*46T2X Thus. the elastic properties vary with direction. >S.Sec. Jyz 7zx = = = = = = SllCTx S2lO'x SsiCTx Suffjc SblOx SeiCx + + + + + + Sl2(^y S^lCy S320^y Si2Cry Sh20'y S62cry f Sn(72 S230'z f SuTxy SiiTxy »Sl34Txy + + + + + . terms of the elastic coefficients Cn. 211] Stress and Cc Strain Relationships for Elastic Behavior 41 strain readings €«." McGrawHill Book Company. . 1955.. . ^x €y Cz 7x2. R. Generalized StressStrain Relationships stressstrain relationships given in the previous two sections conThese are the only material constants needed to describe the elastic stressstrain behavior provided However. A metallic single crystal is an extreme example of an anisotropic elastic material.e. . in the usual case with engineering materials the grain size is small enough and the grains are in a sufficiently random arrangement so that the equations based on isotropic conditions may be used. S^^a^ + + \ *S430'j j SuTxy SbiTxy S^iTxy Sb^O'z SeaO'z + + + + + + + SnTyz S^bTyz \ SieTzx 'S26Tzi S^bTyz Si^Tyz Sf. However. in order to calculate the stress it is general circumstances. 21 1 . G. Hooke's law can be written in completely general terms as a linear relationship between strain and stress. are anisotropic i.5Tyz SebTyz + + + + + Ss^Tzx S^^Tsx /o oo\ S^GTzx SeeTzx The constants ^Sn. In order to consider the possibility of elastic constants which vary with orientation in the material. . and for the three gages in the rectangular and delta rosettes shown in Fig.. tain three elastic constants E. Perry Inc. while coldworked metals may also exhibit anisotropic elastic behavior. considerations can reduce considerably the number of necessary constants. Lissner. 211. The derivation of the equations and graphical solutions for these equations are discussed in texts^ on strain gage technology. Note that these equations indicate that a shear stress can produce a linear strain in an elastically anisotropic material. ei>. Fortunately.
the compliance S^i is the reciprocal of exist *J13 the shear modulus. the maximum number of constants that need to be considered is 12. even for the least symmetrical crystal structure (triclinic) the number all of independent constants is reduced to 21. Thus. 2 with unequal subscripts are equivalent is when the order of the subscripts reversed.50 Sn Su Su Sn Sn 1. Eqs.24 0.52 1. Sij — Sji Ci \jji Therefore.03 2.49 0.14 0.49 0. For metals which 'Jll in one of the Ssi cubic crystal O44 structures *^55 = *J22 = AJ33.26 2. *J12 = = S2I = 023 = = 7x2/ 032.82 0. thing for the z direction.33 012 012 012 012 0. for a material with a cubic crystal structure there are three independent elastic constants. and that Szi represents the same Also.77 +0. (233) can be written as ex = SnCx + Su(Ty + + Snf^z Jxp = — SuTxy (235) SeeTzx 62 = Si\(^x + Sz2(^v SzZ(^z Izx By comparing that *Sii is these equations with Eqs. that S21 determines the component of linear strain produced in the y Table 23 Elastic Compliances for Metal Crystals Units of 10~^2 cm^/dyne Metal Sii ^12 s.28 0. which have crystal structures of relatively high symmetry.07 0.62 0. For an isotropic .58 0.66 6.23 0.80 0. 013 S3 Aluminum Copper Iron Iron (polycrystalline) Tungsten Magnesium Zinc 1. (223) and (224) it can be seen the reciprocal of the modulus of elasticity in the x direction. and — = 066.64 0.42 Mechanical Fundamentals [Chap.98 2. Therefore.80 1. rial Therefore.48 0. and (224) is apparent. (235) Cx Qy 62 may be written as = Siiax = Snay = 8\\Crz + + + >Si2(o"j^ *Sl2(0'2 + + f (Tz) = SuTxy (236) <Tx) (Ty) lyz *Sl2(o'x Izx — SiiTyz = & \^T zx (223) The identity between these equations and Eqs. Eq.66 direction due to ax.11 3.59 1. For metals. while for a truly isotropic mate there are only two independent elastic constants. equivalent to v/E.
212] Stress and Strain Relationships for Elastic is Behavior 43 material the relationship between these constants 044 given by = 2(oii — 012) Table 23 lists some typical values for elastic compliance. As with strength satisfy the conditions of equilibrium. the ordinary methods of analysis of strength of materials. Forces acting on volume element. = ^^• = dx + dTru = (237) l"+^"+«° from the consideration of the weight of the body.^^Bdy)dx dy dTxy >^^'^/ dy dy) dy aj^dy^ Tjrydyy dy dx TxydX ay *(o>+^<i^)^/ dx Fig. where p is the mass per unit volume and g is the acceleration of gravity. requirement for a solution is to Figure 214 illustrates the forces Taking the acting on an element of the body for a planestress situation. Theory of Elasticity of elasticity entails a The mathematic theory ation of the stresses and strains in a loaded member than more detailed consideris required by icr. 214. arrived at by strength of materials are usually simpler by the assumption of a strain distribution which satisfies the physical situation but may not be mathematically rigorous. 21 2. The solutions made mathematically in the loaded member In the theory of elasticity no simplifying assumptions are of materials. the first made concerning the strain distribution. summation of forces in the x and ij directions results in 2P.Sec. By applying the above relationship compare the difference in isotropy between copper and tungsten. The above equations constitute the equations of equilibrium for plane The term pg arises .
Eq. (a) twice with respect to y. it can be considered that accompany these stresses are compatible and that the body will be preserved. between stress and external load must be such that at the boundary of the body the stresses become equal to the surface forces per unit area. chap. Physically. Equations (237) must be throughout the body. . it must satisfy the boundary conditions. op. (238).. because they are expressed in terms of two displacements u and v.e. (c) with respect to both x and y results in dy^ ax ox dy Equation (238) is the equation of compatibility in two dimensions. 9 . Note that these equilibrium equations do not give a relationship between Instead. (239). cit. dv / V (C) These three equations show that there is a definite relationship between the three strains at a point. this in no gaps occur means that the stresses must vary the material. Differentiating Eq. 27). the relationship of the stresses at any point in the body. '^ = Ty = du Tdy ^^^ Txy +— dx . they give the rate of change the stresses and the external loads.. One of the important requirements of the theory of elasticity is that is the deformation of each element must be such that elastic continuity preserved. 2 tions. Mechanical Fundamentals [Chap. However. The equations of compatibility so that for the twodimensional case can be obtained from the definitions for strain in terms of the displacements u and v (Sec. (223) and substituting into Eq. ay. (h) twice with respect to x. du . W . the strains which continuity of the ^ and r^cy satisfy Eq.44 stress. Timoshenko and Goodier. The equation of compatibility and (224) can be expressed in terms of stress by differentiating Eqs.^ For a threedimensional stress system there will be three equaeach containing one partial derivative of the normal stress and satisfied at all points two partial derivatives of the shear stresses.'W' ^ ^' ' "^' ^ If ^ '^ '^ didy ^^^^^ values of o^. i. they are compatible with each other and the continuity of the body is preserved. and Eq. If the strains satisfy this equation.
and the methods of the theory of elasticity must be used. r^y in terms of the external loads that satisfies the equations of equilibrium [Eqs. (240) boundary conditions of the problem. the stresses are given also satisfy the problem. the equations of compatibility [Eq. In order to satisfy the compatibility equation. Southwell. 212] Stress and Strain Relationships for Elastic Behavior 45 Basically. (240) are differentiated and substituted dx* If 4 = + 2— .12 a stress function can be found for the problem which satisfies Eq. far Generally. and the boundary conditions. "Relaxation Methods in Theoretical Physics. requires the determination of an expression for the stresses <Ty. Continuity is main tained by local plastic yielding. m ^^y . In other problems. localized yielding is important.f— "^ " dy'dy . (237)].Sec.^^j (240) Equations (240) satisfy the equations into Eq. 1862. Airy. One method which of elasticity is is used for the solution of problems by the theory to find a function $ in x and y which satisfies Eqs. Brit.. by Eqs. V. Airy^ showed that stresses at there will always exist a function in x and y from which the any point can be determined by the following equations: ^ = ^ pgy ^^ = ^. the solution of a problem with the theory of elasticity (Tx. New York. continuity of deformation is In solutions by strength of materials not always satisfied. G. (237) and (239) is and also expresses the stresses in terms of the loads. Rept. Assoc. (239)]. and therefore only a fairly limited number of problems have been solved by this technique. (241) 3. this does not have an impor tant effect on the solution because the effects of yielding do not extend beyond the region where they occur. 1946. For problems with complicated geometry and loading systems it is usually necessary to use finitedifference equations and relaxation methods." Oxford University Press. Sci. (241) and the boundary conditions is usually not easy. elasticity arise out of the necessity for satisfying the requirement of continuity of elastic deformation. (240) provided that Eqs. of equilibrium. Such a solution generally involves considerable matheMost of the complications attending the theory of matical agility. Such a function usually called the Airy stress function.for the solution of the (241). R. B. The discovery of a stress function which satisfies both Eq. Eqs. such as the determination of stresses at geometrical discontinuities (stress raisers). Advance.. (239). 1 ^ .
Stress Concentration A geometrical discontinuity in a body. results in a Thus. such as a hole or a notch. Kt = ^nominal (242) localized condition of biaxial or triaxial stress. . Stress distributions due to (a) circular hole and (6) elliptical hole. Figure 215a shows a plate containing a circular hole which is subjected t t LA. With the hole present. uniformly distributed over the cross section of the plate and it would be equal to the load divided by the crosssectional area of the plate. the distribution is such that the axial stress reaches a high value at the edges of the hole and drops off rapidly with distance away from the hole. a stress concentration occurs at the discontinuity. a notch also creates a For example. If the hole were not present. The stress concentration tration factor Kt. nonuniform stress distribution at the vicinity of the disconAt some region near the discontinuity the stress will be higher than the average stress at distances removed from the discontinuity. oj). the stress would be to a uniaxial load. pp. 2 213.46 Mechanical Fundamentals [Chap. for the an axial load. or stress raiser. cit. From elastic analysis. a radial stress is produced as well as a longitudinal stress. ~fd \ \ \ \ {a) Fig. is expressed by a theoretical stressconcen Generally Kt is described as the ratio of the maximum stress to the nominal stress based on the net section. tinuity.^ the stresses produced in an infinitely wide plate containing a circular hole and axially ^ Timoshenko and Goodier. circular hole in a plate subjected to In addition to producing a stress concentration. [b] 215. although some workers use a value of nominal stress based on the entire cross section of the member in a region where there is no stress concentrator. 7881.
Neuber. Inglis. Another interesting case concentration is for is which an analytical solution for the stress available^ the case of a small elliptical hole in a plate. a very narrow hole. 1. 1946. Stressconcentration factors for practical prob lems are usually determined by experimental methods. such as a crack. W. 2 H.. 1951. & Sons. E. Therefore. "Handbook on Experimental Stress Analysis. the ends of the hole is The maximum stress at given by the equation K^ + ^O Note that. Figure 2156 shows the geometry of the hole. "Photoelasticity. Much of this work has been compiled by Neuber. Naval Architects. Therefore. 3(7 stress occurs (Te = (Tmax (244) where a is the uniform tensile stress applied at the ends of the plate. Trans. Edwards. (244). Inc. a compressive stress of equal of the hole at point magnitude exists at the edge B in a direction perpendicular to the axis of loading in the plane of the plate.^ who has made calculations for Eq. when a tensile stress is applied to the plate. Frocht. for a circular hole (a (245) = b).. 1913.Sec." John Wiley M. M. Inst. Figure 216 shows typical 1 C.^ Photoelastic This method is analysis^ of models is the most widely used technique. J. The is (To theoretical stressconcentration factor for a plate with a circular hole 3. Inc. although it is possible to make threedimensional photoelastic analyses. Inc. * New York. normal to the tensile direction will result in a very high stress concentration. pp. '' Ann Arbor. 21 3] Stress and Strain Relationships for Elastic Behavior 47 loaded can be expressed as Examination at point of these equations Q A when = 7r/2 and r = = shows that the maximum For this case a. . therefore equal to Further study of these equations shows that = —a for r — a and = 0.. Mich. Publisher. 219230. Hetenyi. the above equation reduces to Equation (245) shows that the stress increases with the ratio a/b." Enghsh translation. M. especially applicable to planestress problems." John Wiley & Sons. various types of notches. 1950. pt. "Theory of Notch Stresses.. Mathematical complexities prevent the calculation of elastic stressconcentration factors in all but the simplest geometrical cases. New York.
4 P\ *l 2r 1 .ap.o 3.48 Mechanical Fundamentals [a.4 0.0 o.2 (a) 0. 2 t 6' 0.8 1.6 0.
Sec. 213] Stress and Strain Relationships for Elastic Behavior 49 CO ?.^ .
Thus. . a ductile metal loaded statically will not develop the full theoretical stressconcentration factor. 214. redistribution of stress will not occur to any extent in a brittle material. . if the material is sufficiently ductile. (Ti = 2c7i — — — = (72 — Cs ' I' (J. Spherical and Deviator Components of Stress and Strain Experiment shows that materials can withstand very large hydrostatic pressures (spherical state of stress) without undergoing plastic defor mation.. 50 Mechanical Fundamentals [Chap. particularly in the theory of plasticity. Inc. & Sons. in In many problems. therefore a stress concentration of close to the theoretical value will Although stress raisers are not usually dangerous in ductile appreciable stressconcentration materials subjected to static loads. plastic defor is exceeded at the point of maximum Further increase in load produces a local increase in strain at the Because of strain critically stressed region with little increase in stress. is This is known as the stress deviator a'. 2o3 01 — 02 0^3 03 0"3 ^ It 1 can be readily shown that + 03 + o's 0. until. Much of the information on stress concentrations in machine parts has been collected by Peterson. the stress distribution becomes essentially uniform. Peterson. E. O"! = = = 01 — — — (j\ . and result.. 02 02 = 2 o2 01—03 o K^^n /9 A^\ . "Stressconcentration Design Factors. 1953. New R. 12. or hydrostatic. mation occurs when the yield stress stress. 2 curves for the theoretical stressconcentration factor of certain machine elements that were obtained by photoelastic methods. hardening. it is desirable to designate the part of the total stress which can be effective producing plastic deformation." John Wiley York. The other component the spherical. the stress increases in regions adjacent to the stress raiser.. In a ductile material. Stress raisers are very important in the fatigue failure of metals and will be discussed further in Chap. component of stress c" a" = "' + ""' + "' ^ p (246) The stress deviator is given by the following equations: . effects will occur in ductile materials under fatigue conditions of alternating stresses. However.. The effect 1 of a stress raiser is much more pronounced in a brittle material than in a ductile material.
.
(223) and (224)] for the strains in the above expression results in an expression for strain . By the same reasoning. Essentially all the work performed during elastic deformation is stored as elastic energy. the strain energy per ment subjected to pure shear is given by Uo unit volume of an ele = V2T. and this energy is recovered on the release of the load. so that the average energy is equal to onehalf of their product. Since dA dx is the volume of the element. Uq.){dA dx) (256) Equation (256) describes the total elastic energy absorbed by the ment. Strain Energy is the amount of energy expended by the action deforming an elastic body.yy.dx) = y2{(T^e.y = l%= V^llyG (258) The relationship for pure uniaxial deformation and pure shear can be combined by the principle of superposition to give the elastic strain energy for a general threedimensional stress distribution. Energy (or work) is equal to a force times the distance over which it acts. This is also equal to the area under the loaddeformation curve. 2 21 5. f/o = y2{<yxex + Oyey + a^e^ f Txylxy + r^^^xz + ry^iy^ (259) Substituting the equations of Hooke's law [Eqs. the elastic strain energy given by U = KP5 = ma. In the deformation of an elastic body the force and deformation increase linearly from initial values of zero. The elastic strain energy of external forces in U = For an elemental cube that sc yiPb (255) is subjected to only a tensile stress along the is axis. is ele'per given by Uo = y2CTA = l^ = Vze. the strain energy unit volume.52 Mechanical Fundamentals [Chap.'E (257) Note that the lateral strains which accompany deformation in simple tension do not enter in the expression for strain energy because no forces exist in the directions of the strains.dA){e.
Southwell. Inc. 1953. v. A. New York.' + e. C/o expressed in terms of strains and the = V2\^' + G{e. N. Goodier: "Theory of Elasticity. dUa/de^ = XA + 2Gej. 1941. (259) the stresses are elimiis nated. McGrawHill Book Company.." 2d ed. P." McGrawHill Book Company. (231) into Eq. . Thus.Sec..: "Applied Elasticity. H. New York. and J.. = a^ [compare with Eqs. (231)]. R. + Also. Timoshenko. Treatise on the Mathematical Theory of Elasticity. 1951. Oxford University Press.: "An Introduction to the Theory of Elasticity.~) + ^Giyl + tL + ^D (261) any note that the derivative of Uq with respect to component gives the corresponding stress component. S. It is interesting to strain BIBLIOGRAPHY Love.." 4th ed. Wang. E. ^ (jI + tI + tD (260) by substituting Eqs.: "A Dover Publications." 2d ed. 21 5] Stress and Strain Relationships for Elastic Behavior 53 energy per unit volume expressed solely in terms of the stress and the elastic constants. 1949. New York. C. and the strain energy elastic constants.' + e.. T. Inc. New York..
because it is necessary to adopt a yield criterion in the theories of plasticity. but for plastic deformation the plastic strain depends not only on the final load but also on the path by which it was reached. elastic deformation depends only on the initial and final states of stress and is independent of the loading path. This aspect of plasticity 1 The determination of the Hmiting load between elastic and plastic behavior is also generally covered in strength of materials. The mathematical description of the plastic deformation of metals is not nearly so well developed as the description of elastic deformation by means of the theory of elasticity because plastic deformation is much more complicated than elastic deformation. The consideration of small plastic strains allows economies in building construction through the use of the theory of limit design. For example. in the plastic region of strain. From the viewpoint of design. there is no simple relationship between stress and strain as there is for elastic deformation. For example. However. The theory of plasticity is concerned with a number of different types of problems. The analysis of large plastic strains is required in the mathematical treatment of the plastic forming of metals. and the overspeeding of rotor disks. The yield criterion^ must be expressed in terms of stress in such a way that it is valid for all states of stress. The designer is also concerned with plastic deformation in problems where the body is purposely stressed beyond the yield stress into the plastic region. shrink fitting. this topic is covered in the chapter on plasticity.Chapter 3 ELEMENTS OF THE THEORY OF PLASTICITY 31 . plasticity must be considered in designing for processes such as autofrettage. plasticity is concerned with predicting the maximum load which can be applied to a body without causing excessive yielding. Introduction of plasticity deals The theory with the behavior of materials in the region of strain beyond which Hooke's law is no longer valid. Moreover. 54 .
crystal structure.Sec. 3la. It is very difficult to describe. This aspect of plasticity considered in Part Two. unlike the situation in the elastic region. in rigorous analytical way. Hooke's law is followed up to some yield [a] Fig. Another aspect of plasticity is concerned with acquiring a better understanding of the mechanism of the plastic deformation of metals. the behavior of a metal under these conditions. the stress and strain are not related by ' See Chap. However. 32J will Elements oF the Theory of Plasticity in 55 a be considered Part Four. Most metals strainharden in this region. Therefore. and to compare them with the theoretical flow curves The true stressstrain curve for a typical ductile metal.) Beyond cro. stress oQ. The purpose of this section' is to describe typical stressstrain curves for real metals for ideal materials. The Flow Curve stressstrain curve obtained is The by uniaxial loading. 9 for a more complete discussion of the mathematics of the true stress strain curve. strain is (The value of oq will depend upon the accuracy with which measured. [b) Typical true stressstrain curves for a ductile metal. so that increases in strain require higher values of stress than the initial yield stress cro. and lattice imperfec tions on the deformation behavior are is of chief concern. True strain is discussed in the next section. . 32. True stress is given of by the load divided by the instantaneous crosssectional area of the specimen. The effect of metallurgical variables. plotted in fundamental interest in plasticity when the curve is terms of true stress a. certain simplifying assumptions are usually necessary to obtain a tractable mathematical solution. Interest in this field is centered on the imperfections in crystalline solids. such as aluminum. as in the ordinary tension test. 31 . the metal deforms plastically. is illustrated in Fig.and true strain e.
type of behavior is approached by a ductile metal which is in a highly coldworked condition. With a little additional plastic strain. the slope of a loglog plot of Eq. perfectly plastic material. in this field it is common practice to devise idealized flow curves which rigid. given maximum load at which Even the simple mathematical expression for the flow curve that is by Eq. 55). Figure 326 illustrates the flow curve for a This behavior is perfectly plastic material with an elastic region. (31) can result in considerable mathematical complexity when it is used with the equations of the theory of plasticity. the stressstrain curve becomes a continuation of what it would have been had no unloading taken place. is simplify the mathematics without deviating too far from physical reality. a small amount of the plastic strain €2 — €3 will disappear with time. Figure 32a shows the flow curve for a For this idealized material. (31). 3 any simple constant of proportionality. The strain decrease ei — eo is the However. Depending upon the metal and the temperature. the strain remaining is not all permarecoverable elastic strain. . The hysteresis behavior resulting from unloading and loading from a plastic strain is generally neglected in plasticity theories.56 Mechanical Fundamentals [Chap. This Generally the anelastic strain is negis known as anelastic behavior. on reloading the curve will generally bend over (Fig. a tensile specimen cro. 316). as the stress approaches the original value of stress from which it was unloaded. completely rigid (zero elastic strain) until the axial stress equals whereupon the material This flows plastically at a constant flow stress (zero strain hardening). 8. Therefore. approached by a material such as plaincarbon steel which has a pronounced yieldpoint elongation (see Sec. Generally the stressstrain curve on unloading from a plastic strain will not be exactly linear and parallel to the elastic portion of the curve Moreover. A more realistic approach is to approximate the flow curve by two straight lines corresponding to 1 Anelasticity is discussed in greater detail in Chap.0 the strainhardening coefficient. Many attempts have been made to fit mathematical equations to this curve. (31) where is K is the stress at e = 1. lected in mathematical theories of plasticity. This equation can be valid only from the beginning of plastic flow to the the specimen begins to neck down. If the metal is strained to point A. The most common is a power expression of the form ^ a = Ke and n. A true stressstrain curve is frequently called a flow curve because it gives the stress required to cause the metal to flow plastically to any given strain. when the load is released the total strain will immediately decrease from €i to C2 by an amount a/E. nent plastic strain.
namely. "Elemente der technologischen Mechanik.Sec. is satisfactory for elastic strains where AL is very deformation the strains are frequently large. Idealized flow curves. gage length. . rather than to the original in plastic However. (11). Ludwik^ first proposed the definition of true strain. and during the extension the gage length changes considerably. 32c). or natural strain. 1909. the elastic and plastic regions (Fig. AL Lo L — Lo Lo _ L Lo e+l e = L = In ^= In (e 1 1) (34) * P. Ludwik. e = yLid = ^ Lio / dL Jlo This definition of strain small. the change in length referred to the original unit length." SpringerVerlag OHG. True Strain Equation (11) describes the conventional concept of unit linear strain. piecewise linear (strainhardening) material. u . 32. which obviates this difficulty. (c) (a) Rigid ideal plastic material. e. (b) ideal plastic material with elastic region. 33] Elements of the Theory of Plasticity 57 Fig. (32) ^dL = ^ L J(33) In The relationship between true strain and conventional linear strain follows from Eq. l^ or ~ Lo L^ Li Ls — L2 L2 . of curve results in 33. In this definition of strain the change in length is referred to the instantaneous gage length. This type somewhat more complicated mathematics. Berlin.
6 = In ^ = In 4^ summation (35) Also. the of the three princi pal strains is equal to zero. = = In (2Lo/Lo) = In 2. To Lo)/Lo = 1. Maximumshearstress Theory The maximumshearstress theory.(T3. the strain produced by extending its the cylinder to twice length. Eq. strain give nearly identical results up to Because the volume remains essentially constant during plastic deformation.58 Mechanical Fundamentals of [Chap. F(ai. or Guest yield criterion. relationships. (33) can be written in terms of either length or area. sometimes called the Tresca. For compression to half the original length. related to some particular combination of the princi form but there is at present no theoretical way of calculating the relationship between the stress components to correlate yielding in a threedimensional state of stress with yielding in the uniaxial yield criterion can be expressed in the general . A . or a achieve the same amount of negative linear is then e = (2Lo — strain in compression. plastic flow begins at the yield stress. tension test. Coulomb. intuitively we should expect that the strain procompressing a cylinder to half its original length would be the although opposite in sign to. because of constancy of volume. there are two generally accepted theories predicting the onset of yielding in ductile metals.a2. yielding Criteria for Ductile Metals The problem of deducing mathematical relationships for predicting the conditions at which plastic yielding begins when a material is subjected to a complex state of stress is an important consideration in the field of plasticity. duced in as. The advantage original length. In [(Lo/2)/Lo] = In M =  34.) = 0. e € For extension to twice the original length. 3 The two measurements strains of about 0. In uniaxial loading. equivalence is obtained for the two cases. same would have to be squeezed to zero However. states that yielding will occur when the .0. The yielding criteria are therefore essentially empirical for At present.1. In 2. and it is is to be expected that yielding under a situation of com bined stresses pal stresses. of using true strain should be apparent from the followis ing example: Consider a uniform cylinder which extended to twice its The linear strain strain of 100 per cent. .ki. If true strain is used. 61 + 62 + €3 = (36) This relationship is not valid for the principal conventional strains. the cylinder thickness.
(7i — 03 .
The development of this yield criterion is associated with the names of Von Mises. the strain energy of distortion. Hencky." McGrawHill York. Maxwell. 33. (312). Subsequent experiments showed that Eq. and Huber. it is valid to assume that only the stress deviator can produce distortion. and a stress deviator. The total elastic strain energy per unit volume (see Sec. (312) on the basis of distortion energy is given below. Another common physical interpretation of Eq. To illustrate the resolution of total strain energy into its components.y + ((72  a^y + ((73  (7i)2]^^ (312) According to this criterion. basis of the distortionenergy concept. W. Inc. Uq. A to the number of attempts have been made to provide physical meaning Von Mises yield criterion. the 1 1 P.. 1952. 215) can be divided into two components. The derivation of Eq. yielding will occur when the differences between the principal stresses expressed by the righthand side of the equation exceed the yield stress in uniaxial tension. a'. Uq. (311). Theory fit A somewhat better with experimental results is provided by the yield criterion given in Eq. (313) primarily because it was mathematithan the invariant form of the maximumshearstress theory given by Eq. and the strain energy of volume change. (312) is that it represents the critical value of the octahedral shear stress (see Sec. Therefore. yielding will occur On the when the strain energy of distortion per unit volume exceeds the strain energy of distortion per unit volume for a specimen strained to the yield stress in uniaxial tension or compression. or Distortionenergy. Book Company. consider Fig. ^2  fc2 = is (313) J2 is the second invariant of the stress deviator. . (313) provides better overall agreement with combined stressyielding data than the maximumshearstress theory. One commonly accepted concept is that this yield criterion expresses the strain energy of distortion. 214 that a general threedimensional state of stress can be expressed in terms of a spherical or hydrostatic component of stress. (70 = ^ V2 [((Ti  a. New "Studies in Large Plastic Flow and Fracture. Bridgman. Von Mises proposed this criterion in the invariant form given cally simpler by Eq. This figure illus trates the point established in Sec. ao.60 Mechanical Fundamentals [Chap. Because experiments have shown that up to rather large values of hydrostatic pressure a hydrostatic state of stress has no effect on yielding. 3 Von Mises. and k the yield stress in pure shear. 37). a".
1 f/o = 2^ (O"!^ + (Xm 02. The strain energy of distortion will be determined by first calculating the strain energy of volume change and then subtracting this term from Oz Og 02 Fig. / ir >203 ^3 Referring to the definitions for the deviator component of strain given in Sec. It repre sents only the strain energy associated with changing the shape of the specimen and neglects the strain energy associated with changes in volume. from Eq. Uo = H<T. 34] Elements of the Theory of Plasticity 61 strain energy of distortion will be based on the stress deviator.Sec. (260) for the total strain energy C/o. the strain energy will 11 per unit volume associated with a volume change jrii be Uo = . <ti — = tro. (<T3  CT. 33. Eq. (316) reduces to C/'O = —^ [(<^l  ^2)2 + (<T2 = a. since = (oi f 02 f (Ts)/3 and k = E'/[3(l 2v)].1.y] (317) For a uniaxial state of stress. 214 and letting o„i equal the hydrostatic component of stress.r 03 + 0.03^) — p (olO2 + 02(r3 "I" Oldi) " 9 12 "^ — (316) However. Referring again to Fig. (314) becomes (315) U'o'^l"^ Since . 33. (229) A = (Xm/n. or average stress. / n n /'201 ei + I i / It n >^02 62 + 1 .(ei + 62 + 63) = V2<rmA (314) However. Resolution of stress into hydrostatic stress and stress deviator. the total strain energy. so that Eq. U'o = Uo — U[' the strain energy of distortion can be determined by using Eq. cr2 = C/'o ^'^o^ (318) .
50(7o.62 Mechanical Fundamentals [Chap. if the distortionenergy theory is a valid yielding criterion.5 and 0. 3 Therefore. 1 + V 2 1 + '^ 2 TO = ^(70 = V3 0. For the combined axial tension and . the ratio for uniaxial stress (318) system yielding begins when the strain energy of between this critical value and pure shear can be obtained by equating Eqs. the distortionenergy yield criterion can be written or <T0 = ^ of [{<Jl  <T2y + (<r2  a. Combined Stress Tests The conditions for yielding under states of stress other than uniaxial and torsion loading can be conveniently studied with thinwall tubes. average occurring close to the predicted value. O"! fO Cr2 = Cj ^ — (To is Therefore. and (320).577(70 (321) Thus.577 times the tensile yield strength. Axial tension can be combined with torsion to give various combinations of shear stiess to normal stress intermediate between the values obtained separately in tension and torsion. the strain energy of distortion for this state of stress given by Uo=^^<^'=^^r^ If for (320) any type of stress distortion reaches a critical value. such as occurs in torsion. should be Actual data show that the shear with the between 0. 35. yield stress falls from a torsion test.6 of the tensile yield stress. t ^= = a.y + ((73  (7l)2]^^ (319) For a condition pure shear. agreement shown by the distortionenergy theory for these two different types of tests is one reason for preferring the distortionenergy theory of yielding. the yield strength in shear. It should be noted that The better the maximumshearstress theory predicts that tq = 0. as determined 0.
3 0.Sec.^ radial direction is negligible (03 = at the outer free surface).1 0. Comparison between maximumshearstress theory and dist9rtionenergy (Von Mises) theory.3 0. this test of yielding provides a biaxial state of stress.9 1.5 0.1 0.4 0. Another type of combined stress test is to subject thinwall tubes to Since the stress in the axial load and internal hydrostatic pressure. the distortionenergy theory can be expressed mathematically by (325) . 0.2 0. (27) are ^1 (T2 = = 2" + (^T + ^j (322) 2 V 4 """7 is Therefore.8 0. Figure 34 shows that the experimental results agree best with the distortionenergy theory. the principal stresses from Eq.5 0.2 0.6 0. For the planestress condition.4 I? ^ 0.7 0. 34.6 0. 35] Elements oF the Theory of Plasticity 63 torsion. the maximumshearstress criterion of yielding given by ^'''^ fey +^ fey =1 and the distortionenergy theory of yielding is expressed by Both equations 1 define an ellipse.0 Fig.
Another contribution of Lode was the introduction of a strain parameter v. Thus. Note that the maximumshearstress theory and the distortionenerg3'^ theory predict the same 35. there should be no yielding. 3 x/^ co and whose minor semiaxis ^/% oq. to account for the influence of the intermediate principal stress. the intermediate principal stress on According to the maximumshearstress law. two yield procedure adopted by determining the effect of plane stress. called Comparison of yield criteria for criteria is the Lode of Lode's stress parameter. biaxial energy theory = The greatest di vergence between the two theories occurs for a state of pure shear ((71 = —02). effect of the value of the intermediate stress 02.3_2g) Aes where Ae is a finite increment of strain. A very sensitive method of dif ferentiating between the Fig. It has already been shown that stress for this state of stress the shearstress law predicts a yield which is 15 per cent lower than the value given by the distorMaximum shear stress theory tionenergy criterion. Lode introduced the parameter n. (312) results in (7 1 — <T0 03 2 (3 + M^) 2\H (327) Experimental data plot much better against Eq. yield stress for conditions of uniDistortion axial stress stress (<Ti and balanced a 2). indicating that the intermediate principal stress has an influence on yielding. V = 2 A62  A63  Aei Aei — . 35. M Solving this equation for 02 = 2(T2 — O"! g"3 — 03 0"] — (326) and eliminating 02 from Eq. (327) than against the maximumshearstress equation. A plot of n against v should .64 This is Mechanical Fundamentals the equation of an ellipse whose major semiaxis is is [Chap. For the distortionenergy theory. (oi — 03) /oq = 1. A convenient is way of comparing yielding criteria for a twodimensional state of stress with a plot such as Fig.
I. and the cosine of this angle is l/\/3.. Ol)]'^ = ^2 —^ (To or oo ~ [(^1  ^^y +  (T^y f  a. Octahedral Shear Stress and Shear Strain The octahedral stresses are a particular set of stress functions which are important in the theory of plasticity." 2d McGrawHill Book Company. (331) is identical with the equation already derived for the distortionenergy theory. Inc. A. The normal octahedral equal to the hydro component of the total stress. the angle the faces and the nearest principal axis is 54°44'. "Theory of Flow and Fracture of Solids. the two yielding theories give the same results. 1 According to this theory. 39). 99105. the octahedral theory can be considered the stress equivalent of the distortionenergy theory. In a sense. . Most metals but systematic deviation from Lode's relationship n = v. In this analogous to the stress de viator.Y + (cr2  (TzY + (C73  (Ti)']^^ (330) Since the normal octahedral stress is a hydrostatic stress. 36] Elements of the Theory of Plasticity if 65 yield a straight line at 45° to the axis the metal behaves according to the LevyVon Mises equations slight show some 36. 1950. pp. [(o"l — 02)^ + (02 — (<^2 03)2 + ((T2 — ((73 ^.y^'' (331) Since Eq. the octahedral shear stress of stress responsible for plastic deformation. of plasticity (see Sec. vol. If it is assumed that a critical octahedral shear stress determines yielding. of between the normal to one a normal octahedral stress oQct and an octahedral shear stress is stress lying in the octahedral plane. the failure criterion can be written as the component it is respect. Nadai. Therefore.Sec. it cannot pro duce yielding is in solid materials. 1 Toot = = r.. static Xoct. They are the stresses acting on the faces of a threedimensional octahedron which has the geometric property that the faces of the planes make equal angles with each of the For such a geometric body. The stress acting on each face of the octahedron can be resolved^ into three principal directions of stress. the octahedral ed. Coot = "^ + "3^ + "^ = a" (329) The octahedral shear root stress r„ct is given by = M[(^l  <J. New York.
y + ((73  a{}T (335) Effective or significant strain ^ = [(ei  62)2 + (62  63)2 + (63  61)^]^^ (336) Note that both effective stress and effective strain reduce to the axial normal component of stress and strain for a tensile test.. . These quantities are defined by the following equations for the case where the coordinate axes correspond to the It is frequently useful to simplify the representation of a of stress or strain by means of invariant functions of stress principal directions: Effective or significant stress ^ = ^ ^ [(<ri  <T2y + ((T2  <x. 8. The octahedral linear strain is given by €oct = by o (333) Octahedral shear strain Toot is given = %[(^l  ^2)' + (62  63)2 + (63  61)^]^^ (334) 37. Invariants of Stress and Strain complex state and strain. Appl. 3 shear stress corresponding to yielding in uniaxial stress \/2 given by Toct= ^(70  0. J. the flow curves obtained in a uniaxialtension test and a biaxialtorsion test of a thin tube with internal pressure will coincide when the curves are plotted in terms of invariant stress and strain functions.471(70 (332) Octahedral strains are referred to the same threedimensional octahedron as the octahedral stresses. stress and strain. These values 1 A. approximately the same curve will be obtained regardless of the state of stress. 1937. 205. Nadai. For example.66 Mechanical Fundamentals is [Chap. Nadai^ has shown that the octahedral shear stress and shear strain are invariant functions which describe the flow curve independent of the type of test. p. If the plastic stressstrain curve (the flow curve) is plotted in terms of invariants of stress and strain. Other frequently used invariant functions are the effective. vol. Phys. or significant.
dcFl (T\ stress components increase _ da 0'2 _ d(Tz O'S /q OQN For this type of loading. Mech. as can be above equations. J. A particular condition of loading which simplifies the analysis tional loading. in terms of the final stress state ' the stress history. 1949. Drucker. Xoct defined below. pp.2 )ec. ing stress Teq. (330) and 0 (334) with the and strain. . Appl. C. D. vol. Basis of the Theories of Plasticity The development of a generalized theory of plasticity. is plotted against octahedral shearing strain Toct instead of being plotted against — Toct 0^:)' (338) where J 2 and J3 are invariants of the stress deviator. is The final value of a plasticstrain component given by the integral of the increments of the plasticstrain component is over the loading history that the material has undergone. Therefore. There appears to be no theoretical or experimental justification for choosing invariant stress and strain parameters other than the closeness of agreement with the data and mathematical convenience. elastic deformation depends only on the initial and final states of stress or strain and therefore the results are independent of the path along which the load is reached. 349357. (337) Toct = —o— \/2 Toct = V2 /e Drucker^ has pointed out that there are a large number of different functions of stress and strain which might serve as invariant stress and For example. The inherent difficulty in developing a simple mathematical description of plasticity lies in the fact While that plastic deformation is essentially an irreversible process. in plastic deformation. wnth the same wide applicability as the theory of elasticity. the strains can be expressed because the final stress state specifies 16. the type of load cycle deter mines the strain increment. he shows that combined stress data for aluminumalloy tubes show better agreement when the eciuivalent shearstrain parameters. in 'propor For proportional loading. in plastic deformation the total plastic strain depends not only on the final load but also on the path by which it was reached. has not progressed rapidly because of the complexity of the problem. the constant ratio to each other. 381 Elements oF the I heory of Plasticity 67 are also related to the octahedral shearing stress seen by comparing Eqs. 38..
. 3 Mathematical theories types.5 for all values of plastic strain. while flow theories relate the stress to the strain rate. 26. and strain are assumed to coincide at all Time effects are usually neglected. Budiansky. Flow theories consider a suc cession of infinitesimal increments of distortion in which the instantaneous stress is related to the increment of the strain rate. tropic. i. This also leads to the useful For the values of stress usually encountered. The formation of many plasticity theories that the stress deviator is is proportional to the strain increment. One way to take strain hardening into consideration is to use experimental data plotted as invariant stressstrain functions. 1959. it is a very good assumption to consider that volume remains constant. but it is is not generally considered to be reliable^ when the direc tion of loading changed during the test. An obvious simplification is to assume that plastic flow occurs at a constant value of flow stress. Appl. no. plastic deformations. pp. J. 2. Deformation theories utilize an averaging process over the entire deformation history and relate the total plastic strain to the final stress. This tj^pe of theory is valid when the material is subjected to proportional loading. of plasticity can be divided roughly into two Deformation theories relate the stress to the strain. there by using an average value of yield stress. so that visco elastic materials are excluded from the theories presented in this chapter.68 Mechanical Fundamentals [Chap. that there is no strain hardening. vol. based on the premise This is equivalent to saying that Lode's stress and strain parameters are equal. . of general is The metal assumptions which are common to all considered to be continuous and iso The principal axes of plastic stress times.. Mech. or the velocity of strain. it is common practice to allow for strain hardening Unfortunately. 61 + €2 + 63 = must increase from Experiments is Constancy its elastic of volume also requires that Poisson's ratio value to a value of 0. 259264.e. In analyses of forming operations involving large plastic strains. Plasticity theory based on ideal plastic behavior has been developed further than theories which consider the strain hardening of the metal. but frequently the incorrect assumption made that = 0. relation that the sum of the principal true strains is equal to zero. it is Because a flow theory considers the instantaneous deformation. in the plastic region is no simple relationship between stress and strain such as exists with elastic deformation. show that Poisson's V ratio increases with progressive plastic strain to this limiting value.5 for the plastic condition. better able to describe large There are a number plasticity theories. ^ Arguments to show that deformation theories of plasticity should be valid for loading paths other than proportional loading have been given by B.
To provide additional simplification to the analysis. the plastic component of the hydrostatic component of strain must be equal to zero. Levy.e. e. and Von Mises. Note that a dot over the symbol for strain indicates the time derivative of strain. Flow Theories Rigid Ideal Plastic Material A by flow theory for a rigid ideal plastic material based on the propor between stress deviator and strain rate is the outgrowth of work Venant. strained only slightly beyond the yield stress. the Levy. the coefficient .^ = er €2^ er = e^ (341) 39. These equations are similar to the equations of viscosity for an incomThe important difference is that for the case of the fluid is the proportionality constant X a true material constant. X is a proportionality tionality St. However.. all elastic strain is neglected. Cx ~ CFm a'y a^ = = = fx — 2Xex Txy Ty^ Txz 2Xey 2X63 = = = ^^jxy Xjy. so that the elastic strains are negligible in many problems the body is when the plastic strain is by comparison. it is necessary that the elastic strains be considered in the analysis. Xjxz (342) In terms of the principal stresses. and the total strain is considered to be n entirely plastic. strain. it is often assumed that the body acts as a rigid plastic material. so that the elastic and plastic strains are of comparable magnitude.Von Mises equations can be written a^ = 2Xei ffo = 2X€2 a"3 — 2Xe3 (343) pressible fluid. it appears that the proportionality between stress deviator and strain increment is a reasonably good approximation.Von Mises equations are given below for a general coordinate system. constant. the plasticstrain deviator e/' equal to the plastic strain. The Levy. The total strain is then the sum of the elastic and plastic ^ij = 4+ 4 (340) However. the strain rate. 39] Elements of the Theory of Plasticity 69 = V.Sec. This is a suitable assumption large. For this situation. Although deviations from Lode's relationship have been shown by experiment. i. With this assumption. and <rm is the hydrostatic component of stress. e^" = H(^/ + = 62^ is + 63^) = Therefore. because of the assumption of constancy of volume.
3 of viscosity. = {a[Y + {a'.Y^ ^' (344) Substituting Eqs. (344) results in X2(6i2 + e^^ + h') = ^' of strain rate.70 Mechanical Fundamentals [Chap. Assuming that . (345) The quantity ei^ + €2^ + > €3^ is an invariant Substituting Eq. For the case of the plastic the vakies of stress and strain.Y + {c'. — ~ — 02 — 03 = 37 dei at 2o2 <^i ~ — <^3 = "^ — at '^^2 (347) Z(T3 (Ti (T2 —r. (345) back into Eqs. the value of X depends on when the yield cri established. must be integrated over a particular stress path or strain path for the solution of a particular problem. (343) gives = [3(6i2 V^ opei (o_Aa\ ^ ^ ' + e^^ H k. plus the constancyofvolume relationship constitute a system of differential equations that + €2 + €3 = 0. terion is X can be evaluated body. The Von Mises yield criterion is given by or 2J. Ci^3 Eliminating Q\/dt from these equations results in 2ai 2a2 2cri 2(73 — — — — (T2 cs 02 Ci — — — — (7z _ _ dei de2 o"! (348) dei C?€3 Cs (72 The above two ei equations.^W a'^ Completely analogous equations follow for Equations (343) can be written Z<Ti and 03. (343) into Eq. between the has been primarily the work of Prandtl In discussing this theory it will be necessary to differentiate plastic strains and elastic strain e^ and the plastic strain e^. ElasticPlastic Material The extension of of the LevyVon Mises equations to the consideration both elastic and Reuss.
a" = 3/ce^" ^ This derivation follows the procedure given by Prager and Hodge.r This expression can be used to eliminate the proportionality constant X from Eq.P' 2^6/' = \a[ 2^62^' = \<j'. (250)] gives the corresponding equations for elastic strain.r f a[(a'. 2G€[ If it is = (Ti + XcTi 2G'e2 = 0^2 + Xo2 26^63 = ^3 "f \a^ (351) there is assumed that the Von Mises no strain hardening. (351). U'o dis = o'ifi + aU'^ + cr'^e's (353) By using Eqs.{a'. (349) and (350) results in expressions for the time derivative of total strain. . = \a'. (352) and (353) and the yield criterion J2 possible to obtain the relationship = fc^ it is 2GUo = The 2XA. However.Sgc. To to remember that ai = a[ get the rate of change of stress. pp. op. (344) J2 = a[{a[r + a'. J2 criterion of yielding applies and that = F J2 = = (352) From Eq. 3L^'o (355) A it is These equations give the rate of change of the stress deviator as long as J 2 = k"^ and £/[. (352) and solving for the stress rate. as U'o is the rate of change of strain energy involved in opposed to the strain energy required for volume change. the quantity Uq is introduced. f necessary a^ .2 (354) stressstrain relations of Reuss's equations are obtained by substi tuting Eq. > 0. (349) The time derivative of Hooke's law expressed in terms of stress and strain deviators [Eq. 39] Elements oF the Theory of Plasticity is 71 the rate of change of plastic strain results in proportional to the stress deviator 2Gi... From Eq. = SG'f^l^^. (251) (356) cit. to simplify the algebra. (354) into Eq.) = 2G (. a'.' tortion. 2729. 2G'6i^' = &[ 2G'62^' = &2 2(?e3^' = &'s (350) Combining Eqs.
"Plasticity. of Ep from an invariant stressstrain curve is shown in Fig. Eqs." pp. . Gp is a plastic shear modulus which varies depending upon the values of stress and strain.72 Mechanical Fundamentals the stress is [Ch ap. (358) and the familiar equations of elasticity expressing strain in terms of the principal stresses [Eqs. (J = 2Gp^' (357) Elastic strains are neglected in Eq. eters. 1 A. Deformation Theories Hencky proposed that for small strains the stress deviator could be considered proportional to the strain deviator. Nadai^ has developed relationships on the equality of Lode's stress and strain paramsimilar to Eqs. elasticity equations like The proper equations unloading from the plastic are given by the Eq. e (359) ff Definition of Ep. 310. (250). Inc. (357) can be expanded in terms of principal stresses and strains to give 2(Ti ei — — a2 — — — ^3 _ 1 O"! — 2 (^^2 + 03) Ep J_ Ol — — 2 (^^2 + 03) €2 = 2a'2 Ci CTs 1 QGp 2cr3 — (Ti ^2 es QGp ^ 3Gp _ 1 ~ 3^ Ep 02 2 ^^^ "*" '^^^ os — 2 (<^i + ^2) J_ Ep 03 — 2 ^^^ '^ '^^^ (358) The analogy is apparent between the righthand side of Eqs. 7779. Because of the assumption of constancy of volume. (358) based The fact that 11 = v leads to the conclusion that the ratios of the principal shearing stresses to the principal shearing strains are equal. or in region. (223)]. (355) do not apply. J_ ^ Ep Significant strain Fig. Nadai. Therefore. New York. 1931. McGrawHill Book Company. Ep can be considered to be a plastic modulus that is actually a variable depending upon The evaluation the stress and strain. Eq.. 36. Poisson's ratio has been taken equal to 3^. and e = e. e" = 0. 36. For the plastic case. When in the elastic range. (357).
or it can arise from a situation where only part of the material is deformed and the rigid material outside the plastic region prevents the spread of deformation (Fig. The Hencky theory gives results in agreement with the incremental or flow while in a flow theory. 37 a).• ^  ^v'Jv Rigid // // metal merai // // ^iPlastic^^ v//////y//?7/?///////// <i^////////////////////////A. displace ments can be considered to be limited to the xy plane. but it is often used for small plastic strains because it offers certain mathematical conveniences. dition of ylane strain. In a deformation theory. 311. for this reason that relationships like Eqs. 37. so that strains in f^=0 Punch Punch Plastic Rigid ^^ •^ . (358) are sometimes Nadai's equations. (a) Fig. Twodimensional Plastic Flow In — Plane Strain all many practical problems. such as rolling and drawing. Kb) Methods of developing plastic constraint. such as theory provided that the principal axes of stress and strain remain coincident during the straining process and provided that proportional loading The Hencky theory is not satisfactory for large deforis maintained. the increments of plastic strain are proportional to the stress deviator. 311] Elements of the Theory of Plasticity 73 It is and from these three relationships the equations can be derived. to develop a necessary to constrain the flow in one direc Constraint can be produced by an external lubricated barrier. the total plastic strain is called is given by the Hencky or Nadai proportional to the stress de viator.)ec. mations. it is all directions. a good indication of the deformation and forces required can often be obtained by consideration of the analogous planestrain problem. such as a die wall (Fig. such as equations. is given by the Reuss equations. This is known as a con When a problem is too difficult for an exact three dimensional solution. 37b). Since a plastic material tends to deform in planestrain condition tion. Even though the strain in one of the principal directions is equal to . the z direction can be neglected in the analysis.
. Sachs. (321)] Von Mises stress criterion of yielding the relationship However. (. stress in shear [Eq. (360) becomes — = 2k. designated a and . value substituted into the expression for the Von Mises criterion of yielding. Thus. 1]8. 1953. with the planestrain condition defining the minimum principal stress will be 02 and the shear stress criterion should be written CTi — (72 = CTo = 2k (361) In Eq.74 Mechanical Fundamentals it [Chap. the critical shear stress k will first This condition is shown in Fig. direction. 386). These lines of maximum shear stress are called slip lines. "IntroductiontotheTheoryof Plasticity for Engineers. 2k. it should be carefully noted that the slip lines just referred to tion of the principal stresses. " McGrawHill Book Company. respect to arbitrary cartesian coordinates. determine the principal planes such that the shear stresses vanish (Fig. (361) k is the yield stress in pure shear.15oo The maximumshearstress (Ti criterion of yielding can be expressed by — cTs = (To = 03. The slip lines have the property that the shear strain is a maximum and the linear strain is zero tangent to their direction. Equation (361) is equally valid when written in terms of the stress deviator. but a" varies from point to point. Hoffman and G. Inc. 38c. fc is Thus.8. principal stresses are simply functions of the spherical stress. based on the between the tensile yield is and the yield given by ca = s/S k. New York. = ao = 2k (362) 312. reach its value on these planes. However. 02 Therefore. and the shearing stress k. The maximum shear stress will occur on planes 45° to the direc a" . 3 is o^ zero for plane strain. the two yield criteria are equivalent. (Ts does not follow that there ^ zero stress in this It can be shown If this that for plane strain is = (a^ + (Ty)/2 or = (oi + c72)/2. The of component a constant throughout the plastic region if strain hardening is neglected. the yield criterion for plane strain becomes (360) (Ti — a2 = —27= Co V3 = 1. where it is seen that the maximum shear stress occurs in two orthogonal directions.. and it can be considered that twodimensional flow will occur when the shear stress reaches a critical value of k. ' O. for planestrain conoi ditions. Slipfield Theory Consider a volume element in plane strain within a plastic region of a Figure 38a represents the twodimensional state of stress with It is possible to body. Eq. the value of However. p. .
However. 312] Elements of the Theory of Plasticity 75 are not the slip lines. since there can be no resultant tangential force at a . the a< be straight lines. The values can a" is known since cTi (72 = — a" 0" \ k (363) — K slip lines will If a" is constant throughout the region. 386 and c. o] =a +/[ ic) Fig. 39. hold: if the slip Unes curve by an angle 0. will This latter type of slip lines be discussed more fully in the next chapter. the following relationships a" a" j 2k(^ 2k4) — — = constant along a line (364) constant along (3 line The slip lines at a free surface must make an angle of 45° with the surface (Fig. 38.Sec. By comparing be determined if Fig. it is seen that the principal stresses have of the principal stresses a direction 45° to the slip lines. observed under the microscope on the surface of plastically deformed metal. or slip bands. =0 Fig. 39). Slipline field at free surface. Twodimensional state of stress in plane strain.
"The Hardness of Metals." pp. (/> at R is <yin = a" ^ k = (k fj ^ 2k l) + k or a. 310. free surface. stress at a free surface. York. the slip line turns through an angle ^ = — 7r/2 so that its equation at point Q is a" — 2/c(7r/2) = k. New .. where the slip line deviates from a straight line. the equation of this slip line may be written a" \.76 Mechanical Fundamentals Since there (363) is is [Ch ap. As a further example of the use of slip lines. 02 Fig. consider the deformation of an ideal plastic metal by a fiat punch. mation will first start at the corners of the punch and of the slipline field Plastic deforwill result in a such as is shown in Fig. ^ The friction between the face = and by Eqs. Since no further change takes place in in going to point R. = 2k(^l+ = 2k and ^2« If we trace out any of the other slip lines. the principal stress normal to the surface is Since this at a free surface.2k(f> = k. punch and the metal is considered to be negligible. the pressure uni form over the face of the punch and equal to (71 = 2k[l { i) (365) 1 D. In accordance with Eqs. Slipline field produced by indentation of a punch. is we shall find in the same way is that the normal stress 2k{l \ x/2). Consider the point is M. Tabor. There is no change in the value of a" until we reach point A^. the normal stress zero and a" = k. 0" = —k. 1951. and the transcompressive with a value of 2k. 3437. In going from A'' to Q. Therefore. (364). Oxford University Press. 01 no resultant normal verse principal stress = —2fc. 310. is Therefore.
1. Mech." Oxford University Press. Hill. Inc. the theory predicts that indentation... Inst. 10211030.. (London). W. the method of slip fields. J. Meiallurgia. 1959. will occur fullscale plastic flow. with the resulting when the stress across the face of the punch reaches three times the yield strength in tension. 1954. Proc. Mech.: "The Mathematical Theory of Plasticity. and P. Stockholm. is relatively simple and represents an However. Roy. TechnoL. sometimes called the Hencky plasticsection method.: "Theory of Flow and Fracture of Solids. Jr. R.. Quart. Phillips. vol. A. Hundy. vol. J. The example described above overidealized situation. G. A. Trans.^ given general procedures for constructing slipline fields. 169. Thomsen. 1949. O. 293." AddisonWesley Publishing Company. 1950. Inc. " . and G. Mass. New York. However. 8184. vol. no." The Ronald Press Company. 1955. New York. "Theory of Perfectly Plastic Solids. Alexander. Inst. 1950. New York. is an important analytical tool for attacking difficult problems in plasticity. 1956. : 1 R. pp. 65. 49. 24. B. no. Mech.." McGrawHill Book Company. C. 1953. Appl. B.: "An Introduction to Plasticity. vol. Inc. Prager. Engrs. 1953. ^E.Sec." 2d ed. pp. 1951. McGrawHill Book Company. 4052. 1957. pp. Math. pp.: "Introduction to Plasticity. I. Hoffman. 109118.. Reading. 2 M." John Wiley & Sons. It has been used in the analysis of twodimensional problems such as the yielding of a notched Prager^ and Thomsen* have tensile bar^ and the hot rolling of a slab. BIBLIOGRAPHY Hill. mental verification of theoretically determined slipline fields has been obtained for mild steel by etching techniques^ which delineate the plastically deformed regions. New York. Prager. Sachs: "Introduction to the Theory of Plasticity for Engineers. vol. Appl. 312] Elements of the Theory of Plasticity 77 Since k = ao/y/S. * W.. Hodge. <Ji = cr^ax = ^ ^1 + I ) = 3(ro (3_66: Thus. there Partial experiis no easy method of checking the validity of a solution. /.. New York. Nadai.
.
Part Two METALLURGICAL FUNDAMENTALS .
.
Nielsen. rays by metallic crystals Following the discovery of the diffraction of by Von Laue in 1912 and the realization that metals were fundamentally composed of atoms arranged in specific geometric lattices there have been a great many investigations of the relationships between atomic structure and the plastic behavior of metals. assumption that we are dealing with an isotropic homogeneous medium becomes less tenable. Trans. 4. to rely heavily on experimentally determined data. 147. 2 81 .Chapter 4 PLASTIC DEFORMATION OF SINGLE CRYSTALS 41 .. 42. Lawson and S. Much of the fundamental work on the plastic deformation of metals has been performed with singlecrystal X specimens. 13. Met. Honeycombe. Within the plastic range the theories describe the observed behavior. ASM. W. Holden. pp. Inc. 1959. 1950. pp. New York. for finegrained metals sub logical description of the elastic and plastic behavior of metals. jected to static loads within the elastic range the theories are perfectly adequate for design. "Preparation of Single Crystals. 1958. N. D. K. Techniques for preparing single crystals have been described in a number ' of sources. Introduction three chapters have been concerned with the The previous phenomeno It has been shown that formal mathematical theories have been developed for describing the mechanical behavior of metals based upon the simplifying assumptions that metals are homogeneous and isotropic. However. A.^"* R. 3 W. our ability to predict the behavior of metals under stress by means of the theories of elasticity and plasticity decreases. As the in general. 319346. although not with the precision which is frequently desired." Academic Press. vol. vol. no. so as to eliminate the complicating effects of grain boundaries and the restraints imposed by neighboring grains and secondphase particles. Reviews. That this is not true should be obvious to anyone who has examined the structure of metals under a microscope. For conditions of dynamic and impact loading we are forced.
. since the origin of the plane. Plane fore. z ABCD . Crystallographic planes and direc tions will be specified with respect to these axes in terms of Miller indices. Concepts oF Crystal Geometry Xray diffraction analysis shows that the atoms in a metal crystal are arranged in a regular. Three mutually perpendicular axes are arbitrarily placed through one of the corners of the cell. 41).82 Metallurgical Fundamentals [Ch ap. The most elementary crystal structure is the simple cubic lattice (Fig. To Fig. the plane Thereaxes and intersects the y axis at one interatomic distance ao. This is the type of structure cell found for ionic crystals. chapters specifically devoted to these subjects. 41 . 6. Primary consideration will be given to tensile deformation. the reciprocals of these intercepts are used. effects. EBCF would be designated as the (TOO) plane. deformation in single crystals will be This subject will be extended in the next chapter to a consideration of plastic deformation in polycrystalline specimens. They are reduced to a lowest common denominator to give the Miller indices (hkl) of the in Fig. coordinate system can be moved to G because every point in a space . which plays such an important part in presentday concepts of plastic deformation. 42. 1/ oo or (hkl) = (010). A in crystallographic plane is specified terms of the length of its intercepts on the three axes. repeated threedimensional pattern. formulas. The atom arrangement of metals is most simply portrayed by a crystal lattice in which the atoms are visualized as hard balls located at particular locations in a geometrical arrangement. such as NaCl and LiF. A more detailed consideration of dislocation theory will be found in Chap. This will be followed by a chapter on the fundamental aspects of fracture and a chapter on internal friction and anelastic basic of plastic The mechanisms discussed in this chapter. The fundamental deformation behavior in creep and fatigue will be covered in The dislocation theory. measured from the origin of the coordinate axes. but not for any oi' the metals. 41 is parallel to the x and For example. simplify the crystallographic Simple cubic structure. will be introduced in this chapter to the extent needed to provide a qualitative understanding of modern concepts of plastic deformation. the indices of the plane are 1/ oo 1/1.
there are four facecentered cubic structure (% f ^^). As an ex [WO]. cell for a facecentered cubic is crystal structure. this direction of are A family crystallographically equivalent directions would be des ignated {uvw. (OlO). Many of the common metals have only. copper. Crystallographic indicated [uvw]. and nickel are common facecentered cubic metals. as is the atom located at the center of the cell. structure cell for the bodycentered Typical metals which have this crystal structure are alpha iron. depending upon the choice of axes. molybdenum. (001). atoms per structure Aluminum. (TOO). there are two atoms per cubic structure tungsten. 42. (010). directions are [\00). lead. Each corner atom is surrounded by eight adjacent atoms. Fig.D 42] lattice Plastic Deformation of Single Crystals 83 has the same arrangement of points as every other point. The bar over one of the integers indicates that the plane intersects one of the axes in a negative direction. tantalum. Figure 42a shows a bodycell centered cubic structure with an atom at each corner and another atom at the body center of the cube.HBCG [\\\)GEC [\\2)GJC ample. There are six crystallographically equivalent planes of the type (100). Therefore. columbium.) tice For the cubic latis always perpendicular to the plane having the same indices. a direction either a bodycentered cubic (bcc) or facecentered cubic (fee) crystal structure.HADG by integers in brackets: Reciprocals are not used in determining directions. the direction of the line F is obtained by moving out from the origin a distance ao along the x axis and moving an equal distance positive y direction. Since these latter atoms cell in belong to two unit silver. (6) facecentered cubic structure. . there an the atom at the center of each of the cube faces. (a) Bodycentered cubic struc ture. gold. of in the The then indices [110]. and {% \ I). The used when they any one are to be considered as a group. chromium. atom at each corner. or family of planes. (OOT) notation J 100} is of which can have the indices (100). Figure 426 shows the structure In addition to an cells.
The first third index is related to the i two by the relation = —{h\k). 44.GHJ Type I. four axes the three axes ai. tions in the hep structure. Order 1 (lOTl) .GHL Type n. The first layer of spheres is arranged so that each sphere is surrounded by and just touching six other spheres. 43). Basal plane (0001)  ABCDEF Prism plane (1010) . The facecentered cubic and hexagonal closepacked structures can both be built up from a stacking of closepacked planes of spheres. Stacking of closepacked spheres. The first is possibility results in a stacking sequence ABC ABC • • . a^. 44. These indices are based on .KJH . 43. Figure 44 shows that there are two ways in which the spheres can be stacked. 43. packed plane. This corresponds to the solid circles in Fig. 44).FEJH Pyramidal planes Type I. planes in axis is normal to the These axes and typical the hep crystal structure c are given in Fig. Order 2 (101_2) .84 Metallurgical Fundamentals [Ch ap. Order (1 21 Type n. ture. A second layer of closepacked spheres can be placed over the bottom layer so that the centers of the atoms in the second plane cover onehalf the number of valleys in the bottom layer (dashed circles in Fig. which found for the . they may lie either over the valleys not the first plane (the dots in Fig.KHL Digonal axis [ll20] . 44). valleys in covered in There are two ways of adding spheres to give a third closeAlthough the spheres in the third layer must fit into the the second plane. is The packed third common metallic crystal structure the hexagonal close (hep) structure (Fig. it is In order to specify planes and direcconvenient to use the MillerBravais system with four indices of the type (hkil). and the vertical basal plane.FGC 1 1 ) Fig. Order 2 (1122) . az are 120° apart in the basal plane. Hexagonal closepacked struc Fig. 44) or directly above the atoms in the first plane (the crosses in Fig.
43] Plastic Deformation of Single Crystals 85 {111} planes of an fee strueture. The other possibility results in the stack ing sequence or 1.568 1. Table 42 Atomic Density of Lowindex Planes .623 1 633 1 856 1 886 . the ratio of c/a is \/%.587 1. . on a hard sphere model. the hep strueture.Sec. is found for the (0001) basal plane of For the ideal hep packing. Table 41 Axial Ratios of Some Hexagonal Metals Metal c/a Be Ti Mg Ideal hep Zn Cd 1. The fee and hep structures are both closepacked cell is four per cent of the volume of the unit Plastic deformation is generally confined to the lowindex planes. This is contrasted with 68 per cent packing for a bcc unit cell and 52 per cent of the volume occupied by atoms in the simple cubic unit cell. in the fee and hep structures. Seventyoccupied by atoms.633. Table 41 shows that actual hep metals deviate from the ideal • ABAB . which have a higher density of atoms per unit area than the highindex planes. Note that the planes of greatest atomic density also are the most widely spaced planes for the crystal structure. structures. . which c/a ratio. Table 42 lists the atomic density per unit area for the common lowindex planes.
may be divided into line defects and surface.^ exists (Fig. The term defect. 5). or imperfection. London. 410) is of the crystal that have alternate stack also a surface defect. Point Defects Figure 45 illustrates three types of point defects. . for a better understanding of the structuresensitive properties defects. it the defect extends through microscopic regions of the crystal. in relatively recent times. defects. At equilibrium. it The itself has been necessary to consider a number of types of lattice description of the structuresensitive properties then largely to describing the behavior of these defects. In pure metals. Only since the realization of this fact. Structuresensitive reduces Structureinsensitive Elastic constants Electrical conductivity Melting point Density Specific heat CoeflScient of thermal expansion Semiconductor properties Yield stress Fracture strength Creep strength As is suggested by the above brief tabulation. practically all the mechanical properties are structuresensitive properties. or plane. Lattice imperfections is called a lattice imperfection. is generally used to describe any devi ation from an orderly array of lattice points. Line defects obtain their name because they propagate as lines or as a twodimensional net in the crystal. A vacancy. 45a). or point imperfection. 1958. The stacking fault between two closepacked regions ing sequences (Sec. or vacant when an atom is missing from a normal lattice position lattice site. the fraction of lattices that are vacant at a given temperature is given approximately by the equation ^ 1 n = lis exp^ /A (41) i\ 'Vacancies and Other Point Defects. However." Institute of Metals. The edge and screw dislocations that are discussed in this section are the common line defects encountered in metals. have really important advances been made in understanding the mechanical behavior of materials. small numbers of vacancies are created by thermal excitation. When the deviation from the periodic arrangement of the lattice is localized to the vicinity of only a few atoms if it is called a point defect. Lowangle boundaries and grain boundaries are surface defects (see Chap. Surface defects arise from the clustering of line defects into a plane.' 86 Metallurgical Fundamentals [Chap. and these are thermodynamically stable at temperatures greater than absolute zero. 4 of the perfect lattice is adequate for explaining the structureinsensitive properties of metals.
number of vacancies at room temperature. (6) interstitial. Vacancy. Table Equilibrium Vacancies in a Metal . When the density of vacancies becomes relatively large. 4ii Plastic is Deformation of Single Crystals 87 where n required to Table 43 number of vacant sites in N sites and E^ is the energy move an atom from the interior of a crystal to its surface. it is possible to trap in a greater than equilibrium oooo oooo oo o oooo ia) Fig. (c) impurity atom. oooo oooo oooo o oooo [b] (a) ooooo ooooo o^ooo ooooo [c) Point defects. By rapid quenching from close to the melting point. Higher than equilibrium concentrations of vacancies can also be produced by extensive plastic deformation (cold work) or as the result of bombardment with highenergy nuclear particles.bee. illustrates how the fraction of vacant lattice sites in a metal the increases rapidly with temperature. it is possible for them to cluster together to 43 form voids. 45.
the area Slip has occurred in the direction of the slip vector over ABCD. . it is generally agreed that Fig. The boundary between the righthand Note that the parts slip slipped part of is the crystal and the lefthand part which has not yet slipped the line AD... . slip occurs in the area over which it moves. of the crystal slip. The two basic types of dislocations are the edge dislocation and the screw dislocation. fatigue. the atoms above move . 47. creep. atoms above area C have been displaced one atomic distance in D have not yet slipped.. but they are also intimately connected with nearly all other mechanical phenomena such as strain hardening. slip the edge dislocation. 46. helps explain i why real crystals . which was originally suggested by Orowan. The amount of displacement equal to the For a pure edge dislocation such as is shown here. Figure 47 shows the slip that produces an edge dislocation for an element of crystal having a simple i cubic lattice. or TaylorOrowan dislocation. Although the exact arrangement of atoms along AZ) is not known. It is assumed that slip is advancing to the right.• ^ r ^ ^i r expected tor a crystal with a perfect lattice. a dislocation can easily on the application of only a small this force. AB represents a dislocation lying in the slip plane. The simplest type of dislocation. As the dislocation moves. and brittle fracture. slip plane Not only are dislocations important for explaining the slip of crystals. . 48 closely represents the atomic arrangement in a plane normal to the edge dislocation AD. . The plane of the paper in this figure corresponds to a (100) plane in a simple cubic lattice and is Burgers vector b of the dislocation. deform much more readily than would be in a Fig. the magnitude of the Burgers vector is equal to the atomic spacing.88 Metallurgical Fundamentals [Chap. 46. A defining characteristic of an edge dislocation is that its Burgers vector is always perpendicular to the dislocation line. and Taylor. A dislocation . Polanyi. AB is then the boundary between the slipped and unslipped regions.. In Fig. the yield point. in the direction of plane by an with respect to origi the part of the crystal below the the shaded area in Fig. is called the edge dislocation. which is the plane of the paper... All the the slip direction. It is shown shaded to indicate that for a few atomic distances on each side of the dislocation line there is a region of atomic disorder in which the slip distance is between zero and one atomic spacing. 4 disturbance separating the slipped and unslipped regions of a crystal. amount indicated by which were is All points in the crystal nally coincident across the slip plane have been displaced relative to each other by this same amount. In the absence of obstacles. above the plane have been displaced.
1953. plane of atoms above the slip plane. Read. Jr. vector Fig. is called by convention a positive edge dislocation and is frequently indicated by the symbol X. T. distorted in the region of the dislocation. perpendicular to slip direction. Dislocation lies along AD. Slip has occurred over area ABCD. Slip vector ' . Edge dislocation produced by slip in a simple cubic lattice.. 2. 43] Plastic Deformation of Single Crystals 89 equivalent to any plane parallel to the front face of Fig. the lattice is Note that There is one more vertical row of atoms above the slip plane than below it. p. The atomic arrangement results in a compressive stress above the slip plane and a An edge dislocation with the extra tensite stress below the slip plane. as in Fig..Sec. 48. r . 47. McGrawHill York. New . "Disloca in Crystals. tions {W. 47.) Book Company.." Inc.
planes /'Slip ^ plane in Therefore. New shown in Fig. has taken place to the Thus. Fig. and therefore the motion of a screw dislocation is less restricted than the motion of an edge dislocation. parallel to slip area ABCD. as an edge dislocation has. McGrawHill York. dislocation line. tions in Crystals. has occurred over the (W. A screw have a preferred slip plane. or slip vector. the experimental evidence for dislocations. Deformation by Slip The usual method of plastic deformation in metals is by the sliding of blocks of the crystal over one another along definite crystallographic . we one atomic plane behind that conIn making this circuit we have traced the path of a righthanded screw. 410. is The second J'igure 49 basic type of dislocation the screw. The upper part of the crystal to the right oi left of AD has moved slip relative to the lower part in the direction of the slip vector. 6. dislocation. made around the dislocation displaced one slip the end point is plane parallel to the the lattice. and the interaction between dislocations. the atomic are arranged around the vector dislocation in a spiral staircase or screw. 49. Among the topics covered will be the effect of crystal structure on dislocation geometry. location lies along direction. X and comarrive at X'. shows a simple example of a screw dislocation. we shall return to a detailed discussion of dislocation theory in Chap.) Book Company. or Burgers.90 Metallurgical Fundamentals [Ch ap. the discussion of dislocations will be limited to the geometrical concepts presented in this section. The open circles represent atoms just above the slip is atoms just below the slip plane. pleting the circuit. In this figure we are looking down on the slip plane in Fig. and therefore AD is a. on the front face of the crystal." Inc. T. plane. Jr. 44. After a more complete discussion of the plastic deformation of single crystals and polycrystalline specimens. Burgers vector. cuit is line. 49. No AD. Slip The arrangement of atoms (in two dimensions) around a screw dislocation in a simple cubic lattice AD.. and by definition this Consider the trace of a circuit around the Starting at dislocation line. movement by climb is and the solid circles are dislocation does not not possible with a screw dislocation. Read.. Slip that produces a screw disDis location in a simple cubic lattice. Every time a cirtaining X. 1953. For the present. However. the dislocation line is parallel to its must be a screw dislocation. "Dislocap. 15.
AD New York. 17. called slip planes. stress exceeds a critical value. Atomic arrangement around the screw dislocation shown in Fig. the slip line will disappear (Fig. or can he considered analogous to the distortion produced it in a deck of cards when is pushed from one end. of a crystal As a very crude approximation. 41 la. 44] Plastic Deformation of Single Crystals 91 planes. 49.." p. so that the step is removed. Jr. "Dislocations in Crystals. If the surface is then repolished after slip has occurred. (W. T. The plane of the figure is parallel to the slip plane. When we . Slip occurs The atoms move an and a step integral when the shear number of is produced in the view the polished surface from above with a microscope. glide. 410.Sec.) to a metal cube with a top polished surface. a shear stress is applied ® <ii — ®—®—®—®— 6) ii ®=^ ii ^i «>=^ <i^ «) «5 6) «J 6N^ «? 65 ^) 6) ^) 5 6 J «5 i) 65 6^ 6^ (5 65 6 5 ^5 6 5 65 65 «5 65 6^ 6^ $ 65 65 65 65 6) 65 65 65 65 6 5=:^ (5 6) 65 65 6^ (5 65 6 5 6 5 ^ ® 65 65 6) 6) 6 5 ® 65 6 5 6 5 65 6 5 65 6 5 65 6>=. trates this classical picture of slip. 65 6 5 6 5 65 65 6^ ^5 65 6) 65 6^ Slip (5 65 ® ® 6^ vector ® ® ® ® Fig. the step shows up as a line. McGrawHill.. Book Company. Figure 411 illusIn Fig. which w^e call a slip line. The single atomic distances along the slip plane. ABCD is the slipped area. 1953. polished surface (Fig. 41 Ic). Inc. Open circles represent atoms in the atomic plane just above the slip plane. the slip. and is the screw dislocation. 4116). Read. and the solid circles are atoms in the plane just below the slip plane.
92 crystal Metallursical Fundamentals is still [Chap. 4 a single crystal after slip has taken place provided that the deformation was uniform. surface elevation Note that slip lines are due to changes in and that the surface must be suitably prepared for Polished surface .
ratios. and cobalt slip occurs on the (0001) plane in the (1120) directions. Schematic drawing of the fine structure of a sKp band. the fee lattice has 12 reverse directions being neglected). the slip plane. Slip distance ^Interslip / Lamella spacing region id) Fig. The digonal axes (1120) are the closecadmium. structure is Accordingly. magnesium. In the hexagonal closepacked metals. For zinc. Slip occurs most readily in specific directions on certain crystallographic Generally the slip plane is the plane of greatest atomic density planes. the {lllj octahedral planes and the (110) directions are the closepacked systems. on the prism and . The bcc structures. planes in the fee unit cell. the resistance to slip most widely spaced planes is generally less for these planes than for any other set of planes. found for alumieven when viewed at high magnification. The ' {110} planes have the highest atomic density in the bcc structure. (a) Small deforma tion. the hep structure The limited number of slip systems is the possesses three slip systems. In the facecentered cubic structure. as (111) in the fee structure and (0001) in the hep structure. packed directions. slip primarily Zirconium and titanium. reason for the extreme orientation dependence of ductility in hep crystals.} Sec. The slip plane together with the slip direction establishes the slip system. there not a closepacked structure like the fee or hep is no one plane of predominant atomic density. 41 3. but in alpha brass there only one slip line. (6) large deformation. There are eight 111 However. (Table 42). the planes at opposite corners of  the cube are parallel to each other. so that there are only four sets of plane contains three (110) directions (the Therefore. and the slip direction is the closestpacked direction within Since the planes of greatest atomic density are also the in the crystal structure. { Each 111 } possible slip systems. which have low c/a pyramidal planes in the (1120) direction. octahedral planes. 44] Plastic Deformation of Single Crystals is 93 there are many slip lamellae comprising the slip band is num and copper. the only plane with high atomic density is the basal plane (0001).^ Since there is only one basal plane per unit cell and three (1120) directions.
Steijn and R. J. vol. M. MchI. but they differ from most other metals by not having a definite single slip plane. G. R. Slip in bcc alpha iron has concluded that the the [111] zone. J. Vogel and R. M. Home. Wavy alip lines in alpha iron. vol. p. 4 but they are not greatly superior in this respect to several other planes. Guard. 140G1448. 49. 1954. while the slip direction is always There are 48 possible slip systems. in the bee structure the (111) direction is just as closepacked as the (110) and (1120) directions in the fee and hep structures. W. Brick. Trans. which supports the belief that slip in alpha iron is stress axis with respect to the crystal axis { noncrystallographic. but since the the [111] direction. Low and R. 700. 197. Therefore. L. AIME. ASM. T. the bcc metals obey the general rule that the slip direction is the closepacked direction. €P Fig. 150 X. . Brick. {112}. J. vol. Acta Mel. Trans. Slip in bcc metals is found to occur on the {110}.. These studies have shown that observed deviations from the lowindex planes {110}. {Cuuiicsy J. R. Certain metals show additional systems with increased temper 1 F. pp. 171179. its been particularly well studied.) planes are not so closepacked as in the fee structure. 2 J. Cox. P. higher shearing stresses are usually required to cause slip. Cox. and R. ASM. 4G. However.94 Metallurgical Fundamentals [Chap. 414. and {123} planes. 7. 1957. 414).^ It has been may occupy any position in position being determined by the orientation of the slip plane in alpha iron and the variation in the shearing strengths of the planes in the slip zone. pp. have shown that curved slip linos are produced in iron by screw components of the dislocation loop but that the slip lines are straight when viewed normal to the edge component of the dislocation. {112}. 1958. Further evidence for noncrystallographic slip is the fact that slip lines in alpha iron are wavy^ slip (Fig. 1959. 118131. Trans. vol. F. and 123} are real effects.
Aluminum deforms on 'he {100} plane at elevated temperature. 45] Plastic Deformation of Single Crystals 95 ature.Sec. while in 45. magnesium the {lOllj pyramidal plane plays an important role in deformation by slip above 225°C. Slip in a Perfect Lattice to occur If slip is assumed it is by the translation of one plane of atoms over another. possible to make a reasonable estimate^ of the shear o o o o o o o [a) / . In all cases the slip direction remains the same when the slip plane changes with temperature.
Hooke's law should apply. provide such a mechanism. As a first approximation. (46). As a rough approximation. Eq. Thus. the relationship between shear stress and dis placement can be expressed by a sine function T = Tm Sni —r ' (42) where r^ is the amplitude of the sine wave and h is the period. r = Gy Gx =~ (43) For small values of x/h. a. This is still at least 100 times greater than the observed shear strengths of metal crystals. (46) predicts that the theoretical shear stress will be in the range 10^ to 10® psi. At small values of displacement. In the next section. the theoretical shear strength of a perfect crystal is with the result that approximately equal modulus divided by Tm^§^ The shear modulus 10^2 (46) for metals is in the range 10® to 10'^ psi (10^^ to dynes/cm^). Therefore. it seems reasonable to expect that the theoretical shear strength of most metals lies between G/10 and G/50. 4 symmetry position. Eq.96 Metallurgical Fundamentals [Chap. the value of r^ cannot be reduced by more than a factor of 5 from the value predicted by Eq. to the shear h can be taken equal to 2x. It can only be concluded that a mechanism other than the bodily shearing of planes of atoms is stress required to 10* psi. while actual values of the shear produce plastic deformation are in the range 10^ to refined calculations are used to correct the sinewave assumption. Even if more responsible for slip. Between these positions each atom is attracted toward the nearest atom of the other row. (43) and slip (44) provides an expression for the maximum shear stress at which should occur. (42) can be written 27ra. so that the shearing stress is a periodic function of the displacement. (44) Combining Eqs. it is shown that dislocations .
at large distances This illus trates a positive edge dislocation. for example. 91." p. if the top row of atoms is displaced toward the right relative to the bottom row. Inc.Sec. Seilz. all the atoms There minimum positions in the energy curve. the consider pairs of atoms.. Now consider the situation when the crystal contains a dislocation (Fig.) crystal move a dislocation through a very low compared with the theoretical shear stress. Now energy curve. The atoms from the center of the of the dislocation are at positions corresponding to the atoms at the center are not. minimum . or slip band. (a) Energy field in perfect crystal lattice. "The Physics of Metals. For the normal arrangement lower plane are at of a perfect crystal. Slip by Dislocation Movement The concept of the dislocation was first introduced to explain the discrepancy between the observed and theoretical shear strengths of metals. 1943. 416. 1 23456789 id) Fig. {F. New York. with the extra plane of atoms between 4 and 5. 4166). 416. in the fore. 46] Plastic Deformation of Single Crystals 97 46. This is the situation described in Sec. Schematic diagram illustrating the fact that a dislocation moves easily through a crystal lattice. it is necessary to demonstrate (1) that the passage of a dislocation through a crystal lattice requires far less than the theoretical shear stress and (2) that the movement of the dislocation through the lattice produces a step. For the concept to be useful in this field. each atom encounters the same force opposing the displace ment. 45. McGrawHill Book Company. (6) lattice containing an edge dislocation. Figure 416a represents the atoms in two adjacent planes illustrate that the stress required to is To which does not contain a dislocation. we shall use Fig. at the free surface. The top curve of the figure represents schematically the energy of an atom in the lower plane of atoms as a function of its position relative to the upper in a perfect crystal lattice plane.
Therefore. rch ap. 145A. In general. is needed to drive a dislocation through the lattice. 145A. vol. the PeierlsNabarro force. or slip band. Soc. 1934. 362. Roy. 1934. The lattice offers essentially no resistance to the motion of a dislocation only when the dislocation lies at a position of symmetry with respect to the atoms in the slip plane. p. one atomic distance The slip plane is location reaches the right side of the crystal. based on the original work by Taylor. The top series of sketches shows a positive edge dislocation moving to the right in a simple cubic lattice.^ illustrates that the movement of a dislocation results in a surface step. (G. (London). I.) the displacement is zero. is and the stress required to move the dislocation very small. ' shown dashed. the accurate calculation of this force is difficult because it depends strongly on the relatively uncertain atomic arrangement at the center of a dislocation. Soc. Taylor. Roy. 3 and 6. if the atoms near the center of the dislocation are displaced by equal distances. etc. Fig. a small force. located symmetrically on opposite sides of the They encounter forces which are equal and As a result. (London). Figure 417.98 4 and Metallurgical Fundamentals 5.. When the disassumed to be a free surface. . Proc. While it is well established that the value of the PeierlsNabarro force is much smaller than the theoretical shear stress for a perfect lattice. Movement of edge dislocation in a simple cubic lattice. to a first approximation. p. G. Proc. onehalf will encounter forces opposing the motion and onehalf will encounter forces which assist the motion. Taylor. the net work required to produce center of the dislocation. 41 7. I. 369. opposite. vol.
critical resolved shear ^ This value is really the singlecrystal equivalent of the yield stress of an ordinary stressstrain curve.)ec. 418). the first orientation with respect to the tensile axis of the plane on which slip appears and the slip slip direction. The angle between the normal to the plane and the tensile axis is 0. 41 8. series of sketches The bottom shows that the same surface step edge dislocation to the left. or one atomic distance for the simple cubic lattice. and the component of the axial load acting in the slip plane in the slip direction is P cos X. solved shear stress from a single crystal tested in tension.. The value of the critical resolved shear stress Slip Slip plane direction depends chiefly on composition tensile loads and temperature. the critical shear stress determine the stress at which the first slip bands is obtained by the intersection of the extrapolated elastic 2 and plastic regions of the stressstrain curve. Elektrochem. 37. of slip in a single crystal shearing stress produced and depends on the magnitude of the external loads. crosssectional area A Consider a cylindrical single crystal with (Fig. Schmid. the geometry of the crystal the orientation of the by active slip planes with respect to the shearing stresses. from Xray diffraction. Critical Resolved Shear Stress for Slip The extent structure. E. Therefore. Z. p. the critical resolved shear stress is given by '^ ^ P COS X 1\ <t>) ~T7? A/(cos = "T COS P A (^ COS X (47) 1 In practice it is very difficult to are produced. 1931. is pro duced by the movement of a negative 47. Diagram for calculating Schmid. . 47] Plastic Deformation of Single Crystals 99 it of produces a shift with respect to the planes on each side of the slip plane one Burgers vector. The area of the slip plane inclined at the angle 4> will be A /cos 0. vol. by a w^as critical resolved first shear this recognized by Fig. Slip begins when the shearing stress on the slip plane in the slip direction reaches a threshold value called the stress. To calculate the critical re critical resolved shear stress. it is necessary to know. In most cases. The fact that different are required to produce slip in single crystals of different orientation can be rationalized stress. 447. and the angle which the slip direction makes with the tensile axis is X.
Jr. J. p. D. Mg Cd. F. "R. /J. p.100 Metallurgical Fundamentals of the critical resolved shear stress. between the plane and the results in so many equivalent slip In fee metals the high symmetry systems that it is possible to get a Table 44 Roomtemperature Slip Systems and Critical Resolved Shear Stress FOR Metal Single Crystals Critical Metal Crystal structure Purity. vol. 191.9 99.400 9. ASM. Burke and W. <* « A. 49. Churchman. 216. Cox.99 99. AIME. Trans. C. fee fee Fe. variation in the yield stress of only about a factor of 2 because of differ The slip plane with the tensile axis. London. F. slip is known as Schmid's best demonstrated with hep metals. Chen. Schmid. E. Hibbard. 1954. p. Trans. demonstration of the resolvedshearstress law is even less favorable in bcc metals owing to the large number of available slip systems." vol. vol. p. Table 44 gives values of critical resolved shear stress for a number of metals. Trans. 295.996 99. T. 1009. K. available data indicate that Schmid's law is obeyed for cubic metals as well as hep metals. Home. The law law. C. also [Ch ap. 1935. bee 65 94 580 2.98 99. where the limited in orientation number slip of systems allows large differences tensile axis (see Prob. AIME. Soc. ences in the orientation of the . 1951.190 48 73 131 Cu.999 99. 194. 1954. 188. 1129. and G. R. Proc. 200. T. Ti. Trans.8 99. 48). p.000 D. Jillson. g/mm^ 18 77 Zn.97 99. 118. Roy. vol. Trans.96 plane shear stress. direction Ref. E. 937.99 99. vol. AIME. Rosi.93 99. Physical Society. 2. R.. "International Conference on Physics.996 99. 1952.800 Mo « ^ '^ bee (110) [111] 5. Slip Slip % 99. vol.999 99. 1957. However. Mehl. 58 1. p. 1950. The importance of small amounts of impurities in increasing the critical resolved shear stress is shown by the data for silver and copper. 226A. hep hep hep hep fee (0001) (0001) (0001) [1120] [1120] [1120] (1010) (1010) (111) (111) (111) (111) (111) (111) (110) (112) (123) [1120] [1120] [110] [110] [110] [110] [110] [110] [111] Ag. Maddin and N. (London). vol. AIME. Ni.
Physik.Sec. F. 62. 473.) they form a solid solution over the complete range of composition. crystals. Z. 47] Plastic Deformation of Single Crystals 101 Alloyingelement additions have even a greater effect. Trans. In solid solutions.) 28 I 8 24 J 20 \\ l\ o t 16 \ • 2? 12 NN. T. Cox. and G. °C 50 100 150 . vol. alloy single crystals. . 40 Atom 60 00 Au % Au 419. R. where the solute atoms differ considerably in size from the solvent atoms. as shown by the data for goldsilver alloys in Fig. Note that a large increase in the resistance to slip is produced by alloying gold and silver even though these atoms are very much alike in size and electronegativity. 419. an even greater increase in critical resolved shear stress would be observed. Mehl. Variation of critical resolved shear stress with temperature for iron single (J. p. Variation of critical resolved shear stress with composition in silvergold((?. vol. — ° 250 200 150 100 50 Temoerature. Weerts.CO Fig. and hence 20 Ag Fig. 49. ASM. Sachs and J. 1957. 123. p. Home. J. 1930. 420.
7 = cos Xi . 14841491. the glide strain y can be obtained^ from Eq. On the basis of this reasoning. provided that the total number of imperfections is not zero. . after deformation. This stress is. of course. London. "PlasticiU Hughes & Co. While the strain curves may be plotted in terms of average uniaxial stress vs. 48. 4 The magnitude of the critical resolved shear stress of a crystal is determined by the interaction of its population of dislocations with each other and with defects such as vacancies.. Schmid and W. pp. 2 For a derivation of Eqs. average linear strain (AL/Lo). Testing of Single Crystals Most studies of the mechanical properties by subjecting the crystal to simple uniaxial of single crystals are made stress tension.102 Metallurgical Fundamentals [Chap. see E. Qx sm and xi final sm (48) Xo where xo and xi are the initial angles between the slip plane and the tensile axis and Xo and Xi are the initial and final angles between the slip direction and the tensile axis. of Crystals. slip direction the orientation of the slip plane and the with respect to the tensile axis are known both before and (48) . 1956. the critical resolved shear stress should decrease as the density of defects decreases. sm Phtjs. . greater it is dislocation. S. (47)] against the shear strain or glide strain.sin^ (49) or 1 T S. F. Boas. 5860. shown by the fact that the critical resolved shear stress of soft metals can be reduced to less than onethird by increasing the purity. sity is can be grown essentially dislocationfree. interstitials. microndiameter singlecrystal filaments. a more fundamental way of presenting the data is to plot resolved shear stress [Eq. Glide strain is the relative displacement of two parallel If slip planes separated at a unit distance. A. cos Xo . with out requiring information on the final orientation of the glide elements. and impurity atoms. vol. or whiskers.. When should the rise last dislocation is eliminated. /. — (48) sin^ Xo} — cos \o (410) Brenner. Tensile tests ^ on these filaments have given strengths which are approximately equal to the calculated strength of a perfect crystal. and (49). the critical resolved shear stress abruptly to the high value predicted for the shear strength of a perfect Experimental evidence for the effect of decreasing defect dencrystal. 27. 1950. The glide strain can also be expressed in terms of the axial change in length and the original orientation. At the other extreme. AppL pp. xo)'^'^ 1^ t>0 = (1 + 27 sin xo cos Xo 2 + 7." English translation. but than the stress required to move a single appreciably lower than the stress required to pro duce slip in a perfect lattice.
"The Structure of Metals. the change in the angle between the slip plane and the tensile axis is related to the change in gage length in the axial direction by tensile axis increases sm xo (411) sin XI A men 1 convenient way of recording this reorientation is by following the In Fig." 2d chap. (6) Fig. McGrawHill Book Company. Therefore. is Near the extended. 42 la. of (a) Tensile deformation of constraint.. slip due to con slip planes rotate. . 2. Near the center of the gage length the sV (a) Fig. In tensile deformation. 422.Sec. the movement of the crosshead of the machine constrains the specimen at the grips. 4216. see C. since the grips must remain in line.. so as to align themselves grips bending of the slip planes superimposed on the rotation. rotation straint. manner shown in Fig. Stereographic triangles show single crystal without planes ing lattice rotation of fee metal during tensile elongation. [b] 421. as the crystal is parallel with the tensile axis. Instead. the specimen deforms in the as is pictured in Fig. axis of the specimen on the unit stereographic triangle. 1952. New York. ed. 48j Plastic Deformation of Single Crystals 103 testing In the ordinary tension test. S. Barrett. Inc. The amount of rotation toward the with the extent of deformation.^ the initial orientation of the axis of an is fee singlecrystal tension speci plotted on the unit stereographic triangle at P. 422. the specimen is not allowed to deform freely by uniform glide on every slip plane along the length of the specimen. The slip plane is For a description of stereographic projection.
This method of testing has the advantage that the crystal can be oriented so that the maximum shear stress occurs on any desired slip system. With even greater rotation. Parker and Washburn have described a method of loading single crystals in pure shear so that the shear strain is applied by a couple acting parallel to the active slip system. An excellent method of studying the deformation behavior of single crystals is by loading in shear..104 (111). of intersecting slip lines. Adv. axis initial or primarij slip system occurs. 2 For a complete review of this subject. a new set of slip lines appear on the specimen surface and the axis rotates toward the [112]. it is geometrically possible for a fourth slip system (Ill)[011] to become operative. is The twinned portion of the crystal 1 a mirror image of the parent crystal. Washburn. DeFormation by Twinning is The second important mechanism by which metals deform known as twinning. "Twinning and Diffusionless Transformations in Metals. London. it is increasing on another set of planes. the value of cos Therefore.^ Twinning results when a portion crystal takes up an orientation that is related to the orientation process rest of the of the of the the untwinned lattice in a definite. vol. In the fee metals. When the resolved shear stress on the new slip system is equal or about equal to the shear stress on the old slip system. However. "Modern Research Techniques Metal lurgy. The appearance of more than one slip system during deformation is often described under the general heading of duplex or multiple slip." Butterworth & Co. (Publishers). R. moves along a great circle passing through P and the the specimen direction As the deformation continues and rotation of the slip [101]. Cahn. . symmetrical way. 3. Parker and J. O. which are being rotated closer to a position 45° to the tensile axis. 4 and the slip direction is [101]. 49. Metals Park. Hall. in Phys. ^ of test. strain hardening neg lected. this slip system is usually not found to be operative in fee metals. Cross slip on the (111)[101] system may also This slip system has the same slip direction as the primary slip occur. a greater tensile load must be applied to maintain the value of the While cos <^ cos X is decreasing on the primary slip system owing to rotation. Metallurgical Fundamentals [Chap. W. in Physical E. even if 4> cos X for the priis mary slip system decreases. Resolved shear stress and shear strain are measured directly in this type resolved shear stress on this slip system. During elongation of the crystal. the primary slip lines. pp. 1953. In the microscope cross slip usually appears as short offsets to system. or R. 1954.. 363445. Ohio. Ltd. see E. the new slip lines occur on the conjugate slip system Under the microscope conjugate slip appears as another set (111)[011]. 1954." American Society for Metals.
4236. of symmetry between the two portions is called the twinning Figure 423 illustrates the classical atomic picture of twinning. paper. It should be noted that twinning differs from In slip. while in twinning Slip occurs the atom movements are much less than an atomic distance. In a simple lattice such as this. the orientation of the crystal above and below the slip plane is the same after deformation as before. ning plane (Fig. Classical picture of twinning. the planes of atoms have sheared in such a way as to make the lattice a mirror image across the twin plane. To the left of this plane. the crystal will twin about the twin The region to the right of the twinning plane is undeformed. 4236). Slip is usually considered to occur in discrete multiples of the atomic spacing. and solid circles are the final positions of these atoms in the twinned region. dashed circles indicate the original positions in the lattice of atoms which change position. the difference in elevation would be eliminated but the twin would still be visible after etching because it possesses a different orientation from the untwinned region. Note that the twin is visible on the polished surface because of the change in elevation produced by the deformation and because of the difference in crystallographic orientation between the deformed and undeformed regions. while twinning results in an orientation difference across the twin plane.Sec. If the surface were polished down to section A A. lattice Figure 423a represents a section perpendicular to the surface in a cubic with a lowindex plane parallel to the paper and oriented at an angle to the plane of polish. each atom in the twinned region moves by a homogeneous shear a distance proportional to its distance from the twin plane. 49] Plastic Deformation of Single Crystals 105 The plane plane. . slip in several specific respects. In Fig. 423. If The twinning plane is perpendicular to the a shear stress is applied. Polished surface Fig. open circles represent atoms which have not moved.
Twins may be produced by mechanical deformation or as the result of annealing following plastic deformation. a slip band is formed. while for slip there is a delay time of several milliseconds before certain conditions. The first mechanical twins.106 Metallurgical Fundamentals [Ch ap. If twinning occurs during a produces serrations in the the stressstrain curve. Table 45 lists the common twin planes it click or loud report (tin cry) Twii . ing (shock loading) and decreased temperature. Twinning occurs in a definite direction on a specific crystallographic plane for each crystal structure. Under form with a tensile test. Twins can form in a time as short as a few microseconds. and mechanical twins have been produced in copper by tensile deformation at 4°K. twins can be heard to . but in the twinned region of a crystal every atomic plane is involved in the deformation. on relatively widely spread planes. the latter are called annealing twins. although goldsilver alloys twin fairly readily when deformed at low temperature. type are known as Mechanical twins are produced in bcc or hep metals under conditions of rapid rate of load Facecentered cubic metals are not ordinarily considered to deform by mechanical twinning.
However. such as the hep metals. 49] Plastic DcFormation of Single Crystals 107 is only 7. twinning is important in the overall deformation of metals with a low number of slip systems.Sec.39 per cent. it should be understood that only a relatively small fraction of the total volume of f'" . The important twinning in plastic deformation comes not from the strain produced by the twinning process but from the fact that orientation changes resulting from twinning may place new slip systems in a favorable orientation with respect to the stress axis so that additional slip can take place. into a twin on the {1012} plane role of Thus.
AB AB AB. 3. The stacking sequence then becomes it Fairly recently ing sequence can be produced in most metals ABC A CAB. Slip has occurred between an A and a B layer. has been realized that errors. E. (c) Facecentered cubic packing.^ A I o^ OB o A BC'A CB'CA (c) ABABAB {d) (a) Fig. 4 410. Faulted structures. or faults. 1955. Very precise Xray diffraction measurements are needed to detect the presence of For example. For the fee structure. . (Fig. Acta Met. (6) deformation fault in fee. the stacking sequence of the planes of atoms is given by In an earlier section. in the stackby plastic deformation. vol. 473. P. ^ stacking faults. moving each atom layer above the slip plane one identity distance to the right. Stacking Faults it was shown that the atomic arrangement on the plane of an fee structure and the {0001 plane of an hep structure could be obtained by the stacking of closepacked planes of spheres. For the hep structure. 4255. 4256) with the stacking sequence for an hep structure without faults shows that the deformation stacking fault contains four Therefore.^ Slip on the {111} plane in an fee lattice produces a deformation stacking fault by the process shown in Fig. 425. Warren and E. see B.108 Metallurgical Fundamentals [Chap. id) hep packing. the stacking sequence is given ^o ^o o^ So off Bq tOA ~^C^ Op ^° Z^'^ Cq^ Be o o A B CA (a) BCA A B C A\C A B {b] TO 'o— ^oo o° oc o o o OB o o A ^^^ ^o o I oa ^o o °5 A Ar. } {111 } by ABC ABC ABC. Warekois. the formation of a stacking fault layers of an hep sequence. 425d) Comparison of this faulted stacking sequence (Fig. p. twin fault in fee..
19601961. As a result. :. The possibility of stacking faults in the {112} planes has been investigated theoretically and demonstrated by Xray diffraction. if somewhat different from that we straight Hne we will not come ^ to another atom on the next A to a B layer. 427434. vol.i' //r: :.3.. 1958. It is more difficult to form stacking faults in a bee lattice than in the closepacked fee and hep structures. Fourdeux and A.^_>. /. stacking faults in fee metals can also be con sidered as siibmicroscopic twins ot nearly atomic thickness. Inst. Guenter and B. slip can occur between tw^o of the planes so that the stacking seciuence becomes ABABACBCBC. Appl.•^^J^>>^ Z Fig. Phtjs. vol. it is now known that differences in the deformation behavior of fee metals can be related to differences in staekingfault behavior. J. The lower the locations tend to repel each other. The three layers ACB constitute the twin. 4048.. E.\>. why mechanical readily faults is when The reason twins of microscopically resolvable width are not formed fee metals are deformed is that the formation of stacking hep structure is so energetically favorable. and they have been most extensively studied for this crystal structure. From the point of view of dislocation theory.Sec. a stacking fault in an fee metal can be considered to be an extended dislocation consisting of a thin hexagonal region bounded by partial dislocations^ (Fig. 2 P. 426). or twin.. but this stacking fault energy. 89. The found situation for the in fee metals. J. Figure 425(i shows that.^ Stacking faults occur most readily in fee metals. vol. Acta Met. 5. 410] in Plastic Deformation of Single Crystals 109 an fee metal is equivalent to the formation of a thin hep region.. 29. vol. pp. . Acta Cryst. layer. Otte.. pp.. 1957. The splitting of dislocations into separated partials has been observed with the electron microscope C. Thus. 10. N. order. 3132. stacking fault. Warren. is The nearly parallel dis balanced by the surface tension of the stacking fault pulling them together. 6. / 426. Thus. M. O. and H. a stacking fault in an hep metal is equivalent to the formation of a thin fee region. Metals. 425c. Another way in which a stacking fault eould oeeur in an fee metal is by the sequence' shown in Fig. J. The stacking sequence ABC\ACB\CA is called an extrinsic." Stacking faults have been observed w^ith thinfilm electron microscopy in columbium. Hirsch in stainlesssteel foils. on going from an continue in a stacLing f^uit^ Partiol ^ ) A layer plane ^siip '^ disiocationsr^^y / / However. 1957. * Partial dislocations will be considered in more detail in Chap. pp. the greater the separation 1 between the partial dis Wagner. Schematic model of a stacking the straightline fee stacking fault. B. four layers of atoms BACB are in /^\ J ^<ui^ i^sA^. Berghezen. pp. For example. 44745. 3 A.
Since it requires energy to produce a constriction in the stacking fault. and show a different temperature dependence Figure 426 helps of flow stress from metals with narrow stacking faults. nickel. On this basis. twin easily on annealing. the slip lines are removed by subse quent polishing of the surface. the stackingfault energies for copper. Deformation bands generally appear irregular in shape but are elongated in the direction of principal strain. to illustrate why cross slip is more difficult in metals with wide stackingfault ribbons. the estimates of stackingfault energy are in qualitative agreement with metallographic observations of the fre quency faults in of annealing twins.g. it is not possible for them to transfer from one slip plane to another except at a point where the partial dislocations come together. indicating a general . the process of cross slip is more difficult in a metal with wide stacking faults than in a metal with narrow stacking faults. When slip occurs without restraint homogeneous fashion. can be observed even after repeated polishing and etching because they repreIn single crystals. however. Metals with wide stacking faults strainharden more rapidly. 80. DeFormation Bands and Kink Bands Inhomogeneous deformation in a perfectly of a crystal results in regions of different orientation called deformation bands. and this is in agreement with occurrence of annealing twins. Since the lower the energy of the twin boundary. aluminum rarely shows Xray work has shown that the energy of stacking brass decreases with zinc content. and aluminum are approximately and 200 ergs/cm^. The tendency for the formation of deformation bands is greater in polycrystalline specimens because the restraints imposed by the grain boundaries make it easy for orientation differences to arise in a grain during deformation. deformation bands several millimeters wide may occur. For example. 40.110 locations Metallurgical Fundamentals [Chap. The outline of the bands is generally indistinct and poorly defined. the fact that alpha brass forms a greater copper. the activation energy for cross slip is about 1 ev in aluminum and approximately 10 ev in copper. Stackingfault energies in have been estimated on the assumption that the stackingenergy is equal to twice the energy of a coherent boundary of an annealing twin.. sent regions of different crystallographic orientation. the greater the tendency for the formation of annealing twins. while in polycrystalline specimens microscopic observation is needed to see them. e. 4 fee metals fault and the wider the stacking fault. Because dislocations in the slip plane are extended. Deformation bands. number of annealing twins than of Stacking faults enter into the plastic deformation of metals in a number ways. 411.
It is known that the number with strain over the number present Thus. Hess and C. or work hardening. Frank and W. S. The horizontal lines represent basal planes.Sec. Strain Hardenins oF Sinsle Crystals One is of the chief characteristics of the plastic deformation of metals the fact that the shear stress required to produce slip continuously increases with increasing shear strain. is nearly parallel to the crystal axis. T. or kinking. Further study of kink bands by Hess and Barrett^ showed that they can be considered to be a simple type of deformation band. 599. 185. Barrett. Consideration of the equation for critical resolved shear stress shows that it will be difficult to deform a hexagonal crystal when the basal plane fading out of the orientation difference. Hardening due to dislocation interaction is a complicated problem because it involves large groups of dislocations. Read conceived a of dislocations in a crystal increases in the annealed crystal. and the 427. Kink band. 149. where a nonuniform distribution of slip can produce a bending moment which can cause kink formation. E. C. Distortion of the crystal is essentially confined to the kink band. 643. 1949. p. behavior in Fig. Nature. F. but not in hep metals. 427. Kink bands have also been observed in zinc crystals tested in tension. illustrated Fig. vol. cent from strain hardening metals. AIME. 1 2 J. planes designated p are the kink planes at which the orientation suddenly changes. The increase in the stress required is to cause slip because of previous plastic deformation known as strain hardening. . 412. 1942. of Orowan^ found that if a cadmium crystal this orientation w^re in compression it would deform by a localized region of the crystal suddenly snapping into a tilted position with a sudden short loaded ening of the crystal. p. A. The is buckling. and it is difficult to specify group behavior in a simple mathematical way. vol. 412] Plastic Deformation of Sinsle Crystals 111 Deformation bands have been observed in both fee and bee metals. Orowan. the first requirement for understanding strain hardening was the development of a logical mechanism for the generation of dislocations. Trans. is An increase in flow stress of over 100 per not unusual in single crystals of ductile Strain hardening is caused by dislocations interacting with each other and with barriers which impede their motion through the crystal lattice.
289347. Trans. back Parker. vol. 1525. and hence complications due 428. Edwards. J. H. p. One of the earliest dislocation concepts to explain strain hardening was the idea that dislocations pile up on slip planes at barriers in the crystal. pileups produce a hack stress which opposes the applied stress on The slip the slip plane. strain y 428. The mechanism is consistent with the experi by one mental observation that slip slip is concentrated on a relatively few active is planes and that the total slip on each slip plane of the order of 1. Note that on reloading the crystal yields at a lower shear stress than when it was first loaded. R. it may not always be of 1 E. pp. dislocations of opposite sign could be created at the dislocations responsible for strain in the same sources that produced the first slip direction. p. 1526.) AIMS. J. and then reloaded in the direction opposite to the original slip direction. 1953.. Since dislocations of opposite sign attract and annihi late each other. Zivilingur.. Washburn. H. J. 2 Bauschinger. vol. demonstrated experimentally by shear tests on zinc single crystals. This is because the back stress developed as a result of dislocations piling up at barriers during the first loading cycle is aiding dislocation movement when the direction of slip is reversed. unloaded. This explains the fact that the flow curve in the reverse direction lies below the curve for continued flow in the original direction. 1881. the net effect would be a further softening of the lattice.^ While all metals exhibit a Bauschinger effect. . Direct experimental evidence for the existence of the FrankRead source in crystals has been developed in recent years. A method is also provided in the concept of the FrankRead source for immobilizing the source after slip of this order of magnitude has occurred. Washburn. Effect of complete reversal of direction on stressstrain curve. when the slip strained to point 0. Slip direction J80 ° to original direction . In Fig. Parker. and E. 197. 1953. 197. The lowering of the yield stress when deformation in one direction is followed by deformation in the opposite direction is called the Bauschinger effect. direction is reversed.000 atomic spacings. 27. 6 for details) dislocation. Edwards. Trans. AIME. The existence of a (E. provides a method by which the dislocations initially present in the crystal as a result of growth can generate enough dislocations to account for the observed strain hardening. Shear Fig. R. Furthermore.US logical Metallutgical Fundamentafs [Qap. and E. the crystal is to duplex slip are easily avoided.^ stress was Zinc crystals are ideal for crystalplasticity experiments because they slip only on the basal plane. 4 mechanism by which a large amount of shp couJd be produced The FrankRead source (see Chap. vol.
the flow curve below the original flow curve for metals. with one another to produce a new dislocation that is not in a slip direction. jogs formed after intersection. Schematic representation of intersection of two screw dislocations. Since sessile dislocations do not lie on the slip plane of low shear stress. after reversal of direction does not fall all Moreover. in the dislocation line. existence of back stress and its importance to strain hardening in metals having been established. the . Microscopic precipitate particles and foreign atoms can serve as barriers. in addition to that due to the back stress resulting from dislocation pileups at barriers. the next step is to identify the barriers to dislocation motion in single crystals. The most important dislocation reaction. they act as a barrier to dislocation motion until the stress is increased to a high enough level to break down the barrier. (6) (a) Before intersection. is believed to occur when dislocations moving in the slip plane cut through other disbarriers in fee metals locations intersecting the active slip plane. Such barriers arise from the fact that glide dislocations on intersecting slip planes may combine Fig. but other barriers which are effective in pure single crystals The must be found. Any further move ment screw dislocations along the line A A would require the newly formed edge components to move out of their slip planes. The dislocation of low mobility that is produced by a dislocation reaction is called a sessile dislocation. 429. is and this referred to as the intersection of a forest of Figure 429 shows that the intersection of dislocations results in the formation of jogs. is the formation of CottrellLomer by slip on intersecting {111} planes. The dislocations threading through the active slip plane are often called a dislocation forest. The jogs formed of the in this case are edge dislocations because their Burgers vec tors are perpendicular to the original dislocation line. or offsets. Thus. 41 2] Plastic Deformation of Single Crysta! 113 the magnitude shown here for zinc crystals. Another mechanism of strain hardening.Sec. which leads to the formation of sessile dislocations. strainhardening process dislocations.
Inc. the dislocations are able to move over relatively large distances without encountering barriers. help of thermal fluctuations. is a stage in which the crystal undergoes little strain hardening. vol. and therefore ratedependent. During easy glide. S. 1 and strain dependence rate. it is temperature.^ the flow curve for puremetal single crystals can be divided into three stages (Fig. 430. ^2 ^3 Resolved shear strain y Fig. ser. All these processes require an increased expenditure of energy. 393432. formation of jogs in screw dislocations impedes their motion and may even lead to the formation of vacancies and interstitials if the jogs are forced to move nonconservatively. 1957. "Dislocations and Mechanical Properties of Crystals.and strain On the other hand. 1959.. Accordingly. Basinski. During easy glide. Generalized flow curve for fee single crystals. and therefore they contribute to hardening. certain generalizations can be made for all fee metals. Strain hardening due to a dislocation cutting process arises from shortrange forces occurring over distances less than 5 to 10 interatomic disThis hardening can be overcome at finite temperatures with the tances. 8.' John Wiley & Sons. pp. 2 Z. Mag. .. A. 4. When the stressstrain curves for single crystals are plotted as resolved shear stress vs. shear strain. The low strain hardening produced during this stage implies that most of the dislocations escape from the crystal at the surface. 114 Metallurgical Fundamentals [Ch ap. Jogs in edge dislocations do not impede their motion. New York.. Seeger. is and therefore it relatively independent of temperature data on the temperature and strainrate of strain hardening can be used^ to determine the relative contribution of the two mechanisms. 430) Stage I. the region of easy glide. Following the notation proposed by Seeger. strain hardening arising from dis location pileup at barriers occurs over longer distances. Phil.
an orientation favorable for simple slip. particularly during the early stages. the orientation of the crystal. During stage II. 430 represents a general behavior for fee Certain deviations from a threestage flow curve have been For example. A region of easy glide in the flow curve is favored by slip on a single system. In this stage. T3. The fact that the slope of the flow curve in stage II is nearly independent of temperature agrees with the theory that assumes the chief strainhardening mechanism to be piledup groups of dislocations. strain. Stage III is a region of decreasing rate of strain hardening. When is the tensile axis is parallel to a (Oil) direc one slip system carrying appreciably more shear stress than any other and the flow curve shows a relatively large region of easy glide. usually show only a very small stage II region at room temThe shape and perature because they can deform so easily by cross slip. the temperature at which it is tested. is strongly temperature dependent. The curve shown metals. and approximately independent of crystal orientation and purity. planes. the stresses are high In dis enough so that locations can take part in processes that are suppressed at lower stresses. The stress at which stage III begins. Cross slip is believed to be the main process by which dislocations. of fee single crystals. metals with a high stackingfault energy. stage I slip a nearly linear part of the flow curve where strain hardening increases r. like aluminum. tion.Sec. The easyglide region is much more prominent in hep crystals than in fee metals. high purity. sometimes called laminar flow.apidly. piled up at obstacles during stage II. The proc esses occurring during this stage are often called dynamical recovery. the stress on not very different and the flow curves show rapid rates of strain hardening. II. low temperature. depends upon the purity of the metal. and the rate at which it is strained. . and a method of Figure 431 shows testing which minimizes extraneous bending stresses. slip occurs on more than one set of The length which is of the active slip lines decreases with increasing consistent with the formation of a greater number of CottrellLomer barriers with increasing strain. in Fig. the ratio of the strainhardening coefficient (the slope of the curve) to the shear modulus is nearly independent of stress and temperature. this region of the flow curve. Stage II is For this reason. magnitude of a singlecrystal flow curve. absence of surface oxide films. When the tensile axis is is several slip systems close to a (100) or (111) direction. can escape and reduce the internalstrain field. the flow stress of a crystal strained into stage III if it is more temperaturedependent than had been strained only into stage This temperature dependence suggests that the intersection of forests of dislocations is the chief strainhardening mechanism in stage III. 412] slip is Plastic Deformation oF Single Crystals 115 always occurs on only one slip system. Also. that crystal orientation can have a very strong eff"ect on the flow curve observed.
perature Ti and then the temperature is increased to T2 without any Resolved Fig. H. Effect of specimen orientation on the shape of the flow curve for fee single crystals. A233. (London). Starting as close to absolute zero as is practical. 17. Shear strain Fig. vol. the value of the resolved shear stress at a given shear strain decreases with increasing If fee crystals are strained to the end of stage II at a temtemperature. . 1955. The state of strain hardening reached at Ti is unstable at T2. p. Flow curves exhibiting work softening. 432. the flow stress release at T2 of dislocation pileups locations 1 may produced at Ti. 432). Cottrell Stokes. The release of disbe due to easier cross shp at the higher temperature or and R. A. shear strain 431. drops from ri to t2 (Fig. Work softening is the result of the change in strain.116 Metallurgical Fundamentals [Ch ap. This behavior is called^ work softening. Proc. J. Roy. Soc. and a recovery process sets in which tends to reduce the strain hardening to what it would have been if all the straining had been accomplished at T2.
Ltd. and N. London. L. H. Hargreaves: Work Hardening of Metals. Clarebrough. New . L. in "Progress in Metal Physics. 1950.. 412] Plastic Deformation of Sinsle Crystals is 117 the fact that the size of a stable dislocation pileup of increased smaller at T2 because thermal fluctuations.: "Introduction to Solids. Boas: "Plasticity of Crystals. E. E. Barrett. "The Structure of Metals." McGrawHill Book Company. Chen: Geometrical Aspects of the Plastic Deformation of Metal Single Crystals. Inc. New York. R. 1953." 2d ed. BIBLIOGRAPHY Azaroff.. C. A. A. V. York. 1954.Sec.: Inc.. Cottrell. 1959.. Pergamon Press. F. Ltd.. 1960. New York. 5. and M.. 1952..: "Dislocations and Plastic Flow in Crystals. S. in "Progress in Metal Physics." English translation. Maddin. 8. Pergamon Press. London. Schmid. Hughes & Co. M. K.." vol. London. and W. McGrawHill Book Company." vol." Oxford University Press..
size. 118 . Introduction single crystals in The previous chapter considered the plastic deformation of metallic terms of the movement of dislocations and the basic Singlecrystal specimens deformation mechanisms of slip and twinning. Wherever possible. Other topics covered tions.Chapter 5 DEFORMATION OF POLYCRYSTALLINE AGGREGATES PLASTIC 51 . annealing. single crystals are rarely used for practical purposes because of limitations involving their strength. strain aging. which describe plastic deformation in single crystals because Therefore. results The simplification which from the singlecrystal condition materially assists in describing the deformation behavior in terms of crystallography and defect structure. a gap between fundamental deformation mechanisms determined from single crystals and the prediction of the plastic behavior of a polycrystalline aggregate from these basic concepts. and the development of preferred orientabe appreciated that not all these topics are solely restricted However. with the exception of electronic and semiconductor Commercial devices. However. represent the metal in its most ideal condition. primarily in These factors will each terms of how they influence the tensileflow curve. and be dispersion of in secondphase this particles. the bulk of the experimental to polycrystalline materials. Other factors which also have an important effect on mechanical properties are the presence of subgrain boundaries within the grains. metal products are invariably made up of a tremendous number of small The individual grains of the polycrystalline aggregate do not deform in accordance with the relatively simple laws individual crystals or grains. solidsolution alloying additions. considered chapter. Grain boundaries exert a considerable influence on the plasticdeformation behavior of polycrystalline metals. cold work. there is of the restraining effect of the surrounding grains. qualitative explana tions of these processes will be given in terms of dislocation theory. It will in this chapter include yieldpoint behavior. and production.
and precipitation reactions. and therefore they are considered in this chapter. and for relatively fast strain rates (so that recovery effects are not great). New York 19. 52] Plastic Deformation oF Polycrystalline Aggregates 119 data on these phenomena have been obtained from polycrystalHne materials. rate of strain. and the purity of the metal. phase transformations. At high temperatures and slow strain rates (conditions of creep deformation) deformation is localized at the grain boundaries. 2. In the general case. Ordinary highangle grain boundaries are boundaries of rather high For example. so that fracture occurs in an intergranular rather than transgranular fashion. while the energy of a twin boundary is only about 25 ergs/cm^. Because of their high energy. Oxford University Press. depending upon the temperature." chap. Grainboundary sliding and stressinduced migration can The occur.57 . temperature. grain boundaries increase the rate of strain hardening and increase the strength. Grain boundaries may serve to either strengthen or weaken a metal. At temperatures below approximately onehalf of the absolute melting point. the state of order in the grains on each the boundary increases.Sec. a grain boundary in copper has an intersurface energy. For the limiting case of a lowangle boundary where the orientation difference across the boundary may be less than 1° (see Sec. 52. 53). facial surface energy of about 600 ergs/cm^. This makes it difficult to separate the pure mechanical effect of grain boundaries on properties from an effect due to impurity segregation. Grain Boundaries and Deformation The boundaries between grains in a polycrystalline aggregate are a region of disturbed lattice only a few atomic diameters wide. is called the equicohesive The 1 principal difference between the roomtemperature deformation of singlecrystal and polycrystalline specimens is that polycrystalline For a review of the proposed models of grain boundaries see D. "Grain Boundaries in Metals. An important point to consider is that the high energy of a grain boundary usually results in a higher concentration of solute atoms at the boundary than in the interior of the grain. grain boundaries serve as preferential sites for solidstate reactions such as diffusion. McLean. The ordinary highangle grain boundary represents a region of random misfit between As the difference in orientation between the adjoining crystal lattices.^ side of the boundary decreases. and eventually fracture takes place at the grain boundary. fairly narrow temperature region in which the grain boundaries become weaker than the interior of the grains. the crystallographic orientation changes abruptly in passing from one grain to the next across the grain boundary. the boundary is composed of a regular array of dislocations.
p. The fact that slip lines stop at grain boundaries can be readily observed with the light microscope. Soc. microscope in a thin R. High shear stresses are developed at the head of a dislocation pileup. Hardening due to dislocation pileup . as observed with the electron foil of stainless steel. and eventually this becomes high enough to produce dislocation movement in the neighboring grain across the boundary. 5 The stressstrain material exhibits a higher rate of strain hardening. Associated with the increased strain hardening is usually an increase in yield stress aries and tensile strength. 1957. and W. 62) and highmagnification electron microscopy of thin films it is possible to establish that dislocations pile up along the slip planes at the grain boundaries (Fig. J. W. Only stage II and stage III deformation are obtained with polycrystaUine specimens. by means of special etchpit techniques (Sec. Of greater importance is the fact that the requirement for continuity between grains during deformation introduces complex modes of deformation within the individual grains.120 Metallurgical Fundamentals [Chap. B. This will reduce the dislocation pileup and mini back stresses which oppose the generation Read more sources within the grains. Proc. The first is the fact that Fig. 51).500 X. Roy. Dislocation pileups produce of new dislocations at FrankWith increasing applied stress more and dislocations pile up at grain boundaries. Dislocations piled up against a grain boundary. Home. 240A. mize hardening from this effect. 51 . Hirsch. The effects of grain boundon strength are due to two main factors. P. (London). Bollman. However. curve for polycrystaUine material shows no stage I or easyglide region. 524. vol. [M. 17. Slip on multipleslip systems occurs very readily in polycrystaUine specimens. Whelan.] grain boundaries are barriers to slip.
no grain can be very unfavorably oriented with respect to the applied stress. cit. slip can be initiated in a neighboring grain at only a little higher stress than was required to initiate slip in the most favorably oriented grains. is therefore important in the early stages of deformabut not at large strains. The effect of crystal orientation on the flow curve of fee single crystals was illustrated in Fig. . 62. the early stages of deformation. than in fee or bee metals. However. Inst.. yield stress is more dependent on grain size than tensile strength. with only one easy slip plane. For the latter case. vol. Grain size has a measurable effect on most mechanical properties. From purely geometrical considerations the grains of a polycrystalline metal must remain in Taylor^ has shown that five independent contact during deformation. I. there may be a very unfavorable orientation difference between neighboring grains so that an appreciably higher stress Therefore. It will be more effective in an hep metal. Orientations which produce many favorably Multiple slip oriented slip systems readily deform by multiple slip. Since slip on only two or three systems. Thus..Dy^ = yield stress = friction stress opposing motion of a dislocation = measure of extent to which dislocations are piled up at barriers D = 1 grain diameter G. slip systems must operate in each grain in order to maintain continuity. and impact resistance all increase with decreasing grain size. crystalline hep metals show a very much higher rate of strain hardening compared with single crystals. occurs for multiple slip in single crystals. polyis required to initiate slip in the neighboring grain. chap. J. 431. op. always results in a high rate of strain hardening.Sec. Taylor. slip in polycrystals is more complex Greater strain hardening is usually observed in polycrystals than can be accounted for on the basis of multiple slip in single crystals and by grainboundary barriers. For example. 2 McLean. 1938. p. 307. fatigue strength.^ than in single crystals oriented for multiple slip. for hep metals. 521 Plastic Deformation of Polycrystalline Aggregates 121 at grain boundaries tion. on the average. hardness. For most metals the yield stress is related to the grain size by (51) where an (Xi K. Metals. with many equivalent slip planes. For the later stages of deformation the strength is controlled chiefly by complex dislocation interactions occurring effect of grain size is largest it is The within the grains. at room temperature. cro = ai + K. In fee and bcc metals the difference in the flow curve between polycrystals and single crystals is not nearly so great. for on properties which are related to at this stage that grainboundary barriers are most effective. and grain size is not a controlling variable. depending on orientation. 6. yield strength. tensile strength. so that.
Mag. 2 3 J. 8. 173. the United States is of polish A very to common method measuring grain size in com pare the grains at a fixed magnification with the American Society for Testing Materials (ASTM) grainsize charts. This can be related^ to the grainboundary surface area S to the volume of the grains. vol. It is essentially independent of temperature. 866. E. p. 259266. O. Fetch.. J. p. S. 1960. appears possible to obtain an estimate of the lattice resistance to dislocation motion from an analysis of the grainsize depend ence of yield stress. J. Hall. pp. Trans. Only moderate agreement has been obtained. vol. and the PeierlsNabarro force. Soc. F. a measure of the extent to which dislocations are piled up at grain boundaries. . Phys. ^ Ada L. pp. 3. The analyses of this problem which have been made^ consist essentially in averaging the singlecrystal curves over different orientations. op. 1. vol. 197. 1951. Bishop. 81. The slope of a plot of oq versus D~^^ is Ky. W. vol. cit. by the equation where and A I is the total length of grain boundary on a of is the total area of the grains on a random plane random plane of polish. 1955. (London). The ASTM grainsize 1 N. 25. it the other resistances to dislocation motion are approximately tempera tureindependent. but it is independent of the applied Since the PeierlsNabarro force is temperaturedependent and stress. p. Iron Steel Inst. or by comparison with standard The average grain diameter lines D can be determined from measure(52) ments along random by the equation where L is ^ the length of the line and N D= is the number of intercepts which the grain boundary makes with the ratio of the line." The problem of determining the flow curve of polycrystalline material from singlecrystal data is difficult. subgrain boundaries.. Taylor. vol. (London). U. C. Smith and Guttman. Solids.. Phil. The intercept ai is a measure of the stress needed to drive a dislocation against the resistance of impurities. 1953. AIMS. 1953. Mech. Kocks. 1956. Grain size is measured with a microscope by counting the number of grains within a given area. This term depends on both the composition and the temperature. 64B. V. intersect a given length of charts. by determining the number of grains that random line. Proc. vol. 5 Equation was first proposed for lowcarbon steeP and has been extensively applied to tests on this material. Fetch. Met. Heslop and ISl J. and Phys. 345352. precipitate particles. p.122 Metallurgical Fundamentals (51) [Chap. J. F. 747. /.
Table 51 Comparison of Grainsize Measuring Systemsj ASTM No. . magnification of lOOX by the relationship N* = 2' (54)' Table 51 compares the ASTM grainsize numbers with several other useful measures of grain size. 53] Plastic is Deformation of Polycrystalline Aggregates 123 number n related to N*.Sec. the number of grains per square inch at a s .
52. between the grains is indicated by the In Fig. 5 slight difference in orientation angle 6. so that the spacing between dislocations is large. Therefore. there is an edge disthe grain boundary. evidence for the dislocation nature of lowangle boundaries comes from metallographic observations. The found validity of the dislocation model of the lowangle boundary is in the fact it is possible to calculate the grainboundary energy as a function of the difference in orientation between the two grains. an array of edge dislocations. their position by localized deformation to produce a location. good agreement is obtained between the measured values of grainboundary energy and Other the values calculated on the basis of the dislocation model. 250 X. If the angle is low. elastic deformation to cannot from one accommodate all the misfit. 536 the two crystals have been joined to form a bicrystal Along the boundary the atoms adjust smooth transition grain the other. However. containing a lowangle boundary. Substructure network in iron3 per cent silicon alloy. it is often possible to observe that the . relationship between d From the geometry of Fig.124 Metallurgical Fundamentals [Chap. 536 the and the spacing between dislocations is given by ^ = 2tani2^ = ^ (55) where h is the magnitude of the Burgers vector of the lattice. so that some of the atom planes must end on Where the atom planes end. So long as the angle does not become greater than about 20°. ^ Fig. lowangle tilt boundaries can be considered to be A ».
Jr. (b) two grains joined together to form a lowangle grain boundary made up of an array of edge dislocations. Read. Etchpit structures along lowangle grain boundaries in ironsilicon alloy. McGrawHill Book Company." p.Sec. T. ^'Dislocations in Crystals. 1953. ..) Fig. 157.. 1. 53] Plastic Deformation of Polycrystalline Aggregates 125 ia) Fig. (a) Two grains having a common and angular difference in orientation of 6. New [001] axis York.000 X. {W. 54. id) Diagram of lowangle grain boundary. Inc. 53.
or polygonization. Metals Park. (E. during hightemperature creep deformation. The amount of deformation and temperature must be low enough to prevent the formation of 1 new Sec. Fig. 5 pit boundary to the site of composed of a row of etch pits. They may be produced during crystal growth. When the crystal is heated. "Impurities and Imperfections. grains 512). 55. W. The veining in ferrite grains is a wellknown example of a substructure of ways. R. 1954. H. Ohio. ^ R. Cahn. id) Movement of dislocations to produce polygonization (schematic). 54). corresponding in a Subboundaries or lowangle boundaries can be produced number (a) Fig. Ohio.) when a single crystal bent to results sign. with each an edge dislocation (Fig. 556). Parker and T. "Relation The term occurs polygonization was used is originally to describe the situation that of Properties to Microstructure.126 Metallurgical Fundamentals is [Chap. 1955. . for Metals. the dislocations group themselves into the lowerenergy configuration of a lowangle boundary by dislocation climb.^ from the internal stresses accompanying a phase transformation. in Bending the introduction of an excess number of dislocations of one These dislocations are distributed along the bentglide planes as shown in Fig. 2 3 4 5 6 Density of subboundaries (arbitrary scale) by recrystallization (see This process has been called recrystallization in situ. or as the result of a phase transformation. The resulting structure is a polygonlike network of lowangle grain boundaries (Fig." American Society for Metals. 55a. 56." American Society Metals Park. Perhaps the most general method of producing a substructure network is by introducing a small amount of deresulting formation (from about 1 to 10 per cent prestrain) and following this with an annealing treatment to rearrange the dislocations into subgrain boundaries. a relatively small radius of curvature and then annealed. Effect of density of subboundaries on yield stress. Hazlett.
53] Plastic Deformation of Polycrystalline Agsregatcs 127 Since lowangle boundaries consist of simple dislocation arrays. expected be would what with ment also been found that the boundary angle decreases with increasing as a unit when subjected 80 n 70 60 o 850 i:40 30 20 10 . a study of their properties should provide valuable information on dislocation Parker and Washburn demonstrated that a lowangle boundbehavior. in complete agreeary moves It has for a linear dislocation array. ^ to a shear stress.
as approximately indicated by the lattice parameter. hydrogen. oxygen. Solidsolution Hardening solution in the solventis The introduction of solute atoms into solid atom lattice invariably produces an alloy which There are two types of solid solutions. Brick. much more soluble in copper than is copper in zinc. Roy. metal. The relative valence of the solute and solvent also is important. Trans. ASM. P. differ by less than 15 per cent. Trans. the extent of solid solubility to less than affinity for usually restricted 1 per cent. Faraday Soc. W. the ductility of the material containing a substructure is almost as good as the ductility of the annealed steel. 675698. the size factors is The factor favorable for solidsolution formation. vol. ^ For example. than both the annealed material and the material which Moreover. Angier. 19. Soc. D. HumeRothery. 54. (London). The is solubility of a metal with higher valence in a solvent of lower valence This relativevalence effect can be rationalized to a certain extent in terms of the electronatom more extensive than for the reverse situation. 506600. Proc. so as to produce a substructure. If the sizes of the two atoms. atoms are roughly similar.and solvent lattice points in atoms will occupy is the crystal lattice of the solvent atoms. they occupy interstitial positions in the solvent lattice. zinc is For certain solvent metals. M. and boron are the elements which commonly form interstitial solid solutions. Early studies'' of the increase hardness resulting from solidsolution additions showed that the hard For example.3. L. an alloy of 30 atomic per cent Zn in Cu has an electronatom ratio (3 X 2) + (7 X Ij = 13 valence electrons per 3 + 7 = 10 atoms. pp. 1924. vol. . When is the size factor is greater than 15 per cent. 181. H. L.128 Metallurgical Fundamentals [Chap. nitrogen. Norbury. 2 A. and R. R. Carbon. Metals which do not have a strong chemical each other tend to form solid solutions. Finally. Frye and ^ of 1. tional solid solutions which control the tendency for the formation of substituhave been uncovered chiefly through the work of HumeRothery. vol. has a higher yield point and tensile strength was only coldreduced. Martin. the limit of solubility occurs at approximately the same value of electronatom ratio for solute atoms of different valence. ratio. atoms. 5 annealed. 1942. 1943. The in acquisition of fundamental information about the causes of solid solution hardening has been a slow process. This called substitutional If the solute atoms are much smaller than the solvent solid solution. the solute stronger than the pure If the solute. for complete solid solubility over the entire range of composition the solute and solvent atoms must have the same crystal structure. J. while metals which are far apart on the electromotive series tend to form intermetallic compounds. pp. pp. 114. 31.
1. For a given atomic per cent {W. 1959. E. ^ 5 ^ ' and M.. Soc. cit. . C. Soc. 3. R. even in a perfect lattice the atoms would not be completely random. 215. 58. 1953. 1 J. Ainslie. Lacy Jr. and T.* copper. The tendency for clustering or shortrange order increases with increasing preferentially at dislocations. Met. Gensamer.^ For the case electronatom ratio R. grain boundaries.10 the flow curve in tension have been 1. However. vol. Systemic studies solidsolution of the effect of alloying additions on 1. or with the change in lattice parameter resulting from the solute addition. it is apparent that size factor alone cannot explain solidsolution hardening.. 2 3 W. 32.)ec. 1958. F. 212. solute additions. vol. E. pp. of solute the increase in strength varies inversely with the limit of solubility. vol.^ aluminum. pp. pp. Met. pp. 933 943. R. R. Trans. 188. 1958.^ of iron the is E!ectrono!om ratio Fig. 624628. Pietrokowsky. E. lattice parameter. V. if a given B atom is preferentially surrounded by A atoms. Ada Met. and Tietz. 188. Trans. pp. pp. 88110. Ji. G. AIME. and electronatom ratio have the same initial yield stress. solidsolutionstrengthened alloy a on the yield tion alloys. W. 15. 212. P. Hibbard. Dorn. Tietz. 1944. Effect of nickel. N. 4248. Zackay and T. 54] Plastic Deformation of Polycrystaliinc Aggregates 129 ness increase varies directly with the difference in the size of the sohite and solvent atoms.15 1. stacking faults. stress of power function of the alloy addition. vol. R. Trans. Met. The imporcopper tance of valence plotted ratio. p. For a solid solution of A and B atoms. vol. ASM. Hazlett. vol. Figure 59 shows the increase in tensile strength due to alloying additions in iron. vol. Soc. Pietrokowsky. but the flow curves differ at larger strains. op. 1950. copper solidsoluHibbaid.. 58. S. the solid solution exhibits shortrange order.) AIME. 1950. if B atoms tend to group themselves preferentially around other B atoms. 5358. Trans.' is shown in Fig. R. French and W. AIME. H. Trans. Dorn. Trans. Guard. AIME. is where the yield stress of alloys of constant lattice parameter against the electronatom Further results^ show that alloys with equal grain size. An improvement in correlation of data' results when the relative valence of the solute and solvent are considered in addition to the latticeparameter distortion. However. lowangle boundaries.20 made and for iron. There is growing evidence that solute atoms group and However. The distribution of solute atoms in a solvent lattice is not usually completely random. Hibbard. thp situation is called clustering. Hibbard.. Jr. AIME. and W.
acting on the The net stress will be nearly zero. atomic per cent of alloy added. Trans. 5 is hardening not simply the result of internal stresses due to the local lattice disturbance from randomly Consider a dislocation line in a perfectly random On the average. vol. Since the atoms in the region above a positive edge dislocation are compressed and below the large plane are stretched.) Lacy and M.^ it is generally held that hardening from solute atoms results from the interaction of solute atoms.4 0. H. p. Interstitial atoms collect in the expanded A.130 It is Metallurgical Fundamentals likely that solidsolution [Chap. % solute Increase in tensile strength of iron due to solidsolution alloy additions vs. New . positive and negative stress fields. and the dislocation dislocation line. 59. York. ASM. 0. (C." with dislocations. dispersed solute atoms. Gensamer. Cottrell. the strain energy of distortion can be reduced by atoms collecting in the expanded region and small atoms collecting in the 1 compressed region. in the form slip of "atmospheres. due to solute atoms.7 Atomic Fig. E. Following the ideas of Cottrell. 32. will move through the lattice almost as easily as through the lattice of a pure metal." Oxford University Press. there will be equal numbers of solidsolution lattice. "Dislocations and Plastic Flow in Crystals. 88. 1944. 1953.
. American Society for Metals. alloys this chemical interaction is While for most weaker than the interaction force due to Cottrell locking. 1952. Sci. there is a "chemical greater than the average bulk concentration. . but in an alloy with shortrange order slip will partially destroy the The same 1 Hazlett. 9. When free to move at a lower stress. Metals Park. H. Fisher. Sons. Research Insts. Parker and T.has pointed out the Thermodynamic reasoning existence of a third type of interaction. 1957. pure metal does not change the internal energy of the lattice. J. R. C. 361. 1954. a higher stress is required to make the dislocation move than would be required if there were no interaction between the dislocation and the solute atoms. solid solutions. Cottrell. the dislocation is interaction wj^^h a soluteatom atmosphere is the existence of an upper and lower yield point in iron and known to be associated with interstitial solute atoms (see Sec. between dislocations and solute atoms is given by A. the dislocation can be this away from its atmosphere. much as in the case of Cottrell Slip in a Fisher^ has pointed out that the existence of shortrange order or clustering in an alloy will produce a strengthening effect. 455463. The bestknown case of dislocation torn happens.." p.Sec. Metals Park.^ interaction between the solute due to elastic is atoms and the is dislocations. 131162. H. Inc. 1954. because the configuration of atoms across the slip plane is the same after slip as before. Ohio. Suzuki. American Society for A fairly mathematical discussion of the interactions Metals. estimates show that electrical interaction only about onethird to one seventh as strong as elastic interaction. "Relation of Properties to Microstructure. 1954. Tohoku Univ. shows that the concentration of solute atoms at a stacking fault will be Thus. The upper yield point corresponds to the stress required to tear dislocations away from other metals. 5.. 2 H. solidsolution strengthening." pp. vol. situation would exist for a completely random solid solution. such as described above for interstitial atoms. no. 3 John Wiley & New York. electrical interaction certainly an important factor in also be considered. is In view of the valency effects observed in must However. "Dislocations and Mechanical Properties of Crystals. Theories of solidsolution strengthening are reviewed by E." pp. If the stress becomes high enough. "Relation of Properties to Microstructure. 5053. Suzuki. Ohio. 55). 2. Repts. the force due to chemical interaction does not decrease with increasing temperature nearly so locking. 54] Plastic Deformation of Polycrystalline Aggregates 131 region below the slip plane of a positive edge dislocation. of a yield point in iron is The occurrence their atmospheres of interstitial atoms. pp. Ada Met. interaction" between these regions and dislocations. 4A. vol. local Because the energy is lower when a dislocation is surrounded by a solute atmosphere. p. Cottrell locking A number of types of soluteatom interaction must be considered in explaining solidsolution strengthening.
Ordered alloys with a fine domain size (approximately 50 A) are stronger than the disordered state. and this results in an increase in the stress required to produce slip. and then rises with further strain. 55. The deformation occurring throughout the yieldpoint elon13531354. Ohio. but which are out of step with the order in the neighboring domains. vol. what is equivalent. This arises from the fact that the dislocations in a wellordered alloy are grouped into pairs.132 Metallurgical Fundamentals slip plane. 693704. Ordered alloys with a large domain size generally have a yield stress lower than that of the disordered state. P. 1954. American Society for Metals. A. Flinn. 510. . pp. Because more domain boundaries are produced as slip continues. each pair having a Burgers vector twice as large as that for the disordered lattice. 131162. N. An ordered crystal will contain domains within which the order is perfect. vol. 218. In concentrated solid solutions strengthening from shortrange order should predominate. 145154. particularly lowcarbon steel. In a binary alloy with longrange order each of the constituent atoms is In effect. Mag. fluctuates about some approximately constant value of load. 1960. a loadelongation diagram similar to Fig. this results in a superlattice occupies special sites in the lattice. there is an interaction between dislocations and these antiphase boundaries. 1959. the rate of strain hardening is higher in the ordered condition than in the disordered state. vol. 31. Phil. AIME. The chemical interaction of Suzuki would be expected to predominate over shortrange order in dilute solutions. Herman. AIME. 1954. Metals Park. Trans. Brown. 4. "Relation of Properties to Microstriicture. where the stackingfault energy decreases rapidly with concentration. Yieldpoint Phenomenon Many metals. Since the domain boundaries are a highenergy interface. 1 N. such as was shown in Fig. The stress required to produce slip varies inversely with the distance between domain boundaries. 206. A. [Chap. pp. Rather than having a flow curve with a gradual transition from elastic to plastic behavior. drops suddenly. Brown and M. The load at which the sudden drop occurs is called the upper yield point. H. Cottrell. pp. with a larger unit cell and a new crystal structure. Trans. heterogeneous type of transition from elastic to plastic deformation which produces a yield point in the stressstrain curve. The interaction of dislocations with longrange order^ results in a strengthening effect." pp. and the elongation which occurs at constant load is called the yieldpoint elongation. metals with a yield point have a flow curve or.. The constant load is called the lower yield point. The load increases steadily with elastic strain. 5 pattern of order across the An internal surface of increased energy produced at the slip plane. show a localized.
and aluminum afloys and in cadmium. the the propagated to cover the entire length of This marks the end manner. is However. Typical yieldpoint behavior. yield points have been observed in recently the yield point has since it polycrystalline single crystals of iron. Low and M. zinc. the first Liiders R. often readily along the length of the specimen.207. A number of experimental factors affect the attainment of a sharp upper yield point. very careful axial alignment of the specimen. has been shown^ that almost complete removal of carbon and nitrogen from lowcarbon steel by wethydrogen treatment will remove the yield point. generally at approximately 45° to the tensile axis. the flow curve during the yieldpoint regular. When several Liiders bands are Elongotion Fig. and coincident with the formation of the band the load drops to the lower yield point. Gensamer. 1944. testing at If. Upper yield point They hands. titanium. alloys. A yieldpoint phenomenon was found originally in lowcarbon steel. type of deformation is sometimes referred to as the Piobert effect. the use of specimens free from stress concentrations. The band then propagates deformed metal.001 per cent of either of these elements required for a reappearance of the yield point. will increase with strain in usual flow the of the yieldpoint elongation. frequently. alpha and beta brass. appears at a stress concentration such as a fillet. of stress concentrations. formed. specimen test section. A sharp upper yield point is promoted by the use of an elastically rigid (hard) testing machine. . and alumiUsually the yield point can be associated with small amounts of For example. called lines. molybdenum. 1 subambient temperatures. p.. through careful avoidance J. causing the yieldpoint elongation. come to be accepted as a general has been observed in a number of other metals and In addition to iron and steel. In the usual case several bands will form at several points of stress These bands are concentration. are usually Luders Hartmann and this or stretcher strains. high rate of loading. num. only about 0. each jog corresponding After the Liiders bands have to the formation of a new Liiders band. 158. Trans. pronounced upper and lower yield point and a yieldpoint elongation of over 10 per cent can be obtained wdth this material under proper conditions. it interstitial or substitutional impurities. AIME. vol. The More phenomenon. . and. At the upper visible jdeld pomt a discrete band of with the eye.55] Plastic Deformation of Polycrystalline Aggregates 133 gation is heterogeneous. elongation will be ir 510.
This releases an avalanche of dislocations into the slip plane. and these pile up at the grain boundary.. Solute atoms diffuse to dislocations because this lowers the strain energy of the crystal. vol. G. 1955. The local concentration of solute atoms near the dislocation. (London). A. Acta Met. because a screw dislocation ordinarily has no tensile component. More recently the theory has been modified to show that there is a strong interaction between interstitial atoms and is nonsymmetrically deformed by the component of stress is developed. 2 W. Phijs. . 4962. The stress at which the dislocations break away from their atmosphere corresponds to the upper yield point. interaction energy has a value between 0. Cottrell and B. A. Bilby. Wiedersich. 3. but when a certain breakaway stress has been reached. Cochardt. Soc. pp. upper However. it is more usual to obtain an upper yield point 10 to 20 per cent greater than to form at the middle the lower yield point. and H. The stress concentration at the tip of the pileup combines with the applied stress in the next grain to unlock the dislocations in that grain. the band can be made yield point can be roughly twice the lower yield point. c. Cottrell's concept that the yield point is due to the interaction of solute atoms with dislocations was introduced in the previous section. and in this way a Liiders band propagates over the a dislocation will tend to return to specimen. the solute atmosphere becomes more concentrated and below a critical temperature the atmosphere condenses into a line of solute atoms. 1949. vol. is related to the average concentration co by the relationship screw dislocations when the lattice solute atoms so that a tensile c = Co exp —T— (56) For carbon and nitrogen in iron the and 1. 1 A. These atoms occupy a position of maximum interaction energy just below the center of a positive edge dislocation running parallel where U is the interaction energy. movement of the dislocation becomes easier with increasing distance from the atmosphere.5 to the length of the dislocation. The shear stress required to tear away its a dislocation from its atmosphere Therefore.0 ev. 62A. goes through a maximum when plotted against displacement. atmosphere for small displacements. Schoek. The dislocations are then anchored in position by an atmosphere The original theory^ considered that solute atoms would of solute atoms. 134 Metallurgical Fundamentals [^hap. Proc. segregate only to edge dislocations. 533537. pp. 5 of the test specimen. As the temperature decreases. H.
in which the strength of a metal Strain Fig. the Moreover. point is due to the diffusion of carbon and nitrogen atoms to the dislocations during the aging period to form new atmospheres of interstitials anchoring the dislocations. Strain Asing is Strain aging ductility a type of behavior. flow curve of a lowcarbon steel. immediately retested after Region C. yield point will reappear. which schematically describes the effect of strain aging on the is point phenomenon. Consider now that the specimen is strained to point Y and unloaded. Note that on reloading the yield point does not occur. through yield point. Support for this mechanism is found in the fact that the activation energy for the return of the yield point on aging is in good agreement with the activation energy for the diffusion of carbon in alpha iron. 511 A strain curve for a lowcarbon steel strained plastically and nitrogen atoms. 511. 56] Plastic DeFormation of Polycrystalline Aggregates 135 56. 511. since the dislocations have been torn away from the atmosphere of carbon Region of Fig. reappearance and increase in yield point after aging at curves for lowcarbon steel showing strain aging. Region A. the yield point will be increased The reappearance of the yield to Z. Region B. Fig. then unloaded and retested without appreciable delay or any heat treatment (region B). Stressstrain original material strained reaching point A'. usually associated with the yield is increased and the decreased on heating at a relatively low temperature after This behavior can best be illustrated by considering cold working. by the aging treatment from F . If it is reloaded after aging for several days at room temperature or several hours at an aging temperature like 300°F. shows the stressthrough the yieldThe specimen is point elongation to a strain corresponding to point X. 300°F.Sec.
there is no commercial lowcarbon steel which is completely nonstrain aging. and boron have been added for this purpose. Aluminum. or repeated. so the existence of strain aging has come to be recognized in metals other than lowcarbon steel. ASM. also called the PortevinLe and aging while the specimen This phenomenon is due to successive This results from is being tested. vol. Discontinuous yielding is observed in aluminum3 per cent magnesium alloys. While a certain amount of control over strain aging can be achieved. and plaincarbon steel. 1952. 9). yielding. This temperature range is also the region in which steels show a minimum in strainrate sensitivity and a maximum in the rate of strain aging. Just as the existence of a yield point has become recognized as a general metallurgical phenomenon. This temperature region is known as the blue brittle region because steel heated in this temperature region shows a decreased tensile ductility and decreased notchedimpact resistance. 44. it is usually desirable to lower the amount of carbon and nitrogen in solution by adding elements which will take part of the interstitials out of solution in the form of stable carbides or nitrides. columbium. The strainrate sensitivity is the change in stress required to produce a certain change in strain rate at constant temperature (see Chap. The phenomenon of strain aging should be distinguished from a » J. The usual industrial solution to this problem is to deform the metal to point by roller leveling or a skinpass rolling operation and use it immediately before it can age.136 Metallurgical Fundamentals [Chap. 643666. Trans. In addition to the X return of the yield point and an increase in the yield point after aging. D. To control strain aging. . 5 Nitrogen plaj^s a more important role in the strain aging of iron than carbon because it has a higher solubihty and diffusion coefficient and produces less complete precipitation during slow cooling. alpha brass. of serrations in the stressstrain It is curve is known yielding as discontinuous. is All these facts point to the realization that blue brittleness not a separate process phenomenon but is just an accelerated strain aging. vanadium. duralumin. For plaincarbon steel discontinuous yielding occurs in the temperature region of 450 to 700°F. it the fact that in the range of temperature in which occurs the time is required for the diffusion of solute atoms to dislocations much less than the time required for an ordinary tension test. Lubahn. titanium. The occurrence Chdtelier effect. pp. it has been suggested^ that a serrated flow curve and a minimum in the variation of strainrate sensitivity with temperature are characteristics of strain aging. From a practical standpoint it is important to eliminate strain aging in deepdrawing steel because the reappearance of the yield point can lead to difficulties with surface markings or "stretcher strains" due to the localized heterogeneous deformation.
D. Therefore. 1954. In a multiphase alloy. It is almost impossible to vary these factors independently in experiments. as in the age hardening Plastic deformation is not necessary to produce of aluminum alloys. ductility. the structure may consist phase in a ductile matrix. or somewhat above. and strainhardening behavior of the matrix and second phase. Quench a type of true precipitation hardening that occurs on quenching from the temperature of maximum solubility of carbon and nitrogen in ferrite. and the interfacial energy and interfacial bonding between the phases. and it is very difficult to measure many of these quantities with any degree of precision. produces an increase in hardness and yield stress. quench aging.^ The two phases may be ductile and present in the microstructure in relatively massive form. the presence of secondphase particles in the continuous matrix phase results in localized internal stresses which modify the plastic properties of the continuous phase. Only a relatively small solubility as in alphabeta brass. A number of different conditions may be encountered. most commercial alloys contain a heterogeneous microstructure consisting of two or more metallurgical phases. American Society for Metals. 7194. and only a relatively small hardening effect can be produced in most alloy systems by solidsolution additions.Sec. the crystallographic fit between the phases. brittle On the other hand. and distribution of the secondphase particles. Therefore. of a hard. each phase contributes certain things to the overall properties of the aggregate. For twophase alloys produced by equilibrium methods. The strengthening produced by secondphase tive to the solidsolution strengthening usually addi produced in the matrix. Metals Park. the existence of a second phase ensures maximum solidsolution hardening because its presence resulted from supersaturation of the continuous phase. number." pp. 57] Plastic Deformation of Polycrystalline Aggregates 137 known aging is as quench aging. ' If the contributions from each A review of the effect of secondphase particles on mechanical properties has been given by J. Subsequent aging at room temperature. Many factors must be considered for a complete understanding of strengthening from secondphase particles. Starr. as in spheroidized steel or WC particles in a cobalt matrix in a cemented carbide cutting particles is tool. which occurs in lowcarbon steels. shape. the strength. Strengthening from Secondphase Particles number of alloy systems permit extensive solid between two or more elements. Moreover. These factors include the size. . 57. E. Ohio. "Relation of Properties to Microstructure. Dorn and C. our existing knowledge of the effect of second phases on mechanical properties is mainly empirical and incomplete.
Dorn and C. (b) equal (From J.) Fig. D. The average strain in the alloy at a given stress is then given by = /lei + fo^i (58) Figure 5126 shows the flow curve for a 0." pp. 5 phase are independent. Estimate of flow stress of twophase alloy. 7778. then the properties of the multiphase alloy will be a weighted average of the properties of the individual phases. the strain in each phase phase. 1954. (a) Equal strain. For example. of an alloy consisting of two ductile phases depends upon the total deformation and the volume fractions of the phases. is equal. Starr. and if very little of the stronger usually lie The deformation .5 volume fraction of phase 2 on the basis of the equalstrain hypothesis. the average stress in the alloy for a given strain will increase linearly with the volume fraction of the strong Oavg The volume fraction of phase 1 + /2Cr2 is/i. American Society for Metals. Slip will occur first in the weaker phase. "Relation of Properties to Microstrudure.138 Metallurgical Fundamentals [Chap. and/i + /a = = /lOl (57) 1.5volumefraction alloy on the basis of the equalstress hypothesis. Metals Park. structuresensitive mechanical properties the properties of the aggregate are generally influenced by interaction between the two phases. the density of a twophase alloy will be equal to the sum of the However. Ohio. Both these hypotheses are simple approximations. 512. simple hypotheses may stress. Figure 512a shows the calculation of the flow curve for an alloy with 0. E. An alternative hypothesis is to assume that the two phases are subjected to equal stresses. and the strengths of alloys containing two ductile phases somewhere between the values predicted by the two models. for the volimie fraction of each phase times its density. Two be used to calculate the properties of a twophase It it is assumed that alloy from the properties of the individual phases.
pp. the second phase in dispersionhardening systems has very little solubility in the matrix. Sci. This is the situation in heattreated steel with a tempered martensitic structure. most of the deformation will continue in the softer phase. Repts. Because of the finely dispersed secondphase particles these alloys are much more resistant to recrystallization ' L.3. it is at least theoretically possible to produce an almost infinite number of dispersionhardened systems by mixing finely divided metallic powders and secondphase particles (oxides. Australian J. 57] Plastic Deformation of Polycrystalline Aggregates 139 is present. 7290.) and consolidating them by powder metallurgy techniques. as in oxygenfree copperbismuth alloys or hypereutectoid steel. precipitation hardening. M. or age harden is produced by solution treating and quenching an alloy in which a second phase is in solid solution at the elevated temperature but precipitates upon quenching and aging at a lower temperature. Usually there is atomic matching. By contrast. the alloy If the brittle phase is present as a grainboundary in the is envelope. 1950. . The strengthening produced by a finely dispersed insoluble second phase in a metallic matrix similar strengthening ing. vol. between the lattices of the precipitate and the matrix. they deform to about the same extent. On the other hand.^ The mechanical properties of an alloy consisting of a ductile phase and a hard brittle phase will depend on how the brittle phase is distributed in phase the microstructure. If the brittle phase is tinuous particles at grain boundaries.. At greater reductions the two phases deform more uniformly. 3. the soft phase deforms more than the hard phase for reductions of up to 60 per cent. Advantage has been taken of this method to produce dispersionhardened systems which are thermally stable at very high temperatures. When the phases are present in about equal amounts. The requirement of a decreasing solubility with temperature places a limitation on the number of useful precipitationhardening alloy systems. Clarebrough. A very phenomenon. For precipitation hardening to occur. as ness of the alloy when oxygen form added of discon to copper bismuth alloys or with internally oxidized copper and nickel. even at elevated temperatures. At large deformations flow of the matrix will occur around the particles If the volume fraction of the harder phase is less of the harder phase. is known as dispersion hardening. than about 0. the second phase must be soluble at an elevated temperature but must exhibit decreasing solubility with decreasing temperature. When the brittle phase is present as a fine dispersion uniformly distributed throughout the softer matrix. the brittleis reduced somewhat. carbides. borides. while in dispersionhardened systems there generally is no coherency between the secondphase particles and the matrix. The agehardening aluminum alloys and copperberyllium alloys are common examples.Sec. is brittle. etc. or coherency. a condition of optimum strength and ductility is obtained. nitrides.
Variation 01 yield stress with aging time (schematic). 86. As aging continues. After quenching from solid solution the alloy contains regions of solute segregation. 1 (J. the slip bands become finer and more closely spaced. the particles growth or overaging to a much greater extent than the secondphase particles in a precipitationhardening system. 1960. and therefore the hardness is low^er than at the stage when coherent Q' was present. so that the hardness of GP[1] Loss of coherency is higher than for the solid solution. With still further aging the equilibrium phase CUAI2. 51 3. Acta Met. The formation of a coherent precipitate in a precipitationhardening system. Nutting. 172176. Melals. pp. Continued aging beyond this stage produces particle growth and further decrease in hardness. ^ definite precipitate platelets (Particle size r. 4^4^ 4^ CUAI2. Thomas. is formed from the transition lattice Q' These particles are no longer coherent with the matrix. and therefore this structure is known as a GP zone. B. The clustering may produce local strain. such as AlCu.. [Chap. Thomas and J. . Figure 513 illustrates the way in which strength varies with aging time or particle size. With additional aging is the hardness increased further Equilibrium precipitate by the ordering of larger clumps of copper atoms on the {100} planes Qf ^j^g matrix. Although other precipitationhardening sysstages. or 0. vol. and Nutting. Nicholson. For most precipitationhardening alloys the resolution with the light microscope of the first precipitate occurs after the particles are no longer coherent with the matrix. fewer and fewer slip bands can be observed with the electron microscope. rig. J. The sequence of events in the AlCu system is creased strain field in the . as This structure or is known Aging time „ . lose it is quite common for a coherent coherency when the particle grows to a tems may not have so many precipitate to form critical size. 8. GP[2]. R. Q" . 19571958. 5 and grain growth than singlephase resist Because there is very little solubility of the secondphase constituent in the matrix. particularly complicated. • ^) • r 4o Tr ij it. G. occurs in a number of steps. As GP[2] zones form and the alloy proceeds toward peak hardness.^ In the asquenched condition slip bands are broad and widely spaced. or 0. The co herent precipitate produces an inmatrix and a further increase in hardness. or clustering. Insl. J. 714. and then Metallographic observations of deformation mechanisms in precipitationhardening systems require very careful techniques. pp.140 Metallurgical Fundamentals alloys. vol. which are coherent with the matrix. form on the {100} ' ' ' planes of the matrix. of . Next. special Guiner and Preston first detected this local clustering with Xray techniques.
86. Particle shape is of greater importance in notched impact. Shaw. 212. M. E. In addition to shape. where a spheroidized structure will be tougher than a lamellar structure.. Pearsall. vol. pp. Trans. Confirmation of this relationship has been found for tempered martensitic structures' and overaged AlCu alloys. For example. for a given carbon content. for a given volume fraction of second phase. reducing the particle size decreases the average distance between particles. 46. R. Gensamer. For a given size particle. S. J. . Detailed quantitative metallography on steels heattreated to provide has shown the relationship between Gensamer and coworkers. the strength will be higher for a fine carbide spacing than with a coarse interparticle spacing. E. the distance between particles decreases with an increase in the volume fraction of second phase. These factors are all interrelated so that one factor cannot be changed without affecting the others (see Prob. D. Jr. the slip once more can be observed. In general. 1958. /. 57] Plastic Deformation of Polycrystalline Aggregates 141 When lines the alloy begins to overage and coherency breaks down. Metals. Low. Quantitative relationships between strength and the geometrical factors have not been determined to any extent for real alloys. R.found that the flow stress. B. vol. Trans. Dorn. Turkalo and R. pp. although for a given volume fraction of carbides lamellar carbides will be stronger than spheroidized carbides. 10751088. B. W. 55). at a true strain of 0. 1 P. Trans. However. Particle shape has a less important effect on tensile properties. was inversely proportional to the logarithm of the mean interparticle spacing (mean free ferrite path) for pearlite and spheroidite structures (Fig. lines do not cut through the Instead. Met. For a noncoherent precipitate the particles.2.Sec. but eventually the dislocations shear Ihrouj^h the particles. the secondphase dispersion can be described by specifying the volume fraction. and 1954. and J. pp. Inst. the of dispersion hardening can be con common situation of carbide particles in ferrite being used as an example. ^C. pp. Electronmicroscope studies have dislocation motion is shown that impeded by slip fully and partially coherent precipitates. J. the dislocation lines bend to avoid the particles. Starr. 2 M. Soc. ASM. A.^ The degree of strengthening resulting from secondphase dispersions depends upon the distribution of particles in the soft matrix. ASM. AIMS. 30. Low. Further.. The figure shows the variation of different interparticle spacings strength and structure. Pellini. the hardness and strength increase with carbon content or volume fraction of the carbide phase. 195758. 750758.'' Figure 515 illustrates the marked strengthening produced by CuAl2 particles in an AlCu alloy. 3 9831020. 1314. Hirsch. the qualitative aspects sidered. B. vol. probably by a process of cross slip. Jr. 514). 1942. and mean interparticle spacing. vol. average particle diameter.
5 200 .142 Metallurgical Fundamentals [Chap.
Table 52 gives the variation of proportional limit and tensile strength sion.>ec. the Table 52 Variation of Tensile Properties with Volume Fraction OF Second Phase for CoWC Alloys! Volume . and the bottom curve is for an alloy without dispersed particles which contains the same amount of copper in solid solution as the top two alloys. The top curve i»s for a fine CuAU disper middle curve for a coarse dispersion. 57] Plastic Deformation of Polycrystalline Aggresates 143 flow stress witk temperature for an alloy containing 5 of volume per cent CuAU in three conditions.
p. At stage 2 the line is beginning to bend. leaving them encircled by small loops of dislocations. In '^ Ci>< Ci> c ® ® (5) (1) Fig. (59) the stress needed to force the dislocation line between the obstacles is given by Gb A (510) At stage 4 the dislocation has passed between the obstacles. This continues until the shear stress developed by the loops is high enough to shear the particles or the surrounding matrix. According to the theory developed by Fisher. Orowan.. ^ E. 516. 1953. 451. 1 Fig. vol. which tends to keep length." p. 1947. The smallest radius of curvature to which a dislocation line can be bent under the influence of an internal stress field Ti is given by R = Orowan' suggested that the sion of fine particles is Gb (59) 2Ti yield stress of an alloy containing a disper dislocation line between determined by the shear stress required to force a two particles separated by a distance A.144 Metallurgical Fundamentals it [Chap. H. Schematic drawing of stages in passage of a dislocation between widely separated obstacles —based on Orowan's mechanism of dispersion hardening. Fisher. Hart. London. E. discussion in Metals. and Pry. the presence of dispersed particles leads to increased strain hardening during the period when loops are building up around the particles. These dislocation dislocations loops exert a back stress which must be overcome by moving on the This requires an increase in shear stress to continue deformation. 1. Pry. C.^ the increase in shear slip plane. Every dislocation gliding over the slip plane adds one loop around the obstacle. (2) (3) (4) 516. stage particles separated by a distance shows a straight dislocation line approaching two A. Thus. Institute of 2 J. and at stage 3 it has reached the critical stage. 33G. from Eq. Ada Met. . W. Hart. Since A equals twice the critical radius of curvature. at its shortest any bending or increase in length of a dislocation line requires the expenditure of extra energy. 5 dislocation possesses line tension. and R. "Symposium on Internal Stresses.
Hardening Due to Point Defects Vacancies and interstitials are produced by the bombardment of a metal with highenergy nuclear particles. by the relationship TH = 3rJ" (511) where n is between 1 and 1 . It is difficult for the dislocation line to bend sharply enough to pass between the particles." pp. A situation in which the only point defects are vacancios can be produced by rapidly quenching a pure metal (so that there can be no 'A. 139.5. According to Eq. and so it shears through them instead. "Vacancies and Other Point Defects in Metals and Alloys. the dislocation line becomes more rigid. 58] stress Plastic DeFormation of Polycrystalline Aggregates 145 due to fine particles.Sec. is related to the volume fraction of the second phase. Cottrell. pro longed neutron irradiation can appreciably raise the transition temperature. such as steel. two point more mobile are even than vacancies. There are indications that in the region of small particle spacing the yield stress direct function of the radius of the particles. Hart. Neutron irradiation increases the hardness and yield strength of most metals. the critical radius of curvature is increased and the stress required to bend the dislocation line is decreased. ta. so that quite low temperatures are required to prevent them from interacting with other lattice defects. and the shear strength of a dislocationfree matrix. When the distance between particles is decreased. H. Tc. Institute of Metals. and vacancies are left behind. The Fisher. /. is a 58. . (510) the shear strength of a dispersionhardened alloy will have a maximum value when A has a value that makes it e(}ually likely that dislocations will pass between the particles or cut through them. 1958. Orowan's relationship between strength and particle spacing has been experimentally confirmed for most systems containing overaged or noncoherent particles. As the distance between particles is increased. In copper single crystals a dose of 10'^ neutrons per scjuare centimenter increases the yield strength by a factor of 10 and changes the deformation characteristics so that they are similar to alpha brass.' In metals which show a ductiletobrittle transition. and Pry equation for the contribution to strain hardening from dispersed particles also appears to be approximately verified. London. The structural changes producing radiation hardening and radiation damage are difficult to study in detail because at least Interstitials defects are acting simultaneously. The bombardment of the lattice with fast neutrons having energies up to 2 million ev knocks atoms into interstitial positions in the lattice.
5 precipitation of a second phase) from a temperature near point. a dispersion of point defects can produce hardening. and therefore. can be achieved by quenching. It is likely that the aging permits the vacancies to migrate to dislocations. vacancies are important in the creep of metals. tion appears to be particularly important in the fatigue of metals. by analogy with the hardening produced by a dispersion of secondphase particles. Because of the mutual interference of adjacent grains in a polycrystalline specimen multiple slip occurs readily. and it will be considered from this standopint in Chap. 13. Much remains to be learned about the interaction of point defects with each other and with line defects and how these interactions affect the mechanical properties. and with polycrystalline specimens it is not observed. in yield stress and a decrease in the rate of strain hardening. Vacancy formaa discussion of this topic will be deferred until Chap. 6. chiefly vacancies. Thus. parallel planes. and there is appreciable strain hardening. Therefore. copper. However. point defects are created by the intersection of dislocations. So long as slip takes place on only a single set of locations with other dislocations with single crystals of hep metals. At elevated temperatures vacancies become very important in controlling diffusion and in making possible dislocation climb. as of strain . 612). hardening results if an aging treatment is interposed between the quench and the measurement of the tensile properties. can be hardened by introducing a randomly distributed populaQuench hardening results in an increase tion of vacancies in this way. Soft metals. 59.146 Metallurgical Fundamentals its [Chap. 4 strain hardening was attributed to the interaction of disand with other barriers to their motion through the lattice. just as is produced by radiation hardening. 12. These Plastic deformation produces point defects. and they will be considered in greater detail in Chap. Strain Hardening and Cold Work In Chap. even with single crystals extensive easy glide is not a general phenomenon. Plastic deformation which is carried out in a temperature region and over a time interval such that the strain hardening is not relieved is called cold work. such as aluminum. The mechanism by which this There is some evidence that at this stage occurs is not yet established. melting At room temperature or below the metal contains a supersaturated solution of most of the vacancies that existed in equilibrium at Vacancy concentrations of up to about 10~^ the higher temperature. Plastic deformation produces an increase in the number of dislocations. only a small amount hardening occurs. and zinc. where they interact and impede the movement of dislocations (see Sec. A greater quench the single vacancies have migrated into clusters.
However.. Kelly. Warren and B. line Xray broadening can be due to both a decrease in size of the diffraction Groin boundary would occur if the grains were fragmented by cold work. 517. 147202. 59] Plastic Deformation of Polycrystallinc Asgregates 147 which by virtue stress. Pergamon Press. 41. London. Appl. while a severely plastically deformed metal contains about 10^^ dislocations per square centimeter. IB. niques for analyzing the entire peak profile of Xray lines and unit. Averbach. Gay. J. Averbach.>ec. "Modern Research Techniques in Physical Metallvirgy. A fairly reliable model of the structure of coldworked metal has developed from microbeam Xray studies^ and from electron microscopy of thin films. 2 P. E. 1953. Figure 517 is a schematic drawing of the coldworked structure that occurs within a single grain.. "Progress in Metal Physics. Strain hardening or cold work can be readily detected by Xray diffraction. Ohio. B. and an increase in lattice strain due Techto dislocation interaction. as Distorted regions of high dislocation density separating out the contribution due to lattice strain and particle Fig. B. and temperature. although detailed analysis of the Xray patterns in terms of the structure of the coldworked In Laue patterns cold work produces a state is not usually possible. The study of the dislocation structure of coldworked metal with thinfilm electron microscopy is a very active area of research which should provide valuable information about how these networks vary with composition. Phijs. Metals Park. Warren. 595. roughly about 10 per cent of the expended of relatively perfect regions of the lattice other by boundaries of dislocation networks. blurring. or asterism. vol. . Hirsch." vol. 1954. 7. Warren and B. L. An For DebyeScherrer patterns the Regions of relatively perfect lottice lines are broadened by cold work. Most of the energy expended in deforming a metal by cold working is converted into heat. 1950. 1959. L. and A.^ worked metal (schematic). Ltd. likely that improvements in this method and more widespread application of the technique will result in better understanding of the structure of coldworked metal. vol." American Society for Metals. 21. P. pp. Acta Cryst. Model of the structure of coldIt is size have been developed.. It is a celllike structure consisting which are connected with each According to this model the dislocation density varies drastically from a high value in the distorted boundaries to a low value in the relatively perfect regions. B. of the spots. deformation. E. p. p. of their interaction results in a higher state of internal annealed metal contains about 10® to 10^ dislocations per square centimeter. 8. E.
the final almost always greater than that of the pure metal coldworked to the same extent. Ltd. Bever. The stored energy increases with strain up to a limiting value corresponding to saturation. % the rate ot straui hardenmg may Fig. The rate of strain hardening can be gaged from the slope of the flow curve. 7. London. of the stored energy for metals Vacancies account for part deformed at very low temperature. Titchener in Metal Physics. wire drawing vs. deformation.148 energy is Metallurgical Fundamentals [Chap. pp. e. Stacking faults and twin faults are probably responsible for a small fraction of the stored energy. The major part of the stored energy is due to the generation and interaction of dislocations during cold working. ^e either increased or decreased compared with the behavior for the pure metal." vol. decreasing temperature of Very careful calorimeter measurements are required to measure the small amounts of energy stored by cold working.g. ture also lowers the rate of strain hardening. Figure 518 shows the typical variation of strength and ductility and M. Pergamon Press. Gener the rate of strain hardening Increasing tempera lower for hep metals than for cubic metals. L. A reduction in shortrange order during the deformation of solid solutions may also contribute to stored energy. the rate of strain hardening can be is expressed by the strainhardening coefficient ally. • alloys i strength ened 10 20 30 40 50 Reduction by cold work.01 to 1. "Progress For a comprehensive review of the stored energy of cold work see A. B. The magnitude of the stored energy increases with the melting point of the metal and with solute additions. 247338. (31). 518. It increases with.. is n in Eq. ' . 5 stored in the lattice as an increase in internal energy. For r . is strength of a coldworked solidsolution alloy However. tension. Elastic strain energy accounts for only a minor part of the measured stored energy.. However. 1958.. Strain hardening or cold working is an important industrial process that used to harden metals or alloys that do not respond to heat treatment. by . For a given metal the amount of stored energy depends on the type of deformation process. solidsolution i additions • 60 70 . Variation of tensile properties with amount of cold work. Reported values of stored energy^ range from about 0. vacancies are so much more mobile than dislocations that they readily escape from most metals deformed at room temperature. In mathematical terms.0 cal/g of metal.
51 9. If now a compressive stress is applied. Because of the increased internal energy erties. small elastichysteresis effects being neglected. 511). point B on the dashed curve.510] Plastic Deformation of Polycrystallinc Aggregates 149 parameters with increasing amount of cold work. the yield strength comwould be approximately the same. of the coldworked state chemical reactivity is increased. cold work produces Severe deformation produces a reorientation of the grains into a preferred In addition to the changes in tensile properties orientation (Sec. 510. Since in most coldworking processes one or two dimensions of the metal are reduced at the expense of an increase in the other dimensions. an appreciable decrease in electrical conductivity due to an increased number of scattering centers. initial yield stress of the ma in tension is A. cold working produces changes in other physical propelongation of the grains in the principal direction of working. shown in Fig. If the same ductile material were tested in pression. and a small increase in the thermal coefficient of expansion. Now. 518. is ure 519 an example Bauschinger of the is of stressstrain curve that type obtained is when The terial the effect considered. consider that a new specimen is loaded in Fig. which is appreciably lower than the original compressive yield stress of the material. There is usually a small decrease in density of the order of a few tenths of a per cent. was decreased. the yield stress in compression effect. path CD. This is the Bauschinger The phenomenon . If the specimen is then unloaded. Fig The is directionality of strain hardening called the Bauschinger effect. it will follow the Bauschinger effect and hys teresis loop. This leads to a general decrease in corrosion resistance and in certain alloys introduces the possibility of stresscorrosion cracking. Bauschinger Effect In an earlier discussion of the strain hardening of single crystals it was shown that generally a lower slip stress is required to reverse the direction of on a certain slip in slip plane than to continue the original direction. While the yield stress in tension was is increased by strain hardening from A to C. tension past the tensile yield stress to C along the path OAC. plastic flow will begin at the stress corresponding to point E.
New York. Therefore. is by fatigue is likely to occur." Elsevier Publishing Company. However. 57). when the load is removed. or texture. as in develop a preferred orientation. This is in agreement with the type of flow curve usually observed for the Bauschinger effect. (510). stronger and closer obstacles. precipitate particles. One way of describing the 13 amount This of is Bauschinger effect is by the Bau schinger strain (Fig. Obstacles to dislocation motion are considered to be other dislocations. then unloading. it encounters. inclusions. 1959. For a it given shear stress a dislocation line will move over the slip plane until meets a row of obstacles that are strong enough to resist shearing and close enough to resist the dislocation loop from squeezing between them. in which certain crystallographic planes tend to orient themselves in a preThe ferred manner with respect to the direction of maximum strain. Preferred Orientation metal which has undergone a severe amount of deformation. will 1 E. and those which do show only a small effect. the cycle repeated many times. not all metals show a permanent softening after strain reversal. in "Internal Stresses and Fatigue in Metals. Orowan. The area under the loop will depend upon the initial overstrain If beyond the is yield stress and the number failure of times the cycle is repeated. when the direction of loading is reversed. so that the shear stress continuously increases with strain.150 Metallurgical Fundamentals [Chap. 519). the dislocation line will not move appreciably unless there are very high back stresses. the dislocation line can move an appreciable distance at a low shear stress because the obstacles to the rear of the dislocation are not likely to be so strong and closely packed as those immediately in front of the dislocation. on the average. a mechanicalhysteresis loop is obtained. As the dislocation line moves. 519 is completed by loading further in compression to point F. the difference in strain between the tension and compression curves at a given stress. the flow curve after reversal of strain ought always to be softer than the flow curve for the original direction of strain. A . and reloading in tension. If the loading cycle in Fig. if the Bauschinger effect due solely to the effect of back stresses. for had the specimen compression. 511. etc. Now. The stress required to move a dislocation through these obstacles is given approximately by Eq. Causes and Effects of Internal Stresses. the yield stress in tension would have been decreased. However. Orowan^ has pointed out that. Orowan considers that the Bauschinger effect can be explained by the same mechanism which he proposed for dispersion hardening (Sec. tendency for the slip planes in a single crystal to rotate parallel to the axis rolling or wire drawing. 5 originally been stressed plastically in reversible.
" American Society for Metals. York. The Xray pattern of a finegrained randomly oriented metal will show rings corre sponding to different planes where the angles satisfy the condition for Bragg reflections. a compilation of pole figures describing the deformation textures in metals. 511] Plastic Deformation of Polycrystalline Aggregates in the previous chapter. A preferred orientation can be detected with X rays after about a 20 to scatter decreases with 30 per cent reduction in crosssectional area by cold working. 2 A. is essentially complete. If the grains are randomly oriented. Geisler.. Other factors which may be important are the temperature of deformation and the type of texture present prior to deformation. which is brought about by the alignment of inclusions. but if a preferred orientation be broken up into short arcs.Sec. New . the reader is many The current use of Geigertechniques'has made it possible to deterXray diffractometer counter mine pole figures with greater accuracy and less labor than with older referred to Barrett. until at about 80 to 90 per cent reduction the pre The type of preferred orientadeveloped depends primarily on the number and type of slip systems available and on the principal strains. which It is similarity to the arrangement in naturally fibrous materials. Inc. of the grains of a particular crystallographic orientation The with respect to the principal directions of working is best shown by means of a pole For a description of the methods of determining pole figures and figure." 2d ed. 151 of principal strain was considered The same situation exists in a polycrystalline aggregate. H. and secondphase constituents in the 1 C. but the complex inter actions between the multiple slip systems crystalline situation makes analysis of the poly much more difficult. lattice bending and fragmentation will occur. the rings will angles. Barrett. is deformation texture. the intensity of all the rings will be uniform for exists. The simplest deformation texture is produced by the drawing or rolling This is often referred to as a fiber texture because of its of a wire or rod. McGrawHill Book Com pany. or spots. 9..^ film methods. chap. The dense orientation areas of the Xray photograph indicate the orientation of the poles of the planes corresponding to the diffraction ring in question. important to note that a distinction should be made between the crystallographic fibering produced by crystallographic reorientation of the grains during deformation and mechanical fibering. 1953. At this stage of reduction there is appreciable scatter in the orientation of individual crystals about the ideal orientation. Preferred orientations are determined by Xray methods. "Structure of Metals. Metals Park. 1952. cavities. S. or The increasing reduction. Since the individual grains in a polycrystalline aggregate cannot rotate freely. "Modern Research Techniques in Physical Metallurgy. Ohio. ferred orientation tion.
and the texture is symmetrical around the wire or Several types of deviations from the ideal texture are observed. is There often considerable deviation from the ideal texture.^ This texture changes to the more common {110} (112) texture by the addiIn bcc metals the {100} planes tion of solidsolution alloying elements. resulting from deformation is strongly and twinning systems available for deformation. vol. cit. or swaging. The formation of a strong preferred orientation will result in an The preferred orientation dependent on the slip ^ It has been suggested that a (111) texture is favored by easy cross sUp. and reduction per pass.^ Precision determination of the rolling texture in fee metals has shown that the texture may be described best by the {123} planes lying parallel to the plane of the sheet with the (112) direction parallel to the rolling direction. In an ideal wire texture a definite crystallographic direction lies parallel to the wire axis. 236238. See N. 1961. Bodycentered cubic metals have a simple (1 10) wire texture. In facecentered cubic metals a double fiber texture is usually observed. pp. op. roll speed. 221. 3 R. 84. J. vol. The wire texture in hep metals is not so simple. tend to be oriented parallel to the plane of the sheet with the (110) direcFor hep metals the tion within a few degrees of the rolling direction. but it is not generally affected by processing variables such as die angle. For example. 5 Mechanical and crystallographic producing directional mechanical properties of plastically worked metal shapes such as sheet and rods. roll The direction of flow is the diameter. while above this is temperature (2110) is parallel to the fiber axis. direction of mechanical working. A large number of pole figures for rolling textures are given by Barrett. most important process variable. 2 AIME. This will be discussed further in Chap. ^ For moderate amounts of parallel to the fiber axis.. fibering main are important factors in The grains have either (111) or (100) parallel to the wire axis and have random orientations around the axis. so that pole figures are useful for describing the degree of preferred orientation.152 Metallurgical Fundamentals [Chap. E. basal plane tends to be parallel with the rolling plane with (2110) aligned in the rolling direction. which occurs most readily in metals with high stackingfault energy. Brown. Metals. Inst. the same deformation texture is produced whether a rod is made by rolling. 1018. is is about 20° from the wire parallel to the wire axis for deformation For magnesium and its alloys (1010) is below 450° C. pp. 195556. 18. 9. drawing. while for severe deformation the deformation the hexagonal axis (0001) of zinc hexagonal axis axis. Smallman. chap. . The deformation texture of a sheet produced by rolling described by the crystallographic planes parallel to the surface of the sheet as well as the crystallographic directions parallel to the direction of rolling. fiber axis. Trans.
mechanical properties in different directions can result in uneven response of the material during forming and fabrication operations. 512. and the metal to revert to the strainfree condition. Interscience Publishers. The overall process of annealing can be divided into three fairly distinct processes. ever. Recovery is help to distinguish between these processes. and grain growth. by interposing annealing operations after severe deformation it is possible to deform most metals to a very great extent. 3. Adv. A. Beck. 3. there a tendency for strainhardened  Reco\/ery i Recrystollization Grain growth Temperature Fig. Although the individual grains of a metal are anisotropic with respect to mechanical properties.* )ec. recrystallization. New York. 245Burke and D. the grain alignment that accounts for the preferred orientation Different again introduces an anisotropy in mechanical properties. Turnbull. Schematic drawing indicating recovery. vol. J. The overall process by which this occurs is known as annealing.. . 1954. "Progress in Metal Physics." vol. recrystallization. Inc. E. when these grains are combined in a random fashion into a polycrystalline aggregate Howthe mechanical properties of the aggregate tend to be isotropic. con For detailed reviews of annealing. recovery. metal softens and reverts to a strainfree condition. — and grain growth 520. in Phys.. 324. With increasing temperature Eventually the the coldworked state becomes more and more unstable. Figure 520 will usually defined Electrical as the restoration of the physical properties of the coldworked metal without ^ any observable change in microstructure. Annealins of Coldworked Metal The coldworked undeformed metal. state is a condition of higher internal energy than the is Therefore. pp. chief property changes in each region. 1952. see P.' Annealing is very important commercially because it restores the ductility to a metal that has been severely strainhardened. Therefore. 512] Plastic DcFormation of Polycrystalline Aggregates 153 anisotropy in mechanical properties.
53) can be considered an intermediate situation between recovery and recrystallization. As the temperature is increased. Figure 521 shows the progression from a coldworked microstructure to a fine recrystallized grain structure. Recrystallization is readily detected by metallographic methods and is evidenced by a The density decrease in hardness or strength and an increase in ductility. driving force for both recovery and recrystallization. as measured with X rays. and lattice strain. (a) recrystallization. Recrystallization is the replacement of the cold worked structure by a new set of strainfree grains. are those which are sensitive The strength which are controlled by locations. 5 ductivity increases rapidly toward the annealed value during recovery. and finally to a larger grain size by grain growth.) Fig. and all effects The stored energy of cold work is the of strain hardening are eliminated. by recovery properties. of dislocations decreases considerably on recrystallization. The fraction of the microstructure that has recrystallized in a time t can be represented by an of initially .154 Metallurgical Fundamentals [Chap. {Courtesy L. 150X. Recrystallization is the reversion by thermal activation of the coldworked structure to its original strainfree condition. Coldworked 40 per cent. 15 min. An exception to hep metals which have deformed on only one set For this situation it is possible to recover com pletely the yield stress of a strainhardened crystal without producing (a) (b) (D 521. (6) 400°C. 15min. are not affected at recovery temperatures. If the new strainfree grains are heated at a temperature greater than that required to cause recrystallization. Changes in microstructure of coldworked 7030 brass with annealing. Polygonization (Sec. The driving force for grain growth is the decrease in free energy resulting from a decreased grainboundary area due to an increase in grain size. (c) 575°C. A. The dis properties that are most affected to point defects. Monson. there will be a progressive increase in grain size. is appreciably reduced. the dislocation networks tend to contract and the regions low dislocation density begin to grow. this is single crystals of of planes (easy glide) .
" pp. Values of between and 2 indicate one dimensional recrystallization. The recrystallization temperature decreases with increasing purity of the metal.. Ohio. (5) composition. The smaller the degree of deformation. 2.. it is not a fixed temperature in the sense of a melting temperature. Solidsolution alloying additions always raise the recrystalrecjuired to lization temperature. the higher the temperature the required to cause recrystallization. Six (1) main amount They of prior deformation. recrystallized grain size. Increasing annealing time decreases the recrystallization temperature. F. different metalworking processes. such as rolling. the greater the amount of cold work produce an equivalent recrystallization temperature. Doubling the annealing time is approximately equivalent to increasing 4. Recrystallization. while values between 2 and 3 denote twoIt is convenient to consider the process of dimensional recrystallization. 6. For practical considerations a recrystallization temperature can be defined as the temperature at which a given alloy in a highly coldworked state completely The relationship of the above variables to the recrystallizes in 1 hr. produce somewhat different effective lization behavior increases with increased 8. 7. Sec. The final grain size to a lesser extent of depends chiefly on the degree of deformation and on the annealing temperature. (3) time. recrystallization in terms of the rate of nucleation A^ and the rate of growth G of new strainfree grains. However. recrystallization process can be summarized^ as follows. the annealing temperature 10°C. etc. sites of The relative values of N and G determine the many are If A^ is large with respect to G. the smaller the 5. (2) temperature. and (6) amount of recovery or polygonization Because the temperature at which recrystallization occurs depends on the above variables. » R. there are nucleation and the grain size will be relatively small. 259268. tion. . recrystallized grain size. The greater the degree deformation and the lower the annealing temperature. The larger the original grain size. (4) initial grain size. 1948. American Society for Metals. Metals Park. Mehl. The amount of deformation required to produce equivalent recrystal temperature of working. drawing. in "Metals Handbook. 1 A minimum amount of deformation is needed to cause recrystallizaprior to the start of recrystallization. 3. 512] Plastic Deformation of Polycrystalline Aggregates 155 equation of the form X= where 1  exp iBt"') n' (512) 1 B and n' are constants. variables influence recrystallization behavior. temperature is far more important than time. For a given reduction in cross section.
the grain size. which restricts grainboundary movement. grain is strongly temperaturedependent. grain growth. at a temperature at which recrystallization occurs readily grain growth will occur slowly. Anneal{ ingtexture formation depends on a number of processing variables. the amount and type deformation preceding annealing. 5 deformations. or recrystallization texture. the annealing temperature and time. not stored energy.2 to temperature. depending on the metal and the Under lized certain conditions. identical recrystallization behavior may not be Because the driving force for grain growth is appreciably lower than the driving force for recrystallization. This phenomenon is known as exaggerated. Grain growth is inhibited considerably by the presence of a fine dispersion of secondphase particles. However. Therefore. most experimental data agree best with an equation where n varies from about 0. For the usual type of grain growth. theory predicts that at a given temperature the grain size Z> at a time t is given by growth will /)2 . This is called an annealing texture. but because the phenomenon shows kinetics similar to those of recrystallization it is often called secondary recrystallization. An outstanding example is the cube texture in copper. Annealing Textures recrystallization of a coldworked metal is The may orientation which different from that existing in the produce a preferred deformed metal.156 Metallurgical Fundamentals [Chap. or abnormal. Generally the factors which favor the formation of a fine recrystallized grain size also favor the formation of an essentially random orientation of recrystallized grains. and a graincoarsening region soon be reached in which the grains increase in size very rapidly.D^ = Ct (513) However. 0. where the 100} plane lies parallel to the rolling plane with a (001) direction parallel to the direction of rolling. the composition of the alloy. The driving force for exaggerated grain growth is the decrease in surface energy. The existence of a recrystallization texture depends on a preferential orientation of the nuclei of the recrystallized grains. and the preferred orientation produced by the deformation. some of the grains of a finegrained recrystal metal will begin to grow rapidly at the expense of the other grains when heated at a higher temperature. Moderate cold reductions and low annealing of .5. obtained. 51 3. where the grains increase in size uniformly.
chap. York. C.Sec. New York. 1959. To obtain a nearly perfect recrystallization texture. C. 1959." 2d ed." American Society for Metals.: "Physical Metallurgy. Sometimes the formation of a strong recrj stallization texture is beneThe best example is cubeoriented siliconiron transformer sheet. A. B. where the grains are oriented in the easy direction of magnetization. New York. Inc. Chalmers. 1959." McGrawHill Book Company. Mass. Inc. 15. 1952... G. 513] Plastic Deformation oF Polycrystalline Aggregates 157 temperatures are beneficial. "Relation of Properties to Microstructure. S. ficial.: "Elements of Physical Metallurgy. Guy. Ohio. Metals Park. Reading..: "Structure of Metals. 1954. AddisonWesley Publishing Company. Birchenall.. lowed by enough added cold reduction to break up this orientation and produce a fine recrystallized grain size at a low temperature.. followed by long annealing at a high temperature to allow selective grain growth to produce a strong texture." 2d ed. it is necessary to produce This is a high degree of preferred orientation in the coldworked metal. A good way of minimizing a recrystallization texture is first to produce a strong preferred orientation by a heavy This is folinitial reduction and then use a high anneaUng temperature. New . Inc.: "Physical Metallurgy. BIBLIOGRAPHY Barrett.." John Wiley & Sons. McGrawHill Book Company. E.
by means of the FrankRead source is given particular 62. 4. 1934. Proc. Polanyi. E. in Chap. 362. p. intended to present a more complete and somewhat more The rapidly improving tech niques for detecting dislocations in real metals are considered. Soc. 660. Z. Interaction of dislocations with other dislocations. Orowan. A general picture has been given of the interaction of dislocations with foreign atoms. Physik. and Polan^d^ in 1934. 5. and foreign atoms is discussed in some detail. or hep crystal structures is considered. p. vacancies. and strain aging. to help explain solidsolution hardening. Methods of Detectins Dislocations The concept of the dislocation was proposed independently by Taylor. 158 . 1934. precipitate particles. This has been used to give a qualitative picture of the strain hardening of single crystals and. I. Roy. 1934. is was shown that the existence of a dislocationlike defect necessary to explain the low values of yield stress observed in real crystals. 614. introduced in Chap. vol. Orowan. 89. and other dislocations. Taylor. bcc. where the geometry of This concept was edge and screw dislocations It was presented for the case of a simple cubic lattice. but the idea lay relatively undeveloped until the end of World War II There followed a period of approximately 10 years in which the theory of dislocation behavior was developed 1 G.Chapter 6 DISLOCATION THEORY 61. The important problem of dislocation multiplication attention. 634. (London). 89. pp. and experi mental evidence to support the theory is given wherever possible in subsequent portions of the chapter. Z. yieldpoint behavior. dispersedphase hardening. Introduction A all dislocation is the linear lattice defect that is responsible for nearly aspects of the plastic deformation of metals. vol. The effect on dislocation behavior of considering real fee. USA. 605. M. This chapter is rigorous treatment of dislocation theory. Physik. vol.
W. p. 14. there certainly will be other changes in current concepts of dislocation theory. 1957. Fortunately. A. W. better experimental techniques. vol. ing dislocations in real materials. Seeing Dis locations. J.. real materials. Johnston. which has a very large lattice spacing (12 A) [J. New York. and D. See P. 62] Dislocation Theory 159 extensively and applied to practically every aspect of the plastic deformation of metals. 236A. J. ^J. The resolving power of the best electron microscope would have to be improved by a factor of 5 to 10 in order to observe directly the distortion of the individual lattice planes around a dislocation in a metal crystal. Interscience Publishers." Institution of Metallurgists. p. Inc.. G. (London). 345. ^ Several excellent reviews of experimental techniques have been published. of Oilman and W." John Wiley & Sons. "Dislocations and Mechanical Properties Crystals. vol. Bassett. Nutting. Because there were no really reliable methods for detectit was necessary to build much of this theory on the basis of indirect observations of dislocation behavior. . no. A great deal of information about dislocation behavior in the ionic crystal LiF has been obtained in this way by Oilman and Johnston. New York.'^ Important information about gories. Soc. Proc.Chemical methods include etchpit techniques and precipitation techniques. applicable to a wider variety of materials. 119.Sec. 1959. J. 1959. B. Reviews. be modified and some abandoned. and those changes at the site of a dislocation. These experimental techniques can be roughly classified into two cateutilize the strain field chemical reactions with the dislocation. Menter. 1958. Soc. 1956]. W. Pashley. pp. Roy. 101140. Hirsch. will be developed As more information is obtained on dislocation behavior in in the future. those involving utilizing the physical has been possible by means of an electron microscope to observe this lattice disan organic crystal of platinum phthalocyanine. 246A. See G. are formed at dislocation sites because the strain field surrounding the dislocation causes preferential chemical attack. Today. (London). in "The Structure of Metals. Methods based on the physical structure at a dislocation site include transmission electron microscopy of thin films and Xray diffraction techniques.^ Practically all the experimental techniques for detecting dislocations around a dislocation to increase its effective size. Proc. An indication of the lattice distortion at a dislocation in metals has been obtained by ' It tortion in making use of the magnification resulting from moire patterns produced by electron transmission through two thin overlapping crystals with slightly different orientations or lattice spacings. vol. Inc. no question as to the existence of lattice defects with properties to similar to those ascribed to the dislocation. Menter. The simplest chemical technique is the use of an etchant which forms Etch pits a pit at the point where a dislocation intersects the surface. while others have had Undoubtedly.. 4. Met. Roy. since 1955 improved techniques have made it possible to observe dislocations as they actually exist there is in many materials. Many of the theoretical predictions have been confirmed by experiment.
. 61 . etchpit formation at dislocations appears to be dependent on purity. l'*"*Vt a ^. 75... 740. F. care must be taken to ensure that pits are formed only at dislocation sites and that all dislocations intersecting the surface are revealed. 1959. pp. and J.:. pp. L.) Fig. 2 A summary . Figure 54 shows an etchpit structure in an ironsilicon alloy which was made visible by diffusion of carbon atoms to the dislocations.. Wernick.^ Because of solute segregation to the dislocation. 1 J.000 X. •> . 5. H. Wilsdorf. G. Figure 61 shows the excellent resolution obtainable from etchpit studies Pits only 500 A apart have been resolved. vol. {J D. 218. ' * "' "* Etch pits on slip bands in alpha brass crystals.( . Lowell. form the precipitate after suitable heat treatment. Trans.'V'. »/j. Meakin and around the dislocation becomes anodic to the surrounding metal. the region ^J>. Etchpit techniques are useful because they can be used with bulk samples. vol. '*•}. H. D. and consequently preferential etching occurs at the dislocation. of etchpit techniques in metals is given by L. F. . In the on alpha brass. Trans.. F. 218..* i •*. Vogel. Wilsdorf. AIME. 6 dislocations in metals has also been obtained with etchpit techniques. . p. However. 737745. Metal Prog. In metals.. Meakin and H. 9696D.. 1960. M^ . A similar method of detecting dislocations is to form a visible precipitate along the dislocation to Usually a small amount of impurity is added The procedure is often called "decoration" of dislocations. C.. AIME. 1960.160 Metallurgical Fundamentals [Chap. : '*» ir. vol.. ^ region of heavy slip shown in this electron micrograph the dislocation density is 10^" cm^^. This technique was first used lines. G. i' f*J 1 .
By means of this technique has been possible to observe dislocation networks (Fig. 1957. stacking faults. 1. dislocation pileup at grain boundaries (Fig.. W. 62] Dislocation Theory 161 in AgBr with photolytic has since been used with many other ionic crystals. p.^ such as AgCl. Hexagonal network of dislocations in XaCl detected by a decoration technique. 63). some '^ X^''^^ Fig 62. and CaF2.. p. 1953. New York. NaCl. vol. W. 1956. Mag. 51). J. and M. (S. 1588. the dislocation lines. Amelinckx. Home. 5 P. J. less than 1 thick. Hirsch. is thinned after deformation by electropolishing to a thickness of about 1. in "Dislocations and Mechanical Properties of Crystals. 1957.. p.* Thin sheet. With these optically transparent crystals this technique has the advantage that it shows the internal structure of by Hedges and MitchelP to decorate dislocations It silver. Inc. 103. B.) work has been done along these lines with the AlCu precipitationhardening system and with silicon crystals. The most powerful method available today for the detection of dislocations in metals is transmission electron microscopy of thin foils.. Although dislocation decoration has not been used extensively with metals. W. Phil. Mag. vol. vol. 677. S." John Wiley & Sons. 223. 44. crystal lattice cannot be resolved. KCl. "Dislocations and Mechanical Properties of Crystals." John Wiley Sons. Inc. Hedges and J. and many other structural features 1 of dislocation theory. Amelinckx. 1956. R.000 A. . New York. Dislocation ^ & M. Phys. Phil. Rev. Whelan.Sec.. Mitchell. BoUmann. At this thickness the specimen Although the is transparent to electrons in the electron microscope. CottrellLomer barriers. individual dislocation lines can be mm observed because the intensity of the diffracted electron beam is altered it by the strain field of the dislocation. tions in a Figure 62 shows a hexagonal network of disloca NaCl crystal which was made visible by decoration.
Whelan. by Xray diffraction microradiographic techniques. with the resolution at present available it is limited to crystals of low dislocation density (approximately lO^cm^). Hirsch. Phys.162 Metallurgical Fundamentals [Chap. In general. a dislocation will be partly edge and partly screw in character. Burgers Vector and the Dislocation Loop The Burgers vector b direction of location. vol. 32. 6 foil movement has been observed by generating thermal stresses in the thin with the electron beam. 30. p.) advantage of being nondestructive and giving information on a bulk sample. the most characteristic feature of a dis has already been shown that for a pure edge dislocation the Burgers vector is perpendicular to the dislocation line. . Phil. 677. /. (P. 62 and 63. Dislocation network in coldworked aluminum. dislocations will A. 8.. The method has the Fig. 63. 63. 1956. However.. dislocations in real crystals are rarely straight lines lie and rarely in a single plane. Appl. vol. while for a pure screw dislocation the Burgers vector is parallel to the dislocation line. W. R. and M. B. It Actually. It is expected that much more information will be gained with this method as techniques for preparing and deforming The dislocation structure of a crystal can be detected thin foils are improved. J. 1. is the vector which defines the magnitude and it is Therefore.^ The strain field at the dis location results in a different diffracted intensity. slip. Lang.500 X. ser. R. pp. Home. 1 As shown by Figs. 1959. Mag. 17481755.
Burgers vector equal to one lattice Because of energy .spacing Therefore. 64. In general. Dislocation loop lying in a slip plane (schematic). if the path encloses a dislocation. Fig. 63] Dislocation Theory 163 ordinarily take the form of curves or loops. node can be considered as two dislocations with Burgers vectors bi and b2 combining to produce a resultant dislocation bs. A convenient way of defining the Burgers vector of a dislocation is by means of the Burgers circuit. around several dislocations vectors. topographic considerations demand that it either must be a closed loop or else must end at the free surface of the crystal. A is at a node. dislocation loop is pure screw at point A and pure edge at point B. The closure failure of a Burgers circuit If the region enclosed the Burgers circuit will close. is equal to the sum of their separate Burgers Because a dislocation represents the boundary between the slipped and unslipped region of a crystal. that the Burgers vector is the same along the entire dislocation If this were not so. 64. which in three dimensions form an interlocking dislocation network.Sec. Note. which shows the atomic arrangement around an edge dislocation. however. while along most of its length it has mixed edge and screw components. Since the periodic force field of the crystal lattice requires that atoms must move from one equilibrium position to another. 48. the crystal structure will determine the possible Burgers vectors. in Fig. The closure failure of the Burgers circuit is the Burgers vector b. Consider Fig. a dislocation line cannot end inside of a crystal. . would have to slip by a different amount the relative to another part of Slip plane Burgers vector would mean that another dislocation line would crystal and this run across the slipped region. imagine a path traced from atom to atom. it follows that the Burgers vector must always connect one equilibrium lattice position with The exception another. part of the crystal above the slipped region loop. is A dislocation with a said to be a dislocation of unit strength. the resolved into edge and screw components. However. In considering a dislocation loop in a slip plane any small segment of the dislocation line can be For example. where three or four dislocation lines meet. by the path does not contain a dislocation. an equal distance in each direction. The vector bs is given by the vector sum of bi and b2. always in the direction of one of the vectors of the unit cell. Starting at a lattice point. the Burgers circuit will not close.
below the allel slip plane within the dislocation loop. A A dislocation of unit strength. Therefore. The shortest lattice vector is (ao/2) [110]. p. consideration of the atomic arrangement on the {111} slip Figure 65 represents plane shows that slip will not take place so simply. Physica. where a translation of one Burgers vector does not produced by the For a stacking fault to be stable. A unit dislocation parit to the slip direction cannot dissociate further unless becomes an imperfect dislocation. the Burgers vector for axes of the crystallographic structure cell. which connects an atom at a cube corner with a neighboring atom at the center of a cube face. It has already been the atomic packing on a closepacked (111) plane. For example. The Burgers vector is therefore (ao/2) [110]. The criterion for deciding whether or not dissociation will occur is based on the fact that the strain energy of a dislocation is proportional to the ^ square ^ of its Burgers vector. C. has a minimum energy when in Burgers vector is parallel to a direction of closest atomic packing This agrees with the experimental observation that A unit discrystals almost always slip in the closepacked directions. 15. ABC ABC sequence shown that the {111} planes are stacked in a • • . 131. region. ao/2. 0]. as The strength of a dislocation generally written. vol. location of this type is also said to be a perfect dislocation because translaFor a tion equal to one Burgers vector produces an identity translation. lattices cell. ao/2. but not if 61" < 62^ bs will occur when br^ > hi^ bi— Dislocations with strengths less than unity are possible in closepacked b2 + + + where the equilibrium positions are not the edges of the structure Burgers vector is specified by giving its components along the Thus. The Burgers vector is [ao/2. the decrease in energy due to dissociation must be greater than the increase in interfacial energy of the faulted result in an identity translation. perfect dislocation there is perfect alignment of atom planes above and its the lattice. b = (ao/2) [110]. Dislocations in the Facecentered Cubic Lattice Shp occurs in the fee lattice on the {111} plane in the (110) direction. 6 considerations dislocations with strengths larger than unity are generally unstable and dissociate into two or more dislocations of lower strength. Frank. . slip in a cubic lattice from a cube corner to the center of one face has the components an/2. or unit dislocation. The vector 1 bi = (ao/2) [lOl] defines one of the observed slip directions. 1949. with Burgers vector aaluvw] is \h\ — ao[w^ + w^ + w'^Y'. 0. the dissociation reaction 63^. A stacking fault is dissociation of a unit dislocation into two imperfect dislocations. or. However. 64. F. the magnitude of the Burgers vector given above is \h\ = ao/\/2. 63^.164 Metallurgical Fundamentals [Chap.
1948. As Fig. Slip in a closepacked (111) plane in an fee lattice. As was discussed in Sec. 67. z components righthand side of the equation must add up to the x. AIME. 37. F. Trans. London. J. 65. the stackingfault energy Thompson and W. The combination of the two partials AC and is known as an extended dis ones which do not produce Figure 66 represents the situation looking AD location. The above strain reaction is energetically Fig. the atoms are considered as hard spheres. C. X component y z M ¥2 = %+M 76 component component — 2. p. The partial dislocations will fault 1 an equilibrium separation determined primarily by the stackingenergy.)ec. 410. This dissociates according to the above reaction into partial dislocationswith Burgers vectors b2 and bs. Slip New York. 64] if Dislocation Theory 165 an However. AB represents the perfect dislocation line having the full slip vector bi. y. R." p.f[2TT) + '112] To check this reaction. since the dislocations are imperfect complete lattice translations. However. Oxford University energy proportional change ao~/2 —> air/3. Mathewson. the summaof the tion of the X. 109. 1944. 1953. 2 H. Iron Steel Inst. 66 shows. This dislocation reaction was suggested by Heidenreich and Shockley. D. z components of the original dislocation. Press. {After A. Because b2 and bs are at a 60° angle. Heidenreich and W. The region between them is a stacking fault representing a part of the crystal which has undergone slip intermediate between full slip and no slip. Millington.) by this twostage process creates a stacking fault ABC A'C ABC in Burgers the stacking sequence. in Crystals. the dislocation with vector bi has been dissociated into two partial dislocations bo and bs. . W. Physical Society. vol. E. and therefore this dislocation arrangement is often known as Shockley partials. down on (111) along [111]. favorable since there CottreU. (London). vol. there will be a repulsive force between them (Sec. p. 69). C." p. ^ it is easier for atom on a type B plane to move along a zigzag path b2 + bg in the valleys instead of moving over the hump that lies in the path of the vector 61. 1924. The dislocation reaction is given by bi—> b2 + bs "[10Tj. y. 32. the surface tension of the stacking fault tends to pull settle at them together. "Report on Strength of Solids. Shockley. is a decrease in to the "Dislocations and Plastic Flow 73. 38. H.
The nature of the dislocation network in fee in difficult in ^ metals changes with the stackingfault energy.166 Metallurgical Fundamentals [Chap. the more difficult it is to produce constrictions in the stacking faults. with a stackingfault energy around 13 ergs/cm^. 66. cit. Figure 426 shows that this requires the formation of a constriction in the stacking The greater the width of the stacking fault. to two shp planes. Austenitic stainless steel. These ideas are borne out by electronmicroscope transmission studies of dislocation networks in thin foils. Gold. This is the process However. shows dislocation net works only along slip planes. andSOergs/cm^. . which have no fixed glide plane. which has a wide stackingfault ribbon. where the energy is about 30. respectively. may surmount obstacles by gliding onto another slip plane having a common slip direction. must first recombine into perfect dislocations since an extended disloca tion cannot glide on any plane except the plane of the fault. while it is copper. or the lower the stacking fault energy. which has a very narrow stackingfault ribbon. Stacking faults can be readily detected in these thin films. aluminum. 6 can vary considerably for different fee metals and alloys and this in turn can have an important influence on their deformation behavior. even for large deformations. the extended dislocations of cross slip. op. 40. This may explain why cross slip is quite prevalent fault ribbon. in order to do this. and nickel. Dissociation of a dislocation into two partial dislocations. A characteristic of the fee lattice is that any Burgers vector is common This presents the possibility that screw dislocations. ^ Hirsch. copper. Extended dislocation C [121] foi] to = 3 = ^[2Tl] Fully slipped No slip >^1=^[10T] Fig.
. E. This is called a Frank partial dislocation. vol. ° i here! ore. 897.. 64] Dislocation Theory 167 show the low dislocations arranged in complex threedimensional networks at deformations. Mag. p. (London). and screw dislocations can easily arrange themselves into a network of lowangle boundaries. and at high strains they try to form lowangle boundary networks to lower their strain energy. are called glissile. P. A method by which a missing row of atoms can be created in the (111) Eviplane is by the condensation of a disk of vacancies on that plane. A Frank partial dislocation or {After A. 1953. Sec. shows almost perfect subboundaries. . the Frank partial dislocation cannot move by glide. cross J* sessile dislocation. R. i. This picture of a graded transition in the way the dislocations are arranged is in agreement with the intiuence strains.. Since glide must be restricted to the plane of the stacking fault and the Burgers vector is normal to For this this plane. obtained has been dence for the collapse of disks of vacancies in aluminum by transmission electron microscopy. with a stackingfault energy of 200 ergs/cm. Frank' pointed out that another type of partial dislocation can exist in Figure 67 illustrates a set of (111) planes viewed from the edge. H. by the diffusion of atoms or vacancies to or from the Because climb is fault. copper. Silcox. . H. Cross slip is energy on the undergo cross very difficult in strains. New . Phys. vol. York. 62A. 3. The center part of the middle A plane is missing.^ 75^ Oxford University Press. cross slip rj^. slip ^ oi SCrew dislocations occurs. process of climb.i vi jpi Crysand rtastic blow in n Dislocations ^^/^ . B. Soc. reason it is called a sessile dislocation. stainless steel. but probably only at highly stressed nickel. Its Burgers vector is perpendicular to the central stacking fault. cross slip is very prevalent. and K. Sessile dislocations are produced in the fee lattice by the glide of dis locations on intersecting (111) 1 ' planes. This changes into poorly developed subboundaries at higher Aluminum. Phil. 202. Hirsch. even at high so that the dislocations are confined to the slip planes. p. Smallman. Dislocations which glide freely over the slip plane. These sessile dislocations are F. An edge dislocation is formed in this region with a Burgers vector (ao/3)[lll]. Frank. sessile dislocations provide obstacles to the movement of other dislocations. r. Westmacott. n ' 1 regions. 67.) In aluminum. by the not a likely process at ordinary temperatures. such as perfect dislocations or Shockley partials. '"'s and is possible. A sessile dislocation can move only the fee lattice. of the stackingfault ability of a metal to slip.e. Proc. C. 1958. 1949. Cottrell. In gold. J.
vol. 2 A. p. These dislocations attract each other and move toward the intersection point 0. CottrelP showed that the product of Lomer's reaction could be truly immobile made by the following dislocation reaction I [110] The products ^ ^ [112] + ^ [112] + ^ [110] edge dislocations of this dislocation reaction are imperfect which form the boundaries of stacking faults. M. Since Fig. three dislocations must be parallel 68. [lIO]. "Dislocations and Plastic Flow in Crystals. B glides in a (111) plane with a Burgers vector (ao/2)[011]. the dislocation (ao/6)[112] glides in the (111) plane and forms a stacking fault bounded and the line of the dislocation. p. It is repelled from the line and forms a stacking fault bounded by two [110] lines. The third dislocation with by the line where the two stacking Burgers vector (ao/6)[110] hes along the line » W. 42. Cottrell. Phil. Dislocation reaction to the line of intersection of the slip plane.. Mag. Since (001) is not a common slip plane in the fee lattice.: 168 Metallurgical Fundamentals [Ch ap. it is not a true sessile dislocaFrank partial. tion. {After A.n imperfect dislocation.) Press. Dislocation Disloca (111) plane with a Burgers vector (ao/2)[101]. New vector [110] and the line [lIO]. vol. 1951." Oxford York. 1327. which is the intersection of the two Burgers vectors along the direction [110]. 43. The plane (001) contains both the Burgers leading to CottrellLomer barriers. the edge dislocation formed by Lomer's reaction has a slip plane (001). p. Cottrell. At this point the // two dislocations react according to Lomer's reaction f [101]+ I [Oil] to ao [110] form a new dislocation all of reduced energy. in the sense of the However. known as CottrellLomer harriers. moving on intersecting slip planes will attract Lomer^ pointed out that and combine their Burgers vectors two dislocations A is moving in a tion moving on the have suitable orientations. H. 1952. H. . Lomer. Phil. Figure 68 illustrates slip planes of an fee lattice.. University 1953. the dislocation formed from Lomer's reac tion should not glide freely. because it is not d. of the strain They are an important element in the mechanism dislocations if hardening of metals. Mag. the line and the line of the dislocation. In a similar way. The dislocation (ao/6)[112] is a Shockley partial which glides in the (111) plane. 171. 645.
1957. CottrellLomer locking provides an effective barrier to slip. Diehl. . Roy. although in other bcc metals predominantly on the {110} planes. 1956. It will be recalled that shp lines in iron have been found to occur on {110}. vol. Dislocations in the Bodycentered Cubic Lattice in the (111) direction in the bcc lattice. J. Dislocations in the basal plane can reduce their energy by dissociating into Shockley partials according ao[1120]^ ao[10lO] + ao[OlTO] The stacking fault produced by this reaction lies in the basal plane. Studies by transmission electron microscopy of dislocation interaction in thin foils have confirmed the existence of interaction that is in agreement with the model of Cottrell Lomer locking.^ temperatures.Sec. Whelan. the closepacked (1120) direction. by cross slip before the stress is high enough 65. the Burgers vector is (ao/2)[lll].. the Burgers vector to the reaction ao[1120]. ser. Phil. lattice vector Shp occurs {112}. 1961. 700706. p. Mader. 43). S. and R. Phil. A. 1958. Mag. vol. N. Slip occurs on the basal plane (0001) in the <1120> direction (Fig. 1 CottrellLomer barriers can be overcome at high stresses and/or A mathematical analysis of the stress required to break down a barrier either by slip on the (001) plane or by dissociation into the dislocations from which it was formed has been given by Stroh. Hirth. Dislocations in the Hexagonal Closepacked Lattice is The basal plane of the hep lattice a closepacked plane with the stacking sequence ABABAB ao is • • • .. However. Stroh. Therefore. The shortest extends from an atom corner to the atom at the center of the unit cube. 8. 66. p. 1 slip appears to occur M. 2. and {123}. Seeger. all possible dis location reactions in the fee lattice have been worked out Phys. Rebstock. J. 2 A. 3 by J. Proc. it tions piled has been shown'' that for the important case of screw dislocaup at CottrellLomer barriers the screw dislocations can generally escape the pileup to collapse the barrier. (London). Therefore. 1. 114. 489. pp. 249A. 66] Dislocation Theory 169 faults join. Appl. This combination of three dislocations produced by the CottrellLomer reaction forms an isosceles triangle which is locked rigidly in place and cannot glide. Mag. J. vol. P. p.. and the extended dislocation which forms it is confined to glide in this plane. The smallest unit vector for the hep structure has a length and lies in Therefore. 32. vol. Soc. 323.
New p. and because of the geometry of the situation. Consider Fig. 212. This dislocation reaction may be important to the brittle fracture of bcc metals. Cottrell. Soc. Dislocation A. Met. they are not completely confined to the (112) are slip plane. 1958. is gliding in the Dislocation B. vector perpendicular to the It is also its Burgers an imperfect sessile dislocation that forms the The dislocation (ao/6)[lll] the Shockley partial of boundary of a stacking fault in the (112) planes. The two (ao/2)[lll]. "Dislocations Flow in Crystals. A dislocation in the (112) plane energy by dissociating according to the reaction may also lower its f [lll]>°[lll]+f [111] As discussed above. Acta Met. which appears to lead to the formation of immobile dislocations in the bcc lattice. with Burgers vector (ao/2)[lll]. J. 7. Plastic A. M. Silcock. However. 2 3 York. with Burgers vector (101) plane. 192. While the existence of stacking faults has been demonstrated by Xray diffraction. vol. detailed studies of the dislocation reactions discussed in this paragraph have not yet been made. An analysis^ of the atomic positions giving rise to stacking faults on {112} planes shows that there two types which may result. is gliding in the intersecting slip plane (101). CottrelP has suggested another dislocation reaction. p. A. . AIMS. 6 Dislocation reactions have not been studied so extensively in the bcc lattice as in the fee lattice. both the partial dislocations formed by this reaction are pure screw. 1959. 359. 1 of this reaction is a Since this is not a and pure edge dislocation which lies on the common slip plane in the bcc lattice. H. Trans. this dislocation can line of intersection of three planes of the glide out of the plane of the stacking fault too easily to be part of a true extended dislocation. H. vol." Oxford University Press. because [111] is the type {112}. Cottrell^ has suggested that a perfect dis location in a (112) plane can dissociate according to the reaction f [111] ^ 1° [112] + The dislocation (ao/3)[112] lies is 1° [111] a pure edge dislocation since slip plane. 1953.. 69a. Cottrell. dislocations come together and react to lower their strain energy by the in the bcc lattice reaction f [ni]f[lll]^ao[001] The product (001) plane. is an imperfect glissile dislocation similar to the fee lattice.170 Metallurgical Fundamentals [Chap.
67] Dislocation Theory is 171 the dislocation immobile. Inc. H. vol. Read. This crack can then grow by additional dislocations gliding While this particular dislocation reaction has not Fi9. 1953. 695). AIME. Jr.Sec. However. However. 1958. Read. (.. For the case of a perfect dislocation a good approximation of the stress field can be obtained from the mathematical theory of elasticity for continuous media. Nabarro. Stress Field of a Dislocation produces between dislocations and solute atoms. Trans. pp. The stress around a straight dislocation will be a good approximation to that around a curved dislocation at distances that are small 1 compared with the radius of curvature. the equations obtained dislocation is A surrounded by an elastic stress field that forces on other dislocations and results in interaction are not valid close to the core of the dislocation line. 67. T. cylinder without a dislocation 1 is shown by the dashed line. N. ate in cubic ionic crystals such as LiF it has been found to oper and MgO. 212. the (001) plane is the cleavage Cottrell suggests that the plane along which brittle fracture occurs. no. vol. D." pp.) on intersecting (110) planes. 1953. Eshelby. 69. R. New York. 271395. and W. 3.^ Appreciably greater complexity crystal with anisotropic elastic Figure 610 represents the cross section of a cylindrical piece of elastic parallel material containing an edge dislocation running through point undistorted original The to the z axis (normal to the plane of the figure).. W. Acta Met. W. 1.. 351359. 114123. pp. McGrawHill Book Company. Shockley. "^ . The equations given below apply to straight edge and screw dislocations in an isotropic crystal. 1. formation of a dislocation on the cleavage plane by slip on intersecting {110} planes is equivalent to introducing a crack one lattice spacing thick (Fig. over the {110} planes. Advances in Phys. 1952. Slip p. The dis For derivations see F. results from the consideration of a constants. J.4. been established by experiment in bcc metals. T. "Dislocations in Crystals.. vol. 196. Cottrell.
Since the dislocation line parallel to the z axis. H. Chaps. dislocation in an elastically isotropic the stresses. sliding the cut surfaces along each other the distance again. A A'. = (65) For polar coordinates. New York. 1934. are given The notation is the same as that used in by the following equations. "A Treatise on the Mathematical Theory of Elasticity. while ae acts in a plane perpendicular to stresses r.172 Metallurgical Fundamentals [Chap. and joining them back together This sequence of operations^ produces a positive edge dislocation running along the z axis with a strain field identical with that around a dislocation model such as that of Fig.. 1 and 2. is Fig. Deformation of a circle con that direction are zero and the prob taining an edge dislocation. Love. 6 location was produced by making ?/ a radial cut along the plane (line = OA). Cambridge University Press. 48. lem can be treated as one strain. . The unline. the equations are CTr — — (Te — —Toh r sin Q (66) TrB TBt — h cos d To (67) ar acts in the radial direction. long before the concept of dislocations was originated. by(3xTo + ij^) (61) by(x'  y^) To (62) (63) where To = 27r(l G  v) Txy Txz — bx{x^ To — y'^) (64) (x^ \ y'^'Y Ty. E." pp. in plane strained circle is shown by a dashed The solid line represents the circle after For the case material of a straight edge in the dislocation has been introduced. . terms of three orthogonal coordinate axes. Note that the vary inversely with distance from the dislocation 1 It is interesting that this problem was analyzed by Volterra in 1907. 221^ 228. The mathematical details may be found in A. strains in 610.
1956. 1 U = n\ 2 Jro /''' Trebdr = 2 Jro 0. The total energy of a crystal containing 1 W. 1211. Dislocation Theory 173 Since the stress becomes infinite at r = 0.. the strain energy per unit length of dislocation is proportional to Gh'^. Phys. 101. L. In ^ ro is (612) v) In the same way.. vol.^" 47r(l  . cylindrical Since there is ^ of a. the strain energy of a screw dislocation given by (613) U= ^ 2 \ Te^h dr = Jro ^\n ' ro 47r Note that. 1 /"'' rob" cos dr ^— r (611) But cos 6=1 along the slip plane y = so that the strain energy is given by U= . a small cylindrical region r analysis. 69). p.Sec. Andrus. This strain energy corresponds to about 10 ev for each atom plane threaded by an edge dislocation (Prob. = ro around the dislocation line must be excluded from the A straight screw dislocation in an isotropic medium has complete symmetry. The screw dislocation. The strain energy involved in the formation of an edge dislocation can be estimated from the work involved in displacing the cut OA in in tions for a stress field Fig. = ^^ in a silicon crystal (610) The strain field around an edge dislocation has been observed' by means of polarized infrared radiation. 67] line. For a rectangularcoordinate system only two components of stress are not equal to zero. 610 a distance h along the slip plane. r. intensity is The variation in agreement with what would be expected from the equaaround an edge dislocation in an isotropic medium. . in accordance with our assumption up to this point. . Bond and J. no extra half plane atoms in a are no tensile or compressive normal stresses. there stress field is simply is one of shear. Rev. The is radial symmetry of this stress field apparent when the shear stress expressed in a polarcoordinate system.
the dislocations move and produce is a force acting on a dislocation line which tends to drive it slip. the force is per unit length acting on the dislocation line F = This force is Th line at (614) normal to the dislocation every point along its length and is directed toward the unslipped part of the glide plane. The area swept out by the line element is then ds dl. f 6 many dislocation lines is the si^t^ the strain energies of the individiml dislocations. where A is the area of the slip plane. Thus. . Force acting on a dis location line. 68. line moving Figure 611 shows a dislocation in the direction of its Burgers t.174 Metallurgical Fundamentals [Chap. it is convenient to consider that a dislocation possesses a line tension which attempts to minimize its energy by shortening its length. the line tension produces a restoring force which tends to straighten it out. Because the strain energy of a dislocation line is proportional to its work must be performed to increase its length. The direction of this force is perpendicular to the dislocation line and toward the center of curvature. For a curved dislocation line. plus a term describing the internal stresses developed by the external forces. The line tension has the units of energy per unit length and is analogous to the surface tension of a liquid. plane of an amount ds dlb/A. The equilibrium condition length. there forward. The magnitude of this force is T/R. a dislocation line can have an equilibrium curvature only if it is acted on by a shear stress. Forces on Dislocations When an external force of sufficient magnitude is applied to a crystal. ing on the element ds in the direction of This corresponds to a force dW/dl actits normal. This corresponds to an average displacement of the crystal above the slip plane to the crystal below the slip 611. where r is the line tension and R is the radius of curvature of the bent dislocation line. vector under the influence of a uniform shear stress tion line ds An element of the disloca is displaced in the direction of slip normal to ds by an amount dl. Therefore. The work done by the shear plane is stress acting in the slip dW = TA{ds dlh)/A. plus terms expressing the interactions of the stress fields of the dislocations. Fig. Therefore. Because of line tension.
610. Referring to Fig.000 A and ro = 2 A are used." pp. vol. This is obtained from Eq.5Gb^. New York. American Institute of Mining A. N. The Consider ^ and Metallurgical Engineers. Proc.h = ^ same (616) The force is attractive for dislocations of opposite sign for dislocations of the (antiparallel screws) the and repulsive sign (parallel screws). (613) provides a good approxilocation line. 2 E. Orovvan. 67B.Sec. 69. the edge dislocations are at P and Q. Forces between Dislocations same slip plane will attract each and annihilate each other. Stroh. run together. Soc. the shear stress needed to maintain a dislocation line in a radius of curvature R is r = i^ (615) Orowan^ has pointed out that the determination of this stress bears an analogy with the problem of blowing a bubble from a nozzle submerged The line tension will vary from point to point along a disin a liquid. Stroh^ has shown that Eq. 427. An approximation often used is F ~ 0. p. 69] Dislocation Theory is 175 for this to occur Therefore. The simplest situation to consider is the force between two parallel screw dislocations. . now the forces between two parallel edge dislocations with same Burgers vectors. The force between them is not a central force. (London). 48). This can be seen readily for the case of an edge dislocation (Fig. 99102. with their Burgers vectors along the X axis. Fr = re. Since the stress field of is a screw dislocation is radially symmetrical. dislocations of like sign on the same slip plane will Dislocations of opposite sign on the other. where the superposition of a positive and negative dislocation on the same slip plane would eliminate the extra plane of atoms and therefore the dislocation would disappear. Phys. the force between them a central force which depends only on the distance that they are apart. 1953. mation of the line tension. and so it is necessary to consider both a radial and a tangential component. (613) when typical values ri = 1. parallel to the z axis. repel each other. 1954. Conversely. "Dislocations in Metals.
{After "Dislocations and Plastic Flow in Crystals. which e is the slip direction. Graphical representation of Eq.176 Metallurgical Fundamentals is [Chap. locations of same A. Solid curve A is for two edge disDashed curve B is for two unlike edge dislocations. 1953. Press. is of most F:c = _ ~ Fr cos 27r(l  Fe sin 6 Gh^x(x^ — y'^)  p)(x^^ + yT (618) a plot of the variation of F^ with distance x. 6 force per unit length given by^ Gb'1 Gh' 27r(l sin 20 (617) v) 27r(l  v)r  r slip plane. 46. 612. New York. "Dislocations and Plastic Flow in Crystals. when x > y The reverse (6 is < 45°) and attract each true for dislocations of A. H. 48. 1953. direction." p. expressed in units of y. (618).) component along the x interest. Oxford University sign." p. Cottrell. Cottrell. H.3 Fig. 0. New York. where x is Curve A is for dislocations of the same sign. Because edge dislocations are mainly confined to the the force. Oxford Uni versity Press. curve B is for dislocations of opposite sign. . Note that dislocations of Figure 612 is the same sign repel each other other 1 when x < y {d > 45°).
The situation of two parallel dislocations tors can be rationalized by considering their relative energies. Fx is zero at x = where the edge dislocations lie vertically above one another.Sec.f^' 47r(l — . 1941. However. . Koehler^ has shown that this force is approximately equal to the force which would be exerted in an infinite solid between the dislocation and one of opposite sign located at the position of its image on the other side of the surface. theory predicts that a vertical array of dition of equilibrium. the positive direction of climb are taken the direction in which atoms layer. so that it is The motion and therefore of a screw dislocation always involves not involved with climb.  (619) v) r an edge dislocation.. glide. J. Phys. edge dislocations of the same sign is in stable equilibrium. They will repel if it is less than 90°. op. the dislocations will attract if the angle between their Burgers vectors is greater than 90°. 610. it is is Climb requires mass trans port by diffusion. represents the situation of dislocations on two intersecting In general there will be no stable position. as for the previous case. This image force is equal to F = for . opposite sign. 610] Dislocation Theory 177 and x = y. = 0. it should be noted that metal surfaces are often covered with thin oxide films. This is the arrangement of dislocations that exists in a lowangle grain boundary of the tilt variety.62^ and will repel if 63^ > bi^ + 62^ Expressed another way. 2 Koehler. S. By convention. 131. p. a thermally activated process. A dislocation approaching a surface with a coating of an elastically harder material will encounter a repulsive rather than an attractive image force. which may or may not attract and combine into bs. away from p. The situation a. . To move an edge dislocation in a direction perpendicular to the slip plane requires the tion line An process of climb. 397. Dislocation Climb edge dislocation can glide only in the plane containing the dislocaand its Burgers vector (the slip direction). dislocations either will try to come together or will move far apart.' with different Burgers vecThis slip planes. A free surface exerts a force of attraction on a dislocation.. Rev. cit. vol. The two dislocations will attract if 63^ < 61^ \. the extra half plane of atoms in an edge dislocation so that this extra half plane 1 moves up one atomic The usual way Read. 60. The Con sider two parallel dislocations bi and b2. since escape from the crystal at the surface would reduce its strain energy. is a conThus.
. 61 1 . it creates a step. possible. j. or jog. lying on plane Pad.178 Metallurgical Fundamentals is [Chap. The intersection of two edge dislocations is illustrated in Fig. 614.^ edge dislocations on slip planes that produces lowangle grain boundaries by the process (a) of polygonization.. It is also but not energetically favorable. Dislocation climb is also a very important factor in the creep of metals. the extra half plane and become an interstitial atom. The intersection produces a jog PP' in dislocation AD. 613). as was shown earlier.. in Fig. onstrated the existence of this phe nomenon. However. Jogs can be produced by the There plane. . the formation of jogs in edge dislocations will not impede their motion. Since selfdif fusion occurs by the movement of vacancies.. 429. 61 3. . When intersection of dislocations. or. this implies that dislocation climb is involved in creep. . (o) dislocation climbs up one lattice spacing. which creates vacancies. in the dislocation line. An edge dislocation with a Burgers vector bi is moving on plane Pxy. and therefore it can readily glide with the rest of the dislocation... \. . . i^ i JtLitchpit techniques . edge dislocation. with Burgers vector h. The resulting jog is parallel to bi..f . atoms must be added to the extra half plane of atoms. j on u bent and annealed crystals have amply demt?. Hence. where the activation energy for steadystate creep is equal to the activation energy for selfdiffusion in pure metals. Diffusion of vacancy to . less probably. but it has a Burgers vector b since it is part of the dislocation line APP'D. or a jog can be produced during climb owing to the failure of climb to occur along the entire length of the extra half plane of atoms. J. . it requires energy to cut a dislocation because the formation XY . Jogs is in Dislocations no requirement that a dislocation must be confined to a single a dislocation moves from one slip plane to another. It can be seen that the jog resulting from the intersection of two edge dislocations has an edge orientation. t ^ . • • • • • • V^ ' • • • • • « Dislocation climb necessary to bring about the vertical alignment of (^) Fig. This can occur by atoms from the surrounding lattice joining the extra half plane. for the atom to break loose from To produce negative climb. 6 diffusing to the dislocation for this to occur by a vacancy and the extra atom moving into the vacant lattice site (Fig. by an interstitial atom is diffusing to the dislocation. It cuts through dislocation AD. The length of the jog will be equal to the length of the Burgers vector bi.
612] Dislocation Theory 179 .Sec.
New York. and "Symposium on Vacancies and Other Point Defects in Metals and Alloys. 509512. 1. Mott." pp. Cottrell. motion of a screw dislocation containing jogs in a direction normal to its axis can occur only by climb. pp." Institute of Metals. John Wiley & Sons. 43. Mag. 1957. 1165. Friedel. it is a favorable center for the absorption and annihilation of vacancies." GauthierVillars & Cie. Cottrell^ has shown that the jogs formed by intersecting screw dislocations will generally produce interstitials. There For reviews of this subject see T. 1958." pp. and Cottrell. "Dislocations and Mechanical Properties of Crystals. Advances in Phys. Phil. p. H. 1955." pp. 1956. not vacancies. 1958. 1952. "Vacancies and Other Point Defects in Metals and Alloys. it generates vacancies. vol. Advances in Phys. 1954. N. 5 J. "Les Dislocations. Inc. 3 J. 6 mental evidence' based on deformation at low temperature (so as to suppress the mobility of vacancies) followed by the measurement of electrical resistivity and mechanical strength before and after annealing treat found that about half the increased resistivity due to cold welldefined temperature ranges and with activation energies which generally agree with the temperatures and activation energies observed for the annealing of quenched and irradiated Moreover. Other mechanisms for the generation of vacancies by jogs on dislocations have been proposed by Friedel. Friedel. Institute of Metals. annealing experiments show that vacancies rather than interstitials are the predominant point defect in coldworked metals. It is also generally considered that vacancies can be generated at jogs. As was pointed out in the previous section. It is work anneals out over responsible for the resistivity changes. 2 F.180 Metallurgical Fundamentals is [Chap. p. However. 3.. two points of doubt have been raised about this mechanism. . London. Mott. ^ A.. Jogs in dislocation lines can act as sources and sinks for point defects. New York. vol. 46. As the jog climbs. 2683. vol. Paris. Because of the reentrant corner at a jog.. A. indicating that dislocations are not ments. Friedel^ has pointed out that there is no reason why a jog should not glide along a screw dislocation without producing vacancies so long as it can shortly attach itself to an edge component of the dislocation line. However. The generation of point defects due to deformation has been demonstrated in ionic crystals by measurements of conductivity and density and by the observation of color centers. Broom. there is little question that jog formation due to the intersection of dislocations is involved. the changes in resistivity are accomplished with samples. Inc. Seitz.. The usual mechanism^ involves the jogs formed by the intersection of screw dislocations. 469471.. 1957. Cottrell. little change in mechanical strength. An 1 attractive force exists between vacancies and dislocations.^ While the exact details for the mechanism of vacancy formation during cold work have not been established. H. F. "Dislocations and Mechanical Properties of Crystals. John Wiley & Sons. 2829. London.
When r.Sec. vol. o„. p. Under these conditions interaction occurs between solute atoms and both screw and edge dislocations. = — Is{(Tx + c^ + (7^) is the hydrostatic component of the stress The volume change is given by AV = %wea^ (621) a) /a is the strain where a is the radius of the solvent atom and e = (a' — produced by introducing a solute atom of radius a'. 63A. 1950. . Dislocation Theory 181 in the vacancies should be able to form atmospheres around dislocations same way as solute atoms. For simplicity. 61 3] fore. The segregation of An atoms to dislocations lowers the strain energy of the system. Bilby. other to form vacancy pairs (divacancies). Dislocation — Foreignatom Interaction The presence of a large foreign atom produces a dilation of the matrix. atom is located at a point given by the polar coordinate dislocation. solute atom produces an unequal distortion of the matrix atoms can interact with the shear component of the stress field as well as the hydrostatic component. Ui = a^AV (620) where field. (London). oversized atom will be attracted to the tension region and repelled from the compression region of an edge dislocation. Proc. 191. the solute 6 from an edge = ^^^^ = 4Gbea^'^ r r (622) The force force. Phys. For the case of carbon and nitrogen atoms in iron the tetragonal symmetry around the interstitial sites leads to a shear component of the stress field. it is assumed that the solute atom produces a symmetrical solute If the solute atom occupies a volof the matrix atom it replaces. The between an edge dislocation and a solute atom is not a central radial and tangential components are given by When the solute lattice in different directions. Soc. and there 61 3. the energy than the volume ume AV greater local stress field of the dislocation and the the of interaction between will be foreign atom hydrostatic distortion of the matrix. the interaction energy is given by^ U. In fee alloys the dissociation of dis1 B. Vacancies may also interact with each is some evidence to support the hypothesis that they collect into larger groups or clusters. A.
When the B. (622) diffusion coefficient of solute atoms at temperature T In the derivation of this equation the dislocation line serves as a soluteatom sink which captures any passing atom but does not obstruct the entry of other atoms. The local concentration strain aging. In this way the dislocation lines fi^ are freed to act as sinks for a longer period of time. If the interaction between the solute atoms is strong. . and a steadystate concentration gradient develops. This concept is valid during the early stages of D = found to hold. n{t). vol. However. 64. and the assumption that it acts like a sink can no longer be valid. Leak. Iron Steel Inst. ship will remain valid until solute atoms. two elastically bound dis Cottrell and Bilbj^ have shown that in time t the number of solute atoms. p.182 Metallurgical Fundamentals [Chap. f'^ where the relationship is toward the later stages of strain aging the sites c is related to the average concentration co by the relationship c = Co exp j~ (625) It has been suggested^ that solute atoms can diffuse along dislocations until they meet a barrier. Now the probability of atoms leaving the center equals the probability of atoms flowing in. on the dislocation line become saturated. A. all and the relation dislocation lines have been saturated with When the concentration of solute atoms around the dislocation becomes high enough. 6 locations into partial dislocations produces locations with a substantial edge component. (London). The steadystate distribution of solute atoms around the dislocation is referred to as an atmosphere. the atmosphere will condense into a single line of solute atoms parallel to the dislocation line at the position of maximum binding about two atomic spacings below the core of a positive edge dislocation. (622) and site of ro « 2 X 10"^ cm is the distance from the dislocation core to the 1 the line of solute atoms. a fine precipitate can be formed. /. 184. 1956. M. Bilby and G. The breakaway of solute stress required to pull a dislocation line is away from a line atoms at 0°K (626) 6Vo^ where A is given by Eq. that migrate to a unit length of dislocation line from a solution containing initially no solute atoms per unit volume is ^(0 = 3(^y''(^^y'no (624) where A = interaction parameter of Eq.
pro duced as the result of the growth of the crystal from the melt or the vapor phase. even with the assistance of thermal fluctuations. the maximum velocity at which a dislocation can move and still drag its atmosphere with it is line If the dislocation line is moving faster than this velocity.Sec. Completely annealed material will contain about 10^ to 10* dislocation lines per square centimeter. Experimental evidence for dislocations in crystals solidified under carefully controlled conditions has been obtained by etchpit studies and by Xray diffraction methods. When an external force tries to move a dislocation line away from its atmosphere. atoms. the atmosphere exerts a restoring force that tries to pull it back to it its original position. 614. It is generally believed that all metals. (626). with the exception of tiny whiskers. while heavily coldworked metal will have a dislocation density of about of dislocations in The density ingly small. location line pulling Serrations in the stressstrain curve are the result of the dis away from the solute atmosphere and then slowing down and allowing the atmosphere to interact once again with the dislocations. If the speed of the dislocation line is slow. thermal equilibrium with a crystal is vanishThere is no general relationship between dislocation density and temperature such as exists with vacancies. 10'^ dislocation lines per square centimeter. it will be necesis left sary for the restoring force to be overcome and the atmosphere behind. For crystals grown by vapor . a metal can have widely different dislocation densities depending upon processing conditions. may be able to move by dragging the atmosphere along behind According to Cottrell. it. initially contain an appreciable number of dislocations. 614] Dislocation Theory is 183 dislocation line slip pulled free from the field of influence of the solute can proceed at a stress lower than that given by Eq. Since dislocations are not affected by thermal fluctuations at temperatures below which recrystallization occurs. results in an important difference between line defects and point defects. This is the origin of the upper yield point in the stressstrain curve. it tion is so high as to make completely annealed crystals and in The line energy of a dislocavery unlikely that stresses of reasonable magnitude can create new dislocations in a region of a crystal where no This dislocations exist. Dislocation Sources The low yield strength of pure crystals leads to the conclusion that dislocation sources must exist in crystals carefully solidified from the melt.
184 Metallurgical Fundamentals [Chap. the density of dislocations in Thus. dislocation loops have been ration techniques. there must be a method of generating dislocations a single crystal. and produces cit. the dislocation line bulges out ^ a shear stress t acts in the slip slip. Phys. Silcox. T. 79. Read. 6 deposition it has been shown that nucleation of the soUd phase occurs around screw dislocations emerging from the surface of the solid substrate. 61 5a). The plane of the figure is the slip plane. These loops are believed to originate from the collapse of disks of vacancies and correspond to prismatic dislocations. . much more information is needed about the mechanism. by which they are produced and the way in which they are arranged in the metal. unworked crystals. op. vol. 2 Frank and W. Ample evidence in works ^ 61 5. Such a mechanism is when it is realized that due to the movement of about Thus. or of multiplying the number initially present to produce the high dislocation density found in coldworked metal. This could occur if D and D' were nodes where the dislocation slip planes. 722723. the number of dislocation 1. Smallman. cold work should decrease. of the existence of threedimensional dislocation netannealed ionic crystals has been provided by dislocation decoIn annealed metals. While there is little doubt that dislocations exist in annealed or carefully solidified metal. Rev. The scheme by which dislocations could be generated from existing dislocations was proposed by Frank and Read^ and is commonly called a FrankRead source. For a given Hirsch. observed by transmissionelectron microscopy of thin films. if there were no source generating dislocations. rather than increase.. 1950.000 dislocations over the slip plane. or If in the plane of the paper intersects dislocations in other the anchoring could be caused by impurity atoms. F. C. Moreover. plane. so that it is immobilized at these points. Multiplication of Dislocations — FrankRead Source required is One of the original stumbling blocks in the development of dislocation theory was the formulation of a reasonable mechanism by which sources originally present in the metal could produce new dislocations by the process of slip. pp. sources initially present in a metal could not account for the observed slipband spacing and displacement unless there were some way in which each source could produce large amounts of slip before it became immobilized. the surface displacement at a slip band Consider a dislocation line DD' lying in a slip plane (Fig. and Westmacott. There is some evidence to indicate that these loops can grow and join up to form dislocation networks in annealed. The dislocation line leaves the slip plane at points D and D'. which are then responsible for the formation of dislocations. There is also some evidence to suggest that some of the condensed vacancies form voids.
61 5rf the dislocation has almost doubled back on itself. each time prowill . 1953. plane with increasing stress.. becomes a semicircle so that R has the minimum value From the approximation that V ~ Q. 1/2 (615). Inc. In Fig. {W T. 6156). . [b) . This produces a complete loop and reintroduces the original dislocation line DD' The loop can continue to expand over the slip this critical value. The section DD' will soon straighten out under the influence of applied stress and line tension. which has started to double back on itself. while in Fig.'' McGrawHill Book Company. "Dislocations in Crystals.Sec. Schematic representation of the operation of a FrankRead source. Tb k Tb III [a) . Figure 61 5c shows the expanded loop.Tb (d) [e) Fig. When the stress is raised becomes unstable and expands indefinitely. New York.bGh'^ and Eq.) (615) it can be readily seen that the stress required to produce this is configuration Gh (628) where above I is the distance DD' between the nodes. the dislocation . 61 5. The maximum value required when the dislocation bulge (Fig. This process can be repeated over and over again at a single source.. Read. 615e the two parts of the loop have joined together. 61 5] stress the dislocation line will Dislocation Theory 185 assume a certain radius of shear stress is of curvature given by Eq. Jr. and the Frank Read source then be in a position to repeat the process.
57. John York. 1957. When the back stress diicing a dislocation loop the slip plane. Inc. New and Mechanical Properties of Crystals. C. "Dislocations Wiley & Sons. Dislocation Pileup Frequent reference has been made to the fact that dislocations pile up slip planes at obstacles such as grain boundaries. [W. Inc. Evidence has also been found by precipitation techniques in aluminum alloys and in ionic crystals and by means of thinfilm electron microscopy in stainless steel.. 616.) equals the critical stress given operate. . 616." p. Dash. Dash.. Mechanical Properties of Crystals. in "Dislocations and Sons. up along the slip plane opposes the applied stress. FrankRead source in silicon crystal. the source will no longer The most dramatic evidence for the existence of a FrankRead source has been found by Dash^ in silicon crystals decorated with copper. and sessile dislocations. The dislocations in the pileup will be on ^ W. C. 6 which produces slip of one Burgers vector along However. 1957. once the source is initiated. (628). Fig. Figure 616 shows a FrankRead source in a silicon crystal as photographed with infrared light. by Eq.186 Metallurgical Fundamentals [Chap." John Wiley & New York. secondphase particles. it does not conThe back stress produced by the dislocations piling tinue indefinitely.
calculations for . 1 J. F. Phil. = Gh (629) the average resolved shear stress in the slip plane and k is a factor close to unity. Mag.Sec. op. Head. 1951. p.. Mag. 617).^ The number of dislocations that can Source / Fig. cit. When number n the source is located at the center of a grain of diameter D. R. The distribution of dislocations of like sign in a pileup along a single slip plane has been studied by tightly Eshelby. pp. C. Frank. vol.. Eshelby. 616] Dislocation Theory 187 packed together near the head of the array and more widely spaced toward the source (Fig. 42. 351. total slip The produced by a pileup can be considered that due to a single dislocation nb moving a distance 3L/4. Nabarro.. more complicated types of pileups have been given by A. occupy a distance L along the obstacle is slip plane between the source and the klTTsL n where r. Very high forces act on the disD. experimental confirmation of theory has been obtained by Meakin and Wilsdorf. while for a is screw dislocation A = 1. pp. and Nabarro. N.. For an edge dislocation k = I — v. 745752. 4. 1959. 295302. Frank. 617. Dislocation pileup at an obstacle. K. At large distances from the array the stress due to the dislocations can be considered to be due to a dislocation of strength nb located at the center of gravity three quarters of the distance from the source to the head of the pileup. and F. vol. Phil. the of dislocations in the pileup is given by = ^^ many (630) The factor 4 stress is used instead of the expected factor of 2 because the back on the source arises from dislocations piled up on both sides of the source. A piledup array of n dislocations can be considered for purposes to be a giant dislocation with Burgers vector nb.
From size Eq. the above expression becomes )  p)(ao <r.5°. and the temperature. Roy. 7r(l = oe/2. p. S. 404414.. of Breakdown of a barrier can occur by slip on a new by climb of dislocations around the barrier. 223. in addition. Fetch's equation that expresses the dependence of yield stress on grain can be developed from the concepts discussed above. for If. vol. pointed out that large tensile stresses of the order of nr will be produced Stroh. the material. 85. Koehler. 2 A.has made a somewhat more detailed analysis of the stress distribution at the head of a dislocation pileup. For this (632) The shear stress acting in the plane OP is given by (633) T = ^Ts (jY where will The number the slip an orientationdependent factor which is close to unity. (London). Yielding is assumed to occur when a critical shear stress Tc is produced at the head of the pileup. equal to nbrg. Stroh. . tensile stresses. 480. This stress is assumed independent of grain size. where Koehler^ has at the head of a pileup. plane. N. vol. Soc. 1952. T. Rer. pp. shear stresses are Tc converted to uniaxial example. (630) we get nTs = 7r(l  pWD 4G6 assumed that the resolved shear stress is equal to the applied stress minus the average internal stress required to overcome resistances to disIt is location motion. of dislocations which can be supported by an obstacle depend on the type of barrier.)£ _ 8Gb 1 J. is This force is the average resolved shear stress on the slip plane. 617. 1954.188 Metallurgical Fundamentals [Chap. 6 locations at the head of the pileup. Phys. the orientation relationship between /S is plane and the structural features at the barrier. Proc. stress Using the coordinate system given in Fig. he showed that the tensile normal to a line OP is given by m ) Ts sin d cos 2 (631) The maximum value situation of a occurs at cos 9 = i^ or 6 = 70. or by the generation high enough tensile stresses to produce a crack.
G. (ed. W. York. 1956.. M. New York. Inc.Sec." American Institute of Mining and MetalJ.  = " + VJ^) 5 = . Fisher. in F. A. (eds. H. and W. Burgers: Dislocations in Crystal Lattices. R. 1956. I. Academic Press. Inc." vol. ." Oxford University Press. G.. J. W. New York. G. Vreeland. 1960. Cohen.): "Dislocations in Metals.. York. Schoek.. Johnston. (ed.+ "'"' BIBLIOGRAPHY ^"''^ Burgers.." Interscience Publishers.: "Dislocations in Crystals. Academic Press Inc. and T." John Wiley & Sons. lurgical Engineers. G.: "Dislocations New 1957. 616] Dislocation Theory 189 This can be rearranged to give the desired relationship between yield stress (To and grain diameter D. R." McGrawHill Book Company." vol.). and Plastic Flow in Crystals. 1953. Van Bueren. Cottrell.: Dislocation Theory of Plasticity of Metals. Read. IV. Jr. Inc. 1953. New York.): "Dislocations and Mechanical Properties of Crystals. H. Jr.: "Imperfections in Crystals. New York.. in "Advances in Applied Mechanics. C. 1953. New T. M. Thomson. New York. Inc.. Eirich "Rheology.
increasing strain rate. The two . Introduction is the separation. 14. An appreciable amount of gross deformation is is usually present at the fracture surfaces. Fracture occurs in characteristic ways. with It is is no gross deformation and very microdeformation. and triaxial stress conditions (usually pro duced by a notch). and creep (Chap.Chapter 7 FRACTURE 71 . (Chap. 12). The process of fracture can be considered to be made up of two components. because it occurs without warning and usually produces disastrous consequences. crack initiation and crack propagation. This chapter will present a broad picture of the fundamentals of the fracture of metals. 13) and lowtemperature brittle fracture. and rate of loading. Unless otherwise stated. this topic will be given considerable prominence. temperature. and the temperature. ductile fracture and brittle fracture. Since most of the research has been concentrated on the problem of brittle fracture. A ductile fracture is characterized by appreciable plastic deformation prior to and during the propagation Fracture or of the crack. Brittle fracture is to be avoided at all cost. 72. or fragmentation. fatigue (Chap. depending on the state of stress. Fractures can be classified into two general categories. ized Brittle fracture in metals character by a rapid little rate of crack propagation. it will be assumed in this chapter that fracture is produced by a single application of a uniaxial tensile stress. The engineering aspects of brittle fracture will be con sidered in greater detail in Chap. akin to cleavage in ionic crystals. of a solid body into two more parts under the action of stress. state of stress. Metals can exhibit 190 many depending on the material. Fracture luider more complex conditions will be conTypical examples are fracture due to torsion (Chap. increased with decreasing temper The tendency for brittle fracture ature. or hydrogen embrittlement sidered in later chapters. Types of Fracture in Metals different types of fracture. temper embrittlement. the rate of application of stress. 14). 10).
plastic to describe fractures as follows: Behavior described . A brittle fracture (Fig. and the appearance Gensamer^ has summarized the terms commonly used of the fracture. 7la) is sidered. like may actually be drawn down to a point before they rupture gold or lead. characterized there is by separation normal of deformation. Tj^pes of fractures observed in metals subjected to uniaxial tension. Polycrystalline specimens very ductile metals. 7lc). to the tensile stress. of 716). Single crystals of hep metals may slip on successive basal planes until finally the crystal separates I {a) Fig. (d) In the tensile fracture ductile fracture in polycrystals. crystallographic mode of fracture. (a) by shear (Fig. moderately ductile metals the deformation eventually produces a necked region (Fig. 7 Id). ductile shearing fracture in ductile single crystals. Outwardly no evidence it although with Xray diffraction analysis is possible to detect a thin layer of deformed metal at the fracture surface.O's to grainboundary embrittlement. Ductile fractures can take several forms. familiar "cupandcone" fracture. (c) completely fracture in polycrystals. but not in fee metals unless there are factors contributing . Brittle fractures have been observed in bcc and hep metals. such as strain to fracture. (Fig.72] Fracture 191 broad categories of ductile and brittle fracture have already been conFigure 71 schematically illustrates some of the types of tensile fractures which can occur in metals. Fracture begins at the center of the specimen and then extends by a This results in the shear separation along the dashed lines in Fig. 7 Id. of Brittle fracture of single (6) crystals and polycrystals. Fractures are classified with respect to several characteristics. 71.
. while a cleavage fracture appears bright or granular. Fracture surfaces frequently consist of it is a mixture of fibrous and granular fracture. which equal to the theoretical cohesive strength of the material. The boundary between a and brittle fracture is For example. coeffi high melting points. a deeply notched tensile specimen will exhibit little gross deformation. If shows the variation of the cohebetween two atoms as a function of the separation between This curve is the rethese atoms. Theoretical Cohesive Strength oF Metals arbitrary and depends on the situation being considered. and customary to report of these categories. This corresponds maximum in the curve. X Fig. A point force reached decreas where the repulsive force to the is negligible and the attractive is ing because of the increased separation of the atoms. Metals are of great technological value. In general. sultant of the attractive and repulThe sive forces between the atoms. high cohesive forces are related to large elastic constants. nodular cast iron is ductile when compared with ordinary gray iron. yet it would be considered brittle when compared with mild steel. A ductile fracture is one which exhibits a considerable degree of ductile deformation. yet the fracture could occur by a shear mode. As the tensile load increased is is still further. primarily strength combined with a certain measure of plasticity. As a further example. *^^he Cohesive force as a function of separation between atoms. The repulsive force decreases more rapidly with is increased separation than the attractive force. the separation between will be increased.192 Metallursical Fundamentals [Chap. because of their high In the most is basic terms the strength due to the cohesive forces between atoms. 73. so that a net force between atoms balances the tensile load. interatomic spacing of the atoms in 72 sive force the unstrained condition is indicated by ao. the repulsive force continues to decrease. 7 by shear appears from the flat at low magnification to be gray and fibrous. the percentage of the surface area represented by one Based on metallographic examination. Separation Figure between atoms. and small cients of thermal expansion. owing to reflection of light cleavage surfaces. atoms the crystal is subjected to a tensile load. 72. fractures in polycrystalline sam ples are classified as either transgranular (the crack propagates through the grains) or intergranular (the crack propagates along the grain boundaries).
Eq. = —^^ ^A TT f/o = / <T„ax sm (72) J3 assumed that new surface is y. take the (71). is The work done during fracture. O" (77) . (74) can be differentiated to give dx ao (73) Equating final (75) and (76) and substituting into Eq. sm — (71) where omax is the theoretical cohesive strength. produces the expression for the theoretical cohesive strength of a crystal. Eq. 73] Fracture 193 A obtained good approximation to the theoretical cohesive strength can be if it is assumed that the cohesive force curve can be repre sented by a sine curve. = 27r'Y — (73) Since Hooke's law holds for the initial part of the curve. the _ ~ 2wx '"'"="' y ^°^ IT of x is approximately unity for the small values above expression can be written as which Also. (72) can be written unit area required to produce a all The energy per it is new or (T^a. per unit area. da dx Since cos (27rx/X) are involved. the area under the curve.Sec. dx. If the work involved in fracture goes into creating two surfaces.a. o = o„. (73). derivative of Eq. the stress can be written as <T = — do first 2tt (74) In order to eliminate X from Eq.
Congr. while it cannot be applied directly to metals. Soc. and so the problem can be treated as one in plane stress. Griffith's ideas have had great influence on the thinking about the fracture of metals. 55. Griffith crack model.000 times greater strength of the order of 2 X 10^ psi. Roy. or. The stress 1 A. First Intn. Only the fracture strength of dislocationfree metal whiskers approaches the theoretical cohesive strength. expressed in another way.^ theory in its original form is applicable only to a perfectly brittle material such as glass. The effect of both types of crack on the fracture behavior is the same. size to propagate as a brittle fracture. pp. The cracks are assumed to have an elliptical cross section.. The thickness of the plate is negligible. 73. For a crack at the interior the length is 2c. A. However. of the crack. Consider the crack model shown in Fig. 221A. which When one it of the cracks spreads into a brittle fracture. 163198 1920. London. Delft. Mech. Trans. pro duces an increase in the surface area of the sides This requires energy to overcome it the cohesive force of the atoms. Appl. This criterion can be used to determine the magnitude of the tensile stress w^hich will just cause a crack of a certain create the new crack surface. 71) results in the prediction of a cohesive This is 10 to 1. Griffith. while for an edge crack it is c.194 Metallurgical Fundamentals [Ch ap. Griffith proposed that a brittle material contains a population of that the theoretical cohesive strength in localized regions at a is fine cracks which produce a stress concentration of sufficient magnitude so reached is nominal stress well below the theoretical value. The substitution of reasonable values for the quantities involved in the above expression (see Prob. p. . is requires an increase in surface enof the increased surface is The source energy the elastic strain energy which released as the crack spreads. will propagate Griffith established the following : criterion for the propagation of a crack A crack when the decrease in elastic strain to energy is at least equal the energy required to Fig. Griffith Theory of Brittle Fracture The by first explanation of the discrepancy between the observed fracture Griffith's strength of crystals and the theoretical cohesive strength was proposed Griffith. Phil. 73. than the observed fracture strengths of metals. 74. ergy. 1924. vol.
the increased surface energy is compensated by a decrease in elastic strain energy. Trans. by a factor of 4 reduces the fracture stress is by onehalf. Therefore. The surface energy due to the presence of the crack Us = 4c7 The total is (79) change in potential energy resulting from the creation of the crack AU = Us+ Ue According to Griffith's criterion. i. stant applied stress a an incremental increase dAU _ dc = — (^^^ dc . Sack.. Proc. where the crack is a very flat oblate spheroid. (London). 55. Inst.results only in a modification to the constant in Griffith's equation. The elastic strain is energy per unit of plate thickness equal to U. R. pp. 74] Fracture 195 distribution for an elliptical crack in strain was determined by Inglis. A. E..^^) Equation (711) gives the stress required to propagate a crack in a brittle material as a function of the size of the microcrack. vol. For a plate which thick compared with the length of the crack (plane is strain) the Griffith equation given by 2Ey (1  (712) j')Vc Analysis of the threedimensional case. p. Inglis. . Soc. Note that this equation indicates that the fracture stress is inversely proportional to Thus. 1946.Sec. = where a is "^ is (78.^ A decrease energy results from the formation of a crack. the tensile stress acting normal to the crack of length 2c. 58.e. pt. increasing the crack length the square root of the crack length. 2 I. the if (710) crack will propagate under a con in crack length produces no change in the total energy of the system. Naval Architects. 729. 1913. Phys. An alternative way of rationalizing the low fracture strength of solids 1 C. the simplification of considering only the twodimensional case introduces no large error. vol. 219230.
123. 38. Early experiments on the fracture of glass freshly fibers showed that strengths close to the theoretical fracture strength could be obtained with fibers drawn from the melt. L. temperature of the melt. 353. 27. vol. and amount and rate of drawing from the melt. Orowan. the theory predicts a crack length of several millimeters. Brenner. This average crack length could easily be greater than the thickness of the specimen. m' equation. Recent results^ indicate that there is no dependence of strength on diameter when differentsize glass fibers are prepared under nearly identical conditions. pletely brittle material such as glass. reasonable values of crack length of about 1 n are calculated from Eq. Am. 2 pp.. with a radius of curvature p at the end of the crack. The strength This is of a metal whisker varies inversely with the its diameter. Making this substitution in Eq. Experiments on metal whiskers* have also demonstrated fracwith the smallestdiameter fibers. 331J. on the average these fibers would However. can affect strength. in "Fracture. p. Appl. Phys. For zinc. p combining it = ao. 1955. Griffith Criterion for Glass Fracture. ^ Griffith's (714) Within the accuracy of the estimate. J. 1959.^ Inghs showed that the stress at the end of an ellipsoidal crack of length 2c. The sharpest radius of curvature at the end of the crack should be of the order of the interatomic spacing. the whisker con tains a certain 1 number The of dislocation sources.." pp. S. New York. . 7 with the high theoretical cohesive strength was proposed by Orowan. and therefore the theory cannot apply. vol. the length of the most E. (711). The highest fracture strengths were found have the shortest microcracks. 1956. if the type of size dependence that would be expected the strength were controlled if by number of surface defects. Ceramic Soc. Otto. 3 < John Wiley & Sons. is given by (713) where o is the nominal stress when no crack is present. Inc. Anderson.. Welding O. other factors besides diameter. (713) and with Eq. p. F. 157s160s. such as method of preparation. 34. 1484. (77) results in an expression for the critical stress is to cause brittle fracture which similar to Griffith's equation. S. since ture strengths close to the theoretical value. On the other hand.196 Metallurgical Fundamentals [Chap. this equation predicts the same stress needed to propagate a crack through a brittle solid as the Griffith theory satisfactorily predicts the fracture strength of a comIn glass. vol. /. 1955.
solution of the salt crystal in the water. 1951. K.Other indications that brittle fracture in metals is always preceded by a small amount of plastic deformation. D. since estimates of the plasticwork term are about 10^ to 10^ ergs/cm^. /. 34. 2E(y + p) ire m' (715) The surfaceenergy term can be neglected. 570s575s. 935957.. Inc. ionic crystals has 75. Welding J. 3. is touched with a hard object. 1950. Felbeck and E. P. ModiFications of the Griffith Theory Metals which fracture in a brittle manner show evidence of a thin layer deformed metal when the fracture surface is examined by Xray diffraction. extremely sensitive to surface defects. pp. the strength will instantly decrease to a low value. NewYork. Joffe. Trans.Sec. Orowan. Chang. the Joffe effect in these crystals cannot always be explained simply by surface dissolution. vol. it appears that Griffith's form. "The Physics of Crystals. 212217. 75] Fracture 197 extended source will vary directly with the diameter and the strength again be inversely related to the whisker diameter. under the influence of atmosof glass fibers is The strength If the surface of a freshly drawn fiber pheric attack. However.. Thus. pp. 77. andPhys. Orowan. Even the strength of a fiber which has not been handled will. decrease to a low value within a few hours of being from the melt. should not be expected to apply to the brittle fracture of metals. of Mech. Inc. L. 1955. New York. ' Klier. Orowan' suggested that the Griffith equation could be made more compatible with brittle fracture in metals by the inclusion of a term p expressing the plastic work required to extend the crack wall. are given in Sec. This attributed to the healing of surface cracks by the has been effect The fracture behavior of other been shown to depend on the environment in contact with the surface. and Fracture of Metals. vol. compared with values of y of 1 A. in "Fatigue Institute of Technology. 2E. Joffe^ drawn greatly increased Joffe showed that the fracture strength of NaCl crystals could be when the test was carried out under water. 1955. F. ASM. E. vol. pp. 43. Solids." symposium at Massachusetts John Wiley & Sons." McGrawHill Book Company. 1928. theory. C.. on a microof plastically scopic scale.. it is not possible from the size dependence of strength to establish whether the high strength of whiskers is due to a freedom from surface defects or will dislocations. . in its original Therefore.
in Irwin's modification of this theory 9 is taken as For a finite plate of width L an experimentally determined parameter with a central crack of length 2c or two edge cracks of length c. 1960.000 to 2. To measure 9c. Naval Research Lab. J. the fracture toughness. January and February. and H. temperature. There is some experimental evidence that p decreases with decreasing temperature. value of 9 for this condition is equal to 9c. 2 Detailed procedures for measuring 9c in tension have been presented in the Bulletin. 4763. depending on temperature and composition. May. a crack was assumed to propagate rapidly when 9 = 27. R. 58. pp. Proc.^. the crackextension force for tensile loading is given by = ^ E (1  . The quantity ^ of interest is the crackextension force. H.^) tan (^) (717) ^ G. PB 121224. ASTM ASME. rate. 1956. It represents the fraction of the total work expended on the system which is irreversibly absorbed in local plastic flow and cleavage to create a unit area of fracture surface. Irwin. the geometry of the specimen. this quantity reaches a critical value £c. Wundt. ASTM. Methods using a notchedbend test and a highspeed rotating disk have been given by D. Smith. 1958. vol. Irwin. Trans. R.lb/in. also called the strainenergy release The crackextension force 9. Kies. However. the crack will 9c is propagate rapidly. is the quantity of stored elastic strain energy released from a cracking specimen When as the result of the extension of an advancing crack by a unit area. An extension of the Griffith theory into the area of fracture mechanics The objective is to find a reHable design crihas been made by Irwin.198 Metallurgical Fundamentals [Chap. G. 1958. L. 1643. (716) with the modified Griffith equation (715) shows that 9 is analogous to Orowan's plasticwork factor p. lb/in. and rate of loading.000 ergs/cm.For a crack of length 2c in an infinitely wide plate the relationship between the stress and 9 is given by '  &r Comparison of Eq. available from Office of Technical Services. A. 2. microstructure. Winne and B.^ The specimen is then loaded until a stress is reached at which the crack which was The calculated initially present in the specimen propagates rapidly. In the original Griffith theory. it is necessary to have a reliable mathematical expression for 9 in terms of the crack dimensions. 640660. 80. effects. terion for predicting the stress at which rapidly propagating fractures This is essentially a macroscopic theory that is concerned will occur. 7 about 1. and the nominal applied stress. the elastic constants. Rept. 9c appears to be a basic material property which is essentially independent of size It does depend on composition. measured in units of in. with cracks that are tenths of an inch in length or greater. Values of 9c for steel vary from about 100 to 600 in. . vol. p. M.
The area of the cleavage plane is A /(cos <^). 76] Fracture 199 76. fracture occurs considered to be related to the Sohncke's law states that when the resolved normal stress reaches a critical value. Therefore. Considering the situation used to develop the resolved shear stress for slip (Fig. 418). P The cleavage cos </) A/(cos = P r cos" 4>) A (718) planes for certain metals and values of the critical normal stress are given in Table 71. the component of the tensile force which acts normal to the cleavage plane is P cos where is the angle between the tensile axis and the normal to the plane. Table Critical 71 Normal Stress for Cleavage of Single Crystals t Metal . Fracture of Sinsle Crystals The brittle fracture of single crystals is resolved normal stress on the cleavage plane.Sec. the critical normal stress for brittle fracture is </>.
1 N.Verlag OHG. interesting point that the change from brittle to ductile fracture change in orientation of only about 2°. The mode of fracture in bcc iron crystals is strongly dependent on ^ Crystals temperature. purity. Soc. An excellent correlation between plastic deformation. Metallosraphic Aspects of Brittle Fracture Because metals. 2 J.A. 234A. I. More usually. very sharp. . 7 modes of ductile fracture in single crystals are shown in Fig. microcracks. Proc. it has been natural for metallurgists to use their microscopes in a search for Griffith cracks in However. heat treatment." p. Several 71. and crystal orientation. Springer. 60. which is very low in inclusions. P. Allen.M. E. slip ture will then occur by "shearing off" (Fig.T. (London). will necks occur on systems other than the basal plane. and J. based on observations up to the magnifications availis no reliable evidence that There is. they do contribute to the observed anisotropy in the ductilefracture strength. p. however. "Deformation and Flow of Solids. (if multiple slip by slip on one set of planes until The crystal can draw down to a chisel edge or a point continues to fracture). 1956. shows a reduction in the fracture anisotropy supports the idea of microcracks being formed at secondphase particles. The fact that vacuummelted steel. B.U. occurring over a 77. Griffith cracks exist in metals in the unstressed condition. The best stress criterion for followed ductile fracture in fee metals appears to be the resolved shear stress on the fracture plane (which is usually the slip plane). E. 1956. Low. 221. Hopkins. 716). Madrid Colloqium.200 Metallurgical Fundamentals [Chap. An is while crystals closer to and [Oil] orientations may rupture by drawing down to a chisel edge is when tested at the same temperature. able with the electron microscope. However. McLennan. there Metallographic evidence of the formation of microcracks at nonmetallic inclusions in steel as a result of plastic deformation has existed for a num These microcracks do not necessarily produce brittle fracture. Berlin. vol. and brittle fracture was made by Low. Roy. located near the [001] corner of the stereographic triangle show no measurable ductility [111] when tested in tension at — 196°C. Under certain conditions hep metals tested at room temperature Fracor above will shear only on a restricted number of basal planes. R. The usual mode of fracture in fee crystals is the slip.^ He showed that for mild steel of a ber of years. of the prominence of the Griffith theory. a growing amount of experimental evidence to show that microcracks can be produced by plastic deformation. formation of a necked region due to multiple fracture occurs. so that the crystal down and draws down almost to a point before rupture occurs.
and M. in the neighborhood of room temperature. Cohen. A transition from ductile to brittle fracture occurs at the ductility transition temperature Td. In region A. G. men before it The correlation between the temperature dependence of yield stress.Sec. Hahn. Hahn. In region B the but the outer rim of the fracture contains cleavage facets. and ductility and microcrack formation is shown in Fig. the ductility transition occurs when the conditions are suitable for the growth of microcracks into propagating The initiation of microcracks is not a sufficient criterion for S. 77] Fracture 201 given grain size tested at — 196°C brittle fracture occurs in tension at the same value of stress that is required to produce yielding in comMicrocracks only one or two grains long were observed. the fracture stress. ^^^0^ /'^ Fig. T. detailed studies of the conditions for microcrack formation have been made^ with peratures. Owen. 250 X {Courtesy G. Averbach.. T. tensile tests on mild steel at carefully controlled subzero temFigure 74 illustrates a typical microcrack found in a specifractured. W. 367s376s. 75. to practically a zero value. 38. Microcracks produced in iron by tensile deformation at — 140°C. vol.) fracture stress. The reduction of is fracture 50 to 60 per cent. pp. . However. 74. 1959. microcracks are found above fractures. • Td. Welding J. a tensile specimen fails with a ductile cupandcone fracture. B. Therefore. Accompanying this is a large decrease in The percentage of grains containing microcracks increases rapidly in region C just below Td. L. The existence of a transition temperature is indicated by the drop in the reduction of area at the fracture area at fracture is still of the order of ductile. More pression.
Presumably fracture occurs at the first spot to undergo discontinuous yielding. In region E cleavage fracture occurs abruptly before there is time for discontinuous yielding.present in the material but are produced by the deformation process. The fact that at appropriate temperatures appreciable numbers of microcracks are . 75. Cohen. eventually the fracyield point. B. Finally. at very low temperatures in region F fracture is initiated by mechanical twins. Temperature dependence of fracture stress. °C Fig. L. Hahn.) has undergone some discontinuous yielding. yield stress. and microcrack frequency for mild steel. 1959. Microcracks occur only in regions which have undergone discontinuous yielding as a result of being loaded through the upper As the temperature drops in region C. 7 brittle fracture. Detailed experiments such as these demonstrate that the cracks responsible for brittlecleavagetype fracture are not initiallj'. ture stress drops to a value equal to the lower yield stress. S. 372s.. In region D the lower yield stress and fracture stress are practically identical. p. T. Welding J. vol. 38. Averbach. Mechanical twins are observed at temperatures as high as Td. and M. Fracture occurs at a value equal to the lower yield stress after the material Lower yield stress  100 Reduction in area CJ» t o o tl^icrocraclfs Temoeroture.^ 202 Metallurgical Fundamentals [Chap. but it is only in region F that they appear to be the source of initiation of fracture. {After G. Owen. The fracture stress increases because the yield stress is increasing with decreasing temperature. W.
O. (2) crack initiation. Worden. nitro gen. segregation at the grain boundaries can lower the surface energy sufficiently to cause intergranular failure. and (3) crack propagation. (1) plastic deformation. Zappfe and C. The process of cleavage fracture should be considered to be made up of three steps. 76. Cleavage steps and river pattern on a cleavage surface. Apparently. or carbon. ' C. Sometimes a considerable amount of information can be obtained by examining the surfaces of the fracture at fairly high magnifications. austenitic stainless steel or molybdenum alloys containing oxygen. ASM. 42. Inter granular failure can also occur without the presence of a microscopically visible precipitate at the grain boundaries. vol. However. if the grain boundaries contain a film of brittle constituent. the fracture will occur in an intergranular manner. .Sec. as in sensitized 7^^ Fig.^ At high magnification. Most brittle fractures occur in a transgranular manner. The are indications of the absorption of energy by local deformation. The embrittlement produced by the addition of antimony to copper and oxygen to iron and the temper embrittlement of alloy steels are good examples. This type of examination is known as fractography. 77] Fractu re 203 present shows that the conditions for the initiation of a crack are not necessarily the same as conditions for the propagation of a crack. 1950. transgranularcleavage surfaces usually contain a large number of cleav These age steps and a "river pattern" of branching cracks (Fig. A. 577603. pp. 76). Trans.
/.. vol. 3 N. 2 A. J. {London).204 Metallurgical Fundamentals [Chap. Petch^ has found that experimental data agree best with an equation of the type as ^ for iron and steel = a + KD^'^ (723) C. Dislocation Theories of Fracture The idea that the high stresses produced at the head of a dislocation pileup could produce fracture stress acting was first advanced by Zener. Stroh^ has proposed that a cleavage crack can form when n dislocations piled up under the action of a resolved shear stress r^ satisfy the condition nbTs = 127 is (719) where b is the Burgers vector and y the surface energy. the energy than for transgranular cleavage. 25. with a From the appearance absorbed in an intergranular fracture surface. (London). vol. 1948. Fetch. general absence of cleavage steps. 7 surfaces of intergranular brittle fractures are mvich smoother. Zener. in "Fracturing of Metals.^ on the slip plane squeezes the dislocations together. 968. The Micromechanism of Fracture. Phil. Soc. 1953. 404." American Society for Metals. . is The length of the slip plane that the pileup will occupy given by (720) nbG 7r(l — v)ts Eliminating n from these equations gives rs^L = ^^ 7r(l — (721) V) When a specimen of grain size D is tested in tension. p. vol. Proc. N. 223 A. p. Ohio. 46. The fracture stress in tension can be expressed in grain size (t/2 and terms of by QGy 0/ ' 7r(l  D''v) = KD^'^ (722) However. Roy. The shear At some critical value of stress the dislocations at the head of the pileup After analyzing the stresses at a dislocation pileup are pushed so close together that they coalesce into an embryonic crack or cavity dislocation. Metals Park. Mag. Iron Steel Inst. of the fracture is much lower 78. 1955. Stroh. 1954. and making use of the Griffith criterion. p. Ts = L = D/2. 174.
the stress 1 is the controlling step factor. The fact that brittle fracture can occur in single crystals suggests that the role of grain boundaries as barriers for dislocation pileup may be overemphasized in current theories. Trans. If the propagation of microcracks. (722). 212. Also. the fracture equation stant The constant K in given approximately by Eq. 1959. 78] Fracture is 205 This equation very similar to the equation expressing the grainsize dependence of the yield strength. not the hydrostatic component of stress. AIME. Acta Met. In both equations ai is the frictional resisting the motion is of an unlocked dislocation. The con equation is a measure of the localized stress needed to unlock dislocations held up at a grain boundary so that yielding can be transmitted to the next grain by the propagation of a Liiders band. H.. 1118. vol. 1958. facts of fracture has led Cottrell and Fetch independently to conclude that the growth of a microcrack into a selfpropagating fracture is a more difficult step than the nucleation of microcracks from glide dislocations. Soc. 8. + KyD^'^ (724) This similarity stress is to be expected in view of the fact that yielding and brittle fracture are closely related. it is questionable that the necessary stress concentration can be produced at the head of a pileup before slip occurs in the neighboring grains to relieve the high localized stresses. pp. crack nucleation by dislocation coalescence should depend only on A consideration of the known the shear stress. normal to the crack would be an important 2 3 H. vol. according to a Griffithtype criterion. vol. Met. .there also evidence to show that brittle fracture can be produced in the absence of mechanical twins. 69). Birnbaum. D.. Acta Met. the strong orientation dependence of fracture of iron single crystals can be explained' on this basis. is While there experimental evidence to indicate that twin intersections is may initiate brittle fracture. but not for an fee lattice.Sec. pp. A. This term increases with decreasing temperature of testing. K. in fracture. But there is ample experimental evidence that fracture is strongly influenced by the hydrostatic component of stress (see Sec. Hull. This mechanism is energetically favorable for a bcc and hep metal. the brittle deformation twins may act as barriers for For example.. This quantity is important in current theories of fracture. pp. Moreover. Ky in the yieldstrength It is possible that dislocation pileup. 7. Support for this viewpoint is found in the fact that many nonpropagating microcracks are observed. (TO = cr. 716). 516517. 66 and Fig. in agreement with the fact that fee metals do not undergo brittle fracture. 192203. Cottrell. 1960. mechanism by which cracks can form is the glide of Another dislocations on inter secting slip planes according to the hypothesis of CottrelP (see Sec.
206 Metallurgical Fundamentals [Chap.^ where n is the number and y slip is of dislocations of Burgers vector h that coalesce into the crack the surface energy of the crack. To evaluate nh. When the glide dislocations coalesce into a crack or a cavity dislocaTherefore.2 (^^y (730) Equations (728) and (729) express the limiting conditions for the formation of propagating crack from a pileup of glide dislocations. for tension jS = 3^^. If conditions are such that the lefthand side of the equation is less than the » Ihid. . + k. n is the frictional resistance. and for the plastically constrained region at the root mum of a notch j8 « 3^. 7 This should lead to a strong dependence of fracture on the hydrostatic component of stress. the frictional resistance to glide equals zero.D'^^ (727) Writing Eq. Utilizing the Griffith criterion. L is The shear displacement at the center of the length If L is given by (r — Ti)L/G. (725) one arrives at an expression for the stress required to propagate a microcrack of length D. The effective shear stress on the plane is given by r — n. Cottrell^ has shown that the stress required to propagate a microcrack is given by (725) . Equation (724) can be written in terms of shear stress as TO = r. where nh. assume that a r plane of length L is acted on by an applied shear stress slip ~ a/2. and this is approximately equal to taken as about onehalf the average grain diameter D. .)k. by making substitutions from the above equations into Eq. (725) as nhro above equations results in = y and substituting for nh and to from the inD'^' + k. tion. = (?7/3 (728) or the equivalent relationship TokyDy^ = Gyl3 (729) In the above equations /3 is a term which expresses the ratio of the maxiFor torsion /3 = 1. shear stress to the maximum normal stress. .
cannot grow beyond a certain This is the case of nonpropagating microcracks. is When the left hand side of the equations greater than the right side. friction stress. since it Table 72 gives some typical values of ky obtained from measurements of Table 72 Values of ky/Gf Material . 75 for tension tests Therefore. these equations predict a ductiletobrittle transition. and k. a crack can form but length.78] Fract ure it 207 righthand side. The determines the number of dislocations that are released into a pileup when a source is unlocked. such as was shown on mild steel at decreasing temperature. state of stress. yield stress. surface energy. a propagating stress.. brittle fracture can be produced at a shear stress equal to the yield in Fig.. parameter ky is very important. The equations in describing the ductiletobrittle transition are expressed terms of the following metallurgical or mechanical factors: grain size.
7 From Eq. The effect of a notch in decreasing the ratio of shear stress to tensile stress is covered in Cottrell's equations by the constant /3.208 Metallursical Fundamentals (729).4 o 0. 712. High values of surface energy tend to promote ductile fracture. for a notch to produce the plastic ." This branch of the curve extrapolates to the fracture stress for a single crystal. Strain rate or rate of loading does not enter explicitly tendency for into Cottrell's equations. 5 6 X Fracture stress o Yie/d stress o Strain to fracture 0. Low. It is well known that the presence of a notch greatly increases the brittle fracture. r G.) (/. Ar = ^ DH (731) This predicts that fracture stress extrapolates to zero at D~5^^ ship is satisfied. 1954. although various environmental and metallurgical conditions may lower the surface energy. No. this is not a factor which is readily increased. T/ = To + 0. Microstructure. The embrittlement of steel due to hydrogen has been attributed to this factor. 77. which Figure 77 shows that this relation In the region of grain size for which cracks propagate as completely brittle fractures. film Intergranular fracture due to an embrittling may also be explained in this way. the fracture stress equals the yield stress. However. Effect of grain size on the yield and fracture stresses for a lowcarbon steel to R. Metals Park. tested in tension at — 196°C.6 ^ 0. the necessary value of shear stress is [Chap. Ohio. in "Relation of Properties American Society for Metals. Approximate 3 200 I 13 i ASTM. Unfortunately.2 I ^ 1 2 ^ _i "^ 1 5 ^2 6 (Grain diameter)" ^2 ^ rnm~ Fig.S. The complicated effects of a notch will be considered more fully in Sec. is = a linear function of Z)~5^^.
increasing the strain rate raises the 79. . transition temperature. are rapidly loaded to a constant 10 10' 10 10"2 10"' 1 10 10^ 10^ 10" Delay time. S. such as occur in a notchedimpact test. ASM. 1954.^ found that a certain delay time is required Figure 78 shows that the delay time increases with decreasing temperature at a constant stress. notably mild steel. Clark. yielding will have to occur more rapidly. vol. S. For a constant temperature. 46. p. Clark. Trans. Therefore. 1954. it is necessary for the material At high rates of strain. c^ 3^'. The above the yield stress. Trans. (732) in the next section. 49. it is lower limiting stress shown by the horizontal portion of the curves corre sponds to the upper yield point for tests carried out at slow speeds. When certain metals. (D.79] Fracture j8 209 constraint that results in a value of to yield locally. 46. sec Fig. Delay time for initiation of yielding in mild steel as a function of stress. Delayed Yielding A phenomenon which is important to brittle fracture is delayed yielding. ASM. the delay time increases with decreasing stress. p. 34. vol. 78.) stress before plastic yielding occurs. The temperature dependence of the delay time may be expressed by an exponential relationship t = to exp kT (732) 1 D. this can occur at the same value of ro if the temperature is increased. As is indicated by Eq.
approximately 10~^^ sec Boltzmann's constant stressdependent activation energy is Cottrell' has estimated that Qia/ao). p. 1957.38. The term c is cg is the length of a Griffith crack. (733) approaches the limiting value Bvq. The elastic energy that is released by the movement of the crack is the driving force. — o"/ao)^ where a is the applied stress and the yield The fact that brittle fracture occurs when plastic deformation fails to keep the stress below a critical value indicates that there should be a connection between delayed yielding and brittle fracture. delayed yielding probably plays the important role of localizing the slip by preventing nearby dislocation sources from operating. and the actual crack length. vol. 1948. on Properties Materials at High Rates of Strain. K. The delay time is quite temperaturedependent. F. 16. At temperatures where the metal fractures in a ductile manner the delay time is so short that slip occurs around the pileups and the high localized This agreement is supstresses are dissipated by plastic deformation. velocity v is given by V = Vo Bvo ( 1  — is (733) jrhere B is a constant and = {E/p)'' the velocity of sound in the material. isotropic medium.9(1 stress. A. 2 » London. 710. Conf. N. Engineering. p. Mott. H. Roberts and A. The constant has been evaluated^ for the planestress condition and found tohe B c^ 0. In the temperature region where brittle fracture is caused by an avalanche of dislocations breaking away from a barrier and running together to form a crack. Institu tion of Mechanical Engineers. When c is large by compared with Cg. Q{a/ao) = = = = delay time a constant. . in electron volts. ported by the fact that metals which have a ductiletobrittle fracture transition also have a delayed yield phenomenon. 1 A. and so is brittle fracture. 178. 7 ^0 A. (711). Eq. Wells. Cottrell.) 210 where t Metallurgical Fundamentals [Chap. as evaluated Eq. D. 165. given approxioq is mately by 0. 820. Velocity of Crack Propagation Brittle fracture is not possible unless the cracks which are nucleated can propagate at a high velocity throughout the metal. This must be balanced by the surface energy of the new surface that is created and the kinetic energy associated with the rapid The crack sidewise displacement of material on each side of the crack. Engineering. vol. Mott^ has made an analysis of the velocity of a crack in an ideal elastic. 1954. Proc.
711] Fracture 211 in brittle Table 73 shows that experimental values for the crack velocity crack velocity is materials agree quite well with the theoretical prediction that the limiting given by V = 0.38 Kf) ( )  ^" (734) Table 73 Velocity of Propagation of Brittle Fracture Material .38^0 = 0.Sec.
This crack grows in a direction perpendicular to the axis of the specimen until it Shear Fibrous (^1 Fig. 7 along the axis of the specimen at the center of the necked region. 79d). Extensive localized deforand because the . 796). The outer cone a region of highly localized shear. 79c).212 Metallurgical Fundamentals [Chap. approaches the surface of the specimen. When the central "cup" region of the fracture is viewed from above. fine cavities Many form in this region (Fig. the central crack has a zigzag contour. such as would be proof the duced by tearing between a number fracture is of holes. 79. It then propagates along localized shear planes at roughly 45° to the axis to form the "cone" part of the fracture (Fig. When the fracture is sectioned longi tudinally. much as if the individual elements of the specimen were split into longitudinal fibers and were then drawn down to a point before rupture. it has a very fibrous appearance. ie) Stages in the formation of a cupandcone fracture. mation occurs by the sliding of grains over one another. and under continued straining these grow and coalesce into a central crack (Fig.
Fetch. 710). This suggests that the voids are nucleated ^ the ductile fracture of iron has the by dislocation pileups at grain boundaries. The occurrence 1 of this state of stress can be explained by the conof straints to plastic flow which a notch 8. the stress concen tration reduced. 1. 186. Fetch has shown that the fracture stress (corrected for necking) for same dependence on grain size as is found for brittle fracture. 71 2. 1956. E. ^ Even nucleating secondphase particles original rolling or extrusion direction of than when tested in the direction working. This is borne out by the fact that the fracture stress and reduction in area can be appreciably lower when tested in a direction perpendicular to the itself the inclusion fractures. The introduction of a notch results in a stress concentration at the root of the notch. 8. sets up. vol. Under tensile strain either the metal separates from the inclusion.. Puttick. 71 2] Fracture 21 shear fracture propagates rapidly compared with the fibrous fracture. However. it appears that fracturenucleating elements are present before deformation. The changes produced by the introduction of a notch have important The presence of a notch will very appreciably increase the temperature at which a steel changes from ductile to brittle fracture. Phil. Mag. Instead. Mag. even though all microstructural evidence of working has been removed by heat treatment and there is no strong crystallographic texture. impurity phases. It must be assumed that working elongates these "sites" so that they open into voids more readily when the tensile stress is applied perpendicular to their length. Notch Effect in Fracture implications for the fracture of metals. However. ser. or in metals for which no crackcan be observed. there is appreciable localized heating.3 Sec.. When is yielding occurs at the root of the notch. . 2 K. This stress drops from a high value at the notch root to a lower value at the specimen axis. p. it is extremely unlikely that dislocation pileups large enough to produce cavity dislocations can be produced in ductile fee metals like aluminum and copper. The transverse acts in the circumferential direction of a cylindrical specimen. 4. Figure 710 shows the nonuniform distribution of the longitudinal tensile stress in a notched tensile bar. For an equilibrium N. 1959. transverse and radial notch (Fig. ser. J. voids in these metals appear to be nucleated at foreign particles such as oxide particles. vol. Phil. 964. but the interior of the specimen and then drops stress (tt to a high value in off again. p. or secondphase particles. stresses are set <tr is up in the vicinity of the The radial stress it rises zero at the free surface at the root of the notch.
consider that yielding occurs at a critical shear stress Tc. F. forces to be maintained in the notched bar. = produce yielding. stress. necessary that no stresses act normal to the free surfaces of the notch. a relatively large mass of unstressed metal exists around a central core of highly stressed son effect. J. When plastic deformation occurs at the root of the notch. taken by the metal in the core of the notch. mass of ma hoop of unstressed material The resistance of the unstressed terial to the deformation of the central core produces radial and transverse stresses. All the tensile load must be Therefore. Orowan. plastic deformation produces plastic constraint at the root of the notch. London. the elasticstress concentration is reduced to a low value. raised over which the flow curve is raised because of the notch can be expressed q. H. The amount by by a plasticconstraint factor from elasticstress concentration in a basic way. Stress distribution Since the critical shear stress for yielding is produced in notched cylinder under uniaxial loading. or the same for both cases. .214 Metallursical Fundamentals it is [Ch ap. Nye. and W. J. In contrast to elasticstress concentrations. From elastic considerations the stress concentration at the root of a notch can be made extremely high as the radius at the root of the notch approaches zero. The existence of radial and transverse stresses (triaxial stress state) raises the value which yielding occurs. Cairns." vol. but to a it is material. The central core tries to contract laterally because of the Pois restrained by what amounts around it. 1. it is apparent from or = ra these equations that the existence of transverse stresses requires a higher longitudinal stress to transverse dial stress. "Strength and Testing of Materials. M. 710. The is entire flow curve of a notched specimen that for an unnotched specimen because of this effect. For an unnotched tension specimen this critical value is of longitudinal stress at given by Tc  2 this For a notched tension specimen becomes Fig. However. Plastic constraint differs ' E. ctl = longitudinal stress. For simplification. no matter how sharp the notch the plasticconstraint factor^ cannot exceed a value of about 3. Stationery Office. 1952.
Since stress is proportional to strain. Concept of the Fracture Curve it In earlier chapters plastic flow at was shown that the flow curve. In an analogous was proposed by Ludwik^ that a metal has a fracture stress BritfleS^—". 71 3. can be considered to represent the stress required to cause manner it any particular value of plastic strain. 713] Fracture 215 A third important contribution of a notch the local strain rate. at a point near the notch is rapidly increasing with time because of the sharp gradients. The stress picture changes from one of high elastic stresses to a lower plastic constraint.>cc. When yielding occurs. and in so doing a high plastic strain rate develops near the notch.— —— Fraclurecurje__^ . or the true stress strain curve. the plastic flow tends to relieve the stresses. there is a high local elastic strain rate. the stress is to produce an increase in While the notch is still loaded in the elastic region.
in a qualitative sense. a point on the fracture curve is obtained by plastically and then introducing embrittling parameters (low temperature or a notch) so that the specimen is stressed to failure without added strain. where there is no ductiletobrittle transition. Further. For all metals the yield With fee stress or flow stress increases with decreasing temperature. where it is useful for describing certain aspects of fracture. In principle. off rapidly over temperature range at which this The called the transition temperature. the fracture stress for ductile fracture is very difficult to measure accurately because the ductile fracture is initiated at the interior of the specimen. Classical Theory of the DuctiletoBrittle Transition Brittle fracture of stress.216 Metallurgical Fundamentals [Chap. However. Temperature has a strong effect on the basic flow and fracture properties of the metal. (2) a is promoted by three main (3) factors. the yield stress increases by a factor of 3 to 8 over the same temperature range. In the previ was shown that the presence of a notch provides condition 1 and contributes to condition 3. The separation between the two fracture curves and their relative height will be different for other conditions of fracture. the fracture curve of metals. By repeating this process with different specimens stressed to different values of plastic strain it would be possible to construct the entire fracture curve. . the increase in yield stress on going from room temperature to liquidnitrogen temperature (— 196°C) is about a factor of 2. However. this does not prohibit using the straining a specimen to a given point on the flow curve concept of the fracture stress. metals. since the embrittling effect of a notch is limited to a plasticconstraint factor of about 3. it is generally more effective to attempt to resist any further deformation by carrying out the test at a very low temperature. 714. (1) a triaxial state low temperature. It also shows that the ous section it reduction of area at fracture in a tensile specimen drops a narrow temperature transition occurs is interval. with most metals this is not possible since a slight amount of deformation invariably results on straining at low temperature. In view of the evidence that fracture is initiated by plastic deformation. Actually. and the stress distribution is complicated by necking of the Therefore. there is no reliable method for determining tensile specimen. 7 are for the ordinary tensile fracture of a ductile metal in which a shear type of fracture takes place. Figure 75 illustrates the trends in fracture stress and yield stress with temperature. and a high strain rate. In bcc metals. which show a ductiletobrittle transition. it would appear that the fracture stress measured by this technique does not measure the true resistance of the metal to fracture.
Sec. but for a notched test the transition temperature is much closer to room occurs of flow stress intersects the cleavage strength. . the curve marked cto gives the temperature dependence of yield stress ture for simple tension. The ciu've qao.^ without at the same time raising Decreasing the tempera the fracture stress will favor brittle fracture. Inst. While this picture of the ductiletobrittle transition does not provide embodied in the dislocation theory. In Fig. the existence of a transition temperature is due to the difference in the way the resistances to shear and cleavage change with temperature. and increasing the strain rate both have this effect. A For an unnotched tension specimen this occurs at a quite low temperature. Davidenkov and F. Felbeck and Orowan^ were unable to produce cleavage fracture in steel plates unless the crack reached a high velocity. 4. The curve marked aj is the fracture strength or cleavage strength as a function of temperature. The relative values of these two parameters determine whether the fracture Above the transition temperature the flow will be ductile or brittle. when a curve temperature. ^ Felbeck and Orowan. it does give an easily grasped working model of the phenomenon. Extensive plastic deformation was present at the base of the crack in all cases. Tech. is the temperature dependence of the yield stress in the presence of the plastic constraint at a notch. while below the transition temperature the fracture stress is reached first.R. Transition temperature simple tension Notcti transition temperature where q « 3.). this classical theory ascribes no major effect to the role of strain rate. Phys.S. 1937. N. Temperature—>Fig. Schematic description of tran sition temperature. cit. yet recent experiments have indicated that strain rate may be more important than plastic constraint in producing brittle fracture. In agreement with available data. 71 3.S. it is drawn as a less sensitive functransition temperature tion of temperature than the yield stress. As originally proposed. Using sharp cleavage cracks as notches. Wittman. Factors which increase the critical shear stress for slip t suggested by Davidenkov and Wittman. 714] Hracture 217 The socalled classical theory of the ductiletobrittle transition was According to this concept. 300. 713. stress is reached before the fracture stress. vol. (U. p. op. These experiments could be interpreted only by confor the structural details 1 N.
a threedimensional fracture surface in terms of the three principal stresses For any combination of principal stresses the metal will (Ti. pp. or distortionenergy.. Dorn. Enough experimentation fracture when the limiting surface is reached. Ohio. For accurate results bulging or necking during the later This makes it difficult to obtain stages of the test must be avoided. A. 12. fracture is controlled only by the magnitude of the greatest principal Available data on ductile metals such as aluminum and magstress. Metals Park. 1945. esting to note that the yield stress of mild steel However. Most experimentation in this field has been with biaxial states of stress where one of the principal stresses is zero. Figure 714 illustrates the fracture criteria which have been most frequently proposed for fracture under a biaxial state of stress. 2 E. E. is it is inter very sensitive to Also. 3 with regard to the prediction of yielding under complex states The problem of determining general laws for the fracture of stress. nesium alloys^ and steel^ indicate that the maximumshearstress criterion 1 J. and 03. 7 is raised to the value of the fracture stress." American Society for Metals. This same approach was discussed in Chap. vol. brought about by using a notchedimpact test can be understood on this basis when it is considered that the strain rate in the impact test is about 10^ times greater than in the ordinary tension test.218 Metallurgical Fundamentals [Chap. 02. A13A24. 1948. Tubular specimens in which an axial tensile or compressive load is superimposed on the circumferential stress produced by internal pressure are ordinarily used for this type of work. strength of metals is quite difficult because fracture is so sensitive to In principle we can conceive of prior plastic straining and temperature. J. Davis. . not by plastic constraint. and additional experiments would be very worthwhile. good data for very ductile metals. Appl. Mech. The maximumshearstress criterion and the Von Mises. Fracture under Combined Stresses to fracture is The phenomenological approach concerned with uncover ing the general macroscopic laws which describe the fracture of metals under all possible states of stress. 71 5. but by the effect of high strain rate on increas sidering that the jaeld stress ing the yield stress. criterion have already been considered previously in the discussion of The maximumnormalstress criterion proposes that yielding criteria. the large increase in transition temperature that is strain rate. It is difficult to separate these two effects. has been done to realize that the fracture surface cannot be rigid but must be considered as a flexible membrane which changes shape with changes in stress and strain history. "Fracturing of Metals.
iron^ shown in Fig. . 715. Note followed in the tensiontension region and that the fracture strength increases significantly as one of the princiTwo theories^ which consider the pal stresses becomes compressive.' )ec. 716] Fracture 219 for fracture results in the best agreement. Agreement between experiis ment and theory is not nearly so good The fracture criterion for a brittle cast that the normal stress criterion is as for the case of yielding criteria.
" vol. 1949. pp. C." John Wiley & Sons. G. and D.. 12. L. 5. London.. 418465.. 232.): "Fracture. Ltd. 1957. Parker. loaded with superimposed pressure to a point short of fracture and then tested at atmospheric pressure. "Fracturing of Metals. K. vol. pp. B. 1957. which are completely brittle under ordinary conditions. : vol. Felbeck. Inc. Thomas (eds. in general. the amount of deformation required to produce fracture after removal of the hydrostatic pressure increases with an increase in the magnitude ture is of the pressure. 1959. Further.." Technology Press and John Wiley & Sons. 1954. These facts indicate that. Stroh. like limestone or rock salt. actually necked down when pulled in tension with superimposed It was also found that. E. 7 Materials static pressure than when tested with simple uniaxial loading." American Society for Metals. Metals Park. 49. R. The Fracture of Metals. Repts. 1948. D. 53117.. New Patch. described his results. J. fracturing. A. if a tensile specimen was hydrostatic pressure. New York. A. 6. frac not determined completely by the instantaneous state of stress or Bridgman was able to find no simple stress function which strain. 1957. vol. N. ASM.: "Brittle Behavior of Engineering Structures.: Fracture and Strength of SoUds. in "Progress in Metal Physics. E. Campbell Memorial Lecture. Barrett. 1956. Perga^ : mon Press. Orowan. S. in Phys. Progr. York..: Advances in Phys. Hahn. BIBLIOGRAPHY Averbach. even if it required further deformation before the elongation under pressure was greater than the metal could v/ithstand when ordinarily tested at atmospheric pressure. T. Metallurgy at Low Temperatures. 185 Trans.220 Metallurgical Fundamentals [Chap. Ohio. . Inc. pp. N.
Introduction ability of a vibrating solid its which is completely isolated from its mechanical energy of vibration into heat is called internal friction. damping effects or at low stress levels it may be due to thermal. Certain static effects such as the elasticaftereffect are concerned with Internal friction resulting from cold work is strongly anelastic behavior. the fact that can be observed at stress levels far below the macroscopic elastic limit indicates that metals have a very low true elastic limit. spond to a phase lag between the applied stress and the resulting strain. there The would be no internal friction. magnetic. therefore. This may be due simply to plastic deformation at a high stress level. "Elasticity and Anelasticity. Zener. 221 1948. Chicago. Studies of internal friction are primarily concerned with using damping as a tool for studying internal structure and atom movements diffusion. effects correindeed. and magnetic interactions. if. in solids. The method has provided information on sity of dislocations. The former term is preIf ferred by physicists. or damping capacity. stress relaxation across grain boundaries." University Chicago Press. Much of our present knowledge of the mechanisms which contribute to anelasticity . and the latter is generally used in engineering. metals behaved as perfectly elastic materials at stresses below the nominal surroundings to convert elastic limit. one exists at all. amplitudedependent and. ordering. be due to thermal diffusion. important division of the field of nonelastic behavior is called This subject is concerned with internalfriction effects which Anelastic behavior can are independent of the amplitude of vibration. Internalfriction. atomic diffusion. . or atomic rearrangements. 1 and solu bilities of interstitial elements and has been used for estimating the den The vibration amplitudes employed in this type of of C. stressinduced ordering. or damping. is not an anelastic phenomenon.Chapter 8 INTERNAL FRICTION 81 . However. An anelasticity is due to Zener^ and his coworkers.
The simInternal friction is measured by a number of techniques. If the damping is amplitudedependent. The amplitude at any time. of defining internal friction or damping capacity is with the logarithmic decrement 5. or ultrasonic energy. the strain must lag behind the applied stress. a~4^ where eg' (81) ei = = nonelastic strain elastic strain in component 90° out of phase with stress phase with stress Internal friction is frequently measured by a system which is set into motion with a certain amplitude ^o and then allowed to decay freely. For higherfrequency measurements the specimen is excited by an electromagnetic drive. the decrement is given by the slope of the curve at a chosen amplitude. 8 stresses are work are usually quite aspect of this field is small. The phase angle. American Society Metals Park. Phenomenological Description of Internal Friction For energy to be dissipated by internal friction. 82. a plot of In A versus be linear and the slope of the curve is the decrement. and the very low." pp.222 Metallurgical Fundamentals [Chap. The logarithmic decrement is related to the lag angle by If the internal friction is the number of cycles of vibration will 5 = Tza (84) For a condition 1 of forced vibration in which the specimen is driven at C. a can be used as a measure of internal friction. with determining the damping capacity of a material at the relatively large amplitudes encountered in engineering practice. At. 1953. a piezoelectric crystal. can be expressed by an equation At = Aoexp(/30 The most common way (82) where /3 is the attenuation coefficient. "Modern Research Techniques for Metals. h = In 4^ (83) independent of amplitude. in Physical Metallurgy. The logarithmic decrement is the logarithm of the ratio of successive amplitudes. 225250. Ohio. Wert.^ plest device is a torsional pendulum for use in the lowfrequency region around 1 cps. or lag angle. Another the determination of engineering data on the dissi This work is usually concerned pation of energy in vibrating members. .
2 . 35. The fact that the is dynamic modulus effect. vol. is larger than the static elastic modulus of called the AE A number ' behavior of materials. 81. vol. Physik. 81. Phil. 1892. In a forcedvibration type of experiment customary to determine a resonance curve such as that of Fig. Mag. The logarithmic decrement for a resonance curve is given approximately by it is 8 = 7r(bandwidth) fr 7r(/2 — fr /i) (86) A is measure often ir/8. symbol Q^^ has been adopted as a measure of internal Q' = hh fr (87) Under conditions of cyclic excitation the dynamic elastic modulus will be greater than the static elastic modulus because of nonelastic internal friction. Vibrational energy propor tional to the square of the amplitude. while the static modulus relaxed modulus Er. The modulus under dynamic conditions is frequently termed the unrelaxed elastic modulus Eu. 671. 821 Internal Friction is 223 a constant amplitude a measure of internal friction the fractional is decrease in vibrational energy per cycle. p. p. theory the reciprocal of this value called the Q of the circuit.Sec. is called the The unrelaxed modulus 01 is given by (88) Eu ei' where ei^ is the elastic and ei^ is the plastic strain component in phase with the stress. Maxwell. 18G8. Voight. so that the logarithmic decrement can be expressed by i = AW 2W (85) where at AW is the energy lost per cycle and of W is the vibrational energy the start the cycle. C. 47. Frequency Q = is Since in electricalcircuit Resonance curve. models have been proposed to describe the nonelastic The models suggested by Voight^ and Maxwell^ W. the friction. J. of internal friction that used is the Q. where Fig.. Ann. 134.
the models do not account for the dependence of internal friction. can be duplicated by a mechanical model composed of springs (elastic component) and dashpots (viscous component). The behavior of a material. 8 Both models consider that the material has component coupled with a viscous component. Further. but they are of limited usefulness in dealing with metals. together with the equations which the models predict. Figure 82 illustrates the composition of a Voight and Maxwell solid. Various modifications of the models have been useful in studying the mechanical properties of polymers. which is observed with real metals. elastic an models. Voight solid dynamic modulus on Maxwell solid cr= F^e + 7j€ . For real metals the frequency dependence of internal friction does not agree with the equations predicted by the are frequently mentioned.224 Metallurgical Fundamentals [Chap. with the properties attributed to it by the theory.
metals at room temperature. (T + T. 83] in Internal Friction 225 many higher temperatures. approaching its amount which remains and slowly original value. 84. aid \ aib = hie { 626 (89) A material which obeys this type Time known as a standard The mechanical model for this material is shown in Fig.Sec. 84. The general equation terms of three independent constants. ship stress is A realistic relation obtained by equating the its first and derivative with respect to time to the strain and Time the strain rate. This time dependence of strain on loading and unloading has been called the elastic aftereffect. Note that the time dependof equation is linear solid. — = Eu there Ej (812) VEuEr For a standard linear solid is only a single relaxation time .& = ER(e + T^e) (810) where = = Er = t^ Ta time of relaxation of stress for constant strain time of relaxation of strain for constant stress relaxed elastic modulus The relationship between the relaxation times is and the relaxed and unrelaxed modulus given by Er Eu (811) A dimensionless combination of elastic constants. it is In considering the stressstrain relationship for an anelastic material apparent that a constant linear relationship between these two fac tors will not adequately describe the situation. is a measure of the total relaxation E. ence of strain closely duplicates the for a standard linear solid can be rewritten in behavior of a material with an elastic aftereffect. called the relaxation strength. although the effect is greater at When the stress is removed. Mechanical model of standard linear solid and associated time dependFig. ence of stress and strain. the strain will decrease but there will be a certain decreases with time.
usually easier to deter constant and varying the In many materials. [Chap.. fusion of interstitial (814)] the activation energy Aiif can be determined. measurements are well suited for studying the difatoms in bcc metals. for a material which behaves like an anelastic standard linear solid^ an internalfriction peak will occur at an angular frequency which is the reciprocal of the relaxation time of the process causing the relaxation. For a given relaxation time the diffusion atoms is given by coefficient of the interstitial D = where is ^ dor (816) ao is the interatomic spacing. The temperature dependence of D given by D = 1 D. Equation (813) is symmetrical in both co and r and has a maximum when cor = 1. The lag angle.226 T Metallurgical Fundamentals (r. 8 is = + T<. given by the following equation:^ a = E. S. Relaxation peaks arise owing to diffusion of interstitial atoms to minimumenergy positions in the For a given frequency the relaxation stress fields of the dislocations. 2'' . From the temperature dependence of relaxation time [Eq. pp. time is expressed by r = 1/co. exp  1^ (817) A. Ltd. . mine the relaxation spectrum by holding relaxation time r. Pergamon Press. 2 2 (813) where co = 2x/is the angular frequency of vibration. Internal Friction in Metals. It is often difficult experimentally to vary the angular frequency by a it is co factor much greater than 100." vol. 1516. 4. Nowick. London.)/2. in "ProgjRss in Metal Physics. to determine the relaxation spectrum. 1953. Therefore. including metals. and the peak occurs at a temperature Ti. r varies expo nentially with temperature so that r = TO exp ^ all (814) Therefore. Internalfriction that is necessary is to measure a as a function of temperature for constant angular frequency. on the basis of this model. Therefore. At another value of frequency the relaxation peak will occur at a temperature To..
Sec. vol. Worrell. and all internalfriction effects were independent of amplitude. p. T. This variation of internal friction with frequency can be considered as a relaxation spectrum which is characteristic of a particular material. 87. 1947. This type of internal friction is considered in Sec. 1951. Work in this area has led to the conclusion that grain boundaries behave in some ways like a viscous material. Phys. vol. Ke found that a broad peak occurred in the region of 300°C in polycrystalline aluminum. 2 T.. lowenergy twin boundaries due to stress is believed to produce relaxation effects. 533. Appl. Phys. p. while no internalfriction 1 F.. 41. 1948. At the low torsional strains used in this work the was completely recoverable.' This type of deformation is also responsifriction is discussed in The movement of ble for anelastic effects found in movement in ferromagnetic materials. vol. p. 71. p. Relaxation produced by thermal fluctuations will be considered in Sec. the internal friction cannot be due to the viscous slip associated with incoherent boundaries. vol. is quency investigated. 1257. 85] Internal Friction 227 84. the behavior of the metal in the region of the The stress application of stress to a substitutional solid solution can produce ordering in an otherwise random distribution of atoms. S. Ke. Studies of this relaxation process have provided data on the solubility and diffusion of interstitial atoms. 929. 22. (810) with suitably determined constants. An alternating between pairs of solute atoms. Rev. conjunction with domainboundary Since twin interfaces are crys tallographically coherent boundaries. The relaxation peak due to preferential ordering of interstitial atoms one of the bestunderstood relaxation processes. 19. Ke. Grainboundary Relaxation An important source of internal friction in metals is stress relaxation along grain boundaries. This interesting aspect of internal rise to relaxation can give more detail in the next section.first demonstrated the strong internalfriction purity strain peak due to grainboundary relaxation by experiments on highaluminum wires. Relaxation Spectrum A number occur in metals. rated. so that of internalfriction peaks can be found if a wide range of freof relaxation processes Each will Provided that the peaks are sufficiently sepapeak can be expressed by Eq. stress is in the lattice from an applied 85. a number with different relaxation times can occur in a different frequency region. 86. A large and broad internalfriction peak is produced in polycrystalline specimens by the relaxation of shear stress across grain boundaries. 72. J. .
02 . 8 peak was observed in aluminum single crystals (Fig. 85). U.06 §0.04 CZ 0.228 Metallurgical Fundamentals [Chap. measurements of the modulus (which is proportional to the square of frequency) at different temperatures showed a sharp drop for the polycrystalline specimen which was not found with the singlecrystal specimen This behavior is consistent with the assumption that grain (Fig.08 Or i 0. In addition.I u 0. 86). boundaries behave to a certain extent in a viscous manner at elevated temperatures.
No energy loss or damping occurs under adiabatic conditions. When the stress is removed. When the stress fluctuations occur at a very high frequency. In the region of intermediate frequencies the conversion of energy into heat is Nonuniform 1 not reversible. 8. 1941. p. if a stress is applied along the y axis. 1939. perature. so that there is not time for appreciable heat flow to take place during a stress cycle. However. stress can result in macroscopic thermal currents which 6. Thermoelastic Internal Friction The thermal and the mechanical behavior of materials are interrelated. 1942. the crystal will have tetragonal symmetry because of the distortion produced by the interstitial atoms. 87. As long as no stress is applied. 9. while a contraction produces a temperature is called the thermoelastic effect. either tending toward or tending away from a preferred orientation. and no energy or heat is lost. Even though no external forces are applied. 87J Internal Friction 229 86. This is an isothermal process. On the other hand. vol. and internalfriction effects are observed. 711. vol. J. A similar but weaker relaxation peak can be observed due to shortrange order in substitutional solid solutions. Snoek. a fluctuating temperature gradient is produced. An extension of the specimen will result in a decrease in temrise. the atoms will migrate toward a random distribution. the interstitial atoms will migrate to octahedral positions which tend to give a preferred alignment in the y direction. at very low frequencies there is adequate time for heat flow. Under the oscillating stresses imposed by an internalfriction apparatus the interstitial atoms will be in continuous motion. and this strain will be accompanied by a small change in temperature. the distribution of atoms among the octahedral sites is random and the tetragonal axes of the unit cells are randomly oriented with respect to the specimen axes. 862. The application of a small stress to a metal will produce an instantaneous strain. Physica. studied in iron containing small amounts of either carbon or nitrogen in solid solution. and an equilibrium temperature is maintained in the specimen. A strong relaxation peak results. If the applied stress is This behavior not uniform throughout the specimen. If the nonuniform stress varies periodically with time. a temperature gradient will be set up and additional nonelastic strain will result. the process is adiabatic. 591.Sec. The Snock Effect Internal friction resulting from preferential ordering of interstitial atoms under applied stress was first explained by Snoek' and is known as This type of relaxation has been most extensively the Snoek effect. vol. p. Interstitial carbon atoms in bcc iron occupy the octahedral holes in the lattice. . p.
vol. the peak would occur in the region of 1 to 100 cps. For specimens of ordinary thickness. A poly crystalline specimen which is subjected to completely uniform stress can show relaxation due to intergranular thermal currents arising = = = thickness of specimen from the fluctuations in stress from grain to grain. 1937. 88. p. Zener. provided that the frequency is is such that there enough time for thermal currents to flow back and forth Zener^ and effect a partial neutralization of the temperature gradient. It is theoretically possible for a specimen vibrated in a longitudinal mode to show relaxation from macroscopic thermal currents. macroscopic thermal currents occurs in a specimen subjected to torsional vibration.. therefore. and Therefore. the method by which the deformation was intro1 C. an the tension side will undergo a decrease in temperature. A relaxation process occurs. The frequency at which the relaxation will occur is related to the grain size of the metal. The effects are very complex and depend on variables such as the amount of plastic deformation. The localized stress differences from grain to grain are due to the elastic anisotropy of individual grains. However. It is important that effects from this source be considered in experiments where the prime interest is in damping from other sources. Dislocation Damping The internal friction of metals is quite sensitive to plastic deformation. . A rectangular bar which will increase in is vibrated transversely behaves like a standard linear solid (single relaxation time). Rev. represent a single relaxation time. Phys. 8 produce internalfriction peaks. The compression side of the specimen temperature. alternating temperature gradient is produced across the thickness of the bar.230 Metallurgical Fundamentals [Chap. has shown that the relaxation time is irWt (818) where h Dt thermaldiffusion constant thermal conductivity/ (specific heat) (density) The frequency at which this relaxation peak occurs can be determined from the relationship w^ = 1. 230. which No relaxation from is well beyond the range of normal observations. The relaxation peak due to intergranular thermal currents will not occur at a sharp frequency and. because shearing stresses are not accompanied by a change in temperature. the frequency region where the peak would occur would be of the order 10^" to 10^^ cps. 52. Internal friction due to inter granular thermal currents can occur for all types of stressing.
9. F. New York. 144. 30. 1941. Granato and K. & Sons. Bordoni. Read^ demonstrated that internal friction arising from cold work is strongly amplitudedependent.^ modulus defect which is based on the bowing out under stress of a network of dislocation lines anchored at nodes and impurities. Z. 1 T. S. Appl. 27. AIME. 32. Metallk. 88] Internal Friction 231 duced into the metal. vol. However. vol." John Wiley Inc. the frequency of vibration.Sec. 583. p. mechanism is for the internalfriction effects observed in The theory due to Koehler* and Granato and Liicke^ assumes that amplitudedependent internal friction is due to a stressstrain hysteresis arising from the irreversibility of dislocation lines breaking away from pinning impurity atoms. Mott. 1953. Trans. Phys. G. the dynamic modulus returns to its steadystate value. in agreement with observed results. . vol. 1940.. Mag. 5 A. or the Koster effect. amplitudeindependent internal friction is assumed to result from a viscouslike damping force acting on the bowedout segments of the discoldworked metals not well established. Koehler. p. 143. 282. 2.. the purity of the metal. is tively low temperatures called the modulus defect. ® P. 1950 ^ J. The decrease in modulus due to cold work which can be eliminated by annealing at relathe dynamic modulus. Koster. Read. 43. vol. 7. p. A freshly coldworked metal has a relatively high internal friction recrystallization. The only relaxation process which gives an internal peak that is is definitely ascribable to dislocations the Bordoni peak^ found in fee metals at very low temperatures in the region of 30 to 100°K. Phil. 1956. vol. 1952. even for strain amplitudes as small as 10^^. N. A.. Nuovo cimento. 2 3 W. p. ^ The dislocation cc NU A^" (819) In a coldworked metal typical values of ^^ 10' and L c^ 10~^ cm would lead to values of I\E/E of 10 per cent. 1151. Lucke. in which anneals out very rapidly at temperatures well below those required for The high damping is also accompanied by a decrease As the internal friction anneals out.. ser. The theory predicts that the modulus dislocation for the Mott has proposed a modeP defect is proportional to the products of the dislocation length per cubic centimeter and the square of the effective loop length of a dislocation segment. and the time between the deformation and the measurement of internal friction. J. location lines. suppl. p. "Imperfections in Nearly Perfect Crystals. are indications that the Bordoni peak of dislocations is There due to some intrinsic property and is not involved with the interaction of dislocations with impurity atoms and other dislocations.
with the exponent varying between 2 and energy. The damping capacity Turbine blades. strongly dependent on the stress or strain amplitude. and aircraft propellers are typical applicawhere damping capacity is important. June. Metallurgical Fundamentals [Chap. Specific damping energy is approximately a power function of stress level. engineering damping properties of materials hy the specific damping This quantity represents the area inside a stressstrain hysteresis loop under uniform stress conditions and is a true material property. (83) and Ai — amplitude of vibration of first cycle An = amplitude of vibration of nth cycle N = number of cycles from Ai to An 6 Values of these damping parameters depend not only on the condition of the material but also on the shape and stress distribution of the specimens. Rept. . available from Office of Technical Services. J. Methods of converting logarithmic decrement and damping capacity to specific damping energy have been published. crankshafts. Lazan. 56180. 5644. They are. Tech. Rept. however. Tech. Since these conditions are often not specified. 8 Dampins Capacity is concerned with the engineering aspects of internal fricof structures and machine elements is concerned with the internal friction of materials at strain amplitudes and stresses which are much greater than the values usually considered in A high damping capacity is of practical internalfriction experiments. = I^L^ (85)] (820) = specific damping capacity = logarithmic decrement [see Eqs. Podnieks and B. June.^ Engineering dampingcapacity measurements are not very dependent upon frequency of vibration. J. This section tion. R. 1956. The damping properties of materials are frequently expressed in 6 terms of the logarithmic decrement or the specific damping capacity \l/. Demer. WADC WADC 1957. ^ = where \l/ 2. Damping capacity can be defined as the amount of work dissipated into heat per unit volume of material per cycle of completely reversed tions stress. ' L. Analytical Methods for Determining Specific Damping Energy Considering Stress Distribution. there is considerable variation and contradiction in the published literature^ on the dampThe proposal has been made to express the ing properties of materials. engineering importance in limiting the amplitude of vibration at resonance conditions and thereby reducing the likelihood of fatigue failure. Bibliography of the Material Damping Field. 2 E.232 89.
12% Cr stainless steel — 188 stainless steel Cast iron Yellow brass t S.86 4. ed. that high damping capacity correlates with a low notch sensitivity. L.Sec.0 1. While in certain cases it appears Table 81 Damping Capacity of Some Engineering MATERiALsf Specific damping capacity APF/PT 11..500 psi 6.200 psi at various stress levels Material 4.16 40.38 8.49 8. 89] Internal Friction 233 3 for most materials." rev. 2.0 0. This is attributed to energy losses in the graphite flakes. 1952.700 psi Carbon steel (0. the magnitude of the effect The damping capacity for a given metal and test condition depends on the type of stress system.8 Hoyt.78 0. The damping behavior is a function of the number Generally. One important contribution to damping by the in many alloys used for turbineblade applications comes from This has been demonstrated^ which showed high damping had much decreased damping capacity when tested in a magnetic field.28 0. Trans. distribution produced by different methods. The lower damping in the magnetic field can be attributed to the fact that the domains are lined up in the direction of the field and cannot move freely under stress. AIME. Cochardt. the motion of ferromagnetic domain fact that a ferromagnetic alloy 1 A. there limit. pp. walls. i. This is the result of differences in stress increasing with stress level. A number of attempts have been made to relate damping behavior with other properties such as fatigue strength and notch sensitivity. . 206. whether tested in torsion or tension.0 0. of reversed stress cycles. W. Furthermore.76 28.0 0. Cast iron has one of the highest damping capacities of these materials. is no general relationship between damping capacity and fatigue Table 81 lists some values of damping capacity for a number of engineering materials at several stress levels. vol. Reinhold Publishing Corporation. New York. "Metal Data.. 1956. there is no general relationship between these properties.50 2.16 0. 12951298.0 3. the damping capacity increases with number of cycles of stress reversal.e.1 % C) NiCr steel quenched and tempered.70 8.
in "Progress in Metal Physics. Niblett." vol. S. of Metals. 4.. (eds. Nowick. Wilks: Dislocation Damping in Metals Advances inPhys. 1953. H. K. The Damping Capacity of Metals.. Edward Arnold & : Co. and J. Zener. A. .234 Metallurgical Fundamentals [Chap. 1960. 9.). M. 1948. London. Ltd." University of Chicago Press. D. 188.: Internal Friction in Metals. Quarrell "The Physical Examination of Metals.. vol. 8 BIBLIOGRAPHY Entwistle." 2d ed. 1960. Pergamon London. Press. G.. Chalmers and A.. in B. C: "Elasticity and Anelasticity Chicago. pp.
Part Three APPLICATIONS TO MATERIALS TESTING .
.
obtained by dividing the load by the original area of the cross section of the specimen. while the appearance of a yield point in the stressstrain curve was covered in Sec. the tension test' In is a specimen subjected to a continually increasing uniaxial tensile force while si multaneous observations are made of the elongation of the specimen. An is engineering stressstrain curve constructed from the loadelonga tion measurements (Fig. longitudinal in specimen.Chapter 9 THE TENSION TEST 91 . Troxell. 91 ) .. New York 1955." 2d ed. 24. P_ (91) The 1 strain used for the engineering stressstrain curve is the average H. 237 . The engineering stressstrain curve stress It is is the average the tensile curve. T. Inc. McGrawHill Book Company. The stress used in this significant points stressstrain Conventional strain e Fig. Engineerins StressStrain Curve The engineering tension test is widely used to provide basic design information on the strength of materials and as an acceptance test for the specification of materials. "The Testing and Inspection of Engineering Materials. G. E. Wiskocil. and C. 91. E. chaps. 15. 55. Davis. The on the engineering stressstrain curve have already been considered in Sec..
and reduction of area. au = ^ (93) The tensile strength is the value most often quoted from the results of a it is a value of little fundamental significance with regard to the strength of a metal. Tensile Strength r The tensile strength. The parameters which are used to describe the stressstrain curve of a metal are the tensile strength. and as such it is a very useful identification of a material in the same sense that Further. and state of stress imposed during the testing. the last two indicate ductility. length of the specimen. The two curves frequently used interchangeably. yet in reality tensile strength and properties such as hardness and fatigue strength are . it is useful for the purposes of specifications and for quality control of a product. which is obtained by dividing the elongation of the gage 8. suitably reduced by a factor of safety. The first two are strength parameters. heat treatment. For many years it was customary to base the strength of members on the tensile strength.238 Applications to Materials Testing [Chap. or ultimate tensile strength (UTS). It will be shown that this value bears little relation to the useful strength of the metal under the more complex conditions of stress which are usually encountered. by its original length. The current trend is to the more rational approach of basing the static design of ductile metals on the yield strength. is the maxi mum load divided by the original crosssectional area of the specimen. Since both the stress and the strain are obtained by dividing the load and have the are elongation by constant factors. However. the loadelongation curve will same shape as the engineering stressstrain curve. because of the long practice of using the tensile strength to determine the strength of materials. 9 linear strain. because the tensile strength is easy to determine and is a quite reproducible property. yield strength or yield point. the chemical composition serves to identify a metal or alloy. it has become a very familiar property. Extensive empirical correlations between tension test. The shape and magnitude of the stressstrain curve of a metal will depend on its composition. prior history of plastic deformation. For ductile metals the tensile strength should be regarded as a measure of the maximum load which a metal can withstand under the very restrictive conditions of uniaxial loading. per cent elongation. temperature. and the strain rate.
. ASME. 10311064. . The elongation of the specimen is uniform along the gage length up to the point of maximum load. the yield strength and yield point are the preferred engineering parameters for expressing the start of plastic deformation. (95) is simply the total measured elongation of the specimen. When the design of a ductile metal requires that plastic deformation be prevented. H. the yield strength strength of the metal.Sec. the tensile strength is a valid Yield Strength The yield strength is the load corresponding to a small specified plastic strain divided by the original crosssectional area of the specimen. (94) Because of the practical difficulties of measuring the elastic limit or pro portional limit. the determination of the elasticbreakdown pressure of thickwall tubes subjected to internal pressure from the results of a of this An example tension test. 34). vol. c^o = — ^6=0. This uniform strain is of more fundamental importance than total strain to fracture. For brittle materials.p. is the ratio of the increase in the length of its the gage section of the specimen to original length. expressed in per % elongation where L/ Lo = ^ = e/ (95) gage length at fracture original gage length conventional strain at fracture €f The numerator in Eq.002 2 . in heat treatment and method of Percentage Elongation The percentage elongation cent. and the deformation is no longer uniform along the length of the specimen.. more complex conditions of stress An by means of the distortionenergy yielding criterion is (Sec. This value is influenced by the deformation during the necking of the specimen. Faupel. and it is also of 1 J.^ The yield strength and yield point are more sensitive than the tensile strength to differences testing. and hence the value of per cent elongation = = = depends somewhat on the specimen gage length. 78. Beyond this point necking begins. 91] The Tension Test 239 often quite useful. pp. criterion for design. Trans. is the appropriate criterion of the important feature of the yield strength is that the value determined from the tension test can be used to predict the conditions for static yielding under other. 1956.
240 Applications to Materials Testing practical use in predicting the formability of sheet metal. . the uniform elongation not usually determined in a routine ten sion test. a qualitative indication of formability of a metal can sometimes be obtained from these values.4). is a measure of the stiffness of the material. of the reduction of area in thin sheet specimens is and for this reason it is usually not measured in this type of specimen. A high reduction of area indicates the ability of the metal to deform extensively without fracture (see Prob. of Elasticity is Modulus The slope of the initial linear portion of the stressstrain curve the modulus of elasticity. so that. flat. the area after fracture may be approximated by A = ^{a^2d) where h. 9. the percentage elongation ifi always based on the total elongation. Reduction of Area The percentage reduction expressed in per cent. is [Chap. However. Therefore. unless specifically stated. 9 some How ever. rectangular tensile specimens. Reduction of area = q — An — ^ At (96) The determination difficult. (97) a d = = = width of specimen thickness at center of specimen thickness at ends of cross section of specimen of area are usually not directly useful to The elongation and reduction the designer. or Young's modulus. A decrease in reduction of area from a specified level for which experience has shown that good service performance will result is a warning that quality is substandard. There appear to be no quantitative methods for determining the minimum elongation or reduction of area which a material must have for a particular design application. For thicker. percentage elongation values. It is determined by putting the broken tensile specimen together and measuring the change in gage The original gage length should always be given in reporting length. of area is the ratio of the decrease in the cross sectional area of the tensile specimen after fracture to the original area. The reduction of area is the most structuresensitive parameter that is measured in the tension test. its most important aspect is that it is used as an indication of material quality. The modulus of elasticity The greater the modulus.
Sec. tions of Since the modulus of elasticity is needed for computing deflecbeams and other members. The modulus Table 91 Typical Values of Modulus of Elasticity at Different Temperatures Material .^ However. one of only slightly affected by alloying additions. The modulus of elasticity is determined by the binding forces between Since these forces cannot be changed without changing the basic it atoms. it is an important design value. the most structureinsensitive of the mechanical properties. of elasticity is nature of the material. follows that the modulus It is is usually measured at elevated temperatures by a dynamic method^ which measures the mode and period of vibration of a metal specimen. increasing the temperature decreases the modulus of elasticity. Typical values of the modulus of elasticity for common engineering metals at different temperatures are given in Table 91. 91] TheT ension Test 241 the smaller the elastic strain resulting from the application of a given stress. heat treatment. or cold work.
9 low modulus of Table 92 gives some values of modulus of resilience for different materials.242 Applications to Materials Testing elasticity. [Chap. Table 92 Modulus of Resilience for Various Materials .
Also. and this produces the fallofT in the stressstrain curve beyond the point of maximum load. of the Because the crosssectional area specimen is decreasing rapidly at deformation falls off.Sec. The true stress is the load at any instant divided by the crosssectional area of the specimen at that instant. if the load is removed at this point and then reapplied. If the strain measurement is also based on instantaneous measurements. the area under the curve can be approximated by either of the following equations: or ~ GuCf Ut ^ —2 Ut brittle materials the stressstrain (99) ^f (910) For curve is is sometimes assumed to be a parabola. 0 = ^ (912) .3V„e/ All these relations are only approximations to the area strain curves. ductile metal which is pulled in and necks down during the course of the test. and the area under the curve given by (911) C/r~2. based on original area likewise decreases. and these dimensions change continuously during the tension becomes unstable test. the metal continues to strainharden all the way up to fracture. since they are based on the original area of the specimen. 92J The Tension Test 243 stressstrain curve like that of the structural steel. by the amount shown on the Thus. If the true stress. the load required to continue The average stress curve (Sec. it represents the basic plasticflow characteristics Any point on the flow curve can be considered as the yield stress for a metal strained in tension curve. the material will behave elastically throughout the entire range of reloading. the curves do not represent the true behavior in all the plastic range. This is also known as a flow this stage in the test. so that the stress required to produce further deformation should also increase. 92. under the stressFurther. TruestressTruestrain Curve The engineering stressstrain curve does not give a true indication of it is the deformation characteristics of a metal because based entirely on the original dimensions of the specimen. is used. the curve which is obtained is known as a truestresstruestrain curve. it is found that the stressstrain curve increases continuously up to fracture. based on the actual crosssectional area of the specimen. Actually. 32) since of the material.
33 as e = In ^^ = Lo m— A 7 (913) This definition of strain was proposed by Ludwik^ near the beginning of 140 130 120 110 100 90 'in o80 o o .70 cn tn ?i 60  50 40 30 20 10 .244 True Applications to Materials Testing strain [Chap. 9 was defined in Sec.
True Stress at Maximum Load maximum will The true stress at load corresponds to the true tensile strength. the true stresspressed the elastic region into the y axis. by the constancyof volume relationship.. strain axis. the Thereshape of the flow curve from maximum load to fracture depends on the rate of development of the neck. while Am represents the cross good approximation. should be attached to this linear region of the flow curve. and therefore there is no assurance that the flow curve will be linear in this region. strain curve is linear from the maximum load to fracture. 7~="r = ^+l cr L = £ (e + 1) (915) Figure 93 compares the true stressstrain curve for a nickel specimen with its engineering stressstrain curve. necking sectional area of the specimen at maximum — ^ load. Then. Ai Lo From Eq. fore. has com Frequently. As a occur at a value of strain where the Let am and e„ denote the true stress equals the slope of the flow curve. The following parameters are usually determined from the true stressstrain curve. When necking occurs. true stress and strain at maximum load. 92] The Tension Test 245 But. while in other Little significance cases its slope continuously decreases up to fracture. (914) e = In ^ L = In (e + 1) or Ao .j Sec. Note that the large scale on the which was used to emphasize the plastic region. tensile strength to the true stress . the triaxial state of stress that is created in this region increases the average longitudinal stress needed to continue plastic flow. This can be different for materials with different strainhardening behavior. For most materials necking begins at maximum load. O^u — = * max — Ao (Tm P Cm P max iu —A — i /im m r— Am ^^0 An (916) and o„ exp (€„) Equation (916) relates the ultimate and strain at maximum load.
Since the data required for this correction are often not available. True Fracture Strain The true fracture strain e/ is the true Ao and the area after fracture. stress values are frequently in error.Since Eq. e/ strain based on the original area = ln^^ (917) This parameter represents the maximum is true strain that the material analogous to the total strain to fracture of the engineering stressstrain curve. It may be calculated from either the specimen crosssectional area Au or the gage length L„ at maximum load. (914) is not valid beyond the onset of necking. Af. it is not possible to calculate e/ from measured values of e/. (920) The usual method of determining a true stressstrain curve is to meas . q. truefracturetrue fracture stress is The sectional area at fracture.246 Applications to Materials Testing [Chap. The uniform strain is often useful in estimating the formability of metals from the results of a tension test. However. is related to the true fracture strain by the relationship can withstand before fracture and 5=1True Uniform Strain exp (e/) (918) The up to true uniform strain e„ is the true strain based only on the strain maximum load. en = In 4^ . tion (914) e„ may = In ^ (919) True Local Necking Strain The local necking strain e„ is the strain required to deform the speci men from maximum load to fracture. Equa be used to convert conventional uniform strain to true uniform strain. for cylindrical tensile specimens the reduction of area. . 9 True Fracture Stress the load at fracture divided by the crossThis stress should be corrected for the triaxial state of stress existing in the tensile specimen at fracture.
but beyond equations implies that the axial strain this stage the major portion of the strain is localized at the neck and the equations do not apply. MacGregor. W. = = = 21„g! '^T '"t^ (921). Appl. be obtained at high strain rates and at Diameters at various positions along tapered specimens are measured before and after The true stress acting at each location on the specimen is the testing. ' = '4. Care should fracture. (914) and (915). the true strain may be readily calculated from the original diameter Do and the instantaheous diameter A. 32. The simplest power curve described earlier in Sec.0. the profile of the contour of the neck. 94). 1939. the curve can be extended to the fracture stress by linear extrapolation. vol. stress True stress and strain may also be determined from the conventional and strain by means of Eqs. coefficient and order to 1 make the data fall closer to a straight line. (7 the = Xe" strain (922) where n is the strainhardening this A loglog plot of true stress if straight line linear slope of this and K is the strength coefficient. Stresses and strains determined from these equations are accurate up to the beginning of necking. it is usually desirable C. including the region after necking has occurred. 92] The Tension Test 247 ure the crosssectional area of the specimen at increments of load up to Micrometers or special dial gages can be used. . and K is the true stress at e = 1. precisely for the However. limited to fairly slow rates of strain If necessary to know This method of determination is it and to tests at room temperature. It is usually desirable to be able to express the true stressstrain curve useful expression is by a mathematical relationship. J. up to maximum load will give a equation is satisfied by the data (Fig. Mech. A156158. lb correct is complex stresses at the neck. 6.. This gives the flow curve from the state of yielding to the point of maximum True stressstrain curves may elevated temperatures by using the twoload method. since their derivation is based on the constancyofvolume relationship. If the load at fracture is measured. The use of these is uniformly distributed over the gage length of the specimen. maximum load divided by the area at that point after testing.^ load. be taken to measure the minimum diameter of the specimen. the specimen has a circular cross section.Sec. pp. This method is applicable over the complete range of "^treag to fracture. The In line is n.
so that frequent deviations from this relationship are observed. to subtract the elastic strain from the total strain. (922)."T K and K are listed in Table 93. Equation (923) is typical of the more complicated relationships which have been suggested^ to provide better agreement with the data. 94. There is nothing basic about Eq. Some typical values of n . loglog plot of true stress while in other cases a curve with con strain curve. tinuously changing slope is obtained. is One for common 0.248 Applications to Materials Tcsti.001 0.01 0. to result in two straight lines with different slopes.1 True strain f type of deviation (922) a loglog plot of Eq. = Ci  (Ci  (72) exp (A) (923) Table 93 Values of n and Metal K for Metals at Room Temperature . [Ch ap. Fig.
gradual decrease in the crosssectional area of the specimen as it elongates. where the increase in stress due to decrease in the crosssectional area of the specimen becomes greater than the increase in the loadcarrying ability . 93] TheT ension Test 249 of strain hardening. which tends to increase the loadcarrying capacity This effect is opposed by the the specimen as deformation increases. Necking or localized deformation begins at maximum load.Sec.
Plastic instability is often important in forming operations with sheet metal since the strain at which the deformation becomes localized constitutes the forming limit of the metal. and a sheet subjected to a hydrostatic bulge test. + If a„ (Tm _ A _ Ao Lq L From the definition of conventional linear strain. 9 strain curve strain. for according to Eq. plotted in terms of true stress against conventional linear Let point . is the flow curve for a material given by the power law of Eq. 171. a thin. 562573. Sec. T. 95 shows that Eq. possible readily to determine the strain at which necking occurs.— 250 Applications to Materials Testing is [Chap. Lankford and SaibeP have determined the criteria for localized deformation for the case of a sheet subjected to biaxial tensile forces (stretching). e). 1 of a neck in the tensile triaxial state of stress in that region. (926) is satis fied If it is when OD is the tensile strength. (922). pp. the tensile The stress at this point is the true stress at maxiwe had plotted average stress. (T. Lankford and E. As was discussed in W.wall tube subjected to internal pressure and axial loading. vol.0. (925) the slope at this a /{I mum load. Stress Distribution at the Neck specimen introduces a complex The necked region is in effect a 712. Trans. . A line drawn from point the point of point is which maximum tangent to the stressstrain curve will establish load. a = = /ve" 0 ^ de 6„ = Ae" = nKe^~^ (927) = n is Therefore. AIME.4 A represent a negative strain of is 1. a notch under tension pro The formation mild notch.„. Lo 1 L so that a„ 1 +e (926) = am :7—7 A study of the similar triangles in Fig. Saibel. the strain at which necking occurs numerically equal to the strainhardening coefficient. this would have been The relation between these two stresses is strength a„. 94. 1947.
Geometry of necked region. higher o> than the stress which would be required to cause flow if simple tension prevailed. This analysis was based on the following assumptions 1 The contour of the neck is approximated by the arc of a circle. 4. *ar <^t a tapered. 3. Figure 96 illustrates the geometry at the necked region and the stresses developed by this localized deformation. Table 94 Correction Factors to Be Applied to Average True Stress to Compensate for Transverse Stresses at Neck of Tensile Specimen a/R . the average true stress is at the neck. 94] The Tension Test 251 stress required to cause plastic flow. The cross section of the necked region remains circular throughout the test. The Von Mises criterion for yielding applies. duces radial and transverse stresses which raise the value of longitudinal Therefore. conical radius Fig. 2.. which determined by dividing the axial tensile load by the area is minimum crosssectional of the specimen at the neck. 96. : )ec. analysis which provides a correction to the average axial stress to compensate for the introduction of transverse stresses. which can be measured either by projecting the contour of the necked region on a screen or by using gage. (a) Bridgman' made a mathematical (6) stresses acting on element at point 0. The strains are constant over the cross section of the neck. R is the radius of curvature of the neck.
In order to help with this situation. pirical Experimental values^ for copper and steel are also included in this figure.6. ASM. 46. 716. Their given by 1 1 a + a/4R (929) These two equations differ by less than 1 per cent for values of a/R less than 0. Trans. C. relationship converted into the variation of a/a^y with true strain. N.. Aronofsky. p. Marshall correction factor a/aav strain. 1147. pp. 1952. Appl. This investigation showed that Bridgman's equation provides better agreement with experiment than Davidenkov's. p. vol. I. Spiridonova. of the radius are given in Table The determination of curvature of the neck during the is progress of the test certainly not a routine or easy operation. 44. R. Trans. E. Marshall and J. M. 1952. Davidenkov and N. 1951. Relationship between Bridgman (E. based on about 50 steel Figure 97 shows this specimens. Shaw. The dashed curve in Fig. ASTM. 1946. Tj^pical values for these corrections 94. vol. Mech. Shaw. R. 93 is the true stressstrain curve of nickel adjusted for necking by means of Bridgman's correction factor. 705725.) relationship between the neck contour (a/R) and the true strain. The problem of the stress distribution at the neck of flat tensile specimens has been considered by Aronofsky. 44.^ 95. . 7584. pp. vol. ASM. particularly in metals which show pronounced necking before 1 2 3 N. J. and true tensile and M. o 9 According to this analysis the ratio of the true axial stress average axial stress oav is to the (7av (1 + 2R/a)\ln (1 + a/2R)] (928) Davidenkov and Spiridonova^ determined a correction for necking based on somewhat different assumptions from those expression is of Bridgman. Strain Distribution in the Tensile Specimen The strain distribution along the length of a tensile specimen is not uniform. Bridgman determined an em Fig. 18. Proc. 97.252 Applications to Materials Testing [Chap. vol. C.
The British standards specify 10.e.51 y/A.. and the tion of strain will shape of the cross section of the In general. i. 95] fheT ension Figure 98 shows in a schematic Test 253 fracture. the specimens must be geometrically In similar.505in. the softer test section. Schematic drawing of variation Therefore. or L = 4. for a given material. the greater of local elongation with position along gage length of tensile specimen. diameter has specimen a the standard tensile the United States This is the and a 2in. L/D = 3. It is for this reason that the specimen gage length should always be reported with the percentage elongation. basis for the dimensions of the ASTM tensile specimens listed in Table ~ 95. 0. L/D 4. while the German standards use L/D = Table Dimensions of Diam. Gage length total elongation of the gage length. 98. the gage length. the percentage elongation. the ratio of gage length to diameter must be constant. Also.54. and more greater the ductile the metal.Sec. the shorter the gage length. way the distribution of local elongation along the length of a tensile specimen. The exact distribu depend upon the metal. the away of deformation necked region. gage length. . Thus. amount the from the greater the influence of localized deformation at the neck on the Fig. 95 ASTM Tensile Specimens in. It is generally recognized that in order to compare elongation measurements of differentsized specimens. the shorter the gage length.
9 The uniform up to elongation is not affected by specimen geometry. since maximum load the specimen elongates and contracts in diameter The specimen changes from a cylinder of a certain length and diameter to a cylinder of longer length and smaller diameter.254 Applications to Materials Testing [Chap. Effect of Strain Rate on Tensile Properties The roomtemperature stressstrain curve is not greatly influenced by changes in the rate of straining of the order obtainable in the ordinary ^40 o o o sz a> c= ^ 10 n . 96. ductility than the conventional percentage elongation. For this reason the uniform elongation is a more fundamental measure of uniformly.
' A. Tests in which the specimen is pulled at a constant true strain rate are not readily performed on conventional Although it is fairly easy to maintain a constant rate testing machines. e = ki + ki log (934) I. this does not ensure a constant rate of strain in the specimen since the rate of straining in the specimen increases with load. pp. 96] The Tension Test 255 and the strain rate is difficult because of the many experimental problems associated with measuring tensile properties at very rapid rates Among the experimental problems is that an adiabatic of deformation. New York. particularly during necking. For a cylindrical specimen the true strain rate is related to the instantaneous diameter Di d[2 In (Z)o/A)] dt by (932) de dt 2 d{D^) Di dt The true strain rate is related to the conventional strain rate by the following equation . there is not enough time for the heat of plastic deformation to be dissipated. condition is created at high strain rates. The strain rate expressed in terms of stress conventional linear strain is e. The true strain rate e is given by d[\n (L/Lo)] dt 1 ^e dt dL dt v L L (931) This equation indicates that for a constant crosshead speed the true To maintain a constrain rate will decrease as the specimen elongates.. the crosshead velocity to the increase in length of the specimen. Nadai. causing the temperature of the specimen to increase. ^ dt dt U dt U ^^'^^^ Thus. of crosshead movement. Inc. in proportion strain must increase stant true rate. Nadai^ has presented a mathematical analysis of the conditions existing during the extension of a cylindrical specimen with one end fixed and The the other attached to the movable crosshead of the testing machine. '~L~Ldt~l+edt~l+e Strainrate experiments with mild steel have relationship between the lower yield point (To _ v^ _ Lode _ 1 de _ e ^^"'^'^^ shown a semilogarithmic and the strain rate.: >ec. 1950. the conventional strain rate is proportional to the crosshead velocity." vol. crosshead velocity is y = dL/dt. The equation is applicable up to the onset of necking. Hill "Theory of Flow and Fracture Book Company. McGraw . of Solids. 7475.
Phy&. at a given strain and temperature. . 2232. or recrystallization may ranges to alter this general behavior. pp. log (q2/oi) log (€2/61) (936) most metals increases with temperature and with 97. strength decreases and ductility increases as the test temperature is increased. The maximum in strength is accompanied by a mini1 C. The strain rate sensitivity m may be defined as the ratio of the incremental change in A <T to the resultant change in log e. elevated temperature occur over certain temperature Further. of steel with temperature. stress 9 However. a more general relationship between flow rate. Temperature Fig. /. 910. — 910.* 256 Applications to Materials Testing ^ [Chap. H. Effect of Temperature on Tensile Properties In general. vol. 911. The change with temperature in the shape of the engineering stressstrain curve in mild steel is shown schematically in Fig. at constant temperature and strain. Changes in engineering stress Variation of tensile properties strain curves of mild steel with temperature. App/.T (935) m is a coefficient known as the strainrate sensitivity. 'Room temperature en f Tens. 91 1 .. However. m— The strainrate sensitivity for strain. The variation of the tensile properties of steel with temperature is shown in Fig. /96°C •O) O o Strain e Fig. Hollomon. seems to be C{ey where log and strain W. extended exposure at may cause creep. 15. The strength of steel increases as the temperature is raised above room temperature. structural changes such as precipitation. strain aging. value for this parameter can be obtained from a test where the strain rate is rapidly changed from one value to another. Zener and J. 1944.
Sec.100 Sso £ 60 40 20= 200 . Figure 912 shows the variation of yield strength with temperature for bodycentered cubic tantalum/ tungsten. molybdenum. 140 120 This difference in the o o S. which occurs in the vicinity of 400°F. and iron and facecentered cubic nickel. 97] TheL Test 257 mum in ductility. Note that the yield strength of nickel increases with decreasing temperature to a lesser extent than in the bodycentered cubic metals. due to strain aging or blue brittleness.
cal/g mole universal gas constant. 1. exp RT an activation energy for plastic flow. °K If this equation represents the data. where Q = R = 100 60 40 20 .987 cal/(deg)(mole) T = testing temperature.:9 The temperature dependence of the flow stress at constant strain and strain rate can be generally represented by (937) a = C2. a straight line is obtained for a plot The activation energy is obtained from the slope of of In a versus 1/T.258 Applications to Materials Testing [Chap.
the MacGregor and Fisher. vol. J. pp. = t(\  k In ) (940) eo are constants related to reactionrate constants. 14691475. MacDonald. and temperature regardless of the previous temperature and rate of straining. cit. strain rate. 1116. (940) k and T. vol. Bechtold. pp. m is the strainrate sensitivity. 3 ^ W. E.It has been shown that the same functional relationship is obtained between stress and strain for a constant value of Z. Dorn. /. 49. Trozera. It was suggested that Eq. AIME. vol. ASM. where In Eq. D. 173188. and C. if a metal did not undergo a phase change or gross change in metallurgical structure. (938) represented a mechanical equation of state. vol. and Lankford. (938) was it is first proposed. AIME. 535. Zener and Hollomon originally based this relationship on the fact that the yield strength and tensile strength of steel and copper correlated well with Z over rather wide ranges of e and T. Carlson. 2 T. 1947. Appl. Sherby. p. The concept of the mechanical equation of state^ indicated that the flow stress of a metal was a function only of the instantaneous values of strain. H. When Eq. U.^ They proposed is that strain rate and temper ature could be combined into a velocitymodified temperature. C. MacGregor and J. 1953. pp. Trans. 171. is The quantity A/7 in parentheses in (938) often called the ZenerHollomon parameter Z.. 197. O.. 13.Sec. it w^as interpreted much more broadly than today. This equation was originally verified for data on steel and aluminum over a large temperature range but only a small range of strain rates. Hollomon. H. . 98] The Tension Test 259 Eq. flow stress at a particular strain a function of the velocitymodified temperature T^. In other words. it was considered that a metal would arrive at the same final conditions of flow • J. 1957. analogous to the equations of state for a perfect gas. Trans. Z = e exp ^ (939) plot of In e versus 1/T should result in a straight line. More recently it has been found to hold for true stress data on molybdenum^ and pure aluminum. 1946. stress this relationship does not uniquely describe the flow this A slightly different type of approach to problem was taken by Then. A. Mech. Trans. More recently* it has been verified for lowcarbon steel over a greater range of strain rates. op. but since AH is not inde A pendent of curve. Fisher.
Dorn. p. ." pp. (93S) was satistied. of the notch produces a condition of biaxial stress at the root of the notch and triaxial stress at the interior of the specimen. specimen will show whether or not a material is notchsensitive and prone to brittle fracture in the presence of a stress concentration. a tension test with a notched P^ D ^ a ^ P Fig. Tietz. The introduction As was shown previously in Sec. The impact test has the advantage of ease in preparing specimens and testing over a wide range of temperature. extensive experiments' on aluminum. American Society for Metals. 163179. "Cold Working of Metals. AIME. Figure 914 shows the geometric details of a notched tensile specimen. Tietz and E. Notch Tensile Test The ordinary tension test on smooth specimens will fail to indicate notch sensitivity in metals. and temperature. 99.  T. and T. Trans. this test has been widely used for mild steels and as a test for temper embrittlement. Dorn. However. vol. Ohio. 1949. 205. copper. A. J. Metals Park. but it lacks the advantage of the notch tensile test of more basic interpretation of test results because of a betterdefined state of stress. as described in Chap. 9 stress and strain by dilYerent paths of strain rate and temperature provided that Eq. 14. and lowcarbon steel have shown appreciable deviations from the beha^•ior predicted by the mechanical It is now established that the flow stress depends on equation of state. the presence of transverse stress at the notch increases the resistance to flow » and decreases the ratio of J. E. 1949. for studying hydrogen embrittlement in steels. 914. E. 180. strain rate. Details of notched tensile specimen. stainless steel. The notch tensile test has been used for testing high strength steels. 260 Applications to Materials Testing [Chap. However. and for investigating the notch sensitivit}' of hightemperature alloys. the previous conditions of temperature and strain rate as well as on the instantaneous values of strain. Goldberg. E. The failure of the mechanical equation of state is due to the fact that the structural changes occurring during plastic deformation are dependent not solely on strain but on the strain rate and temperature as well. Notch sensitivity can also be investigated b^y means of the notchedtensile impact test. 712.
. reducing the notch sharpness increases the notch strength and the notchstrength ratio.000 psi. The notch usually sensitivity of steel is evaluated strength. is 200 The term notch ductility to indicate the reduction of area at the notched region of . is The shape. sharpness a/r and the notch depth.Sec. Below this point the NSR is about Note that the notch ductility decreases to very low values for tensile strengths over 200.000 psi there is no effect on the notch strength of increasing the radius from 0. where ductility is low. At strength levels below about 200.050 in. The most commonly used notch has 50 per cent of the area removed at the notch. of curves which are The drops off strength notch sharply at about the strength level. The than used NSR tivity.001 to 0. The amount is is notch ductility often very 100 small and therefore difficult to determine accurately. 200. Notch depth = 1 characterized by the notch  r. with a radius of 0.001 and a 60° notch angle. unity. of The effect changes in notch radius on notched tensile properties depends on the strength level of the steel. the material notchbrittle.2 The notch strength is defined as the maximum load divided by the original ratio crosssectional area at the notch. is a measure of notch sensiIf the NSR is is less ^ S. notchstrengthtensilestrength curve a function of the notch Increasing the notch radius reduces the elasticstress concentra tion but has little effect on the degree of triaxiality of stress. above this strength level. For most heattreated steels the NSR falls off below 1. in. The notchstrength '^^^ (NSR) is the ratio of the notch strength to the ultimate tensile strength. by measuring Figure 915 the notch strength as a function of tensile shows the type obtained.000psi indi cating that the steels are notchbrittle 1.5 when the notch ductility drops below about 6 per cent.5. At high strength levels. 99] The Tension Test is 261 A notch shear stress to tensile stress.
1948. Differences in an annealed due to the shape of the cementite particles are shown in where the tensile properties of a spheroidized structure are compared with a pearlitic structure for a steel with the same carbon content. 9 On triaxiality the other hand. American Society for Metals. The literature on notch tensile testing has been described in a number of reviews.262 Applications to Materials Testing [Chap. Ohio. elements in solid solution (see Fig. 90132. NSR = At higher 1 + notch depth. Empirical correlations have been worked out^ between composition and cooling rate for predicting the tensile properties of steels with tensile properties Fig. Notch Tensile Testing. 916. 244272. subject. It has been clearly demonstrated that microstructure is the chief metallurgical variable which controls the tensile properties of steels. 3 1. 1947. Kramer. 59) The carbon content has a very strong effect amount of cementite A normalized steel will have higher strength than steel because the more rapid rate of cooling used in the normalizing treatment causes the transformation to pearlite to occur at a lower temperature. Gorsuch. pp. 79. Because of the wide variety of microstructures which are possible with changes in composition and heat treatment. yet somewhat complex. R. 100 % strengths. a great deal work has been done in correlating their tensile properties with composition and microstructure. Tensile Properties of Steels Because of the commercial importance of ferrous materials. D. . Trans. and distribution of the cementite. ^'^ 910. At low is strength levels the notchstrength ratio a linear function of the notch depth. 111115. pearlitic structures. "Fracturing of Metals. P. Lubahn. changing the notch depth produces large changes in with only small changes in stress concentration. Newhouse. D. pp. The strength of the ferrite depends on the amount of alloying and the ferrite grain size. 172. vol. L. because it controls the amount of cementite present either as pearlite or The strength increases and ductility decreases with as spheroidite. D. AIMS. of shape." pp. Metals Park. where the ductility is low. Trans. and D. The tensile properties of annealed and normalized steels are controlled by the flow and fracture characteristics of the ferrite and by the amount. Lubahn. and a finer pearlite spacing results. ij. increasing carbon content because of the increased in the microstructure. this is a very interesting. vol. 1957. ASME. the notch strength is dependent on the notch ductility. 2 J.
910] TheT ension Test 263 .Sec.
" which showed that the tensile strength varied linearly with the variation of tensile properties for a NiCrMo eutectoid is thermalreaction temperature.^ This logarithm of the tures.264 Applications to Materials Testing [Chap. mean free ferrite path in isothermally transformed strucIn the region 1300 to 1000°F the transformation product is 300 250  8 o &200 ^ 150 100 50 . 9 certain microstructures on the properties of steel. Figure 917 shows the steel with isoa recent extension of Gensamer's work.
1 0. be commercially /I 99. sitic Because structure. 497. Hardenahility. importance of obtaining a fully martenit is desirable that the steel have adequate hardenability.6 0. in many cases. ability may be increased by altering the transformation kinetics by the addition of alloying elements.. Asquenched hardness of steel as a function of carbon content for different {ASM Metals Handbook. measure of the strength of quenched and tempered steels.9% martensite 95 % martensite C D  90% 80% martensite E .3 0. 1948 percentages of martensite in the microstructure.) impractical. p.5 wt.8 Carbon content._ Sec. 11.7 0. while hardenability Hardenis connected with the transformation characteristics of a steel. ed. the attainment of a comture of 100 per cent martensite.) Hardness is associated with strength. Figure 918 shows the hardness of martensite as a function of carbon content for different total amounts of martensite in the microstructure.2 0. These curves can be used to determine whether or not complete hardenHardness is used as a convenient ing was obtained after quenching. while the hardness of a steel with given transformation kinetics is controlled primarily by the carbon content. The best combination of strength and ductility is obtained in steel which has been quenched to a fully martensitic structure and then tempered.50 % martensite martensite _ 0. pletely martensitic structure may. 918. The validity of this procedure is based on the excellent correlation which exists between . distribution of from hardness.4 0. % Fig. should be differentiated which is the property of a material which represents its of the resistance to indentation or deformation. (This subject is discussed in Chap. The best criterion for comparing the tensile properties of quenched and tempered steels is on the basis of an asquenched strucHowever. the property of a steel which determines the depth and hardness induced by quenching. 0. 910] TheT ension changes in microstructiire is Test 265 of the reduction of area to well illustrated by these results.
altered The mechanical properties of a quenched and tempered steel may be by changing the tempering temperature. 919). This is the typical behavior for heattreated steel. annealed.266 Applications to Materials Testing [Chap. . 9 tensile strength and hardness for heattreated. Figure 920 shows how an hardness and the tensile properties vary with tempering temperature for SAE 4340 steel. and normalized steels (Fig.
000 to 200^000 psi. 43. 2 W. that certain generalities can be altoy steels containing 0.. 1939.5 ^100 30. indicated logical to will The expected scatter in values by the shading. Patton.^carbon which are quenched to essentially 100 per cent martensite and then tempered back to any given tensile strength in the range 100. vol.diameter bars. it is why so many 14. . as be seen in Chap. °F 1400 Fig.Sec. ' and they may differ considerably in these E. ^Utal Prngr. lowalloy steels do not have the same impact resistance or notch sensitivity. 515518." pp. r. '200 50 15040. the mechanical properties of this the tensile strength. tant to note that this generalization does not say that two alloy steels given the same tempering treatment will have the same tensile properties.. Figure is 921 shows this relationship between the mechanical properties of steels with tempered martensitic structures. Ohio. 20 50ion E\o^ . Tensile properties of quenched and tempered tempering temperature.3 to 0. 726733. all different alloy steels are used. pp. Baeyertz. the other common tensile properties will have a relatively fixed value depending only on In other words. J. Janitsky and M. 1943. For fully SAE 4340 steel as a function of hardened lin. carbon It is imporcontent within the above limits.10uj 200 400 600 800 1000 1200 Tempering temperature. because different tempering temperatures would quite likely be required to bring two different alloy steels to the same tensile strength. Actually. Metals Park. American Society for Metals. important class of steels do not depend basically on alloy content. or tempering temperature. "Metals Handbook. 920.5 per cent made about 300 250 60. 910] TheT« Test 267 For lowtheir properties. ask Because of this similarity in properties.
psi.000 Further.268 respects Applications to Materials Testing [Ch ap. For with closely spaced carbon contents. steels are available carbon content consistent with the required asquenched hardness. . when heattreated to tensile strengths in excess of 200. it is an advantage to use a steel with the lowest this reason. to minimize processing difficulties such as quench cracking and weld embrittlement.400 300 200 70 60 S 50 e 40 30 .
a severe plastic deformation which produces a strong preferred orientation will cause a polycrystalline specimen to approach the anisotropy of a single crystal. and M. the The shorttransverse direction minimum dimension The of the product. Averbach. Two is transverse directions must be considered. The dependence of properon orientation is called anisotropy. the deviation in the properties of slackquenched steel from those of tempered martensite becomes smaller. As the tempering temperature is increased. The effect of slack quenching is greatest at high hardness levels. The yield strength. Cohen. the thickness of a plate. Crystallographic anisotropy may also result in the elliptical defor mation of a tensile specimen. products are not the same in ties Since the strength of a single crystal is highly anisotropic. and second phases in the direction of working. segregation. especially when they have been severely worked into sheet. > L. is for example. although the formation of a recrystallization texture can cause the reappearance of a different type of anisotropy. Crystallographic anisotropy results from the preferred orientation of the grains which is produced by severe deformation. This type of anisotropy is important in forgings and plates. 911] The Tension Test 269 while impact strength can be very greatly reduced. depending upon the type of preferred orientation which exists. Anisotropy of Tensile Properties frequently found that the tensile properties of wroughtmetal It is all directions. are the properties most affected." or nonuniform deformation in deepdrawn cups. This is the long axis of a bar or the rolling direction in a sheet or plate. Castleman. B. ASM. Two general types of anisotropy are found in metals. This type of anisotropy is most frequently found in nonferrous metals.Sec. pp. The principal direction of working is defined as the longitudinal direction. L. . and to a lesser extent the tensile strength. 240 263. vol. 44. Mechanical fihering is due to the preferred alignment of structural discontinuities such as inclusions. In steels with sufficient hardenability to form 100 per cent martensite it is frequently found that not all the austenite is transformed to martensite on quenching. 1952. 91 1 . longtransverse direction perpendicular to both the longitudinal and shorttransverse directions. S. Crystallographic anisotropy can be eliminated by recrystallization. voids. Trans. Studies' have shown that the greatest effect of retained austenite on tensile properties is in decreasing the yield strength. is A practical manifestation of crystallographic anisotropy the formation of "ears. dicular to the The yield strength in the direction perpen main (longitudinal) direction of working may be greater or less than the yield strength in the longitudinal direction.
while in a sheet the properties in the shorttransverse direction cannot be measured. In a round or square. pal cause of directional properties. and highest in the longitudinal direction. of area are most affected.270 Applications to Materials Testing [Ch ap. In wroughtsteel products mechanical fibering is the princi Measures of ductility like reduction In general. both these transverse directions are equivalent. 70 . intermediate in the longtransverse direction. reduction of area is lowest in the shorttransverse direction.
911] TheTc Test 271 scatter in measurements of RAT it is necessary to use statistical methods. with strength In the region of tensile strength between 80.5 per cent for each 5.000 and 180.000 psi increase in Figure 923 shows the way in which the longitudinal tensile strength. The degree of anisotropy in reduction of area increases and transverse reduction forging ratio is of area varies with reduction by forging.000 psi the RAT decreases by about 1. level. The the ratio of the original to the final crosssectional area 72 .Sec.
8. ASM.: "Engineering Materials. p. it was considered that they could be initiated at inclusions and secondphase particles. it bar when was untwisted. If the specimens were twisted to this strain and then untwisted. in "Properties of Metals in Materials Engineering. 1954. and with tensile properties. Inc. New York.. John Wiley & Sons. 1940. York.: "Theory of Flow and Fracture of Solids. ductility. 9 from a cupandcone fracture to a fracture on a 45° plane. 1954." chaps. In interpreting these results it was suggested^ that the twisting produced a preferred orientation of initially randomly oriented The cracks were presumed to become oriented along the microcracks. solidification of the ingot and perhaps during plastic deformation. when they are oriented in the principal working direction. 1952 Nadai." American Society for Metals.: "Strength and Resistance of Metals. 3 and 4. New York. N. Hollomon. B. and B." chap. this orientation have little effect on the BIBLIOGRAPHY J. Englewood Cliffs. The cracks were assumed to become reoriented in the longitudinal direction they would Although there was no real experimental evidence for the existence of microcracks. pp. J. vol. A. Trans. 163.. Proc. 33. confirmed and extended Swift's observations." vol. 1950. ASTM. 1. or type of fracture. 1949.J. McGrawHill Book Company. John Wiley & Sons. Ohio.. H. similar experiments^ with OFHC copper. . R. 655680. there was little effect of the torsional deformation on the fracture stress. 46. A. However. The mechanical anisotropy which was observed was explained on the assumption that the metal contained a fibrous flaw structure. vol. D.. Inc. 1 and 11. 501609. Metals Inc. Low. 104).. vol. ASM. 1947. helical surface. Hollomon." chaps. 40. A.272 Applications to Materials Testing [Chap. J. I. Zener and J. Symposium on Significance of the Tension Test of Metals in Relation to Design. J. Shaler. Lessells.Hall. 1 C. Jr. Jaffee: "Ferrous Metallurgical Design...: New Park. 1944. 2 W. which is in compression during twisting (see Fig. Marin. which has the characteristics of submicroscopic There is some indication that these cracks originate during cracks. Backofen. in which there were no secondphase particles and no preferred orientation. and L. Hundy. Prentice. Inc. Trans. H. Separation occurs along this 45° plane of the when axial tension is applied. chap. John M. Behavior of Metals under Direct or Nonreverse Loading. pp.
sional loading in service. with a chuck for gripping the specimen and for applying the twisting moment to the speci men. axles. The twisting moment resisted by shear stresses 273 . perature twist test to evaluate the forgeability of materials.Chapter 10 THE TORSION TEST 101. Torsion tests are made on materials to determine such properties as the modulus modulus of elasticity in shear. Mechanical Properties in Torsion is Consider a cylindrical bar which subjected to a torsional is moment at one end (Fig. since this represents the simplest Since in the elastic range for the calculation of the stress. be carried out on fullsized which are subjected to torIt is frequently used for testing brittle mateand has been used in the form of a hightem may and twist drills. Introduction met with the wide acceptance and the use However. 102. the shear stress varies linearly from a value of zero at the center of the bar to a maximum value at the surface. A torsion specimen geometry generally has a circular cross section. it is useful in many engineering applications and also in theoretical studies of plastic flow. rials. Determination is made of the angular displacement of a point near one end of the test section of the specimen with respect to a point on the same longitudinal element at the opposite end. or torque. and a weighing head. the torsional yield strength. which grips the other end of the specimen and measures the twisting moment. of rupture. it is frequently desirable to test a thinwalled tubular specimen. This results in a nearly uniform shear stress over the cross section of the specimen. The torsion test has not been standardized to the same extent as the tension test and is rarely required in materials specifications. 101). and the Torsion tests also parts. such as tool steels. The deformation of the specimen is measured by a twistmeasuring device called a troptometer. Torsiontesting equipment consists of a twisting head. torsion test has not The that have been given the tension test. such as shafts.
from Fig. polar moment a Since the shear stress cylindrical specimen where / '" maximum at the surface of the bar. Mt = But jr"^ rr = a TfdA = /r fa r r'dA jo (101) =o dA is the polar moment of inertia of the area with respect to the axis of the shaft. Thus. psi torsional radial distance Mt = = J = r moment. for a solid = 7ri)V32.^ is of shaft. in. measured from center of inertia. used to determine the angle of twist. .274 set Applications to Materials Testing [Chap. d. Torsion of a solid bar. the maximum shear stress is ^ MtD/2 ^ \QMt .7rZ)V32 " ttD^ stress (103) For a tubular specimen the shear on the outer surface is ^ where Di IQMtDi (104) = outside diameter of tube inside diameter of tube is D2 = The troptometer expressed in radians. usually If L is the test length of the specimen. 10 up in the cross section of the bar. in. Ibin. 101. Equating the Fig. twisting moment to the internal resisting moment. Mt = or rJ r — (102) MtT where r = shear stress. The shear stress is zero at the center of the bar and increases Hnearly with the radius.
The elastic properties in torsion may be obtained by using the torque at the proportional limit or the torque at some offset angle of twist. or the specimen will by buckling rather than torsion. frequently 0. Thus. 102. and calculating the shear stress corresponding to the twisting moment from is the appropriate equa tions given above. first onset of yielding is generally not readily apparent with the instruments ordinarily used for measuring the angle of twist. deg Offset Fig. (103) or (104) does not strictly apply. as shown in Fig.001 radian/in. A torquetwist diagram is usually obtained. The use of a thinwalled tubular specimen minimizes this effect because the stress the gradient is practically eliminated. an ultimate torsional shearing strength. that fail the wall thickness not reduced too greatly. however. Angle of twist. 102J The Torsion be seen that the shear strain is Test 275 101 it will given by (105) 7 = tan = rd — During a torsion test measurements are made of the twisting moment Mr and the angle of twist. 6. or modulus of rupture.Sec. is frequently determined by substituting the maximum measured torque into these . However. Torquetwist diagram. 102. Experience has shown that for determi nations of the shearing yield strength and modulus of elasticity the ratio of the length of the reduced test section to the outside diameter should be about 10 and the diameterthickness ratio should be about 8 to 10. Because of the stress gradient across the diameter of a solid bar the surface fibers are restrained from yielding by the less highly stressed inner fibers. Once the torsional yield strength has been exceeded the shearstress distribution from the center to the surface of the specimen is no longer linear and Eq. is Care should be taken. of gage length. A tubular specimen usually required for a precision measurement of the torsional elastic limit or yield strength.
276 Applications to Materials Testing [Chap. and the angle of twist." 2d ed. To simplify the analysis. Referring to Eq. Nadai^ has presented a method for bar is calculating the shear stress in the plastic range is the torquetwist curve known. . Proc. 1957. the shear 7 = rd' (108) in a cross section of the bar. 10 equations. for the purpose of comparing and For the determiselecting materials it is generally sufficiently accurate. G. McGrawHill Book Company. The results obtained by this procedure overestimate the method of calculating this value be discussed in the next section. Nadai. pp. is the modulus to the shear strain. A. "Theory of Flow and Fracture of Solids. is related to the shear strain by the stressstrain T = /(t) Introducing this equation into Eq. Inc. (102) and (105) into Eq. (106) gives an expression modulus in terms of the geometry of the specimen. (109) and changing the variable from A. (103) and (104) do not apply. 1950. 57. Within the elastic range the shear stress can be considered proportional The constant of proportionality. or the modulus T of rigidity. p. we shall consider the angle of twist per unit length. vol. the torque. I. 6'. strain will be where 6' = d/L. S. ASTM. Fields and W. (105). ^ Backofen. vol. =Gy (106) for the shear Substituting Eqs.. Torsional Stresses for Large Plastic Strains Beyond the of the torsional yield strength the shear stress over a cross section no longer a linear function of the distance from the axis. Equation (101). 12591272. G = ^ MtL if (107) 103. 347349. and Eqs. A generalization of this analysis for strainrate sensitive materials has been given by D.. the ratio of ultimate shear stress. will A more precise gage length to diameter should be about 0.5 and the diameterthickness about 10 to 12. New York. nation of the modulus of rupture with tubular specimens. for the resisting torque can be expressed as follows: Mn = 27r y"^" rr^ dr (109) Now the shear stress curve in shear. ratio of elasticity in shear. Although the procedure just described results in considerable error.
103. Figure 103 illustrates how this is done. (1010) with respect to ^ (Mre") But. 103 as follows: it can be written in terms of the {BC 27ra' + 3CD) (1012) It will also torque dMr/dd' be noticed from Fig. Examination of Eq. diMrd") dd' = = 27ra»(0')V„ SMrier + Therefore. 103] r The!orsion of Test 277 to 7 by means Eq. Angle Fig. (108) gives Mt = Mrid'Y = where 7a 27r ^^^' {d'Y e' 27r / "f(y)Ydy 6' (1010) = ad'. Therefore. of twist per unit lengtti d Method of calculating shear stress from torquetwist diagram. (1011) shows that geometry of Fig. the shear stress can be calculated with the above equation. 103 that at the maximum value of = 0. dMr (G'y^ 2ira\d'YTa (1011) If a torquetwist curve is available. the ultimate torsional shear strength . Differentiating Eq.Sec. the = 2wafiad')a'id'r = 2TaHe'mae') is maximum shear stress in the bar at the outer fiber Ta = /(ad') Therefore.
State of stress in torsion. Since this plane bisects the angle between the two planes of maximum shear stress and makes an angle of 45° with the longitidunal and transverse directions. and 03 is The intermediate stress a 2 is zero. A along a plane perpendicular to the direction of the maximum tensile stress. perpendicular to the longitudinal axis ^max '73= C^1 Fig. {i>) 105. (a) Fig. ^^ (1013) 104. Types of Torsion Failures Figure 104 illustrates the state of stress at a point on the surface of a bar subjected to torsion. is a tensile stress. of the fracture is normal to the longitudinal axis (see Fig. cTi yy and parallel with the longitudinal axis xx. 10 modulus can be expressed by r. Torsion failures are different from tensile failures in that there is little A ductile metal fails by shear along one of the planes of maximum shear stress. = 3M. Typical torsion failures. (a) Shear (ductile) failure. 1056). Generally the plane localized reduction of area or elongation. oi in magnitude to the shear an equal compressive stress. (b) tensile (brittle) failure. 104.. The maximum shear stress occurs on two mutually perpendicular planes. The principal stresses and 03 make an angle of 45° with the longitudinal axis and are equal stresses.278 or Applications to Materials Testing of rupture [Chap. it results in a helical brittle material fails in torsion fracture (Fig. 105a). test section of the Fractures are sometimes observed in which the specimen breaks into a large number of fairly small .
tests can be made fairly easily at constant or high strain rates. Moreover. . there will be a steep stress gradient across This will make it difficult to make accurate measurethe specimen. obtained in torsion without complications such as necking in tension or barreling due to frictional end effects in compression. ASTM. 46. unless angleoftwist data into shearstressstrain curves. considerable labor is involved in converting torqueFurthermore. On the other hand. ^2 — W Q Traax Sinh^ Tmax €1 — 63 = 2ei i ff = [H(e. 320. The tension test and the torsion test are compared below in terms of the state of stress and strain developed in each test. 2. pp. pp. directly a shearstressshearstrain curve. Trans.' + 62^ + e^r^^ = (71 & = V 2 3(ri 7 1 2 R. Wessel. Hull. €2 __«i — — ~ K €3 2 max ^1 ^max __. This type of curve has more fundamental significance in characterizing plastic behavior than a stressLarge values of strain can be strain curve determined in tension. 105. a tubular specimen is used. 38. 0^2 0"! = Cs = O"! = ~0'3'. In these cases started on a plane of maximum can usually be determined that the fracture shear stress parallel with the axis of the steel as a function of specimen. Proc. Cr2 So"! = 0max ' max 2 Cmax _ — = Cl. C. ASM. vol. D._n — = Cl 2 — ^3 .Sec. Sauveur. T. Torsion Test vs. pt. ments of the yield strength. Tension 0"! test Torsion test = (Tmax'. 8799. and F. A study of torsion failures in a tool hardness^ showed that fracture started on planes of maximum shear stress up to a Vickers hardness of 720 and that above this hardness tensile stresses were responsible for starting fracture. in torsion. E. 1938. 105] The Torsion it Test 279 pieces. Tension Test A good case can be made for the position advanced by Sauveur^ that the torsion test provides a more fundamental measure of the plasticity For one thing. vol. Olleman. 1954. A. the torsion test yields of a metal than the tension test.
(After Gensamer. 10 be twice as great in torsion as in first approximation it can be considered that plastic deformation occurs on reaching a critical value of Tmax and brittle fracture occurs on reaching a critical value of omax. 106.280 Applications to Materials Testing rmax will [Chap. This comparison shows that Since as a representative of the condition for a brittle material such as hardened Critical r^ax for plastic flow Critical ! CTmax for fracture Fig. This is illustrated schematically in Fig. Effect of ratio Tmax/cmax in determining ductility. the opportunity for ductile behavior is greater in torsion than in tension.) . which can be considered tension for a given value of omax. 106.
a straight line is obtained for torsion data when the logarithm of significant stress plotted against the logarithm of sig and n obtained from these curves agree The values of with comparable values obtained from the tension test. H. 577586. Englewood Cliffs. . 10331051. Metals Park. "The Testing and Inspection of EngiMcGrawHill Book Company. A. G.: "Engineering Materials. 111112. Trans. the literature.. 106. 43.. 1940. 43. P. Ohio.. When both curves are plotted cant strain (the tension curve within fairly close limits. 1941. E. ASM.Sec. 62.: "Strength of Metals under Combined Stresses. American Society for Metals. 2d ed. Ohio. Trans. Klier. Wiskocil: neering Materials.*'^ nificant strain. the figure in torsion shows that the amount in tension. pp. Marin.J. M. J. PrenticeHall." pp. 2. N. Davis. ASME. Troxell. test. 5. Inc. 1 E. Figure 107 shows a truestresstruestrain curve from a tension test and the shearstressshearstrain curve for the same material in torsion. E. ASM." chap. of plastic deformation is greater sion The tensile stressstrain curve can be derived from the curve for torwhen the stressstrain curve is plotted in terms of significant stress and strain or the octahedral shear stress and strain (see Prob. 2 3 H. 1948. 9931012. Even for a metal stress is where the than critical normal which pushed is ductile in the tension far to the right in Fig.. 1951. pp." chap. Marin. Inc. "Metals Handbook. Larson and E. Trans." American Society for Metals. vol. Gensamer. T. pp. while in tension the critical normal stress is reached before the shear stress reaches the shear tool steel. H. Metals Park. 1955. 1952. 105] The Torsion Test 281 In the torsion test the critical shear stress for plastic flow is reached before the critical normal stress for fracture. J.^ fairly well A terms of significant stress and signifiunchanged). and C. Faupel and J. 1951. vol.4). New York. 10. K BIBLIOGRAPHY Davis. stress for plastic flow.. vol. the two curves superimpose number of examples of this can be found in in is is Also.
easily and fied to the design engineer often means an measured and speci quantity which indicates something about the strength and heat treatment of the metal. general. the Mohs scale. but does not lend itself to high reproducibility or extreme is accuracy. or dynamic. major engineering interest for metals. Hardness is measured according to ducted. 176. order of their ability to be scratched. Bergsman. Most hard metals fall in the Mohs hardness range of 4 to 8. 1951. is This consists of 10 standard minerals arranged in the The softest mineral in this scale 1). A about 2. and martensite a hardness of 7. B. 3743. In dynamichardness measurements the indenter 1 usually dropped E. rials testing. (2) indentation hardness. annealed copper has a value of 3. A different type of scratchhardness test^ measures the depth or width of a scratch fingernail has a value of made by drawing a diamond load. Scratch hardness is of primary interest to mineralogists. Introduction The hardness of a material is a poorly defined term which has of the many In meanings depending upon the experience metals the property plastic deformation. and Only indentation hardness is of (3) rebound. talc (scratch hardness while diamond has a hardness of 10.Chapter 11 THE HARDNESS TEST 111. With this measure of hardness. The Mohs scale is not well suited for metals since the intervals are not widely spaced in the highhardness range. ASTM Bull. stylus across the surface under a definite This is a useful tool for measuring the relative hardness of microit constituents. hardness usually implies a resistance to deformation. hardness. 282 . pp. is person involved. various minerals and other materials are rated on their ability to scratch one another. a measure of their resistance to and for permanent or To is a person concerned with the mechanics of matelikely to hardness most it mean the resistance to indentation. September. There are three general types of hardness measurements depending upon the manner in which the test is conThese are (1) scratch hardness.
(111) d^) diameter of ball. of rebound of the indenter. and the hardness is expressed as the energy of The Shore sceleroscope. and the diameter of the inden measured with a lowpower microscope after removal of the load. It will be noticed that the units of the are kilograms per square millimeter. kg mm mm BHN BHN indentation. It is assumed that the diameter of the indentation is the same as the diameter when the ball was in contact with the metal. d = diameter of indentation. 112] The Hardness Test 283 onto the metal surface. This is expressed by the formula^ is The average BHN = (tD/2){D where ^ . BHN .VD^ . measures the hardness in terms of the height impact. It has been shown that in order to obtain the same Brinell hardness number geometrical similitude must be maintained. The surface on which the indentation is made should be relatively smooth and free from dirt or scale. and for very hard metals a tungsten carbide ball tation is used to minimize distortion of the indenter. However. at a nonstandard load This requires that the ratio To a first of the indentation to the indenter.000 kg. The Brinell hardness number (BHN) is expressed as the load P divided by the surface area of the indentation. (111) does not give the mean pressure over the surface of the P = D = applied load. d/ D. Brinell in 1900. The Brinell hardness test consists in indenting the metal surface with a 10mmdiameter steel ball at For soft metals the load is reduced to 500 kg to a load of 3. A.Sec. approximation this can be attained when P ID^ is kept constant. 112. remains constant. Brinell Hardness The first widely accepted and standardized indentationhardness test was proposed by J. the radius of curvature of the indentation will be larger than that of the spherical ^ Tables giving as a function of d for standard loads may be found in most of the references in the Bibliography at the end of this chapter. The load is applied for a standard time. which is the commonest example of a dynamichardness tester. of two readings of the diameter of the impression at right angles should be made. owing to elastic recovery. the is not a satisfactory physical concept since Eq. The greatest error in Brinell hardness measurements occurs in measuring the diameter of the impression. the Brinell hardness number of a material is constant only and diameter of ball. avoid too deep an impression. However. for one applied load In general. usually 30 sec.
It is frequently desirable to increase the sharpness of definition of the impression so that the diameter can be measured more accurately. 111. The Fig. 1908. Ver. that this mean pressure should be taken as the measure It is referred to as the Meyer hardness. Cross sections drawing on the bottom shows ''sinking in. deut. the greater the elastic recovery. This type of behavior is comridging and (6) sinking mon with annealed metals having a high rate of in. The sketch "ridging. Ing. However. pp. so that this does indenter. but it will have only a negligible effect on the chordal diameter of the impression. strain hardening. The true diameter of the impression can sometimes be obtained by coating the ball with bluing or dye before making the indentation. Meyer hardness = 1 —r^ 4P (112) E.284 Applications to Materials Testing still [Chap. . The mean pressure between the surface of the indenter and the indentation is equal to the load divided by the projected area of the indentation. Meyer proposed of hardness. This behavior is most common in coldworked metals with little ability to strainharden. vol. Meyer Hardness Meyer^ suggested that a more rational definition of hardness than that proposed by Brinell would be one based on the 'projected area of the impression rather than the surface area. 11 be symmetrical. 645654. not in general influence Brinell hardness." in which a lip of metal forms around the edge of the impression. 52. Meyer. The harder Elastic recovery will affect measurements of the depth of indentation. 113." in through Brinell indenwhich there is a depression of the metal at the rim tations illustrating (o) of the indentation. although the indentation will the metal. but since the ridge carries part of the load. The measured diameter is greater than the true diameter of the impression. two types of anoma lous behavior can occur as a result of localized deformation of the metal at the indentation. ically in These are shown schematat the top illustrates cross section through the indentation in Fig. 111. Z. lightly etched steel ball or by coating the This can sometimes be done by using a surface with a dull black pigment. it is customary to base the hardness measurement on the diameter d shown in the sketch." or ''piling up.
113] The Hardness Test 285 Like the Brinell hardness. while the Brinell hardness decreases as the load increases.. The exponent in Meyer's law 2. metals have a value of n' of about 2. The Meyer hardness is a more fundamental measure of indentation hardness yet it is rarely used for practical hardness measurements. A:ii)i"'' P = Meyer found indenter = k. however. is approximately equal to the strainhardening coefficient plus When indentations are n' will made with balls of different diameters. this equa (115) . tion can be written First. Meyer hardness has units of kilograms per The Meyer hardness is less sensitive to the applied For a coldworked material the Meyer load than the Brinell hardness. This can be expressed empirically by a relationship of the form D C = The kiDi''~ = A:2/)2"' = ksDs'^ • general expression for Meyer's law then becomes ^ Cdi"' Z)i"'~2 ^ Cd^ ^ i)2"'2 Cd^'^' ^^^~^' Z)3"'2 Several interesting conclusions result from Eq. Sec. Meyer proposed an empirical relation between the load and the size of the indentation.D2"^' = ksDs"^' ' ' that n' was almost independent of the diameter of the but that k decreased with increasing values of D. Meyer hardness increases continuously with the load because of strain hardening produced by the indentation. and k is the value of P at c? = 1. a material constant related to strain hardening of metal a material constant expressing resistance of metal penetration n' is mm to The parameter the slope of the straight line obtained when log P is plotted against log d. first increases with load and then decreases for still higher loads. The Brinell hardness. (114). Fully annealed fully strainhardened metals. P = where kd'^' (113) P = d n' fc applied load. For an annealed metal the square millimeter. This relationship is usually called Meyer's law. different values of k and be obtained. kg = = = diameter of indentation.5. while n' is approximately 2 for This parameter is roughly related to the strainhardening coefficient in the exponential equation for the truestresstruestrain curve. hardness is essentially constant and independent of load.
» D. Tabor. deformation around the indentation is not fully plastic and Eq. the a lower limit of load below which Meyer's law is not valid. (113) is not obeyed. 47.500 kg.) (a) an ideal plastic metal with no strain hardening the highest pressure occurs immediately below the surface of contact at a depth of about d/2. Oxford University Press. Beginning of plastic deformation at point 0. However. "The Hardness of Metals. {After D. "The Hardness of Metals. and for steel with a exceed 50 kg for copper with a For balls of different diameter of 400 the load should exceed 1. F jd} is proportional to the Meyer hardness. Tabor. geometrically similar indentations give the same Meyer hardness number. Analysis of Indentation by a Spherical Indenter Tabor^ has given a detailed discussion of the mechanics of deformation metal surface with a spherical indenter. the same ratio P/D^ must also provide the same result. 1951. 1'4. The pressure at this point is about 0A7pm. For a 10mmdiameter ball the load should of 100. of a flat •0 (a) Fig." p. The elements of this Figure 112 illustrates the process. (b) full plastic flow. BHN BHN the critical loads will be proportional to the square of the diameter. 1951. Equation (114) can also be rearranged to give (116) Remembering again that geometrically similar indentations are obtained when d/D is constant. Plastic deformation of an ideal plastic material by a spherical indenter. 1 D F Idr must be constant for geometrically similar indentations.1 286 Since dl ratio Applications to Materials Testing [Chap. New York. . There is If the load is too small. For analysis will be described here. This load will depend upon the hardness of the metal. the must also be constant. we see that the above equation shows that the Therefore." Oxford University Press. Therefore. 112. where pm is the mean pressure over Assuming that the maximumshearstress theory is the circle of contact. New York. hardness values will be obtained when the ratio F/D^ is kept constant.
6776. . 1. Since real metals strainharden. plastic elastic until the mean At about this pressure deformation begins in the vicinity of point further increased. the (Fig. sary condition for Meyer's law to be valid.66cro. stress distribution at the hardness indentation pre cludes a straightforward relationship with the stress distribution in the However. 1951. Meyer hardness plasticity occurs tests on severely coldworked metal indicates that Pm when « 2. The method is based on the fact that there is a similarity in the shape of the flow curve and the curve obtained when the Meyer hardness is measured on a number of specimens subjected to true stressstrain curve increasing since the amounts complex of plastic strain. ll2a).Sec.loo (117) where the yield stress in tension or compression. 115] The Hardness Test 287 the criterion for plastic flow. /. the deformation under the indenter pressure reaches about 1. where co is not possible to measure The true stress (flow stress) is obtained from Eq. Meyer hardness = Pm = 2. Inst. Relationship between Hardness and the TensileHow Curve Tabor^ has suggested a method by which the plastic region of the may be determined from indentation hardness measurements.. vol. (118). 79. to be considered the flow stress at a given value of is true strain. we can write 0.8(ro (118) For an ideally value if plastic metal the pressure would remain constant at oo this the load were increased further. is Therefore. As the load is mean it pressure increases and the plastically deformed region grows until 1126). the method has been shown to give good agreetension test. An analytical solution for the pressure between the spherical is indenter and the indentation under conditions of full plasticity The best analysis of this problem indicates that pm « difficult. Metals. Tabor concluded that cit. as the indentation also a neces the pressure would increase owing to an increase in process was continued. The method is basically empirical.5(To 0A7pm = or Pm oo is « l. This is 115. Tabor. Most Brinell hardness tests are carried out under conditions where full plasticity is reached. pp. p.8cro From 1 a study of the deformation at indentations. contains the entire region of contact (Fig.1 times the yield stress. op. ment for several metals and thus should be it of interest as a means of obtaining flow data in situations where tensile properties. very full 2.
Figure 113 shows the agreement which has been obtained by Tabor between the flow curve and hardness versus d/D curve for mild steel and annealed copper. There is a very useful engineer ing correlation between the Brinell hardness and the ultimate tensile strength of heattreated plaincarbon and mediumalloy steels. A brief consideration will show that this is in agreement with Tabor's results. then the tensile strength is equal to the yield stress and Eq. that its limitations should be in 1951. . Tabor's results have been verified by Lenhart^ for duralumin and OFHC copper. "The Haidness of Metals. However. mined (circles. If we make the simplifying assumption that this class of materials does not strainharden.) vestigated for new applications. = 500(BHN). it is possible at least to approximate ' the tensilefiow curve. if the Meyer hardness is varies from the smallest value •"" measured under conditions such that d/D for full plasticity up to large values and Eqs. Tabor. 55114. This work should not detract from the usefulness of this correlation but. (Tu = TynVm = 0. New York. should serve to emphasize mined from compression test (solid lines). Tabor's analysis did not predict the flow curve for mag nesium. WADC Tech. Comparison of flow curve deterfrom hardness measurements crosses) with flow curve deter Lenhart to the high anisotropy of deformation in this metal. Lenhart. 11 the true strain was proportional to the ratio d/ D and could be expressed as e = d 0. rather.288 Applications to Materials Testing [Chap. Oxford University Press. Rept. June. 1955. {D. Upon converting to engineering units the expression becomes au 1 = 515(BHN) R.36p„j kg/mm^ The Brinell hardness will be only a few per cent less than the value of Meyer hardness p^. Ultimate tensile strength. in pounds per square inch." p. (118) and (119) are used.2 D (119) Thus. 1 1 3. which was attributed by Fig. (118) applies. 74. E.
described in the next section. for annealed copper the assumption that For a metal strain hardening can be neglected will be grossly in error. for a mm extremely hard With the Rockwell hardness test. so that measurements at one extreme of the scale cannot be strictly compared with those at the other end. requires careful surface preparation of the specimen. it is usually given load. 116] The Hardness Test 289 the same relationship does not hold for example. ness test has not been widely accepted for routine testing because it is slow.500.. Because shape of the indenter this test. The loads ordinarily used with this test range from 1 to 120 kg. or VPH). The Vickers hardness test uses a squarebase diamond pyramid as the The included angle between opposite faces of the pyramid is 136°. This angle was chosen because it approximates the most desirable of the ratio of indentation diameter to ball diameter in the Brinell hardness test. for personal error in the determination of the diagonal A perfect indentation made with a perfect diamondpyramid indenter .>cc.0) P = L = 6 = applied load. tionality will be greater than that used for heattreated steel. except at very light loads. angle between opposite faces of diamond = 136" The Vickers hardness test has received fairly wide acceptance for research work because it provides a continuous scale of hardness. is frequently called the diamond The diamondpyramid hardness number (DPH). calculated from microscopic measurements of the lengths of the diagonals may be determined from the following The DPH equation. DPH = where ^^ %W2) ^ 1^ („. or Vickers hardness number (VHN. pyramid hardness of the impression. from very soft metals with a of 5 to DPH necessary to change either the load or the indenter at some point in the hardness scale. Because the impressions made by the pyramid indenter are geometrically similar no matter should be independent of load. 116. or the Brinell hardness test. is defined as the load In practice. kg average length of diagonals. with greater capability for strain hardening the "constant" of proporIt shovild now be apparent why P'or other metals. Vickers Hardness indenter. depending on the hardness of In spite of these advantages. This is generwhat their size. the DPH ally found to be the case. and allows greater chance length. this area is divided by the surface area of the indentation. the Vickers hardthe metal to be tested. materials with a DPH of 1.
. The diagonal measurement in this case produces a low value of the contr. under constant load.ct area so that the hardness numbers are erroneously high. results from r 117. so that finished This test utilizes the depth of indentation. p. but unlike the Brinell and Vickers hardness designations.. results in a high hardness number. 14. 1 . F. B. This minimizes the amount of surface preparation needed and reduces the tendency for ridging or sinking in by the indenter. as a measure of hardness. The major load is then applied. ^^ . . This is in agreement with the other hardness numbers described previously. worked metals. and the depth of indentation is automatically recorded on a dial gage in terms of arbitrary hardness numbers. 100. and the small United States is the Rockdue to its speed.• i. l4c is found ^. the Rockwell hardness numbers are purely arbitrary. The barrelshaped indenYlg. A minor load of 10 kg is first applied to seat the specimen. . 1 290 Applications to Materials Testing [Chap. This condition is observed with annealed metals and results in an overestimate of the diagonal [a) length. of 60. diameter steel balls are generally used as indenters. called a Brale indenter. anomalies corresponding to those described observed with a pyramid indenter (Fig. Fig. The dial is reversed so that a high hardness. freedom ability to distinguish small hardness differences in test in the acceptance is size of the indentation. Rockwell Hardness Test The most widely used hardness well hardness test..indentation Perfect indentation i . 1946. Inst. each division representing a penetration of 0. Crowe and J. Empirical corrections for this effect have been proposed. pincushion (c) up of the metal around the faces of the indenter. Its general from personal error. which corresponds to a small penetration. 1 would be a square. Types (a) of diamondpyramid . 114). it is necessary to specify the combination 1 heattreated parts can be tested without damage. 1146 is the of the result of sinking in fiat metal faces of the pyra mid. 72. It t. vol. Since the Rockwell hardness is dependent on the load and indenter. T. 1 14. inVb) ^^^:^^^ . Metals. which have units of kilograms per square miUimeter. and }/{&.00008 in. The dial contains 100 divisions. earlier for Brinell impressions are frequently However.and Major loads 3/^in. /. in cold dentations. and 150 kg are used. Hinsley. The pincushion indentation around the in Fig. One combination of load and indenter will not produce satisfactory results for materials with a wide range of hardness A 120° diamond cone with a slightly rounded point.^ ridging or piling iin due to sinking barreled indentation due to ridging. hardened steel.
118] is The Hardness Test This is 291 done by prefixing the hardness number with a combination of load and indenter for the employed. This done by adjusting the dashpot on the Rockwell tester. October. Hardened is tested on the C scale with prefix is steel letter 150kg indenter and a major load. made on only a single thickness of material. is and empirical^ corrections for this effect have been published. Variations in hardness can be appreciable in very soft materials unless the rate of load For such materials the operating application is carefully controlled. dry. The A scale (diamond penetrator. should be 6. The indenter and anvil should be clean and well seated. 1953. 7. E. Standard El 8. R. The useful range for the diamond materials are usually about 20 to Softer is from 70. which used. ASTM Bull. 2. is should be standardized. Tests on cylindrical surfaces will give low readings. which is usable for materials from annealed Many other scales are available for special brass to cemented carbides. Heyer. diameter steel ball and a 100kg major load. 1 1 8. 3. retieal^ 5. 39.^ The Rockwell hardness test is a very useful and reproducible one provided that a number of simple precautions are observed. handle of the Rockwell tester should be brought back as soon as the major load has been fully applied. 1939.: >ec. vol. The thickness of the specimen should be such that a mark or bulge It is recommended that not produced on the reverse side of the piece. scale this Re Re tested on the B scale with a 3^i6in. letter indicating the particular purposes. the determination of the hardness of individual con» See ASTM S. The 4. load. 2 3 W. A Rockwell hardness number without the hardness scale meaningless. pp. pp. The surface to be tested should be clean. Proc. and free from oxide. smooth. A roughground surface is usually adequate for the Rockwell test. The spacing between indentations should be three The speed of application of the load to five times the diameter of the indentation. and hardness of the material. 193. the error dependTheoing on the curvature. Microhardncss Tests metallurgical problems require the determination of hardness Many over very small areas. H. . The measurement of the hardness gradient at a carburized surface. indenter. surface should be flat and perpendicular to the indenter. 4041. Sutton and R. 60kg major load) provides the most extended Rockwell hardness scale. Most of the points listed below apply ecjually well to the other hardness tests 1. 12811291. ASTM. Ingerson. Tests the thickness be at least ten times the depth of the indentation. The range of this scale is from Rb to Rb 100.
Thibault. The long diagonal of the Knoop impression is essentially unaffected by elastic recovery for loads greater than about 300 g. Hardnessconversion Relationships From ^ a practical standpoint it is important to be able to convert the Since results of one type of hardness test into those of a different test. 331353. 4. 11 stituents of a microstructure. Met. However. ber (KHN) is the applied load divided The Knoop hardness numby the unrecovered projected area of the indentation. 38. For a review of microhardness testing see H. L. ASM. 1959. 2 Tarasov and N. vol. 119. for lighter loads Further. The use of a scratchhardness test for these purposes was mentioned earlier. W. Metallographic polishing is usually required. 49100. kg Ap — unrecovered projected area of indentation. Reviews. P. or the checking of the hardness of a delicate watch gear might be typical problems. so that it is usually observed that the Knoop hardness number increases as the load is decreased below about 300 g. no. mm^ L = length of long diagonal. 1947. the small amount of elastic recovery becomes appreciable. pp.292 Applications to Materials Testing [Chap. . Buckle. Work hardening of the surface during polishing can influence the results. with the very small indentations produced at light loads the error in locating the actual ends of the indentation become greater.have shown that if corrections are made for elastic recovery and visual acuity the Knoop hardness number is constant with load down to 100 g. but an indentationhardness The development of the Knoop test has been found to be more useful. Both these factors have the effect of giving a high hardness reading. ™N = £ = ^ where (llU) P = applied load. thirtieth of the length of the longer diagonal. The Knoop indenter is a diamond ground to a pyramidal form that produces a diamondshaped indentation with the long and short diagonals The depth of indentation is about onein the approximate ratio of 7: 1. 3. vol. Trans. The surface of the specimen must be carefully prepared. pp. Tarasov and Thibault. mm C = a constant for each indenter supplied by manufacturer The low load used with microhardness tests requires that extreme care be taken in all stages of testing.^ indenter by the National Bureau of Standards and the introduction of the Tukon tester for the controlled application of loads down to 25 g have made microhardness testing a routine laboratory procedure.
1027. ASM 1110. Brinell. Proc. p. copper. yet the former has a Rockwell B hardness of 31 which is compared with a hardness of Rb 7 for the coldworked aluminum. However. On the other hand. 3 The Wilson Mechanical Instrument Co. vol. 1944. of the care As an extreme example required in using conversion charts for soft metals. or with greater Heyer^ has shown that the indentation hardstrainhardening capacity. different conversion tables are required for materials with greatly different elastic moduli. 13. it is possible for Armco iron and coldrolled aluminum each to have a Brinell hardness of 66. and 188 Metals Handbook. normalized. 101) is based on tests on these metals.Sec. ASM Metals Handbook. The most reliable hardness harder than 240 Brinell. be discussed in Chap. 44. p. and many other standard references. it is not surprising that no universal hardnessconversion It is important to realize that hardrelationships have been developed. hardness with hightemperature strength properties.. 1948 ed. Chart 38 for metals softer than BHN 240 (see ASM Handbook. and SAE (Society of Automotive Engineers) have agreed on a table for conversion between Rockwell. Heyer. ness conversions are empirical relationships. and tool steels in the and quenched and tempered conditions. 1 . which in turn is dependent on the previous degree of strain hardening of the material before the test. SAE Handbook. 1110] The Hardness Test 293 a hardness test does not measure a welldefined property of a material and since all the tests in common use are not based on the same type of measurements. Hothardness testers using a Vickers indenter made of sapphire and with provisions for testing in either in correlating hot This will This table may be found in ASTM Standard E4847. ASM. such as tungsten carbide. Special hardnessconversion tables for coldstainless steel are given in the worked aluminum. metals such as yellow brass and lowcarbon sheet steel have a wellbehaved BrinellRockwell conversion^* relationship for all degrees of strain hardening. good indication of the potential usefulness of an alloy for hightemperSome degree of success has been obtained ature strength applications. annealed. The ASTM. ASTM. H. ness of soft metals depends on the strainhardening behavior of the material during the test. 2 R. and diamondpyramid hardness which is applicable to heattreated carbon and alloy conversion data exist for steel which is ^ steel and to almost all alloy constructional steels asforged. Hardness at Elevated Temperatures Interest in measuring the hardness of metals at elevated temperatures has been accelerated by the great effort which has gone into developing Hot hardness gives a alloys with improved hightemperature strength.
ASM. ASM. 115. pp. vol. 1 The change in slope occurs at a tem F. vol. °K Temperature dependence vol. H. = Aexp (BT) (1112) H= T = B = hardness. H. kg/mm^ °K constants Plots of log H versus temperature for pure metals generally yield two straight lines of different slope. S. 830837. 57. Trans.) Westbrook^ showed that the temperature dependence of hardness could be expressed by H where A. Smith. Westbrook. 8 00 Temperature. of the hardness of copper. P. 1953. 1953. 3 J. 5358. 233. 1953. pp. 1957. In an extensive review of hardness data at different temperatures 400 Fig. and G. test temperature. Westbrook. 377396. Semchyshen and C. pp. 221248. 1958. ASM. (/. 2 M. Trans. 1960. Westbrook. 873897. Torgerson. 45. V.294 Applications to Materials Testing or an inert atmosphere [Chap. ASTM. Garofalo. . J. 50. ASTM Bull. pp. 11 vacuum have been developed/ and a hightem perature microhardness test has been described. H. p. Trans. vol. ASM. Proc. pp. R. Trans. 45. vol. 45. Malenock. 246.
vol. the bodycentered cubic lattice is always the softer structure when it is perature which tested. ASTM. 1 W. vol. Fe. is the temperature This constant was related in a rather complex coefficient of hardness. New York. Hardness measurements as a function of temperature will show an abrupt change at the temperature at which an allotropic transformation Hothardness tests on Co. AIME. with the heat content of the liquid metal at the melting point and with the melting point. 1956. D. pp." Reinhold Publishing Corporation. 93105. Ti. Trans. Metals Park. 803856. 1951. Symposium on the Significance of the Hardness Test of Metals in Relation to Design. that is. 203. "Metals Handbook. 1948. The constant A derived from the lowtemperature branch of the curve can be considered to be This value would at 0°K. deformation mechanism at higher temperature. 1949. It is likely that this change in slope is due to a change in the copper. The constant B.Sec. Ohio. be expected to be a measure of the inherent strength of the binding forces Westbrook correlated values of A for different metals of the lattice. W.: "The Hardness of Metals. These while highly complex crystal structures give even higher hardness. V. E. Chubb. Mott. 1943. the intrinsic hardness of the metal." Oxford University Press. derived from the slope of the curve.. and Zr have shown^ that occurs. Tabor. Lysaght. The facecentered cubic and hexagonal closepacked lattices have approximately the same strength. 189192." pp. With these correlations it is possible to calculate fairly well the hardness of a pure metal as a function of temperature up to about onehalf its melting point. 1955. Similar behavior H involved in an allotropic transformation. 1110] The Hardness Test is 295 about onehalf the melting point of the metal being is found in plots of the logarithm of the tensile Figure 115 shows this behavior for strength against temperature. 43.: "Microindentation Hardness Testing. pp. This correlation was sensitive to crystal structure. American Society for Metals. London. New York. (Publishers) Ltd. BIBLIOGRAPHY Hardness Tests. Proc. B. .: "Indentation Hardness Testing. results are in agreement with the fact that austenitic ironbase alloys have better hightemperature strength than ferritic alloys." Butterworth & Co. way to the rate of change of heat content with increasing temperature. U.
Frequently the progress of the fracture is indicated by a series of rings. there is no obvious change in the structure of a metal which has failed in fatigue which can serve as a clue to our understanding of the reasons for fatigue failure. These are 296 . presumably It it because fatigue is generally observed that these failures occur only after a to "crystallization" of the metal. compressors. namely. which shows a smooth region." progressing inward from the point of initiation of the failure. until today it is often stated that fatigue accounts for at least 90 per cent of all service failures due to mechanical causes. where the member has failed in a ductile manner when the cross section was no longer able to carry the load. 121). with no gross deformation at the fracture. In fact. Introduction or fluctuating stress will has been recognized since 1850 that a metal subjected to a repetitive fail at a stress much lower than that required to cause fracture on a single application of load. Failures occurring under conditions of dynamic loading are called fatigue failures. pumps. Three basic factors are necessary to cause fatigue failure. due to the rubbing action as the crack propagated through the section (top portion of Fig. subject to repeated loading and vibration. Fatigue has become progressively more prevalent as technology has developed a greater amount of equipment. etc. but this view can For a long time the notion persisted that no longer be considered in the light of concepts which hold that a metal is crystalline from the time of solidification from the melt.Chapter 12 FATIGUE OF METALS 121. A fatigue failure can usually be recognized from the appearance of the fracture surface. because it occurs without any obvious warning. Figure 121 also illustrates another characteristic of fatigue. and a rough region. that a failure considerable period of service. turbines. Fatigue results in a brittle fracture. A fatigue failure is particularly insidious.. On a macroscopic scale the fracture surface is usually normal to the direction of the principal tensile stress. was due usually occurs at a point of stress concentration such as a sharp corner or notch or at a metallurgical stress concentration like an inclusion. or "beach marks. aircraft. such as automobiles.
Sec. overload. (3) a sufficiently large In addition. 122] (1) Fatigue of Metals tensile stress of sufficiently high value. number of cycles of the applied stress. corrosion. and combined stresses. there are a host of and other variables. residual stresses. (2) a large 297 enough a maximum variation or fluctuation in the applied stress. metallurgical structure. temperature. o . such as stress concentration.
stress can be considered to be made up of two com(Xm. R. at the Spec. In keeping with the the maximum and minimum stresses are equal. In this illustration they are both tension. plicated stress cycle which might be encountered in a part such as an aircraft wing which is subjected to periodic unpredictable overloads due to gusts. Figure 1226 illustrates a repeated stress cycle in which the mp^ximum stress omax and minimum stress a^nin are not equal. a mean. 12 illustrate typical fatigue stress cycles. This is an idealized situation which is produced by an R. 1949. Publ. or steady. . Common end of this chapter types of fatigue machines are described in the references listed. but a repeated stress cycle could just as well contain maximum and minimum stresses of Figure 122c illustrates a comopposite signs or both in compression. 2 the minimum stress is the lowest Fig. (c) 122. pletely reversed cycle of stress of sinusoidal Figure 122a illustrates a comform. random stress cycle. (6) repeated stress.298 Applications to Materials Testing [Chap. Tensile stress is considered positive. Typical fatigue stress cycles. A fluctuating 1 stress cycle ponents. ASTM 91. irregular or (a) Reversed stress. and compressive stress is negative. Tech. algebraic stress in the cycle. conventions established in Chap. Moore rotatingbeam fatigue machine^ and which is approached in service by a rotating shaft operFor this type of stress cycle ating at constant speed without overloads. or variable. and an alternating. and in the Manual on Fatigue Testing.
which represents the dependence of the A'^. is onehalf the range of stress. 123] stress (Xa. on the maximum applied stress 0. ^T O'max O^n (121) The alternating stress. Most investigations of the fatigue properties of metals have been made by means of the rotatingbeam machine. is sometimes used in presenting fatigue data is is Stress ratio defined as R = 123.Sec. where the mean stress is zero. <Tm — ^ Another quantity which the stress ratio R. As can be seen from Fig. then. (7„  2 of the (122) The mean stress is the algebraic mean maximum and minimum (126) stress in the cycle. Fatigue of Metals 299 We must also consider the range of stress is Cr. of the specimen. Figure 123 gives typical SN curves for this type of test. 1226. the range of stress the algebraic difference between the maximum and minimum stress in a cycle. in number of cycles to failure. The >SA^ Curve (124) The of the basic method of presenting engineering fatigue data is by means life SN curve. Cases 60 t/i 8 o 50 !^'40 (/) 1^30 ^20 "O O o o .
This is made up of the number of cycles to initiate a crack and the number of cycles to propagate the crack completely through the specimen. at about twothirds the static tensile strength of the material. These will be discussed later in the chapter. T. 364365. which is usually at least 10'^ cycles. magnesium. F. if several specimens are tested at the same stress. vol. Fatigue tests at low stresses are usually carried out for 10^ cycles and sometimes to 5 X 10^ cycles for nonferrous metals. 123. the material can presumably endure an infinite number of cycles without failure. 185. which is called the fatigue limit. In such cases it is common practice to characterize the fatigue properties of the material fatigue strength at an arbitrary by giving the limit number of cycles. 10* cycles. 1 where the mean stress is not zero are of considerable practical interest.. Trans. or endurance limit. Usually no distinction is made between these two factors. . For materials without a fatigue limit the test is usually terminated for practical considerations The SN at a low stress where the life is about 10* or 5 X 10* cycles. The usual procedure for determining an SN curve is to test the first specimen at a high stress where failure is expected in a fairly short number of cycles. Below this limiting stress. the SN curve becomes horizontal at a certain limiting stress. although it can be appreciated that the number of cycles for crack propagation will vary with the dimensions of the specimen. and copper alloys. curve is usually determined with about 8 to 12 specimens. have an SN curve which slopes gradually downward with increasing number of cycles. The highest stress at which a runout (nonfailure) is obtained is taken as the fatigue limit.2 300 Applications to Materials Testing [Chap. for example. e. frequently as much as one log cycle between the minimum and maximum value. 1949. like aluminum. As can be seen from Fig. the number of cj^cles of stress which a metal can endure before failure increases with decreasing stress. These materials do not have a true fatigue limit because the SN curve never becomes horizontal. Most nonferrous metals. Ransom and R. have a fatigue are not known. although a hypothesis regarding this imporcertain materials The reasons why tant question will be discussed later in the chapter. although a smooth curve can usually be drawn through the points without too much difficulty. pp.g. N is taken as the number of cycles of stress to cause complete fracture of the specimen. there is a great amount of scatter in the observed values of number of cycles to failure. Mehl. For a few important engineering materials such as steel and titanium. The test stress is decreased for each succeeding specimen until one or two specimens do not fail in the specified number of cycles. However. it has been shown that ^ the fatigue limit of steel 1 is subject to considerable variation and that a J. Further. It will generally be found that there is a considerable amount of scatter in the results. Unless otherwise indicated. AIMS.
34.. 2 H. Several specimens are tested at different values of stress increase per cycle. and T. 124. T. it must be realized that considerable devi be given in Chap. 440. Dolan. 1213. Proc. A modification of the Prot method is sometimes used w^hen a special machine equipped to provide a continuously increasing stress is not availwhen the number of specimens is not large. However. 875902. This topic is discussed in greater detail in Sec.Sec. Results obtained by this step method and the Prot method may not produce values of fatigue limit in agreement with those obtained from testing at constant stress. and the process is continued until failure occurs. Dimoff. the stress is raised by a certain amount. 54. J. Profs method has undergone considerable investigation and modification' and appears useful for the rapid determination of the fatigue limit of ferrous materials. Prot. The fatigue limit of the specimen is taken as the stress halfway between the breaking stress and the highest stress at which the specimen survived. vol. 1 M. ASTM. . Statistical Nature of Fatigue A considerable analysis of fatigue data test results. 124] Fatisuc of Metals 301 fatigue limit determined in the in error. 10^. Rev. p. Since fatigue life and fatigue limit are statistical quantities.^ In this method. Corten. because of changes which can occur in For example. pp. the metal during testing at stresses below the fatigue limit. for example. manner just described can be considerably will The statistical nature of fatigue be discussed in the next of the fatigue section. The fatigue limit is obtained from this plot by extrapolation to y/ a = 0. mH. vol. The test is run for a fixed number of cycles. each specimen is started at an initial stress below the expected value of the fatigue limit. 1954. T. Another unit of cycles is applied at this stress. Prot suggested that a linear relationship should exist between the stress at which fracture occurs and y/ a. will A amount of interest has been shown in the statistical and in the reasons for the variability in fatiguemore complete description of the statistical techniques 16. w^here a is the stress increase per cycle. and the stress is progressively increased at a constant rate until fracture occurs. An limit interesting test for obtaining a more rapid estimate than is possible by conventional means was proposed by Prot. The initial stress taken at about 70 per cent of the estimated fatigue limit. 1937. and if failure does not occur. certain metals can be strengthened b}^ "coaxing" at stresses below the able or level is fatigue limit. it is important here to gain an acquaintance with the concept of the statistical approach so that existing fatigue data can be properly evaluated.
302 Applications to Materials Testing [Chap. 12 is ation from an average curve determined with only a few specimens to be expected. It is necessary to think in terms of the probabihty of a specimen attaining a certain Hfe at a given stress or the probability of failure at a given stress in the vicinity of the fatigue limit. To do this requires the testing of considerably more specimens than in the past so that the statistical parameters^ for estimating these probabilities can be determined. The basic method for expressing fatigue data should then be a threedimensional surface representing the relationship between .i .
. An example of the errors which can be introduced by ordinary testing with a few specimens is illustrated in Fig.000 to 52. This figure summarizes^ ten SN curves determined in the conventional manner for the same bar of alloy steel. Sec. 125. Am. no. a stress above which will but below which it will not fail. 2 3 W. variability. J. T. We can only test a specimen at a particular stress. For example. each curve being based on ten specimens. 1954. Ransom. 1951. For engineering purposes accurate to assume a logarithmic normal distribution of life life fatigue at constant in the region of the probability of failure of P = 1 0. In determining the fatigue limit of a material. Appl. 121. discussion in ASTM 3. and there was no excessive scatter or uncertainty as to how to draw the SN curves. vol. distribu the fatigue sufficiently life was expressed as log N. Spec. 1 A. or normal. although it Alternative approaches have been the use of the extremevalue distribution ^ or Weibull's distribution. there is considerable difference in the measured values of the fatigue limit for the steel due to the fact that the curves were based on insufficient data. below which all specimens w^ould presumably have infinite lives. 124] Fatigue oF Metals 303 frequency distribution of tion it is if N followed the Gaussian. it is now recognized that the fatigue limit is really a statistical quantity which requires special techniques for an accurate determination. 293297. vol. pp.. frequently used. It is known that inclusions in steel have an important effect on the fatigue limit and its but even vacuummelted steel shows appreciable scatter in The statistical problem of accurately determining the fatigue limit is complicated by the fact that we cannot measure the individual value of the fatigue limit for any given specimen.90. pp. fatigue limit. WeibuU. 0. Gumbel. Freudenthal and E. 1952. Assoc. that each specimen has fail its it should be recognized it own fatigue limit. 5963. J.10 to P = less. and if the specimen fails. pp. Puhl. The fatigue limit of steel was formerly considered to be a sharp threshold value. 575597. even if it did not fail at the test stress. 18. Statist.000 psi. and that this critical stress varies from specimen to specimen for very obscure reasons. For the statistical interpretation of the fatigue limit we are concerned with the distribution of stress at a constant fatigue life. in a heattreated alloy forging steel the stress range which would include the fatigue limits of 95 per cent of the specimens could easily be from 40. 49. as can be seen from the figure. J. life tion of a lognormal distribution of is life is no longer justified. However. Since the specimen cannot be retested. M. then the stress was somewhere above the fatigue limit of the specimen. to predict the fatigue per cent or it is frequently important to be able corresponding to a probability of failure of At this extreme limit of the distribution the assump However. Tech. Yet. The specimens were as identical as it was possible to make them. Mech. J.
Tech. Publ.000 70. Summary (J. p. 61.000 60.000 Cycles to failure Fis. T. stress. Fatigue has certain things in Only a small fraction of the effort . at each making a statistical estimate of the fatigue limit are called probit analysis 6" 2 !0 9175384 3. ASTM and the staircase method. specimens at several stresses to see how many fail all that we and proposition.304 Applications to Materials Testins to estimate the statistics of the fatigue limit [Chap. 1952. each based on 10 specimens. 12 we have by testing groups of Thus. near the fatigue limit fatigue is a "gono go" means by specimens of can do is to estimate the behavior of a universe The two statistical methods which are used for of a suitable sample. drawn from the same Spec. of B'N curves. 1 25. Ransom.000 50. The procedures for applying these analysis to the determination of the fatigue hmit will methods of be given in Chap. 121. Structural Features of Fatigue devoted to fatigue research has been concerned with the study of the basic structural changes that occur in a metal when it is subjected to cyclic stress.) bar of steel.000 40. 16. 125.
Fatiguecrack propagation ordinarily transgranular. However. A. discussed in the previous paragraph. J. Metals. it appears that intrusions and extrusions are produced at local soft spots in the crystal. p. bands which are more persistent than when the other slip lines have been polished away.^ These persistent slip bands are embryonic fatigue cracks. Roy.'* Therefore. 1959. which is observed as distorted regions Cracks are usually found to occur in the regions of heavy of heavy slip. * W. Soc. A. 3 P. This H. Once formed. 242A. 19551956. pp. In many metals the increase in visible slip soon reaches a saturation value.. G. Inst. 1 An 2 . 33. the occurrence of slip during fatigue does not in itself mean that a crack will form. thousand Successive cycles produce additional slip few bands. in "Fracture. Forsyth and C. and this suggests that cross slip is needed for their formation. unidirectional deformation slip grains. Such slip bands have been observed after only 5 per cent of the total life of the specimen. pt. vol. Smith. Slip bands have been observed at stresses below the fatigue limit of ferrous materials. New York. 125] Fatigue of Metals 305 common with plastic flow and fracture under static or unidirectional The work of Gough^ has shown that a metal deforms under cyclic strain by slip on the same atomic planes and in the same deformation. Proc. although they may later take a normal to the is maximum applied tensile stress. 1933. 83. J. vol. J. The study of slipband intrusions and extrusions has been undertaken too recently to uncover all the factors responsible for their formation. Wood. vol.Sec. Slip lines are generally formed during the first cycles of stress. fatigue cracks tend to There will generally be several slip the rest and which will remain visible propagate direction initially along slip planes." John Wiley & Sons. crystallographic directions as in unidirectional strain. E. 1957. since they open into wide cracks on the application of small tensile strains. Therefore. Some Basic Studies of Fatigue in Metals. Proc. but the number of slip bands is not directly proportional to the give no evidence of number of cycles of stress. A study of crack formation in fatigue can be facilitated by interrupting the fatigue test to remove the deformed surface by electropolishing. these structural features are the origin of the persistent slip bands. deformation parallel to what was originally a slip band. Inc. (London). in fatigue is Whereas with all usually widespread throughout the some grains will show slip lines while other grains will slip. Gough. 3114. 2. Stubbington. ASTM. pp. C. or fissures. 189196. important structural feature which appears to be unique to fatigue deformation is the formation on the surface of ridges and grooves called slipband extrusions and slipband intrusions J Extremely careful metallography on tapered sections through the surface of the specimen has shown that fatigue cracks initiate at intrusions and extrusions. 395.
1957. 2 3 R. Phil. . W. which are oriented to deform only in easy glide. vol. There is essentially no strain hardening or distortion in the Xray diffraction pattern. Mag. ASTM.. A study of the dislocation structure in thin films of aluminum^ has shown that for high fatigue stresses dislocation networks are formed similar to those formed on unidirectional loading. Soc. 160166. G. L. Coarse slip bands are formed. West. This sug1 L. When the copper is fatigued in the lowstress region. H. and 4. Hargreaves. and tests carried out at low stresses. Polakowski and A. K. and there is appreciable asterism in the Xray diffraction pattern. the stored energy is released over a fairly narrow temperature range during annealing. However. M. 1959. (London). N. in the lowstress region slip lines are very fine and are dif&cult to distinguish by ordinary metallographic techniques. Clarebrough. In considering the structural changes produced by fatigue. p. 1954. 701. E. as would occur if only recovery took place. G. Palchoudhuri. An annealed metal usually undergoes moderate strain hardening with increasing cycles in the highstress region. The fact that initially coldworked copper becomes softer as a result of fatigue^ can be explained by the generation of point defects which allow the metal partly hypothesis ^ to recover by permitting dislocations to climb out of the slip plane. This represents energy release due to both recovery and recrystallization. where failure occurs in more than 10^ cycles. the stored energy is released over a wide range of temperature. M. Proc. 242A. Proc. pp. Partridge. Segall and P. Head. Roy. pp. This is a good indication that large numbers of point defects are produced during fatigue. vol. Structural features produced in the highstress region of the SN curve bear a strong resemblance to those produced by unidirectional deformation. The difference in the release of stored energy between fatigued and coldworked copper is in line with what would be expected from a large concentration of point defects. where failure occurs in less than about 10^ cycles of stress. 912919. Agehardening aluminum alloys in the precipitationhardened condition can be overaged by fatigue deformation at room temperature. 54. just as would be expected for a metal plastically deformed in tension. A. For copper specimens tested in the highstress region. There are a number of other indications that cyclic deformation results in a higher concentration of vacancies than cold working by unidirectional deformation.2 306 Applications to Materials Testing [Chap. vol. At low fatigue stresses the metal contains a high density of dislocation loops similar to those found in quenched specimens. 1 is borne out by the fact that fatigue failure is difficult to produce in certain ionic crystals which do not easily undergo cross slip and by the fact that it is not possible to produce fatigue failure in zinc crystals. it is advisable to differentiate between tests conducted at high stresses or strain amplitudes.
3 * A. 242A. 113126. ing after only a small fraction of the expected total fatigue It of a crack. 1951. Molineux. Soc. ^ 1. K. and N. Sec. It extends from the initial widespread strain hardening to the formation of a visible crack. 1956. 1957. Many of the theories that exist have been qualitative and base their acceptance mainly on the fact that the analysis yields a stresslog A'' relationship similar to the observed SN curve. The third stage of fatigue con the propagation of the crack to a size large enough to cause failure. Thompson. Mech. pp. N. M. p. J. Rosenberg. and V. N. J. The second stage comprises the major part of the fatigue life of a specimen. for many assumed mechanisms can lead to a prediction of the general shape of the fatigue curve. Solids. although the crack cannot be readily detected. this may not necessarily be a satisfactory criterion. Whittaker. Metals. Inst. However. There is considerable evidence that a fatigue crack is formed before life about 10 per cent of the total until of the specimen has elapsed. N. D. Widespread bulk deformation occurs until the metal strain hardens to the point where it can withstand the applied stress. 1 T. 357 363. Roy. 203. except by repeated electropolishing. 2 R.' The primary stage occurs only in metals where the applied stress level is above the initial static yield stress. Depending upon the stress. and Phys. 126. the first stage will last for 10^ to 10* cycles. 1. 134141. 126] Fatigue of Metals 307 gests that vacancies produced by fatigue are available to accomplish the Moreover. Broom. G. First Natl. many cycles later. .^ — 190°C. 195556. Louat. vol. McCammon and H. Sinclair and T. The principal evidence for this'*'^ is that anneallife does not has been concluded that the damage produced by this small number of cycles must be in the nature significantly increase the fatigue life. Proc. vol. pp. vol. 1953. 84. it is unlikely that our knowledge of the structural changes produced by fatigue is at all complete. M. Mag. (London). J. For one thing. Theories of Fatigue It is perhaps unnecessary to state that no mechanism or theory has been proposed which adequately explains the phenomenon of fatigue. Head. Proc. vol. H. Wadsworth. Mech. pp. However. is The process of the formation of a fatigue crack often divided into three stages. Appl. /. the diffusion of vacancies is not essential for fatigue failure. the fatigue where vacancy movement is negligible. the fact that fatigue fracture can be produced at 4°K indicates that a temperatureactivated process such as strength increases markedly on going from 20 to diffusion required for the overaging process. 647651. Congr... pp. Phil. During the second stage of fatigue the crack sists of is initiated. Dolan.
corresponding to movements of the order of 10~^ cm rather than steps of 10"^ to 10~* cm. However.3. 1939. which may be areas of favorable orientation for slip or areas of high stress concentration due to metallurgical notches such as inclusions. A. Soc. 79106. It in increases with an increase in the stress applied to the specimen. a crack is formed. has evolved a concept of fatigue failure which does not require localized strain hardening for fatigue defor mation to occur. Inst. September. He further showed that the total plastic strain (sum of positive and negative strains) converges toward a finite value as the number of cycles This limiting value of total plastic strain increases toward infinity. Metals. 171A. pp. and this forms a new localized plastic This process is repeated over region in which the process is repeated. W. cycle) fine slip might by deformation continued trates Wood's concept of how lead to a fatigue crack. A. vol. 1955. but it does not depend on any specific deformation mechanism other than the concept that fatigue deformation is heterogeneous. 1 The figures illustrate schematically the fine 2 E. was assumed that these small regions could be treated as plastic regions an elastic matrix.2 308 Applications to Materials Testing [Chap. The crack creates a stress concentration. The metal is considered to contain small. if the stress and over theory is until the crack becomes large enough so that fracture occurs on the cycle. Proc. Orowan showed that for repeated cycles of constant stress amplitude the plastic regions will experience an increase in stress and a decrease in strain as the result of progressive localized strain hardening. which are observed for static slip bands. He interprets microscopic observations of slip produced by fatigue as indicating that the slip bands are the result of a systematic buildup of fine slip movements. Roy. weak of fatigue^ Orowan's theory was one regions. 56. {London). Wood^s Concept of Fatigue W. the application of the full tensile stress of The essence of this that localized strain hardening uses up the plasticity of the metal so that fracture takes place. The existence of a fatigue limit hinges stress the total plastic strain upon the fact that below a certain fracture. pp. of the earUest generally accepted This theory leads to the prediction of the general shape of the SN curve. Orowan. . 1 Orowan's Theory explanations for the fatigue process. . Such a mechanism is believed to allow for the accom modation of the large total strain (summation of the microstrain in each Figure 126 illuswithout causing appreciable strain hardening. Bull. vol.^ who has made many basic contributions to the under standing of the mechanism of fatigue. cannot reach the critical value required for is such that the total plastic strain in the weak region exceeds the critical value. Wood. Wood.
195197. tals^ tend to support the Mott mechanism and to disprove the CottrellHull model. in "Fracture. see also A. The notch would be a stress raiser with a notch root of atomic dimensions. crack. while Mott^ has suggested one involving Fatigue experiments on ionic crysthe cross slip of screw dislocations. Cottrell and D. pp.. 1961. Wood's concept of microdeformation leading to formation of fatigue (6) fatigue deformation leading to surface notch fatigue deformation leading to slipband extrusion. vol. and E. 126c) at the surface. 1 26] Fatigue of Metals 309 structure of a slip band at magnifications obtainable with the electron static microscope. Roy. Machlin. Cottrell and HulP have suggested a model involving the interaction of edge dislocations on two slip systems. trast. . the backandforth fine slip movements of fatigue could build up notches (Fig. vol. Phil. 8. ser. .Sec. Kennedy. Soc. 1266) or ridges (Fig. N. 1958. 3 4953. F. Mag.. [c) W. 1957. Slip produced by deformation would produce a con In contour at the metal surface similar to that shown in Fig. Inc. 6. S.. Such a [b] Fig. H. (c) A. pp. J. McEvily. 242A. 1959. while other metals do not have a fatigue limit. vol. A. situation might well be the start of a fatigue crack. 211213." John Wiley & Sons. New York. Proc. (intrusion) This mechanism agreement wdth the facts that fatigue cracks start at surfaces and that cracks have been found to initiate at slipband intrusions and extrusions. pp. Acta Met. Critical Experiments on the Nature of Fatigue in Crystalline Materials. Hull. 6. J.. Mott. Theory of the Fatigue Limit One of the puzzling questions in fatigue is why certain metals exhibit an SN curve with a welldefined fatigue limit. The answer to this question appears to have been 1 2 A. (London). (a) Static deformation. 126. is for the initiation of a fatigue crack in Dislocation Models for Fatigue The growing awareness of the role played by subtle changes in surface topography in initiating fatigue cracks has led to several dislocation models for the generation of slipband intrusions and extrusions. 126a. Jr.
C. molybdenum. 1957. Their tests with mild steel showed that as the total carbon and nitrogen content was decreased. yet it does not ordinarily show strain aging in the However. pp. usually a V notch or a circular notch. only very localized strain aging is required to tension test. (London). 7 that the presence of a notch in a specimen F. C. ASTM. see also J. inclusions. ^ Levy and S. screw threads. vol. Eng. 57. titanium.. stress concentrations resulting from geometrical discontinuities. J. usually start at such geometrical irregularities. Heattreated steel exhibits a definite fatigue limit. Rally and G. T. Lipsitt and G. 1947. The effect of stress raisers on fatigue is generally studied by testing specimens containing a notch." They proposed that the fatigue limit represents the stress at which a balance occurs between fatigue damage and localized strengthening due to strain aging. . 296300. 1 27. Lowcarbon steel. press fits. Since actual machine elements invariably contain stress raisers like fillets. Sinclair. and decarburization. and aluminum7 per cent magnesium^ alloy are good examples. L. the SN curve flattened out and the knee occurred at a larger number of cycles than if the carbon content were higher. see J. so that the tendency for strain aging decreased.310 Applications to Materials Testing [Chap. 118119. M. 197. Caswell. op. Similar results were found by Lipsitt and Home. and it is quite likely that the fatigue test is more sensitive to strain aging than the tension test. local overheating in grinding. Kanitkar. and Whittaker. keyways. University of 1 Illinois. and it is holes. Iron Steel Inst. S. "Influence of Strain Aging on the Shape of the SN Diagram." Department of Theoretical and Applied Mechanics. Prod. It has been shown in Chap. Effect of Stress Concentration on Fatigue raiser such as a Fatigue strength is seriously reduced by the introduction of a stress notch or hole. 12 given by Rall}^ and Sinclair/ who noted that metals which undergo strain aging have an SN curve with a sharp knee and a welldefined fatigue limit. A. pp. not surprising to find that fatigue cracks in structural parts One of the best ways of minimizing fatigue failure is by the reduction of avoidable stress raisers through careful design^ and the prevention of accidental stress raisers by While this section is concerned with careful machining and fabrication. These factors will be considered in other sections of this chapter. 587600. affect fatigue properties. The correlation is fairly good between materials which show both strain aging and a fatigue limit. Broom. Home. 111. Urbana. pp. January. stress concentration can also arise from surface roughness and metallurgical stress raisers such as porosity. Molineux. 1955. Proc. 1961. cit.. vol. 2 H. * For examples of good design practice.
Peterson. For materials which do not exhibit a fatigue limit the fatiguenotch factor is based on the fatigue strength at a specified number of Values of Kf have been found to vary with (1) severity of the cycles.. stress to the The maximum nominal stress is the theoretical stressconcentration factor Ki. notched specimens are usually plotted in terms of nominal stress based on the net section of the specimen. (4) the type of loading. and should be carefully examined for their However. or fatiguenotch factor. (2) the type of notch. very sharp notches (high Kt) have less effect on as Kt increases. the ratio of Kf/Kt decreases Thus. If we divide both sides the nominal stress. Inc. subject to considerable scatter limitations and restrictions. two general trends are usually observed for test conditions of completely reversed loading. First. values of Kt can be computed from the theory of elasticity for simple geometries and can be determined from photoelastic measurements for more complicated situations. The effectiveness of the notch in decreasing the fatigue limit is expressed by the fatiguestrength reduction factor. The distortionenergy criterion of yielding for biaxial stress can be expressed a. we get the expression Kt' Eq. and. factors Most of the available ^ data on stressconcentration have been collected by Peterson. The effect of notches on fatigue strength is determined by comparing The data for the SN curves of notched and unnotched specimens. It is often desirable to include the effect of the biaxial state of stress at the root of a notch in the value of the stressconcentration factor. and The values of Kf published in the literature are (5) the stress level. second. 127] Fatisue of Metals effects: (1) there is 311 under uniaxial load introduces three an increase or concentration of stress at the root of the notch. (3) a triaxial state of stress ratio of the is produced. Kf. Kf is usually less than Kt. E.C + Cy^ (126) where Kt' is the stress concentration factor including both combined stress and stress concentration.)ec. This is in agreement with observations that fatigue cracks can exist in a 1 New R. This factor is simply the ratio of the fatigue limit of unnotched specimens to the fatigue limit of notched speci mens." John Wiley York. . fatigue strength than would be expected from their high value of Kt. (125) by = Kt{l . (2) a stress gradient is set up from the root of the notch in toward the center of the specimen. "Stressconcentration Design Factors. by C2)^^ = (7i(l .C+ (125) of where C = crz/ai and ao = 0. 213. (3) the material. notch. 195:1 & Sons. As was discussed in Sec.
312 Applications to Materials Testing ^ [Chap. and the nominal stress. The notch sensitivity of a material in fatigue is usually expressed by a notchsensitivity index q. vol. However. The stress gradient. while a material in which the notch has its full theoretical effect has a notchsensitivity index of g = 1. 12 specimen for millions of cycles without propagating. 501. » N. Fig. because of increased q it is possible tion size and tensile strength. this should not be interpreted as license to assume that a sharp notch or crack can be tolerated in a structure. However. the size of the specimen. 1955. This stress produced by the notch is omax. Frost. E. has an important influence on the notch sensitivity. q is not a true material constant since it varies with the severity and type of notch. Engineer. or slope of the stressdistribution curve near the root of the notch. The unnotched fatigue limit of the material is reached at a depth 5 below the root of the notch. stress neglecting the notch. under certain circumstances to decrease the fatigue performance of a theoretical factor member by increasing the hardness or tensile strength. 464. The notch sensitivity increases with secThus. Q = Ki Kt 1  or 1 Kt Kt'  1 (127) 1 where q = notchsensitivity index Kf — Kt Kt' notchfatigue factor = = = fatigue limit unnotched/fatigue limit notched theoretical stressconcentration factor = omax/cnom which combines Kt and a biaxial stress factor Equation (127) was chosen so that a material which experiences no reduction in fatigue strength due to a notch has an index of g = 0. Stress distribution at a notch in bending. is o„. 200. pp. . 1 27. and the type of loading. The maximum is (Tg. Figure 127 illustrates the stress distribution in a notched bar in bending.
Sec. 127] Fatigue of Metals 313 The stress gradient can then be written dy 5 This expression can also be written HO" in terms of the notch radius O^inax r. ^ dy r Combining these two expressions and assuming that in faikire occurs when .
if not altogether impossible. Changing the size of a fatigue specimen usually results in a variation in two factors. A precise study of this effect is difficult for several reasons. does not exceed about 10 per increases the . of the stress. say. increasing the diameter increases the volume or surface area of the specimen. Experience has shown that in most cases a size effect exists. Size Effect An important practical problem is the prediction of the fatigue performance of large machine members from the results of laboratory tests on small specimens. section size. to prepare geometrically similar specimens of increasing diameter which have the same metallurgical structure and residual stress distribution throughout the cross section. Another viewpoint is that the stress gradient across For a fine grain size the stress gradient is Geometrically similar notches will not if and the value produce the same will of q is large. For tests in reversed bending and torsion. a grain low. is the critical factor.314 Applications to Materials Testing [Chap.1 to 2 in. First. Several hypotheses have been sis made to explain the variation of notch sensitivity with notch radius. stress gradient across the grains the grain size is equal in differentdiameter specimens. effect of grain size In addition. for plain or notched specimens loaded in bending or torsion. The change in amount of surface is of significance. i. an increase in diameter usually decreases the stress gradient across the diameter and volume of material which is highly stressed. the fatigue strength of large members is lower than that of small specimens. maximum This involves a statistical argument that the probability of finding a flaw or critical crack nucleus increases with the volume of highly stressed material. Experimental data on the size effect in fatigue are contradictory and not very complete. Second. 12 is increase with specimen diameter. It is extremely difficult. The specimen with the larger diameter have the lower stress gradient across a grain. since fatigue failures usually start at the surface. while more commonly it is observed that the fatigue limit decreases with increasing diameter. Fine grain a higher q than coarse grain size. 1 28.e. some investigators have found no change in fatigue limit with specimen diameter. 5 per cent.. One hypothe assumes that failure is determined by the volume of material that is stressed to within a small percentage. The problems in fatigue testing largesized specimens are considerable. there a measurable size results in on notchsensitivity index. and there are few fatigue machines which can accommodate specimens having a wide range of cross sections. and grain size. For mild steel the decrease in bending fatigue limit for diameters ranging from 0.
121) show that the Horger's data' for steel shafts tested in reversed bending (Table fatigue limit can be appreciably reduced in large section sizes. Table IN 121 Fatigue Limit of Normalized Plaincarbon Steel Reversed Bending Specimen diam. . 129] Fatigue of Metals 315 cent.Sec.
316 Applications to Materials Testing (2) [Chap. the surface is subjected to oxidation dition of the surface.000 PSI^^ SAE Type of finish . has been recognized machining procedures fatigue performance. give the highest values in in Such carefully polished specimens are usually used Table 122 Fatigue Life of 3130 Steel Specimens Tested under Completely Reversed Stress at 95. that different surface finishes produced can appreciably affect mens. 12 roughness or stress raisers at the surface. Smoothly polished speciscratches (stress raisers) are oriented parallel with it by different stress. and corrosion. Surface Roughness Since the early days of fatigue investigations. in which the fine the direction of the principal tensile fatigue tests. and (3) changes in the residual stress conIn addition. changes in the fatigue strength of the surface metal.
Surface Residual Stress of a favorable compressive residualstress pattern at the probably the most effective method of increasing fatigue performance. the changes in dimensions when a thin layer of material is removed from a part which the surface. since favorable compressive residby these processes. 1 Thus. Reviews. since large changes in the residual stress. rather than at the surface. and hardness of the plate can be produced. 15. The amount of strengthening depends on the diameter of the part and the depth of surface hardening. it cannot be considered that the higher fatigue properties are due exclusively to the formation of higherstrength material on the surface. which act over regions which are large compared with the grain size. Electroplating of the surface generally decreases the fatigue limit of steel. 129] Fatigue of Metals ' 317 However. while a softer believed to have effect on fatigue strength. ual stresses are produced in the surface of fatigue in surfacehardened parts that the failure initiates at the inter face between the hard case and the is softer case. WiUiams. 1957. is not subjected to an external force. 165223. The effectiveness of carburizing and nitriding in improving fatigue performance is greater for cases where a high stress gradient exists. the regions which have been plastically deformed prevent the adjacent elastic regions from undergoing complete elastic recovery to the unstrained condition. Consider a metal specimen where the surface has been deformed in tension by bending so that part of it has undergone plastic deformation. cadmium plating is The particular plating conditions used to produce an electroplated surface can have an appreciable effect on the fatigue properties. . 5. 1960. The subject of residual stress will be considered in greater However. The greatest percentage increase in fatigue performance is found when notched fatigue specimens are nitrided. Metals Park. are They can be measured by Xray methods or by noting considered here. Ohio. 2 A detailed review of the effect of electroplating on fatigue strength is given by R. A. Improvements in fatigue properties similar to those caused by carburizing and nitriding may also be produced by It is a general characteristic flame hardening and induction hardening. When the external force is removed. carburizing and nitriding. sidered that residual stresses are lockedin stresses which are present in The formation is surface Only macrostresses. porosity. pp.^ Sec. than in an axial fatigue test. the elastically "Fatigue Durability of Carburized Steel. R. as in bending or torsion." American Society for Metals. Residual stresses arise when plastic deformation is not uni form throughout the entire cross section of the part being deformed. it can be condetail in Chap. Chromium plating particularly difficult to accomplish without impairment of fatigue little properties. vol. Hammond and C. Met. for the present discussion. adhesion.
318 Applications to Materials Testing [Chap. the region which was plastically Tension Connpression ^crmo ^^/? ^l ^o> id) Fig. FigFigure 129 illustrates this effect. while the region tensile residual stress maximum For compression will removed. cross section plastically part of the is deformed while the where situation rest undergoes elastic deformation. deformed have a in tension will ing. beam with no residual A typical residualstress distribution. The value of residual stress which can be produced is equal to the in have a compressive residual which was deformed plastically stress after unload when the external force is elastic limit of the metal. Superposition of applied and residual stresses. 12 deformed regions are left in residual tension. and the regions which were plastically deformed must be in a state of residual compression to balance In general. to an externally applied tensile stress on that surface decreases the likelihood stresses of fatigue failure at that point. 1 29. for a the stresses over the cross section of the specimen. which exists at a point on the surface. Thus. the addition of a compressive residual stress. such as would be produced . many ure 129a shows the elasticstress distribution in a stress. purposes residual stresses can be considered identical to the produced by an external force.
the metal due to strain hardening occur during these processes. Care must be taken to ensure uniform coverage over the stresses distribution due to the algebraic summation and the residual stresses is Frequently an additional improvement in fatigue by carefully polishing the shotpeened surface to reduce the surface roughness. presumably because the surface is such a potential source of weakness. 129] Fatigue of Metals is 319 by shot peening. particularly adapted to large parts. The severity of the stress produced by shot peening is frequently controlled by measuring the residual deformation of shotpeened beams called Almen strips. shape. Shot peening consists in projecting fine steel or castiron shot against the surface at high velocity. However. properties can be obtained resulting It is from carburizing. it is believed that the improvement in fatigue performance is due chiefly to Surface rolling is the formation of surface compressive residual stress. It should also be apparent that the improvements in fatigue performance which result from the introduction of surface compressive residual stress will be greater when the loading is one in which a stress gradient exists than when no stress gradient is present. It is possible to damage the surface by excessive peening or rolling. The peak tensile stress is displaced to a The magnitude of this stress point in the interior of the specimen. It is particularly adapted to massproduced parts of fairly small size. bending shown. conditions. some improvement in the fatigue performance of axial loaded fatigue specimens results from surface compressive residual stresses. nitriding. The principal variables in this process are the shot velocity and the size. 129c the stress of the external stresses over the interior of the cross section. Experience and testing are required to establish the proper conditions .Sec. It is frequently used in critical regions such as the fillets of crankshafts and the bearing surface of railroad axles. shown in Fig. depends on the gradient of applied stress and the residualstress distriThus. Note that the maximum tensile stress at the surface is reduced by an amount equal to the surface compressive residual stress. The chief commercial methods of introducing favorable compressive residual stresses in the surface are by surface rolling with contoured Although some changes in the strength of rollers and by shot peening. and hardness of the shot. Other methods of introducing surface compressive residual stresses are by means of thermal stresses produced by quenching steel from the tempering temperature and from stresses arising from the volume changes accompanying the metallurgical changes area to be treated. important to recognize that improvements in fatigue properties do not automatically result from the use of shot peening or surface rolling. residual stresses at the surface Note that the high compressive must be balanced by tensile residual In Fig. 1296. subsurface initiation of failure is possible under these bution. and induction hardening.
Corrosion Fatisue The simultaneous action as corrosion fatigue. pp.000 to 12. the higher the testing speed. a very pronounced reduction in fatigue properties results which is greater than that produced by prior often produces pitting of metal surfaces. Corrosionfatigue tests may be carried out in two ways. ASTM. when tests are made in a corrosive environment there is a definite dependence on testing speed. In the usual method the specimen is iL. Thus. However. 2 W. . vol. tested in air at Materials which show a definite fatigue limit when room temperature show no indication of a fatigue limit carried out in a corrosive environment. P.320 Applications to Materials Testing [Chap. 57. Letner. vol. and this stress pattern may persist at preparation It is is responsible for much of the scatter in fatiguetest results. while the possibility of fading of residual stress during service should be recognized. important to realize that residualstress patterns may be modified by plastic deformation or by thermal activation. 1957. T. 12 which produce the optimum residualstress distribution. In general. the smaller the damage due to corrosion. S. cer tain metallurgical processes yield surface tensile residual stresses. corrosion of the surface. the chemical attack greatly accelerates the rate at and fatigue occur simultanewhich fatigue cracks propagate. 601622. for preparing fatigue specimens can result in appreciable surface residual It is quite likely that lack of control of this factor in specimen stress.^ of cyclic stress and chemical attack is known Corrosive attack without superimposed stress The pits act as notches and produce a reduction in fatigue strength. by P. Met. Proc. over a range from about 1. surface tensile stresses are produced Further. when corrosive attack occurs simultaneously with fatigue loading. Gilbert. the polishing^ methods ordinarily used the grinding conditions. compressive surface residual stresses can be produced. Tarasov. 1210.000 cycles/min. When corrosion ously. Hyler. An extensive review of the literature on this subject has been prepared 1. depending upon Further. The data on "fading" of residual stress are very meager and not too reliable. when the test is While ordinary fatigue tests in air are not affected by the speed of testing. it is possible for periods of overload or periods of increased temperature to result in some relief of residual stress. it does not prohibit the use of compressive residual stress as the most effective method of combating fatigue failure. by quenching deephardening low tempering temperatures. 379417. R. and H. pp. Thus. Grinding of hardened steel requires particular care to prevent a large It has been shown ^ that either tensile or decrease in fatigue properties. 1956. Reviews. Since corrosive attack is a timedependent phenomenon. steel.
R. aided by the disruption of the oxide film by cyclic strain. Wadsworth. 1."^ Separate tests in oxygen and water vapor showed little decrease over the fatigue strength in vacuum. p.^ The action of the cyclic stress causes localized disruption of the surface test is interrupted after a certain period is duced Many more small pits occur in corrosion fatigue than in corrosive attack in the absence of stress. vol. Tests of the last type have helped to establish the mechanism of corrosion fatigue. . Metals. Simnad. 188A. vol. and N. Gough and D. There is evidence to indicate that even fatigue tests in air at room temperature are influenced by corrosion fatigue. 113126. Zinc and cadmium coatings on steel and aluminum coatings on Alclad aluminum alloys are successful for many corrosionfatigue applications. the choice of a material for this type of service should be based on its corrosionresistant properties rather than the conventional fatigue properties. Louat. Inst. U. N. G. more anodic than the rest of the metal so that corrosion proceeds inward. stainless steel. T.. or beryllium copper would probably give better service than heattreated steel. Proc. Subsequent work with copper^ showed that the fatigue life was much longer in oxygenfree nitrogen than in air. (London). bronze.)CC. Mag. The cyclic stress will also tend to remove or dislodge any corrosion prodThe bottoms of the pits ucts which might otherwise stifle the corrosion. 1946. pp. oxide film so that corrosion pits can be produced. N. In general. Cracking will occur when the pit becomes sharp enough to produce a high stress are concentration. Phil. 72. 1 medium. J. H. J. indicating that the relative humidity may be a variable to consider in fatigue testing. pp. and the damage which was proevaluated by determining the remaining life in air. Thompson. It was concluded that water vapor acts as a catalyst to reduce the fatigue strength in air. vol. 372. Protection of the metal from contact with the corrosive environment by metallic or nonmetallic coatings is successful provided that the coating does not become ruptured from the cyclic strain. Nitriding is particularly effective in combating and shot peening has been used with success under cer 2 3 Evans and M. Sopwith. 1956. Fatigue tests on copper showed that the fatigue strength was higher in a partial vacuum than in air. 1210] Fatigue of Metals 321 continuously subjected to the combined influences of corrosion and cyclic In the twostage test the corrosion fatigue stress until failure occurs. 415421. Soc. 1947. Roy. Thus. even though these ducted when tests are concompressive residual stresses tends to keep surface notches from opening up and giving ready access coatings may cause a reduction in fatigue strength in air. A number of methods are available for minimizing corrosionfatigue damage. Metallographic observation showed that the development of persistent slip bands was slowed down when tests were made in nitrogen. The formation of surface to the corrosive corrosion fatigue.
tion of the coefficient of friction between the mating parts may be beneficial. ing the force normal to the surfaces may accomplish this. but this is frequently difficult to do related to wear than to corrosion fatigue. Teed. Oxidation of the metal surface occurs and the Although oxide film is destroyed by the relative motion of the surfaces.^' 1 Several excellent reviews of this R. fatigue must be considered.2 322 Applications to Materials Testing [Chap. of stress concentrators by careful design is very important when corrosion tain conditions. 11571172. when two surfaces in conThe phenomenon is more However. 1955. Proc. Exclusion of the atmosphere from the two surfaces will reduce fretting. Fretting is caused by a combination of mechanical and chemical effects. 1960. pp. Fretting Fretting is the surface damage which results tact experience slight periodic relative motion. as is demonstrated by relative motion between two nonoxidizing gold surfaces. oxidation is not essential to fretting. vol. when conditions are such that oxidation can occur fretting damage is many times more severe. Waterhouse. then fretting will not occur. Met. 267295. since the chief problem is maintaining a lubricating film for a long period of time. {London). subject have been published. Mech. . Reviews. by the facts that the relative velocity of the with a high degree of effectiveness. There are no completely satisfactory methods of preventing fretting. pp. although they may be obscured from observation by the surface debris. Inst. 1 In closed systems it is possible to reduce the corrosive Finally. Surface pitting and deterioration occur. usually accompanied by an oxide debris (reddish for steel and black for aluminum). Metal is removed from the surface either by a grinding action or by the alternate welding and tearing away of the high spots. The removed particles become oxidized and form an abrasive powder which continues the destructive process. Increasing the wear resistance of the surfaces so as to reduce surface welding is another approach. P. Fatigue cracks often start in the damaged area. it differs from wear two surfaces is much lower than is usually encountered in wear and that since the two surfaces are never brought out of contact there is no chance for the corrosion products Fretting is frequently found on the surface of a shaft to be removed. B. Encjrs. the elimination attack by the addition of a corrosion inhibitor. 169. then reducstopped. Solid lubricants such as MoS are most successful. but the damage increases with the normal force up to the point where relative motion is If relative motion cannot be completely eliminated. vol. with a pressfitted hub or bearing. L. 2 5. IncreasIf all relative motion is prevented.
^.1211 Fatigue of Metals 323 ^.^ .
discussion in Proc. Basically. this diagram shows the variation of the limiting range of stress. by drawing straight lines from the fatigue limit for completely reversed stress (which is usually available from the literature) to the tensile strength. stress (To for practical purposes testing is usually stopped lie exceeded. . 2 New York." 9th ed. common type Note that as the mean stress becomes more tensile the allowable range of d„ stress is reduced. 1954. T. 1211 omax and a^^^ lines. Compression^ > Tension Mean stress ct^^j ^ Completely reversed stress data plotted here Fig.324 Applications to Materials Testing [Chap. pp. 54. Goodman is diagram. 1210. The test data usually below the Fig. omax — o'min with mean stress. respectively. "Mechanics Applied to Engineering. steel tested in axial fatigue indicate that the allowable stress range increases with increasing compressive mean stress up to the yield stress Fig. J. A diagram similar to 1211 may strength at any given number of where the mean stress cycles. Inc. in lieu of actual test data. 1930. vol.. 12 Early contributions to this problem were made by Goodman. may actually be curves. until at the tensile strength the stress range is zero.. 1 21 1 . Green & Co. ^ John Goodman.. 847848. Goodman diagram may be constructed for the fatigue Very few test data exist for conditions Data^ for SAE 4340 is compressive. Ransom. so when the yield somewhat above and that these lines shown on A conservative approximation of the be obtained. Figure 1211 shows one of Goodman diagram which can be constructed from fatigue data of the type illustrated in Fig. However. Longmans. ASTM.^ so that curves which show the dependence of limiting range of stress on stress are frequently called mean Goodman diagrams.
alternative This is method of presenting meanstress data is shown sometimes known as the HaigSoderberg diagram. while the parabolic curve was proposed by Gerber. the linear relationship is usually preferred in engineering These relationships Co may be expressed by the following equation. Figure 1212 is obtained for alternating axial or bending stresses with static tension or compression or alternating torsion with static tension. (1210) — Ce fe)' . 1955. R. is alternating stress plotted against the relationship follows the suggestion of The mean stress. ^ residual stress increases the fatigue limit.r where Oe is a: = 1 for the Goodman line. line. Sines.has shown that these results can be rationalized 1 C. NACA Tech. for alternating torsion with static torsion or alternating bending with static torsion there is no effect of the static stress on the allowable range of alternating stress provided that the static yield strength is not exceeded. Failure of Materials under Combined Repeated Stresses with Superimposed Static Stresses. G. 1211] in compression. 1930. Sines. 1212. An 1212. Note 3495. 2 . Test data for ductile metals generally fall However. 52. (1210). 1212. However. vol. ASME. the design is based on the yield strength. because of the scatter in the results and the fact that tests on notched specimens fall closer to the closer to the parabolic curve. Goodman design. Gerber parabola Goodman line Compression Fig. Trans. Soderberg. and If the fatigue limit for completely reversed loading. Alternative method of plotting the Goodman diagram. APM522. = 2 for the Gerber parabola. A straightline Goodman.Sec. Fatigue of Metals 325 This is in agreement with the fact that compressive in Fig. as indicated by the dashed Soderberg line in Fig. then ao should be substituted for o„ in Eq.
ble to predict the fatigue performance of a combined stress system by substituting the fatigue strength for simple types of loading into the Many equation of the failure criterion. pp. The constants ae and C2 can be evaluated at any value of cycles to failure. W. the peris Finally. 417440. . vol. vol. For brittle ing under a complex state of stress. stress fatigue failure are fewer ^ materials a of failure. N. (1211) is greater than the right side. Fatigue tests with different certain generalizations can be made. ^2)2 + = = = =  a^y + (c73  ai)Y' =  C^iS. it is the slope of the Goodman line in Fig. 1212 Fatigue failure will occur if the left side of Eq. the static stress. it should be possialternating and steady components of stress. 14.. Proc. Engrs. 1956. 1949. op. 35cit. Sy. The effect of residual stresses can be introduced by adding these terms to the static stresses on the right side of the equation.) (1211) a 2. (London). combinations of bending and torsion show that for ductile metals the distortionenergy criterion provides the best overall fit. 1. Proc. no. J. Exptl. When *Si + S2 is positive. Fatisue under Combined Stresses machine parts must withstand complex loadings with both Ideally.e. " Gough. Inst. Mech. (1211) shows that compressive residual stress will allow a greater alternating stress for the 1 same fatigue life.326 Applications to Materials Testing : [Chap. Eq. H. pp. Stress Anal. Thus. 4= [(^1 V2 where ai. Sz (Te fatigue strength for completely reversed stress C2 a material constant giving variation of aa versus am'. Sines. Mathur. 160. just as the yield stress in tension can be used with the distortionenergy criterion of failure to predict static yield Although the data on combinedand less reliable than for static yielding. 1 21 2. manner The planes of maximum alternating shear stress and the static normal stresses Si and S2 on these planes are established. 12 in the following are determined. 03 alternating principal stresses static stresses Sx. When Si + S2 is negative. N. maximum principal stress theory provides the best criterion Sines^ has suggested a failure criterion that accounts for the effect of combined stresses and the influence (cr2 of a static mean <^e stress. ^2 has no effect on Ca regardless of the applied static + zero. + Sy + S. Findley and 46. Soc. an increase in static stress reduces the permissible value of missible value of aa stress aa. i. when Si is increased. P.
in other applications complex loading conditions. Overstressing and Understressing The conventional fatigue test subjects a specimen to a fixed stress amplitude until the specimen fails. Thus. 1213] Fatigue of Metals 327 1213. Proc. On the other hand. for a constant difference test stress. Further. for a for a cycle ratio at the prestress of is AB/ AC EF/DF. Special fatigue tests which apply a random load have been developed for these cases. 122c. 1 2 J. Conf. B. 1956. reduction in fatigue life at the test stress. Proc. 532541. ^ Overstressing is the process of testing a virgin specimen for some reFig. Tests may be made at a number of different values of stress to determine the SN curve. and then subsefailure at ing fatigue stressing. The gin fatigue mean virThe damage stress produced by a given cycle ratio of overstress is frequently evaluated by the ratio of the number life of cycles of overstress at the prestress to the at this same is called the cycle ratio. and the loading cannot be assumed to vary sinusoidally. Inter. pp. For these conditions it is difficult to arrive at an Prestress average stress level. many practical applications where the cyclic stress does not remain constant. K. but instead there are periods when the stress is either above or below some average design stress level. 1 21 3. vol. . Fatigue of Metals. log /!/ number of cycles less than that Test procedure for determin quired for failure. London. such as are illustrated in Fig. 45. The initial stress to which the specimen is subjected is called the prestress. the reduction in life at the test stress will usually be a lower percentage than the cycle ratio. damage produced by over quently running the specimen to another stress. 1213. EF represents the given cycle ratio at a prestress lower than the test stress. 1945. at a stress above the fatigue limit.Sec. A. but each time the maximum there are stress is held constant until the test is completed. pp. However. The higher the value of the test stress. are encountered. 301303. Kommers. the between prestress and more nearly the damage at the test stress equals the cycle ratio at the prestress. ASTM. Head. Extensive experience^ with this type of test has shown that a high prestress followed by a lower test stress causes a higher percentage of damage at the test stress than the reduction in life used up by the cycle ratio. referring to Fig. while the final stress is the test stress. the damage at the test stress amount which the virgin fatigue life at the test stress has been reduced by the overstress.
2
328
Applications to Materials Testing
of
[Chap.
1
measuring the damage due to overstressing is to subspecimens to a certain cycle ratio at the prestress and then determine the fatigue Hmit of the damaged specimens. Bennett^ has shown that increasing the cycle ratio at the prestress produces a
ject a
Another way
number
of
greater decrease in the fatigue limit of damaged specimens, while similar experiments^ which employed a statistical determination of the fatigue
limit
showed a much greater reduction
In line with this point,
statistical
in the fatigue limit
it
overstressing.
is
it
because of the
used.
nature of fatigue,
due to the important to note that, is very difficult to reach
reliable conclusions
by overstressing
tests, unless statistical
methods are
If a specimen is tested below the fatigue limit, so that it remains unbroken after a long number of cycles, and if then it is tested at a higher stress, the specimen is said to have been under stressed. Understressing frequently results in either an increase in the fatigue limit or an increase in the number of cycles of stress to fracture over what would be expected for virgin specimens. It has frequently been considered that the improvements in fatigue properties due to understressing are due to localized strain hardening at sites of possible crack initiation. A different interpretation of the understressing effect has resulted from experiments on the statistical determination of the fatigue limit. ^ Specimens which did not fail during the determination of the fatigue limit showed greater than normal fatigue lives when retested at a higher stress. By means of statistical analysis it was possible to show that the observed lives at the higher stress were to be expected owing to the elimination of the weaker specimens during the prior testing below the fatigue limit. Thus, it was concluded that understressing was at least partly due to a
statistical selectivity effect.
If a specimen is tested without failure for a large number of cycles below the fatigue limit and the stress is increased in small increments after allowing a large number of cycles to occur at each stress level, it is found that the resulting fatigue limit may be as much as 50 per cent greater than the initial fatigue limit. This procedure is known as coaxing. An extensive investigation of coaxing"* showed a direct correlation between a strong coaxing effect and the ability for the material to undergo strain aging. Thus, mild steel and ingot iron show a strong coaxing effect, while
aluminum alloys, and heattreated lowalloy improvement in properties from coaxing.
brass,
steels
show
little
1
J.
2 3 '
A. Bennett, Proc. ASTM, vol. 46, pp. 693714, 1946. G. E. Dieter, G. T. Home, and R. F. Mehl, NACA Tech. Note 3211, 1954. E. Epremian and R. F. Mehl, Spec. Tech. Publ. 137, 1952. G. M. Sinclair, Proc. ASTM, vol. 52, pp. 743758, 1952.
ASTM
Sec. 1214]
Fatigue of Metals
329
1214. EFfect of Metallursical Variables on Fatigue Properties
The fatigue properties of metals appear to be quite structuresensitive. However, at the present time there are only a limited number of ways in which the fatigue properties can be improved by metallurgical means. By far the greatest improvements in fatigue performance result from design changes which reduce stress concentrations and from the intelligent use of beneficial compressive residual stress, rather than from a change in material. Nevertheless, there are certain metallurgical factors which must be considered to ensure the best fatigue performance from a particular metal or alloy. Fatigue tests designed to measure the effect of some metallurgical variable, such as special heat treatments, on fatigue performance are usually made with smooth, polished specimens under completely reversed stress conditions. It is usually assumed that any changes in fatigue properties due to metallurgical factors will also occur to about the same extent under more complex fatigue conditions, as with notched specimens under combined stresses. That this is not always the case is shown by the notchsensitivity results discussed
previously.
Fatigue properties are frequently correlated with tensile properties.
In general, the fatigue limit of cast and wrought steels
is
approximately
is
50 per cent of the ultimate tensile strength.
the fatigue ratio.
The
ratio of the fatigue
limit (or the fatigue strength at 10^ cycles) to the tensile strength
called
Several nonferrous metals such as nickel, copper, and
fatigue ratio of about 0.35.
is
magnesium have a
lations of this type
While the use
of corre
should be clearly understood that these constant factors between fatigue limit and tensile strength are only approximations and hold only for the restricted condition of smooth,
convenient,
it
polished specimens which have been tested under zero
mean
stress at
room temperature.
steel will
For notched fatigue specimens the fatigue
ratio for
be around 0.20 to 0.30.
Several parallels can be drawn between the effect of certain metallurgical variables
on fatigue properties and the
effect of these
same variables
on
on
tensile properties.
The
effect of solidsolution alloying additions
the fatigue properties of iron^ and aluminum^ parallels nearly exactly
their effect
on the
tensile properties.
The
fatigue strength of nonferrous
size.
metals^ and annealed steel increases with decreasing grain
size
Grain
does not have an important influence on the unnotched fatigue
1
2
J.
E. Epremian and E. F. Nippes, Trans. ASM, vol. 40, pp. 870896, 1948. W. Riches, O. D. Sherby, and J. E. Dorn, Trans. ASM, vol. 44, pp. 882895,
1952.
3
G.
M.
Sinclair
and W.
J.
Craig, Trans.
ASM,
vol. 44, pp. 929948, 1952.
330
Applications to Materials Testing
[Chap. 12
Gensamer^ showed that the fatigue hmit of a eutectoid steel increased with decreasing isothermalreaction temperature in the same fashion as did the yield strength and the tensile However, the greater structure sensitivity of fatigue properstrength. ties, compared with tensile properties, is shown in tests comparing the
properties of heattreated steels.
fatigue limit of a plaincarbon eutectoid steel heattreated to coarse
pearlite
and to spheroidite
of the
same
tensile strength. ^
Even though
the steel in the two structural conditions had the same tensile strength, the pearlitic structure resulted in a significantly lower fatigue limit due to the higher notch effects of the carbide lamellae in pearlite.
mum
In general, quenched and tempered microstructures result in the optifatigue properties in heattreated lowalloy steels. However, at a
hardness level above about Re 40 a bainitic structure produced by austempering results in better fatigue properties than a quenched and
tempered structure with the same hardness.^ Electron micrographs indicate that the poor performance of the quenched and tempered structure
150
140
130
120
'no
100
90
80
70
60
50
40
Sec. 1214]
Fatigue of
Metah
331
i
.
.
332
Applications to Materials Testing
[Chap. 12
tests
direction of wrought products.
However,
have shown^ that the
It
may be only has been established that practically all the fatigue failures in transverse specimens Nearly complete elimination of inclustart at nonmetallic inclusions. sions by vacuum melting produces a considerable increase in the transfatigue Hmit in the transverse direction of steel forgings
60 to 70 per cent of the longitudinal fatigue
limit.
verse fatigue limit (Table 123).
The low transverse SAE
fatigue limit in
Table 123 Influence of Inclusions on Fatigue Limit of
4340 Steelj
Electricfurnace
Vacuummelted
139,000 120,000 0.86 29
melted
Longitudinal fatigue limit, psi Transverse fatigue limit, psi Ratio transverse/longitudinal. Hardness, Re
.
116,000 79,000 0.68 27
t
Determined
in repeatedbending fatigue test
{R
=
0).
Data from
J.
T. Ransom,
Trans.
ASM,
vol. 46, pp.
12541269, 1954.
steels containing inclusions is generally attributed to stress concentration
at the inclusions, which can be quite high
stringer
is
when an elongated
inclusion
oriented transverse to the principal tensile stress.
However,
melt
the fact that nearly complete elimination of inclusions by
ing
still
vacuum
results in appreciable anisotropy of the fatigue limit indicates
that other factors
subject have
limit
may
be important.
Further investigations^ of this
in the transverse fatigue
shown that appreciable changes
which cannot be correlated with changes in the type, number, or Transsize of inclusions are produced by different deoxidation practices. verse fatigue properties appear to be one of the most structuresensitive
^engineering properties.
1215. Effect of Temperature on Fatigue
Lowtemperature Fatigue
^how that the
atures, there
Fatigue tests on metals at temperatures below room temperature fatigue strength increases with decreasing temperature. Although steels become more notchsensitive in fatigue at low temperis
no evidence to indicate any sudden change
in fatigue
properties at temperatures below the ductiletobrittle transition temper1
J. J.
T.
Ransom and R.
F. Mehl, Proc.
ASTM,
vol. 52, pp. 779790, 1952.
2
T. Ransom, Trans.
ASM,
vol. 46, pp.
12541269, 1954.
3 G. E. Dieter, D. L. Macleary, and J. T. Ransom, Factors Affecting Ductility and Fatigue in Forgings, "Metallurgical Society Conferences," vol. 3, pp. 101142, Interscience Publishers, Inc., New York, 1959.
Sec. 1215]
atiire.
Fatigue of Metals
333
The
fact that fatigue strength exhibits a proportionately greater
increase than tensile strength with decreasing temperature has been inter
preted as an indication that fatigue failure at
room temperature
is
associ
ated with vacancy formation and condensation.
Hightemperature Fatigue
In general, the fatigue strength of metals decreases with increasing temperature above room temperature. An exception is mild steel, which shows a maximum in fatigue strength at 400 to 600°F. The existence of a maximum in the tensile strength in this temperature range due to strain aging has been discussed previously. As the temperature is increased well above room temperature, creep will become important and at high temperatures (roughly at temperatures greater than half the melting point) it will be the principal cause of failure. The transition from fatigue failure to creep failure with increasing temperature will result in a change in the type of failure from the usual transcrystalline fatigue
the
At any given temperature with increasing mean stress. Ferrous materials, which ordinarily exhibit a sharp fatigue limit in roomtemperature tests, will no longer have a fatigue limit when tested at temperatures above approximately 800°F. Also, fatigue tests at high temperature will be dependent on the frequency of stress application.
failure to the intercrystalline creep failure.
amount
of creep will increase
It is
customary to report the
total time to failure as well as the
number
of cycles to failure.
In general, the higher the creep strength of a material, the higher its hightemperature fatigue strength. However, the metallurgical treatment which produces the best hightemperature fatigue properties does not necessarily result in the best creep or stressrupture properties. This
has been shown by Toolin and
number
ties
MocheP in hightemperature tests of a Fine grain size results in better fatigue properat lower temperatures. As the test temperature is increased, the
of superalloys.
difference in fatigue properties
between coarse and
fine grain material
decreases until at high temperatures, where creep predominates, coarse
grain material has higher strength.
In general, wrought alloys show
while castings are often more
somewhat superior
resistant to creep.
fatigue
resistance,
Procedures which are successful in reducing fatigue failures at room temperature may not be effective in hightemperature fatigue. For example, compressive residual stresses may be annealed out before the operating temperature is reached.
Thermal Fatigue
The
1
stresses
necessarily need to
P. R. Toolin
which produce fatigue failure at high temperature do not come from mechanical sources. Fatigue failure can
L. Mochel, Proc.
and N.
ASTM,
vol. 47, pp. 677694, 1947.
334
Applications to Materials Testing
[Chap. 12
be produced by fluctuating thermal stresses under conditions where no
produced by mechanical causes. Thermal stresses result in dimensions of a member as the result of a temperFor the simple ature change is prevented by some kind of constraint. case of a bar with fixed end supports, the thermal stress developed by a temperature change AT is
stresses are
when the change
a
= aE AT
(1212)
where a
= — E
linear thermal coefficient of expansion
elastic
modulus
is
If failure
occurs by one application of thermal stress, the condition
called thermal shock.
However,
if
failure occurs after repeated appliit is
cations of thermal stress, of a lower magnitude,
called thermal fatigue.^
Conditions for thermalfatigue failure are frequently present in hightemperature equipment. Austenitic stainless steel is particularly sensi
phenomenon because of its low thermal conductivity and high thermal expansion. Extensive studies of thermal fatigue in this material have been reported.^ The tendency for thermalfatigue failure appears related to the parameter a/k/Ea, where o/ is the fatigue strength at the mean temperature and k is the thermal conductivity. A high value of this parameter indicates good resistance to thermal fatigue. An excellent review of the entire subject of hightemperature fatigue has been prepared
tive to this
by Allen and
Forrest.^
BIBLIOGRAPHY
Battelle
Memorial
Inc.,
New
:
Cazaud, R. London, 1953. "The Fatigue of Metals," Institution of Metallurgists, London, 1955. Freudenthal, A. M. (ed.): "Fatigue in Aircraft Structures," Academic Press, Inc., New York, 1956. Grover, H. J., S. A. Gordon, and L. R. Jackson: "Fatigue of Metals and Structures," Government Printing Office, Washington, D.C., 1954. (Contains very complete compilation of fatigue data.) Pope, J. A. (ed.): "Metal Fatigue," Chapman & Hall, Ltd., London, 1959. Proceedings of the International Conference on Fatigue of Metals, London, 1956.
Sines, G.,
Inc.,
Institute, "Prevention of Fatigue of Metals," John Wiley & Sons, York, 1941. (Contains very complete bibliography up to 1941.) "Fatigue of Metals," translated by A. J. Fenner, Chapman & Hall, Ltd.,
and
J.
L.
Waisman
J.
(eds.):
"Metal Fatigue," McGrawHill Book Company,
vol. 7, no. 25,
New
N.,
York, 1959.
Thompson,
and N.
Wadsworth: Metal Fatigue, Advances inPhys.,
pp. 72169, January, 1958.
^
Failure of metals like
coefficients
uranium which have highly anisotropic thermalexpansion mnder repeated heating and cooling is also called thermal fatigue.
2L. F. Coffin, Jr., Trans. ASME, vol. 76, pp. 931950, 1954. " N. P. Allen and P. G. Forrest, Proc. Intern. Conf. Fatigue of Metals. London, 1956, pp. 327340.
Chapter 13
CREEP
AND
STRESS RUPTURE
131. The Hightemperature Materials Problem
has been mentioned that the strength Since the mobility of atoms increases rapidly with temperature, it can be appreciated that diffusioncontrolled processes can have a very significant effect on highIn several previous chapters
it
of metals decreases with increasing temperature.
temperature mechanical properties.
greater mobility of dislocations
High temperature will also result in by the mechanism of climb. The equi
librium concentration of vacancies likewise increases with temperature.
deformation mechanisms may come into play at elevated temperIn some metals the slip system changes, or additional slip systems are introduced with increasing temperature. Deformation at grain boundaries becomes an added possibility in the hightemperature deformation of metals. Another important factor to consider is the effect of prolonged exposure at elevated temperature on the metallurgical stability For example, coldworked metals will recrystallize of metals and alloys. and undergo grain coarsening, while agehardening alloys may overage and lose strength as the secondphase particles coarsen. Another important consideration is the interaction of the metal with its environment at high temperature. Catastrophic oxidation and intergranular penetration of oxide must be avoided. Thus, it should be apparent that the successful use of metals at high temperatures involves a number of problems. Greatly accelerated alloydevelopment programs have produced a number of materials with improved hightemperature properties, but the everincreasing demands of modern technology require materials with even better hightemperature strength and oxidation resistance. For a long time the principal hightemperature applications were associated with steam power plants, oil refineries, and chemical plants. The operating temperature in equipment such as boilers, steam turbines, and cracking units seldom exceeded 1000°F. With the introduction of the gasturbine engine, requirements developed for materials to operate in critically stressed parts, like turbine
atures.
New
335
336
Applications to Materials Testing
[Chap. 13
of more powerful Rocket engines and ballisticmissile nose cones present much greater problems, which can be met only by the most ingenious use of the available hightemperature materials and the development of still better ones. There is no question that the available materials of construction limit rapid advancement in
buckets, at temperatures around 1500°F.
The design
engines has pushed this limit to around 1800°F.
hightemperature technology. An important characteristic of hightemperature strength is that it must always be considered with respect to some time scale. The tensile properties of most engineering metals at room temperature are independent of time, for practical purposes.
results
if
It
makes
is
little
difference in the
it
the loading rate of a tension test
test.
such that
requires 2 hr or
2 min to complete the
Further, in roomtemperature tests the
is
anelastic behavior of the material
of little
practical consequence.
However, at elevated temperature the strength becomes very dependent on both strain rate and time of exposure. A number of metals under these conditions behave in many respects like viscoelastic materials. A metal subjected to a constant tensile load at an elevated temperature will creep and undergo a timedependent increase in length. The tests which are used to measure elevatedtemperature strength must be selected on the basis of the time scale of the service which the material must withstand. Thus, an elevatedtemperature tension test can provide useful information about the hightemperature performance of a shortlived item, such as a rocket engine or missile nose cone, but it will give only the most meager information about the hightemperature performance of a steam pipeline which is required to withstand 100,000 hr
of elevatedtemperature service.
Therefore, special tests are required to
evaluate the performance of materials in different kinds of hightemperature service.
The
creej) test
measures the dimensional changes which
test
occur from elevatedtemperature exposure, while the stressrupture
measures the effect of temperature on the longtime loadbearing characteristics. Other tests may be used to measure special properties such as thermalshock resistance and stress relaxation. These hightemperature tests will be discussed in this chapter from two points of view. The engineering significance of the information obtained from the tests will be discussed, and information which is leading to a better understanding of the mechanism of hightemperature deformation will be
considered.
132. The Creep Curve
The
creep.
progressive deformation of a material at constant stress
is
called
The
simplest type of creep deformation
is
viscous flow.
A
mate
Sec. 132]
rial is
Creep and
if
Stress
Rupture
337
is
said to experience pure viscous flow
the rate of shear strain
proportional to the applied shearing stress.
^/«
When
by a simple constant, the material
T
is
(131)
the proportionality between these two factors can be expressed
said to
show Newtonian
viscosity.
=
dy
V
(132)
It
where
??
is
the coefficient of viscosity.
Most
liquids
obey Newton's law
of viscosity,
but
it is
only partially followed by metals.
Secondary creep
Tertlory creep
Primary creep
I
H<
n
y Fracture
Time
Fig. 1 31
.
t
Typical creep curve showing the three stages of creep.
Curve A, constant
load test; curve B, constantstress test.
To determine
the engineering creep curve of a metal, a constant load
is
applied to a tensile specimen maintained at a constant temperature, and the strain (extension) of the specimen is determined as a function of time.
Although the measurement
in practice
it
of creep resistance
is
quite simple in principle,
requires considerable laboratory equipment. ^^
The elapsed
time of such tests may extend to several months, while some tests have been run for more than 10 years. The general procedures for creep testing are covered in
ASTM
Specification E13958T.
Curve
A
in Fig. 131 illustrates the idealized
e) is
The
^
slope of this curve (de/dt or
referred to as the creep rate.
INIetals
shape of a creep curve. Follow
Creep and Creep Rupture Tests,
ASM
Handbook, Supplement, August,
1955, Metal Progr., vol. 68, no. 2A, pp. 175184, Aug. 15, 1955.
U.
A. Fellows, E. Cook, and H. S. Avery, Trans.
AIME,
vol. 150, p. 358, 1942.
338
ing an
Applications to Materials Testing
initial
[Chap.
eo,
1
3
rapid elongation of the specimen,
the creep rate decreases
with time, then reaches essentially a steady state in which the creep rate changes little with time, and finally the creep rate increases rapidly with time until fracture occurs. Thus, it is natural to discuss the creep curve
in
terms of
its
three stages.
It
should be noted, however, that the degree
to which these three stages are readily distinguishable depends strongly
on the applied stress and temperature. In making an engineering creep test, it is usual practice to maintain the load constant throughout the test. Thus, as the specimen elongates
>
and decreases in crosssectional area, the axial stress increases. The initial stress which was applied to the specimen is usually the reported value of stress. Methods of compensating for the change in dimensions
of the
specimen so as to carry out the creep test under constantstress
Sudden
strain
Creep curve
Transient creep
Viscous creep
+
Time
Fig. 1 32.
+
Time
Time
Time
Andrade's analysis of the competing processes which determine the creep
conditions have been developed. ^'^
it
When
constantstress tests are made,
is
frequently found that no region of accelerated creep rate occurs
(region III, Fig. 131)
and a creep curve similar to
is
B B
in Fig.
131
is
obtained.
Accelerated creep
found, however, in constantstress tests
when
metallurgical changes occur in the metal.
Curve
should be con
sidered representative of the basic creep curve for a metal.
Andrade's pioneer work on creep^ has had considerable influence on
the thinking on this subject.
He considers that the constantstress creep curve represents the superposition of two separate creep processes which occur after the sudden strain which results from applying the load. The
first
component
is
a transient creep, which has a decreasing creep rate
to this
is
with time.
creep.
Added
a constantrate creep process called viscous
132.
1
The superposition of these creep processes is illustrated in Fig. Andrade found that the creep curve could be represented by the
E. N. da C. Andrade and B. Chalmers, Proc. Roij Soc. (London), vol. 138A, p. 348,
1932.
2
R. L. Fullman, R. P. Carreker, and
J.
C. Fisher, Traris.
vol.
AIME,
vol.
3
97, pp.
657
659, 1953.
^ E. N. da C. Andrade, Proc. Roy. Soc. (London), "Creep and Recovery," pp. 176198, American Society
1957.
for Metals,
90A, pp. 329342, 1914; Metals Park, Ohio,
Sec.
1
32]
Creep and
Stress
Rupture
339
following empirical equation,
L =
where
Lo,
/3,
Lo(l
+
131''^)
exp
Kt
(133)
L = K =
length of specimen at time
t
empirically determined constants
The constant Lo approximates the length of the specimen when the sudden strain produced by the application of the load has ceased. The transient creep is represented by the constant /3. Thus, when k = 0, Eq. (133)
yields a creep rate
which vanishes at long times.
L =
Io(l
"^^ dt
+
0t''^)
=
yiU&t^'^
(134)
When
/3
=
0,
L
y
= exp
Kt
dL
dt
kLq exp
Kt
=
kL
The exponent
creep.
k
therefore represents an extension, per unit length, which
It represents
proceeds at a constant rate.
rate of change of length
(132).
is
the viscous component of
Strictly speaking, k represents quasiviscous flow because the
not proportional to stress as required by Eq.
is
Sometimes transient creep
is
referred to as
/3
flow,
and viscous
(steadystate) creep
referred to as k flow in keeping with Andrade's
analysis of the creep curve.
Andrade's equation has been verified for
conditions extending up to several hundred hours, which result in total
extensions in excess of
1 per cent. Modifications of these equations will be considered in another section of this chapter. The various stages of the creep curve shown in Fig. 131 require further explanation. It is generally considered in this country that the creep curve has three stages. In British terminology the instantaneous strain designated by eo in Fig. 131 is often called the first stage of creep,
so that with this nomenclature the creep curve
stages.
is
considered to have four
by eo occurs practically instantaneously on the application of the load. Even though the applied stress is below the yield stress, not all the instantaneous strain is elastic. Most of this strain is instantly recoverable upon the release of the load (elastic), while part is recoverable with time (anelastic), and the rest is nonrecoverable
strain represented
(plastic). Although the instantaneous strain is not really creep, it is important because it may constitute a considerable fraction of the allowable total strain in machine parts. Sometimes the instantaneous strain
The
340
Applications to Materials Testing
[Chap. 13
is subtracted from the total strain in the creep specimen to give the strain due only to creep. This type of creep curve starts at the origin of
coordinates.
The
first
stage of creep,
decreasing creep rate.
sient creep in
known as primary creep, represents a region of Primary creep is a period of predominantly tran
virtue of
its
the creep of
which the creep resistance of the material increases by For low temperatures and stresses, as in deformation. lead at room temperature, primary creep is the predominant
own
creep process.
is
The second
stage of creep,
known
also as secondary creep,
a period of nearly constant creep rate which results from a balance
between the competing processes of strain hardening and recovery. For this reason, secondary creep is
usually referred to as steadystate
creep.
The average value
of the
creep rate during secondary creep
is
called the
minimum
creep rate.
Thirdstage or tertiary creep mainly
occurs in constantload creep tests
at high stresses at high temperatures.
Time, hr
The reasons
for the acceler
ated creep rate which leads to rapid
failure are not well
Schematic representation of effect of stress on creep curves at constant temperature.
Fig.
133.
known.
is
It is
unlikely that tertiary creep
since
due
solely to necking of the specimen,
many
materials
fail in
is
creep
at strains which are too small to produce necking.
Tertiary creep
more
probably the result of structural changes occurring in the metal. Evidence has been found for void formation and extensive crack formation during this stage. Figure 133 shows the effect of applied stress on the creep curve at constant temperature. It is apparent that a creep curve with three welldefined stages will be found for only certain combinations of stress and temperature. A similar family of curves is obtained for creep at constant stress for different temperatures. The higher the temperature, the greater the creep rate. The basic difference for this case would be that all the curves would originate from the same point on the strain axis. The minimum creep rate is the most important design parameter derived from the creep curve. Two standards of this parameter are commonly used in this country, (1) the stress to produce a creep rate of 0.0001 per cent/hr or 1 per cent/10,000 hr, or (2) the stress for a creep rate of 0.00001 per cent/hr or 1 per cent/ 100,000 hr (about 11 years). The first criterion is more typical of the requirements for jetengine alloys, while the last criterion is used for steam turbines and similar equipment.
^
133]
Creep and
Stress
Rupture
341
A
its
loglog plot of stress vs.
minimum
is
creep rate frequently results in a
straight line.
This type of plot
very useful for design purposes, and
use will be discussed more fully in a later part of this chapter.
133. The Stressrupture Test
is basically similar to the creep test except that always carried to the failure of the material. Higher loads out the test with stressrupture than in and therefore the test a creep test, used are Ordinarily the creep test is carried out at the creep rates are higher. Emphasis in the relatively low stresses so as to avoid tertiary creep. creep test is on precision determination of strain, particularly as to the determination of the minimum creep rate. Creep tests are frequently
The
stressrupture test
is
conducted for periods of 2,000 hr and often to 10,000 hr. In the creep test the total strain is often less than 0.5 per cent, while in the stressrupture test the total strain may be around 50 per cent. Thus, simpler Stressrupture strainmeasuring devices, such as dial gages, can be used. equipment is simpler to build, maintain, and operate than creeptesting
equipment, and therefore
units.
it
lends
itself
more readily
to multiple testing
The
higher
stresses
and
creep rates of the stressrupture test
cause structural changes to occur in metals at shorter times than would
be observed ordinarily in the creep
test,
and therefore stressrupture
tests
can usually be terminated in These factors have con1,000 hr.
test.
tributed to the increased use of the
stressrupture
ularly
well
It
is
partic
0.001 001
,000
Rupture time,
tir
suited
to
determinFig.
1
ing the relative hightemperature
34.
Method
of
plotting
stress
strength of
new
alloys for jetengine
rupture data (schematic).
applications.
for applications where creep deformation can be tolerated but fracture
Further,
must be
prevented,
it
has direct application in design.
from the stressrupture test is the time to cause failure at a given nominal stress for a constant temperature. The elongation and reduction of area at fracture are also determined. If the test is of suitable duration, it is customary to make elongation measurements as a function of time and from this to determine the minimum creep rate. The stress is plotted against the rupture time on a
The
basic information obtained
loglog scale (Fig. 134).
A
straight line will usually be obtained for
in the slope of the stressrupture line
each test temperature.
Changes
are due to structural changes occurring in the material, e.g., changes from
3
,
342
Applications to Materials Testing
[Chap.
1
transgranular to intergraniilar fracture, oxidation, recrystallization and grain growth, or other structural changes such as spheroidization, graphiIt is important to know about the tization, or sigmaphase formation.
existence of such instabilities, since serious errors in extrapolation of the data to longer times can result if they are not detected.
1
34.
Deformation
at
Elevated Temperature
slip,
The
principal deformation processes at elevated temperature are
subgrain
formation,
is
deformation
and grainboundary sHding. Hightemperature Measurecharacterized by extreme inhomogeneity.
ments of local creep elongation^ at various locations in a creep specimen have shown that the local strain undergoes many periodic changes with
time that are not recorded in the changes in strain of the total gage In largegrained specimens, local regions may length of the specimen. undergo lattice rotations which produce areas of misorientation. A number of secondary deformation processes have been observed in
metals at elevated temperature. These include multiple slip, the formation of extremely coarse slip bands, kink bands, fold formation at Many of the deforgrain boundaries, and grainboundary migration. mation studies at elevated temperature have been made with largegrain(Aluminum is favored for this type size sheet specimens of aluminum.
of
study because
lead.
It is
its
thin oxide skin eliminates problems from oxidation.)
Studies have also been
and
made of creep deformation in iron, magnesium, important to remember that all the studies of hightemperature deformation have been made under conditions which give a creep rate of several per cent in 100 or 1,000 hr, while for many engineering applications a creep rate of less than 1 per cent in 100,000 hr is Because the deformation processes which occur at elevated required. temperature depend on the rate of strain as well as the temperature, it is not always possible to extrapolate the results obtained for high strainrate conditions to conditions of greater practical interest.
Much
of the
work on deformation processes during creep has been reviewed by SuUy^ and Grant and Chaudhuri.^
Deformation hy Slip
New
slip
systems
may become
operative
when metals
are deformed at
{ }
,
elevated temperature.
Slip occurs in aluminum* along the
or {211} planes above 500°F.
1
lU 100} Zinc shps on the nonbasal {1010} planes
{
H. C. Chang and N. J. Grant, Trans. AIME, vol. 197, p. 1175, 1953. A. H. Sully, "Progress in Metal Physics," vol. 6, pp. 135180, Pergamon Press, Ltd., London, 1956. 3 N. J. Grant and A. R. Chaudhuri, Creep and Fracture, in "Creep and Recovery," pp. 284343, American Society for Metals, Metals Park, Ohio, 1957. * I. S. Servi. J. T. Norton, and N. J. Grant, Trans. AIME, vol. 194, p. 965, 1952.
2
Sec.
1
34]
Creep and
Stress
Rupture
343
in the (1210) directions above 570°F, and there is evidence of nonbasal hightemperature slip in magnesium. The slip bands produced at high temperature do not generally resemble the straight slip lines which are usually found after roomtemperature deformation. Although hightem^
perature slip may start initially as fairly uniformly spaced slip bands, as deformation proceeds there is a tendency for further shear to be restricted The tendency for cross slip and the formation to a few of the slip bands. of deformation bands increases with temperature. Fine slip lines, which are difficult to resolve with the optical microscope, have been found
between the coarse
slip
bands
in creep
specimens of aluminum.^
These
of
represent the traces of slip planes on which only very small
amounts
shear have occurred.
will
The
significance of fine slip to creep deformation
be discussed
later.
investigations of creepdeformation processes, Hanson and Wheeler^ established that the slipband spacing increases with either an increase in temperature or a decrease in stress. Subsequent work on aluminum and its alloys^' ^ showed that the slipband spacing was
first
In one of the
inversely proportional to the applied stress but independent of temperature.
If
These observations
may
be interpreted in the following way:
is
aluminum with a
certain initial grain size
tested at a certain stress,
If
there will be a certain characteristic slipband spacing.
is
the grain size
smaller than the slipband spacing, the slip bands will not be visible
in the
specimen after deformation. Deformation of the grains will occur by shear along the grain boundaries and by the breakup of the grains into "cells," or subgrains.^ Deformation at high temperatures and/or low
for
which it is difficult to detect slip lines but which there is abundant evidence of grainboundary deformation. This condition has often been called "slipless flow." Complex deformation processes occur in the vicinity of the grain boundaries. While grain boundaries restrict deformation at high temperature to a lesser extent than at room temperature, they still exert a restraining influence on deformation.
strain rates are conditions for
Suhgrain Formation
Creep deformation
is
quite inhomogeneous, so that there are
many
opportunities for lattice bending to occur.
Kink bands, deformation
bands, and local bending near grain boundaries are
1
known
to occur.
A. R. Chaudhuri, H. C. Chang, and N.
J.
Grant, Trans.
AIME,
vol. 203, p. 682,
1955.
2 3 ^
6
«
D. McLean, J. Inst. Metals, vol. 81, p. 133, 19521953. D. Hanson and M. A. Wheeler, J. Inst. Metals, vol. 55, p. 229, 1931. I. S. Servi and N. J. Grant, Trans. .4 /il/E, vol. 191, p. 917, 1951. G. D. Gemmell and N. J. Grant, Trans. AIME, vol. 209, pp. 417423, 1957. W. A. Wood, G. R. Wilms, and W. A. Rachinger, J. Inst. Metals, vol. 79, p. 159,
1951.
344
Applications to Materials Testing
[Chap. 13
Polygoiiization can take place concurrently with lattice bending because
dislocation climb can occur readily at high temperature (see Sec. 610).
The formation of cells, or subgrains, as creep progresses has been observed by means of X rays and metallographic techniques. The size of the subgrains depends on the stress
cells,
are produced
and the temperature. Large subgrains, or by high temperature and a low stress or creep rate.
The decreasing creep rate found during primary creep is the result of the formation of more and more subgrains as creep continues. The increased number of lowangle boundaries provides barriers to dislocation movement and
results in a decrease in creep strain.
Grainboundary Deformation
It
has already been shown in Sec. 85 that the grainboundary relaxis
ation which
measured by internal
Therefore,
it is
friction at elevated
temperature
indicates that the grain boundaries have a certain viscous behavior under
these conditions.
not surprising that the grain boundaries
behave in a manner to indicate considerable mobility when creep is produced at high temperature. The main grainboundary processes which
are observed in hightemperature creep are grainboundary sliding, grain
boundary migration, and fold formation. Grainboundary sliding is a shear process which occurs in the direction of the grain boundary. It is promoted by increasing the temperature and/or decreasing the strain rate. The question whether the sliding occurs along the grain boundary as a bulk movement of the two grains or in a softened area in each grain adjacent to the grain boundary ^ has not been answered. Grainboundary shearing occurs discontinuously in time, and the amount of shear displacement is not uniform along the
^
grain boundary.
that grainboundary sliding
grain past another because
Although the exact mechanism is not known, it is clear is not due to simple viscous sliding of one
it is
preceded by appreciable amounts of
plastic flow in adjacent crystals.
Grainboundary migration
direction which
is
is
a motion of the grain boundary in a
It may be considered Grainboundary migration is a creep recovery process which is important because it allows the distorted material adjacent to the grain boundary to undergo further deformation. The wavy grain boundaries which are frequently observed during hightemperature creep are a result of inhomogeneous grainboundary deformation and grainboundary migration. For grainboundary deformation to occur without producing cracks
inclined to the grain boundary.
to be stressinduced grain growth.
at the grain boundaries,
'
it
is
necessary to achieve continuity of strain
2
H. C. Chang and N. J. Grant, Trans. AIMS, vol. 206, p. 169, 1956. F. N. Rhines, W. E. Bond, and M. A. Kissel, Trans. ASM, vol. 48, p. 919, 1956.
Sec.
1
35]
Creep and
Stress
Rupture
345
A common method of accommodating grainalong the grain boundary. temperature is by the formation of folds at the boundary strain at high
Figure 135 shows a sketch of a fold. boundary. of slip and grainboundary displacement to importance The relative has been investigated by McLean for alumideformation the total creep this relatively low temAt num at 200°C. only shown that a small has perature he deformation is due to total the of fraction grainboundary displacement, about half the total deformation is due to slip which is
end
of a grain
^
readily attributed to coarse slip bands, while
the remainder of the total deformation cannot be attributed to any microscopic defor
mation mechanism.
McLean
attributes this
"missing creep" to deformation by fine slip, which is very difficult to detect with the
microscope.
It is believed that
I
deformation by fine slip can explain the observations Fig. 135. Fold formation at a ^^P^^ P^^'^t (schematic). of earlier workers that creep deformation Greater occurs without slip (slipless flow). contribution to the total deformation from grainboundary displacement
would be expected at higher temperatures and lower
35. Fracture at Elevated Temperature
stresses.
1
It has been known since the early work of Rosenhain and Ewen' that metals undergo a transition from transgranular fracture to intergranular When transgranular fracture fracture as the temperature is increased. occurs, the slip planes are weaker than the grain boundaries, while for
intergranular fracture the grain boundary
Jeffries''
is
the weaker component.
introduced the concept of the equicohesive temperature (EOT), which was defined as that temperature at which the grains and grain boundaries have equal strength (Fig. 136a). Like the recrystallization
temperature, the
stress
ECT
is
not a fixed one.
In addition to the effect of
and temperature on the ECT, the strain rate has an important influence. Figure 1366 shows that decreasing the strain rate lowers the ECT and therefore increases the tendency for intergranular fracture.
The
effect of strain
rate on the strengthtemperature relationship
is
believed to be
1
much
larger for the grainboundary strength than for
2 3
H. C. Chang and N. J. Grant, Trans. AIME, vol. 194, p. 619, 1952. D. McLean, J. Inst. Metals, vol. 80, p. 507, 19511952. W. Rosenhain and D. Ewen, J. Inst. Metals, vol. 10, p. 119, 1913.
Jeffries,
Z.
Trans.
AIME,
vol. 60, pp. 474576, 1919.
346
Applications to Materials Testing
[Chap. 13
the strength of the grains.
Since the
size,
amount
of
grainboundary area
decreases with increasing grain
a material with a large grain size
will
have higher strength above the
Grain boundary
ECT
Temperature
[a]
has shown that large tensile stresses should be developed at a triple point due to shear stresses acting along the grain boundaries. C. L. W. 137. where the total short. Trans." p. Balluffi and L. Machlin. AIME. pp. 1 36] Creep and Stress Rupture 347 reasonable models for grainboundary fracture have been suggested. C. p. 544550. Grant. 449.. 1959. Grainboundary migration displaces the strained grain boundary to a new unstrained mits the region of the crystal. Ohio. Chang and N. vol. Chen and E. whether voids are initiated at grain boundaries by a process of vacancy condensation or as to explain the sintering of metals. Seigle. Zener. Zener. the tendency for grainboundary fracture is diminished. Metals Park. vol. Two meet on a plane of polish. vol. 1956. 1 36. MachHn. Theories of Lowtemperature Creep overcome Creep is possible only because obstacles to deformation can be of thermal fluctuations and stress. cracks may be initiated at triple points where three grain boundaries conditions. pp. AIME. Ada Met. 206. at present. while the formation of folds per relief of stress concentration at grain corners by plastic defor mation within the grains. Micromechanism of Fracture. At lower temperatures recovery proc by the combined action H. 1948. Several methods by which cracks form' as the result of grainboundary sliding are shown schematically in Fig. It is uncertain. The second type of intergranular fracture is characterized by the formation of voids at grain boundaries.Sec. the result of localized plastic yielding. Ballufh and Seigle* have advanced a theory for the growth of voids based on the ideas used On the other hand. Acta 140. J. Met. W. " R. J. 5. S. This type of grainboundary failure life is fairly is preva lent for high stresses. 1957. This type of fracture is most prevalent when low stresses result in failure in relatively long times. Trans. particularly those which are perpendicular to the tensile stress. Intrater and E. p. 1 ' 'C. When grainboundary migration and fold formation can occur. S. vol.. The voids grow and coalesce into grainboundary cracks. types of intergranular fracture have been observed under creep Under conditions where grainboundary sliding can occur. there are experiments" which show that grainboundary voids are not formed unless there is grainboundary sliding. 3. . none are capable of predicting all the details of grainboundary fracture. 7. in "Fracturing of Metals. 1957. 209. One mechanism is based on the idea that the voids are formed by the condensation of vacancies and grow by the diffusion of vacancies. 829835. American Society for Metals. At least two mechanisms have been suggested for this type of fracture. Diffusioncontrolled processes are important chiefly at temperatures greater than about onehalf the melting point.
where It is believed to be a true exhaustion process in which the ratedetermining step is the activation energy to move a dislocation. However. creep predominates. . and Phys. N.3 348 esses Applications to Materials Testing [Chap. Solids. 2 When = ^. A and ior are described . (136) creep rate constant and Eq. 5 N. 1 which are not dependent on diffusion. 1952. An 1 exhaustion theory does not adequately describe the behavior during A.e. Eq.. When n' = 1. the is (136) represents steadystate creep. 66B.4 Mott and Nabarro* and Lowtemperature logarithmic creep obeys a mechanical equation of i. while below Tml'^ transient. (136) becomes Andrade's equation e for transient creep. 2 ^ Cottrell. tant roles in the creep process. play imporHightemperature creep is predominantly steadystate or viscous creep. Nabarro. 1953]. Soc. depending by upon the value of n' If n' = 0. op. the rate of strain at a given time depends only on the instantaneous values of stress and strain and not on the previous strain history. ' e = At''' (136) where Different types of creep behavEq. As these easytomove dislocations are exhausted. H. F. creep at higher temperatures is strongly dependent on prior strain and thermal history and hence does not follow a mechanical equastate.^ discussed in Sec. by the movement rate decreases. Physical Society. the dislocations with the lowest activation energy move first to produce the initial creep strain. the activation energy for the process continuously increases. 132. vol. [O. 1. Phys. of dislocations of higher activation energy. (London). Cottrell. such as cross slip. creep can continue only recovery cannot occur. p. 5363. or primary. Andrade's equation for describing transient and steadystate creep was An alternative general equation for the time laws of creep was suggested by Cottrell. = I3t'^^ (138) Logarithmic creep occurs at low temperatures and low stresses. R. pp. 1. vol. Therefore. Wyatt. and the creep Theoretical treatments of exhaustion creep that result in a logarithmic equation have been proposed by Cottrell. Mech. cit. Logarithmic creep has been observed for copper below 200°K. On the initial application of stress. e = a In f (137) where a is a constant. This n' is the logarithmic creep law found at low temperatures. "Report on Strength of Solids. 1948. (136) becomes n' are empirical constants. London." p. /. Proc. 495. Eq. tion of state. Mott and F. H.
389. J. J. slip Since dislocation it will be the rate controlling step in steadystate creep. vol.first suggested that steadystate creep could be treated as a is the slope of the balance between strain hardening and recovery. West Scot. Metals. in "Creep and Recovery. 54. The effects of strain hardening are recovered by the escape of screw dislocations from pileups by cross about T'm/2. Schoeck. or secondary. 1957.. 37. d<r = ^dt\^de = ot (139) oe . p. then requires that the flow stress must remain constant. Analysis of existing data^ indicates that the escape of screw dislocations from pileups by cross may be the chief recovery mechanism in fee metals. k is Cottrell process of (k flow) and r and ^ are defined and Aytekin^ have shown that the thermally activated recovery can be expressed by the steadystate creep rate r= where da = ^ Cexp —{H — ^ qa) /iQ 1 1\ (1311) H = an activation energy to atomic stress concentration q = an activation volume related C = a constant k = Boltzmann's constant 1 G. Aytekin.Sec. 1 37] Creep and Stress Rupture 349 transient creep. Although there is is state condition achieved for all and the escape of edge dislocations by climb. there is ample experimental evidence to indicate that approximate steadystate conditions are achieved After a short period of testing in the hightemperature region. Iron Steel 3 . H. 1947. creep predominates at temperatures above some question whether a true steadycombinations of stress and temperature. vol. = _fZ' = r 0(T/oe <p (1310) where above. Ohio. pp. 1950. The decreasing creep rate during transient creep arises from the increasing dislocation density and the formation of lowangle boundaries. Metals Park. 4596. Inst. 77. The recovery mechanisms operating during slip transient creep are not well established. If = da/dt is the rate of stressstrain curve at the applied stress a and r the steadystate condition recovery of the flow stress on annealing. Theory of Creep." American Society of Inst. Orowan. = . Metals. Theories of Hishtemperaturc 1 Creep Steadystate. Steadystate creep arises because of a balance between strain hardening and recovery. 2 E. A. climb has a higher activation energy than cross slip. Orowan. Cottrell and V.
. This is a conventional method The corresponding equation e of plotting engi = Ba' (1316) is often used in the engineering analysis of the creep of structures and machine elements. (1311) into Eq. S.^_^. Feltham and D. 5783. creep resulted from the shear of "flow units" past each other through the periodic potential field of the The metal was assumed to behave like a viscous mateand for creep to occur the flow unit had to acquire sufficient energy The theory to surmount the potential barrier of the activated state. 3 P. A relationship very similar to Eq. Dushman. AIME.^ e = ^oexp^ sinh^ q are If (1314) The parameters Ao. neering creep data. Trans. € = C exp j^ sinh ^ 5 is (1313) where AF is the free energy of activation and a constant describing the size of the flow unit. Huthsteiner. 108. p. ^ = ^"^P C (H kf qa) (l^l^) . vol. Appl. pp. H. (1313) has been used by a number crystal structure. . However. 143. rial. J. /. vol. so that this theory of investigators for describing steadystate creep data of annealed metals in terms of applied stress and temperature. 15. Meakin. (1310) gives an equation for the steadystate creep rate. and H. 1959. qa/kT is greater than the above equation simplifies to e = HAo exp —^^^^^ (1315) log Equation (1315) implies the existence of a linear relationship between A linear relationship is also often obtained when log e is e and (T. Kauzmann. Theories of steadystate creep based on Eyring's chemical theory of reaction rates were at one time very prominent in the thinking on this subject. W. L. plotted against logo. and only 2. vol. q varies sig nificantly with temperature. 1 "^ W. 1944. predicted a steadystate creep rate of the form crystal lattice. Acta Mel.. Dunbar. The flow units were not defined in terms of it had little physical basis. 7. 1 Substituting Eq. 1941.3 350 Applications to Materials Testing [Chap. and independent of a. 614627. was an important advance in creep theory because it emphasized that creep was a thermally activated process. Phys.. pp. i'^ According to this concept.
Fig.000 psi 4x10"'^ e^te^"/RT. E. where A' is Avogadro's number. /.ac) o'c = constant Figure 138 shows a typical correlation between e and d for creep data obtained at several temperatures. in "Creep and Recovery. . and Phys. Dorn." pp. Creep strain vs. the total creep strain function of d.) been found that creep can be correlated compensated time parameter 6. e = = R = T = t time of creep exposure activation energy for creep universal gas constant* is a simple (1318) = f(d. in "Creep and Recov ery.60 0. 137] Creep and Stress Rupture 351 Dorn^'^ has made valuable contributions to the knowledge of hightemperature creep by putting the reactionrate theory of creep on a In his creep experiments on highpurity polybetter physical basis." American Society for Metals. " 0. RT (1317) Ai/c absolute temperature For constant load or stress conditions. 1954. Ohio. crystalline aluminum. hr {J. Solids. 138. Metals Park. The Spectrum of Activation Energies for Creep. 2 J. vol. Metals Park. E.987%) Creep under constant stress of 3. American Society for Metals.40 0. careful attention has been given to changes in It has lattice distortion and subgrain formation during the creep test.30 0. 85. Dorn. p. 3.)ec. in terms of a temperature = where t exp AH. 1957. An important characteristic of this ' J. 1957. and the gas constant R are related through the equation R = kN. = t %x^ {—l^H /RT). E. Mech.20 0. Dorn. 255283. Ohio.10 t Pure aluminum (99.50 0. ^ Boltzmann's constant A.
352 Applications to Materials Testing is [Chap. The activation energy for creep can be evaluated from two creep curves determined with the same applied stress at two different temperatures. Eq. (1319) at the im minimum creep rate im gives = ner. Not only can Or = tr exp ^^ = = F{a) (1322) where U is the time to rupture. Therefore.. di = ti exp = $2 = h 2 exp (1325) AHc = R J'^' J — i 1 Inf^ tl . this equation be used for describing the relation between temperature and strain rate for the relatively high strain rates of the tension test. dm is only a function of the creep stress so that <Tc = F (im exp H^ = F{Z) (1321) Z is the ZenerHollomon parameter considered previously in Sec. but it appears to be reliable for the very low strain rates found in the creep test. 13 correlation that equivalent structure. (1318) is differentiated with respect to time. for equal values of total creep strain obtained at the two temperatures. For engineering analysis of creep. aTJ (1320) However.a) exp ^ ac. 98. the values of d are equal. as determined by metallography and If X rays. is Evaluating Eq. obtained for the same values of e and 6. Thus. the stress dependence of the time to rupture is often expressed by the empirical equation tr aa"" (1323) An empirical relationship also exists between minimum creep rate e^ such that log tr the rupture life and the f C log e^ = iC (1324) where C and K are constants for a given alloy. The effect of stress on time to rupture can also be correlated with the timetemperature parameter 6.
and no correlation exists with the activation energy for selfdiffusion. At temperatures below about onehalf the melting point A/fc is a function of temperature.137] Creep and Stress Rupture 353 A rather extensive correlation' of creep and diffusion data for pure metals shows that the activation energy for hightemperature creep is approximately equal to the activation energy for selfdiffusion (Fig. 139). The excellent correlation between the activation energies for creep and selfdiffusion indicates that dislocation climb is the rate 80 .
This dislocation model leads to an equation for the steadytheir annihilation. H. 26. 2 J. ASME. by using the assumption that dislocation The obstacles to dislocation move ment 64).n = iS exp smh .^ t. 2124. . 1955. F. = . Stationery Office. the dislocations climb into adjoining slip planes. (1328) the structure term S is assumed constant at the minimum creep rate and the constant <Tc is dependent on temperature. Weertman^ has derived an equation for the steadystate creep rate which climb is is similar to Eq. (1327) the ratecontrolling process. vol. 81. are considered to be immobile CottrellLomer dislocations (see Sec. Trans. "NPL Sj'inposium on Creep and Fracture at High Temperature.. ser. Weertman. creep proposed state creep rate similar to Eq. There is no other recourse than to make an intelligent selection of design stresses from the existing data. 617627. forming lowangle boundaries. Phijs. J.000 psi) this equation becomes A f/ 6„. = N. 1955. pp. Fortunately. to make from existing experiment and 1 38. 12131217.3 354 Avhere S' Applications to Materials Testing [Chap. For low stresses (below about 5. A steady state is established between the generation of dislocations and An alternative dislocation model of steadystate by Mott^ considers that the ratecontrolling process is the formation of vacancies due to the movement of jogs on screw dislocations.(T) (1328) In Eq. The choice of the proper functional relationship between is difficult creep rate and stress theory. Conrad. (1312) in which the stress is present as a cr/r term. In order to escape from these obstacles. a large 1 number of reliable hightemperature strength data H. The activation energy for this process is the activation energy for selfdiffusion. 1959. pp. Appl. Presentation of Engineering Creep Data It should be apparent from the previous sections that knowledge of the hightemperature strength of metals has not advanced to the point where creep and stressrupture behavior can be reliably predicted for design purposes on a theoretical basis. vol. London.S" exp ^jr c^"' (1327) Another equation which gives good agreement between minimum creep rate and stress has been suggested by Conrad. D. M." pp. 1 is a structuresensitive parameter and B' is a constant which is independent of temperature and structural changes. Mott.
Sec. A common method of presenting creep data is to plot the logarithm of the stress against the logarithm of the minimum creep rate (Fig. ^ ^ 6 5 k 4 . 1310). 1 38] Creep and Stress Rupture 355 have been collected and published by the ASTM^ and by the producers of hightemperature alloys. but discontinuities due to structural instabilities will Values of minimum creep rate lower often occur at higher temperatures. With this type of plot straight lines will frequently be obtained for the lower temperatures.
1311). to produce different convenient to present the data as a plot of the time amounts of total deformation at different stresses Each curve represents the stress and time at a fixed (Fig. In designing missiles and highspeed aircraft data are needed at higher temperatures and stresses and shorter times than are usually determined tor creep tests. temperature to produce a certain per cent of total deformation (sudden strain plus creep strain). data is by the Creep tests are conducted at 10 . A separate set of curves is required for each temperature. A common method of presenting these use of isochronous stressstrain curves.356 Applications to Materials Testing it is [Chap. 1 3 design criterion.
^ Several suggestions of timetemperature parameters for predicting longtime creep or rupture behavior from the results of shorter time tests at high temperature have been made. times is is required.000 hr is probably safe and extrapolation even to 100. in such situations extrapolation of the data to long the only alternative. although the alloy has been in existence for only Obviously. Reliable extrapolation of creep and stress rupture curves to longer times can be made only when it is certain that no structural changes occur in the region of extrapolation which would produce a change in the slope of the curve.Sec. While this parameter has been useful for correlating creep and stressrupture data for pure metals and dilute solidsolution alloys. °F time. hr a constant with a value between about 10 and 30 In the original derivation of Eq. . the LarsonMiller parameter should give a master plot which represents the high temperature strength of the material for all combinations of temperature and time. (1329) on the basis of reactionrate theory a value of C = 20 was used.000 hr may be possible. the LarsonMiller parameter is not very sensitive to small changes in rupture life due to structural changes in the material. (1317)] is another timetemperature parameter. For example. Miller. While this is a good approximation when other data are lacking. However. Once the proper constant has been established. 74. A logical method of extrapolating stressrupture curves which takes into consideration the changes in slope due to structural changes has been proposed by Grant and Bucklin. Trans. if in 1. 765771. G. An extension of the LarsonMiller parameter has been suggested by 1 N. ASME. 2 F. The LarsonMiller parameter^ has the form (7^ + 460) (C f log t) = constant (1329) where T = = C = / temperature. vol. 720761. 1952.000 hr no change in slope occurs in the curve at 200°F above the required temperature. Trans. one way of checking on this point is to examine the logstresslogrupture life plot at a temperature several hundred degrees above the required temperature. The parameter 6 suggested by Dorn [see Eq. Bucklin. 1 39] Creep and Stress Rupture 357 (11 years) 2 years. 42. pp. for best results C should be considered a material constant which is determined experimentally. Larson Grant and A. ASM. since the temperature term in this parameter is given considerable weight. R. pp. 1950. Since structural changes generally occur at shorter times for higher temperatures. extrapolation of the lower temperature curve as a straight line to 10. and J. J. it has not been used to any great extent with engineering hightemperature alloys. then a plot of log stress vs. vol.
°F time. 1 Manson and Haferd. log ta = = = test temperature.^ The MansonHaferd parameter has the form T log t Ta log ta — = constant (1330) where T t Ta.3 358 Applications to Materials Testing [Chap. hr constants derived from test data t A straight line is obtained when log 100 is plotted against T for data obtained 10 .
127129. which in certain cases may be a more controlling factor than the strength. Hightemperature Alloys Hightemperature alloys are a particular class of complex materials developed for a very specific application. On the other hand. each point tensile strength or creep strength and hot hardness. how metallurgical variables influence creep The development of hightemperature alloys has.Sec. 266271. pp. 45. Just as a linear relationship exists between stress and the logarithm of the rupture time. Inst. vol. along the stress and a value of the parameter we can establish values of temperThis procedure has been verified for both singleature and rupture time. in the main. Considerable use has been made of these parameters for predicting longtime data on the basis of shorttime While there is no assurance that erroneous predictions will results. timetemperature parameter. 1310] Creep and Stress Rupture 359 has been found to provide satisfactory correlation of the data. 1310. Materials & Methods.^ However. a similar relationship holds between hot hardness and the logarithm In Sec. . it is generally considered that the use of these parameters for this purpose will probably give better results than simple graphical extrapolation of loglog plots by one or two logarithmic cycles. It is a A shortcut approach to hightemperature properties test. not result from this procedure. Figure 1312 illustion exists between hot hardness and tensile strength. is through the There are certain parallelisms between hightemperature strength results and hothardness data. J. The nominal compositions 1 of a number of hightemperature alloys are E. In general. empirical investigation. 1957. Underwood. use of the hothardness useful method for obtaining hightemperature properties of comparatively brittle materials. pp. Metals. tests of this type give no indication of the ductility of the material. with a value righthand ordinate. 1110 it was shown that a linear relaof the indentation time. 19591960. trates a similar relation between hot hardness and hightemperature From the lefthand ordinate we get the relationship between strength. 88. these parameters provide useful methods for presenting large numbers of data through the use of master plots of log stress vs. and it is really only in retrospect that principles underlying these developments have become evident. Some appreciation of the metallurgical principles behind the development of these alloys is important to an understanding of behavior. E. phase alloys and complex multiphase hightemperature alloys. been the result of painstaking. vol. LarsonMiller parameter of establishes a certain value the on the curve of hightemperature Thus.
with increased chromium and molybdenum to form complex carbides. The ferritic alloys Table 131 Compositions of Typical Hightemperature Alloys Alloy C Cr Ni Mo Co W Cb Ti Al Fe Other Ferritic Steels 1.10 316 347 16256 A286 Inconel Inconel X B Nimonic 90 Hastelloy Ren^ 41 Udimet 500 Vitallium (HS21. Because of given in Table 131... X40 (HS31) N155 (Multimelt) S590 S816 K42 B Refractaloy 26 . Molybdenum is particularly effective in increasing the creep resistance of steel. 5 Cr. 360 Applications to Materials Testing [Chap.20 0.. Mo Mo "1722A" S 410 0. 1 3 available alloys' could be were developed first to meet They are increased temperature requirements in steam power plants.. Only a few of the many included in this table. essentially carbon steels.30 0. .10 0.25 Cr. which resist softening.
Ti). The perature strength depends on the ability of the system to resist overaging. Since this compound produces greater hardening than NisAl. it is customary to add both Al and Ti to these nickelbase alloys. However. 1310] Creep and Stress Rupture 361 oxidation and the instability of the carbide phase these alloys are limited about 1000°F. tung and columbium are added to these alloys to is form stable complex carbides. are introduced into a metal matrix by artificial The prototype for this development was the sintered aluminumpowder (SAP) alloy. and S816 the Molybdenum. Certain carbide systems undergo a series of complex aging reactions which may result in be strengthened by the addition of small intermetallic compounds NisAl or NiaTi are formed by these additions. The superalloys for jetengine applications are based on either nickel or cobalt austenitic alloys or combinations of the two. The carbon content of these alloys usually critical and must Increasing the carbon content up to a certain limit increases the amount of carbide particles. Up to three out of every five atoms of Al in NisAl may be replaced by Ti. This can be destroyed if the service conditions fluctuate above the optimum aging temperature. pp. 209. Si02. Preston. Nickelbase alloys containing Al and Ti are true agehardening systems. ^ metallurgy methods. therefore. and hence the rupture strength is increased. In the cobaltbase alloys and certain complex NiCoCrFe alloys like N155. means. Owing to the increased oxidation resistance of austenitic stainless steels. vol. AIMS. additional strengthening. so that their hightemNickelbase superalloys of Al may amounts and/or Ti. if too much carbon is present. 1957. Superalloys. chiefly and Zr02. these alloys extend the useful stressbearing in use to range to about 1200°F. » Mixtures of fine oxides and metal powders are N. in which fine AI2O3 particles were dispersed in an aluminum matrix owing to the breakoff of surface oxide during the extrusion of sintered aluminum powder. massive carbide networks form and reduce the rupture strength. are multiphase alloys which attain their strength primarily from a dispersion of stable secondphase particles. Singlephase such as Nichrome (NiCr) and austenitic stainless become weak above about 1300°F. the carbides will no longer be present as discrete particles. . Trans. secondphase particles are complex metal carbides. A new class of dispersionstrengthened hightemperature alloys are being developed in which thermally stable secondphase particles. Instead. S590. The heat treatment must be carefully controlled to put these alloys in the condition of maximum particle stability. At present most developments are centered in the preparation of a dispersionhardened alloy by powderAI2O3. to form Ni3(Al. In general the> contain appreciable solidsolution alloys steel chromium for oxidation resistance.Sec. be controlled carefully. 349356. Grant and O. J. sten.
DMIC Memorial Institute. The hightemperature strength of metals is approximately related to by powdermetallurgy methods. 300 500 1000 Temperature. nickel alloys and extruded into useful shapes. Moon and W. . in which ceramic particles such as borides. By comparing on the basis of rupture time at a given stress. Simmons.000 per cent have been reported for dispersionstrengthened alloys. since in principle maximum resistance to growth. and silicides are combined with a and amount An metallic binder Cermets have shown very high strength at 1800°F. Mar.000 psi have been found for nickel. analogous approach is the development of cermets.000 hr versus temperature for Memo 92. Battelle 1313. 1961. P. selected engineering alloys. while increases in rupture stress at 100 hr of over 10. sintered. Data for copper and prepared in this way show that the stressrupture curves drop off for a dispersionstrengthened alloy much less rapidly with time than for the same metal without secondphase particles. but their use has been limited owing to poor ductility and insufficient thermal. permits the use of a secondphase and at the same time the of the particles can be controlled.362 Applications to Materials Testing [Chap. °F 1500 2000 Fig. 23. (Z). carbides. Typical curves of stress for rupture in 1. improvements of 1.) The production particle with size of artificial dispersionstrengthened alloys offers conit siderable promise. 13 pressed. F.and mechanicalshock resistance.
ASM. purifying. R. aluminum has better hightemperis better than aluminum. which shows the stress for rupture in 1. of increasing melting point. a certain extent. and tungsten.000 hr at different temperatures for a number of engineering materials. molybdenum and its alloys have the best hightem common metals available at the present time. Parker' has proposed that the observed dependence of hightemperature strength on grain size is basically incorrect. Formerly it was suggested that the lower creep strength of fine grain material was due to greater grainboundary area available for grainboundary sliding. in order chromium. and the creep rate is high. which in turn depends upon the rate at which vacancies can diffuse to edge dislocations. the use of molybdenum is limited undergoes catastrophic oxidation at elevated temperOxidationresistance coatings for molybdenum are being used to ature. Parker. cobalt. . Titanium. EFfect of Metallursical Variables The hightemperature creep and of course.Sec. it by the fact that includes a search for methods of increasing their oxidation resistance and hightemperature strength by alloying. columbium. tantalum. 1311. Since hightemperature creep depends on dislocation climb. and copper For example. Parker feels that the grainsize effect effect of on creep is due to the grainboundary structure on vacancy diffusion. and nickel. molybdenum. higher for coarsegrain material than for a finegrainsize metal. Vacancy diffusion is more rapid along highenergy grain boundaries than through the bulk lattice. However. 1311] Creep and Stress Rupture 363 their melting points. 1958. work which indicates that grainboundary sliding can account for only about 10 per cent of the total creep strain has cast doubt on this explanation. This work ority of molybdenum. ature strength than zinc. with finegrain material with many highangle grain boundaries dislocation climb will be rapid. Extensive development of ways of preparing. is stressrupture strengths are usually This. of grain size The basic effect on hightemperature strength is clouded because it is practically impossible to change the grain size of complex superalloys without inadvertently changing some other factor such as the carbide spacing or the aging response. 52104. and fabricating the other refractory metals is under way. the principal constituents of presentday superalloys. are the relatively common metals with melt ing points greater than iron. 50. Figure 1313. pp. Therefore. In fact. vol. illustrates the superiperature strength of any Unfortunately. Although the hightemperature strength of titanium has not proved to be so great as would be expected on the basis of its melting point. in contrast to the behavior at lower temperature. 1 E. Trans. where a decrease in grain size results in an increase in strength.
However. Recent experiments by Parker in Fig. 86. Thus. R. most of the highenergy grain boundaries disappear owing to grain growth. 1314.) in which the grain size of copper was changed without changing the proportion of lowangle and highangle grain boundaries have showai that the creep rate increases with increasing grain size. 1 31 4. Note the change in shape of the creep curves and the decreasing creep rate as more and more lowangle grain boundaries were introduced. The boundaries which remain are mostly lowerenergy grain boundaries. 50. Parker. The fact that lowangle grain boundaries improve creep properties is Networks of lowangle boundaries were produced amounts in tension at room temperature and annealing at 800°C.10 Fig. shown in nickel by straining small 0. it appears as if the fundamental dependence of creep rate on grain size is no different from the dependence of strength on grain size at low temperature. for which vacancy diffusion is relatively slow. when the same material is heated to a high temperature to size. dislocation climb will be slower in the coarsegrain material. coarsen the grain Therefore. vol. Comparison of the properties of cast and forged hightemperature alloys in the same state of heat treatment shows that cast alloys generally have somewhat higher hot strength and creep resistance than forgings at temperatures above the equicohesive temperature.364 Applications to Materials Testing [Chap. ASM. secondary effects which accompany the change in grain size when it is accomplished by commercial annealing procedures are responsible for the effect which is usually observed. p. The reasons for . Trans. Effect of substructure introduced by prestrain and anneal treatments on the creep curve of highpurity nickel. with greater prestrain. 1958. 13 However. {E.
Deformation in this temperature region for these highly alloyed materials is about equivalent to working mild steel at room temperature. between roomtemperature For example. strength and hightemperature strength. and S816 are appreciably improved by controlled amounts of reduction in the range 1200 to 1700°F. there is little correlation . Most hightemperature alloys are hotworked only with difficulty. which cannot be hotworked. It should be recognized that working highly alloyed materials like superalloys does not constitute true hot working since residual strains will be left in the For this reason the resulting properties of agehardenable alloys lattice.Sec. Although a quenched and tempered martensitic structure of fine carbides has the best strength at room temperature and may have good strength for short times up to 1100°F. the incorporation of a dispersion of fine oxide particles in the metal matrix may produce only a modest increase in roomtemperature strength. particularly cobaltbase alloys. production the conditions must be carefully controlled to provide ings. are dependent on the hotworking conditions. Creep and Stress Rupture 365 but it appears that it is related to a strength ening from the dendritic structure of the casting. The importance of using a thermally stable metallurgical structure for longtime hightemperature service is well illustrated by the case of lowalloy steels. forgings are more ductile than castings. and generally they show fatigue properties which are superior to those of castings. The rupture properties of certain nonaging solidsolution alloys like 16256. uniform properties. N155. This procedure is known as hotcold working. There are certain alloys. perature In is increased. or warm working. Because of their worked structure. but the improvement in rupture time at elevated temperature may be a thousandfold. 1311] this are not well established. Warm working and cold working at room temperature of less highly alloyed materials produce higher creep strength because of the energy stored in the lattice as a result of the plastic deformation. Variations in grain size due to changes in section size may be a problem with castings. Advantages can be ascribed to both casting and forging as a method of production for high temperature alloys. and therefore they must be produced as Although castings show somewhat better strength than forgcastings. Hot working of superalloys for ingot breakdown is generally done in the range 1700 to 2200°F. on long exposure at elevated temperature the carbides grow and coalesce and the creep general. Since hightemperature alloys are designed to resist deformation at high temperatures. For a given operating temperature there will be a critical amount of cold work beyond which the increased lattice strain causes rapid recovery and recrystallizaThe critical amount of cold work decreases as the operating temtion. it is not surprising to find that they present problems in mechanical w^orking and fabrication.
3 366 Applications to Materials Testing [Chap. of the oxidation can have an important influence on the hightemperature properties. 1959. When mate 2 M. Much better creep properties are obtained if the steel is initially in the stable annealed condition. creep began again at nearly the original rate.^ When the copper was electrolytically tests removed. The environment surrounding the specimen can have an important influence on hightemperature strength. AIMS. tals Creep tests on zinc single crys showed that creep practically stopped when copper was plated on the surface of the specimens. refine the grain size. Considerable improve ment in the hightemperature properties of superalloys results from vacuum melting. Not only does aluminum also accelerates spheroidization and graphitization. but intergranular penetration of oxide will usually lead to decreased rupture 1 life and intergranular fracture. while the reverse is true at low temperatures and high strain rates. The creep properties of steels are particularly subject to variations in properties. 244255. type of mechanical working. Assum ing that the transformation characteristics for the alloy are completely known. life to of the decrease in the number and size of inclusions. A thin oxide layer or a finely dispersed oxide will usually lead to strengthening. . Parker. Aluminum but it deoxidized steels generally have poorer creep properties than silicon deoxidized steels. Hightemperature properties can show considerable scatter. P. Shahinian and M. and therefore more uniform response to heat treatment. which lead to intergranular fracture. vol. R.^ At high temperatures and low strain rates these materials are stronger in air than in vacuum. One of the difficulties which must be guarded against is the formation of grainboundary precipitates. 51. Trans. is obtained with vacuum melting. The problem of selecting the best heat treatment for a difficult. vol. R. pp. Pickus and E. 1951. and frequently measurably different results are obtained between different heats of the same material or even between different bars from the same heat. Fabricability is also improved. ASM. the selection of the heat treatment would be based on the expected service temperature and the required service life. Achter. melting practice. pp. 1 properties are very poor. This behavior is attributed to the competing effects of strengthening resulting from oxidation and weakenThe nature ing due to lowering of the surface energy by absorbed gases. A dispersion of complex superalloy is frequently compromise must be reached between the fineness of the secondphase particles and their thermal stability. 792796. which is usually not the case. rupture and ductility at fracture are both increased by vacuum melting. Better control of composition. R. presumably because In general. Trans. and microstructure. Stressrupture on nickel and nickelchromium alloys show a complex dependence on atmosphere. 191. which are related in complex ways to the composition.
When stress relaxation occurs. (335) and (336)]. 82. The assumption of incompressible material leads to the familiar relationship The assumption that principal shearstrain rates are 61 4.Sec. Soderberg. Trans. vol. .«2 + fs = 0. ASME. ser. Met. D. For engineering purposes the stress dependence of the creep rate can be expressed by Eq. 453461.^ 1 31 3. R. 447506. For combined stress conditions e and a must be replaced by the effective strain rate e and the effective stress a [see Eqs. D. creep of thickwalled tubes under and tubes stressed in biaxial tension. we can write i = Ba"' (1333) Combining Eqs. E. 1936. Thus. the stress The original analysis of this vol. 1 31 2. For a critical review see A. Trans. ASME. E. plotted on loglog coordi they give a straightline relationship. vol. (1316). 462464. 1960. Finnie. 733. (1332) €1 and (1333) results in = Ba"''[ai . rials 1 31 3] Creep and in Stress Rupture 367 in cor must operate an atmosphere of hot combustion gases or life is rosive environments. Reviews. pp. Johnson. Trans. effective stress and the effective strain rate are useful parameters has been for correlating steadystate creep data. problem was given by C. tests. In the absence ^ assumptions of plasticity theory (see Sec. Creep under Combined Stresses combined Considerable attention has been given to the problem of design for stress conditions during steadystate creep. 1960. of metallurgical changes. p. 82. 1960. Correlation obtained between uniaxial creep internal pressure. ser. Subsequent analysis has been made by I. Davis. vol.H(^2 + When C73)] (1334) The nates. 2 5. pp. A. the service materially reduced. 58. 38) hold reasonably well for these conditions. Stress Relaxation Stress relaxation under creep conditions refers to the decrease in stress at ^ constant deformation. the basic simplifying proportional to principal shear stresses gives ji^lj2 ^ 0"! — ^^^^ 0'2 0'2 — (73 ^ hj^^ ^ ^ — (Ti (T\ ^^3_3j^ Combining these eciuations results in ei =^kl  K(CT2 + CT3)] (1332) Similar expressions are obtained for ez and is. ASME. pp.
but it levels off because the stress level decreased and the transient creep rate decreases with time.368 Applications to Materials Testing [Chap. These type can be used to estimate the time required to If Eq. blies may lead to loose joints and leakage. The initial rate of is decrease of stress is high. L. 1 3 needed to maintain a constant total deformation decreases as a function Consider a tension specimen which is under a total strain e at an elevated temperature where creep can occur. Relaxation properties of steels and superalloys are given in 187." H. (1316) can be used to express the stress dependence of creep. 214. Spec. Tech. vol. 1948. Robinson. 2 E. = where e 6e + (1335) €e Cp = = = total strain elastic strain plastic (creep) strain remain constant as the material creeps. Stressrelaxation curves. 1956. The relaxation of stress in bolted joints and shrink or pressfit assemtotal strain to For the sary for the elastic strain to decrease. it is necesThis means that the stress required to maintain the total strain decreases with time as creep increases. 1956.^ commonly made on bolting materials for hightemperature Figure 1315 shows the type of curves which are obtained. stressrelaxation Time Fig. Proc. then the time required to relax the stress from the initial stress at to a is given by^ of curves also relieve residual stress by thermal treatments. t = BE{n'  1 l)(r"' ©" ASTM (1336) BIBLIOGRAPHY "Creep and Fracture London. Stationery Office. of time. Therefore. ASTM. M. 48. . 1 31 5. tests are service. PuU. p. ^ of Metals at High Temperatures.
1951. Inc." McGrawHill Book Inc." McGrawHill Book Company. Inc.: Recent Advances in Knowledge Concerning the Process of Creep in Metals. 1959. Ohio. (eds. 6. New Ohio. Heller: "Creep of Engineering Materials. 1956. Ltd... Ohio. Metals Park. 1959. Metals Park. "Utilization of Heat Resistant Alloys. Ault & Sons..Creep and Stress Rupture 369 Finnie. G. and G.): "High Temperature Materials. "Progress in Metal Physics. Company. H." vol.. 1950.: "Properties of Metals at Elevated Temperatures." John Wiley York. New York. R. Metals Park.. "Creep and Recoverj^. Smith. F.. Hehemann. 1957. R. v. . M." American Society for Metals." American Society for Metals. I. 1954. Pergamon Press. "High Temperature Properties of Metals. and W. New York." American Society for Metals. London. Sully. A.
A Critical Survey of Brittle Failure in Carbon Plate Steel Structures Other than Ships. Analysis of Brittle Behavior in Ship Plates. while. A They ' M. Shank.Chapter 14 BRITTLE FAILURE AND IMPACT TESTING 141. it is important to understand that this is not the only application where brittle fracture as the year 1886. E. The Brittlefailure Problem During World brittle failure of War II a great deal of attention was directed to the welded Liberty ships and T2 tankers. pipelines. (2) a low temperature. and bridges have been documented^ as far back Three basic factors contribute to a brittlecleavage type of fracture. A broad research program was undertaken to find the causes of these failures and to prescribe the remedies for their Failures occurred both future prevention. seas Most of the failures occurred during when the ships were in heavy and when they were anchored at dock. 158. Many of the results of this basic work are described in Chap. In addition to research designed to find answers to a pressing problem. L. in other instances. ASTM Spec. are (1) a triaxial state of stress. Williams. other research was aimed at gaining a better under standing of the mechanism of brittle fracture and fracture in general. pressure vessels. and (3) a high strain rate or rapid rate of loading. 7. Brittle of Temperature on the Symposium on Effect Behavior of Metals with Particular Reference to Low Temperatures. ASTM Spec. 45110. which should be reviewed before proceeding further with this chapter. These calamities focused attention on the fact that normally ductile mild steel can become brittle under certain conditions. the winter months. 158. Publ. 370 . 2 M. All three of these factors do not have to be present at the same time to produce brittle fracture. Tech. 1954. pp. While the brittle failure of ships concentrated great attention on brittle failure in mild steel. 1954. Brittle failures in tanks. Tech. Publ. the fracture did not completely disable the ship. is a problem. pp. 1144.^ Some of these ships broke completely in two.
Designation E2356T. and low temperature are responsible for most service failures of the brittle type. A is great deal of research has since demonstrated that welding. 142] triaxial state of stress. It is important to eliminate all stress raisers and to avoid making the structure too rigid. much work has been devoted to the development of additional tests for defining the tendency for brittle fracture. since these effects are accentuated at a high rate of loading. so that However. The design of a welded structure is more critical than the design of an equivalent riveted structure. However. . while the Izod specimen is favored in Great Britain. and much effort has been expended in correlating the results of different brittlefracture tests.Sec. se. strict quality control is stress raisers or notches. many types of impact tests have been used to determine the susceptibility of materials Steels which have identical properties when tested in to brittle fracture. Charpy bar specimens are used most commonly in the United States. Further no general agreement on the interpretation or significance of results obtained with this type of test. if a brittle failure did occur. struction. However. Notchedbar Impact Tests Various types of notchedbar impact tests are used to determine the tendency of a material to behave in a brittle manner. 142. This type of test will detect differences between materials which are not observable in a tension test. The results obtained from notchedbar tests are not readily expressed in terms of design requirements. per not inferior in this respect to other types of construction. were incorporated in some of the wartime ships so that. Brittle Failure and Impact Testing 371 such as exists at a notch. needed to prevent weld defects which can act as New electrodes have been developed that make it possible to make a weld with better properties than the mildsteel plate. To this end. their tendency certain there are disadvantages to this type of test. Two classes of specimens have been standardized^ for notchedimpact testing. known as crack arresters. and much effort has gone into the development of safe designs for welded structures. 3. 1958. pt. it would not propagate completely through the structure. there is 1 ASTM Standards. riveted sections. The Charpy specimore. A large number of notchedbar test specimens of different design have been used by investigators of the brittle fracture of metals. since it is not possible to measure the components of the triaxial stress condition at the notch. torsion at slow strain rates can show pronounced differences in tension or for brittle fracture when tested in a notchedimpact test. Since the ship failures occurred primarily in structures of welded conit was considered for a time that this method of fabrication for service was not suitable where brittle fracture might be encountered.
Sketch showing method of loading in Charpy and Izod impact tests. beam and is struck with the pendulum at the opposite end. depend strongly on the dimensions of the bar and the For this reason it is important to use standard The value of the transverse stress at the base of the notch depends chiefly on the relationship between the width of the bar and the notch radius. The load is applied by the impact of a heavy swinging pendulum (approximately men length. In Europe impact results are frequently expressed Very in energy absorbed per unit crosssectional area of the specimen. specimens. The relative values of the three Impact load 7777777. impact velocity) applied at the midspan of the beam on the side The specimen is forced to bend and fracture The Izod specimen is at a strain rate on the order of 10* in. 7/7777/ Charpy Vnotch Fig. often a measure of ductility. /(in.) (sec). Plastic constraint at the notch produces a triaxial state of 16 ft /sec opposite from the notch. Izod 141.372 Applications to Materials Testing [Chap. stress similar to that shown Impact load in Fig. Figure 141 illustrates the type of loading used with these tests. Note that the notch is subjected to a tensile stress as the specimen is bent by the moving pendulum. usually measured by For metals this is usually expressed in footpounds and is read directly from a calibrated dial on the impact tester. The Charpy specimen is supported as a beam in a horizontal position. is used to supplement this information. either circular or square in cross section and contains a V notch near one The specimen is clamped vertically at one end like a cantilever end. Figure 142 illustrates the appearance of these two types of fractures. The wider the bar in relation to the radius of the notch. the greater the transverse stress. principal stresses details of the notch. 14 has a square cross section and contains a notch at the center of its Either a V notch or a keyhole notch is used. failure) or granular (cleavage fracture). 710. examine the fracture surface to determine whether it is fibrous (shear The response of a specimen to the impact test is the energy absorbed in fracturing the specimen. such as the per cent contraction at the It is also important to notch. The notchedbar impact test is most meaningful when conducted over a range of temperature so that the temperature at which the ductileto .
yet its transition ing on roomtemperature results. The material with the lowest transition temperature is to be preferred. Frazier.S factor is the transition temperature. selecting a material from the stand4020the point of notch toughness or tend ency for portant trates brittle failure im. right. Figure 143 illustrates the type of curves which are obtained. Fracture surfaces of Left. 142. 212°F. center. decrease does not occur sharply at a certain temperature. Publ. 100°F. Spretnak. Note gradual decrease in the granular region notch with increasing temperature. W.)ec. it difficult Note that the energy absorbed decreases with decreasing temperature but that for most cases the This makes In 4:60o to determine accurately the transition temperature. and F. Boulger. 40°F. Charpy specimens tested at different temperatures. and increase in lateral contraction at the brittle transition takes place can be determined. °F can be misleading. 1954. 143. ASTM Spec. J. W. Steel A shows Fig. . pp. 286307.^ particu larly in the region of the transition temperature. temperature is higher than that of steel B. showing fallacy of dependtemperature. 142] Brittle Failure and Impact Testing 373 Fig. Tech. Figure reliance 143 illus on impact 80 40 +40 +80 resistance at only one temperature Temperature. H. 1 Most of this scatter is Effect of R. how Notchedbar impact tests are subject to considerable scatter. Transitiontemperature curves higher notch toughness at room for two steels. 158. Symposium on Tem perature on the Brittle Behavior of Metals.
"F Fig.1 8 percent C 60 50 Energy transition 0. 7 44. p. while in [Chap. 222. Pellini. fracture appearductility.54 percent Mn 0. ance. 1954.374 due to Applications to Materials Testing local variations in the properties of the steel.) and notch {. 14 difficulties some is due to preparing perfectly reproducible notches. Transitiontemperature curves based on energy absorbed. Publ. as is the proper placement of the specimen in the impact machine.07 percent Si ^ r 40 Charpy keyhole =^30 a> i5 20 10 100 80 fe"60 to 40 20 60 40 20 20 40 60 100 120 140 Ductility transition 20 40 60 100 120 140 160 Temporcture. The shape of the transition curve depends on the type of test and also on the material. For example. . 158.W S. Both notch shape and depth are critical variables. keyhole Charpy specimens usually give Semikilled mild steel . Tech.0. ASTM Spec.
and G. E. 158. 143. as measured by the contraction at the root of the notch. 2 G. energy transition temperature for Vnotch Charpy specimens is frequently set at a level of 10 or 15 ftlb. the width produced during the bending of an unnotched much greater than the thickness. with decreasing temperature. it is impor tant to understand the criteria which have been adopted for its definition. The most suitable criterion for selecting the transition temperature is whether or not it correlates with service performance.. change in the appearance of the fracture. A biaxial state of stress is is beam when ratio of axial.^ ' The unnotchedbend Pellini. Exptl. When the width to thickness is close to unity. 143] Brittle Failure and Impact Testing 375 For a tough Vnotch Charpy specimens generally give somewhat higher values than keyhole specimens. Soc. Sangdahl. The ductility transition temperature is sometimes arbitrarily set at 1 per cent lateral contraction at the notch. S. the condition for a state of plane stress. pp. S. W. section. This figure also illustrates the relative shapes of the The curves obtained with keyhole and Vnotch Charpy specimens. pp. Tech. 118. the stress is essentially unibut as the width increases. Proc. Aul. vol. or a transition in the ductility. the ratio of the transverse to longitudinal stress approaches a value of }'2. the transition temperature is frequently selected to correspond to a temperature where 50 per cent fibrous (shear) fracture is obtained. 1. Symposium on Brittle Behavior of Metals with Particular Reference to Low Temperatures. One characteristic of these criteria is that a transition temperature based on fracture appearance always occurs at a higher temperature than if based on a ductility or energy criterion. temperatures measured in different ways will be discussed in a later steel a sharper breaking curve than Vnotch Charpy specimens. criteria for determining the transition temperature are based on a transition in energy absorbed.Sec. 1948. Publ. Effect of Temperature on the no. ASTM Spec. 1954. . Stress Anal. 6. Sachs.' Figure 144 shows that the same type of curve is obtained for each criterion. In general. The transition temperature for a given steel will be different for differentshaped specimens and for different types The correlation of transition of loading with different states of stress. Slowbend Tests The slow bending of flatbeam specimens in a testing machine is sometimes used as a method of determining the transition temperature. L. test represents a condition of severity inter Evaluation of the Significance of Charpy Tests. Where the fracture appearance changes gradually from shear through mixtures of shear and cleavage to complete cleavage. 216261. Because the transition temperature is not sharply defined.
Both specimens are bent with the load applied opposite to the notch. 45° notch O.05"deep Weld bead. The Lehigh specimen provides are obtained duplicate tests. 14 mediate between that of the tensile test and a notchedimpact test. In a comparison of the transition temperature measured with a slowbend test and a Charpy impact test with identically notched specimens it was found that the ductility transition was raised by impact but that the fracture From this and other work. Usunotch is used to introduce triaxial stress. or total energy absorbed. /^ 4" KINZEL BEND SPECIMEN Notch radius 1 mm  0. Lateral contraction at the notch and bend angle are also measured. Notchbend test specimens. The effect of adding the variable of high strain rate is complex.Ol" radius 0. Both specimens incorporate a longitudinal weld bead which is notched so that the weld metal. 1 45. appears as if the fracture transition temperature is not sensitive to ally a strain rate.080" deep LEHIGH. it transition was lower for the impact test. BEND SPECIMEN Fis. energy absorbed after maximum load to fracture.376 Applications to Materials Testing [Chap. The Kinzel and Lehigh (Fig. the heataffected zone. and the unaffected base metal are exposed to the stress at the root of the notch. in which case the transition temperature is raised. in and the data plotted Loaddeflection curves terms of energy absorbed to maxi mum load. . 145) notchbend specimens are frequently used for studying the effect of welding and metallurgical variables on notch toughness.
A. pp. vol. p. interesting of these tests. 146. The advantage of a combined tension plus bending load over one of bending alone is that by suitably increasing the tensile load the compression region developed by the bending load can be eliminated. 146a. Robertson. which give indication of attaining more general steel A number acceptance. Welding J. 1949. J. 153s165s. Specialized Tests for Transition Temperature of new tests for determining the transition temperature of have been developed as a result of the research on the brittle failure Space will permit only a brief description of several of the most of ships. {London). Imbemo. vol. of tests subject the notch to simultaneous tension and This can be done by eccentrically loading a notched tensile Sow Impact cut Weld ^ (^ A_ Heat applied Liquid Ng coolant [a) Fig. 29. specimen or by using a specimen such as shown in Fig. (6) specimen used in Robertson test.. . 175. A. vol. pp. Since a highcompression region will retard crack propagation. a test which combines both bending and tension aids in crack propagation. 445448. S. Robertson^ devised an interesting test for determining the temperature at which a rapidly moving crack comes to rest. 1951. N. Iron Steel Inst. A number bending. 172. (a) Specimen used in Navy tear test. 1953.^ It employs the full thickness of the steel plate. A uniform tensile stress 1 2 T. Engineering. Kahn and E.Sec. 1 44] Brittle Failure and impact Testing 377 144. 361. This specimen is used in the Navy tear test.
Northrup.. a temperature gradient is maintained across the width perature. J. 1954. A crack is started at the cold end by an impact gun. pp. 228231. the of The energy available from the impact is not sufficient to make the crack grow very large. Welding J. Ibid. specimen. but the applied tensile stress tends to keep it growing.square plate. which the crack will not propagate. 2 3 cit. R. Feely.378 is is Applications to Materials Testing [Chap. Pellini. Hrtko. S. op. 1466. S. cracklike notch. pp. for further spread of the crack cannot be supplied by the stored elastic For each applied tensile stress there is a temperature above energy. A The crackstarter weld is applied to the center of a 14in.^ The dropweight test was developed by the Naval Research Labocooled with liquid nitrogen. The highest temperature of fracture is termed the nilductility transition temperature. in that at a given This is a "gono go" type of test temperature the specimen either fractures completely or remains intact. 3in. This test has been modified for use without a temperature gradient. 233235. Robertson's data showed that this arrest temperature decreases sharply for most mildsteel plates when the applied tensile stress is lowered to about 10. The welded side of the specimen is placed face down over end supports. pp. 33.. cut with a jeweler's A starter crack This side is kept at a higher temThus. D.long bead of hardfacing in the presence of the cracklike notch. Only 3° of bend is needed to produce a crack in the brittle weld bead. . 1 F. and the other side is saw at one side of the specimen. the bending fixture is designed so as to limit the deflection of the specimen to 5°. producing a sharp. The additional 2° of bend provides a test of whether or not the steel can deform the presence of a cracklike notch. and M. 14 applied to a specimen of the type shown in Fig. and the center of the specimen is struck with a 60lb falling weight. For these steels the crack will not be arrested if the stress exceeds this value and the temperature is below room temperature.. ratory^ to measure susceptibility to the initiation of brittle fracture in The specimen is a flat plate with a metal applied at the center and notched to half depth. vol. The crack travels across the width of the specimen until it reaches a point where the temperature is high enough to permit enough yielding This occurs when the plastic deformation required to stop the crack. Since the purpose of the dropweight test is to see whether or not fracture will occur at a sharp notch when the amount of yielding that can occur is restricted. The bead of hardfacing metal cracks in a brittle manner. The explosionbulge test was developed by the Naval Research Laboratory^ to measure susceptibility to propagation of brittle fracture.000 psi. 99sllls. Kleppe. This test provides a sharp transition temperature and is quite reproducible.
Concept of two temperatures. because fracture occurs by the shear mode with appreciable absorption of energy. transition the Temperature Fig. Actual materials do not have two distinct transition temperatures such as were shown in Fig. 145. Instead. 147. but once initiated they propa gate rapidly with little energy absorption. Figure 147 shows the transitiontemperature curves for such an ideal material. The dropweight test establishes a temperature below which the material is very susceptible to fracture initiation. The ductility transition temperature usually occurs at . ature is The fractureappearance which cracking transition temperis selected as the temperature at confined to the bulged region of the plate. Above the fracture — transition temperature cracks do not propagate catastrophically. SisniFicance of the Transition Temperature The notch toughness of a material should really be considered in terms of two distinct transition temperatures. The frac tureappearance ature is transition temperof related to the crackpropcharacteristics agation material. placed over a circular die and subjected to the force of a The explosion produces a compressive shock wave which test is is reflected from the bottom of the plate as a tensile wave. 147. For most steels this transition temperature will fall 40 to 60°F above the nilductility transition of the dropweight test. Fig. At still higher temperatures the plate bulges considerably more. The two tests supplement each other. a higher temperature the plate bulges. tility The ducis transition temperature ret Ductility lated to the fractureinitiation tendencies of the material.> 145] Brittle Failure and Impact Testing 379 specimen is controlled explosion. while the explosionbulge test establishes a temper ature above which the material is immune to brittlefracture propagation. but the cracks still run to the edges of the plate. pletely brittle Comfracture transition tennperature cleavage occurs readily below the ductility transition temperature. and the crack becomes a shear crack which is confined to the center of the specimen. 148 is more characteristic of the type of curves that are obtained with Charpy Vnotch tests on mild steel. In the region between these two transition temper atures fractures are difficult to initiate. This At interpreted in terms of the appearance of the fracture in the plate.
14 ftlb is an energyature 20 ftlb. Usually it is taken at the temperature at which 50 per cent fibrous fracture is obtained. The greatest number of data exist for failed hull plates in welded ships. Ductility transition Shear fracture Ductile behavior in service Difficult crack initiation and propogation Temperature Fig. the ductility transition temperature depends very strongly on the testing conditions. Part of the difficulty in correlating notchimpact data has been caused by failure to recognize and distinguish between the two general types of . The ductility transition temperature is usualso be may the root of the notch. and rate of loading. Experience with rimmed and semikilled mildsteel plates in thicknesses up to 1 in. Tests on these steels showed that they all had Charpy Vnotch values of 11. The fracture transition temperature always occurs at a higher temperature than the ductility transition temperature. not necessary to worry about its propagation.^ 380 Applications to Materials Testing level of 5 to [Chap. indicates that a minimum Charpy Vnotch value of 15 ftlb at the lowest operating temperature should prevent brittle fracture if the nominal stresses are of the order of onehalf the yield point. ally more pertinent then to service performance. For a given material the fractureappearance transition temperature is fairly constant regardless of specimen geometry. Frequently a value of 15 used to establish this transition temperature. because it is if it is difficult to initiate a crack.4 ftlb or less when tested at the temperature at which failure occurred. The ductility transition temper determined from measurements of the contraction at The fractureappearance transition temperature is measured by the per cent shear in the fracture surface. Significance of regions of transitiontemperature curve. On the other hand. There is no general correlation between any of the brittlefracture tests and service performance. — 148. notch sharpness. For higher alloy steels a higher value minimum impact resistance may be required.
John Wiley & Sons. 1957.1 per cent manganese. it is not generally advisable to use steel made by the Bessemer process for lowtemperature applications. 146] Brittle Failure and Impact Testing 381 Comparisons should not be made of transitiontemperature criteria. R. 43. good correlation has been obtained' between the nilductility transition measured by the dropweight test and the Charpy Vnotch test. Good correlation is found between the two specimens when they are both evalu men but ated with common ductility criteria. Good correlation has been fractures. ASM. Metal Progr. 3 New York.1 per cent carbon. 1957. "Brittle Behavior of Engineering Structures. This transition temperature is lowered about 10°F for each increase of 0. shown between the dropweighttest nilductility transition and service The correlation problem is well illustrated by the work of the Ship Structure Committee. 71. between test results where the two criteria have been mixed. 1951. 1 2 H.01 per cent phosphorus. Further. no correlation is found between the keyhole and Vnotch Charpy tests when the transition temperature is measured at a level correspondThis is because the 50 per ing to 50 per cent of the maximum energy. Metallurgical Factors Affecting Transition Temperature temperature of over 100°F can be produced by chemical composition or microstructure of mild steel. vol. vol. Jr.. Greenberg. Rinebolt and W.^ The 15ftlb transition temperature Vnotch Charpy specimens (ductility transition) is raised about 25°F for each increase of 0." chap. Inc. Phosphorus also has a strong effect in raising the transition temperThe 15ftlb Vnotch Charpy transition temperature is raised about 13°F for each 0. toughness. pp. pp. A maximum decrease of about 100°F in transition temperThe practical ature appears possible by going to higher Mn C ratios. . 7581.2 per cent carbon is needed to maintain the required tensile amount : : properties. Harris. to assess because of its interaction with other elements..^ 146. E. The largest changes in transition temperature result from changes in the in transition Changes changes in the for of carbon and manganese. June. while about 0. cent energy level is close to the ductility transition for the keyhole specinear the fracture transition for the Vnotch specimen.Sec. Since it is necessary to ature. Trans. Parker. For example. 11751214.4 per cent lead to trouble with retained austenite. A. Increasing the carbon content also has a pronounced effect on the maximum energy and the shape of the energy transitiontemperature curves (Fig. J. control phosphorus. The Mn:C ratio should be at least 3:1 for satisfactory notch 149). 6. J. is difficult The role of nitrogen It is. limitations to extending this beyond 7 1 are that manganese contents above about 1..
. which are deoxidized with silicon.057 per cent. Effect of carbon content on the energytransitiontemperature curves for {J. Iron Steel Inst. Aluminum also has the beneficial effect of combining with nitrogen to form insoluble aluminum nitride. P. 200 Temperature. In view of these results. 43. p. vol.001 per cent to the high value of iron^ Corbon 280 200 Fig. Rimmed steel. J. (London). and H. Molybdenum raises the transition almost as rapidly as carbon. B. Tipler. 0. it is not surprising that deoxidation practice has an important effect on the transition temperature. steel. vol. 1952. E.382 Applications to Materials Testing [Chap. while chromium has little effect. 172 pp. For highpurity it was found that oxygen contents above 0. Trans. A. Nickel is generally accepted to be beneficial to notch toughness in amounts up to 2 per cent and seems to be particularly effective in lowerSilicon. Semikilled steels. Hopkins. while for steels which are fully killed with silicon plus aluminum the 15 ftlb transition temperature will be around — 75°F. appears to raise the transition temperature.) 0.003 per cent produced intergranular fracture and corresponding low energy absorption. generally considered to be detrimental to notch toughness. Jr.25 per cent. generally shows a transition temperature above room temperature. the transition temperature was raised from 5 to 650°F. Rees. The use of a fully killed deoxidation practice is not a completely practical answer to the problem of making steel plate with high notch toughness because there is only limited capacity for this type of production. 1951. Harris. ASM. with its high iron oxide content. 14 however. An increase W. in amounts over ing the ductility transition temperature. /. Rinebolt and W. R. Notch toughness is particularly influenced by oxygen. have a lower transition temperature.. When the oxygen content was raised from 0. °F 400 149. 1197. size Grain 1 has a strong effect on transition temperature. 403409.
The phenomenon of blue hrittleness. is due to strain aging. allowfine pearlite and grain size in a thick section. vol. Quench aging is caused by carbide precipitation in a lowcarbon steel which has been quenched from around 1300°F. Averbach. this treatment results in reduced transition temperature. in. in which a decrease in impact resistance occurs on heating to around 400°F. Decreasing the grain diameter from ASTM grain size 5 to ASTM grain size 10 can change the 10ftlb Charpy Vnotch similar effect of is transition temperature^ from about 70 to — 60°F. . in thickness. the transition temperature will be appreciably higher in thick hotrolled plates than in thin plates. For a given chemical composition and deoxidation practice. cooling rate from the normalizing treatment are variables which also The and the deoxidation practice must be considered. The best notch toughness is obtained with a microstructure which is A completely pearlitic structure has poor notch toughness. Quench aging results in less loss of impact properties than strain aging. D. will perature at 25 ftlb was 350°F lower for the tempered martensitic strucFurther discussion of the notch toughness of heattreated steels be found in Sec. Whitmore. Lowcarbon steels can exhibit two types of aging phenomena which produce an increase in transition temperature. usually around 40 to 60°F. L. Strain aging occurs in lowcarbon steel which has been coldworked. structure on transition temperature.. possible finishing temperature for hot rolling of plate also beneficial. Air cooling and aluminum Using the lowest is deoxidation result in a lower transition temperature. rolling results in a grain refinement. and a structure which is predominately bainite As an example of the effect of microis intermediate between these two. Many of the variables concerned with processing mild steel affect the ferrite grain size and Since normalizing after hot therefore affect the transition temperature. ance for this effect must be made in plates greater than ^ The notch toughness of steel is greatly influenced by microstructure. 1957. ture. Cohen. H. This is due to the difficulty of obtaining uniformly Generally speaking. 146] of Brittle Failure and Impact Testing 383 one ASTM number in the ferrite grain size (actually a decrease in grain diameter) can result in a decrease in transition temperature of 30°F for mild steel. 36. S. it was found that the Charpykeyhole transition temcompletely tempered martensite. 1 W. 503s511s.Sec. and B. Cold working by itself will increase the transition temperature. but strain aging results in a greater increase. Welding J. 148. pp. A decreasing transition temperature with decreasing austenite grain size observed with higher alloyed heattreated steels. in an SAE 4340 steel for which a tempered martensitic structure and pearlitic structure were compared at the same hardness. if not carried out at too high a temperature. M. Owen.
it seems that specimen and notch orientation are not a very important variable for this criterion. but the specimen is oriented transverse to the rolling direction. Specimens A and B are oriented in the longitudinal direction of the plate. Fig. 14 The notchedimpact properties of rolled or forged products vary with orientation in the plate or bar. The orientation of the notch in specimen A is generally preferred. °F Fig. while in B the notch lies parallel to the plate surface. If. 147. In specimen A the notch is perpendicular to the plate. Transverse specimens are used in cases where the stress distribution is such that the crack would propagate parallel to the rolling direction. Figure 1410 shows the typical form of the energytemperature curves for specimens cut in the longitudinal and transverse direction of a rolled plate. orientation can greatly affect the results. 1410 shows that quite large differences can be expected for different specimen orientations at high energy levels. Since ductility transition temperatures are evaluated in this region of energy. + 80 +120 1410. however. Effect of specimen orientation on Charpy transitiontemperature curves. materials are compared on the basis of roomtemperature impact properties. . In specimen C this notch orientation is used. Effect of Section S ize Difficulty with brittle fracture usually increases as the size of the struc ture increases. but the differences become much less at energy levels below 20 ftlb. Reference to 80 :? 60 S40 l5 20 +40 Temperature. This is due to both metallurgical and geometrical factors.384 Applications to Materials Testing [Chap.
J. These models were similar in all details and were made from the same material by the same welding procedures. 206. tests on specimens of varying size but identical metallurgical structure and geometrically similar notches show that there is a size At some temperature the largest specimens will be completely brittle. 9 was shown that the tensile properties of tempered martensites of the same hardness and carbon content are alike. and onequarter scale models were tested. 1956. 148. The due to larger structure can contain a stress raisers. The frac tures for inbetween specimens will vary from almost fully ductile to almost fully brittle. terion requires that the elastic strain energy Since the Griffith must provide the surface energy for the formation of the fracture surface. Full scale. AIME. When fracture strength was measured in terms of pounds per square inch of the net crosssectional area. the fullsized specimen had only about onehalf the strength of the quarterscale model. . but not valid for the variation of impact resistance with temperature. Even greater spread in transition temperature would be obtained if the tempering temperature were adjusted to give a higher hardness.4 per cent carbon and with a tempered martensite structure produced by quenching and tempering to a hardness of Re 35. Charpy effect. more unfavorable state of stress and it also procri vides a large reservoir for stored elastic energy. The higher structures is transition temperature or lower fracture stress of large due to two factors. structure produces the best combination of strength and impact resist ance of any microstructure that can be produced in it In Chap. pp. French.^ Slack all 1 number H. all having about 0. Trans. the easier it is for the attainment of an uncontrollable. Note that a maximum variation of about 200°F in the transition temperature at the 20ftlb level is possible. Fig ure 1411 shows the temperature dependence of impact resistance for a of different alloy steels. However. while the small specimens will be completely ductile.Sec. rapidly spreading crack. Notch Toughness of Heattreated Steels It has been demonstrated many times that a tempered martensitic steel. irrespective of the amount of other alloy additions. A dramatic demonstration of size effect was obtained in tests of ship hatch corners carried out at the University of California. the greater the available stored energy. 148] Brittle Failure and Impact Testing 385 In the previous section it was shown that the transition temperature of a given steel usually decreases with increasing plate thickness because of the increased grain size produced in hotrolling thick plates. 770782. This generalization holds approximately for it is the roomtemperature impact resistance of heattreated steels. vol. onehalf scale.
bainite. and 80 70 60 50 S40 °30 20 10 . ature. 14 quenching so that the microstructure consists of a mixture of tempered martensite. and peariite results in even greater differences between alloy steels in a general increase in the transition temperature.386 Applications to Materials Testing [Chap. in the impact test of an alloy steel at a given absorbed The energy increases with increasing tempering tempergenerally temperature test minimum in the curve in the general region is a there However.
but they severely reduce the impact resistance. 40 35 30 25 . Ch Bornett. These platelets have no effect on the reduction of area of a tensile specimen. They can be formed at temperatures as low as 212°F and as high as 800°F.149] Brittle Failure and Impact Testing 387 from ecarbide during the second stage of tempering. Silicon additions of around 2. depending on the time allowed for the reaction.25 per cent are 50 4330 8 4340: 45 Klingler.
The presence of temper embrittlement is usually determined by measuring the transition temperature by means of a notchedbar impact test.388 Applications to Materials Testing [Chap. The kinetics of the process produces a Cshaped curve when some parameter of embrittlement is plotted on temperaturetime coordinates. while the brittle fracture of a nonembrittled steel is transgranular. 1410. Therefore it is generally hypothesized that temper embrittlement is due to the segregation of impurities to the grain boundaries without the formation of an observable precipitate phase. Hydroscn Embrittlement Severe embrittlement can be produced in many metals by very small of hydrogen. The effect of various alloying elements on this embrittlement can then be explained on the basis of their rates of diffusion and relative solubilities at the grain boundaries and within the grains. Molybdenum is the only alloying element which decreases the susceptibility to temper embrittlement. 14 This can become a particularly important problem with heavy sections that cannot be cooled through this region rapidly enough to suppress Temper embrittlement also can be produced by isoembrittlement. The fracture of a temperembrittled steel is intergranular. The hardness and tensile properties are not sensitive to the embrittlement. Tempering at a higher temperature for a short time may be better than a long tempering treatment at a lower temperature.0001 weight per cent of hydrogen can cause cracking in As steel. Much more information is needed before a detailed mechanism of temper embrittlement can be determined. no evidence for a grainboundary film or precipitate has been uncovered from studies by means of the electron microscope of the microstructure of temperembrittled steel. A water quench from the tempering temperature will serve to minimize embrittlement on cooling. Temper enibrittlement can be completely eliminated from an embrittled steel by heating into the austenite region and cooling rapidly through the embrittling temperature region. little as Face centered cubic metals are not generally susceptible to hydrogen embrittle . This suggests that temper brittleness is due to a grainboundary weakness. but the transition temperature can be increased around 200°F by ordinary embrittling heat treatments. Bodycentered cubic and hexagonal closepacked amounts metals are most susceptible to hydrogen embrittlement. except for very extreme cases. More rapid embrittlement results from slow cooling through the critical temperature region than from isothermal treatment. The best solution to the problem is to avoid tempering in the region of greatest susceptibility to embrittlement. 0. thermal treatments in this temperature region. However.
and observe the time to chief characteristics of sensitivity. it diffuses very rapidly at temperatures above room temperatvire. or perhaps simply regions of high dislocation density.^ during solidification. or Hydrogen may be introduced during melting and entrapped it may be picked up during heat treatment. and its susceptibility to delayed Unlike most embrittling phenomena. electroplating. steel The hydrogen content of be reduced by "baking. These voids might be true voids. The simi larity of the delayed fracture curve to the fatigue SN curve has led to the use of the term "static fatigue" for the delayedfracture phenome non. detect hydrogen embrittlement by a drastic decrease in ductility. due to the reaction of by hydrogen at elevated hydrogen with oxygen to form internal pockets . At low^ temperatures and high temis peratures hydrogen embrittlement is negligible. hydrogen embrittlement fracture." or heating at around 300 to 500°F. but notchedimpact tests are of no use for detecting the phenomenon. Because it is a small interstitial atom. its failure.1410] Brittle Failure and Impact Testing 389 ment. As hydrogen diffuses 1 may The familiar example of the embrittlement of copper is temperature of steam. or "static fatigue limit. For steel the region of greatest susceptibility to hydrogen embrittlement is in the vicinity of room temSlow bend tests and notched and unnotched tension tests will perature. microcracks. A commonly held concept of hydrogen embrittlement is that monatomic hydrogen precipitates at internal voids as molecular hydrogen. A common method of studying hydrogen embrittlement is to charge notched tensile specimens with know^n amounts of hydrogen. 1413. fracture wnll not occur. hydrogen embrittlement are its strainrate temperature dependence. Note that the notched tensile strength of a charged specimen may be much lower than the strength of a hydrogenfree specimen. but it is most severe in some intermediate temperature region. Hydrogen is present in solution as monatomic hj^drogen. The minimum critical stress. or welding." increases with a decrease in hydrogen content or a decrease in the severity of the notch. The A typical delayedfracture curve is shown in Fig. acid pickling. There is a re gion in which the time to fracture depends only slightly on the applied There is also a minimum critical value below which delayed stress. load them to different stresses in a deadweight machine. enhanced by slow strain rates.
Further insight into the mechanism has resulted from the work of Troiano and coworkers. Dana. 14 up and produces fracture. S. Pearson. ^ S. J. 5480. E. Proc. the lower the necessary critical stress. vol. very high amounts of hydrogen can be tolerated without the transformation stresses are minimized by decom posing the austenite above the Ms temperature before cooling to room temperature. Troiano. Flow and Fracture under Very Rapid Rates of Loading The mechanical properties of metals can be appreciably changed when they are subjected to very rapidly applied loads. of for flaking varies widely with composition. in large ingots and forgings during cooling or roomtemperature aging has long been attributed to the Studies^ of flake formation have shown that in presence of hydrogen. On the other hand.390 Applications to Materials Testing [Chap. it is not in agreement with all the experimental facts. The main effect of stress is to assist in the accumulation of this concenThe minimum critical stress in Fig. Troiano. 2 A. of the crack tip. J. 203. Duwez and D." Metals Park. to link up with the previous main crack. The formation of hairline cracks. pp. higher the average hydrogen content. a small crack forms and grows. 47. While this concept explains the general idea of hydrogen embrittlement. 1413 can be interpreted tration. vol. The hydrogen content necessary and segregation. . 1947. Trans. or flakes. pp. The fact that the time for the initiation of the first crack has been found to be insensitive to applied stress supports the idea that the process depends on the attainment of a critical hydrogen concentration. The as the stress needed to cause a critical accumulation of hydrogen. pp. size. 895905.^ By into the voids. Ohio. "Behavior of Metals under Impulsive Loads. Rinehart and ASTM. American Society for Metals. and A. R. J. the diffusion of hydrogen to the region of high triaxial stress just ahead When a critical hydrogen concentration is obtained. vol. the pressure builds determining the rate of crack propagation by means of resistivity measurements they were able to show that the crack propagates discontinuThis indicates that the rate of crack propagation is controlled by ously. 1955. Clark. 1411. AIME. W. 52. F. 1954. addition to containing hydrogen the steel must be subjected to transfor mation stresses for flaking to occur. 3 P. Shock loading can be produced by highvelocity impact machines^ or by shock waves from the detonation of explosives. ASM. 1960. Shortsleeve. R.* In considering dynamic loading of this type it is important to consider effects due to stresswave propagation within 1 A. 502532. Trans. Flakes have been observed causing flakes if in steel with as low as 3 ppm hydrogen.
de = Jo ~ ("— Co (143) P Jo These equations can be used to determine the stress or strain in a dynamically loaded metal provided that the wave and particle velocities can be determined.' This all is because a rapidly applied load not instantaneously transmitted to undisturbed. The wave velocity and particle velocity are in the same direction for a compressive wave. so that the elastic limit is not exceeded. When a bar is subjected to tension impact. Eq. = ('^' (141) where Co da/de p If = = = velocity of wave propagation slope of stressstrain curve density of metal is the wave amplitude low. (141) can be written Co = (T (142) Corresponding to the wave velocity co. By combining Eqs.Sec. The particle velocity Vp is related to the wave velocity co by the following equations: Vp = f\. 1411] Brittle Failure and Impact Testing is 391 the metal. but they are in opposite directions for a tensile wave. 1953 . at a brief instant after the load has been applied the remote portions of the body remain The deformation and stress produced by the load move through the body in the form of a wave that travels with a velocity ot the order of several thousand feet per second. Rather. Kolsky. Compression waves are generated in a metal when it is subjected to an explosive blast (impulsive loading). (141) and (143) the equation for the critical velocity is obtained. "Stress Waves in Solids. For shock loads below Vu the bar would undergo deformation but would not fracture." Oxford University Press. where €„ is the strain corresponding to the tensile strength of the metal. while tensile waves can be produced by a tensionimpac\ machine. parts of the loaded body. 1 H. The value of critical impact velocity for most metals lies in the range of 200 to 500 ft/sec. it is found that there is a critical velocity which produces rupture at the impacted end at the instant of impact. a certain particle velocity Vp is produced in the metal. New York. The propagation velocity of a compressive or tensile stress wave is given by c.
ticular Reference to Low Temperatures. 1954. Tipper. 1960. no. Appl. E. On Fractures Caused by Explosions and Impacts. M. 1956. and the plate is said to have scabbed. Ohio.4 392 Applications to Materials Testing of the [Chap. 555. Marked differences occur between fracture under impulsive loads and under static loads. Tech. no. There is generally an increase in the energy to fracture with increasing impact velocity up to the point where the critical velocity is reached. "The Embrittlement of Metals." American Society for Metals. p." Welding Research Council.000 to 20. S. New York. R. October. With impulsive loads there is not time for the stress to be disturbed throughout the entire body. 195261. vol. the interference from the incident and reflected wave from the opposite surface will cause a tensile stress to be The tensile stress built up a short distance from the opposite surface. 55. metal. J. 1951. BIBLIOGRAPHY & Sons York. Met. available indicates that for shock loading the stressstrain curve is raised about 10 to 20 per cent compared with the static curve. while the velocity of crack propagation is about 6. and it is this reflected tension wave which in most cases causes fracture under impulsive loading. New : ASTM 1 J. Quart. Therefore. and at discontinuities within the A compression wave is reflected from a surface as a tension wave.: "Control of Steel Construction to Avoid Brittle Failure. 22.: "Brittle Behavior of Engineering Structures. Publ. 7. 4. Phys." John Wiley Inc. Symposium on Effect of Temperature on the Brittle Behavior of Metals with ParSpec. pp. R. C. Parker. B. with impulsive loads it may be found that cracks have formed but did not have time to propagate before the stress Reflections of stress waves occur at free surfaces and state changed. fixed ends. School Mines. . 1957. 1957. at changes in cross section.000 ft/sec. Shank. Queneau. vol. 2. When a thick plate is subjected to explosive loading against one surface. so that fracture can occur in one part of the body independently of what happens in another part. 158. Metals Park. vol. Reviews.. Colo. E. is difficult 1 Measurement dynamic stressstrain curve because of the short time during which events occur and because care must be taken The information which is to consider all wavepropagation phenomena. may be high enough to cause fracture. F. Rinehart. From studying^ the thickness of the scabs it is possible to arrive at values for a critical normal fracture stress. 1957.000 ft/sec. The velocity of propagation of stress waves in solids lies in the range 3.: The Brittle Fracture of Metals at Atmospheric and Subzero Temperatures..
or lockedin stresses. while the center of the sheet strain fibers unchanged. while the surface fibers seek to stretch the central fibers of the sheet The result is a residualstress pattern in. (a) id) in rolling of sheet. For example. Residual stresses are produced whenever a body undergoes nonuniform plastic deformation. consider a metal sheet which is being rolled under conditions such that plastic flow occurs only near the surfaces of the sheet (Fig. the surface and center must undergo a accommodation. (6) resulting distribu Inhomogeneous deformation tion of longitudinal residual stress over thickness of sheet (schematic). 15la). Since the sheet must remain of the sheet a continuous whole. 151. They are sometimes referred to as internal stresses. The surface fibers of the sheet are coldworked is and tend to elongate. the sheet 393 . Origin of Residual Stresses Residual stresses are the system of stresses which can exist in a body when it is free from external forces. The center fibers tend to restrain the surface from elongating.Chapter 15 RESIDUAL STRESSES 51 1 . (a) Fig.
by plastic deformation until reaches the value of the yield It is important to distinguish between macro residual stresses and Macro residual stresses. Residual stresses are to be considered as only elastic stresses. with which this chapter is primicrostresses. 1516 this means that the area under the curve subjected to compressive residual stresses must balance the librium. Frequently. Thus. the sign of the residual stress which is produced by inhomogeneous deformation will be opposite to the sign of the plastic strain which produced the residual stress. This. Another example is the precipitation of secondphase particles from solid "solution. the residual stresses acting across the width and thickness of the sheet should be considered. then each particle in t'^ying to occupy a larger volume is compressed by the matrix.394 Applications to Materials Testing of a high [Chap. develops tensile stresses in the matrix in directions radial and tangential to the secondphase particles. A complete determination is of the state of resid ual stress in three dimensions a very considerable undertaking. The area subjected to tensile residual stresses. Because of the anisotropy of the elastic properties of crystals. for the case of the rolled sheet. because of symmetry. in turn. fibers left in which were elongated in the longitudinal direction by rolling are is a state of compressive residual stress residualstress when the external load removed. The maximum to oppose value which the residual stress can reach is the elastic limit of the material. it.omponents from which they formed. the total force acting on any plane through the body and the total moment of forces on any plane must be zero. Actually. vary continuously through the volume of the body and act over regions which are large compared with atomic dimensions. microThe back stress developed stresses will vary greatly from grain to grain. system existing in a body must be in static equiThus. although their effects may extend throughout most of a grain. for a complete analysis. although measure . by a pileup of dislocations is an example of this type of residual stress. If the precipitate particles occupy a larger volume than the . A stress in excess of this value. with no external force it will relieve itself stress. if the secondphase particles have a lower density than the matrix. For the longitudinal stress pattern in Fig. 1516). The situation is not quite so simple as is pictured in Fig. The experimental determination of these localized stresses in twophase systems is very difficult.. 15 compressive stress at the surface and a tensile In general. and the state of residual stress at any point is a combined stress derived from the residual stresses in the three principal directions.e. or textural stresses. the surface which consists residual stress at the center of the sheet (Fig. act over dimensions as small as several unit cells. marily concerned. i. 151. Microstresses. only the residual stress in one direction need be considered.
1957. N. it will not be considered further in this chapter. Nitriding and carburizing are processes in which a microstress distribution is produced around each nitride or carbide particle. the temperature differences which are present between the surface and the center may be enough to chemical composition or rate of heat transfer. a 59. hot ingot of a metal w^hich shows no phase change.''^ lations of the microstresses existing in twophase systems have been made by Lazlo/ who plastically uses the terminology "tessellated stresses" for this type of residual stress. J. However. there in the distribution of microstresses which will 1 J. R. may members of the assembly. C. J. Trans. The determination of the microstresses is which exist in necessary for an understanding of the mechanism of strain hardening. such as welded structures. Research Natl. Vacher. . the structure structure is not subjected to external loads. 239 243. . In the cooling of a large. principal methods by which this can occur are by inhomogeneous changes in volume and in shape. " For review of Lazlo's exten:sivo and detailed work. there is a nonuniform volume increase in this region. see F. pp.^!etals. Thus. Gurland. a macro compressive residual stress is produced on the surface. Volume changes need not necessarily involve rapid quenching or phase changes to produce residual stresses. 151] Residual Stresses 395 Calcu ments of their average value have been made with X rays. ASM. Newton and H. vol." p. "Sym posiuni on Internal Stresses. Standards. but because these diffusioncontrolled reactions occur only on the surface. it will be considered in a separate section of this chapter. different Even though members of the be under stress due to various interactions between the This type of residual stress is called reaction stress. is an nonuniform volume change which produces very localized micro residual stresses. Estimates of these microstresses can be made from detailed analysis of the broadening of Xray diffraction lines. Because it falls in the area of structural engineering. Bur. In.titute of London. and this is balanced by tensile residual stresses in the interior. if the reaction does not proceed uniformly over the cross section of the body because of differences in either The precipitation of secondphase particles in a metal matrix of a example will be a variation produce macro residual stresses. 50.Sec. 10631071. Residual stresses arise from nonuniform plastic deformation of a body. Gl. pp. deformed singlephase metals Further improvements on the techniques are needed before these measurements can be used without ambiguity. 1958. Because of the technological importance of this situation. vol. Nabarro. 2 C. 1948. A third source of residual stress may The exist in builtup assemblies. The phase transformation from austenite to martensite which occurs during the quenching of steel is an outstanding example of a nonuniform volume change leading to high residual stresses.
(d) ic) 152. the ingot shrinks to relieve some of the stress (Fig. vol. Baldwin. 1 5 center. ASTM. When the center of the slab finally cools. 12. ASTM. l52d). 541. pp. 49. Proc. Shot peening. develop residual stresses. 1526.) Development shown shaded.^ The edges of a hot slab cool faster than the The thermal contraction of the cooler edges produces a strain mismatch between the edges and center of the ingot which results in the it shown in Fig. . Since the hot cannot support the compressive stress imposed on that region and because of plastic deformation the center of distribution of longitudinal stresses center has a lower yield stress. Jr. surface hammering.. portions 1949.396 Applications to Materials Testing [Chap.. 539583. As was noted in Chap. Spot welding and butt welding both produce high tensile stresses at the center of the area of application of heat. Jr. shot peening is an effective method of reducing fatigue failures. operations required to convert metals to finished and semifinished shapes rarely produce homogeneous deformation of the Each particular plastic forming process has a residualstress distribution which certain extent is characteristic of the process and is influenced to a by the way in which the process has been carried out. the total contraction will be greater for the center than the edges because the center contracts owing to both cooling and plastic stressed in residual tension. The center will then be and the edges will be in compression. ia) Fig. {After Cool W. Baldwin. 49. Earlier in this section we have seen how the residual stress in the rolling direction of a sheet results from inhomogeneous deformation through the thickness of the sheet. M. 1949. The forming metal. deformation (Fig. Other metalworking operations which are not generally classed as metalforming processes can produce residual stresses because they involve inhomogeneous deformation. and surface rolling produce shallow biaxial compressive stresses on a surface which are balanced by biaxial tension stresses in the interior. 152c). of residual stresses during cooling of a hot ingot. p. Residual 1 W. vol. M. Proc.
for as will be seen in the next section. keyway in a colddrawn bar. the final deformation process determines the resulting residualstress pattern. bility. However. depending upon the conditions of the plating process. In autofrettage thickwalled cylinders are purposely strained beyond the elastic limit of the material at the bore of the cylinder so that this region will contain compressive residual stresses when the cylinder is unloaded. Warping due to redistribution of residual stresses when surface layers are removed can be exceedingly troublesome. superposition of stress distributions is a valid procedure when one is considering the effect of residual stress on the response of a body For all practical purposes residual stresses to an external stress system. 1 52. machined notches a problem in notch tensile is The effect of residual stresses on fatigue performance recognized phenomenon and was considered in Sec. and residual tensile stresses will increase the ease with which failure occurs. The unknown residual stresses existing at the root of testing. can be considered the same as ordinary applied stresses. the measure . the residual stresses in the material removed are also eliminated. and the body distorts to establish a new equilibrium condition. In a relatively brittle material like highpossibility of is strength steel the presence of tensile residual stresses can cause a decrease in the fracture strength. and zinc creep sufficiently at most of tfiese plating stresses.)ec. In general. 129. This upsets the static equilibrium of internal forces and moments. The superposition of several deformation operations does not produce a final residualstress distribution which is the algebraic sum of the stress distributions produced by the preceding operations. Effects of Residual Stresses The presence rial of residual stresses to externally applied stresses. Conversely. compressive residual stresses will increase the yield stress. If a well Residual stresses are responsible for warping and dimensional instapart of a body containing residual stresses is machined away. This fact is used to strengthen gun tubes and pressure vessels by the process known as autofrettage. However. Thus. there is as in machining a long a useful aspect to this. particularly with precision parts like tools and dies. 1 52] Residual Stresses 397 stresses are developed in electroplated coatings. can have either high tensile or compressive residual room temperature to relieve Hard coatings like chromium and nickel stresses. a compressive residual stress will reduce the effectiveness of an applied tensile stress in producing fatigue failure. Soft coatings such as cadmium. lead. can influence the reaction of a mateIn the tension test. that region of a specimen containing high residual tensile stresses will yield plastically at a lower value of applied stress than a specimen without residual stresses.
Inc. AluminumNaCl Magnesiumpotassium chromate solution 1 53.398 Applications to Materials Testing of dimensional [Chap. below: 1. as Residual stresses cannot be determined directly from straingage is the case for stresses due to externally applied loads. Bauer. In this example the cylindrical bar assumed to contain tensile residual stresses around the periphery and ^A review of this important subject is given by W. Extreme care should be taken to minimize residual stress when these situations are likely to be encountered. residual stresses are calculated from the measurements of strain that are obtained when the body is sectioned and the lockedin residual stresses are released. Heyn and O. Mechanical Methods for Residualstress Measurement measurements. D. New York. pp. 153). Robertson (ed. MetaUog. or boiling 3. MgCl2 Mild steelboiling NaOH solution 4. residual stress in brass cartridge cases 2. The method developed by Bauer and Heyntudinal residual stresses in a cylinder is for measuring the longi a good illustration of the techis niques involved. Intern.). 2 E. 1. The residual stresses in the cylinder can be likened to a system of springs (Fig. 1956. 1911. standardized for detection (ASTM B154) Austenitic stainless steelboiling solution of 10 per cent of H2SO4 and 10 per cent CUSO4. These changes result from the deformation required to maintain equilibrium when the residualstress distribution changes from stress relaxation on longtime roomtemperature aging.. ." John Wiley & Sons.. In fact. Brassmercurous nitrate in water. vol. 5. ment one of the established Stress corrosion cracking^ is a type of failure which occurs when cer tain metals are subjected to stress in specific chemical enviroments. "Stress Corrosion Cracking and Embrittlement. The residualstress pattern in steels may also be altered by the transformation of retained austenite to martensite on aging. Z. Typical solutions which are used for this purpose are listed Residual stress is just as effective as stress load in producing stress corrosion cracking. Rather. accelerated stress corrosion cracking may be used as a qualitative test to indicate the presence of residual stresses. due to an externally applied Examples of combinations which produce stress corrosion cracking are mercury or ammonia compounds with brass (season cracking) and chlorides with austenitic stainless steels and certain agehardenable aluminum alloys. 15 changes when material is removed from a body is methods for measuring residual stresses. 1650. Dimensional instability refers to changes in dimensions which occur without any removal of material.
When the stress distribution approach the distribution is measured by the . 1546 can be determined by the Bauer and Heyn method if the stresses are determined by removing thin layers and measuring the deformation in the remaining portion. Ao is the original area of the cylinAi. is related dei = to the strain through Hooke's law. where Li is the expanded The stress relength of the element. 1546. due to removal of restraint of outer springs. gation experienced by the center springs is directly proportional to the force ex erted on them by the outer springs. 15. then the area of the skin dAi = Aq — The average stress existing in the skin is so that the force in the skin may be written Pskin = asdAi results in Equating the force in the core and the skin an equation for the average stress in the skin. (b) elongation of center portion. (Tc = E dei Since the cylinder was initially in equi librium before the skin was removed. 1546. 1536). of.3a). 154a. if the static equilibrium of forces is upset by removing the outer springs. (a) Heyn's spring model for longitudinal residual stresses in AiE dei = If Pskin a cylinder. lieved by this expansion. the center springs would be compressed and the outer springs elongated (Fig. 1 balance the force in the removed material. Actually. If sufficiently thin layers are removed and the process is repeated enough of longitudinal residual stress is times. the measured stress distribution will shown in Fig. 153] Residual Stresses 399 compressive stresses at the center. Now. the distribution more likely to vary in the continuous manner shown in Fig. The residualstress distribution shown in Fig.Sec. the force in the center core must [a) [d) Fig. drical bar. By the spring analogy. AiE The above equation dei (151) dAi expresses the residual stress when it has the very arbitrary distribution shown in Fig. the compressed springs will elon gate (Fig. The amount of elon ^—1. is as. The is strain experienced by the core dLi/L. 53.
However. (151) gives the residual stress in the first layer removed. the actual stress in the second layer as existed in the original bar is it given by AiE de2 dA2 . Eq.400 Applications to Materials Testing [Chap. 15 successive removal of thin layers. The of the first layer from. because the removal 1— t bar has caused a redistribution of stress in the remainder of the bar. this equation will not give a true indication of the actual stress which o tr originally existed in the second radial layer of the bar.
To eliminate end effects. extreme care being taken to prevent overheating. measurements are made cl and tangential strain ei. Metals Park. A." pp. 1927. the specimen length should be at least three times the diameter. J. About 5 per cent of crosssectional area should be removed between each measurement of strain. 153] Residual Stresses ^ 401 by Sachs. Dodd. Z. A = 6 = The longitudinal. Lynch. Metallk. axial hole. pp. 1952. J. This is not a particularly and tubes made by most forming oper ations have the required symmetrical residualstress pattern. pp. » ^ G.Sec.A)f^ +A 2A e (155) <Jr E' (^i'e) 19. "Residual Stress Measurements. Metallurgia. the longitudinal and tangential strains are combined in two parameters.. 5152. Sachs. technique. tangential. American Society for Metals. {Ao  A) clA A Ao at = = E' iA. Ohio. vol. ' A critical evaluation of experimental procedures has been presented by R. 352357. In using this technique with a solid bar the first step is to drill an Then a boring bar is used to remove layers from the inside diameter of the hollow cylinder. In accordance with the Sachs analysis. vol. cl e< + vet + veL by and radial stresses can then be expressed the following equations. Li eL of longitudinal — Lq Di et Do Do The changes ters. layer strain is After each removed from the bore. longitudinal This method The method is is commonly known as the Sachs boringout residual stresses vary in Hmited to cylindrical bodies in which the the radial direction but are constant in the restrictive condition since bars and circumferential directions. . 45. 109114. in length L and diameter D may be measured with micromebut better accuracy can be obtained by mounting SR4 strain gages^'^ in the longitudinal and circumferential directions of the bar. 1952.
7488.. pp. 22. see also Lynch. To estimate the stress late the along the axis of the bar or at the bore of a tube. The sheet specimen is then released from residualstress state 1 op. . it is necessary to extrapoA versus A and 9 versus A curves to ^ =0. 2 cit. 130134. J. the sheet specimen is cemented to a flat parallel surface. of the bar.^ TreutingRead Method Treuting and Read^ developed a method for determining the biaxial on the surface of a thin sheet. vol. The method assumes that the metal behaves in an elastically homogeneous manner and that the stress varies. these curves should be = ^o. Read. diameter of the axial hole while metal is The Sachs equations for this (TL = E' E' (A^. 1 5 = = E/(l  v"") Ao = A = V original area of cylinder area of boredout portion of cylinder Poisson's ratio In using the above equations it is convenient to plot the strain parameand A as functions of the boredout area A. and the thickness is reduced a certain amount by careful polishing and etching. Treuting and W. pp. Espey. To apply the method. 8792. 1951. not in the plane of the sheet. 1942. Sachs and G.There is a limit to how closely the boredout diameter can approach the outside diameter of the bar without buckling.. Trans. R. 147. Phys. pp. G. AIMS. but only through the thickness. Since the residual stress near the outer surface may be rapidly changing with radial distance an extrapolation in this region may be in error. vol.)^A (A CFt  de An) dA A + Ah e 2A (156) = where Ah E '(^'^) = A = area of boredout hole area of cylinder after each layer removal A method for accurately determining both the longitudinal and tangential (circumferential) residual stresses in thinwalled tubing has been described by Sachs and Espey. T.402 where E' Applications to Materials Testing [Chap. G. The slopes of these ters curves are then used in the above equations. One mine the stresses extrapolated to Jl way bar to get a better estimate of the stresses near the outside surface of the is to measure the changes in removed from the outer surface case are given below. Similarly. Appl. to deter on the outer surface of the bar.
>ec. Figure 155 illustrates the orientation of the principal stresses and the curvature of the sheet. The measured values of radius two parameters Px and Py. Measurements of Rx and Ry are made for different amounts metal removal. the transverse radius of curvature Ry and the thickness t. and measurements are made of the longitudinal radius of curvature Rx. and Px and Py are plotted against the sheet thickness The f / . 154] Residual Stresses 403 the surface. of curvature are expressed in terms of = ^ +^ Rx Rv Py = — +— Ry Rx of t. p.
(Jo) (a) Rolled sheet.. approximate methods which are more rapid but Since the techniques involve a less accurate have been developed. mechanical slitting of the of the slit element. Determination of longitudinal residual stress by deflection method. Actually. . Although this is usually the direction of the maximum stress. . 156a). The formulas that are presented below give the residual stress only at the surface of the body and consider the stress only in one particular direction. or circumference. specimen and measurement of the deflection they are usually called deflection methods.404 Applications to Materials Testing is [Chap. The bending ^ Lynch. op. Deflection it is methods can be applied when reasonable to assume that the stress is varies linearly through the thickness of a plate or tube but constant along the length. 1 56. drawn rod. at. bution through the thickness is rarely linear. specimen. its Rolled Sheet The longitudinal residual stress at the surface of a rolled sheet can be determined by splitting the sheet down its central plane to release the bending moment which existed in the body' (Fig. width. 8182. pp. 1 5 find that nearly 40 hr required for the complete analysis of a single Therefore. it should be realized that the presence of stresses in the other principal directions can affect value. the stress distri ^' 28 "Residual stress distribution (a) ^^' 2S +2r Fig.
J. Inst. 1924. 32. (1511) Thus. With the notation given in Fig. Metals. (^L = ^^ J  n^o^ (159) where c is the distance from the neutral axis to the outer fiber. pp. 3G7383. the maxilinearly over the thickness t. 1566). small compared with the radius of curvature. p. Substituting Eq.. G. .p') = moment of inertia of R = radius of curvature I ^ (158) split section Although the residualstress distribution which went to make up this bending moment is unknown. may be determined^ by tongue from the wall of the tube (Fig. K = g ^. M where E' = = E/(l . in this case t/4. it is assumed for the analysis that it varies Based on this linear relationship.65^^5r y^ (1513) Thinwall Tubing The 1 longitudinal stress at the surface of a tube splitting a longitudinal Ibid. . 1566 the procedure outlined above results in the following equation: (TL = 1. 82. 154] Residual Stresses 405 moment in this sheet may be expressed as follows. * R. (158) into (159) and introducing the value for c results in <TL = ^ (1511) if (1510) The radius of curvature may be expressed in terms of the deflection the deflection 8 is and the length of the curved beam L by Eq. the longitudinal residual stress at the surface aL is given by (1512) = Round Bar The same procedure can be used stress in a to estimate the surface longitudinal round bar> (Fig. Anderson and E.Sec. 157a). Fahlman. vol. J. mum longitudinal stress at the surface is given by the familiar equation from strength of materials.
157. 1 5 With the notation given the following equation. tudinal To determine the slit is made circumferential residual stresses in a tube.2 times the diameter of tion the tube. circumferential stress. the longitudinal stress is given by (TL = TIT (1514) where E' 1 E v' Equation (1514) is the same as Eq. in Fig. and the change in . is obtained when the tongue width is 0. Determination of residual stresses in thinwall tube by deflection method.1 to 0. except that t/2 is used in the latter case because only half the sheet thickness deflects. (1512) for a rolled sheet.406 Applications to Materials Testing [Chap. U) id) Fig. 157 a. a longi the entire length of the tube. (b) (a) Longitudinal stress. Experience with the slitting of tubes has shown that the observed deflecThe maximum deflection is a function of the width of the tongue.
circumferential residual stress at the surface With the notation is given by in the figure. 2491 2 254. Since the Xray method is based essentially on an accurate measurement of the shift in position of the Xray reflection from a given set of lattice planes due to the presence of elastic strain. it is not always necessary to make measurements on the specimen in the unstressed condition. Mechanical methods of measuring localized residual stresses around a small drilled hole have been developed.^ since the Xray beam covers an area approximately 3^ in. it is necessary that the dif Using film techniques to record the specimen yield sharp diffraction lines Since severely if the lattice strain is to be measured with precision. pp. However. 1576). The change in lattice spacing stress. See J. Crampton. Because X rays penetrate less than 0. K. Xray methods have the important advantage that they are nondestructive and involve no cutting or slitting of the object to measure Further. 1934. this strains and therefore only surface and biaxial states of stress. pp. Mathar. Xray Determination of Residual Stress The measurement of residual stress with rays utilizes the interatomic spacing of certain lattice planes as the gage length for measuring strain. this limited to uniaxial For is many applications.Sec. 89. X and for the same material containing residual can be related to residual in. is Since no stress exists normal to a involved. vol. This is often an advantage in making postmortem examinations of parts which failed in service. 233255. the residual stresses in these important classes of specimens cannot be accurately determined by Xray film techniques. fraction lines be accurately located. tive is to characterize the overall residualstress condition in a surface. the introduction of Geigercounter Xray spectrogoniometer equipment Xray reflections requires that the D. 1930. in diameter. the '''\k~wj 1 55. . The Xray method measures the residual stress in a very localized area. the interatomic spacing for a given lattice plane is determined for the stressfree condition stress. This makes the Xray technique useful for measurements of sharply changing stress gradients. AIME. free surface. Trans. 86.001 method is into the surface of a metal. ASME. coldworked material and quenched and tempered steel give broad diffraction lines. Trans. particularly where fatigue failure not a serious disadvantage. Xray methods measure only surface residual stresses can be determined. vol. In essence. 155] Residual Stresses is 407 diameter measured^ (Fig. but it may be a disadvantage where the objecthe stresses.
Z. S. and the measured diffraction angle n\ = 2d sin d (1516) The simplest condition to consider will give values for a uniaxial surface stress or the sum of the principal stresses cri + 02. and the wavelength of Xray radiation. The strain normal to the surface which is measured by the e. d. pp. New York. Physik. " C. "Structure of Metals. 51. Koistinen and R. vol. m. 537555. « G. L. 344358. Bragg's law expresses the relationship between the distance between a given set of lattice planes. McGrawHill Book Company. In the Sachs. 1953.= ^^^ (1519) more general equation for determining the principal stresses by X rays can be obtained by considering the general case for principal stress (Sec. 638676. the is sum of the principal stresses lying in the specimen surface given by a. pp. Weerts. The situation is discussed in detail by Barrett. the strain in the direction normal to the free surface can be expressed as 63 =  i^l + ^ (1518) Therefore.. Trans. 45. pp. 1930. ASM. while the other shot gives do for the stressfree surface. 14. Methods have been developed of measuring with considerable precision residual stresses in heattreated steel by means of X rays." 2d ed. Trans. and n as in a surface 1 A. X. One shot gives the value di for the stressed surface. + a. the order of the diffraction n. E. 64. The normal stress given by the coordinates rp and 4> can be expressed in terms of the three principal stresses and their direction cosines I.. D. 1952. ASM. vol. chap. S. 2). 25). A much .^ The generalized state of principal stress acting on the surface can be represented in three dimensions by an ellipsoid (Fig. vol. 1 has permitted a more accurate measurement of the Xray diffractionline profile than is possible with film techniques. The lattice constant in the stressfree condition is obtained either by removing a small plug of metal from the specimen or by thermally stressrelieving the sample. X rays do is 63. 158). Inc. Marburger. Rowland. = ^±^ (1517) From the theory of elasticity (Chap.5 408 Applications to Materials Testing [Chap. Barrett. Christenson and E.Weerts^ method two Xray determinations are made of the lattice spacing d for the Xray beam oriented normal to the specimen surface. P.^ 6. Sachs and J. 1959.
Representation of principal stresses ai. (1523) leads to the relation _ However. Fig. Since the Xray determination of residual stress considers only the stresses in the surface. 158. = rei \. + nVa (1520) where I = sin \p cos (f). atomic spacing in direction defined by angles \p and <f> . 158. (T = Pai { i/' mV2 </>.2 + n^e^ (1522) If €2 in the values of the direction cosines. (1520) can be written for the e principal strains.V (1524) sin'' sin^ — d^ es — do do di — do d^ — do di (1525) where do di = = = atomic spacing in unstressed condition atomic spacing in stressed metal perpendicular to specimen surface d^. (1520) can be simplified as follows to give the component of stress = 0. together with the values of d and get terms of principal stresses are substituted into Eq. in the direction 4>: 00 = o"! cos^ <^ + 02 sin^ 4> (1521) An equation exactly analogous to Eq. ai. Eq.m^e.Sec. (1522). 03 Referring to Fig. this requires that yp = 90°. 155] Residual Stresses 409 follows. (1521) into Eq. m = sin sin and n = cos ^. e we — 63 = —— ^^ sin x}/ (a I cos^ <^ \ a 2 sin </>) (1523) Substitution of Eq. az by ellipsoid of stress. e e — 63 xj/ E I {. Therefore.
surface. Generally Ap — 45° is used for inclined at an angle this second exposure to give the value of d^. If it is is zero because the measurements are made on a free necessary to determine both the magnitude and direction of a\ necessary to in three and C72. while a third exposure at i/' 1/' in the direction of the other principal stress 03 <^2 determines ai from Eq. (1525) is chiefly determined by the precision with which the numerator is known. orientation of the azimuth angle 4> is related to Xray measurements of atomic lattice spacing by the following equation: C4. determined in addition to a normal exposure to determine It is cuss/' tomary make these exposures at (^. Three arbitrary known <^ directions on the surface must be d\. It is known that for most metals these elastic pp. it is stress components to make four Xray exposures. The in stresses in the three arbitrary directions are calculated from Eq. some uncertainty as to the values of elastic modulus E which should be used in Eq. . 1 Since the accuracy of Eq. (1526) and are then converted into principal stresses by the methods given Chap. the stress in the surface of the specimen for any free samples.] 5 410 Applications to Materials Testing [Chap. In addition to the difficulties of accurately measuring the values of interatomic spacing in highly strained metals which have already been discussed. One per= 45° in the direcpendicular exposure determines d\. one exposure at = 45° tion <^i determines a\ from Eq. = d^ — ^ di E :f— 1 \. as is often the case for residual stresses produced by quenching or plastieforming </>. there is and Poisson's 1 ratio v values of residual Ihid. and another exposure is made for the same value of but with the Xray beam from the normal. (1526). (1526) to calculate stress. For any azimuth angle it is customary to make two exposures to determine the stress in the direction.V I di ^^r sm^ \l/ (1526) The number of Xray exposures that is required depends on the information that is available. If the directions of two principal stresses oi and a<i in the specimen surface are known. One exposure is made normal to the surface {\f/ = 0) to give di. it is permissible to substitute di for do in the denominator without greatly affecting the results. This substitution is an important simplification of the experimental procedure because it eliminates the necessity for making measurements on stressTherefore.. and </> — 60°. (1526). all that is necessary is to make three exposures to determine the complete biaxialstress state at the point on the surface. </> f 60°. with = 45° for each value of 0. 326327. 2 or by the equations given by Barrett^ for the specific conditions established above. \l/ operations.
The development of the stress pattern shown can be visualized as follows: The relatively cool surface of the bar tends to contract into a ring that is both shorter and smaller in diameter than the original diameter. Most investigators who have used Xray methods for residualstress determination have considered that the use of the average value of E and v introduce no serious errors beyond that due to For greatest accuracy. 152. 159a Figure 159c a metal which contracts on cooling. 1 56. will be considered first. Quenching Stresses In the introductory discussion of residual stresses in Sec. The simpler situation. E and v determined from the Experimental data on this point are somewhat contradictory. 156] Residual Stresses 411 Since the constants vary considerably with crystallographic direction. \ loads. these constants should be uncontrollable factors. It is also the situation encountered when steel is quenched from a tempering temperature below the A I critical temperature. shows that the opposite residualstress distribution is obtained if the metal expands on cooling. was shown in Fig. 151 it was of a metal which does not experience a phase change can produce residual stresses because of the strain mismatch produced by the differential contraction between the cooler and hotter parts The development of the longitudinal residualstress pattern of the body. tangential. for this case the residualstress pattern is due to thermal volume changes plus volume changes resulting from the transformation of austenite to martensite. However. where the stresses are due only to thermal volume changes. 159a . Quenching a body from a high temperature shown that cooling an ingot to a lower temperature accentuates the development of residual stresses because the greater temperature differential which is produced between the surface and the center due to the rapid rate of cooling produces a The situation of greatest practical interest greater mismatch of strain. (1526) are based on Xray measure ments it is of lattice strain in certain fixed directions in the crystal lattice. This is the situation encountered in the quenching of a metal which does not undergo a phase change on cooling. This tends to extrude the hotter.Sec. involves the residual stresses developed during the quenching of steel for hardening. residual stresses calculated from Eq. for the usual case of and radial directions is shown in Fig. determined for each material and experimental setup by making Xray measurements on stressrelieved specimens which are subjected to known questionable that average values of tension test should apply. The distribution of residual stress over the diameter of a quenched bar in the longitudinal. more plastic center into a cylinder that is longer and thinner than its in Fig.
The yieldstresstemperature curve strain for the metal is also important. then the mismatch will be small at high temperature because the metal can accommodate to thermally produced volume changes by plastic flow. The stress pattern given in Fig. the higher the possible residual stress. would change dimensions to a shorter and However. On the other hand. 1 59. since the residual stress cannot exceed the yield stress. tangential. it thinner cylinder on cooling. Residualstress patterns (a) (schematic). metal which expands on cooling. like superalloys. k^L Longitudinol Tangential I = longitudinal r= tangential /?. (6) orientation of directions. the higher the yield stress. metals which have a high yield strength at elevated temperatures. longitudinal.412 Applications to Materials Testing If [Chap. If the yield stress decreases rapidly with increasing temperature. Further. 159a results. will develop large residual stresses from quenching. . the higher the residual stress. (c) for found quenched bars. the higher the modulus elasticity of the metal. due to thermal strains For metal which contracts on coohng. For a given strain mismatch. the inner core were free to change shape inde pendently of the outer ring.radial Radia Id) Fig. 15 original dimensions. continuity must be maintained is throughout the bar so that the outer ring drawn in (compressed) in the and radial directions at the same time as the inner core is extended in the same directions. in The magnitude of the residual stresses produced by quenching depends on the stressstrain relationships for the metal and the degree of strain mismatch produced by the quenching operation.
Sec. as determined primarily by the bar diameter. and the heattransfer characteristics of the system. midradius. 1 56] Residual Stresses 413 The following combination match strains on quenching: 1. In the quenching of steels. 15. The surface layers tried to contract against the expanding central core.10a the quenching rate was rapid enough to 0. A A A A low thermal conductivity k high specific heat c high coefficient of thermal expansion. m. 15106). if the bar diameter is rather small and it has been drastically quenched in brine so that the surface and center transform at about the same time. The cooling rates of the outside. the metal expands as the martensite reaction proceeds on cooling from the Ms to Mf temperature. 3. formation characteristics of the steel. The residualstress distribution in a quenched steel bar is the resultant of the competing processes of thermal contraction and volume expansion due to martensite formation. 4. Other factors which produce an increase in the temperature difference between the surface and center of the bar promote high quenching stresses. and the severity of the quench. the transformation had been essentially completed at the surface. convert the entire bar to martensite. However. By the time the center of the bar reached the Ms temperature. Low Figure 1510 illustrates some of the possible residualstress patterns which can be produced by quenching steel bars. Transformation of austenite to bainite or pearlite also produces a volume expansion. the austenitizing temperature. These factors are (1) a large diameter of the cylinder. a high density p These factors can be combined into the thermal diffusivity. of physical properties will lead to high mis 2. This produces a residual stress distribution of the type shown in Fig. as determined chiefly by its composition and hardenability. and the result is tensile residual stresses at the surface and compressive stresses at the center of the bar (Fig. the surface will arrive at room temperature with compressive . and c. and center of the bar are indicated on this diagram by the curves marked In Fig. (2) a large temperature difference between the initial temperature and the temperature of the quenching bath. Since an increase in volume accompanies this transformation. austenite begins to transform to martensite whenever the local temperature of the bar reaches the Ms temperature. but of The resulting stress pattern depends upon the translesser magnitude. 159c. values of thermal diffusivity lead to high strain mismatch. and (3) a high severity of quench. Dt = k/pc. The left side of this figure illustrates a typical isothermal transformation diagram for the decomposition of austenite.
the bar is the time forming on the surface. and the core readily accommodates to the expansion of the outer layers. there is little restraint offered by the hot.414 Applications to Materials Testing If [Chap. 15 slackquenched so that the outside transforms to martensite while the middle and center transform to pearlite (Fig. soft core during residual stresses. 15lOc). The middle and when martensite is Austenite Martensite log time (a) Austenite +cr L/^ Martensite log time ic) .
procedure is sometimes used. these processes produce high surface compressive residual Figure 1511 is typical stresses which aid in preventing fatigue failure. 57. the residual tensile stress may exceed the fracture stress. steels. vol. If stress raisers are present. The reasons for tempering hardened steel are to relieve the high microstresses in the martensite and to reduce the level of macro residual stress. In order to minimize quench cracking in brittle steels. . an interrupted quench. or martempering. 157] Residual Stresses 415 stress pattern and compressive with tensile stresses at the surface and center of the bar stresses at midradius. of the kind of residualstress distribution which is considered in this section. This will cause the surface tensile stress to increase as tempering progresses. there are exceptions to this behavior. Surface tensile stresses equal to the yield stress may be produced by quenching. Boegehold. Normally.^ For steel bars which have been hardened through. Although tempering generally lowers the level of residual stress. If a bar is almost completely through. contraction hardened greater due to tempering of the martensite will occur at the surface than at the center. In relatively brittle materials like tool the yield stress is not far below the fracture stress. The steel is kept at the temperature long enough for it to come to thermal equilibrium. Metal Progr. L.. and then it is quenched to a lower temperature to form martensite. the level of residual stress will decrease uniformly with increas For a bar with surface compressive stresses due to incomplete hardening an increase in tempering temperature will progressively reduce the residual stress. The process of induction hardening consists in inductively heating a this region to martensite with a spray of water. The steel is quenched from the austenitizing temperature into a holding bath which is maintained at a temperature above the Ms temperature of the steel. thin surface layer of a steel part above the transformation temperature and then flashquenching The 1 localized expansion of the martensitic case produces a residualstress pattern of compression stresses at the surface and tension in the interior A. Cracks produced by tensile quenching stresses are called quench cracks. Surface Residual Stresses There are a number of important technological processes which produce high surface residual stresses which have their maximum tvalue either at the surface or just below the surface and which fall off rapidly with distance in from the surface. ing tempering temperature. 183188. 157. pp.)ec. 1950. The steep stress gradient produced by these processes is in contrast with the more uniform residualstress gradient produced by quenching and most mechanical forming operations.
010 in. The impact of the shot on the leaves surface causes plastic stretching of the surface fibers at a multitude of localized regions. the maxi mum compressive residual stress occurs at a depth in from the surface which is just short of the boundary between the carburized case and the lowercarboncontent core. 1956. The maximum residual stress occurs below the surface at a depth of about 0. Typical residualstress distribution produced by surface compressive residual stresses. the carburized surface contains 1511. 1 by changes Ordinarily. the resulting compressive residual stress on the surface which was prestressed in tension will be greater than if the steel was shotpeened in an initially stressfree condition. 1 5 Flame hardening produces the same metallurgical changes and residualstress pattern as induction hardening. is Because there is is no volume expansion occurring in the core. by nitriding Shot peening consists in subjecting a metal surface to the impact of a stream of fine metallic particles (shot). Intern. Conf. Nitriding consists in the diffusion of nitrogen into 4 1 a steel surface at a temperature below the transformation temperature. pp. Mattson. Carburizing consists in changing the carbon content of the surface layers a steel part by the diffusion of carbon atoms into the surface. Both processes not only produce a favorable residualstress pattern but also leave a hardened surface which improves the resistance to wear. Fatigue of Metals. of heattreated steel the In the shot peening maximum compressive residual stress which is about 60 per cent of the yield strength of the steel. When this localized plastic flow is relieved. but considerable treatment like induction hardening. The residualstress pattern consists of compression at the surface and tension at the interior. The process differs from induction hardening in that the entire part is heated into the austenite region during the of Surface Fig. or shot peening.416 Applications to Materials Testing [Chap. Higher values of compressive residual stresses can be obtained by strain peening. but the local heating is produced by a gas flame. as distribution produced possible for carburizing. Proc. it the surface in a state of compressive residual stress.002 to 0. Surface and when the part is quenched volume changes may occur through most of the cross section. London. L. The only volume expansion arises from the formation of nitrides at the surface. The variables of the shotpeening produced is 1 R. 593603. distribution can be produced in processing. variation in residualstress carburizing. For example. the residualstress easier to control. if a flat leaf spring is preloaded in bending. In diffusion process general. .
6 ^ O. pp. Lessells. ASTM. 1 58.. Generally. most of the residual stress will be relieved by timedependent stress relaxation. Mech. (2) the shot hardness. L. 1944. (3) the shot velocity. while shot peening lends itself to highproduction volume parts where broad or irregular surfaces Numerous instances have been reported in the literare to be treated. Horger and T. or reduction in the intensity of residual stress is known as Stress relief may be accomplished either by heating or by mechanical working operations. Proc. Surface working operations such as shot peening and rolling have also been successfully used to mitigate the effects of stress corrosion cracking. F. R. Mattson and W. Fatigue of Metals. 158] Residual Stresses 417 process''^ which mfluence the magnitude and distribution of the residual stresses are (1) the shot size. 62. since the residual stress cannot exceed the yield stress. Ohio. Proc. Conf. The stress relief which comes from a stressrelief anneal is due to two effects. Park. H. Although residual stresses will slowly disappear at room temperature. Horger and H. no. 1935. 1954. First. Intern. New York. pp. ature where the introduction of beneficial compressive residual stresses by flame hardening. * O.^ nitriding. Lessells and R. . 1957. vol.'' carburizing. vol. "Strength and Resistance of Metals." pp. 2. J.. V. Only the residual stress in excess of the yield stress at the stressrelief temperature can be eliminated by immediate plastic flow. Buckwalter. 151.^ Surface rolling areas like fillets is well adapted to the beneficial protection of critical and screw threads of large parts. Neifert. 1315 are useful for estimating stressrelief treatments. vol. vol. London.^ and surface rolling^ has resulted in improved fatigue performance. Appl. 1956. S. 546556. pp. 178189. John Wiley & Sons. A128136. Proc. Horger. 682695. Trans. vol. Relaxation curves such as those of Fig. 1. the process is very greatly accelerated by heating to an elevated temperature. 256259." American Society for Metals. 617627.Sec. Metals 1 Ibid. 1954. 8992. Inc. 8 O. ^ J. and (4) the exposure time (coverage). SAE. Stress Relief The removal stress relief. M.^ shot peening. plastic flow will reduce the residual stress to the value of the yield stress at the stressrelief temperature. 1936. Coleman. Stress Analysis. Since this process is extremely temperR. pp. M. Metal Treatment. Soc. pp. Sutton. High surface compressive residual stress can also be developed by surAn estimate of the loads required face rolling. J. J. Brodrick. * "Fatigue Durability of Carburized Steel. pp. /.. Exptl. 41. 2. for rolling to a given depth may be made by the use of Hertz's theory of contact stresses. as was shown in Fig. 1941. 3 J. 2.
no. Horger. Ohio.5 418 Applications to Materials Testing [Chap. 49. 68." McGrawHill Book Company. Barker. pp.^ Often a compromise of a temperature high enough for the a reasonable length of time and the annealing of the effects of cold work. ASTM. Metal Progr.. For example.: "Evaluation of Residual Stress. John Wiley & Sons. 15. products such as sheet. 8996. Publ. BIBLIOGRAPHY Baldwin. Metals Park. p. K. 1. Willey. 1949. 1955.. 1959. it is important in using this method of stress relief stress distribution may have the to select surface working conditions which will completely cancel the For example. 2 H. and L. Aug. An interesting concept is the use of thermal stresses to reduce quench^ ing stresses. . 459469." American Society for Metals. the time for nearly complete elimination of stress can be greatly reduced by increasing the temperature. only the tensile stresses at the surface would be reduced.: Residual Stresses. C: Bibliography on Residual Stress. vol. Rept. : W. Proc. "Residual Stress Measurements. Inc. Huang. 95. R. Residual Stresses. SP125. 1960." pp. plate. it is conceivable that. Ser. Dangerously high tensile stresses could still exist below the surface. * Internal Stresses in Metals and Alloys.. Since residual stresses result from thermal gradients produced when a part is being quenched. no. Symposium on 1948. and extrusions are often stretched several per cent beyond the yield stress to relieve differential strains by yielding. SP167. Metals Mon. eliminated particular working operation In other cases the residualstress distribution which is characteristic of a may be superimposed on the residualstress pattern initially present in the material. 1948. suppl. However. Inc. vol. S. Twentythird Edgar Marburg Lecture. This concept of an "uphill quench" has been used^ in aluminum alloys to reduce quenching stresses by as much as 80 per cent at temperatures low enough to prevent softening. 1955. Aug. 15. 1952. February. 539583. pp. very light surface rolling were used on a surface which initially contained tensile stresses. Inst. it is possible to introduce residual stresses of opposite sign by subjecting a cold piece to very rapid heating. residual stresses A surface which contains tensile converted into beneficial compressive stresses with a surface working process like rolling or shot peening. ASM. New York. 657674. if only initial stress distribution. O. New York. A. Hill. 1950. vol. Typical stressrelief temperatures and times for many metals will be found in Metal Prog. Residual Stresses in Metals. 1954. vol. SAE Spec. J. The differential strains that produce high residual stresses also can be relief of stress in must be made between the use by plastic deformation at room temperature. 5. "Handbook of Experimental Stress Analysis. M. 2A. 2A. N. Trans. 68.. 52. 1 aturedependent. Heindlhofer. pp. T.
A second reason for considering statistics in conjunction with mechanical metallurgy is that statistical methods can assist in designing experiments to give the maximum amount of information with minimum of experimental investigation. In order to cover the greatest amount of ground. and therefore it is a logical place in which to introduce the elements of statistical analysis of data to metallurgists who may have had no other training in the subject. how Instead. emphasis has been placed on showing can be put to work. The statistical methods which apply in mechanical metallurgy are in general no different from the techniques which are used for analysis of data in other areas of engineering and science. no attempt has been the made to include the mathematical niceties which are usually a part of a course in statistics. or scatter. statistical techniques are useful. Numerous references are included to sources of more complete discussions and to techniques and applications statistics 419 . Finally. and often even structuresensitive properties. for determining the precision of the measurements and for drawing valid conclusions from the data. being Therefore. Mechanical metallurgy is one of the few areas of metallurgy where large numbers of data are likely to be encountered. Why Statistics? is There are at least three reasons why a working knowledge of statistics needed in mechanical metallurgy.Chapter 16 STATISTICS APPLIED TO MATERIALS TESTING 161. statistical methods based on probability theory have been developed to explain certain problems in mechanical metallurgy. The explanation of the size effect in brittle fracture and fatigue is a good example of the use of statistical theory in mechanical metallurgy. mechanical properties. It is recognized that the material which can be included in a single chapter can hardly consist of more than an introduction to this subject. necessary. First. frequently exhibit considerable variability.
16 which could not be included because of space limitations. By be introduced if ing the diameter. Systematic errors arise because of faulty control of the experiment. errors are due to limitations of the in the material being tested. systematic errors (which exert a nonrandom bias). is the collection of all possible tensile specimens which could be cut from this forging or from all other forgings which are exactly identical. and experimental. however. The population. is cut from a steel forging and the reduction determined for this specimen. However. it is obviously impractical to sample and test the entire forging. Experimental. a systematic error could an improperly zeroed micrometer were used for measurrandom errors would result from slight differences in fitting together the two halves of the tensile specimen and from the inherent variability of reductionofarea measurements on metals. the observation represents a sample of the "population from which it was drawn. It is for this When a tensile specimen of area is . or random. errors. in this case.420 Applications to Materials Testing [Chap. will not cancel upon averaging. the sample estimate of the population values becomes better and better. while averaging a number of observations the random error will tend to cancel out. It is hoped that this chapter contributes to a better appreciation of the usefulness of statistics in metallurgical research by providing the background necessary to see a problem from a statistical viewpoint and enough of the standard tools of statistics to allow a valid analysis in straightforward It is important to emphasize that many practical problems are just too complicated to treat by means of the simpler statistical techniques. one of the main purposes of statistical techniques is to determine the best estimate of the population parameters from a randomly selected sample. 162. Therefore. In these cases it is important to acquire the services of a trained statistician for the planning of the experiment and the analysis applications. and therefore the sample provides only an estimate of the population parameters. One of the major objectives of statistical analysis is to deal quantitatively with random error. in the measurement of the reduction of area of a fractured tensile specimen. of the data. As more and more tensile specimens are cut from the forging and reductionofarea values measured. the corresponding parameters calculated from samples contain random errors. or random. However. The systematic error. Errors and Samples The act of making any type of experimental observation involves two types of errors. measuring equipment or to inherent variability As an example. The approach which is taken is to postulate that for each sample the population has fixed and invariant parameters.
Sec. 163]
Statistics
Applied
to Materials Testing
421
reason that statistical methods lead to conclusions having a given probahiliiij of being correct.
163. Frequency Distribution
When
a
is
a large
is
number
of observations are
method
needed to characterize the data.
made from a random sample, The most common method
of equalvalued class intervals
to arrange the observations into a
number
and determine the frequency
interval.
each class In Table 161, out of a total sample of 449 measurements of
of the observations falling within
Table 161 Frequency Tabulation of Yield Strength of Steelj
Yield
422
Applications to Materials Testing
[Chap.
16
70
Sec. 163]
Statistics
Applied
to Materials Testing
is
423
as a
measurements (Fig. 161). histogram. As the number
This type of bar diagram
known
of observations increases, the size of the class
interval can be reduced until
we obtain
the limiting curve which repre(Fig. 162).
sents the frequency distribution of the
sample
of the values of yield strength fall within the interval
Note that most from 126,000 to
is
134,000 psi. If the frequency of observations in each class interval
a percentage of the total
(Fig. 163).
expressed as
number
is
of observations, the area
under a
fre
quencydistribution curve which
plotted on this basis
is
equal to unity
strength will be between the value
is
The probabihty that a single random measurement of yield Xi and a slightly higher value Xi + AX given by the area under the frequencydistribution curve bounded by
two
limits.
these
Similarly, the probability that a single observation
20
424
Applications to Materials Testing
100
fChap.
16
114
118
122
126
130
134
138
142
146
Yield strength, 1,000 psi
Fig.
164. Cumulative frequency distribution.
distribution
it is
is
much
less sensitive
sometimes preferred to a frequency distribution because than the frequency distribution to the choice of
class intervals.
164. Measures of Central Tendency and Dispersion
A frequency distribution such as that of Fig. 162 can be described with numbers which indicate the central location of the distribution and
how
of
The most common and important measure
data
is
the observations are spread out from the central region (dispersion). of the central value of an array the arithmetic mean, or average.
is
The mean
of Xi, X2,
•
•
•
,
Xn
observations
denoted by
X and
is
given by
Ix.
X
(161)
The arithmetic mean is equal to the summation of the individual obserIf the data are vations divided by the total number of observations, n. arranged in a frequency table, as in Table 161, the mean can be most
Sec. 164]
Statistics
Applied
to Materials Testing
425
conveniently determined from
k
X
=
^^^
(162)
where /» is the frequency of observations in a particular class interval with midpoint X,. The summation is taken over all class intervals (see Table 161, column 4). Equation (162) is an approximation to Eq.
(161).
Two
other
common measures
of central
tendency are the mode and the
median. The mode is the value of the observations which occurs most The median is the middle value of a group of observations. frequently.
For a
set of discrete
data the median can be obtained by arranging the
falls in
observations in numerical sequence and determining which value
the middle.
For a frequency distribution the median is the value which The mean and divides the area under the curve into two equal parts. the median are frequently close together; but if there are extreme values in the observations (either high or low), the mean will be more influenced by these values than the median. As an extreme example, the situation sometimes arises in fatigue testing that out of a group of specimens tested at a certain stress a few do not fail in the time allotted for the test and These extreme values therefore they presumably have infinite lives. could not be grouped with the failed specimens to calculate a mean
fatigue
life;
j^et
they could be considered
in
determining the median.
The positions of the mean, median, and mode are indicated in Fig. 162. The most important measure of the dispersion of a sample is given by
the variance
s^.
s'
= '^
w
I
(Xi
 xy
(163)
1
—
The term Xi — arithmetic mean
denominator
is
X is the X of the
deviation of each observation Xi from the n observations. The quantity n — 1 in the
called the
number
of degrees of
freedom and
is
equal to
the number of observations minus the number of linear relations between Since the mean represents one such relation, the numthe observations. ber of degrees of freedom for the variance about the mean is n — 1.
For computational purposes it is often convenient to calculate the ance from the following relation:
vari
n
n{n
—
1)
426
Applications to Materials Testing
[Chap.
16
When
the data are arranged in a frequency table, the variance can be
most readily computed from the following equation:
k \ 2
2
s'
fiXi
d/'^')
n
=
—
(165)
I
it is
In dealing with the dispersion of data
the standard deviation
of the variance.
s,
usual practice to work with
which
is
defined as the positive square root
s
=
I
(X,
 xy
(166)
1
Sometimes
average.
it
is
desirable to describe the variability relative to the
The
coefficient of variation v is
used for this purpose.
(167)
A measure
simplicity
largest
is
of dispersion
which
is
sometimes used because
of its
extreme
the range.
and smallest
The range is simply the difference between the observation. The range does not provide as precise
estimates as does the standard deviation.
165. The Normal Distribution
Many physical measurements follow the symmetrical, bellshaped curve of the normal, or Gaussian, frequency distribution. Repeated measurements of the length or diameter of a bar would closely approxifrequency distribution. The distributions of yield strength, and reduction of area from the tension test have been found to follow the normal curve to a suitable degree of approximation. The equation of the normal curve is
mate
this
tensile strength,
1
viX)
where p(X)
is
aVSTT
mean
^ exp
\{^)
and
<y
(168)
the height of the frequency curve corresponding to an
fx
assigned value X,
is
the
of the population,
is
the standard
deviation of the population.^
from
The normal frequency distribution described by Eq. X = — octoX= +oo and is symmetrical about
It
(168) extends
fx.
The constant
should be noted that
o
is
used for normal stress in the other chapters of this book.
Sec. 165]
l/o
Statistics
Applied
to Materials Testing
427
Figure
1.
\/27r
is
used to make the area under the curve equal unity.
165 shows the standardized normal curve
when
/x
=
and
o
=
The
parameter
z,
defined
by Eq.
(169),
is
called the standard normal deviate.
(169)
z
= ^^^^
normal curve becomes
For
this case, the equation of the
(1610)
The
total area
under the curve
in Fig. 165 is
equal to unity.
The
rela
tive frequency of a value of z falling
between
—
oo
and a
specified value
428
Applications to Materials Testing
[Chap.
16
Table 162 Areas under Standardized Normal Frequency Curve
z
=
X —
3.0
n/a
Area
165]
pies are taken
Statistics
Applied
to Materials Testins
429
from the population, they will each in general yield a mean A" and standard deviation s. The sample values of the mean will, however, be normally distributed about the population mean fx. The centrallimit theorem of statistics provides a method for estimating the population mean n and standard deviation a. According to this theorem, if A" is the mean value of a random sample of n observations A is normally distributed about the population mean with a standdifferent value of
ard deviation (r/\/n. With this as a basis, it is possible to determine the confidence limits within w^hich the sample mean approaches the population
mean
to a certain level of certainty.
99.99
99.9
99
95
.
I I ^
I
=
90 80 70 60 50 40 30 20
10
5 2
1
1 o
t
I
I
0.5
0.2
0.1
0.05
0.01
430
Applications to Materials Testing
fx
[Chap.
16
are normally distributed about
with a standard deviation (x/\/n, we can mean of the population will be within the interval ± 1.96a/\/n. Since the sample estimate of the variance s cannot be substituted for the population variance o^ when the sample size is small, it is necessary to base the calculation^ on the samconclude that 95 per cent of the time the
X
pling distribution of the statistic
calculation, this
known
as
t.
From
the standpoint of
size.
lies
means only that the constant changes with sample
level the true
For a given confidence
the interval
mean
t
of the
population
within
X
±
t
ts/\/n, where values of
are obtained from Table 163.
Table
163
Values of
for Determining the Confidence Limits of
X
and
s
Confidence
Sec. 166]
Statistics
Applied
to Materials Testins
431
central
distribution' without destroying its statistical usefulness.
limit
The
theorem shows that a normal frecjuency distribution is to be expected when the effect being observed results from averaging the effects of a whole series of variables. However, if the effect being observed is due to the smallest or largest of a number of variables, an extremevalue distriIf failure is due to one weakest link or to a bution is to be expected.
number
of more or would apply.
less
equally
weak
links, the
extremevalue distribution
The application of the extremevalue distribution to engineering problems has been largely the result of the work of E. J. Gumbel.^ This theory has been used to predict the maximum wind velocity or the crest
of floods (largest values) of failure
(smallest
value).
and the fatigue life for a 0.1 per cent probability While the mathematical details of the
extremevalue theory are outside the scope of this chapter, an idea of the concept can be obtained from an example. Consider a group of fatigue specimens tested to failure at a constant stress. The distribution
extremevalue distribution
to the
sample can be considered to be an assumed that each specimen which failed represents the weakest out of a large number of specimens tested
of fatigue life obtained
this
if
from
it
is
same particular number
of cycles of stress reversal.
Methods
of
using the extremevalue distribution with fatigue data have been developed,^ and extremevalue probability paper
results.
is
available for plotting the
A
special type of extremevalue distribution
bution.^
For
is
this
is the Weibull distrifrequency distribution, the cumulativedistribution
function
given by
P(X) =
where P(X)
is
1

exp
 ^^^^
=
(^)
X will
m
and Xo and
(1612)
the probability that a value less than
be obtained,
Xn
is
the value of
X for which P(X)
0,
are empirically
determined constants. This equation gives a cumulativedistribution curve which resembles that for a normal distribution except that Weibull's distribution has a finite lower limit X„. Weibull's distribution has been used to describe the yieldstrength distribution of steel and the fatigue
For a discussion of the truncated normal distribution, see Hald, op. cit., pp. 144Am. Statist. Assoc, vol. 44, pp. 518525, 1949. 2 E. J. Gumbel, Statistical Theory of Extreme Values and Some Practical Applications, Natl. Bur. Standards Appl. Math. Ser. 33, Feb. 12, 1954; Probability Tables for the Analysis of Extremevalue Data, Natl. Bur. Standards Appl. Math. Ser. 22, July 6, 1953; see also J. Lieblein, NACA Tech. Note 3053, January, 1954. 3 A. M. Freudenthal and E. J. Gumbel, Proc. Intern. Conf. Fatigue of Metals, Lon1
151; A. C. Cohen, J.
don, 1956, pp. 262271. ' W. Weibull, J. Appl. Meek., vol.
18, pp.
293297, 1951, vol.
19, pp.
109113, 1952.
432
life
Applications to Materials Testing
at constant stress.
It
[Chap. 16
appears to be particularly well suited to
life
describing the distribution of fatigue
of bearings.
is
^
When
a mass of mechanical test data
is
analyzed to select reliable
design values, the design engineer
usually interested in establishing
the smallest values of the property being tested, so that the design can be based on conservative yet realistic strength values. For example, the designer might be interested in knowing the value of yield strength which would be exceeded by 99.9 per cent of the specimens, while the metal
who is trying to improve the yield strength by heat treatment would be mainly interested in knowing the mean and standard deviation of his test values. Of course, the value of yield strength which would be exceeded by 99.9 per cent of the specimens can be calculated from the sample observations provided that the frequency distribution of the observations is known and the observations faithfully follow this distribution. Unfortunately, although most mechanical property data fit a normal or
lurgist
lognormal frequency distribution over the central region of probability,
at the tails or extremes of the distribution appreciable deviations usually
occur.
A mean
of observations
nature of
and standard deviation can be determined from any set by the methods described previously, regardless of the the frequency distribution, and even if the observations deviate
still
considerably from a normal distribution, the powerful statistical tests
described in the next section can
be used with relatively
little error.
On
the other hand, estimates of the probability of failure are not valid
outside the regions in which the data are
distribution.
e.g.,
known
to
fit
the frequency
The
prediction of values at low probabilities of failure,
the 0.1 per cent level, on the basis of the assumption that the data
follow the normal distribution in this region can lead to considerable
error.
Further, because of the large
number
of tests required to
1
make
an experimental determination
of the stress
which gives a
or 0.1 per
cent probability of failure, there are few experimental observations in
the extreme regions which can be used for guidance.
While extreme
value statistics have not yet met with universal acceptance, this approach
is
the most reasonable in deahng with small probabilities of failure.
167. Tests of Significance
important feature of statistical analysis is the ability, through tests make comparisons and draw conclusions with a greater assurance of being correct than if the conclusions had been based only on
of significance, to
An
intuitive interpretation of the data.
of significance are
^
Most
of the
common
statistical tests
it
based on the use of the normal distribution, so that
J.
Lieblein
and M.
Zelen, J. Research Natl. Bur. Standards, vol. 57, pp. 273316,
1956.
a distinct advantage bution.12 at the 5 per cent level and 2.94 at the 1 per cent level of significance. statistics are significantly different it is always necessary to specify the probability of the test predicting an erroneous conclusion. where Si^ > §2^: Let Si F = 'A The F ratio obtained (1613) from is Eci. 1943. The table is entered across the top for the degrees of freedom for the numerator in the F ratio and along the lefthand edge for the degrees of freedom for the denominator. pp. A common made at the 5 per cent level of significance there is 1 chance in 20 that an erroneous conclusion would be reached by the Difference between test. 73. be the standard deviation for a sample size Wi from the first population. as the level of significance is raised (less risk of making an erroneous decision). This is known For example. For example.. Dixon and Massey. Note that. Thompson. the F tables give 2. and let S2 and n^ be the same terms for the second population. if the test of significance is as the level of significance. (161. p. we get the value of the F ratio. (1613) is greater than the F ratio obtained from the table. 167] is Statistics Applied to Materials Testing 433 if the test data conform to this frequency distriproblem which is readily treated by means of tests of significance is to determine whether the means and/or standard deviations of two samples drawn from essentially the same population really differ significantly or whether the observed differences are due to random chance. it may be important to know whether or not a certain metallurgical condition shows more scatter in fatigue life than another treatment or whether or not the scatter is significantly different at high stresses and at low stresses. the two sets of standard deviations are significantly different at the chosen level of significance. 310313.' is The is table corresponding to the selected level of signifi based on the number of degrees of freedom. see also M. the greater must be the F ratio determined from Eq. The test of significance which is used for It is often of interest to of this situation is the F test. . 33. op. Where this column and row intersect. For examcance used. for ni = 1 no = 21. vol. (1613).Sec. cit. If the ratio calculated in Eq. than the sample size. M. Two Sampie Standard Deviations know whether or not the variability of one set measurements differs significantly from that of another set of data.3) is compared with tables of the F distribution. Biometrika. This table which for this case one less ple. The following ratio is determined. Merrington and C. In statistics the term significant difference is used much more In determining whether or not two precisely than in everyday usage.
If the value of t determined from Eq. Only a brief description of some of the nonlations are a 1 Ibid. there is a significant difference between the means. If the F test shows that the standard deviations of the two samples are significantly different. it ^ = = = / 2/ ^ ^ —27~ul (1 (1616) c siV^i d. However. they are less powerful than the methods which have been described above. 307. Techniques which do not depend on the mathematical form of the frequency distribution are available.f. (1617) to use in determining the number of degrees of freedom (d. These nonparametric methods may be useful in cases where the data are too meager to allow an estimate of the frequency distribution to be made. 16 Difference between Two Means Sample Standard Deviations Not Significantly Different. and a value of c is calculated from Eq..n^ — 2. 2 ^ (ni — l)gi^ h (n2 2 l)s2^ ^1 + W2 — (1614) Then a value for the t statistic can be determined. is not correct to determine a common variance. (1616). (1615) t is greater than the value obtained from tables of the distribution^ at the specified level of signifi cance and with the degrees of freedom equal to ni \. Sample Standard Deviations Significantly Different. . the variance of the two samples in common should : be determined.) which is used with the t tables. The value of t is determined from Eq. If the F test shows that the standard deviations of the two samples are not significantly different. because they do not depend on any sampling distribution. the following procedure can be used to determine whether or not there is a significant difference between the means of the two samples Xi and X2 First. p. ./^Y''\/ + S2^/n2 (1617) ni — ^ 1 n2 .cr — 1 (1618) N onparametric The F test Tests of Significance and the t test depend on the requirement that the popugood approximation to the normal distribution.f.434 Applications to Materials Testing [Chap.
Based on a procedure given by C. "Some Rapid Approximate Statistical Procedures.Sec. pp. table 25. we know population. 360. R. while if n = 50. 69. . if n = 15. The number of "runs" from each sample are determined. if two samples A and B are arranged in order to give the series ABA ABB A A ABB. and by comparing with tables. there are six runs ABAABBAAABB in this series. 168] Statistics Applied to Materials Testins 435 should be parametric techniques can be given here." American Cyanamid 2 3 * York. Metal February. made to the original sources. Co. For example. For details. our degree of certainty decreases. Table 164 gives the values of RAT for a hypothetical situation where tensile tests were determined from forgings with 10:1 and 30:1 forging ' F. by using the methods of the previous section. 1949. op. 251255. New Ibid.arranges all values from both samples in ascending order. Wilcoxon. vol. Dixon and Massey. cit. To illustrate the meaning of the term run. The run test".. reference The rank test^ arranges the two samples in numerical order and assigns rank numbers. 8186. p. predicting at various levels of confidence Tables are available^ for and for different sample sizes what values in the array the population median would be expected to lie between. However. Progr. median of a population can be obtained very readily without any consideration of the distribution function of the If the observations are listed in ascending order.. to determine the interaction of variance. As we approach the central value of the sample. interval for the The confidence with great certainty that the median of the population smallest lies between the and largest observation. 1956. the fourth and twelfth observations are confidence limits of more than 95 per cent for the population median. the population median will lie between the eighteenth and thirtythird observation. it can be found whether or not the observations of both samples have been drawn from the same population.. Analysis of Variance The statistical tests of significance discussed in the previous section deal only with the analysis of a single factor at two different levels. one could determine whether the transverse reduction of area is significantly different for different forging reductions or for different heat treatments. The totals of the rank for each sample are compared with tables to determine whether or not a significant difference exists. One method* of analysis of variance will between the two requires the analysis be described below by means of a simplified illustrative example. Smith. pp.. 168. For example.
Is there a significant interaction between forging reduction and of homogenization which affects the level RAT? Table 164 Effect of Forging Reduction and Homogenization Treatment on Transverse Reduction of Area (RATj Homogenization . We 1 interested in answering the following questions. with and without a homogenization heat treatment. Is there a significant difference in RAT for a 10: 1 and 30: forging reduction? 2. Is there a significant difference in RAT between a forging which was given no homogenization treatment and one given the homogenization treatment? 3.436 Applications to Materials Testing [Chap. 1. 16 are reductions.
8)^ + (153.4)^ „_ Betweenrows sum of squares 7^2 (BRSS) (1623) BRSS = ^^'^^ Interaction ^ y /T12 km  fyklm (267.208.9)^ BCSS = ^^^^ 5. (1620) ggg ^ (140.65 = (1624) 35.27 of squares 712 Betweencolumns sum (BCSS) (1622) (561. 168] Statistics Applied to Materials Testing 437 The following computations 1. WSS = TSS . are made from Table 164: Total sum of squares (TSS) where ^X^j is the summation of the square of all readings.5)^ + (280.06 .3)=^ (561. an Values of mean square are obtained by dividing the values of sum of the squares by their respective degrees of freedom. sum ISS ISS = SSS . ^ y 7^2 Im  ifklm (281.5)^ _ (561.^l\ = 422.variance table such as Table 165.7^ klm. 20 ^^ ^^^ " Withincells sum of squares (WSS) (1621) WSS = 4.09 Subtotal sum of squares (SSS) 4^ m .67 .Sec. To determine whether or not there are signifiresults can be entered in Once the computations are made.9)'^ ^ 5 3.BCSS .11 analysisof.6)^^ = + (294.0. the .J^l.09 .9) + (140.7)^ + (126.BRSS = 70.70.35.SSS 422.82 .82 = 351. or TSS = 2. 16.9)^ 2IO of squares (ISS) _ 20^ " 6.
there is no significant interaction between these two variables. Therefore.: 438 Applications to Materials Testing [Chap. 16 cant differences between the columns and rows and interactions between mean squares are used in the F test.61 21. the following F ratio is reduction forging determined the two.11 = 1.49 for the variability to be significant. or not there is a significant interaction between whether to determine and homogenization treatment. Table 165 Analysis of Variance Source of variability . For example.92 A check of tables of the F distribution at the 5 per cent level of significance shows that the F ratio must exceed 4. the respective F = interaction withincells mean square mean square F = 35.
L. can arise over a period of time owing to subtle changes in the characteristics of test. O. Brownlee. For details the reader is referred to several excellent texts on this subject.^ In any experimental program involving a large number of tests. Cox. For two digit numbers. W. In the early stages of a research project it is frequently important to determine what are the important factors in the problem under study and what is the relative importance of each of these factors. "The Design and Analysis of Industrial Experiments. M. bias due to uncontrolled secondorder variaa given test. G. 1954. mix them thoroughly. Only a brief introduction to commonly used statistical designs of experiments can be given in this section. New York. New York. Davies. put a set of numbered tags corresponding to the specimen numbers in a jar. . number in the table will be found on the page corresponding to the first digit of the number in the column given by the second digit.. Edinburgh and London. By randomization we permit any one of the many specimens involved in the experiment to have an equal chance of being selected for In this way. "Experimental Designs." John Wiley & Sons. 1952. Chemical Publishing Inc.. in any extended testing program errors bles is minimized. the objective of the test the jar.. The use of a factorial design in place of the conventional approach of changing each factor one at a time will provide a more efficient answer.. ' In the first trial each factor will Table 166 indicates that three runs would K. Each tag should be placed back is in the jar after it is withdrawn to allow an equal probability of selecting that number. One way of randomizing a batch of specimens is to assign a number to each specimen. Company. Cochran and G. Inc. and then withdraw numbers from bility of the variation of properties considered. 1950. Ltd." Oliver & Boyd. 169] Statistics Applied to Materials Testing 439 with the desired expenditure of time and money. This procedure is repeated over and over until all the specimens are the next selected. labor which table of finger involved in random numbers. randomization of the test specimens will minimize variability due to position in the forging.Sec. be investigated at two levels. A. This can best be done through the combined efforts of a statistician and the engineer during the planning stage of the research project." 4th ed. Assume that the problem is to determine the effect of boron and nickel content on the yield strength of a highstrength steel. it is important to randomize the order in which the specimens are selected for testing. "Industrial Experimentation. the testing equipment or in the proficiency of the operator of the In taking metal specimens from large forgings or ingots the possiIf with position in the forging must be is to measure the average properties of the entire forging. The considerable this procedure can be minimized by using a The first number is selected by placing your on a number with your eyes closed. For example.
one control experiment. run 1. is used to judge content. the boron content at constant nickel content. 16 be required by the conventional approach of varying each factor one at By comparing run 1 with run 2. Comparing run 1 with run 3 gives the effect of varying the nickel content for a constant boron For this design. we can find the effect of varying a time. Table 166 Design of Experiment Based on Conventional Oneatatime Variation OF Each Factor Factor .440 Applications to Materials Testing [Chap.
The very rapid rate at which the number of runs required by a factorial design increases with small increases in the number of either variables or levels is the chief disadvantage of this type of experimental design. the term correlation is used in conjunction with regression. In this section only the used where simplest case of linear regression will be considered. . if included. if we were interested in four variables at four levels. Eugenics.Sec. Loraine. 256 runs would be required. the disadvantages of simple facovercome to a great extent by a modification known The number of runs required is substantially reduced without greatly affecting the power of the method. A. K. Davies and W. 291301. 1948. pts. This is also called the method of least squares because the line which results from the analysis has the property that the sum of the squares of vertical deviations of observations from this line is smaller than the corresponding line. Ill. K. torial design are However. The main sacrifice is in runs which. 1945. 233249. Kelly. vol. squares of deviations from any other Frequently. ' 2 1950. where there are two or more independent variables in the regression equation. Hay. where a curved line is fitted to the data. fractional replicate designs is The construction of covered in the foregoing reference and in a number of others.^ higherorder interactions between the variables. above represents the simplest case of a factorial The number of runs required in a factorial design is equal to the number of levels raised to the power of the number of factors being Thus. 268276. D. investigated. pp. and P.^ 1610. vol. An example might be the determination of the relationship between tensile strength (F) and Brinell hardness (. L. J. 6.Y). A. 12. B. no. pp. IV. sum of the Multiple regression. is a further refinement. 35. while correlation deals with the relationship between two or more sets of data which vary jointlj'. Linear Regression The term regression is of a functional relationship used in statistics to refer to the determination between one or more independent variables Xi and a dependent variable Yi. pp. regression is A clear distinction should be made between and correlation. Biometrika. 16 10] illustration given Statistics Applied to Materials Testing 441 The design. Finney. A. Ann. and curvilinear regression. Regression deals with the situation where there a clear distinction between the dependent and independent variables. Brownlee. K. vol. would provide information on as fractional replication. Regression techniques are ordinarily it is assumed that a fixed but unknown relationship exists between the X and Y populations but the random errors in the measurements prevent a reliable determination of the relationship by inspection. is an extension of this analysis. 3. Biometrics.
'^ 2 — Y. (1629) Ox<>y 1611. L.X) (1625) =n where a = '^ n — = Y X.) (1627) nlX^(l 1=1 The measure estimate analysis. 16 A between the dependent variable given by Y and the independent variable X is Y = i a + b(X . "Statistical Quality Control.J ^^^ ft X. A value of r close to zero indicates that the regression line is incapable of predicting values of Y. . by means of an E.2  b's. Syx..Y. 1952. The control one important method which is used in the branch of applied statistics known as quality control.442 Applications to Materials Testing linearregression equation [Chap.y Y from the value 1=1 of the variation of the observed value of is calculated from the regression equation It plays the role of the called the standard error of standard deviation in regression = ' (s. Inc. Control Charts technique which has proved very useful for routine analydata is the control chart. s.)\l '^^ i=i F.^ The 1 use of the control chart can best be described York.. The control chart is a graphical method of detecting the presence of variation statistical sis of A which chart is is greater than that expected as a result of chance.') (1628) The quantity which determines how well the regression equation fits the experimental data is the correlation coefficient r. while a value close to unity indicates nearly perfect prediction. The use of this technique is based on the viewpoint that every manufacturing process is subject to two sources of variation. (1626) n b = ^^ I . Grant.. New . McGrawHill Book Company." 2d ed. n n n n r = ^^ I X.(I X.F. X. chance variation and variation due to assignable causes.
The mean of each sample and the dispersion is determined by computing the range R. range. Control charts for X and R. R. Van .. Princeton. the control limits are determined. of the The aver age of the sample means. 167. 1611] illustration. A. ' of Manufactured Product.. 1931. Inc. and the means for a 1 per cent chance of the sample value exceeding the control limits due to random chance can be determined from Table 168. 167). "Economic Control of Quality Nostrand Company.Sec. is Points which fall outside the control limits indicate that the variation greater than would be expected solely from chance and that corrective action should be taken to bring the process into control.J. are first determined." D. N. mean and the range (Fig. Shewart. The determination of the upper control limit (UCL) and the lower (LCL) for the mean and the range is greatly simplified by The control limits for the range the use of tables developed by Shewart. A separate X control chart is kept for the /x. Statistics Applied to Materials Testing 443 Consider a commercial heattreating operation where bear ing races are being quenched and tempered in a conveyortype furnace on a continuous 24hr basis. control limit ' W. and the mean sample values of the Rockwell C hardness 47 46 45  UCL 43 42 41 LCL 40 2 4 J L 10 6 8 Batch number I I 12 14 UCL LCL 2 4 6 8 10 12 14 Batch number Fig. Every 2 hr the Rockwell hardness is measured on 10 bearing races to determine whether or not the product conis computed forms to the specifications. Next.
16 For the range UCL = DuR LCL = DlR For the means (1630) UCL = LCL = M M + A^R .A2R 168 (1631) Table Multipliers for Use in Determining Control Limits OF Control Charts f Sample size .444 Applications to Materials Testing [Chap.
where V is the volume of the specimen. Swed. The mathematics for this crack distribution has been worked out by Frenkel and Kontorova. the distribution of crack size is assumed to of crack sizes. Davidenkov. as the volume of the specimen increases number of cracks also increases and therefore the probability of is is encountering a severe crack increased.^ For a Gaussian distribution of the strength of the weakest element the strength should decrease linearly with increasing (log Vy^. 1612] Statistics Applied to Materials Testing 445 is determined. 151. This implies that the probability of having a small defect be Gaussian. A. most dangerous This concept of brittle fracture is called the weakestlink imperfection. ^ increasing volume according to requirement. the brittlefracture strength concept in direct analogy with the fact that the strength of a chain is determined by the strength effect arises quite naturally of its weakest link. E. Roij. pp . where m is the experimentally determined factor in WeibuU's distribution function. J. Appl. but by the one. Research. Eng. WeibuU's distribution function (see Sec. not by an average value of the distribution of imperfections.). 1939.S. 7. as in bending or torsion. If the stress distribution imposed by the test method is nonuniform. Mech. vol. of crack sizes. WeibuU. and F. p{c) =expf^^ (1633) where p(c) a = = probability of occurrence of a crack of length c a constant distribution predicts that the strength decreases linearly The Laplacian with an increase in log V. Therefore. Experiments on the size effect in tension and bending of steel at low temperature show good agreement with WeibuU's prediction. vol. the analysis must be based on the surface area.. Statistical Theory of the Strength of Materials. 3 N. 2 W. Shevandin.Sec. 166) has also been used to describe the distribution is This predicts that the strength should decrease with V~'^''". A Frenkel and T. /. Figure 168 shows the calculated frequency distribution of fracture stresses as a function of the 1 number of cracks N J. 6367. I. If the crack density of assumed constant. pp. is the same as the probability of occurrence of a large defect. 14. The problem of relating the crack distribution to the fracture strength one of finding the distribution function of the smallest value of strength as a function of the number of cracks A'' for a given distribution function Frequently. no.^ It is also argued that a more reasonable distribution of defects decreases in proportion with the The Laplacian distribution adequately expresses this size of the defect. The existence of a size the material the total from this concept. 1946.S. 108114. Inst.R. Phijsics (U. Wittman. Kontorova. 1943.
120 >.446 Applications to Materials Testing of a Laplacian distribution of cracks. but as the number of defects reaches large values.100 o 1 80 1 D 60 40 20 ^ 11) . enough to determine which of these distribution functions for crack size is most generally applicable. 16 on the basis the total Note that the larger stress number of cracks the smaller the scatter in fracture because of the higher probability of finding a critical defect. ^ [Chap. Also. as the number of defects increases. there is less Existing data are not numerous relative decrease in the mean value. the mean value of the fracture stress decreases.
9lA. Consequently. However. The statistical analysis of the fatigue data in the region of the fatigue limit requires special techniques which have not yet been discussed in In making tests to measure the fatigue limit. we can test this chapter. 1613] Statistics Applied to Materials Testing 447 life distribution does not accurately describe the distribution of fatigue at the upper and lower extremes of the distribution. Two particular types of statistical analysis of sensistatistical tivity experiments have been applied to the determination of the fatigue limit. Finney. J. The number of specimens tested at each stress should be apportioned with regard to the expected percentage of survival. 1952.Sec." ASTM Spec. 2 D. A number of specimens are tested at each of four to limit. a is obtained. the test stress. New York. "Probit Analysis. Stepbystep procedures for the analysis will be found in the ASTM Manual on statistical analysis of fatigue data. it is to be expected that the extremevalue distribution function would provide better agreement in these regions. ^ The term probit is an abbreviation of probability unit.^ The percentage of survivors are converted to probit values^ and a regression line is determined according to Finney's procedure for weighting the values at different percentages of survival. The analysis follows the procedures given by Finney in his book on probit analysis. the same type of analysis can be made without resorting to any special techniques.^ The first of the methods for the statistical determination of the fatigue limit is known as probit analysis. Not fewer than 5 specimens should be tested at a stress level. and at least a total of 50 specimens should be used in the determination of the median value straight line test 1 "A Tentative Guide for Fatigue Testing and the Statistical Analysis of Fatigue Data." Cambridge University Press. Tech. a given specimen only at a particular stress. 1958. Since fatigue life depends on the weakest section rather than on average behavior. if the specimen does not fail (a runout) in the prescribed we know that its fatigue limit lies somewhere above Because a given specimen cannot be tested more than once." it is necessary to estimate the statistics of the fatigue limit by testing a large number of presumably identical specimens Experiments of this type are known as sensiat different stress levels. Publ. tivity experiments. even if it "ran out. . and if the specimen fails before the cutoff limit of 10^ cycles then we know that the fatigue limit of the specimen is somewhere below the stress level which we used. six stress levels in the vicinity of the estimated fatigue When is the percentage of the specimens which survived at each stress plotted on normalprobability paper against the stress level. Transformation to probits is a method of eliminating negative values from the standard normal deviate z. The procedures used will be briefly described below. number of cycles. One way of performing the experiment is to 20 specimens at each stress level.
169. The second method of analyzing fatigue data at the fatigue limit is the staircase. method.^ This method provides a good measure of the mean fatigue limit with fewer specimens than are required with the probit method. Mood. and it is possible to use a standard linearregression analysis. The first specimen is tested at the estimated value of the fatigue limit. or upanddown. Dixon and A. M. In the standard linearregression equation X is the stress level. 1948. 43. A^n. median value limit) is of the fatigue limit (often loosely called the mean line at fatigue given by the Its stress which intercepts the regression is 50 per cent survival. but it has the disadvantage that tests must be run in sequence.448 Applications to Materials Testing [Chap. 15 specimens should be tested at stresses which give 15 to 20 and 80 to 85 per cent survival and 20 specimens are required at stresses which result in 10 and 90 per cent survival. standard deviation given by the stress range between 50 and 84 per cent survival. If this specimen fails. The stress applied to the next specimen is then increased by the increment. The testing procedure is illustrated in Fig. 109. The procedure is further continued. Staircase testing sequence for determination of mean fatigue limit. vol. 16 If 10 specimens are tested at stress levels which give between 25 and 75 per cent survival. With this relative distribution of specimens the test values are adequately weighted at the extreme values. The of fatigue limit. O S46 44 42 X = failure o = nonfailure 40 J 2 I I I 1 I I I J \ L 16 J I 4 6 8 10 12 Specimen number 14 18 Fig. the stress for the next specimen is decreased by a fixed increment. 169. 50 . Assoc. and Y is the percentage of survivors converted to values of the standard normal deviate z. The values of z corresponding to a given percentage of survivors p are obtained by entering an expanded version of Table 162 under the column headed Area with 1 — p/100. J. This procedure is continued for each succeeding specimen until a runout is obtained.48 a. 1 W. . p. Statist. J.
psi .1613] Statistics Applied to Materials Testing 449 the stress being increased it fails. Table 169 Method of Analyzing Staircase Data Stress. Fifteen to twentyfive specimens when a specimen runs out and decreased when must be tested.
E. 845899. and F. Dixon. "Methods of Statistical Analysis." 2d ed. G.. Massey. Fisher.: "Statistical Methods for Research Workers. . Inc. 42.. Inc. New York. C.. R.. New York. McGrawHill Book Company. vol. A. Hald. Goulden. J. 1951.. J. Trans.. H. Jr..450 Applications to Materials Testing [Chap. : New York. 1952." 12th ed. Inc. S. ASM. pp. 1950.: "Statistical Theory with Engineering Applications." 1st ed. New York. 1954. 16 BIBLIOGRAPHY W. John Wiley & Sons.. Olds.: "Introduction to Statistical Analysis." John Wiley & Sons. 1952. Hafner Publishing Company. and Cyril Wells: Statistical Methods for Evaluating the Quality of Certain Wrought Steel Products.
Part Four PLASTIC FORMING OF METALS .
.
Bending processes Shearing processes is In directcompression processes the force applied to the surface of the workpiece.Chapter 17 GENERAL FUNDAMENTALS OF METALWORKING 171. pressive forces developed application of bending moments to the sheet. The best example of a tensiontype forming process is stretch forming. 5. Directcompressiontype processes Indirectcompression processes Tensiontype processes 2. 4. ClassiFication of Formins Processes The importance to the ease with of metals in modern technology is due. Figure 171 illustrates these processes in a very simplified way. Hundreds of processes have been developed for specific metalworking applications. in large part. These categories are: 1. 171). The primary applied forces are frequently tensile. The chief examples of this type of process are forging and rolling (Fig. However. but the indirect com by the reaction of the workpiece with the die reach high values. Indirectcompression processes include of a wire and tube drawing. the metal flows under the action of a combined stress state which includes high compressive forces in at least one of the principal directions. where a metal sheet is wrapped to the contour Bending involves the of a die under the application of tensile forces. and the deep drawing cup. Therefore. while shearing involves the application of shearing forces of sufficient magnitude to rupture the metal in the plane of shear. these processes may be classified into only a few categories on the basis of the type of forces applied to the work piece as it is formed into shape. 453 . which they may be formed into useful shapes. 3. extrusion. and the metal flows at right angles to the direction of the compression.
An important purpose of plastic working operations is to break down and refine the columnar or dendritic structure present in cast metals and very precise. objective One is objective is to produce a desired shape. . wire and drawing. Forming methods which produce a part to a final finished shape are called secondary mechanical working processes.454 Plastic Forming of Metals [Chap. this area is not The terminology in Frequently the first category is referred to as processing operations. and the second is called fabrication. of the material The second to improve the properties through the alteration of the distribution of microconstituents. plate. 1 71 Bending Shearing . bar. are called primary mechanical working processes. and the introduction of strain hardening. Most sheetmetal forming operations. Forging Rolling t t A Wire drawing Extrusion Deep drawing ^\ ^ Stretch forming Fig. the refinement of grain size. and tube drawing are secondary processes. Plastic billet to working processes which are designed to reduce an ingot or a standard mill product of simple shape. such as sheet. 17 Plastic forming operations are performed for at least two reasons. Typical forming operations.
Sec. Very large deformations are possible in hot working because the recovery Hot working occurs at an processes keep pace with the deformation. constitutes cold working because this highmelting metal has a recrystallization temperature above this working temperature.000°F. Similarly. tungsten at 2. working It is in the hotworking range for steel. Hot Working Hot working is the initial step in the mechanical working of most metals and alloys. breaking down a cast structure. Once the brittle constituent is broken up. Therefore. the energy required for deformation is generally much less for hot working than for cold working. Effect of Temperature on Forming Processes Forming processes are commonly classified into hotworking and cold working operations. and because the flow stress decreases with increasing temperature. its effect on the mechanical properties is minor and ductility and strength Forging and rolling are the processes ordinarily used for are increased. the flow stress increases with deformation. 1 72] General Fundamentals of Metalworking 455 alloys. Not only does hot working result in a decrease in . the total deformation that is possible without causing fracture is less for cold working than for hot working. esses are not effective. lead and tin recrystallize rapidly at room temperature after large deformations. By compressive deformation it is often possible to fragment a brittle microconstituent in such a into the spaces way that the ductile matrix flows between fragments and welds together to leave a persound structure. cold place simultaneously with the deformation. deformation carried out under conditions where recovery procIn hot working the strain hardening and distorted grain structure produced by deformation are very rapidly eliminated by the formation of new strainfree grains as the result of recrystallization. important to realize that the distinction between cold working and hot working does not depend upon any arbitrary temperature of deformation. so that the working of these metals at room temperature constitutes hot working. Since strain hardening is not relieved in cold working. extrusion is the best method fectly because the billet is subjected to compressive forces only. However. However. Hot working is defined as deformation under cou' ditions of temperature and strain rate such that recovery processes take On the other hand. For most commercial alloys a hotworking operation must be carried out at a relatively high temperature in order that a rapid rate of recrystallization be obtained. essentially constant flow stress. unless the effects of cold working is work are relieved by annealing. Frequently the low strength and ductility of castings are due to the presence of a brittle constituent at the grain boundaries and dendritic boundaries. 172.
will depend upon such factors as the amount of deformation and the time ture. Since the greater the amount of deformation the lower the recrystallization temperature. Since the deformation is always greater in the surface layers. surface reactions between the metal and the furnace atmosphere become a problem. there are certain disadvantages to hot working. The lower temperature limit for the hot working of a metal is the lowest temperature at which the rate of recrystallization is rapid enough to eliminate strain hardening in the time when the metal is at temperFor a given metal or alloy the lower hotworking temperature ature. but the rapid diffusion at hotworking temperatures aids in decreasing the chemical inhomogeneities of the castingot struc Blowholes and porosity are eliminated by the welding together of and the coarse columnar grains of the casting are broken down and refined into smaller equiaxed recrystallized grains. and a considerable amount of metal may thus be lost. Reactive metals like molybdenum are severely embrittled by oxygen. when This is to allow for the possibility of segregated regions of lowermelting . However. oxidation results. the dimensional tolerances for hotworked mill products are greater than for coldworked products. the structure and properties of hotworked metals are generally not so uniform over the cross section as in metals which have been coldworked and annealed. Because high temperatures are usually involved. The upper limit for hot working is determined by the temperature at which either melting or excessive oxidation occurs. the metal will have a finer recrystallized grain size in this region. Rolledin oxide makes it difficult to produce good surface finishes on hotrolled products. These changes in structure from hot working result in an increase in ductility and toughness over the cast state. grain growth can occur in the interior of large pieces. is rapidly deformed and cooled rapidly from temperature will require a higher hotworking temperature for the same degree of deformation than will metal slowly deformed and slowly cooled. which cool slowly from the working temperature. and therefore they must be hotworked in an inert atmosphere or protected from the air by a suitable container. the lower temperature Metal which limit for hot working is decreased for large deformations. Surface decarburization of hotworked steel can be a serious problem.456 Plastic Forming of Metals [Chap. and frequently extensive surface finishing is required to remove the decarburized layer. Generally the maximum working temperature is limited to 100°F below the melting point. Ordinarily hot working is done in air. Further. and because allowance must be made for expansion and contraction. 17 the energy required to deform the metal and an increased abihty to flow without cracking. the metal is at temperature. these cavities. Because the interior will be at higher temperatures for longer times during cooling than will be the external surfaces.
a lowermelting constituent pieces when it is deformed. passes is kept well above the minimum working temperature in order to take advantage of the economies offered by the lower flow stress. Figure 172 schematically illustrates the property changes involved in this cycle. just last pass should be relatively large. or steps. Only a very small amount of a grainboundary film of is needed to make a material crumble into Such a condition is known as hot shortness. cold working of a metal results in ductility. tility which is not possible in hotworking operations. Typical variation of strength and ductility in the coldworkanneal cycle.172] General Fundamentals of Metalworking 457 point material. the metal will fracture before reaching the desired size and shape. Annealing temperature 172. it provides a degree of versaductility. By suitably Cold work Annealing Per cent cold work Fig. 59. in an increase hardness and a decrease When cold working excessive. Since a finegrainsized product is usually common practice is to lower the working temperature for the last pass to the point where grain growth during cooling from the work This finishing temperature is usually above the minimum recrystallization temperature. in order to avoid such difficulties. or burning. coldworking oper ations are usually carried out in several steps. with intermediate anneal ing operations introduced to soften the coldworked metal and restore the This sequence of repeated cold working and annealing is frequently called the coldworkanneal cycle. Cold Working As was shown in strength or is in Sec. are carried out in a number of multiple Generally the working temperature for the intermediate Most hotworking operations passes. the amount of deformation in the ing temperature will be negligible. particularly for reactive metals which must be annealed in vacuum or inert atmospheres. It is likely that some grain growth will occur subsequent to the recrystallization at these temperatures. . desired. Although the need for annealing operations increases the cost of forming by cold working. Therefore. In order to ensure a fine recrystallized grain size.
because the recrystallization process proceeds relatively rapidly and is quite sensitive to small temperature fluctuations in the furnace. then the final operation must be a coldworking step with the proper degree of deformation to produce This would probably be followed by a stress relief the desired strength. Effect of Speed of Deformation on Forming Processes metal to forming operations can be influenced by the deformed. 17 adjusting the eoldwork anneal cycle the part can be produced with any desired degree of strain hardening. halfhard. and spring temper. for certain metals there may be a temperature region below which the metal will shatter when subjected to a highspeed or impact load. threequarterhard. Therefore. Table 171 lists typical values of velocities for different types of testing and forming operations. values of flow stress measured in the tension test are not directly applicable for the computation of forming loads. fullhard. If the finished product must be stronger than the fully annealed material. It is important to note that the forming velocity of most commercial equipment is appreciably faster than the crosshead velocity used in the standard tension test. If it is desired to have the final part in the fully softened condition. Thus. Each temper condition indicates a different percentage of cold reduction following the annealing treatment. remove residual stresses. For cold working a change in the forming speed of several orders of magnitude results in only about a 20 per cent increase in the flow curve.458 Plastic Forming of Metals [Chap. whereas a limited amount of slowspeed deformation can be accomplished in the same temperature of a The response ispeed with which it is range. iron and steel will crack if hammered at temperatures well below room temperature. Exceptions to this statement are the possibility of brittle behavior in a certain temp<jrature region at high forming speeds and the fact that highspeed . then an anneal follows the last coldworking step. depending upon the degree of cold reduction following the last anneal. so that for practical purposes the speed of deformation can be considered to have little effect. It is customary to produce coldworked products like strip and wire in different tempers. quarterhard. Such a procedure to develop a certain combination of strength and ductility in the final product is more successful than trying to achieve the same combinations of properties by partially softening a fully coldworked material. For example. The existence of a transition from a ductile to brittle condition for most bodycentered cubic metals over a certain temperature range has already been discussed in earlier chapters. This transitiontemperature phenomenon is more pronounced for rapid rates of deformation. The coldworked condition is described as the annealed (soft) temper. to 1 73.
aluminum.Sec. 2 T.. ft /sec Tension test Hydraulic extrusion press Mechanical press Charpy impact test Forging hammer Explosive forming 2 X lO"" to 2 X IQ^ 0. depends on J. Hockett. 74. E. DuMond.^ Among the advantages offered are the fact that highstrength materials can be formed without large springback. Phillips. Table 171 Typical Values of Velocity Encountered in Different Testing and Forming Operations Operation Velocity. and mild steel at a series of constant temperatures indicate that the dependence of flow stress on strain rate over a range of e from 1 to 40 sec~' can be expressed by Eq. . pp. 13091319. 59. However. Effect of Metallurgical Structure on Formins Processes The forces required to carry out a forming operation are directly related to the flow stress of the metal being worked. speeds. there is a greater danger of hot shortness. A recent development has been the ultrahighvelocity forming of metals at velocities of 100 to 400 ft /sec by using the energy generated by the detonation of explosives. 1959. C. /. Inst. Proc. A. is quite markedly affected by the speed There are no routine methods for measuring the flow stress during hot working. Highspeed deformation may produce regions of nonuniform deformation (stretcher strains) in a steel sheet which shows no stretcher strains on slowspeed deformation. pp. called the cam plastometer. Because the time at temperature is shorter for high forming (935). and that certain shapes can be produced which cannot be made by any other technique. that metals flow readily into the recesses of the die. 6876. is one device which has given good results. 1958. 81. vol. 8086. ASTM. because the metal retains the heat developed by deformation to a greater extent at high forming speeds. vol. Alder and V. in turn. Data obtained^ for copper. pp. a highspeed compression testing machine. Metal Prog. 1 J. On the other hand. F. For metals which have a rather narrow hotworking range these opposing effects serve to close the gap still further and make it impractical to carry out hot working at very high forming speeds. 1 74] General Fundamentals of Metalworking 459 deformation accentuates the yieldpoint phenomenon in mild steel. 195455. vol. the minimum recrystallization temperature is raised and the minimum hotworking temperature will be higher. The flow stress for hot working of deformation. This.01 to 10 5 to 5 10 to 20 10 to 30 100 to 400 . Metals. 1 74.
If the inclusions or particles are brittle. If the secondphase particles are soft. The presence of a massive. actually have lower flow stresses in the hotworking region than the lower temperature limit for hot the singlephase alphabrass alloys. while if the particles are harder and stronger than the . solidsolution alloying additions. alloys. such as pearlite in mild steel. then difficulties with hot shortness will be encountered. The presence of a large volume fraction of hard uniformly dispersed particles. greatly increases the flow stress and makes working quite difficult. The plastic working characteristics of twophase (heterogeneous) alloys depend on the microscopic distribution of the secondphase particles. which consist of a relatively hard beta phase in an alphabrass matrix. they will be broken into fragments which will be oriented parallel to the working direction. Secondphase particles or inclusions which are originally spheroidal will be distorted in the principal working direction into an ellipsoidal shape if they are softer and more ductile than the matrix. results in less increase in flow stress than for very finely divided secondphase particles. The shape of the carbide For annealed steel. this temperature is about onehalf (As a point. a spheroidization heat treatment. the metallurgical structure and composition of the alloy. 17 For pure metals the ease of mechanical working will.) of the alloy generally raises the flow curve.460 Plastic Formins of Metals [Chap. The addition alloying elements to form a solidmelting point. in general. In the coldworking region the flow stress of alphabeta brass is appreciably higher than that of alpha brass. the upper hotworking temperature must usually be reduced in order to prevent hot shortness. zation temperature is approximately proportional to the melting point. is often used to increase the formability An important exception to the general rule that at room temperature. uniformly distributed microconstituent. work will also increase with melting rough approximation. they have only a small effect on the working characteristics. such as are found in the SAP class of hightemperature alloys. and the forming loads solution Since the melting point is often decreased by increase proportionately. decrease with Since the minimum recrystalliincreasing melting point of the metal. As the result of a mechanical working operation secondphase particles will tend to assume a shape and distribution which roughly correspond to the deformation of the body as a whole. If these particles have a lower melting point than the matrix. particles can be important in coldworking processes. Alloys which contain a hard second phase located primarily at the grain boundaries present considerable forming problems because of the tendency for fracture to occur along the grain boundaries. the presence of a hard second phase increases the difficulty of forming is These brass alloys containing 35 to 45 per cent zinc (Muntz metal). which converts the cementite platelets to spheroidal cementite particles.
fatigue properties. Metals. 174] General Fundamentals of Metalworking 461 matrix. 1958. 189. Inst. there will be an appreciable decrease in ductility. 5X general. it will produce an increase in the flow stress but. Pickering. Unkel. . 1937. If a precipitation reaction occurs in a metal while it is being formed. The forming an alloy can be affected if it a straininduced precipitation or straininduced phase transformation. 174). J. (London). the tensile ductility. Fig. pp. be observed after macroetching (Fig.or from cold 1 H. it usually results when the working is carried out at a temperature just below the solvus line. 148159. J. When brittleness is caused by precipitation. the mechanical An important consequence of mechanical fibering is that properties may be different for different orientations of In the test specimen with respect to the fiber (working) direction. which can result in cracking.Sec. more important. will and impact properties fiber) be lower in the transverse direction (normal to the characteristics of than in the undergoes longitudinal direction. 1711<)G. vol. 173). 2 The boundary on the phase diagram of the limit of solid solubility of a solid solu tion. Iron Steel Inst. of The fiber structure can Microscopic examination wrought products frequently shows the results of this mechanical fiber ing (Fig. F. 173. B. 61. pp. they will be essentially undeformed.' The orientation of secondphase particles during hot or cold working and the preferred fragmentation of the grains during cold working are responsible for the fibrous structure which is typical of wrought products. vol. Fiber structure in wrought product revealed by macroetching.
A most outstanding practical example of a straininduced phase transformation occurs in certain austenitic stainless steels where the Cr:Ni When these alloys are coldratio results in an unstable austenite phase. To facilitate the forming of agehardenable aluminum alloys. While this phase transformation is often used to increase the yield strength by cold rolling. 1 75. Mechanics of Metal Forming of the metal forming is to express forming processes in the mathematical language of applied mechanics so that predictions can be made about the forces required to produce the desired shape. it can also result in cracking during forming if the transformation occurs in an extreme amount in regions of highly localized strain. duces an abnormal increase in the flow stress for the amount of deformation received. Fibered and banded structure in logitudinal direction of a hotrolled mild steel plate.462 Plastic Forming of Metals [Chap. they are frequently refrigerated just before forming in order to suppress the precipitation reaction. Since the forces of research in One prime objectives the forces and deformations involved in . the austenite transforms to ferrite along the slip lines and pro «* r>* Fi3. worked. difficulty from this factor is more likely when forming is carried out at a slow speed at an elevated temperature. 1 74. 17 working after the alloy had been heated to the same temperature region. 100 X. Since precipitation is a diffusioncontrolled process.
. Plasticity theory has made important advances since World War II.>)2 + (a. of usually necessary to This branch mechanics is an outgrowth of the theory of plasticity discussed in Chap. use simplifying assumptions to obtain a tractable solution. 312. 175] General Fundamentals of Metalworking it is 463 and deformations are generally quite complex. and numerous efforts have been made to utilize the new knowledge to improve the accuracy of prediction of the metalforming equations. or Von Mises. many instances the use of these more advanced methods provides a deeper insight into the forming in process. flow criterion [Eq. Hitchcock. and in view of the uncertainties present in the analysis of some of the complex forming operations the tw^o yield criteria can be considered nearly equivalent. 2 appeared weekly in the magazine Steel. the maximumshearstress law will be used in certain cases where it provides appreciable simplification to the analysis. 1933. to May In much of the literature in this field the flow stress is oq is denoted by the symbol k. An 1 important feature of plasticity theory is is the assumption that the An 7. Most of this work involves the use While in of the slipfield theory discussed briefly in Sec. usually this additional information cannot be obtained without considerable increase mathematical complexity. The mathematical descriptions of technologically important forming processes which were developed by SiebeP and other German workers were based on the assumption that the maximumshearstress law was the proper criterion for describing the stress condition for producing plastic flow. excellent review of E. (172)] provided better agreement with experimental data. as was seen in Sec.  as)' + (era  cti)^ = 2^0^ is (172) Therefore. However. which should be reviewed before completing this chapter and the remaining chapters of this book. 16. H. References to more advanced treatments will be given wherever this is applicable. from Oct. Because an analysis based on slipfield made with the background provided in Chap.0"! (171) Subsequent work showed that the distortionenergy. in discussing the mathematical aspects of specific forming processes in succeedtheory cannot be ing chapters only the simpler equations will be derived. 3. (cTi  cr. 34.Sec. Siebel's it work available in an English translation by J. Therefore. the distortionenergy criterion for yielding to be preferred and be used in most of the analyses of forming processes presented in subsequent chapters. 1934. the distortionwill energy and maximumshearstress criteria differ at most by only 15 per cent. 3. In this text k taken to indicate the ^I'eld stress in shear.
the flow curve at For increasing stress. Moreover. strain. for the yield stress strains the hydrostatic stress does have an effect plastic at large However. 716. following equation. To a very good approximation it is permissible to assume that the volume remains constant during deformation.464 Plastic Forming of Metals [Chap. It is mean stress has no effect on considered that only the stress is of importance in producing plastic flow. on the flow was shown in Sec. strain can write a compressive we strain equal to tensile ei €3. container. This leads to the convenient relationships €1 + €2 or dei it + de2 + €3 = + des = ('[7 ^^ can be assumed that the strain increment is proportional This is called proportional straining and leads to the to the total strain. is this stress term borne out by the experimental fact that the assumption is This (342)].= In t . dei ei ^ de2 62 ^ des 63 (\7 A) premise of plasticity theory is that equivalent strain hardening For a is obtained for an equivalent tensile or compressive deformation. or natural. 17 introduction or removal of a hj^drostatic or the flow stress or the state of strain. the ductility raised. Because large deformations occur in metal forming. which is often useful for integrating equations con Frequently taining the strain differential. extruded be often developed owing to the reaction of the workpiece with the extrusion of metals in tension is present. 214) is appreciably increased when a high hydrostatic explains why nominally brittle materials may This pressure since a high hydrostatic compression is successfully. it is important to express stress and strain as true stress and true. ei A basic = —63 = In^ = ~lnr. as strains is large deviator (see Sec. and it which appears in the plasticity equations [see Eq. beginning of flow is independent of the mean stress. hydrostatic stress.
In this test the metal is subjected to plane compression. for calculating existence of a straininduced transformation. IB. {London). degree of strain hardening which occurs for a given reduction is higher than would be determined from a tensileflow curve. just as in the case of a notch in a tension specimen (Sec. pp. vol. The flow curve (true stressstrain curve) determined for either tension is or compression the basic relationship for the strainhardening behavior It is used to determine the value of the flow stress oq forming loads. 175] General Fundamentals of Metalworking 465 Equation (175) expresses the fact that for equal true strains the reduction of area is equal to the reduction in height or thickness. Mech. % 50 True strain € Fig. It is fre quently useful to employ these parameters as a substitute for strain in metalforming experiments. For most commercial forming operations the 10 20 30 40 Reduction of area by drawing. and possibly the of the material. 175. and the flow stress is raised. This is due to the that the metal undergoes nonuniform flow during deformation because it is not allowed to flow freely. Engrs. Fig. 175). Proc. B. . since there is no deformation in the width direction. 195253. stressstrain curves after different Flow curve constructed from amounts 176. 448453. Watts and H. The value of the flow stress will of course depend on the temperature. 712). Method of using average flow for strain hardening. as described in earlier sections of this chapter. The lightly deformed regions provide a constraint to plastic flow. stress to compensate of reduction.^ is to measure the pressure required to produce plastic flow when a sheet is compressed between two rigid anvils.Sec. Ford. One way of determining the flow curve in cases where deformation is nonvniiform is to determine the yield stress after different amounts of reduction in the forming operation (Fig. Inst. the speed of deformation. 1 A. which is used frequently in England. A method for measuring the flow stress fact for coldworked metals.
of the by com Regions metal which have been strained only elastically and regions between the elavsticplastic boundary in which the yield stress has been exceeded but flow is constrained by the elastic region are considered to be rigid. R. but at large strains. 19591960. and the flow stress is constant and independent of the amount of deformation at a given temperature and speed of deformation. it is just as important to describe the geometry of flow in relation to the stress system as it is to be able to predict the stress conditions to produce plastic flow. By using the more advanced slipfield theory. the strains are much greater than can be obtained in a tension or compression test. 76. . J. 88. The work of defor1 2 R. vol. This is a good assumption up to moderate strains. Work of Plastic Deformation The total work required to produce a shape by plastic deformation can be broken down into a number of components. many plastic forming operations. {London). 176. it is possible to consider the stress and strain in both the elastic and plastic regions. vol. it is customary to use a constant value of flow stress which is an average over the total strain. good estimates of the flow stress for reductions greater than 70 to 80 per cent by linear extrapolation when ctq is plotted against the logarithm of the strain or the reduction in area. B. Whitton. as in Fig. where preferred orientations may have been developed.^ For hot working the metal approaches an ideal plastic material. Sims. (31). 145149. 393399. pp. W.. then the plastic stress and strain can be considered coaxial. i. 17 Flow curves In for a number of steels and nonferrous metals have been obtained^ by this method. that stress and strain are coaxial. and 1 also along the elasticplastic boundary. the stress and strain systems usually are not identical. If Lode's stress and strain parameters are equal (see Sec. An alternative.e. To allow for strain hardening in cold working. such as extrusion. which adds to the mathematical complexity. is to include a mathematical expression for the flow curve in the analysis. 177. 35). Metals. Only flow in the completely plastic region of the body is considered in the relatively simple analyses of plastic forming given in succeeding chapters. A basic assumption of plasticity theory which allows this is that at any stage in the deformation process the geometry of strain rates is the same as the geometry of stress. Inst. pp. To describe the plastic flow of a metal. J. 1954. Usually this is limited to a simple power function like Eq. Iron Steel Inst. elastic strains of the order of 0.001 are negligible parison and the metal can be treated as a rigid plastic material. Wilcox and P. Since metalforming problems are concerned with large strains of the It is possible to get order of unity.466 Plastic Forming of Metals [Chap. J.
+ (T3 d€3 de2. J. Wt = Wd + Wr + W. 95 per cent.. wire drawing 50 to 75 per rolling 75 to and sheet work of plastic deformation assumes an ideal plastic material which obeys the distortionenergy criterion of yielding and the LevyVon Mises theory of plastic flow (Sec. Sachs. McGrawHill Book Company. 54. are extrusion 30 to 60 per cent. Iron Steel Insl.for the is The that the metal dWd = (Xl dei + (To de. Efficiency is (177) the work of deformation divided = r. Often part of the total work expended in or internaldeformation. (348) it ((Ti  as) rfei + {ai  <Ji) de2 (1710) follows that de. (179) Since constancy of volume exists. (Lundon). dc2. 1 76] General Fundamentals of Metalworking is 467 mation Wd the work required is for from the initial to final cross section homogeneous reduction of the voUime by uniform deformation.""'  ""'  "' ) (1711) G. Therefore. and des. stressstrain curve multiplied can be seen that the work of deformation be expended to carry out a This is equal to the area under the effective total volume. "Introduction to the Theory of Plasticity for Engineers. Hoffman and G. following analysis'. p. de^ = —dei — dWd = From Eq. dWd is the increment of work per unit volume dissipated during the infinitesimal straining of increments dei." pp. 1955. = deJ . dl Wd = Vja The efficiency of a forming process by the total work of deformation. 39). 180. 1 J. work is redundant work Wr. Inc. Typical efficiencies for form ing processes cent. 1953. Wistreich. The redundant. Finally. part of the total work must be used to overcome the frictional resistances at the interface between the deforming metal and the tools. the total work can be written as the summation of three components. From the above definitions. it (176) represents the minimum energy which must by the particular forming process. 2 . 5254. New York. O. the energy expended in deforming the body which is not involved in a pure change in shape. vol. = W^ ^ W (178) The total work 1 is usually measured with a wattmeter attached to the electrical drive of the forming equipment.Sec.
+ e^)^^ (1718) 77.' + e^e. (1711) into Eq. Formability Tests and Criteria The success of a forming operation depends on the application of high enough forces to overcome the resistance to deformation without exhaust . given in terms of effective stress and strain by _ and since the effective stress a is a dt 3 de equal to the uniaxial yield stress in the ao.)^ + (^€3  de^)']^^ (1716) and from the constancyofvolume relationship des = —dei — de2 Simplifying Eq. (1713). de^ + de^'^f'^ (1717) For the case of proportional straining. (1715) can be integrated to obtain the total ideal work per unit volume. = ^° dWa = From ^^^ = o Zero dt aode (1715) the definition of effective strain rfe a/2 = V o [{dei  de^y + ((ies  de. 2(7i — 'T2 — (Xz Upon referring to Eqs. 17 Introducing Eq. the variable X X is must be known as a function of strain. (1710) and simplifying. we can o ae write (1714) X Substituting into Eq. (312) and (347) it is possible to simplify the above equation to dWa = To ^dt oA  :4713) integrate this equation. Wa= 1 f dWa=^ ^ Toie. (1716). de = —7 2 (dei^ + de. distortionenergy yield criterion. Eq.468 Plastic Forming of Metals [Chap.
at least a qualitative indication of the common metals bend. Because of the difficulty of obtaining reliable data on the deformation resistance of metals during hot working at high deformation rates a number of empirical tests have been developed to evaluate the hot formability The singleblow impact test^ has been used for many years of materials." American Society for INIetals. the forming limit is the ability of the metal to deform without fracture. K. Trans. 943966. 736748. Clark and J. has been obtained with the hot twist test. since eu = n. W. 1937.'^ In this test the metal is stressed in torsion at a high rate. 1 O. The torque required to twist the specimen is a measure of the flow stress. The standard measures forming limits for sheetmetal forming can be obtained from a more detailed analysis of the tensile stressstrain curve. pp.. and dimpling. AIME. only natural that attempts have been made to predict the success of forming operations on the basis of simple laboratory tests. The reduction of area is a qualitative indication of this forming limit. In forming operations where the sheet is restricted from deforming in one area while it is stretched into shape in another region the metal must be able to deform to a large extent without localized deformation. 1946. of ductility which are obtained from the tension elongation and reduction of area. Howtest. A ever. . give only a very crude indication of the ease with which the metal may be formed without cracking. Trans. Ohio. "Working of Metals.Sec. Further. Samples of sheet are bent around successively smaller radii until cracking occurs on the tensile (outside) surface of the The minimum bend radius is taken as the smallest radius which can be used without cracking. 2 H. Russ. 167. vol. roll forming. e. a metal with a high strainhardening coefficient should perform well in this type of forming operation. A measure of its ability to do this is the uniform elongation €». is qualitative test for the evaluation of the formability of sheet the bend test. vol. pp. to evaluate whether or not a material can be hotworked without crackGood correlation with hotworking operations. but there is no convenient test for exactly determining these data at high rates of deformation. 749790. and the number of twists before fracture is a criterion of the ductility. 1946. Ellis. such as forging and ing. L. Ihrig. a 27" bend radius indicates a more ductile sheet than a 4T bend radius. Reliable information on the flow stress required to produce deformation can be obtained with the tension or compression test for slow rates of deformation. piercing of solid rounds to make seamless tubes. AIME. Metals Park. 177] General Fundamentals of Metalworking It is 469 ing the ductility of the metal. The minimum bend radius is usually expressed in multiples of the sheet thickness. 3 C. pp. 167. For sheetmetal forming operations like bending.g.
difficult to They are measure/ and therefore they represent one forming operations. a value of / = 0. steel is between the work and the tools gives rise to shearing The relationship between the shearing stress r. A value of / = 0. Iron Steel Inst. 0. Some typical values^ are / = 0.10 is typical for the cold rolling of most metals with polished rolls. 65.4 is found for rolling steel at a temperature between 700 and 1650°F. Friction forces can materially increase the deformation resistance. T. cit. 2 can be made with a pressuresensitive pin van Rooyen and W. but on rolling above 2000°F the coefficient of order of 0. 235. vol. op. the efficiency of the lubricant. see G. and ' A sensitive measurement of friction inserted at an angle to the interface.05 for the cold rolling of mild steel with flood lubrication and/ = 0. (London). p. the ability to find a of steel suitable lubricant often determines whether or not a forming operation will be successful.05. the speed deformation. . and the temperature. The lowest coefficients of friction are of the excellent lubrication These values occur under conditions of slow speed where the tool surfaces are highly polished. Friction increases with an increase the relative motion between the work and the tools. copper. J. Hoffman and Sachs. 186. Backofen. 1957.05 to 0.15 for cold drawing and deep drawing of steel..470 78. tainties in the analysis of major uncerVarious methods of lubriof the cation are used to minimize friction forces. and the coefficient of friction / is generally expressed by Coulomb's law of sliding friction friction The stresses along the contact surfaces. l=f The value ness of the of in it (1719) of the coefficient of friction will depend upon the material dies. the normal stress on the interface between the work and the tools. but if the roll surface is only a ground finish. Plastic Formins of Metals [Chap. A. In fact. but at high speeds decreases appreciably. 17 1 Friction in Forming Operations An important forces developed consideration in the forming of metals is the friction between the workpiece and the forming tools. being worked. the hot extrusion was not comnot possible mercially possible until molten glass was used as a lubricant in the UgineSejournet process. the material used for the tools or the surface rough work and the tools. For example. A value of / = 0. and brass using hardened polished dies and efficient lubricants.15 is more typical. p. The coefficients of friction are usually higher for hot working because oxidation roughens the work and the tools. and the cold extrusion of without a phosphate coating for a lubricant.01 to 0.
P. 179] friction drops to 0. depending upon the application. capacitance strain gages. greater experimental information on the nature of the deformation in forming processes is needed to allow for the distribution throughout the workpiece. Measurement of the forces presents no fundamental problems. grid networks. Steel Processing and Conversion. However. and then machining to a symmetrical shape. The distortion of the lead grid is obtained by radiographing the deformed billet. Tardiff. 643644. only a limited amount of information can be obtained from grids placed on the surface of the workpiece. for and load cells. . 626632. although considerable ingenuity is required to instrument most pieces of forming equipment. Under certain circumstances the interior deformation can be determined by embedding a lead grid in the casting. vol. 650.2 of the oxide film. The grid network may be applied either by scribing or by photographic methods. An initially banded structure provides a unidirectional internal grid system. Theories of plastic forming which are based on the theory of slipline fields attempt to take the inhomogeneity of deformation into consideration. of identical A number Sometimes plugs are inserted in the workpiece. fastening them back together. General Fundamentals of Metalworking 471 because of a change in the frictional characteristics 179. is that the metal formulation of more realistic slipline fields. pp. Etches are available for most metals which selectively ' H. The direction of flow can be determined from the grain distortion and the preferred alignment of secondphase particles and inclusions. The flow in the interior of the workpiece can be studied by cutting the billet in half. However. However. affixing a grid network to the two faces. Either rectangular or polarcoordinate grids are used. 43. The gross deformation of the workpiece can be determined by means of conventional measuring instruments. the information that is really needed is the strain SR4 strain gages electric Most theories of forming assume deformed homogeneously. yet there are many indications that inhomogeneous deformation is the general situation in commercial forming processes. Most studies^ of the deformation in forming operations have employed specimens are deformed by different amounts. Metallographic techniques are useful for determining regions of heavy deformation.Sec. 1957. Experimental Techniques for Forming Analysis Experimental investigations of forming operations require techniques measuring the forming loads and the deformations. and piezogages are ordinarily used for this purpose. and their distortion is measured by sectioning and examining under the microscope. and the progress of the deformation is determined from the distortion of the network.
BIBLIOGRAPHY Burton. pp. Similar studies using paraffin wax^ have also been made. New York. Ohio." Interscience Publishers. J. Hailing. Iron Steel Inst. vol.: "Fundamentals of the Working of Metals. Use is sometimes made of plasticine models^ for studying inhomogene ous deformation in forming operations. 1 2 J. 1954. Inc. and J. 42.: New Metals Park. G. pp. the examination of the microstructure of a deformed metal after annealing will indicate the presence of nonuniform deformation. C. and only very light equipment is needed to deform the material. P. W. Phil. 1940. Sachs. "Applied Metallurgy for Engineers.. S. works can be readily placed in the interior of plasticine models. and K. 188. 321331. 1951. 1956." American Society for Metals. Mag. Since recrystallization will begin first in the most heavily deformed grains. Green. . York. vol. 17 attack the plastically deformed regions. R. Bodsworth. . Barton.. Deformation in plasticine is similar to deformation in an ideal plastic material therefore plasticine is Grid "netuseful for checking theoretical analyses of plastic forming. (London). Inc. M. A.. 365373.472 Plastic Forming of Metals [Chap. Van Horn: "Practical Metallurgy. 1958." McGrawHill Book Company.
the cylinder flows to a lesser extent at these interfaces than it does at the center. 181). delivers rapid impact blows to the surface of the metal. With impact forging the pressure is at maximum intensity when the hammer touches the metal. while the forging press subjects the metal to a slowspeed compressive force. because there is always friction between the dies the upsetting of a and the metal. and it decreases rapidly in intensity as the energy of the blow is absorbed in deforming the metal. 1816). press forging results in deeper penetration of the deformed zone. The resulting shape is therefore a cylinder which is barreled at the center (Fig. primarily in the surface layers. so that ideally the final shape would be a cylinder of increased diameter and decreased height (Fig. many This forging processes open dies of simple geometrical shape are is true for very large objects or where the number of items produced is small. 473 . 18lc). Today there is a wide variety of forging machinery which is capable of making parts ranging in size from a bolt to a turbine rotor or an entire airplane wing. Therefore. The simplest forging operation is cylinder between two flat dies (Fig. although certain metals may be coldforged. The compressive forces cause the metal to flow out equally in all directions. Classification of Forging Processes Forging pressing. ery to replace the arm of the smith occurred early during the Industrial Revolution. Two major classes of equipment are used for forging operations. Most forging operations are carried out hot. or drop hammer. In used. as the metal is In press forging the pressure increases being deformed. The forging hammer. and its maximum value is obtained just before the pressure is released. However. Therefore. impact forging produces deformation It is the oldest of the metalworking arts.Chapter 18 FORGING 181. This illustrates the general rule that a metal will flow most easily toward the nearest free surface. is the working of metal into a useful shape by hammering or having its origin with The development of machinthe primitive blacksmith of Biblical times.
182c). Examples of (6) ideal flow in upsetting. 18ld).474 Plastic Forming of Metals [Chap. 18le). An example of the use of this type of operation would be in the forging of a connecting rod for in cross section of the an internalcombustion engine. extrusion. Large production runs . punching (Fig. it is Other operations which can be achieved by forging are bending. 1 81 . 182d). rectangular section after upsetting. la) g Pi {b) Fig. Closeddie forging uses carefully machined matching die blocks to pro duce forgings to close dimensional tolerances. (c) barreling due to friction between dies and work. 182a and h. produce a bar of smaller diameter. (a) Cylinder ready to upset. area of a portion of the stock. The reduction is work with concurrent ing out (Fig. (d) (e) thin rectangular section before upsetting. The effect of friction in restraining metal flow is used to produce shapes with simple dies. (Fig. from the center of the fullering die (Fig. If the workpiece has a noncylindrishape (Fig. The metal flow is outward and away V/////A y. the metal is confined by the die from flowing in the horizontal direction but it is free to flow Fullering is used to reduce the crosssectional laterally to fill the die. 18 cal which is the point of least resistance. As is shown in Fig. or is draw the drawingdown operation carried out with concave dies called swaging. Edging dies are used to shape the ends of bars and to gather metal. metal flow during upset forging. 182e) so as to and indenting. 182/). the greatest flow will occur along the narrowest sides (Fig. increase in length If called drawing down. 182gr). twisting. piercing (Fig.
It is important to use enough metal in the forging blank so that the it is die cavity amount completely filled. punching. The greatest change in the shape of the metal usuIt is ally occurs in this step. then transferred to the finishing die. (e) swaging. it is customary to use a slight excess of metal. In closeddie forgto place the metal billet is ing the forging billet is usually first fullered and edged in the correct places for subsequent forging. When the dies come together is of . 1 82.Sec. P^ullering and edging impressions are often placed on the edges of the die block. 181] Forging 475 are generally required to justify the expensive dies. (/) piercing. (a. b) Edging. and the finishing cavity are machined into the same die block. The preshaped then placed in the cavity of the blocking die and roughforged to close to the final shape. (g) Forging operations. (c) fullering. Because it is difficult to put just the right metal in the correct places during fullering and edging. Usually the blocking cavity forged to final shape and dimensions. (d) drawing. where Fig.
The two basic types of forging hammers are the hoard hammer and the steam hammer (Fig.000 lb. Sectional view through the use multiple die cavities in closeddie forging. Upper die In order to prevent the formation of a very wide is flash. Forging Equipment In forging hammers the force is supplied by a falling weight. a single press hammer for drop forgings.2 ft/(sec)(sec) Striking velocities in excess of 25 ft/sec feature of the steam hammer is may be obtained. lb velocity. Since the falling ram is accelerated by steam pressure. known as a flash gutter. it falls owing to gravity. forging 182. 18 for the fini.500lb hammers with a 75in. a ridge. while in drop forging multiple blows are used for each step. the excess metal squirts out of the cavity as a thin ribbon of metal called flash.shing step. or ram. Board hammers are rated by the weight of the falling part of the equipment. and it enters the top to drive the ram down.000 to 50. The energy supplied to the blow is equal to the potential energy due to the weight of the ram and the height of the fall. W where = ^^^^ (181) ^9 M V g = = = falling weight. Steam is admitted to the bottom of a cylinder to raise the ram. ft/sec acceleration of gravity. made on With a press are called press forgings each press forgings. 32. When the ram is released. in contrast to 183. This type of equipment is capable of producing forgings weighing up to about 100 lb. fall. is usually accomplished on a sepaof rate die in separate presses. usually provided (Fig. 184). Closeddie forgings made on a forging hammer are usually called drop forgings. fall to 7.476 Plastic Forming of Metals [Chap. 183). In the board hammer the upper die and ram are raised by rolls gripping the board. is available with the steam hammer (Fig. of the flash while those step Lower die Fig. They range from 400lb hammers with a 35in. the energy supplied to the blow is related to the kinetic energy of the falling mass. 184). in the range from 1. Greater forging capacity. The final step in making a closeddie forging is the removal with a trimming die. An important that the striking force can be readily . In each sequence is usually performed by a single application of pressure.
are rated on the basis of the estimated force or load developed at the . J Board ^/777. y////////////////r7. uses two horizontally opposed rams. Bottom stroke 7777777777777777777777 Hydraulic press Crank press Fig. 182] Forging 477 hammer the mass and height of fall are However. whereas in the board constant. 7ZZZZZZZZZZZ1 Up ^~L upper die Lower die ~L^ 777777777777777777777777> Steam hammer 777777777777777777777777Z. •////////.Sec. Board hammer Intensifier ^Q^ 'V777777777777Z' Top stroke 7/7777777777777. velocity. Presses Forging presses are of either mechanical or hydraulic design. Down Roll d Piston IV////////. Each ram is strikes the work at high and practically all their energy absorbed by the work. where the ram is raised and lowered with compressed air. Schematic drawings of forging equipment. 184. controlled. it is possible to conA unique type of pneumatic forging hammer trol the force of each blow. in the modern version of the board hammer.
although several presses with ratings of 50. and gears are typical parts ing tools which upset the metal. slender sections. he inserts the work. subjects a bar or tube to a series of blows from two dies which rotate around the stock so that it is ham mered from Hammer Fig. 1953. pp. by hammers. D. 18 bottom of the stroke. 184). hardened steel rollers. force. located on top of the press (Fig. Most mechanical presses operate from an eccentric crank (Fig. The remaining portion is reduced in diameter so that the forged part can move freely between the rolls. it is reduced and Different portions of the work can be sucejected toward the operator. Forging machines are rated in terms of the maximumdiameter bar which can be gripped by the machine. Hydraulic presses are generally built to ratings of 500 to 18. the work 14931512. with dies which firmly grip the work around the circumference and formBolts.000 tons have been built. When mers contact the spindle rotates the fly ' the dies are forced together. spring leaves are often ing Long and made on forg Spindle rolls. ^ Large hydraulic presses are particularly adaptable to the opendie forging of steel ingots and the closeddie forging of aluminum and magnesium. . and when the clearance space appears.478 Plastic Forming of Metals [Chap. Stone. High pressures are built up in the intensifier cylinders by using oil or water as the hydraulic medium. 75. Two are dies. backed up in Swaging machine. is subjected to ASMS. Thus. also known as upsetters. rivets. Cage Rotary swaging. spindle. 184). hambut when the the hammers to a position between the rollers. illustrates the operation of the Figure 185 swag ing machine.000 tons are commercially availVertical hydraulic presses usually have the pressure chamber able. The operator stands at the rear of the rolls.000 tons. 185. or rotary forging. the dies open owing to centrifugal M. all sides. When the reducing portion of the rolls comes in contact with the work. or headers. are horizontal presses which are very useful for the highproduction forging of symmetri The machine is basically a doubleacting press cal shapes from bar stock. Trans. The of rotated with the rotates spindle a cage containing a number rollers. Forging machines. Loads of 100 to 7. cessively fed into the rolls or fed through Roller other grooves to produce bolts increased reductions. vol. The rolls use only part of their circumference to reduce the metal. Forging rolls are used for initial forming prior to closeddie forging and for producing tapered sections or long. made with forging machines.
vol. reason small swaging machines have laboratory method of decreasing the and for attaching sleeves to cable ends. friction is present between the dies and the specimen. 1945. Under these conditions the compressive It When force required to cause yielding 1 is no longer given by the simple relation M. for pointing rod prior to wire drawing. C.. 183. J. 71. in actuality a compression experiment is usually complicated by the presence of friction between the dies and the specimen. pp. Metals.Sec. The shaded areas in Fig. The specimen deforms in the inhomogeneous manner shown in Fig. the uniaxial compressive force required to produce yielding is of a flat plate or a cylinder The compression represents the simplest type of forging operation. 186. or natural. when the dies are very well lubricated. the metal adjacent to these regions undergoes little or no deformation. 371390.) ('«^) The true. Deformation in Compression between two flat dies Although at first glance this appears to be a simple type of experiment which is subject to easy analysis.e. entirely compressive. strain in ^ is given by = h Inr n ho = —\nr ho h no K\ (185) can be readily shown that e = In (e + 1). 183] Forsing 479 several thousand blows per minute. 186 represent regions of little deformation owing to the presence of frictional stresses at the die interface. P. The applied stresses are almost For this found considerable use as a diameter of relatively brittle materials. In the absence of friction. Inst. . and the specimen assumes a barreled shape. Larke. = (TO A P is (182) The compressive stress p induced by a uniaxial force given by^ where h = ho = Do = height of cylindrical sample at any instant during compression original height of cylinder original diameter of cylinder The engineering strain in compression is Ah ho _ h — h ~Kr{'compression dh rho F. On a commercial basis swaging is used for reducing the diameter of rod and tubing. Cook and E. i.
'' ' Ihid. Z. Mech. it has The true flow stress or V7/. Solids. 132144. with the same diameter and equal For the two cylinders shown in frictional conditions but different heights.^ planestrain compression of a plate between platens was one of the problems attacked by the method of the slipline field. 226236. M. J. A. H. pp. 1948. H." pp. 228243. 163. Mech. Math." pp. and J. 186. vol. pp. Green. Solids. different diameter. (London). 1923. /. Metals Park. in "Fracturing of Metals. The first Alexander. which was later extended by Hill. 186) as well as the frictional conditions at the interface. P. pp. "The Mathematical Theory of Plasticity. 2 T. Polakowski. 1955. Fig. Ohio. Read. PrandtP made the first analysis. Phil. 186. 1951. Mech. pp. Oxford Uni versity Press. 3 L. 42. angew. W. Mag.^ Green. J. vol. J. the shorter cylinder will require a greater axial force to produce the same percentage reduction in height because of the relatively Similarly. Undeformed regions (shaded) due to friction at ends of a compression specimen. Prandtl.. and Phys. 6 6 > New York. Markus. ^R. 250 276. The force required to produce deformation will be a function of the dimensions of the specimen (Sec. see also N.^ and Bishop. /. vol. Hill. vol.. has also been determined. 18 ship of Eq. 900918. Bishop. McCaughey. A correction factor to apply to the measured uniaxial pressure to correct for barreling. A. 6. u. for cylinders of equal height but larger undeformed region. 3. M.. and Phys. . 1958. American Society for Metals.^ 480 Plastic Forming of Metals [Chap. 233245. 1949.^ Quite reliable approximate methods of calculating the yield stress in compression with friction present will be given in the next two sections. TUexander. vol. Fig. 1950. 3. F. the largerdiameter cylinder will require a higher aver age axial pressure to produce the same reduction in height because a greater surface area in contact with the dies. Iron Steel Inst. yield stress in compression can be found by extrapolating the measured average stress for a number of specimens of different geometry to a diameterheight ratio of zero. 401406. pp. similar to the necking correction for the tension test. (182).
Forsing in Plane Strain with Coulomb Friction The differential equation for the forging of a plate of constant thick ness under conditions of plane strain will be developed with the aid of This figure shows the stresses acting on an element of a flat The two principal stresses plate which is being compressed in open dies. Cx — Cy = 2 7= (To = (t'o = Cx + P (187) By differentiating both sides of Eq. one finds that d<Tx = —dp and rearranging. Taking the equilibrium {<jx of forces in the x direction results in dxTx + — <Tx)h h d(Tx — — 2txu 2tx„ dx dx = = (186) The distortionenergy criterion of yielding for a condition of plane strain (360). 1 87. are the compressive axial pressure on the dies. (188^ dp + ~^dx = . P * I ] Try CFr + day ox dx ttt P Fis. It is assumed that the plate has unit width normal to the plane of the paper and that this width remains constant so that the analysis can be based on the twodimensional planestrain condition. Stresses acting on a plate forged in plane strain. metal flow parallel to the dies. (187). The shearing stresses Txy act at the diemetal interfaces as a result of friction. was given by Eq. which is the stress required to make the Fig.184] Forgin* 481 184. 187. and the longitudinal compressive stress ctx. the general differ and upon substituting into Eq. (186) is ential equation of equilibrium obtained. c^ = —p.
as the ratio of length to thickness. the resistance to compressive deformation increases rapidly.482 If it is Plastic Forming of Metals stress is [Chap. since P = 2pe. This fact used to advantage in closeddie forging. where w is the width in the direction normal to the plane of the paper. Txy = fv. by Eq. Figure 188 shows the pressure distribution and the variation of longiis resistance of the flash . Equation (1813) shows that. P 1 + l^a (a x) (1813) — ^ h fj V The total forging load — . Therefore. { we can make • use of the expansion exp y = y \ y^/2\ + y^/3\ + • • to simplify the above equations. x) (1814) dx 'o exp (2/a//i)  1 (1815) 2/a//i P can then be established. x/h. (188) dv ^"^ dx h = Q Separating the variables results in 2? h (189) Integrating both sides gives In p = _ = C ^+ h In (7 or p integration constant exp 2fx h (1810) The C is evaluated by the boundary condition that at a free surface x = a the longitudinal stress a^ In = 0. 18 assumed that the shearing by Coulomb's law of sliding becomes sure related to the normal pres friction. where the deformation must be very high so that the pressure in the die will be high enough to ensure complete filling of the die cavity.yaw. (187) p = a[ and C = In a[ + 2fa/h. increases. p = ao exp 1 2/(a h x) (1811) — exp 2/(a h x) (1812) Since / is 1 usually a small number. then Eq.
and E. . Trans.provides a better estimate of the forging. center flow. ASM. pp. vol. ing friction at the diemetal interface the shearing stress at the interface is constant and equal to the shearing strength of the metal in plane strain. Thomsen. Forging in Plane Strain with Sticking Friction In the previous section we assumed that the shearing stresses at the diemetal interfaces were related to the normal pressure on the interface by Coulomb's law of sliding friction. Another assumption that could be made is that the die surfaces are so rough that the metal adheres perfectly This is the limiting case known as sticking friction. R. The equations described above are strictly applicable only to open While these equahave been applied with fair accuracy to the prediction of the die forgings. Trans. pp. S. a'fj/2 = cro/\/3. G. G. Trans. L. R. Lapsley. 709728. B. pp. increases rapidly with distance from the free surface. Thomsen. after integration. Substituting this value for r^^ in Eq. compression pressure required for closeddie analysis. 81. 188. 1 85. 1 a more detailed forging pressure for this situation. ser. 2 S. B. vol. The pressure at the also builds up of to a maximum plate. 38. 1947. vol. For stickto them. Kobayashi. Ansel. Forging of a Cylinder in Plane Strain thin cylinder of radius a that ' Figure 189 shows the stresses acting on a volume element cut from a The equiis compressed in the z direction. 1959. 82. T. with the maxipressure at the center of the plate. mum 186. the Note that the longitudinal resistance to ffx. MacDonald. (188) gives.Sec. 246252. Herzog. G. A. and E. Distribution of normal stress stress for tions and longitudinal between plates. Dietrich and G. 228238. 1 86] "orging 483 tudinal compressive stress over a plate of length 2a. ASME. ser. Kobayashi. the following relation: P 0+^^) (1816) pressure distribution over a plate of length 2a Equation (1816) predicts that for a situation of sticking friction the is linear. Fig. J. 1960. ASME.
/. 289294. Webster. Forsrng Defects If the deformation during forging light. solution Fig. Mech. By using specimens of different a/h ratio it is possible to determine the coefficient of friction and the flow stress.9.484 Plastic Forming of Metals in [Chap.2ja/h 2{ja/hy  1 Equation (1820) again demonstrates the fact that higher forming when the height of the body is small compared with its transverse dimensions.f'n dr 2fP h (1817) Since the strains in the radial and transverse directions are equal. . vol. Also. is limited to the surface layers. 18. (1817) yields dp 2/ dr P This is h (1819) the same as Eq.. op. cit. pp. the dendritic ingot structure broken down at the interior of the forging. p = ~2/ ao exp h (a — r) (1820) The average pressure for a cylinder loaded in compression is given by' (1821) exp Pa OQ (2/a//i) . 2 W. 18 Jibrium of forces arV the radial direction gives dd h  ((T. App/. as when will not be hammer blows are used. (189) for Therefore. Substituting these three principal stresses into the equation for the distortionenergy yielding criterion results in P a> + dcr^ (Xr — Co (1818) Differentiating and substituting into Eq. Incomplete forgrapid ing penetration can be readily detected ^ by macroetching a cross section Stone. Studies of this type and analytical expressions for the deformation pressure of thin cylinders under different conditions of die friction have been pressures are required reported. A. 2 187. Schroeder and D. <Tz = p. 1949. the is the rectangular slab. Stresses acting on element of cylinder. o^ = at. + d(Tr){r + dr) d9 h  2fpr dr d9 + ath de dr = which reduces to da/T_ — r rTt (Tt '?. 16.
In the upsetting of bar stock on a forging machine certain precautions must be taken to prevent buckling of the bar. J. ing is more prevalent the thinner the flash in relation to the original thickness of the metal. (a) Cracking at the flash. 18105). (h) cold shut or fold. (c) 1810. Surface cracking can occur as a result of excessive working of the surface at too low a temperature or as a result of hot shortness. pp. 1939. atmosphere can produce hot shortnesfe Cracking at the flash of closeddie forgings is another surface defect.187] of the forging. Typical forging defects. is A cold shut a discontinuity produced when two surfaces of metal fold against each other without welding completely. 18lOa). forgings of dendritic structure. a standard qualitycontrol procedure with large forgings. Inst. made on a forging press. Metals. since the crack generally penetrates into the forging when the flash is trimmed off (Fig. 261283. internal cracking due to secondary tensile stresses. . For upsetting in a single cold shut can occur is operation the unsupported length should be no greater than two to three 1 G. large cross section are usually To minimize incomplete penetration. Forging 485 The examination and cracks is of a deep etch disk for segregation. in closeddie forgings is the cold shut. sulfur concentration in the furnace in steel A high and nickel. vol.^ Flash cracking can be avoided by increasing the flash thickness or by relocating the flash to a less critical region of the Another common surface defect or fold (Fig. body of the This type of crack le) Fi3. Sachs. 64. forging. A ation when a flash or fin produced by one forging operpressed into the metal surface during a subsec^uent operation.
Flaking is associated with the high hydrogen content usually present in steel ingots of large stresses. cylinder or a round (Fig. Ohio. 18 General rules for the optimum dimentimes the diameter of the stock. 1 88. size. Sec. However. upsetting sions for ^ Secondary tensile stresses thus be produced. In order to minimize bulging during upsetting and the of cracking. or flakes. properties in the direction normal to it (transverse direction) To achieve and transamount of an optimum balance between the ductility verse directions of a forging. . the fiber structure results in and is However. it is usual practice to use concave dies." o. in forging machines have been developed. Internal cracking is less prevalent in closeddie forging because lateral compressive stresses are developed by the reaction of the work with the die wall. The deformation produced by forging results in a certain degree of directionality to the microstructure in which second phases and inclusions are oriented parallel to the direction of greatest deformation. appreciable residual stresses and warping can occur on the quenching of steel forgings in heat treatment (Sec. can minimize this type stresses. or fiber structure. Special precautions must be observed during the cooling of large steel forgings from the hotworking temperature. coupled with the presence of residual In order to guard against the development of high thermal or transformation residual stresses. 1948. The existence of a fiber structure characteristic of all forgings not to be considered as a forging defect. can develop during forging. Residual Stresses in Forginss The residual stresses produced in forgings as a result of inhomogeneous deformation are generally quite small because the deformation is usually carried out well into the hotworking region. iO. at the center of the cross section. in the controlled cool used for hotrolled railroad rail and certain forgings. as was discussed in lower tensile ductility and fatigue . 1 "Metals Handbook. 156). large forgings are very slowly cooled from the working temperature. however. Metals Park. as a result of the circumferential tensile Proper design of the dies. Large forgings are subject to the formation of small cracks. 18lOc). it is in the longitudinal often necessary to limit the deformation to 50 to 70 per cent reduction in cross section. 911. this appears as flow is lines.486 Plastic Forming of Metals [Chap. This may be accomplished by burying the forging in ashes for periods ing treatment which is up to several weeks or. development of circumferential tensile stresses. and cracking can Internal cracks can develop during the upsetting of a When viewed at low magnification. by transferring the hot forgin? to an automatically controlled cooling cycle which brings the forging to a safe temperature in a number of hours. American Society for Metals.
1953. 1954.: "The Closed Die Forging Process. and G. P. 1940. New York.Forging 487 BIBLIOGRAPHY O. Van Horn: "Practical Metallurgy." The Macmillan Company." chap. New York. Hoffman." American Society for Metals. . 21.. R." American Society for Metals.. Sachs. and K. 1939. Metals Park. E. Metals Park. Naujoks. Inc. and D. Sachs: "Introduction to the Theory of Plasticity for Engineers. McGrawHill Book Company.. Ohio. C.. Kyle. Ohio. Fabel: "Forging Handbook. G. W.
billets. ClassiFication of Rollins Processes is The process of plastically deforming metal by passing it between rolls known as rolling. or extrusion. as an extrusion billet. nology a billet is any ingot which has received hot working by rolling. forging. A slab refers to a hotrolled ingot with a crosssectional area greater than 16 in. of ingots into is blooms and done by hot This followed by further hot rolling into plate.^ A further reduction by hot rolling results in a billet. Blooms. rails. and sharp limits with respect to dimensions cannot always be made for steelmaking terminology. into the The rolls. and slabs are known as semifinished ^products because they are subsequently formed into other mill products. rod. frictional forces are also responsible for drawing the metal billets is generally The initial breakdown rolling. bar. and the crosssectional area is greater than 36 in. plate has a thickness greater . A bloom is the product of the first breakdown of the ingot. Cold roll and foil with good surface finish and increased mechanical strength. lends itself to high production and close control of the rolls. In general.^ and with a width that is at least twice the thickness.Chapter 19 ROLLING OF METALS 191. strip. sheet. at the same time maintaining close control over the dimensions of the product. The differentiation between plate and sheet is determined by the thickness 488 of the product. final In deforming metal between the work is subjected to high compressive stresses from the squeezing action of the rolls and to surface shear stresses as a result of the friction between the rolls and the metal. The minimum cross section of a billet is about It should be noted that in nonferrous metallurgical termi1/^ by 13^ in. Generally the width of a bloom equals its thickness. This is the most widely used metalworking process it because product. or structural shapes. The cold rolling of metals has reached a position of major importance in industry. ing produces sheet. or the term may refer to a casting which is suitable for hot working. pipe. The terminology used to describe rolled products is fairly loose.
19la). Therefore. . Rollins Equipment A rolling mill consists basically of rolls. Powder rolling is still and produces a sheet with no preferred orienin its infancy. of the rolls for further reduction by hand carrying or by means An a platform which can be raised to pass the work above the rolls. tation. When these requirements are multiplied several times for the successive stands of a large continuous mill. bearings. 192] 3'4 in. but it appears to have defiis nite advantages for reactive metals. process are that the elimination of hot working minimizes contamination. obvious improvement in speed results from the use of a twohigh reversing 1 Storchheim. Rolls of equal size are rotated only in one direction. 70. The stock is returned to the entrance. 191)... Rolling oF Metals 489 although there are exceptions to this Hmit. The advantages claimed^ for this requirement. and a drive for applying power to the rolls and controlling their speed. vol. but this is not a necessary A recent development is powder rolling." which is subsequently sintered to high density. Ordinarily little to decrease increase in width occurs. Melal Progr. Another specialized use of rolling is thread rolling. in which a blank is fed between two grooved die plates to form the threads. is number The simplest and most common type of rolling mill or of the twohigh mill (Fig.. pp. Roll so that the decrease in thickness results in an increase in length. The thickness of the metal is not appreciably changed during this process. Metal powder is introduced between the rolls and compacted into a "green strip. 120126. very rigid construction needed. results in a fine grain size. Generally rolling starts with a cast ingot. greater width. S. Rolling mills can be conveniently classified with respect to the of the rolls (Fig. forming is a special type of cold rolling in which strip is progressively bent into complex shapes by passing it through a series of driven rolls. 1956. molded sections such as irregularshaped channels and trim. The forces involved in rolling can easily reach is many millions of pounds. depending upon Sheet and strip refer to rolled products which generally have In general. strip refers to the rolled product a thickness less than yi in. and very large motors are required to provide the necessary power. In conventional hot or cold rolling the main objective the thickness of the metal. with a width no greater than 12 in. a housing for con taining these parts. rear. Roll forming is particularly suited to producing long. while sheet refers to the product of than the width. 192.)ec. it is easy to see why a modern rollingmill installation demands many millions of dollars of capital investment and many manhours and arrangement of skilled engineering design and construction.
pullover. reversing. (Fig. 192). in rolls. consisting of an upper and lower driven roll and a middle roll which rotates by friction. the strip will . The simplest mill of this type is the by reversing M* ia) [d] ic) id) Fig. The cluster mill which each of the work rolls is supported by two backing which is a typical mill of this kind. is tolerances on a mill with smalldiameter Very thin sheet can be rolled to very close work rolls. 19ld). 19 which the work can be passed back and forth through the rolls their direction of rotation (Fig. twohigh.490 mill. Another solution is the threehigh mill (Fig. have less strength and rigidity than large rolls. For high production it is common to install a series of rolling mills one after another in tandem (Fig. because smalldiameter rolls the use of smalldiameter rolls. (d) fourhigh. (e) cluster. Each set of rolls is called a stand. Typical arrangements of rolls for rolling mills. Twohigh. of the cluster mill The Sendzimir mill is a modification very well adapted to rolling thin sheet or foil from highstrength alloys. Since a different reduction is taken at each stand. fourhigh mill (Fig. (c) threehigh. 1916). 19le). (b) ie) (a) 191. they must be supported by largerdiameter backup rolls. in Plastic Forming of Metalj [Chap. A large decrease in the power required for rolling can be achieved by However. 19lc).
special shapes. called the Steckel mill. Schematic drawing of strip rolling on a fourstand continuous mill. threehigh. 193. 1 Stand 2 Stand 3 Stand 4 Windup reel 192. it is often necessary to reroll billets the blooms on a threehigh. 193] Rollins of Metals 491 be moving at different velocities at each stage in the mill. Hot first Rolling hot working operation for most steel products is The done in the blooming mill (also called cogging mill).to 54in. Because blooming represents the initial breakdown of the ingot structure. This type of mill consists of two rolling mills. may then be rolled into round bars. and the rolls are not driven at the stock to the and Stand Uncoiler Fig. While the amount of reduction per pass that is can be accomplished uses small by means it of the Steckel mill limited. billet mill. all. or continuous. hexagons. one with two horizontal rolls and the other with four vertical rolls so arranged that they can be used for side rolling. The other on a universal mill. the later. force is applied through the power reels. These added horizontal forces have several advantages that will be discussed In a special type of reversible mill. because it work rolls can reduce hard metals to thin gages with close tolerances on thickness. careful steps with repeated reheating. it is done in small. or flats on various types of finishing mills. The speed of each set of rolls is synchronized so that each successive stand takes the The strip at a speed equal to the delivery speed of the preceding stand. and fourhigh mills and then shearing all edges to size. It is not unusual to require 25 passes for the blooming of a large alloysteel ingot. uncoiler and windup rolls reel not only accomplish the functions of feeding coiling up the final product but also can be used to supply a back tension and a front tension to the strip.Sec. Sheared plate is produced by crossrolling billets on twohigh. Blooming mills are usually two high reversing mills with 24. ^ universal mill can take an ingot and roll it directly into plate with of rolling plate is common method . of billet required To produce The the size by smaller finishing mills. diameter rolls.
Fourhigh mills are used in the finishing train. The fourhigh reducing stands usually are equipped with vertical edging rolls to control the width of the strip. the smaller ingot sizes and lower flow stresses found with nonferrous metals generally mean that smaller rolling mills can be used. initially heated to 2200°F. this type of equipment requires a large capital investment and suffers further from lack of versatility. Highspeed fourhigh tandem mills with three to five stands are used for the cold rolling of steel sheet. Generally.and threehigh mills are used for most hot rolling. However. The finishing temperature varies from 1300 to 1600°F. By this method of roUing the ingot merely elongated in the longitudinal direction. aluminum. Two. Coldrolled nonferrous sheet may be produced from hotrolled strip. Further. Highpressure water jets are usually sprayed on the strip to break up the scale. the strain hardening resulting from the cold reduction may be used to give increased strength. A continuous mill has high capacity and results in low labor costs.492 Plastic Forming of Metals [Chap. or in the case of certain copper alloys it is coldrolled directly from the cast state. depending on the thickness and the desired grain size and mechanical properties. Hotrolled steel strip has been produced on tandem continuous mills The starting material is a rolled slab which is since about 1930. The fact that the nonferrousmetals industry deals with a diverse product mix is responsible for the fact that nonferrous hotrolling equipment is generally more versatile and less specialized and mechanized than the equipment used for the hot rolling of steel. rolled example. this type of mill is designed to provide both front and back tension. plate produced with the universal mill will have cross rolling. Cold Rollins Cold rolling is used to produce sheet and strip with superior surface and dimensional tolerances compared with hotrolled strip. The continuous mill consists of a roughing train of four to seven stands and a finishing train of four to eight stands.000 ft/min. In addition. 19 no shearing. The roughing train contains twohigh scalebreaking rolls and a spreading mill for spreading the slab to the desired width. and copper alloys. A greater percentage of rolled nonferrous metals is finished by cold rolling compared with rolledsteel prodfinish The starting material for cold rolledsteel sheet is pickled hotbreakdown coil from the continuous hotstrip mill. straight edges that need is 194. For ucts. Fourhigh singlestand reversing mills with front and back tension are a more versatile . although continuous fourhigh hot mills have been installed for rolling aluminum alloys. the delivery speed of a fivestand continuous mill can reach 6. lower transverse ductility than plate produced by cross rolling. and it thus receives no Therefore.
Rolling of Metals lliis 493 type of mill is used often for the production of spe cialty iienis that varj^ widely in dimensions. Reviews. Since the metal spreads to a much greater extent in the hot rolling of bars than in cold rolling of sheet. vol. in any one pass the metal is usually compressed in one direction only. Rolling of Bars and Shapes Bars of circular or hexagonal cross section and structural shapes like beams. Actually. . 195. 1933. The elimination of the yield point from annealedsteel sheet is an important practical problem since the existence of a yieldpoint elongation results in inhomogeneous deformation (stretcher strains) during deep drawing or forming. which eliminates Temper rolling also results in an improved the yieldpoint elongation. A rolling mill designed to roll bars is known as a bar mill. Most production bar mills are equipped with guides to feed the billet into the grooves and repeaters to reverse the direction of the bar I ^ 1 W. On the next pass it is rotated 90°. The usual practice is to give the annealed steel a final.Sec. the hot breakdown of an ingot into a bloom falls in this category since grooved rolls are used to control the changes in shape during the blooming operation. "Roll Pass Design. 195J installalioii. 193). channels. A typical method of reducing a square billet to a bar is by alternate passes through oval and squareshaped grooves. Bars. Met." 2d ed. pp. The stretcher leveler consists of two jaws which grip the edges of the sheet and stretch it with a are arranged so that the top is passed into the leveler. surface and in improved flatness. Penton Publistiing Company. it is not generally possible to roll metals of widely different rolling characteristics on the same set of bar rolls. or merchant mill. Other methods which are used to increase the flatness of rolled sheet are roller leveling and stretcher leveling. 4. 1959. temper rolling. small cold reduction. Cleveland. The rolling of bars and shapes differs from the rolling of sheet and strip in that the cross section of the metal is reduced in two directions. and Light Sections. The design of roll passes for structural shapes is much more complicated and recjuires extensive experience. R. Stewartson. A rollerleveling machine consists of two sets of smalldiameter rolls which and bottom rows are offset. When the sheet it is flexed up and down and the sheet is straightened as it emerges from the rolls. However. or skin pass. 309379. However. an important problem in designing passes for bars and shapes is to provide allowance for the spreading. and railroad rails are produced in great quantity by hot rolling with grooved rolls (Fig. pure tensile force. Trinks. it cannot compete with the continuoustandem mill where large tonnages are involved.. Ttie Rolling of Rods. Because different metals spread different amounts.
i. 196. by metal sheet with a thickness /lo enters the rolls at the with a velocity vq. several mills are set close together.494 Plastic Forming of Metals [Chap. and a finishing stand.) and feed it ally either back through the next roll pass.or threehigh. so that the vertical compression of the metal is translated into an elongation in the rolling direction. Forces and Geometrical Relationships in Rolling the geometry of the Figure 194 illustrates a number of important relationships between rolls and the forces involved in deforming a metal rolling. a strand stand. A common installation consists of a rough ing stand. To a first approximation no increase in width results.e.. and the rolls in one stand are driven by connecting them to those of the adjacent stand. 19 Fig. It passes through the roll gap and leaves the exit plane YY with a reduced thickness hf. Since equal volumes of metal must pass a given point per unit entrance plane A XX . Rolling of bars and structural shapes. {Courtesy American Iron and Steel Institute. 193. side by side. Mills of this type are genertwo. It is common practice to arrange bar rolls in train.
it is frequently called the separating force. 196] Rolling of Metals 495 time. 194 so as to draw the metal into the rolls. the velocity of the sheet must steadily increase from entrance to exit. The direction of the frictional force is then reversed so that rolls. (191) requires that the exit velocity V. The roll ing load the force with which the rolls press against the metal. such as point A in Fig.and a tangential friction force F. Forces acting during rolling. Because also equal to the force exerted by the metal in trying to force the rolls apart.. Eq. we can b V write bhoVo = bhv = bhfVf (191) where = width of sheet = its velocity at any thickness h intermediate between ho and h/ In order that a vertical element in the sheet remain undistorted. 194. On the exit side of the neutral point the sheet moves faster than the roll surface. The specific roll The contact . This point is called the neutral point. equal to the velocity of the sheet. V. and the frictional force acts in the direction shown in Fig. Fig. it acts to oppose the delivery of the sheet from the The this is vertical is component of Pt is known as the rolling load P. Between the entrance plane and the neutral point the sheet is moving slower than the roll surface. pressure p is the rolling load divided by the contact area. two forces act on the metal.Sec. These are a radial force P. At any point along the surface of contact. or It is indicated in Fig. 194. must be greater than the entrance velocity Vo Therefore. At only one point along the surface of contact between the roll and the sheet is the surface velocity of the roll. noslip point. 194 by point N.
19 area between the metal and the rolls is equal to the product of the width of the sheet h and the projected length of the arc of contact. (1815) can be used to give an approximate value for the mean rolling in Fig. overcome frictional forces between roll and the sheet. The pressure rises to off. Therefore. 194) is made equal to the projected arc of contact. or angle of nip. the coefficient of friction / controls the maximum bite which the rolls can take. re in 195 represents the force required to Entrance Lengtti of contact Exit Fig.  hf)]y^ (192) Therefore. 195 pressure if the factor 2a in Eq. indicates that the neutral point is not really a line on the roll surface but an area. Therefore. pressure distribution does not come to then falls a sharp peak at the neutral point. 188). A similarity should be noted between the the pressure distribution in rolling shown and the pressure distribution for planestrain compression between plates (Fig. between the entrance plane and the center line Metal will not the coefficient of friction be drawn into the rolls rolls when tan a exceeds and the metal. The area under the curve is proportional to the rolling load. while the area under the dashed line AB represents the force required to deform the metal in plane homogeneous compression. JLn — Riho roll hf)  {ho — hf)' [R{h. as required in theoretical treatments of rolling. 19. Use is sometimes made of this analogy to simplify the analysis of the rolling forces. The area shown shaded Fig. 195. For example. Eq. the shape of the pressure distri bution is important because the location of the resultant rolling load with respect to the roll centers determines the torque and power required to produce the duction. Because of this fact. Lp. The angle a of the rolls is called the angle of contact. Lp.496 Plastic Forming of Metals [Chap. the specific pressure is given by P = bLr is (193) The distribution of roll pressure along the arc of contact indicated in Fig. (1815) (Fig. Distribution of roll pres sure along the arc of contact. The fact that the a maximum at the neutral point and which for purposes of calculation acts at the center of gravity of the pressure distribution.5. rolls are often purposely roughened to increase the coefficient of between the .
" American Society of Mechanical Engineers. Since the horizontal forces change direction at the neutral point. 325330. New York. (198) requires a trialanderror procedure. "The Rolling of Strip. and Plate.34 X 16(1 — v'^)/irE is evaluated for the roll material 10* in. Ltd.^ sm which can be expressed /3 iS = —^ a sin^ (a/2) (194) in approximate form as m hj radians (195) Plowever. If it is assumed that the arc of contact remains circular and that the radius of curvature of the roll is increased from R to R'. as a The most result of these pressures the rolls undergo elastic flattening. cit." vol. widely used solution to this problem is the analysis of Hitchcock.196] friction. 1 L. C. London. 286296. 3 E. Actually. "Roll Neck Bearings. I. Since P is a function of R' the exact solution of Eq. (3 is the angle between the neutral point and the center line /S is usually called the noslip angle.. New . pp. the noslip angle can be I expressed by — 2D ho hf ^ (197) f In the previous equations it was assumed that the roll radius remains unchanged under the high pressures developed in rolling. John Wiley & Sons. H. R. 1935. on the deformed roll radius.yton for steel rolls) and P is the rolling load based .^ of Metals. Rolling of Metals rolls 497 for the Blooming same purpose. Chapman & Hall. 1957. Underwood.. Hitchcock. Sheet. 1950. 1516. 2 J. sin a equals the horizontal projection of the arc of contact divided by the roll radius. the angle can be deter mined from an equilibrium of forces in the horizontal direction." pp..^ which assumes that the pressure distribution on the rolls will produce the same distortion as an elliptical pressure distribution. frequently have grooves cut in them The angle of the rolls. the Hertz theory for the elastic compression of two cylinders results in R' = R 1 + h{ho CP  (198) hf) where C (C = = 3. ho Therefore. see Underwood. "The Rolling Inc. pp. op. York. sni a = [Rjho R hf)f' 2{ho  hf) D — 2D (196) where D is the roll diameter. Larke.
the rolls in contact with the sheet are elastically deformed and it is easier for the rolls to deform than for the sheet to deform plastically. When this occurs. The roll diameter has an important influence on the limiting reduction possible with a rolling mill. (2) the deformation resistance of the metal. by the factor 2/\/3. the constrained is flow stress for rolling the value obtained in tension or compression The planestrain flow stress can also be obtained directly from the plane compression test discussed in Sec. Similarly. The compressive force required to deform the cylinder can increase to quite high values compared with the true flow stress when the thickness is a small fraction of the diameter. with smalldiameter rolls. . In this section the effect of these factors will be considered chiefly from the standpoint of their influence on the rolling load. Since sheet rolling multiplied is essentially a planestrain process. 175. The effect of roll diameter on rolling load is eter. Therefore. and (4) the presence of front tension and back tension. Therefore.498 197. for cylinders of constant diameter but varying thickness the average uniaxial compressive yield stress increases with decreasing height of the cylinder. Equation (192) shows that the contact area is directly proportional to Z)''^. the /)'' greater than the direct analog for the case of rolling. 19 Main Variables in Rollins The main variables which control the rolling process are (1) the roll diameter. In the previous chapter it was shown that the average uniaxial stress required to compress cylinders of equal height but varying diameter increases with the diameter of the cylinder. A similar situation exists of the rolls. (3) the friction between the rolls and the metal. However. Increasing the diameter of the rolls results in a large increase in the and frictional conditions. For a given reduction of a sheet of a certain thickness the contact area will be greater for a largediameter roll. Plastic Formins of Metals [Chap. properly stiffened against deflection by backup rolls. it is possible to produce a greater reduction before roll flattening becomes important and no further reduction in sheet thickness occurs. becomes thinner. when the sheet thickness is small compared with the contact The rolling load increases as the sheet entering the area rolls Eventually a point is reached where the deformation resistance of the sheet is greater than the roll pressure which can be applied and no further reduction in thickness can be achieved. for a given pressure required rolling load for given reduction to deform the metal the total rolling load will increase with the roll diamrolling load increases with roll diameter at a rate because of the requirement to overcome the greater frictional forces acting over the larger area of contact. Both the rolling load and the length of the arc of contact decrease for smallerdiameter rolls.
. Proc. On the basis of this assumption. Sj = ^^^^ (1910) where Vf = V = velocity of metal leaving rolls surface velocity of rolls slip is related to The forward the contact angle and the coefficient of High Rates of Strain. Ford. p.P. .^ If back tension is gradually applied to the strip until the neutral point is moved For to the exit of the rolls. very low values of friction. Conf. as has been emphasized earlier. On the other hand. Because the roll pressure distribution is so strongly dependent on the friction. Proc. a constant roll speed and constant reduction. 195 shows that the larger the frictional forces the greater must be the rolling load and the more steeply the pressure builds up toward a maximum value at the neutral point. 1055. extremely difficult to measure this pressure distritheories of rolling are forced to assume a constant coefficient it is of friction. 186. p. London. changes in much on the strain rate However. vol. such as are obtained in cold rolling with lubrication and polished rolls. pp. pressure. J. rolls. Not only does the frictional force pull the impormetal into but it also affects the magnitude and distribution of the roll Reference to Fig. 8697.Sec. Whitton and H. T. 235244. 2 G. 1957. a pressure distribution like that of Fig. {London). ^ all from point to point along the contact arc of the roll. vol. 195 is commonly called the High friction results in high rolling loads and also produces friction hill. 3 W. may lead to difficulties in feeding the metal into the The friction varies However. Mech. Van Rooyen and W. M. Engrs. because bution. on the Properties of Materials at Institution of Mechanical Engineers. rolls. Backofen. 197] Rollins of Metals 499 The flow stress for cold rolling does not depend or roll speed. Inst. pp. the coefficient of friction / can be calculated from the total rolling load P and the torque Mr by (199) f = Mr jr^ is Another way of the of measuring friction in rolling through the determination forward slip Sf. 1957. greater lateral spread and edge cracking. 123. Iron Steel Inst. A. 169. then the friction acts in one direction only.. Cook. coefficients of friction can be obtained from measured values of rolling load and torque. The the friction between the roll and the metal surface is of great tance in rolling. in hot rolling the strain rate can produce significant changes in the flow ^ stress of the metal. (London).
it can be seen that thinnergage sheet is obtainable with cold rolling. Hessenberg and R. C. The reduction in roll pressure due Another important to strip tension decreases the wear of the rolls. With This is increasing rolling speed the thickness of the directly attributable to the decrease in fric sheet is decreased. 3 W. Iron Steel Inst.70. cients for hot rolling of steel can Since the friction coeffi vary from 0. 155 164. vol. ser. Arthur. tion coefficient with increasing rolling speed. 2 Ibid. F. D. 1957. 172. Sims and D. The presence readily of front materially reduce the rolling load. becomes apparent that the deformation resistance ad is reached at a lower value of ai when 03 is present. 168. 681686. vol.^ 500 friction Plastic Formins of Metals relation^ [Chap. J. the and 03.12 are more typical. . is roll pressure.. 81.03 to 0. while for cold rolling values of 0. Sims. are opposite in sign. / = tan a^ax (1912) Values of / obtained by this method are somewhat higher than those obtained by other methods. vol. The minimumthickness sheet that can be rolled on a given mill is directly related to the coefficient of friction. 285 295. The thickness of the sheet produced on a coldrolling mill can also be changed appreciably by altering the rolling speed. ASMS. Pt = P (l''^^^^) (1913) 1 M. 19 by the where r = (/lo — hf)/ho is the reduction. F. the horizontal tension stress. pp. 1951. Since ai. pp. Trans. /. An average vahie of the fric tion also can be obtained by determining the maximum contact angle for which the metal will just enter the rolls. 1952. Iron Steel Inst. (London). R.20 to 0. D. A study ^ of the effect of tension on rolling has shown that back tension is about twice as effective in reducing the rolling load as front tension. and back tension in the plane of the sheet can That this should be so can be seen by assuming that the deformation resistance of the metal is governed by a maximumshearstress law ci — 03 = a'o. Pt. can be calculated from the following equation. pp. (London). B. Stone. advantage of strip tension is that it improves the flatness and thickness uniformity across the width of a sheet. B. it The rolling load when tension applied.
which produces surface defects. Appl. » A. the faster if than the metal and they is over the surface. Nadai. Effect of strip tension on the distribution of pressure. the neutral point will eventually reach the are When slide this happens. While the lateral spread is usually of little importance in rolling sheet and strip. Mech. The amount of lateral spread depends on such factors the width. of contact roll 196. in the rolling of bars and shapes this can result in the formation of flash. schematically in Fig. pp. A54A62. the addition of both front and back tension materially reduces the area under the curve. the neutral point will roll 198. . 6. On the other move toward only front tension entrance. vol. although there is little shift If only back tension is applied. moving hand. the neutral point of the neutral point. used. the reduction in transferred into an increase in length with little increase in Thus. A/o strip tension Front and back tension Back tension only Entrance Exit Length Fi9. 1939.Sec. there is good justification for the use of a planestrain model in the mathematical analysis of rolling. If a high enough back tension roll exit. J. 198] Rolling of Metals rolling load for 501 where P = 06 same reduction but without front or back tension o/ d[ = back tension = front tension = mean value of — = noslip angle planestrain yield stress a j8 angle of contact Nadai' developed a theory of rolling which permits the calculation of As shown the effect of strip tension on the roll pressure distribution. Deformation in Rolling dimensional. is applied.. moves toward the the rolls roll exit. thickness is The deformation produced by rolling can be considered to be twoTo a good approximation in sheet rolling. 196.
1950. J. Trans. practically all investigations of the work by deformation in rolling have been based on surface strain measurements. MacGregor and Coffin^ showed that. According to Trinks. 188. longitudinal stringers of nonmetallic incluin steels are related to melting sions or pearlite practices. that the surface measurements are reliable indications of the interior deformation. 10. Defects in Rolled Products Defects in rolled metal products can result from defects introduced during the ingot stage of production or during rolling. ' W. Distortion of squaregrid netrolling. Appl. pp. and the amount of reduction. the is all stages in the rolling process of con In order to maintain high standards. maximum thickness of Although 197.. is reversed shear near the center of the the bar. studies with grid networks have shown that the surface layers are not only compressed but also sheared.40 times the reduction times the contact length. 199. the surface A13A20. Mech. Trinks. However. 19 as the diameter and condition of the the flow properties of the metal. ^ ratio.502 Plastic Forming of Metals rolls. Jr. Coffin. it appears from investigations^ of the internal strain. L. using the radiography of an embedded lead grid. vol. In comparison with other metalworking processes the deformation produced by rolling is relatively uniform.. pp.^ the lateral spread equals 0. Averbach. Internal defects such as fissures due to incomplete welding of pipe and blowholes are the result of ingot defects. Also. cit. 150153. banding and solidification Because rolled products usually have a high surfacevolume condition of the surface during siderable importance. C. 3 B. AIME. . vol. Figure 197 illustrates the typical grid distortion produced by the rolling of a flat bar. [Chap. F. while the greatest shear strain occurs when a bar is always rolled in one direction. op. MacGregor and L.25 to 0. when at the outside fibers the direction of rolling after each pass the stress occurs End view Fig. 1943.
. To compensate for this.Sec. Scratches due to defective rolls or guides can be a serious quality problem for certain classes of coldrolled sheet. is In spite of the great technological importance of rolling. there ' very Larke. or burning with an oxygen lance to remove surface defects. will depend on the rolling load.. Jr. the rolls so that the working surfaces are parallel when the rolls deflect under load. chap. 161173. thickness.and therefore. pp. Typical surface defects are Laps are surface defects which resemble seams slivers. the longitudinal residualstress pattern will be compressive at the surface and tensile at the midplane of the strip (see Fig. seams. cit.. op. it is customary to camber. as in temper rolling. 151). The prevention ation with railroad of flakes or cooling cracks rail is an important considerSpecial controlled cool and other rolled shapes. Occasionally pieces of refractory and oxide are rolled into the surface. ASTM. Most of the variation in thickness along the rolling direction is due to changes in the rolling speed or strip tension. under load to produce a sheet which is thicker at the center than at the edges. 539583. or camber. or crown. so that the plastically deformed region does not extend much below the outer surface. For this situation the longitudinal residual pattern will consist of tension on the outer surfaces and compression at the interior. Variation in either the thickness or rolls will also hardness of the material entering the cause a variation in Very elaborate devices for continuously measuring the sheet thickness and adjusting the mill operating conditions have been developed to control the thickness in continuous rolling mills. if the rolls are not operated at the conditions for which they were designed. ing schedules are used for these items to minimize the residual stresses. Control of the thickness and flatness of rolled sheet and strip is an important problem. M. .. vol. which are mainly responsible for this problem. sheet with edge or center waves will result. Ibid. Proc. 3. Baldwin. 1949. 2 3 W. and scabs. For the more usual case of large reduction