Professional Documents
Culture Documents
SN (iRINDINd lill 1 ( 1
ON RfSlDli.M .STRliSSl'S DiSIRIHUTION
AND.SURhACEINri (iRITV ()! 1)2
rilRI AD ROLl ING Dil S
Hy
Ol.CiA KARABl-l (illClllKOVA, B.S.
A Til I SIS
IN
INDUSTRIAL ENGINEERING
Submitted to the Graduate Faculty
of Texas Tech University in
Partial Fulfillment of
the Requirements for
the Degree of
MASTER OF SCIENCE
IN
INDUSTRIAL ENGINEERING
Approved
Accepted
ACHKOWI i:[)(iMi-NTS
The completion of this work is due to support and guidance of many people.
First. I would like to express my sincere appreciation and gratitude to my graduate
advisor. Dr. Iris Ri\cro. for her support, guidance, and encouragements throughout my
gradiialc studies and research progress.
I would like to thank Dr. Simon Hsiang and Dr. Hong Zhang for for being a driving
force in challenging m\ grow ili as a student and a researcher. 1 am also very greatfiil to them
for shairing their vision of research and encouraging me throughout my graduate program.
I deeply appreciate opportunities that 1 have received through the Department of
Industrial Engineering at Texas Tech University. Special appreciation goes to Fred
Schneider and Norman Jackson for their assistance in specimens' preparations and
conducting the experiment. Grateful acknowledgement is made to the materials
laboratory of Mechanical Engineering Departments for providing the equipment needed
to perform the experiment.
Thanks to all my friends and especially Steve Kelly, Ron and Jane Baker, who
have believed in me throughout my graduate career. They provided me with confidence
and support needed to achieve my academic goals.
I would like to pay my deepest respect and love to my parents and sister. I am
gratefiil to them for their moral and spiritual support, continuous encouragement, care
and infinite love. I owe them everything I have achieved.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS
11
ABSTRACT
vi
LIST OF TABLES
vii
LIST OF FIGURES
viii
LIST OF NOMENCLATURE
CHAPTERS
I.
II.
III.
INTRODUCTION
LITERATURE REVIEW
11
13
15
17
17
20
21
23
25
27
RESEARCH METHOGOLOGY
29
30
30
31
3.3.1 Hypothesis 1
31
111
3.3.2 Hypothesis 2
32
3.3.3 Hypothesis 3
32
3.3.4 Hypothesis 4
32
3.3.5 Hypothesis 5
33
33
35
36
38
38
3.6.2 BuehlerMetallograph
40
41
3.6.4 PocketSurfProfilometer
42
43
45
IV.
46
46
47
48
48
50
51
3.7.8 Electropolishing
53
55
55
55
60
61
65
66
iv
V.
4.2.2 Hypothesis 2
68
4.2.3 Hypothesis 3
70
4.2.4 Hypothesis 4
72
4.2.5 Hypothesis 5
73
75
76
77
78
78
84
VI.
86
89
90
90
91
100
RESEARCH CONCLUSIONS
102
103
105
106
107
BIBLIOGRAPHY
109
APPENDIX
A. EXPERIMENTAL DATA
113
118
120
129
131
132
ABSTRACT
Three questions were raised in this study: Can residual stresses after multipass
grinding be predicted? What will be their superposition relationship between initial and
final residual stresses distributions? Is there a method to predict and optimize surface
integrity of the material after certain number of passes in order to improve the tools life?
The procedure followed included a nested factorial experiment. The experimental
protocol consisted of six steps, which included microstructural investigation, hardness,
roughness, relative cold work, surface and subsurface residual stresses evaluation. All the
main effects, including heat treatment, type of grinding operation, and multipass grinding
technique and their interactions were found to be significant at a 0.05 level.
Experimental significance was summarized with a second order model representing
the grinding dynamics. The model was selected among four other candidates since it
provided the least predicting errors and the most parsimonious structure with only one
explanatory parameter, the damping ratio. The prediction of the complex nature of the
residual stresses was achieved in two-folds. First, given the multipass grinding operation
preceded by heat treatment, the damping ratio would change based on the experimental
data. Then, this thesis provides the prediction on how the residual stresses pattern at different
depth would change due to the damping ratio parameter. The contribution of this study was
characterization of heat treatment and grinding effects on the surface integrity factor of D2
thread-rolling dies and development of a plausible methodology and potential theory in
describing the memory relationship among multipasses during grinding operations.
vi
LIST OF TABLES
3.1
46
3.2
47
3.3
3.4
52
4.1
61
4.2
64
5.1
83
5.2
98
Vll
finishing.
49
LIST OF FIGURES
2.1
10
2.2
12
2.3
13
2.4
16
2.5
19
2.6
21
2.7
23
2.8
24
2.9
26
3.1
31
3.2
39
3.3
40
3.4
41
3.5
42
3.6
43
3.7
45
3.8
50
3.9
51
3.10
52
3.11
54
4.1
56
4.2
59
4.3
59
4.4
62
4.5
67-68
Vlll
4.6
69
4.7
71
4.8
74
5.1
77
5.2
79
5.3
82
5.4
87
5.5
89
5.6
91
5.7
5.8
93
5.9
94-95
5.10
96
5.11
97
5.12
98
5.13
100
5.14
101
fitting.
IX
93
LIST OF NOMENCLATURE
Partition ratio of the heat flux into the workpiece to the total heat flux
A'u
To
Ti
T?
(yyield
Poisson ratio
Elastic modulus
- y
Depth defined by equality of thermal elastic stress and the materials yield
stress is reached
Ns
y9
NpiNp2
\l/\ , yj2
r]
Ro
Camera radius
SI , S2
Rn
Roughness average
/?,.v
Rz
HT
Quenching
T(u)
Gr
Grinding operation
Stiffness
Viscosity/friction
Inertia
Applied torque
(^
Damping ratio
cOr,
Natural frequency
CDcj
q(i)
q{l)
q(i)
Acceleration term
*max
dmin
Depth corresponding to the right-end point of the residual stresses profile; final
depth point at which the stresses were evaluated
RS
RSmnx
RSmin
Ri,R\p
Magnitude of the residual stresses between the surface compressive values and
the tensile peak for actual and predicted data
Ro,Rop
Actual and predicted residual stresses amplitude between the tensile peak and
the final depth point at which the stresses were evaluated
xi
CHAPTER 1
INTRODUCTION
Fatigue of tools and dies is a crucial problem that has occupied manufacturers for
years. Fatigue is generally described by the behavior of materials under repeated cycles
of applied workload which causes deterioration of the mechanical properties and results
in a progressive failure. The consequence of fatigue includes: (1) loss of strength, (2) loss
of ductility, and (3) increased uncertainty of both strength and service life. The principal
cause of these characteristics is the imperfections in the materials. In addition, fatigue
behavior varies due to service conditions and prior treatments interacting with materials'
characteristics, but is generally common within a particular industrial application in
properties degradation dynamics and type of failure.
Factors influencing fatigue failure may include: (1) design features such as
notches, holes, fillets, uneven surface roughness, or any other feature that tends to create
stress concentrations, (2) fabrication cracks, (3) temperature, (4) speed of loading, and (5)
corrosive environments. Particularly, in thread-rolling cold work application, the dies are
operated at a range of low work temperatures under 400 to 500 F [1]; therefore, high
deformation forces are generated and exerted on the dies. These forces induce great
plastic deformation and, therefore, inevitably lead to the dies degradation and ultimate
failure before the completion the service cycle.
High-carbon high-chromium D2 steel is one of the most commonly used steels for
the thread-rolling dies application. It possesses excellent hardness, strengths, and wearresistance necessary to withstand deformation forces and satisfactory toughness. Despite
its high strength characteristics, there are still some failure related concerns, i.e., spalling
and crumbling. Therefore it remains the goal of metallurgists to work contstantly on
mechanical properties and fatigue life improvement.
Fatigue detection methods may include visual inspection, destructive and
nondestructive testing. However, since characteristic failures of the thread rolling dies
generally occur at surface or near-surface layers, inspection of the surface integrity is
commonly used to monitor the near-surface layers characteristics and their quality by
means of microstructural, hardness, roughness and residual stresses evaluation. Among
the other parameters, the state of the residual stress of a given material is one of the most
descriptive characteristics that can be correlated to many mechanical properties aspects of
the tool. It is mainly defined by a consequence of interactions among service cycle,
temperature, deformation, and microstructure [2]. Thus, studying and optimizing
manufacturing and machining treatments involved in the dies preparation would allow
one to control and predict the tools behavior, and in the long run to prolong dies hfe and
delay the part degradation.
Although heat treatment and grinding operations have been investigated extensively,
the comprehensive analysis of performing multiple grinding passes to obtain desired depth of
cut remains a relatively new paradigm. Researchers, such as Agha and Lui [3] have declared
that final residual sfresses are strongly correlated to initial residual stresses and the number of
cutting passes in machining operation. Their investigation was ftirther supported by Liu and
Yang [4] but limited to establishing the significance of the initial and final residual stresses
relationship at surface level. Multipass grinding operation is commonly applied in the dies
finishing operation, however little information are available in the public domain to broaden
the imderstanding of the multipass grinding dynamic characteristics and their influence on the
fomiation and/or relaxation of the residual stresses. This research was motivated to shed light
on this phenomenon and to assess its true effect by evaluating the resulting residual stresses
after various pass grinding techniques at surface and subsurface levels.
Furthermore, mathematical modeling was incorporated to find a parsimonious
predictive function, in which only one explanatory parameter was used to define the
residual stresses profile. In accordance with the principle of parsimony, there should be a
trade-off between model fit and model complexity. In other words, if the unknown
system can be modeled by more than one model, the simplest one should be preferable.
In this thesis, the emphasis was to determine the state of the residual stresses and to
understand the nature of its profile affected by the preceding treatment operations. For
this reason, the modehng effort of the residual stresses profile and its formation was
postulated to provide a representative, parsimonious, and predictive model, which may
ftirther be used to predict fatigue life of the tool, hi addition, a theory of damping
dynamic system and autoregression system identification technique were used to support
the proposed modeling approaches and theory, and to assign a physical meaning to the
explanatory modeling parameter.
b. The model should predict the residual stresses magnitude, tensile peak
location and subsurface distribution.
c. The mathematical model should establish functional
relationship in
CHAPTER II
LITERATURE REVIEW
Tool steels are used in a variety of industrial applications [5, 6]. The aspects of
their performance affect many other related fields, some of which are safety, continuous
process monitoring, and cost-profit considerations. The parts of a major concern are the
tools used for high duty cycled applications where their failure would cause more
serious damage. Thread-rolling dies are among such critical applications. Being
subjected to great deformation forces exerted on them in service, the dies are operated
at a low range of work temperatures which make them more sensitive to stresses and
more susceptible to sudden fatigue failure.
The life of thread rolling dies is primarily determined by the rate of deterioration
of the die threads profile [1]. Rolling imposes severe on the dies from bending and
sliding action, therefore fatigue failures generally occur from spalling and crumbling. A
failure of one component of the system will inevitably cause collapse of the whole
operation block. Thus the dies were considered as heavy-duty elements of the multicomponent thread-rolling system. This chapter reviews on major manufacturing factors
involved in thread-rolling dies preparation and their effect on mechanical properties and
potential service life of the tool.
D-group steels are the most highly alloyed cold-work steels [7]. Chromium, at a
nonnal concentration of 12% is the major alloying element. All the steels of this group,
except for D3, are hardenable by air cooling from austenizing temperatures. Therefore the
steels are very little susceptible to distortion and cracking during hardening. Excellent
wear resistance makes the steels suitable for long-run dies for blanking, thread-rolling
dies, rolls, shear and slitter knives [1].
Alloying elements of 0-group steels provide sufficient hardenability to make
possible hardening of small-to-moderate sections by oil-quenching. High carbon content
provides the most important service-related property - high resistance to wear at normal
temperatures. Subsequently the steels of this group are extensively used in dies and
punches for blanking, trimming, drawing and flanging [1].
Based on the concepts of productivity and performance, one should select a tool
steel for a particular operation to achieve the desirable service life. In this work, steel
selection and the treatment operation used were made with the key concept - enable steel
required properties for the thread-rolling flat dies application for better performance and
prolong fatigue life.
go
O '
itofionarv
Traversing die -,
Jie
Tro versing
die^
,_f^^^
Sfohonory die
Depending on its geometry, different techniques and processes are used in the
dies preparation. The most significant properties required for this application are
hardness, toughness, and wear resistance [6]. These properties must be high enough to
enable the dies withstand high deformation forces when in service and prevent brittle
behavior. Good wear resistance is the major consideration here, for the reason that the
prime cause of failure in thread-rolling is due to spalling and crumbling.
The most commonly used steel for thread-rolling dies are Ml and M2 high-speed
steels; D2 high-carbon, high-chromium steel and A2 medium-alloy cold-work tool steel [7].
In general, D2 steel possesses the greatest wear resistance among the listed above steels, and
is most commonly used for long production lines in rolling larger parts and alloys of higher
hardness [6]. Despite its versatile use in a range of industrial applications, there are still some
fatigue related problems that metallurgist work on. Answers to these problems would help
10
increase tools life and their efficiency, and therefore would result in economical savings.
Motivated by these reasons D2 steel was chosen for this research.
Due to low work temperatures thread-rolling processes impose severe
defomiation forces and stress on the dies from pressure, sliding and bending [6]. As a
result, the primary causes for the dies failing are spalling and crumbling. These processes
roughen the minor diameter of the product thread and cause the screw to go out of
tolerance. Spalling usually occurs near the edges of the tool, whereas crumbling originates
on the most sfressed rolling areas of crest and gradually propagates over the die threads.
In this work, various treatment combinations of D2 steel for the thread-rolling flat
dies have been extensively investigated (1) to study the effect of preparatory treatment
combinations on the ultimate mechanical properties, (2) to further model the results and
develop parsimonious predictive tool, and (3) to suggest the most advantageous treatment
techniques that would contribute to prolong fatigue life of D2 thread-rolling dies.
11
austenizing
preheating/
1 quenching
1 ""
1
.
^H^
*time
Figure 2.2. Hardening and single tempering practice in D2 dies preparation [8].
12
austenizing
quenching
tempering
tempering
time
Figure 2.3. Hardening and double tempering practice in D2 dies preparation [8].
Machining operation is an integral part in the dies preparation. Flat D2 dies are
usually ground before hardening to avoid grinding cracks, to which D2 is susceptible if
ground improperly [6]. However if tolerances require lower average distortions, dies
must be ground after heat treatment. These facts were kept in mind in selecting the
surface finishing parameters. By performing finishing operations to complete dies
preparation cycle, the author aimed to investigate not only its influence on surface quality
but the tool's properties and service life as well.
13
14
toughen the steel, almost for all industrial applications tempering is used to complete
materials preparation.
Tempering process consists of reheating of hardened steel to the temperatures
below Ai critical point, dwelling time and cooling down. Any temperature up to the
lower critical may be used for tempering; therefore an extremely wide variety of
properties and microstructure can be obtained upon the process [10]. Toughening of the
steel is always accompanied by decrease of hardness. The magnitude of induced
toughness is strongly correlated to the temperature the steel is brought up. The higher the
temperatures the larger degree of toughness the structure of the material will possess.
Therefore for the critical applications where a certain range of work hardness is required,
a selection of tempering procedure should be made very carefully. If the steel can be
tempered to the same hardness at more that one temperature, it is advisable to select the
highest tempering temperature that will produce the desirable hardness. Ultimately, it is
the balance of hardness and toughness required in service that determine the conditions of
tempering for a given application [10]
15
Machining process
Mechanical
Tliermal
Chemical
..^
impact
^
>>..
r fi
Tool
Geometric
Tliermal
machining conditions
properties
Mechanical
Machining parameters
environment
workpiece
Geometric
Thermal
properties
Mechanical
characteristics,
it inevitably
changes
surface
and
subsurface
materials
characteristics. By removing a layer off the materials surface, this procedure imposes local
plastic deformation, which is invariably accompanied by residual stresses formation [12].
The amount of plastic deformation produced on the surface of a ground part is highly
correlated to the parameters of the machining operation. The major factors affecting the
residual stress state of a material has been defined in [11] as the follows:
1. Machining conditions (depth of cut, speed of workpiece, cutting speed);
2. Topography of the grinding wheel (dressing conditions, wear behavior);
3. Specification of the grinding wheel (grit size, bond, hardness);
4. Cooling conditions.
16
17
variables of the process, some of which are depth of cut, feed rate, cutting speed, and use
of coolant [18]. The increase of these parameters result in greater surface temperature,
which, in turn, is directly correlated to the amount of energy entering the workpiece: the
more energy enters the surface, the higher temperature rises. Workpiece temperature and
its distribution under the surface can be expressed as the following [17]
f
'
/
^ ^
k^n
3r
3TK\
yAKtj
2/-2r+
'
1 - erfc
V
(2.1)
where i?. is the partition ratio of the heat flux into the workpiece to the total heat flux q in
the grinding zone, k^ is the workpiece thermal conductivity, K is the thermal difftisivity
of the workpiece, ZQ is the time for the wheel to pass the grinding zone, z is the distance
from the workpiece surface. For the calculation of maximum surface temperature, the /
value should be accepted as -z^, half the time that the wheel takes to pass over the
grinding zone.
Thermally induced residual stresses are determined by this temperature gradient
and materials elastic modulus. The material yield stress is usually a function of
temperature; this underlying relationship was determined by [17] and is schematically
demonstrated in Figure 2.5. The yield stress is constant until transformation temperature,
Tj, is reached corresponding to a ^ y transition. The decrease of the yield stress is lineariy
proportional to the temperature rise until T2 upper limit is reached. In case when surface
temperature does not exceed Ti critical temperature of the material, the following kinetics
of the transformation processes is true. The increase of surface temperature causes
surface expansion, however due to energy dissipation subsurface layers are affected to a
18
(y Yield
temperature
If thermal stress exceeds yield stress of the material, plastic deformation resuUs in
the surface, which leaves permanent deformation. In this case, tensile residual stresses
will be created in the surface upon cooling. This constant near-surface tension degrades
the material's fatigue life and may result in cracking.
It is essential to control grinding parameters as the stresses induced in the process
can either impair or improve the materials performance and service life. By keeping
thermally induced stresses below the materials yield stress, there will not be plastic
permanent deformation in the material and tensile residual stresses can be avoided [17].
The thermal stresses of the process can be determined from elastic stress analysis and can
be expressed as described in Equation 2.2.
19
^yield
a{z) =
d<z <z
2
'
1
l-V
(/ J-rf/-'
(l^
ldl2
(2.2.)
z<z<-d
where cr,,,vw is the yield stress of the workpiece material, vthe material Poisson ratio,
E the elastic modulus, a the coefficient of thermal expansion, T temperature
distribution when grinding surface temperature reaches its maximum, d thickness of
the workpiece and z,. the depth where the thermal elastic stress equals the materials
yield stress.
Final residual stresses created in the ground surface upon cooling can be
calculated from the compatibility Equation 2.3 [17]
EaT]
- | . ( z ) . ^ | =0
(2.3)
where a (z) is the thermal stress distribution at the begiiming of the cooling process
determined through the equation (previous one).
20
120
J-
S 100
I80
1
AAUSlVt
.*.
1 /
0.000
'^.clnrilf.;
fluid
' / ^
7; *"'
a :o
Sitit,
all
bCOj
GOCO
.1X0
,002
?.c.l. o i l
(1;.'0)
D.-y
^v..,_^CONV =^moML
"v
"'^--,.^
GE>rrL
0.002
O.DOO.C'Oe
DEP'H BEtay SJR=ACE
0.005
INCHES
0.010
0.012
Figure 2.6. Residual sfresses distribution induced by diverse grinding conditions [19].
21
AISI 52100 hardened steels. Comparison of the final residual stresses induced in the part
after first and second cut were shown to be different, although the cutting parameters
were identical. Subsequently, it was concluded that final residual stresses depend on preexisting residual stresses, and the authors claimed that it is impossible to predict whether
the residual stress after the second cut would be less or more compressive than that after
the first cut based on machining parameters alone.
A similar study was carried out by Liu and Yang in [4]. In this work, the samples
were subjected to multipass grinding and surface residual stresses were evaluated.
Identical grinding procedures have been performed on samples n times with no stress
relieving procedure in between. The fact that final resuUing residual stress of different
samples differed from each other, enabled the authors to draw a similar conclusion: final
residual stresses depend on initial residual stresses, and that their superposttion is highly
non-linear [4].
Multipass techniques are commonly used in industry however there are only few
studies that have concentrated on investigating in its effect on materials properties. In this
work, multipass grinding carries somewhat of an alternative meaning. Grinding
operations to the same depth were performed by means of 1-pass, 2-pass, and 4-pass
techniques to evaluate the effect of the operation and model the effect on subsurface
residual stresses superposition.
22
1'
^r
Breaking
Hardness /
Toughness
1'
1'
Adhesion
Wear
1
''
'
Tensile
stress
1
1
"
Fat] gu e r iilure
23
The effects of residual stresses have certain common features; however the
magnitude of their action cannot be extended to all materials and to the different
manufacturing processes that induce residual stress. It is accepted, that when applied
sfresses are added to internal tensile sfresses of the material, the part is locally overloaded [2].
To the dies in cold thread-rolling process, these added stresses speed up the deteriorating
processes and may cause sudden failure. If compressive residual stresses are embedded in
the surface, service applied stress is relaxed to some degree and dies fatigue life will be
improved. An example of stresses superposition is given in Figure 2.8.
Figure 2.8. Superposition of residual stress and service stress [2, 22].
Tensile residual stresses may not only facilitate crack initiation but also accelerate
their propagation by increasing mean stress. Fatigue strength of the material is considered
to be a mean or static stress superimposed on the cyclic stress. As the mean stress am
increases, the fatigue stress decreases [22].
The effect of residual stress on the tensile strength was investigated in [23]. After
being subjected to a row of preparation techniques compressive residual stresses were
induced at the surface of 4340 steel samples. Subsequent tensile fatigue testing of all
24
samples revealed increased fatigue life compared to those which were characterized by
tensile surface or lower compressive stresses.
Another manifestation of residuals stresses is their upsetting effect on dimensional
stability. Inequity of residual stresses results in dimensional distortion [24]. These
changes may occur both when residual stresses are relaxed or generated.
In all known up-to-date studies, it has been declared that compressive residual
stresses improve fatigue life; however, the latter carmot be predicted only by the stresses
magnitude and distribution. The authors of [25] demonstrated that it is possible to take
residual stress into consideration when predicting the fatigue life using a global approach.
Some of the aspects, which were involved in the computation, were hardness, strain
hardening and relaxation of residual fatigue stresses. However in real life application,
there are still may be some other factors involved.
The effect of residual sfresses on wear and fiiction properties has also been
investigated although to a much lesser extent [2]. Usually its effect was encountered in
fatigue life through other parameters like toughness, hardness and adhesion which are
affected by the residual sfresses. This effect has been integrated into global parameter of
adhesion, and Bhadeshia in [2] suggested try and determine the real effect of residual sfress.
25
initiated on a microscopically small scale. Then due to continuous cyclic loading the
crack grows to a macroscopic size. Fatigue life of the die is presented in Figure 2.9 and is
usually detemtined by the last cycle when crack reaches the surface and complete failure
occurs [26].
Cyclic
slip
Crack
initiation
Micro crack
propagation
Crack initiation
phase
Macro crack
propagation
Failure
Crack propagation
phase
26
not affected by residual stresses and slip band cracking can still occur. However, if
compressive residual stresses are embedded at the surface, they tend to keep such cracks
closed, which slows down further growth. The author of [26] gives an example of leaf
springs in cars. Shot peening is usually involved in their preparation. Microcracks have
been observed after many years of service; however their accumulation does not lead to
failure. It was concluded that the compressive residual stresses served as an effective
barrier by seazing cracks further growth.
If crack is initiated below the surface, one can be sure it is residual stresses that
caused it. Residual sfresses may have a remarkable influence on the location of crack
initiation [12]. A term "the weakest point" has been defined in [27] as the point of
maximum tensile residuals stress. The authors showed that it could be over four times as
large as that of the average residual stress. Brinksmeier in [11] claimed that fatigue fracture
starts at the weakest point, therefore he emphasized that the knowledge of subsurface
stresses is more important than the average residual stress. If crack propagates into macro
residual stresses field, the behavior of crack propagation can be significantly influenced by
magnitude and distribution of the residual stresses. Thus compressive surface residual
sfress reduce crack propagation rate and, consequently, delay complete failure.
27
Surface integrity factor can be evaluated on three different levels [28]. A basic
level involves microstructure, microhardness and roughness evaluation of the surface
layer resulting from the machining operation under certain process parameters. The
second level includes residual stresses and mechanical properties evaluation, whereas the
top level of the surface intensity factor evaluation consists of actual tests by means of
which the behavior of the part during machining is assessed.
Compressive residual stresses along with flaw-free surface prolong tools life and
delay the components failure [29]; therefore, the main objective of a metallurgical
engineer is to include those mechanical treatment in the part manufacturing, that produce
better surface integrity.
In this work, heat treatment and grinding effects of interest in D2 thread-rolling
dies were studied and evaluated first on a surface level by means of microstructural,
hardness and roughness examination, and then, subsurface residual stresses were
evaluated by x-ray diffraction technique.
28
CHAPTER 111
RESEARCH METHOGOLOGY
Tools life depends upon its manufacturing procedure that enables it with the
required characteristic properties and ensures acceptable longevity. Multipass technique
of the grinding operation is a very important factor to be considered in process
optimization. By selecting specific grinding parameters and techniques an engineer
should fulfill the following industrial requirements:
1. Predict the nature of the surface and subsurface residual stresses along with their
magnitude and in-depth distribution.
2. Optimize cutting conditions to produce the most advantageous
surface
29
30
Heat treaimeni
''
Q+T(1)
Q+T(2)
'
Q+T(2)
'
'
Gr1|
Gr2
1 r
of 1
''
P2
P1
P1
P4
P2
PI
Gr2
Ol/l
P2
P4
Gr1
P4
P2
P2
P1
P2
P4
P4
Figure 3.1. Experimental design and the factors of interest, where the following
notation was accepted: Q + T(i) - quenching followed by i cycles of
tempering, /=1, 2, 3 (Table 2); Gr^ - grinding operation with set y
conditions, 7 = 1, 2 (Table 3); P]< - number of passes employed in
grinding, k= 1, 2, 4.
3.3.1 Hypothesis 1
Hypothesis 1 was concerned with assessing the overall effect of the experimental
structure. Here, the experimental model was evaluated for any significant effect to
establish if any of the variables in the system could be explained.
Ho: None of the operations involved in the dies preparation would be significant.
Hi: At least one of the treatment operations would be significant.
31
3.3.2 Hypothesis 2
The effect of heat treatment {HT) operations was evaluated in Hypothesis 2.
Identically ground specimens across all heat treated groups were analyzed to establish
significance/non-significance of the effect.
HQ- /^ijk ^ /^2jk " /"jjk'
H,: /y,.,^ it ju^^^ ^ ^^.^.
Null hypothesis, if true, would establish that no heat treatment (HT) effect would
be significant, whereas alternate hypothesis would prove the opposite.
3.3.3 Hypothesis 3
Hypothesis 3 tested if grinding conditions (Gr) used in the study produced
significantly different effect. The hypothesis was assessed by analyzing the specimens
within each heat treatment group, which were subjected to the same multipass grinding
operation (P) but of different sets of parameters
JHQ: //,.,,. = /j^,y;
1H,:/^J,^
'^ /"i2k-
Null hypothesis states that no significant effect of grinding (Gr) factor would be
significant, and alternate hypothesis, if true, rejects the first one.
3.3.4 Hypothesis 4
Hypothesis 4 evaluated the effect of heat freatment (HT) and grinding (Gr)
interactions. The samples ground by the same number of passes were evaluated to
32
establish if the effect of each preceding heat treatment {HT) and grinding {Gr) operations
was different from those of other combinations.
|Ho:/A,k =
/"l2k
/^21k =//:2k
= / ' 3 1 k =/^32k;
lu
1,.
.,
k multipass technique, A: = 1, 2, 3.
lH,://nk^
/^,2k
'^ / ' 2 , k ^ / ^ 2 2 k
'^/^Blk
/'32K-
Null hypothesis, if true, would imply that no significant interaction effect of heat
treatment {HT) and grinding {Gr) is significant, while alternate hypothesis, if proven true,
would reveal significant effect of initial residual stresses on final residual stresses.
3.3.5 Hypothesis 5
Hypothesis 5 tested the significance of multipass {P) effect in grinding. The
specimens previously subjected to the same heat treatment {HT) and grinding {Gr)
operations but different grinding technique, i.e., number of passes employed, were
evaluated for significance.
HQ^ /^iji = M\fi = /^ija;
H,: /^ij, ^ /^ij2 '^ Mj3
\i - tempering cycles,; = 1, 2, 3 ;
\j- grinding conditions,7 = 1,2.
Null hypothesis implies that effect of all grinding techniques would be identical
and the mean residuals stresses obtained in various multipass groups would be the same.
Alternate hypothesis would reject null hypothesis.
33
cases due to practical and economical issues a certain range within target vicinity is
acceptable. This target range strongly depends on the nature of the experiment and is
valid within the experimental model. Based on these considerations, there were two
technical limitations that influenced selection of the experimental techniques used in the
work. Due to automatic grinder setup only three levels of multipass grinding technique
could be evaluated; however, this was sufficient to establish the characteristics
relationship of its non-linear superposition effect. Also, the number of replications in
subsurface residual stress measurements was limited to six, due to the nature of labor
intensiveness of the in-depth residual stresses measurements collection [30]. Thus,
repeatability of this experiment was achieved by the set of replications in collecting the
data, which was further used to build pure error term in squared sum of errors in general
linear model when assessing the hypothesis. Validity of this work was mainly concemed
with freedom from bias in the formation of conclusions. In this research the following
types of validity were applicable:
1. Internal validity was assured by use of clear relationship between dependant and
independent variables. The design of experiment was buiU in such a way that the
factors of interest were not confounded with any other effect, which could have
resulted in misleading conclusions.
2. Construct validity deals with assessing how well the experimental model
represents the real worid. The specimens were purchased from MSC fridustrial
Supplier, an official distributor of metal-working components. Selection of heat
treatment and grinding procedure suitable for D2 thread rolling dies was
34
determined by the recommendations of [8, 16] and the techniques used in the
study were kept as close as possible to those of industrial settings.
Also, the experimental set up agreed with the principle of temporal stability. Two
specimens were tested in each treatment combination group with memory-free
independent and repeated measurements to minimize random experimental error.
35
3. Sample material was cut from the adjacent sections of the bar.
4. Samples were tested in randomized order.
5. The number of memory-free replications in repeated measurements was collected
to assess the data variation.
36
tempering. The selection of the heat treatment factor levels was dictated by the a'priori
knowledge of the system; therefore, the levels of HT factor were fixed.
Grinding operation factor {Gr) described the conditions of the machining
operations performed after heat treatment. To prevent crack formation the intensity and
the strength of grinding operation must be kept non-abusive [6]. For this reason, the two
following considerations were kept in mind when selecting the grinding parameters: (1)
non-abusive nature of grinding procedure, and (2) industrial world representativeness.
Depth factor (depth) was introduced in the system with the purpose to evaluate
subsurface residual stress response to the preceding treatments. All hj^otheses being
tested in this research were assessed at each depth level, to establish the range of the
studied effect. Residual stresses distribution throughout the depth is highly non-linear
[20]; therefore, seven levels of the depth factor were studied. An increment in depth layer
and the ultimate depth of the affected layer were determined by considerations of the
previous works in the related studies [11, 28, 32] and the experimental run.
Pass factor {P) described the number of passes used in grinding. Three fixed
levels of the factor were investigated, which included single-, double-, and four-pass
grinding technique. Two studied grinding conditions differed from each other by depth of
cut; therefore the layers of material removed in similar pass techniques were not
identical. In order to ehminate the possibility of drawing any biased conclusion the levels
of pass factor {P) were nested in those of grinding factor {Gr).
Replication factor {rep) was concemed with the replications of the experiment.
Overall, six independent replications in testing each freatment combination were utilized.
37
The levels of this factor were picked randomly from the stand point of data variability
and time issue. Replications were made independently therefore no interactions of this
factor with any other were encountered. Blocking of random replication effect did not
upset any unbiased fixed nature of other factors in assessing the hypothesis.
38
39
tempering procedures were conducted using the lower chamber of the furnace. Safety
interlocks on all doors and access panels were set in such a way, that the power to unit
would be cut every time the furnace was open. Programmable controls allowed the
operator to preset a profile which the furnace automatically followed. Digital displays of
the furnace made it easy to monitor the temperature control. The digital controller was
fully factory-calibrated and ready for configuration. Therefore the only needed
calibration step was setting the minimum and maximum limits of the temperatures used
in the experiment.
40
41
3.6.4 PocketSurfProfilometer
Roughness examination of the samples was assessed by using a PocketSurf
profilometer (Figure 3.5). The body of the PocketSurf roughness gage was equipped with
a diamond-tipped stylus, digital display, parameter selection switch, probe, inch/metric
switch, and transverse length switch. Roughness measurement operations were based on
mechanical-electronical principle: the diamond tipped stylus was mechanically moved
over the surface and deflections of the stylus were collected into electronic signals, which
were then amplified and shown on the display. To ensure validity of the measurement the
probe stylus and skid were in good contact with surface, and the setup was properly
aligned, i.e., the axis of probe transverse were parallel to the surface being measured. Out
of three transverse lengths (1,3, and 5) available in the apparatus, a transverse length of 5
was used to obtain more accurate measurements.
42
Calibration of the profilometer was checked prior to its operation. The gage was
calibrated using the reference specimen and the EPL-1681 Riser Plate supplied with the
PocketSurf kit. Five cutoff lengths were used in assessing proper calibration of the
instrument. The readings from the standard specimen were within 4 |a"/0.1 mm of the
value stated on the label, therefore calibration was assumed to be within tolerance.
43
The following principle lies in the basis of the x-ray diffraction system [33].
Voltage excited tungsten filament, which made electrons hit the chromium anode and
generate x-rays. The geometrical size of x-rays passing through collimator was defined
by the aperture. Detectors measured lattice strains by recording high angle diffraction
lines produced by a collimator x-ray beam. Recorded x-ray diffraction pattern was
converted into optical signal, amplified and transmitted to the computer for processing
and interpretation.
Firstly, the system was initialized and warmed up. The initialization stage was
done with the Z-motor by focusing the x-ray optics, which could be controlled either
through the software or with the remote pendant. The warm-up procedure was used to
bring up the ftiU power from zero to a full operating voltage in a stepwise manner; it
minimized problems related to measurements collection, and in the long run would
increase the life of x-ray tube.
Considering irreversible detrimental effect of radiation the following safety
measures were true. Safety interlocks in the x-ray diffraction system prevented the x-ray
generator from working when the shutters were not properly closed. Another important
safety feature was flow monitor: if the water flow cooling the tube was not sufficient, the
flow monitor automatically enabled the safety interlock to the h.v. generator, thus
disabling x-rays. The x-ray diffraction lab and the operator were constantly monitored
with dosimeters to detect if the radiation exceeded acceptable level.
44
The middle section of the tank was filled with the electrolyte, which was
circulated by pump when running; and the other chambers of the tank were filled with
chilled water to keep the electrolyte cool. The magnet in the system played the role of the
anode. To enable better anode-cathode contact the bottom surfaces of the specimens and
the magnet were polished to remove scale, dirt or any other possible contaminants.
Electropolishing system was equipped with a set of electrode tips of different
apertures. Based on the size and geometry of the specimens a 4 mm electrode tip was
used. Recommended settings of voUage and time for removal 0.001" of the material from
the surface were provided in [34], however prior to the experiment the parameters were
calibrated for D2 steel specimens. Calibration techniques involved trial-and-error testing
of the effects of flow rate and voltage controls in time domain to remove 0.001" off the
45
surface at a time. Once the parameters were determined, this set of data was used at all
experimental runs.
C
1.55
Cr
11.5
Mn
0.35
Si
0.45
Eighteen specimens were cut from the adjacent sections of the same bar. The samples
(1/2" thick and wide, 1" long) were sized for the convenience of the experimental procedures.
Initially, six specimens were randomly selected to undergo one of the three heat treatment
procedures (Table 3.2). Microstructural analysis, hardness, and surface residual stresses
examination were accomplished then. Subsequently, the specimens of each heat freatment
group were subjected to various grinding operation of different pass grinding techniques, and
the characteristics of the specimens were evaluated again.
46
Preheat
Hardening
1500 F, 12min
1500 F, 12min
1500 F, 12min
1850F,40min
1850F,40min
1850F,40min
1 cycle
400 F, 2 hs
960 F, 2 hs
1000F, 2hs
Tempering
2 cycles
900 F, 2 hs
960 F, 2 hs
3 cycles
920 F, 2 hs
47
48
as a reference datum. Grinding operations of interest were performed on the other side of
the specimens. First, the z coordinate of the machine was calibrated to coincide with preground specimen's surface level. Then the parameters of interest, including feed rate,
table speed, and the amount of the material to be removed from the surface at a pass was
set. An effective synthetic coolant system (CIMSTAR 3865 undyed metalworking fluid
concentrate) was used in grinding to reduce work temperature and prevent formation of
tensile surface residual stresses [17]. Grinding parameters utilized in the study are
presented in Table 3.3.
Grinding conditions 1
5416.5 feet/min
3450 rpm
40"/min
0.001"
CIMSTAR 3865
1 pass
Grinding conditions 2
5416.5 feet/min
3450 rpm
40"/min
0.0005"
CIMSTAR 3865
1 pass
49
Among three available roughness measuring techniques, i?, Rmax, and i?-,
roughness average i?, most commonly reported in literature, was used in the study, i.e.,
the arithmetic average height of roughness irregularities measured from a mean line
within the evaluated length, L. This principle can be described in the way of the
following equations:
1^
K=-\\y\dx,
(3.1)
L
R^{approx.) = y.+yi+y^^-yn
50
(3.2)
/v.
X-ray source
Film
.N,P2
Specimen
surface
. ^ - %
Figure 3.9. One-angle arrangement for x-ray diffraction technique [38], where Nsspecimen's normal; fi, angle between incident beam and Ns, Npi, and
Np2 the normals to the different planes 1 and 2 respectively; ^/i and ii/2,
angles between Ns, Npj, and Np2 respectively; rj, the angle between the
incident beam and diffracting plane normals; RQ, camera radius; 1 and
2, two diffracting planes; Si and S2, parameters directly related to the
Bragg's angles, 0i and 02.
Micro- and macrostress deformation principles lie in the basis of x-ray diffraction
techniques. When steel is under applied or residual stress, elastic strains cause change of
interplanar distance in the crystal structure of the material. By measuring this change in
interplanar distance it is possible to quantify the stress. Multiexposure technique (MET)
51
EM
Del 1 Del 2
1.177t.
1.1726
1.1&76
1,1626
0,0
0,1
0.2
0,3
0,4
O.E 0,6
Sin2psi
0,7
0,8
0,9
1,0
Phi=0,00
Figure 3.10. Single exposure method: d versus sin^V|; splitting in residual stresses
measurements on single tempered specimen, subjected to a set of
grinding parameters 1 (Table 3.2) of 1 pass grinding technique.
Summary of the x-ray diffraction measurements parameters are given in Table 3.4.
Table 3.4. X-ray diffraction settings data.
Radiation
Lattice planes - Braggs angle
Specimen's orientation angles, \\i
Irradiated area
Tube voltage
Tube current
Cr-K
(221)- 156
-11.95, 0, 11.95
1 mm^
17 kV
20 mA
52
A 10-second exposure time was selected for the stresses evaluation. Tonshoff and
Brinksmeier in [14] investigated the effect of measuring time on standard deviation of the
produced data and concluded that among the range of 1-30 seconds exposure time per
step, a 10 seconds exposure time produced the smallest standard deviation.
Position of K^ diffraction peak, intensity and peak breadth were assessed by
fitting a Pearson VII 85% distribution function by least square regression [39]. The
following corrections were applied to estimate true residual stresses [34]:
1. Corrections for background - a gain profile collected on glass surface was used in
x-ray diffraction measurements to minimize background noise
2. Correction for factors dependent on 9 and \\i - included Lorentz, polarization, and
absorption factors (LPA)
3.7.8 Electropolishing
Ruud in [38] claimed that electrolytic and chemical polishing are the only methods
for material removal from the surface that do not generate residual stresses in the
component. However the application of these techniques requires subsequent subsurface
residual stresses correction by an amount related to the relaxation due to the removed
layers. The residual stresses correction applied in this work are described by the equations
3.3 - 3.5. The principle of the electropohshing is schematically shown in Figure 3.11.
53
Removed layer
'-1
-1
z7
-^
'-1
-1
dz.
(3.3)
z7
where cr^(Z|) represents true stress in any direction at depth z,, and CJ,. (Z, ) represents the
measured value at that depth, z/ is a distance from the bottom of the specimen to
uncovered depth of interest, H- original thickness of the specimen.
The correction of the residuals stresses at each particular depth was computed as
difference between the true and measured values
^H cr,. (z)
c(z,) = c7,(z,)-cr,. (z,) = 2 | rfz-6z,
'^1
I'I
(3.4)
The integrands can be expanded in a Taylored series in terms of surface values and
integration is performed step by step. However, only the first terms could be accountered
in calculation for shallow depths corrections:
c{z,) =
-4aJH)^.,
(3.5)
54
CHAPTER IV
EXPERIMENTAL RESULTS AND DISCUSSIONS
55
atoms [10] in the body-centered tetragonal crystal lattice of martensite, and the presence
of retained austenite. Therefore steel hardening is always followed by tempering which
causes certain phase transformations, discussed below, and thus promotes more stabilized
materials structure. The results of microstructural analysis of tempered components are
shown in Figure 4.1-b, -c, -d.
\\^^
^ \
,11
,00 n
3w2^ \
'^_
"*
^^^i\.*r>>-'^^. ^'-tAH-
Figure 4.1. D2 steel microstructural examination after various heat treatment conditions.
Specimens were air-quenched from 1850 F, 2% nital etched; lOOx.
56
57
this transfoi-mation occurs only after the transition carbides are well established and
usually takes place in the range of 390-660 F temperatures, which defines the second
stage of tempering [41].
The amount of transfonned retained austenite in the second tempering stage
strongly correlates to the range of tempering temperatures and the number of cycles in
heat treatment. Low tempering temperatures limit the size of the transition carbide
particles and result in very little change in the dislocation substructure of the as-quenched
martensite; therefore very little portion of retained austenite is transformed at lowtempering temperatures [7]. And vise versa, the higher the range of tempering
temperatures, the more retained austenite transforms in martensite. A similar trend was
observed when the number of tempering cycles was increased: the more tempering cycles
were involved in heat treatment operation, the more retained austenite transformed into
martensite.
Carbides precipitation and the retained austenite transformation significantly
increased toughness and lowered the hardness compared to that of as-quenched condition
(Figure 4.2). Due to the microstructure consisting of high-carbon martensite and a high
density of coarse undissolved carbides, the toughness of hardened D2 steel is low [10].
Tempering transforms unstable brittle martensite into a tempered one, thus relieving
some internal stresses. More stabilized tempered microstructures increase plasticity of the
martensitic structure and improve toughness. Being inversely dependent, the increase of
toughness in tempering operations causes decrease of hardness, which was observed on
all tempered specimens.
58
Hardness, HRC
640.43
b5
..aaaaiiM
64
63
62
61
62+0.6
i|:!i;;' 1
60
59
58
57
56
::!iK:l:i::|
iH--E
'!"''='
''!;'! '
-^9+0.43
59^a4
N
U
1i-:''ii.."-','lffi
-''=> '-'=181
'^'''m
In general, the retained austenite of D-group steels is highly alloyed and hence
quite unstable. For this reason, surface hardness remains low until the secondary
hardening temperature is reached [7]. Gill in [42] investigated the effect of tempering
temperatures on the dynamics of the retained austenite. The results of this correlation on
D2 steel are given in Figure 4.3.
Tempering temperature, F
^XS
5O0
700
90Q
1100
13O0
The tiiird tempering stage consists of the precipitation of the cementite and change of
carbides morphology [10]; however, the effect of this stage could not be revealed by the
metallograpliic examination in this work due to microscope resolution limitations.
The results of hardness examination obtained in this work conformed to the graph
of Figure 4.3 and indicated that secondary hardening took place in triple tempering
procedure. In multiple tempering at high temperatures range (900-1000 F), more carbon
and alloying elements are taken into solution, and therefore are available for carbides
precipitation [7]. Roberts et al. in [5] claimed that in high-carbon high-chromium steels a
large portion of retained austenite stays untransformed up to 800 to 1000 F where a
conditioning reaction occurs, which results in large quantities of new martensite to be
formed, and hardness increases again.
60
a=0.05 significance level. Therefore, in the given experimental setup it can be concluded
that surface rougliness depended solely on the amount of the material removed from the
surface during the grinding operation. That is, the smaller the depth of cut, the better
surface finish was produced. The forces generated in grinding to a smaller depth remove
the material in a more even and uniform fashion producing a smoother surface in finishing.
Table 4.1. The resuhs of roughness examination. The data is based on five measurements.
Grinding
technique
1 pass
2 passes
4 passes
Longitudinal
Grinding
operation 1
15 1.58
13.8 1.10
14.41.14
direction
Grinding
operation 2
10.2 0.84
101.58
10.8 1.64
Transverse direction
Grinding
Grinding
operation 1
operation 2
28.8 0.84
29.4 2.88
30.6 1.14
30.6 0.89
29 1.22
29 1.58
61
120 -'
Every tempering cycle of the heat treatment operation employed in the study
advanced residual stresses relaxation, and the more cycles were involved in the heat
treatment, the lower the resulting tensile residual stresses were. Thus, specimens
undergone triple tempering revealed the lowest tensile residual stresses compared to
those of single and double tempering treatments.
Figure 4.4 shows that every tempering practice utilized in the study reduced
tensile residual stresses towards more compressive, however the effect of multiple
tempering operations was non-linear. That is, the most significant amount of residual
stresses relaxation occurred during single tempering, and even though each subsequent
62
tempering cycle also relieved internal stresses, the increment of the stresses relaxation per
each cycle was considerably smaller.
When material is under stress, applied or residual, dimensional changes due to
inhomogeneous plastic deformation and thermal/mechanical relaxation occur [21]. This
deformation may cause a set of macro and micro residual stresses alteration. Micro
residual stresses influence local hardness within component and consequently - the local
strength. Micro residual stress measurements can be collected along with macro residual
sfresses and determined by peak broadening of Full Width Half Maximum (FWHM)
values of the x-ray diffraction profile. Microplastic deformation changes the perfect
crystalline regions between dislocation tangles making them smaller. Upon reaching
nominal 0.1 jam [43], further reduction of these regions along with variation in lattice
spacing of the diffracting crystallites cause peak broadening.
Comparison of near-surface residual stresses and cold work values before and
after single pass grinding operation are given in Table 4.2. The degree to which the
specimens were cold worked demonstrated strong correlation to the amount of material
cut off in grinding: the specimens subjected to grinding operations of a smaller cutting
depth revealed higher degree of surface cold work compared to that of the alternative
grinding. Moreover, the amount of cold work, introduced in grinding, was found to
depend on the surface integrity.
compressive residual stresses were characterized by greater peak broadening, and thus,
more prominent cold work. Specimens finished by grinding operation 2 (refer to Table 3)
demonstrated the maximum relative cold work of all specimens, and the minimum
63
amount of cold work was obtained on non-ground surfaces. Preceding heat treatment
procedures did not show significant effect, i.e. all specimens in each heat treatment group
possessed cold work within the same range.
Table 4.2. Residual stress and relative cold work (FWHM) values on D2 samples of
various surface finishing conditions.
Surface finishing
condition
Preceding heat
treatment
Q
Q + T(l)
Q + T (2)
Q + T(3)
Residual stresses,
ksi
96.46 2.45
45.08 2.65
20.65 2.21
15.17 2.00
Grinding
parameters 1
Q + T(l)
Q + T (2)
Q + T (3)
-53.47 3.26
-55.97 2.92
-59.40 2.35
3.49 0.18
3.74 0.13
3.94 0.21
Grinding
parameters 2
Q + T(l)
Q + T (2)
Q + T (3)
-48.60 3.46
-50.77 3.93
-55.77 2.83
4.76 0.14
4.98 0.17
4.57 0.24
Non-ground
Furthermore, by carefully analyzing the experimental data from the Table 4.2 and
Figure 4.4, the assembled observation clearly shows that grinding procedures produce not
only better surface finish, but also drastically change the stress state of the material. Both
grinding operations of the study converted surface tensile residual stresses towards
beneficial compressive stresses. In particular, grinding operation of set 2 parameters,
which involved smaller depth of cut, tend to produce higher compressive residual stresses
compared to the other grinding operation. However, this difference in surface residual
stresses was not shown to be significant at a=0.05 significance level due to relatively
high variation of the data. This observation corresponds to the results obtained in [19]
64
where Prevey claimed that surface residual stresses may not be representative of the
machining process. Specifically, in case of grinding the surface stresses may be neariy
independent of the grinding parameters [19]. In the same paper, the author also claimed
that machining and grinding practices produce variations in the surface residual stresses
which could be so large that surface results would be of little value. For these reasons,
subsurface residual stress measurements were further conducted in order to correctly
evaluate the effect of grinding on the stress state of the material.
65
And the coefficients of variation of all data sets were within 3-16 percentage range,
which gave strong indication of the reliability of the resuUs.
4.2.1 Hypothesis 1
Hypothesis 1 dealt with assessing the overall effect of the model. First, the
experimental model was evaluated for any significant effect. The results of this
hypothesis (see Table B.l of Appendix B) indicated that all factors of interest have
revealed significant effect, except for the replication factor {rep), therefore shownsignificant factors were further considered individually in the subsequent hypotheses.
Partitioning of sums of squares of the main effects in the experimental model and their
interactions is graphically shown in Figure 4.5-a. More detailed output information is
provided in Table B.2 of Appendix B.
Non-significance of the replication {rep) factor supports previously made
assumption of the identical and independently distributed data and memory-free
replications. This finding promotes unbiased hypotheses evaluation and drawing right
conclusions. Six replications made in residual stresses evaluation were used to build data
variation and error term which was then used in testing the hypotheses. The output results
of this hypothesis revealed significance of all factors of the modeling design; therefore
heat treatment, grinding operation and multipass effects along with their interaction
combinations were considered individually in the following h3^otheses.
66
a: Hypothesis 1
b : Hypothesis 2
Tempering effect
c : Hypothesis 3
Grinding effect
67
d : Hypothesis 4
Tempering and Grindinginteraction effect
numbering of the effects
1 - (HT)* (Or),
2-error term.
e : Hypothesis 5
Multipass effect
numbering of the effects
1 - pass,
2-error term.
4.2.2 Hypothesis 2
The objective for testing this hypothesis was to argue if tempering effect was
significant in differentiy heat treated groups or not. The residual sfresses obtained on the
specimens, whose preparation procedures differed by only number of tempering cycles, were
compared between each other. Evaluation of the tempering effect across all grindmg
operations revealed similar frend in the residual sfresses distributions. In essence, the more
68
tempering cycles were involved in the dies heat freatment, the less tensile subsurface residual
sfresses were. At surface level residual sfresses of such specimens were more compressive;
however the effect was less distinct. An example of the residual sfresses profile describing
the effect of the preceding heat freatment is graphically presented in Figure 4.6.
Grinding operation 1, 2 passes
I -^T(Y)
3
'VI
Figure 4.6. Heat freatment effect on the formation of the residual stresses. The
data points represent averages of six repetitive measurements. Upon
heat treatment, the specimens were finished by grinding operation 1
in a single pass fashion.
Initially, F-test was performed to establish the significance of the effect, which
output results rejected the null hypothesis (Table B.3). Sums of squares partitioning of
the current experimental model is presented in Figure 4.5-b and indicates the successive
efficiency of the model. Subsequentiy, the data was tested using muftiple comparison
tests, which demonstrated significance of the effect at every depth of the residual stresses
profile. To assess the validity of the model, the assumption of normally distributed
residuals was verified. The data was first sorted by grinding {Gr), pass {P), and depth
69
factors and then evaluated. Q-Q plot and several statistical tests, such as Shapiro-Wilks,
Kolmogorov-Smimov and etc. (refer to Tables B.3, B.4 and Figure B.l) let the tests for
normality pass, hence previously made assumption was satisfied.
From the material's standpoint, the tempering effect was mainly determined by
the stress relaxation processes, which occur when the material was heated up to elevated
temperatures and held at a preset level. The dynamics of the phase transformations is
mainly governed by the temperature and the time of the tempering operation. Tempering
procedures relieve internal stresses locked in the material upon hardening by
transforming retained austenite to martensite and changing the morphology of martensite
from brittle to tempered one. Thus more stabilized structures resulting in lower tensile
stresses were achieved. The maximum residual stress relaxation occurred in triple
tempered specimens, which resulted both in a more stabilized microstructure and more
favorable in-depth residual stresses distribution. Single tempering practice, however,
employed the lowest tempering temperatures compared to those of the other two
treatments, therefore resulting residual stresses are more towards tensile.
4.2.3 Hypothesis 3
The effect of grinding operations was evaluated within each heat treated group
across identically muftipass ground specimens. Sums of square partitioning and the effect
of this model are shown in Figure 4.5-c. The output results (Table B.5) indicate that the
effect of both grinding conditions used in the study was significant at all depths below the
surface; however the effect was found to be insignificant at surface layer. The latter
70
finding conform the conclusion made by Prevey in [19], where he claimed that surface
residual sfresses may be nearly independent of the grinding parameters.
Single tempered specimens, 2 passes grinding
--Grl
.^
CO
:i
VI
Hi
on.
(07
Depth
Residual stresses, ksi (Gr 1)
Residual stresses, ksi (Gr 2)
0
-55.97
-50.77
0.001"
34.47
16.16
0.002"
73.73
58.15
0.003"
87.09
79.72
0.004"
67.25
57.73
0.005"
40.79
31.52
0.006"
19.25
16.37
Figure 4.7. The effect of grinding operations on the final residual stresses profiles.
The specimens' preparation included double tempering heat treatment
procedure, following finishing by two grinding conditions in single
pass fashion. The data is based on six observations.
Grinding effect on the residual sfresses profile is shown in Figure 4.7. Analysis of
tests statistics in SAS output and graphical effect presentation (Figure 4.7) indicated that
the effect of the grinding factor was strongly correlated to the type of grinding operation,
i.e., the amount of the material removed off the surface. That is, the specimens finished
by grinding to the smaller depth of cut were characterized by more compressive surface
residual stresses and less tensile stresses in subsurface layers compared to those of the
alternative grinding operation. This trend in residual stresses was mainly due to the
71
amount of energy entering the specimen in grinding, and workpiece surface temperature.
In essence, when cutting depth is increased, the surface temperature grows
proportionally, and due to the number of phase transformations (details in Chapter II) the
residual stresses of more tensile nature are generated.
4.2.4 Hypothesis 4
In the previous two hypotheses heat treatment and grinding effects were found to
be significant. However, in the cycle of the dies preparation none of these effects is
independent: finishing by grinding followed heat treatment specimens' preparation, hence
could be influenced by the initial state of the residual stresses, i.e., heat treatment induced
residual stresses. This hypothesis aimed to investigate the interaction effect of these two
factors and argue if final residual stresses were correlated to initial residual stresses.
The interaction effect was evaluated by contrasting the residual stresses obtained
on specimens within each grinding technique group, previously subjected to various heat
treatment and grinding operations. The F-test output results and R square values for this
model (see details in Table B.7) indicate the significance of this interaction effect.
Therefore it was concluded, that the effect of heat treatment and grinding operations is
strongly correlated, and final residual stresses are pre-defined by the initial residual
stresses induced in the material by heat treatment. Analysis of sums of squares
partitioning (Figure 4.5-d) along with the R square value suggest that the model is
constincted efficiently. The results of the normality testing given in Table B.S and Figure B.3
satisfied the previously made assumption and validated the model.
72
4.2.5 Hypothesis 5
Multipass grinding factor {P) in this hypothesis was evaluated for significance by
contrasting the residual stresses data obtained on identically tempered specimens and
finished by the same type of grinding. By assessing the hypothesis in this fashion,
identical pre-existing residual stresses induced in the preceding treatments were ensured,
and the observed contrasts between the levels of multipass factor were solely due to the
number of passes employed in grinding.
Residual stresses profiles across various combinations of tempering {HT) and
grinding {Gr) differed from each other in magnitude and their distribution throughout the
depths of the affected layer (hypothesis 4), however muUipass effect within each
combination was found to be similar. The significance of this model was established at
every depth of the profiles, and its effecfiveness can be evaluated by the partitioning of
the sums of squares (Figure 4.5-e) and SAS output resuhs in Tables B.9.
The effect of multipass factor on the residual stresses profiles is graphically
presented in Figure 4.9. The major differences in the residual stresses profiles were
observed in the stresses distribution and tensile peak location. And again, residual
stresses were strongly correlated to the amount of material cut off in grinding. Ultimate
depth of cut was identical across all multipass grinding techniques of each set of grinding
condftions. However due to the multipass technique, the amount of material removed in
2- and 4-passes grinding was twice and four time as less as in single grinding. Thus, the
grinding practice with the least cutting depth (4-passes) produced the least tensile
subsurface residual stresses, and single pass ground specimens were characterized by the
73
highest tensile residual sfresses. This effect was found to be significant at all depths at
a=0,05 level of significance.
Double tempered, grinding conditions 1
a
U
0.C07
Figure 4.8. The effect of multipass grinding on the residual stresses distribution in
double tempered specimens after grinding operation 1.
Another distinct difference in the residual sfresses profiles due to the multipass
effect was in the location and magnitude of the tensile peak. Likewise, the previous
discussion, this phenomenon was determined by the amount of the material removed off
the surface at each grinding pass. On single pass ground specimens the tensile peak was
observed at depth of 0.003", whereas specimens ground using both multipass techniques
had tensile peak at the depth of 0.002" below the surface. The latter observation indicates
that in single pass grinding the effect of the applied forces in material removal has much
greater magnitude, therefore the depth of the affected layers and the magnitude of the
induced sfresses are significantly greater compared to those of multipass ground specimens.
74
CHAPTER V
MATHEMATICAL MODELING OF GRINDING DYNAMICS
There are numerous dimensions in any machining process, which may contain
several interactive variables of equal importance, that make the process very complex.
As such, there is a reduced probability for setting up full explanation that can be
asserted to be the "real causation." At best an attractive model which is partial truth
can be aimed. Such model should be able at the very least to describe certain
phenomenon precisely and accurately.
The models that can be proposed depend on the individual's state of knowledge.
Ideally, a model in its mathematical concise form should have predicting power.
Therefore, it is of no utility to construct a mathematical model which is too complex to
support the reason. That is, the individual has to be in a juncture to derive a mathematical
model and hence physical consequences from the model. Some of the purposes for which
models in this chapter were constructed are:
1. Obtain answers about the dynamics of the grinding process;
2. Develop the theory of the grinding process and heat treatment;
3. FaciUtate conceptual progress in the residual stresses formation;
4. Influence further experimental design.
75
76
77
= mq{l)
(5.1,b)
78
applied torque (r). Finally, the inertia term (/;/) dictates the ability to maintain the
momentum or the kinetic energy level. In essence, the physical explanation of the
damping/vibration phenomena involves interchange of potential and kinetic energy of the
components of the system: kinetic energy is released in the form of motion and vibration,
and potential energy is stored in the components of the system.
When describing the effect of grinding operations, i.e., cutting conditions and the
number of passes, both the effects of the grinding wheel and material's surface were
equally important. The lumped dyiiamic system was presented as spring-mass-damper
elements of the grinding machine and damping-stiffiiess of the workpiece. The schematic
of the lumped dynamic system is presented in Figure 5.2.
* xi
f X2
]
Figure 5.2. Simplified mechanical model of wheel-surface system [44].
79
An essential concept derived from Figure 5.2 is that the energy absorption or
damping represents the resistance force of the material to vertical vibration motion in
material removal grinding processes. Damping is always present in the systems where the
energy dissipation occurs [45]. This energy dissipation causes a decrease of vibration
amplitude with depth increase from surface layers downward. According to this concept,
the atomic structure of the outermost layers of the material is subjected to greater
vibration amplitude, whereas its effect on the in-depth layers decays in some exponential
fashion. Internal friction caused by grinding defines the capacity of the material to
dissipate applied mechanical energy in the form of heat. The authors of [21] claimed that
internal fiiction invariably describes the capacity for unconstrained or enforced vibration
and is measured by the rate of decay of free vibration decrement. Under the applied
grinding stress, internal friction and thermal vibration of atoms results in irreversible
plastic flow, which in its turn causes plastic deformation, and thus advances residual
stresses formation.
To summarize all of the above effects to the ground surface, the reaction force of
the material was considered as the sum of static, fiiction, and internal forces.
Furthermore, there should be two categories of friction integrated in the lumped model:
(1) external friction between the grinding wheel and the workpiece, and (2) internal
friction at the atomic level within the material. In evaluating this grinder-workpiece
intercoupled system, it was plausible that the magnitude of the internal friction depended
on the maximal applied stress of the grinding cycle [45].
80
<=J^=
V 4mk
(5.2)
2ylmk
81
damping ratio (Q and natural frequency (fij). Underdamping system possesses null
damping ratio, therefore no damping occurs and the system oscillates continuously
(Figure 5.3-a). Damping ratio within a range of 0 - 1 describes damping dynamic
system, in which the motion decays gradually with depth propagation and eventually
comes to rest (Figure 5.3-b). Damping ratio above critical value of 1 characterizes
overdamping system that comes to rest immediately without oscillation as soon as
stiffness component is released (Figure 5.3-c). The governing equations of these three
dynamic systems are given in Table 5.1.
10
15
20
25
30
-0.5
Figure 5.3. The displacement response for a system under various damping ratio {Q
on the dynamic motion of the system: a - underdamping system (i^= 0);
b - damping system (0 < C< 1); c - overdamping system {C > 1).
82
Time domain
^=0
1 - cos col
-i0j
(b)
1-
0 < <^< 1
Vw^
sm
./VW^ + tan-'
CO.
\.i\
(c)
1+
^>\
a'-\
" -a e
^ =
2a
physical
characteristic
describing
the
potential
change
from
83
,=Vw^ = J-(l-^')>
(5.3)
Vm
where Q)d is damped natural frequency, and ci) is natural frequency of the system, both
are measured in rad/s units.
5.3.2 Data-To-Model Approach
The viscous/friction damping force, considered in the model, is proportional to
the velocity across the damper and always opposes the motion; therefore the damping
force is considered as a linear continuous function of the velocity. Then the complex
motion of the system could be expressed by the following second-order differential
equation
mq{) + bq{i) + k[q{i) - q{0)] = 0,
q{i)
where
q{l) =
q{)-q{-A)
V q{)=
(5.4)
q{i)-q{-M)
q{i) +
q{-2)-2q{-M)
A'
q{-M)-q{'2A)
(5-5)
To apply the Equations 5.4 and 5.5 the basic assumption was made: the motion of
the grinder-material intercoupled system depended on the component's inertia (m),
fiiction {b) and stiffness {k). Velocity of the system q{) was presented as a finite
difference of the current and the previous position in - domain. Further, in this study,
will be transformed into d depth parameter of the residual sfresses profile through linear
84
interpolation. Accordingly, the acceleration term q{i) is found through computing the
rate of the velocity change. Both q{C) and q(0 were assumed to be real and stable
functions, therefore the Equation 5.4 can be re-written in the Newtonian Equilibrium
representation as follows
m
q{f) + q{f-2Af)-2q{iAn
^^
, q{)-q{('-AC)
,,
'- + h''
2^^
l^k[q{(')-q{Q)] = 0.
(5.6)
A/'
or
vq{n +
-2/?;
Af-
Al
q{i-Al) +
m
.Ar
b
q{-2Ai) + kq{0) = 0, (5.7)
Ai + k
(5.8)
By analyzing the equation above, one can conclude that every state of the system
can be expressed through the previous two consecutive positions: if one has two known
consecutive positions of the system in
85
-rz2^+Ai
m
Af
r m
^ A t l ^
m
m
71
-bAl
m
kAf
m
(5.9)
, _ kq{0) _ kq{0)
^ _m_
m Ae
Ae
7 = 1-
Vi^
rsin a)/^\-C
(5.10)
+tan"
J)
To find the maximal and minimal points, the necessary and sufficient conditions
must be satisfied
dY ^
=0
^
and
d'Y ^
r < 0.
dt
86
(5.11)
Accordingly, the following was obtained (for more details see Appendix C):
l = - nn
co..^!^
n e 1,2,3...
(5.12)
Based on the pattern of the residual stresses distribution, the first maximal and
minimal points were the key features of the modeled pattern in this study (Figure 5.4).
>'(0
1.6
n=l
n=2
n=3
1.4
1.2
1
0.8
0.6
0.4
0.2
10
15
20
25
30
When the best fit has been found, the values of the abscissa and ordinate Y{)
were converted to those of the corresponding residual stresses coordinate system, i.e.,
depth d and residual stress {RS), respectively. The conversion was based on linear
interpolation, and the utilized equations were derived from imposing a theoretical curve
onto an actual one and governed by extreme points.
Abscissas coordinate:
87
max
"mill
'^nmx
' = 0-6,
(5.13)
DC
~ ^'^mas
^^-*
^ , )(''^'^max ~'^'^min)
>
max
DC
Kiy^
.
-,
]
I =dr,-d
mill
DC
(^max -^f.X-'^'^max - ^ ' ^ o )
-'f^'Jmax~~
1
RS.1..-RS0
(5.14)
.
-,
T
l = d max
, ^ , - amm
.
88
10
15
20
25
30
35
40
45
50
55
60
Figure 5.5. The effect of natural frequency on the displacement of the damping curve.
The natural frequency defines the rate with which the damping system oscillates and
is determmed by the time interval during which the fimction repeats itself [48].
Mathematically, it does not influence the magnittide of the oscillation, but characterizes the
location of the exfreme maximal and minimal points in the
fitting was done by (1) imposing the first maximal and minimal points of frie damping curve
onto residual sfresses profiles, (2) scaling the coordinates systems in order to convert one into
89
anotlier one, and (3) estimating the lack of fit. Upon scaling, every nattiral frequency-defined
damping curve revealed identical upward and downward slopes. Therefore natural frequency
cy -ylkim
was assumed to be a constant; this implies that the suspension stiffiiess {k) and
mass {m) of tiie given grinder-material intercoupled system tend to scale together.
90
U3
en
lU
^ = 0.41
SSerror = 9 2 4 . 6 8 3 4
surface
3.444
-59.52
-56.34
0.001"
5.1666
69.03
44.09
0.002"
6.8888
87.25
84.92
0.003"
8.6110
62.84
75.91
0.004"
10.3332
40.62
46.81
0.005"
12.0554
24.96
22.35
0.006"
13.7776
15.88
15.88
Figure 5.6. Theoretical damping curve fitting of the residual stresses profile using
three-point fitting approach. Specimens' preparation included
quenching, followed by single tempering, and finishing by grinding
of four-passes in grinding conditions 1.
Again, as seen from the graphical representation, the lack of fit in this method was
large enough to motivate the author to seek another fitting technique within the damping
ratio approach.
91
until equal areas under the theoretical damping curve and actual residual stresses profile
were reached. This approach gave the best fit among all the other utilized approaches,
and therefore is described below in details. The structure of the algorithm used to assess
the best fit can be delineated as the following five steps in the iteration:
Step 1. Based on the a'priori sttidy natural frequency {co,,), number of the exfremum order {n),
and a range of (' values were assigned for each iteration of damping ratio {Q testing.
Step 2. Theoretical peak of the damping curve was subsequently computed.
Step 3. The ratio of R\:Ro segments of the actual residual stresses profile were computed.
The R\ represented the magnitude of the residual stresses between the surface
compressive values and the tensile peak; whereas RQ represented the residual
stresses amplitude between the tensile peak and the final depth point at which the
stresses were evaluated. Schematically these segments are shown in Figure 5.7.
Step 4. For each value within the range of 0
theoretical (i.e. predicted) profile was computed from the following proportion
R
R
5- = ^ , where RQ and Ro are the segments lengths of ordinate range between
the maximal and the right-end (minimal) points of the residual stresses and
the damping curve profiles, respectively; R\ and /?ip are the segments of
ordinate range between the initial and the maximal points. This step
determined the best locations of the initial, maximal and right-end minimal
points of the theoretical curve.
92
Figure 5.7. Defining the .H'l .Ho proportion of the residual stresses profile.
^.pred
pred
93
C=0.1-f0.9
RS. ^=0:2:16
M.. = 0.5=l.
7tn
,...nB
v^fl
^/^?
+ tan'
'1^'
end
Ro = RS {1,3)-RS {1,1)
R, =RS {1,3)-RS {1,1)
Rin
Y,
s]p YmiLX< miLX
R
0.
=Y
pred
max
R,
^^
R,1 .
-N
Qp
-f(y
>^v=l-
VT^
rsm
6 ; / , V r ^ + tan-'
1-r
^
yes
94
B
55 error = 0
-CaX'
/ , V i ^ + tan-'
Y=h
-Sin CO.
1-C^
^ ' '
!=0.1-f0.9
^ _ . = ^ + (/-l)A^
-f((;
i:. = 1 -
sm
/ . . . V l ^ + tan-'
li^
RS,=RS{\,i)
i<4
RS,=RS{\,A)(Y
-Y
){RS{\,^)-RS{\,\)
Y
{Y
-Y
){RS{\,A)-RS{\,1)
-Y
The flow chart presented in Figure 5.9 explains the algorithm used in the equalarea modeling approach. Once the damping ratio coefficient {Q with the best fit was
found, the damping curve coordinate system was converted to that of residual sfresses
profile and the squared sum of error was computed. As in the previous modeling approach,
MATLAB was used to implement the developed algorithm (for more details, see
95
Attachment C). The examples of equal-area fitting approach are demonsfrated by Figures
5.10 and 5.11, while tiie complete information on the fit of all residual stresses profiles is
provided in Appendix D. The residual stresses profiles were collected on the specimens
which undergone similar freatinent, differed from each other only by the type of grinding
operation. The MATLAB code and more detailed information on the output are given in
Appendix C.
(07
^= 0.76
SS.ror= 141.0620
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
value
3.444
5.1666
6.8888
8.6110
10.3332
12.0554
13.7776
Rbact
-59.52
69.03
87.25
62.84
40.62
24.96
15.88
RSpred
-59.52
62.14
87.25
64.14
33.13
18.98
15.88
Figure 5.10. Theoretical damping curve fitting of the residual stresses profile
using equal-area fitting approach. Sample's preparation included
quenching, followed by single tempering, and finishing by grinding
of four-passes in grinding conditions 1.
96
^ = 0.62
Sog
103.7936
surface
0.001"
0.002" ^0.003"
0.004"
0.005"
0.006"
value
4.1
6.0541
8.0081
9.9622
11.9162
13.8703
15.8244
RSact
-62.36
54.00
84.85
62.55
40.61
25.08
12.02
RSpred
-62.36
51.68
84.85
69.56
38.38
18.42
12.02
Figure 5.11. Theoretical damping curve fitting of the residual stresses profile
using equal-area fitting approach. Sample's preparation included
quenching, followed by single tempering, and finishing by grinding
of four-passes in grinding conditions 2.
After determining the curve that provided the best fit for each residual profile, i.e.
every experimental treatments combination, the modeled parameter {Q was plotted
against the factors of interest (see Figure 5.12) to explain the phenomenon and to tailor it
to the effects of the experimental treatment. Numerical values of the damping ratio
coefficients are given in Table 5.2.
97
-Agrinding conditions 1
- grinding conditions 2
4 passes
2 passes
1 pass
2
3
4
number of tempering cycles
Figure 5.12. Correlation of the modeled parameter - damping ratio {Q and the
factors of the experimental model.
Table 5.2. The results of the equal-area modeling approach. The data are the damping
ratio coefficients that provide the best fit to the experimental treatment combinations.
Type of grinding operation (Gr)
Grinding operation 2
Grinding operation 1
Multipass
Heat treatment factor (HT)
Factor
Heat treatment factor (HT)
Q + T(3)
Q + T(2)
Q + T(l)
Q + T(2)
Q + T(3)
Q + T(l)
0.18
0.21
0.11
0.25
0.17
0.1
PI
0.44
0.44
0.51
0.47
0.49
0.56
P2
0.55
0.53
0.62
0.67
0.76
P4
0.76
By analyzing the relationship of the damping coefficients corresponding to the
treatment combinations one can find that there is a limited amount of knowledge and
practice which is significantly manifested in the thoughts and algorithms of this modeling
chapter. The key findings are:
98
1. Multipass grinding factor {P) revealed distinct effect on the residual stresses profile,
i.e., shape and the magnitude of their distribution throughout the affected depth.
That is, regardless of the type of heat treatment and grinding operation, the shape of
all profiles within each level of the multipass factor was similar and therefore
characterized by the damping ratio coefficient within the same range values.
2. Heat treatment effect {HT) on the shape of the residual sfresses profile was found to
be significant on all specimens; however its influence was interactive with the effect
of tiie multipass grinding factor {P). Specifically, in the specimens finished by single
pass grinding the damping ratio coefficient increased with the increase of tempering
cycles. Nevertheless, in four-pass ground specimens this trend was found to be the
opposite: an increase of tempering cycles in heat freatment caused slight decrease in
the characteristic damping ratio coefficient. This finding indicates, that no particular
freatment of the experimental design had independent effect on the resulting residual
sfresses profile, and that every treatment operation used in specimens preparation was
influenced by the precedingfreatmentand, in tum, influences the subsequent one.
3. Grinding effect {Gr) also defined the shape of the stresses profile: specimens
subjected to the grinding operation to a smaller depth of cut (grinding operation 2)
were characterized by greaterer damping ratio coefficient compared to those of
the alternative grinding operation. This effect was more prominent in multipass
grinding rather that in single pass grinding. The fact, that this effect was strongly
correlated to the number of grinding passes employed, conforms to the resuUs of
[4] where high non-linear superposition the residual stresses was declared.
99
Figure 5.13 and can be summarized as follows: increase of damping ratio coefficient
causes the curve smoothing outward, in other words the amplitude of the system's
oscillation is significantly reduced by increase of this coefficient.
=>
Figure 5.13. Correlation of the damping ratio coefficient and the shape/amplitude
of the displacement of the damping system.
100
The inference derived from the model is that the amounts of the material removed
in two- and four-pass grinding teclmiques were 2 and 4 times smaller than those of single
pass techniques of the same type of grinding. Correspondingly, applied force and the
energy entering the workpiece were significantly lower in multipass grinding operation,
which caused less significant internal friction (Figure 5.14). At the atomic level, such
structures would be subjected to a smaller degree of plastic deformation and reach
equilibrium state sooner. Smaller degree of plastic deformation would also result in lower
tensile subsurface residual sfresses, and therefore smaller amplitude of the residual stresses
alteration from surface to subsurface layers within the affected depth of the material.
2F
Figure 5.14. Schematics of the internal friction and displacement of the structural
components under various applied forces (F) in grinding operation.
Along with the observations of the modeling results stated above, the corresponding
relationship of the damping ratio coefficient and multipass technique validated the previous
conjuncttire on the memory related aspects in the system. Specifically, the increase of the
damping ratio coefficient within each of the identically heat freated and ground group
indicated that the information of z pass was transmitted to that of (z +1) pass.
101
CHAPTER VI
RESEARCH CONCLUSIONS
This study evaluated the effect of various combinations of heat treatment and
grinding procedures on D2 thread-rolling dies surface integrity. The hypothesis and the
dies' preparatory treatments were established by means of microstructural examination,
as well as hardness, roughness and residual stresses distribution evaluation. The
experimental results were modeled to develop a predictive tool that would explain
residual stresses superposition due to treatment operations. The final model involved only
one predictive parameter, which reflected any change of the complex experimental
treatment procedures and estimated in a precise and parsimonious way the corresponding
change in subsurface residual stresses distribution. Furthermore, the parameter was
used to elucidate the dynamics of the grinding operation, and to support the theory of
the heat treatment and grinding effects on the materials characteristics. The results of
the experiment and the model helped evaluate conceptual progress in the residual
stresses formation, and can further be used in improving the dies preparation design to
enhance their life.
Modeling of the experimental data is motivated to reveal the underlying structure
and that the observed data represents the reality. If so, the model can further be used to
explore the insight of the nature. Philosophically, there are two opposing views on the
modeling [49]. Niel Bohr suggested that observation of the world, based on certain
modeling perspective, was fundamental to constructing its reality. Any modeling imposes
102
certain views that are projected onto the reality, and therefore the latter can be as
descriptive and complex as the model one can describe it by. Albert Einstein, on the other
hand, posited that objects existed irrespective to observation. That is, reality is always
there regardless of how you view or model it. To imitate their debate, this thesis should
be finalized with the following two indispensable questions:
1. What modeling experience was and could be infonned by the experimental
observation?
2. What observative knowledge was and could be informed by the modeling
experience?
The results of this research work can be viewed both by the model-reality and
reality-model approach and are presented below.
103
However, this dependency was not linear: secondary hardening effect and slight
increase in hardness was observed on triple tempered specimens
3. The effect of grinding conditions on the surface roughness in the longitudinal
direction was found to be significant and strongly dependent on the amount of the
material removed off the surface at each pass. However multipass techniques
produced surface roughness no different from that of the single pass grinding.
Roughness in the transverse direction was neither dependent on the grinding
operation nor on the number of passes employed in the process.
4. Surface residual sfresses upon heat freatment operation were significantly reduced
towards compressive by every additional tempering cycle. This effect, however, was
foimd to be non-linear: the greatest amount of the residual sfresses was alleviated by
the single tempering operation (63%), and even though each subsequent tempering
cycle released additional amount of the intemal sfresses (30% and 7% respectively)
the increment of the stresses relaxation was considerably lower.
5. Surface cold work was strongly correlated to the surface integrity factor: smooth
surfaces with higher compressive surface residual stresses possessed more
prominent cold work. Thus, maximal relative cold work was obtained on the
specimens subjected to a smaller depth grinding, while the minimal cold work
characterized non-ground surfaces.
6. The effect of tempering operation on the residual stresses distribution was
significant at each depth of the affected layer. In essence, the lowest compressive
surface residual stresses and the highest tensile subsurface residual stresses were
104
generated by single tempering heat treatment procedure, while the opposite effect
was observed on triple tempered specimens. In addition, triple tempered
specimens
revealed
significantly
reduce
variability
of
the
measured
105
106
energy expenditure associated with the triple tempering cycle might be justified by
greater hardness and more beneficial residual stresses distribution it generates.
Grinding operation 2 in multipass fashion is strongly recommended. Smaller cutting
depth of such finishing operations promotes formation of smoother surface with greater
degree of relative cold work. These characteristics along with advantageous surface and
subsurface residual stresses distribution would enable D2 thread-rolling dies to sustain
applied forces better and maintain workability throughout greater number of cycles.
In summary, the effect of the manufacturing treatments on the resulting surface
integrity and the residual stresses distribution was weighted by their emphasis on the dies
fatigue life. This research work integrated already existing knowledge of the heat
treatment and grinding operations to the particular application (thread-rolling dies) and
introduced a new perspective on the effect of multipass grinding technique. The analysis
of the experimental results and their further mathematical modeling helped not only
explain the dynamics of the heat treatment and grinding operations and the formation of
the residual stresses, but also predict their superposition and functional relationship.
107
the processes. The latter would be then correlated to hardness and toughness
characteristics of the materials.
2. Fatigue analysis of the dies after their final preparation is to be performed and
correlated to the surface integrity factor and the state of the residual stresses. This
would allow develop functional relationship of the material's characteristics and
fatigue life.
3. The developed two-step model includes: (1) predicting the damping ratio {<^)
based on various experimental interactions (i.e., tempering, grinding operation,
and grinding technique) and (2) predicting the RS patterns based on ^ . The
future modeling effort should expand the current two-step model to encompass
(1) various materials (within cold-work tool steels), (2) various sizes of the same
material, and (3) various types of grinding dynamics (i.e., grit size, griding wheel
and cutting speed).
4. Further optimization of the manufacturing and machining parameters based on the
developed models can be performed. This would enable one to (1) estimate final
mechanical properties of the dies by knowing the parameters of the manufactiiring
freafrnents, and (2) to determine necessary treafrnent combination to produce desired
surface integrity and residual stresses distribution, and thus better fatigue hfe.
5. Cost-Benefit analysis is to be accomplished to estimate the effect of the advanced
mechanical properties benefit of the triple tempered dies over the double
tempered ones scaled to the time and energy cost.
108
BIBLIOGRAPHY
1.
Davis, J.R. ed. Tool materials, ASM International, Materials Park, OH, 1995.
3.
Agha, S. and CR. Liu, Experimental Study of Surface Roughness and Residual
Stress in Superfinish Machining of Hardened Steel, Progress Report DMI96102022, Purdue University, 1997.
4.
Liu, CR. and X. Yang, A New Perspective of the Residual Stress Induced by
Machining and Grinding, Manufacturing Science and Engineering, 10, pp. 807816, 1999.
5.
Roberts, G.A., J.C Hamaker, Jr., and A.R. Johnson, Tool Steels, American
Society for Metals, Metals park, OH, 1962.
6.
Dieter, Jr, G.E., Mechanical Metallurgy. McGraw-Hill, New York, NY, 1986.
7.
Roberts, G.A., G.G. Krauss, and R. Kennedy, Tool Steels, 5th ed., ASM
International, Materials Park, OH, 1998.
8.
Bryson, W.E., Heat Treatment, Selection, and Application of Tool Steel, Vol. 1.
Hanser Gardner PubHcations, Cincinnati, OH, 1997.
9.
Knowlton, H.B., Heat treatment. Uses and Properties of Steel, Vol. 1, American
Society for Steel Treating, Cleveland, OH, 1929.
10.
109
11.
12.
Lohe, D., K.-H. Lang, and O. Vohringer, Residual Stresses and Fatigue Behavior.
In: G. Totten, M.Howes, T.Inoue, eds., Handbook of Residual Stress and
Deformation of Steel, ASM International, Materials Park, OH, pp. 27-53, 2002.
13.
Schwartz, R.W. and S.B. Rao. Wheel Selection Technique for Form Gear
Gnnding. In: American Gear Manufacturers Association Fall Technical Meeting,
San Francisco, CA, USA, pp. 10-16, 1985.
14.
15.
Tonshoff, H.K., H.-G. Wobker, and G. Brunner, CBN Grinding with Small
Wheels, CIRP Annals. Manufacturing Technology, 44(1), pp. 311-316, 1995.
16.
Oberg, E. and F.D. Jones, Machinery's Handbook, 16th ed. Vol. 1, Machinery
publishing, Co., New York, NY, 1959.
17.
Chen, X., W.B. Rowe, and D.F. McCormack, Analysis of the Transitional
Temperature for Tensile Residual Stress in Grinding, Journal of Materials
Processing Technology, 107(1-3), pp. 216-221, 2000.
18.
19.
20.
Erocsson, T., The Effect of Final Shaping Prior to Heat Treatment. In: G. Totten,
M.Howes, T.Inoue, eds.. Handbook of Residual Stress and Deformation of Steel,
ASM hitemational. Materials Park, OH, pp. 150-158, 2002.
110
21.
->7
Lu, J., Handbook of Measurement of Residual Stresses, The Fairmont Press, hic
Upper Saddle River, NJ, 1996.
23.
Pardue, E.B.S. and R.H. McSwain, X-ray Diffraction Stress Analysis as an NDE
Technique. In: AIP Conference Preceedings, TEC publ, Knoxville, TN, pp 14211428,1994.
24.
Lu, J., et al.. Prediction of the Residual Stress Relaxation During Fatigue, hi:
Mechanical Relaxation of Residual Stresses, pp. 75-90, 1998.
25.
Treuting, The Nature, Origin, and Effects of Residual Stresses, hi: Residual Stress
Measurements. American Society for Metals, Cleveland, OH, pp. 1-41, 1952.
26.
27.
Prevey, P.S. and M. Field, Variation in Residual Stress due to Metal Removal,
Annals, of CIRP, 24(21), pp. 497-501, 1975.
28.
Grum, J., Induction Hardening. In: G. Totten, M.Howes, T.Inoue, eds.. Handbook
of Residual Stress and Deformation of Steel, ASM Intemational, Materials Park,
OH, pp. 220-247, 2002.
29.
30.
Ill
31.
32.
33.
34.
Hilley, M.E., et al., eds. Residual Stress Measurement by X-ray Diffraction SAE
J784a, Society of Automotive Engineers, Inc., New York, NY, 1971.
35.
Karabelchtchikova, O.K., I.V. Rivero, and S.M. Hsiang, Evaluation of the Initial
Residual Stresses Distribution and Multipass Grinding Techniques on Final
Residual Stresses Distribution in D2 Steel. In: TMS Annual Meeting and
Exhibition, Chariotte, NC, 2004.
36.
37.
Mahr Federal Inc, Pocket Surf- Protable Surface Roughness Gages. Profilometer
manual, 2000.
38.
39.
Prevey, P.S., The Pearson VII Distribution Function in X-ray Diffraction Residual
Stress Measuremen, Advances in X-ray Analysis, 29. pp. 103-111, 1986.
40.
112
41.
42.
Gill, J.P., High-carbon high-cromium Steels, Trans. ASST, 15, pp. 387, 1929.
43.
44.
45.
Beards, C F., Vibration Analysis and Control System Dynamics, John Wfiey &
Sons, New York, NY, 1981.
46.
47.
48.
49.
Whitaker, A., Einstein, Bohr and the quantum dilemma, Cambridge University
Press, New York, NY, 1996.
113
APPENDIX A
EXPERIMENTAL DATA
Replication
As-quenched
1
2
3
4
5
6
7
8
9
10
11
12
Mean
StDev
C.V.
64.5
64.5
65.0
64.0
64.5
64.5
63.5
64.5
64.5
65.0
65.0
64.5
64.0
0.43
0.007
Quenched
and single
tempered
60.0
62.0
61.5
61.0
62.5
62.0
62.0
62.5
62.0
62.0
62.0
62.0
62.0
0.69
0.011
114
Quenched
and double
tempered
58.5
59.0
59.0
58.5
59.0
59.5
59.0
59.0
59.5
59.0
59.5
60.0
59.0
0.43
0.007
Quenched
and triple
tempered
59.0
59.0
60.0
59.5
59.5
60.0
59.5
60.0
60.0
60.0
60.0
60.0
59.5
0.40
0.007
1 pass
Mean
St.Dev
C.V.
2 passes
Mean
StDev
C.V.
4 passes
Mean
St.Dev
C.V.
Longitudinal direction
Grinding
Grinding
operation 1
operation 2
17
10
16
9
13
11
15
10
14
11
15.0
10.2
1.58
0.84
0.11
0.08
Transverse direction
Grinding
Grinding
operation 1
operation 2
29
29
27
28
34
28
27
30
30
29
29.4
28.8
2.88
0.84
0.10
0.03
14
14
12
15
14
13.8
1.10
0.08
9
10
12
8
11
10.0
1.58
0.16
30
30
31
30
32
30.6
0.89
0.03
31
30
29
31
32
30.6
1.14
0.04
14
16
13
15
14
14.4
1.14
0.08
10
10
9
12
13
10.8
1.64
0.15
29
28
29
31
28
29.0
1.22
0.04
29
31
28
27
30
29.0
1.58
0.05
115
Table A.3. Experimental data on surface residual stress and relative cold work (FWHM)
after various heat treatment operations.
Quenched and
Quenched and Quenched and triple
single tempered
double tempered
tempered
Rep Residual Cold Residual Cold Residual Cold Residual Cold work,
work,
work,
work,
stress,
stress,
stress,
stress, FWHM,
FWHM,
FWHM,
FWHM,
ksi
ksi
ksi
2 deg.
ksi
2 deg.
2 deg.
2 deg.
1
99.52
2.27
41.99
2.71
19.28
3.08
16.27
3.63
2
94.74
2.57
45.51
2.63
18.49
3.51
15.04
4.09
3
96.82
1.89
43.56
3.54
2.95
22.91
3.35
16.26
4
97.98
2.24
3.72
49.08
2.46
23.18
12.03
3.97
5
93.41
2.49
2.94
19.41
3.52
16.27
3.79
45.27
Mean 96.49
3.75
2.29
2.74
20.65
15.17
45.08
3.49
0.21
St. Dev 2.45
0.27
0.21
2.21
0.32
2.00
2.65
0.16
0.13
C.V.
0.12
0.06
0.08
0.11
0.09
0.03
As-quenched
116
sassed #
CN
en oo
o
o O
en en en en in
en
CM
CD
CO
en
to
lO
en
00 1^ o
h*
to
Q:
r^
CO
tn oC O
in CM
CM
117
CM CM
cn
CO
CO
o
en en in en m
cn en
00 CO
CD r ^
CO
en
i^ CD
CO
CM
CD
0-
en
in
CM
00
1**
to
CO
tc
00
CM
O
CO
CM
CM
00
CO
CD
CD
CO
CD
O
CD
00
CO
CO
in CM
-75
-75
-75
00
CO
to
CM
CO
CO
CO
CO
o
^
CM
o
CO
CM
in
CO
CO
00
eo
c;5
CD
CO
CD
CO
1^
to
r-
CM
"3-
CM
CM'
to
eo
^
^
CO
CO
OO
Tt
^
00
CD
00
CD
CO
CM
CO
2.38
-76
cn
CD CM
in
CM
CO
0.60
in
CO
CO
5j oo
CO
CO
en
CM
o
o
CM
in
-75
CO 00
CO
in
T CM
ai
o
^
00
00
^
en en
CM
cn
00
CO
CO
CD
CJ
eo
oo o
to
CO
00
CM
in
oo
in
CM
CD
O
in
in
CO
CO
to
CM
in
r-
<t
CJ
o
tn
to
2.46
26.31
25.15
2.40
44.88
7.92
58.88
12.56
11.05
22.56
24.86
22.39
20.16
43.42
1.46
44.89
1.17
31,74
32.23
34.16
32,34
57.96
0.97
58.93
2.73
44 78
26,45 1 40.37
30.18
-78.69 1 26,79 1 38,22 1 35.50
1.67
11.521 -69.40 1 25.51 1 40.63
1.24 1 10.09
1.89
3.38
1.88
34.36
32.69
25,05
1.88
18.64
0.49
18.15
-75,9
-58,54
0.00
2.51
-60.93
1.73
24.91
2.22
22.69
24.39
21,76
1.89
28.34
1.95
26.39
25.13
2.91 1
15.77
2.78
12.99
10.51
16.72
10.58
16,54
2.16
16.85
2.44
14.41
11.08
16.10
13.98
14.03
42.09
18.02
24.54
29.46
27.74
56.82
-62.74
16.37
17.98
19.67
43.14
11.68
16.26
1.99
30.44
3.18
27.26
44.35
-60.93
1.01
50.57
64.28
53.81
29.67
27.98
59.45
-61.13
-63.40
1.43
8.52
6.39
6.44
7.91
9.24
8.97
1.91
2.54
1.91
24.03
26.15
42.11
15.10
1
1
1
1
1
1
1
48.03
62.38
1.27
51.20
52.54
49,87
60,93
62,71
27,14
28.58
43 97
-58,54
3.07
12.02
10.28
44.18
46.86
67,77
59.96
48,94
62,49
45,62
48 38
47,45
48.61
59,63
20,87
15.98
2.20
21.91
0.64
21.27
18,40
21,44
61.35
50 916
52.89
-57.21
-62,56
2.35
-59.40
21,11
23.01
8.26
to
in
2.07
0-
T-
24.96
r-
22.58
-74
en
-74
CO
CO
21.09
CD
cn
-74
CO
1.50
5 5
cn to
o
ai tn
00
2.51
22.89
^
CM
CD
en
CM
O
00
CM
CO
00
1^
4.26
h-
CO
CM
00
62.84
24.63
CO
21.16
en
24,00
5
CD
CM
t^
^
^
ISt.Dev
CD
en
in
CM
87.25
CM
to
CM
T
CM
eo en C O ai
en eo
CO
-59
00
CO
CM
CO
1.43
CM
CO
co
^ ^
CD
CM
CO
CM
O
61.41
CM
0.95
o en CM CM
eo
86.30
to
21.71
T-
en CM
to
CO
CO
63.71
CO
CM
00
58.06
CM
CM
en CM 00 CO 00
CO CO en CM 00
CM
83.29
o
CD
CO
CO
en
o
in
86.14
m r^ m
CM
62.63
00
CO
CM
en eo o cn
CO
1^ cn
eo 00
CM
-59
en 00 CD en
o en o en
en
en
CO
Mean
CO
CO
^ in
1^
^ in
00
CO
-64
CM
CO
CM
in
CM
en CO
CO o
cn en
CD
1-58
tn
CM
CO
CO
CM
T-
82.77
CM
CM
tn
cn r^
-64
to
CM
92.73
to
CO
-59
s
CO
CO
59.23
64.24
<i-
82.72
CO
90.17
ai CO r-^ CM
eo CM
en en
CM en C O o CO CO o
CO
CO
CO CO CO
-59
o
in
3.73
l~- 00
-64
'J- en
CM CO
CO
1.40
en
30.22
cn en
CM
in
CO
CO
CO
2.47
CO
CD
1.82
CO
0-
115.19 81.70
CM
CO
CO
CO
CO
CM
CM' in CM
-50
<n
in CM
CM
CO
-58
CO
CD
CM
CO
CD
CO
CO
1st. Dev
cn
en
CM o
0.00
o en
m rr ^^ CM
*
2.35
CM
1.41
0.94
CM
cn
0.00
CO
CO
CO
-64
CO
*
27.87
CO
CD
CM
80.29
CO
CD
114.26
CM
CO
-62
14.30
en
CO
CM
CM
12.05
12.86
10.93
27.79
0.94
11,72
CO
o
o
CD
CD
80.26
CO
-61
a-
29.67
"J-
CM
26.83
^
o
81.73
1^ CO
CO
to
CM
1^
79.31
116.83
CO
1^
111 98
CM
-59
-65
to
CO
1^
26.11
en en
o
CM
00
CM
O
112.31
CM
27.45
CO
82.79
62.2
-59.40
19.25
CO
14.35
12.93
4.22
3.16
1.36
17.73
3.81
13.92
14.63
13.96
15.77
1.61
8.18
3.33
4.85
4.69
2.73
6.24
3.48
6.99
4.98
0.92
6.22
2.92
3.30
2.94
4.58
CM
-58
o
o
CM
Mean
CM
CO
CO
CO
tn
2.92
-69
CO
in
CM
CO
en
-57
en
CO
CO
CM
CO
CM
-59
00
00
75.92
CM
CM
CM
CO
CD
CM
114.65
cn
114.86
to
CO
CO
CM
o
o
r^
CM
-59
CD
CO
1.87
tn
29.37
en
1^
00
CO
CO
2.33
0-
oo
CO
00
to
81.73
o cn
o
*
to CM
114.90
^
o
-55
en
52.42
0.00
3.59
-58.8
15.65
-56.19
-62 08
-58,34
24 32
14,47
-58.74
0.006
in
CM
CM
-54
o
CO
3.47
^
CO
CO
CD
0.00
CM
CO
116.49
en 00
-55
a-
2.86
r^
96.05
CO
00
^
to
00
3.26
*
o
en
-59
r^
CD
CO
CM
CO
o> o>
CM o
StDev
en
00
r^
49.56
oo 00
CO
in
1.72
CO
CM
00
114.77
cn o o cn
en
1.14
CO
16.38
15.04
CO
CO
CO
co CO
00
-53
94.91
CM
CM
en
-53
en
i^ CD
en
CM
co CO
CN(
-55
1^
O
00
a>
i~-
CO
0.00
o
^
50.87
CO
r^
CD
CO
CO
-53
CM
CO
51.12
CM
CD
16.20
CD
14.99
CO
115.1
en
CD
111.85
CD
CO
-57
CD
cn
c(z)
00
CO
97.09
en CD
92.58
CD
CD
*
-52
o
en o
oo en r-o CM en
CO
-52
0.004
CD
en
-55
CM
CO
48.96
o
CO
46,08
un
114.4
CO
^
112.74
O
CM
1^
CO
s
00
96.41
CM
O
-61
0.002
50,4
0.003
CO
CM
98,64
118.24
0.001
CO
Mean
o
CO
CO
-55
-47
89.13
.006
m
o
-54
O
o
-53
C/3
49.94
0.005
Q.
0.003
CD
116.29
95.61
Q.
00
0.002
0.0
o
Triple tempering, 1(3)
o
<u
-57.
3
c
-a
-5c
Heat treatme
D.
O
(U
90 09
c
o
"S
00
CO
CM
to
h-
to
CM
T-
CD
<U
(U
CO
o
U
sas set i#
eo
CM
96.31
-58.26 44.23
74.38
72.08
45.01
48.42
46.81
44.69
-62.67 16.52
-61.81 13.69
-68.93 12.02
-63.58 17.73
-64.37 14.29
9.98
11.63
10.90
12.35
10.37
27.52
26.24
28.90
29.15
26.93
0.00
0.47
0.95
1.42
1.90
2.37
2.85
0.00
0.52
1.63
57.73
2.17
55.56
28.8
31.52
2.72
1.03
1.55
58.62
2.06
34.26
16.37
3.26
13.11
2.58
47.13
12.81
17.87
15.09
14.86
0.60
15.46
2.2543
-58.61
-53.55
-56.20
-55.77
0.00
-55.77
3.455
5.59
7,11
8.02
6.17
3.09
0.00
-56.19
-56.07
-57.06
-55.16
-58.31
-54.37
-69.70 28.42
1.269
12.02
25.84
to
52.99 40.83
CO
CD
CO
1 StDev 1 4.323 2.048 4.641 1.9952 3.203 1.542 0.981 6.3156 4.656 3.681 1.557
1.865
15.87
1.193| 6.0614
10.55 i -64.96
43.86
0.003
0.004
0.005
2.169
2.0039
13.42
0.52
1.56 1
5.24
1.04
6.97 1
8.19 1 5.98 1
38.41 1 31.26 1 19.21 1 10.05 1
18.93|
1
1
1
1
2.2975
7.45
2.70 1
2.25 1
16.51 1
1
1
1
1
1
4.7532 1
6.14
2.94
1.79
7.87
6.04
3.74
1.6088|
15.34 1
3.58
11.758]
12.85
14.258
16.25
17.68
13.20
12.94
12.31
13.17
2.3993
29.36
2.98
26.38
28.99
8.61 1 4.53
5.16
33.47] 21.18 1 10.38 1 3.83
31.09| 19.34 1 9.46 1 4.95
1.6294
27.17
1.80 j
25.372
24.17
27.65
23.42
26.91
24.69
25.39
2.1237
48.57
2.39
46.182
28.39| 18.69 1
0.9859
43.78
1.35
42.435
42.32
41.74
41.70
42.20
44.37
42.28
2.1604
62.81
1.79
61.017
47.06
29.011
16.61
9.81 1
44.79 1 25.52 1 13.91 1
37.82 1
34.95 1
44.55 1
11.69
61.61
58.04
41.31 1
35.58 j
10.77
12.85
36.26
13.10
12.90
0.006
12.38
2.0911
1.4557
0.90
49.63
0.45
8.50
48.73
51.34
46.82
48.08
46.01
49.91
50.21
2.4655
47.14
1.19
45.95
46.96
49.38
8.05
8.31
9.06
6.20
10.19
7.65
6.89
45,85
16.81
-54.82
8.14 -56.14
17.23 7.202
16.93
17.11
19.25
15.03
36.68
32.63 16.36
18.71
68.89 60.65
67.64 55.99
77.1
14.82
15.99
58.15 79.72
1.09
55.74 31.64
13.73
14.58
-52.13
0.002
12.00
0.001
-59.32
1 -59.27 45.14 96.64 75.74 48.48 29.99 14.10 -64.47 15.97 78.13 60.16 36.32 19.81 10.30 -56.19
lst.Dev 1 1.605 2.56 1.126 2.1261 1.461 1.162 0.997 2.5279 2.519 7.144 2.141 1.385 1.547 1.112 1.3886
1-65.35 53.98 77.17 59,31 43,91 24.91 9.43 -72.86 31.95 51.04 37.37 22.28 10.14 7.27 1 -58.88
1-63.09 1 55.03 88.43 63.99 37.95 20.67 9.03 -76.7 24.67 58.19 41.94 23.74 13.84 6.19 -72.44
1-64.94 1 54.98 86.16 61.85 35.13 22.89 7.36 -62.36 33.74 48.11 38.94 24.64 15.05 8.72 1 -62.68
1-61.64 1 49.86 88.47 59.76 38.14 23.17 10.31 -61.33 30.21 49.06 39.41 22.53 14.67 5.68 1 -63.76
-71.77 23.48 50.93 38.12 25.51 11.75 6.97 ! -59.58
1-54.03 1 52.35 80.05 62.41 40.53 22.81
1 -65.08 1 54.78 82.86 59.03 36.01 21.06 9.35 -73.2 23.15 53.92 39.15 22.96 13.07 8.39 1 -72.39
1 Mean 1 -62.36
1 83.86 61.058 38.61 22.59 9.027 -69.7 27.87 51.88 39.16 23.61 13.09 7.20 1 -64.96
1 c(z) 1 0.00 1 0.50 1 1.00 1 1.50 r2.oo 2.49 2.99 0.00 0.56 1.12 1.67 2.23 2.79 3.35 ! 0.00
5; CO
OS'ES 1
118
1 Mean -59.27 44.66 95.69 74.322 46.58 27.62 11.26 -64.47 15.46
96.82
-61.30 44.71
77.65
73.46
46.93
78.25
57.06 78.09
2.872 4.774
-65.44 18.48
1.74 3.9314
-50.77 16.16
19.71
12.30
2.558
27.00
0.00
0.54
-50.77 15.62
61.25
60.24 26.38
28.9 11.08
54.08 79.41
49.55
od
94.44
43.93 96.95
1.866
47.62
11.75
17.92
20.7
-57.39
18.87
-0.99
-47.57
21.19
51.14
10,95
17.98
-49.17
19.06 -53.63
20.33
00
-58.23 41.38
-57.48
72.49
75.87
1.9947
42.75
-1.78
44.54
44.89
41.12
TI
49.27 94.86
1.413
94.77
1.546
2.826
-61,13 44.46
-2.55
79.75
102.33 77.20
-2.52
104.85
82.21
48.53
42.49
44.80
T(3)
RS stress at depth or ...X*0.001"
o
-59.24
90.41
-1.47
91.88
-48.60 46.56
-0.38
92.61
0.00
102.40
91.08
-51.21 47.98
46.94
80.07
106.30
1 IVIean | -48.6
T"
78.93
79.11
12.11
0.006
54,07 26.72
0.005
0.004
59,97 73,14
21.14
0.002 0.003
12.27
-49.37
0.006
0.001
107.80
81.26
103.10
o
104.80
0.005
CO
45.38
a
-a
76.94
CO
CO
0.004
0.003
(U
104.70
CD
92.73 1
89.84 1
91.25 1
93.76 1
tn
0.002
cn
(U
0.001
St.Dev
0)
T(2)
-50.961 48.38 1
-43.581 46.13 1
-51.19| 47.06 1
-44.861 47.85 1
-49.80I 44.24 1
00
o
&
Heat treatment
sees
T(1)
c
o
'^
<u
o
APPENDIX B
ANALYSIS OF VARIANCE SAS CODE
* TEMPERING EFFECT;
Proc sort data=profile;
By grinding pass depth;
Proc GLM data=profile;
Class T grinding pass depth;
Model stress = T;
By grinding pass depth;
Means T/SNK;
output out=o2 r=resid;
Proc univariate normal plot data=o2;
Var resid;
by grinding pass depth;
119
* GRINDING EFFECT;
Proc sort data=profile;
By T pass depth;
Proc GLM data=profile;
Class T pass depth grinding;
Model stress = grinding;
By T pass depth;
Means grinding/SNK;
output out=o3 r^resid;
Proc univariate normal plot data=o2;
\'ar resid;
b> T pass depth;
* PASS EFFECT;
Proc sort data=profile;
By T grinding depth;
Proc GLM data=profile;
Class T grinding depth pass;
Model stress = pass;
By T grinding depth;
Means pass /SNK;
output out=ol r=resid;
Proc univariate noiTnal plot data=ol;
Var resid;
by T grinding depth;
run;
120
Table B.l. SAS output on hypothesis 1: assessing the overall effect of the
expenmental model.
Class
Levels
grinding
pass
rep
T
depth
3
6
3
7
Values
12
1 24
123456
1 23
0 123456
Number of observations
756
stress
DF
130
625
755
R-Square
0.996286
Sum of Squares
1411377.097
5261.605
1416638.702
CoeffVar
11.09168
Mean Square
10856.747
8.419
RootMSE
2.901477
F Value
1289.62
Pr > F
<.0001
stress Mean
26.15903
DF
5
2
6
12
5
10
30
60
Type I SS
9.927
76926.729
1217309.680
31134.753
38835.831
1436.375
40093.199
5630.603
121
Mean Square
1.985
38463.364
202884.947
2594.563
7767.166
143.637
1336.440
93.843
F Value
0.24
4568.87
24099.7
308.20
922.62
17.06
158.75
11.15
Pr>F
0.9467
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
<.0001
Table B.3. SAS output on hypothesis 2: testing the heat treatment effect.
grinding=l pass=4 depth=3
Class Level Information
Class
T
grinding
pass
depth
Levels
3
1
1
1
Values
123
1
4
3
Number of observations
18
DF
2
15
17
R-Square
0.964087
Sum of Squares
2503.531378
93.258450
2596.789828
CoeffVar
5.402443
Mean Square
1251.765689
6.217230
RootMSE
2.493437
F Value
201.34
Pr>F
<.0001
stress Mean
46.15389
Source
T
DF
2
Type ISS
2503.531378
Mean Square
1251.765689
F Value
201.34
Pr>F
<.0001
Source
T
DF
2
Type III SS
2503.531378
Mean Square
1251.765689
F Value
201.34
Pr>F
<.0001
Student-Newman-Keuls Test
Alpha
0.05
Error Degrees of Freedom
15
Error Mean Square
6.21723
Number of Means
Critical Range
2
3.0684066
3
3.739284
Mean
61.412
44.358
32.692
122
N
6
6
6
T
1
2
3
g r i n d i n g = l pass=4 depth=3
The UNIVARIATE Procedure
Variable:
resid
Nonnal P r o b a b i l i t y
4.S+
Plot
I
I
1.5-h
++"+-1-
-1.5+
+"-!-"
+ 1'++i'
I
I
-4.5-h
+-
- - + -
123
Statistic
0.973317
0.115231
0.034301
0.209775
p Value
Pr<W
0.8572
Pr>D
>0.1500
Pr>W-Sq >0.2500
Pr > A-Sq >0.2500
Levels
1
1
1
2
Values
2
1
2
1 2
Number of observations
12
DF
1
15
17
Sum of Squares
717.8080083
144.2232167
2596.789828
R-Square
0.832684
Coeff Var
5.861285
Mean Square
717.8080083
14.4223217
Root MSE
3.797673
F Value
49.77
Pr>F
<.0001
stress Mean
64.79250
Source
grinding
DF
1
Type I SS
717.8080083
Mean Square
717.8080083
F Value
49.77
Pr>F
<.0001
Source
grinding
DF
1
Type I SS
717.8080083
Mean Square
717.8080083
F Value
49.77
Pr>F
<.0001
Student-Newman-Keuls Test
Alpha
0.05
Error Degrees of Freedom
10
Error Mean Square
14.42232
Number of Means
2
Critical Range
4.8853767
Means with the same letter are not significantly different.
SNK Grouping
A
B
Mean
72.527
57.058
124
N
6
6
grinding
1
2
S+
ic
I
I
I
++"+"
I
-5+
+
+
-2
-1-1-+'''-h-i-'''
''++++>v
+
.^
^
-1
++++"
"'++'*+++
+*+"++
^
0
_^
+1
_^
_^_
+2
Statistic
0.963241
0.127946
0.032298
0.211147
125
p Value
Pr<W
0.8289
Pr>D
>0.1500
Pr>W-Sq >0.2500
Pr > A-Sq >0.2500
Table B.7. SAS output on hypothesis 4: heat treatment and grinding effects
combination.
= 1 depth-2
Class Level Information
Class
T
grinding
pass
depth
Levels
3
2
1
1
Values
1 2 3
1 2
1
2
Number of observations
36
DF
5
30
35
Sum of Squares
12910.98195
558.79535
13469.77730
R-Square
0.958515
CoeffVar
6.241790
Mean Square
2582.19639
18.62651
RootMSE
i.315844
F Value
138.63
Pr>F
<.0001
stress Mean
69.14433
Source
T*grinding
DF
5
Type ISS
12910.9819
Mean Square
2582.19639
F Value
138.63
Pr>F
<.0001
Source
T*grinding
DF
5
Type I SS
12910.9819
Mean Square
2582.19639
F Value
138.63
Pr>F
<.0001
Level of
grinding
Mean
-stress
Std Dev
94.9100000
91.8783333
72.5266667
57.0583333
52.5443333
45.9483333
126
3.47109781
1.41338483
2.46000542
4.77420325
7.92404144
2.46532283
pass=l depth=3
The UNIVARIATE Procedure
Variable: resid
Normal Probability Plot
5.5+
"+++
+"++
if il it it it
0.5+
it it it it it it
I It It It it it
it I it it Vr it
" + '++
-4.5+
''++++
-+
+
-2
+
-1
+
0
+
+1
+
+2
Statistic
0.983974
0.069969
0.029105
0.219773
p Value
Pr<W
0.8694
Pr>D
>0.1500
Pr>W-Sq >0.2500
Pr > A-Sq >0.2500
127
Corrected Total
DF
2
15
17
Sum of Squares
437.911744
124.0341
561.9458444
R-Square
0.779277
CoeffVar
3.178291
Mean Square
218.9558722
8.26894
RootMSE
2.875576
F Value
26.48
Pr>F
<.0001
stress Mean
90.47556
Source
pass
DF
2
Type ISS
437.911744
Mean Square
218.9558722
F Value
1 26.48
Pr>F
<.0001
Source
pass
DF
2
Type I SS
437.911744
Mean Square
218.9558722
F Value
1 26.48
Pr>F
<.0001
Student-Newman-Keuls Test
Alpha
0.05
Error Degrees of Freedom
15
Error Mean Square
8.268994
Number of Means
Critical Range
2
3.5386643
3
4.3123589
Mean
95.692
91.878
83.857
128
N
6
6
6
pass
2
1
4
5+
+++"++++
-1+
-7++++
++++++++
--++1
--++2
129
Statistic
0.945988
0.154877
0.067391
0.424584
p Valu
Pr<W
Pr>D
Pr > W-Sq
Pr > A-Sq
0.8656
>0.1500
>0.2500
>0.2500
APPENDD( C
MATLAB CODE IN THE CURVES FITTING
Ymax= 1 -e\p(-z. *w. *tmax). *sin(w. *tmax. *sqrt( 1 -z.''2)-i-atan(sqrt(( 1 -z.''2)/z)))/(sqrt{ 1 -z.''2))
k=2
tmin=(pi.*k)/(w.*sqrt(l-z.-^2))
t=0:0.2:tmin
Yt= I -exp(-z. *w. *t). *sin(w. *t. *sqrt( 1 -z.''-2)+atan(sqrt({ 1 -z.''2)/z)))/(sqrt( 1 -z.'^2))
RS=[-55.97
34.47
73.73
87.09
67.25
40.79
19.25]
H0=RS(1,4)-RS(1,7)
H1=RS(1,4)-RS(1,1)
fort=0:0.1:tmax
Yt= 1 -exp(-z.*\v. *t). *sin(w. *t.*sqrt( 1 -z.^2)-i-atan(sqrt(( 1 -z.'^2)/z)))/(sqrt( 1 -z.'^2))
Hlp=Ymax-Yt
HOp=(HO.*Hlp)/Hl
Ypr=Ymax-HOp
x=(tmax-t)/3
t7=tmax+3.''x
ift7>tmin
Y7=0
else
Y7= 1 -exp(-z. *w. *t7). 'sin(w. *t7. *sqrt( 1 -z.'^2)+atan(sqrt(( 1 -z.'^2)/z)))/(sqrt( 1 -z.^2))
end
ifY7>=Ypr
break
dispCfound "perfect fit" curve')
end
end
Y1=1 -exp(-z. *w. *tt). *sin( w. *tt. *sqrt( 1 -z.'^2)-(-atan(sqrt(( 1 -z.'^2)/z)))/(sqrt( 1 -z.'^2))
SSE=0
for 1=1:7
ifi<4
tcorr=tt-i-(i-l).*x
Ycorr=l -exp(-z.*w.*tcorr).*sin(w.*tcorr.'^sqrtC 1 -z.'>2)+atan(sqrt(( 1 -z.'^2)/z)))/(sqrt( 1 -z. 2))
RSact=RS(l,i)
RSpred=RS( 1,4)-((Ymax-Ycorr). *(RS( 1,4)-RS( 1,1 )))/(Ymax-Y 1)
else
\ c ^ 1 -exp(-z. *w. *tcorr). *sin(w. *tcorr. *sqrt( 1 -z.^2)-^atan(sqrt(( 1 -z.'^2)/z)))/(sqrt( 1 -z.'^2))
RSact=RS(l,i)
RSpred=RS( 1,4)-((Ymax-Ycorr). *(RS( 1,4)-RS( 1,7)))/(Ymax-Y7)
end
SSE=SSE-H(RS( 1 ,i)-RSpred).^2
Out=[RSact;RSpred]
end
130
131
=1
ma.x
ml'
Vi-c= ,
'
(-Cii-)c'
1
.
sin
4^
^F-7^+tan f V ^ l
V
f
VI - .;-' + tan
V^
+tan"
f
'
(0(
v ^ ( '1 - 4s "
TAA
:)V'-f' =0
Vl-C' -n
^A
fVi^
= eg
co(yj\-C
^ -
iOlsj] - C' +1
i-C'
' , ; ^ / ^ ^ + tan-'
CO.
sm
x/I - C "
r\\
x/
4^
^'g
clg{A)
x/1 - C '
fx/l-^'
V
'
- tan(/( + S ) =
1 - tan .4 tan S
r
^
1 - tan
(aif^y
V
r
^
I
1 -tan (^x/i-C')
^^r
'
/y
V ^
iV
(c^^V^)^^^^^
x/^
f
^
/
-^ tan tan"'
V^
i(cuf7l-f')-
x A ^
f
7
TAA
V
V
x / ^
TAA
tan
=tan{<.fx/^)
- ^ V ^^
(a>gV'-C')
V^
=tan(a)gV'-^')-
v^
tan
= 0
. n e Z,
Wyl\-C'
132
=;rn,...n eZ,
V r.^7^^
^^ _Vw
'
y;
APPENDIX D
Actual versus predicted RS patterns based on the best fit model - equal-area fitting
C/1
140
120
a> 100
c/l
80
"cS 60
3
40
-a ^0
0)
0
a:
-20
-40
-60
-80
UI
lyi
z=0.1
SS = 2.2723e+003
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
3.2000
-53.47
-53.470
4.2383
51.13
23.7672
5.2766
96.05
85.004
6.3148
116.49
116.49
7.3531
87.49
113.28
8.3914
52.42
79.56
9.4297
26.79
26.79
RSacl
*X.^Died
mi
Z=0.56
SSE = 604.1761
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
RS
3.80
-58.67
-58.67
5.69
54,77
72,64
7.58
115.19
115.19
9.47
81.70
97.28
11.36
50.02
56.03
13.25
15.15
30.22
27.83
18.62
18.62
R^Died
133
CO
.^
UI
I/I
100
80
60
13 40
3
a
^0
UI
UI
ai.
0
-20
-40
-60
-80
Z=0.76
S S = 141.0620
Depth below the surface
t value
suiface
4.9000
-59.52
-59.5200
RSact
RSpred
0.001"
7.2838
69.03
62.1363
0.002"
9.6676
87.25
87.2500
0.003"
12.0514
62.84
64.1432
0.004"
14.4352
40.62
33.1279
0.005"
16.8190
24.96
18.9842
0.006"
19.2027
15.88
15.8800
Figure D.3. Actual versus predicted RS patterns of single temperd specimens, finished by
grinding operations 1 in four-passes technique (Tl Grl P4).
CB
73.
vi
0)
(07
Z=0.11
SS = 2.0684e+003
Depth below the surface
t value
RSpred
surface
2.80
-48.60
-48.60
0.001"
3.97
46.56
18.05
0.002"
5.14
90.41
73.75
0.003"
6.32
102.33
102.33
0.004"
7.49
77.20
98.06
0.005"
0.006"
8.66
42.75
66.05
9.84
19.71
19.71
Figure D.4. Actual versus predicted RS patterns of single temperd specimens, finished by
grinding operations 2 in one-pass technique (Tl Gr2 PI).
134
Residi
40
20
0
-20
-40
-60
-80
Z=0.44
SS = 195.7099
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
3.5000
-59.27
-59.2700
5.2484
45.14
52.8695
6.9969
96.64
96.6400
8.7453
75.74
85.5546
10.4938
48.48
51.9581
12.2422
29.99
24.7420
13.9906
14.10
14.1000
R^act
RSpred
Figure D.5. Actual versus predicted RS patterns of single temperd specimens, finished by
grinding operations 2 in two-passes technique (Tl Gr2 P2).
pred
act
lUU
80
y ^ * * ^ ^
60
^^^^^
40
20
0
)
-20
-40
-60 1
0.001
0.002
0,003
0,004
0,005
0,006
0,( 07
-80
Z=0.62
SS= 103.7936
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
4.1000
-62.36
-62.3600
6.0541
54.00
51.6805
8.0081
84.85
84.8500
9.9622
62.55
69.5593
11.9162
40.61
38.3798
13.8703
15.8244
25.08
18.4234
12.02
12.0200
RSpred
135
s
3
't/i
Oi
0,(07
Z=0.17
SS = 1.4245e+003
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
2.6000
-55.97
-55.9700
3.8587
34.47
7.1045
5.1173
73.73
60.2676
6.3760
87.09
87.0900
7.6347
67.25
83.3664
8.8933
40.79
56.1076
10.1520
19.25
19.2500
RSacI
RSpred
Figure D.7. Actual versus predicted RS patterns of double temperd specimens, finished by
grinding operations 1 in one-pass technique (T2 Grl PI).
ca
-a
ai.
(07
Z=0.49
SS = 21.0527
Depth below the surface
t value
RSa,
RS pred
surface
0.001
0.002"
3.7000
5.4539
7.2078
-64.05
-64.0500
38.24
74.22
74.22
37.3797
0.003"
8.9617
63.35
65.2259
0.004"
0.005"
0.006"
12.4695
14.2233
42.38
23.34
39.9604
20.0367
12.02
12.02
10.7156
136
pred
act
J2
80
'Ol 60
>==-
VI
40
20
T3
1 ,._
^ -20
/ 0,001
0,002
0,00.1
0,004
0,005
0 006
0( 07
-40
-60
-80 '
-100
Z=0.76
SS = 205.7874
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
4.9000
-75.24
-75.2400
7.2838
43.85
35.5662
9.6676
58.44
58.4400
12.0514
46.16
43.2492
14,4352
31.56
22,8591
16.8190
20.84
13.5608
19.2027
11.52
11.5200
KOact
RSpred
0,(07
Z=0.18
SS = 741.3993
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
1.9000
-50.77
-50.7700
3.3958
16.16
2.1718
4.8917
58.15
53.5634
6.3875
79.72
79.7200
7.8834
57.73
74.3360
9.3792
31.54
47.2976
10.8750
16.37
16.3700
R^act
RSpred
137
act
pred
100
80
60
40
20
0
/ /
"
^ ? ^
-20
//OOOI
-40
-JL-
0,002
ooo.-i
3 004
0,005
0 006
Or 07
-60 1
-80
Z=0.44
SS = 594.5753
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
3.5000
-64.47
-64.4700
5.2484
6.9969
8.7453
10.4938
12.2422
13.9906
15.97
38.0962
78.13
78.1300
60.16
69.0202
36.32
41.4112
19.81
19,0454
10.3
10.3000
RSaci
RSpred
Figure D. 11. Actual versus predicted RS patterns of double temperd specimens, finished by
grinding operations 2 in two-passes technique (T2 Gr2 P2).
0.004
0.005
0.006
0,0
Z=0.53
SS= 12.3313
Depth below the surface
surface
0.001"
0,002"
0.003"
0.004"
0.005"
0.006"
t value
4.3000
6.3819
8.4638
10.5457
12.6276
14.7095
16.7914
R'^act
-69.7
-69.700
28.42
27.5574
52.99
52.9900
40.83
42.5474
25,84
24.0856
15.87
13.5121
10,55
10.5500
RSpred
138
pred
act
80
60
^
^
i
^
^
40
3
-a
ai.
20
0
-20
-40
//ooi
0,002
0,00.1
0,004
//
0,005
0,006
0,( 07
-60
-80
Z= 0.25
SS= 1.0930e-^003
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t \alue
3.1000
-59.4
-59.4000
4.2297
24.41
-1.0061
5.3595
53.81
43.0133
6.4892
64.28
64.2800
7.6190
50.57
62.6263
8.7487
30.44
44.0447
9.8785
17.73
17.7300
K b act
IXOpred
Figure D.l 3. Actual versus predicted RS patterns of triple temperd specimens, finished by
grinding operations 1 in one-pass technique (T3 Grl PI).
Z=0.47
SS = 120.0977
Depth below the surface
t value
RSa,
surface
3.6000
-60.93
-60.9300
0.001"
5.3592
18.64
26.3066
0.002"
7.1184
58.93
58.9300
0.003"
0.004"
8.8776
44.89
10.6368
51.4547
29.9309
28.34
0.005"
0.006"
12.3960
16.85
14.1552
6.22
6.2200
Figure D. 14. Actual versus predicted RS patterns of triple temperd specimens, finished by
grinding operations 1 in two-passes technique (T3 Grl P2).
139
act
^
pred
60
Z=0.67
SS= 131.1188
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
4.3000
-69.4
-69.4000
6.3819
25.51
17.8217
8.4638
40.63
40.6300
10.5457
34.36
32.6455
12.6276
24.91
18.5294
14.7095
15.77
10.4448
16.7914
8.18
8.1800
KOact
RSpred
Figure D.l5. Actual versus predicted RS patterns of tiiple temperd specimens, finished by
grinding operations 1 in four-passes technique (T3 Grl P4).
(07
Z=0.21
SS = 490.9401
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
2.5000
-48.6
-48.6000
3.8088
15.38
0.7405
5.1177
46.99
42.1484
6.4265
62.57
62.5700
7.7353
48.26
59.6668
9.0441
10.3530
28.98
40.0736
14.88
14.8800
RSpred
140
(07
Z=0.51
SS = 205.3301
Depth below the surface
surface
0.001"
0.002"
0.003"
0.004"
0.005"
0.006"
t value
3,7000
-56.19
-56.1900
5.5023
8,5
22.0752
7.3045
49.63
49.6300
9.1068
43.78
42.9774
10.9091
27.17
25.4693
12.7114
16.51
12.3257
14.5136
7.45
7.4500
RSaci
RSpred
0,003
0,004
0,005
0,006
0,(|07
Z=0.55
SS = 4.7921
Depth below the surface
t value
RS
RS'pred
surface
3.9000
-64.96
-64.9600
0.001"
5.8065
13.42
14.5606
0.002"
7.7131
39.45
39.4500
0.003"
9.6196
32.82
33.4679
0.004"
0.005"
0.006"
11,5261
21.29
20.1914
13.4326
12.64
15.3392
,2745
8.35
8.3500
141
PERMISSION TO COPY
Student Signature
Date
Student Signamre
Date