Download as pdf or txt
Download as pdf or txt
You are on page 1of 266

ROLL FORCE AND TORQUE IN THE HOT

ROLLING OF MILD STEEL FLAT SECTIONS

by

BRIAN KEITH DENTON, B.Sc.(Eng.), AOROSolio

A Thesis submitted for the degree of Doctor


of Philosophy in the University of London.

December, 1971 John Percy Research Group,


Department of Metallurgy,
Imperial College of Science
and Technology,
London, S.W.7.
2

ABSTRACT

Consideration of the available slip line field


solutions for hot rolling indicates that this method
may be used to predict theoretical loads and torques
with adequate accuracy and simplicity using simple
derived formulae, relating the load and torque functions,
P/kL and GAL2 to the geometry of rolling.

Other theories of hot rolling are compared on


this basis and it is shown that the difference, in certain
cases, is only very slight.
Experimental results, carried out at 900, 1000
and 1100°C, indicate that it is necessary to correct
for
a) the negative contribution to the torque
caused by that part of the roll pressure
distribution which lies beyond the line
of roll centres and
b) the effect of conduction of heat away
from the specimen within the roll gap.
A theoretical method is proposed for the latter.
The corrected results then show that reasonable
agreement with theory is obtained only when use is made
of a theoretical method incorporating:
a) mixed friction conditions,
a variable inhomogeneity factor in the
regions of dynamic friction and
c) a suitably low value for the coefficient
of friction.
ACKNOWLEDGEMENTS

The author wishes to express his deep sense of


gratitude to Dr. F. A. A. Crane for his constant advice,
encouragement and supervision throughout this project
and to Mr. T. Sheppard, for his help in a supervisory
capacity in the early stages.
Grateful acknowledgement is made to the United
Steel Cos. Ltd., (British Steel Corporation, Special
Steels Division), both for their financial support in
this work, and for the provision of the experimental
steel used in the investigation.
Thanks are due to Mr. A. J. Haynes and his
colleagues for their advice and technical assistance
and to all members of staff and fellow postgraduates
in the John Percy Research Group, past and present, who
have contributed to many helpful discussions.
Finally, the author would like to thank his wife,
Jennifer, for typing this thesis and for her continuous
devotion throughout the duration of his research, without
which it would not have been possible.
4

CONTENTS

ABSTRACT 2
ACKNOWLEDGEMENTS 3
FIGURES 8
TABLES AND PLATES 14
NOTATION 15
CHAPTER 1. Introduction 17
CHAPTER 2. The Theory of Rollie 19
2.1 Introduction 19
2.2 Early Assumptions 19
2.3 The Differential Equation
of the Friction Hill 20
2.4 Early Theories 23
2.5 Orowan's Theory of Rolling 27
2.6 Simplified Theories of Hot
Rolling 35
2.7 Roll Force and Torque 41
2.8 Other Theories of Hot Rolling 45
2.8.1 The Slip Line Field Theory 45
2.8.2 The Shear Plane Theory 51
2.9 The Effect of Roll Flattening 54
2.10 The Yield Stress in Hot Rolling 58
2.11 The Mean Strain Rate 62
2.12 Discussion of the Theories of
Hot Rolling 63
CHAPTER 3. The. Role of Frictionin_Hot Rollin 68
3.1 Introduction 68
3.2 The Measurement of Friction in
Rolling 69
3.3 The value of iR in the Hot
Rolling of Steel 70
3.4 The Validity of the Assumption
of Complete Sticking Frcition 71
3.5 A Simplified Rolling Theory
with Mixed Friction Conditions 74
5

CHAPTER 4. Interpretation and Comparison of


Hot Rolling Theories 80
4.1 The Slip Line Field Theory 80
4.2 Other Theories 87
4.3 Some Other Theories incorporating
the Aspect Ratio of the
Deformation Zone 101
CHAPTER 5. Experimental 104
5.1 The Rolling Mill 104
5.1.1 The Limitations of the Mill 104
5.2 Miii Instrumentation 105
5.2.1 The Load Cells 105
5.2.2 The Torque Transducers 105
5.2.3 Load and Torque Recording 105
5.3 Calibration Techniques 106
5.3.1 Calibration of the Load
Measuring Circuits 106
5.3.2. Calibration of the Torque
Measuring Circuits 112
5.3.3 Mill Spring Calibration 112
5.4 Preheating 114
5.4.1 The Preheating Furnaces 116
5.4.2 Temperature Distributions
during Reheating 120
5.5 The Composition and Condition
of the Experimental Steel 124
5.5.1 Grain Size Control 125
5.5.2 Control of Oxidation during
Preheating 126
5.5.3 General Physical Condition
and Surface Finish 130
5.5.4 Measurement and Accuracy of
Specimen Dimensions 131
5.6 Specimen Preparation 131
5*7 Experimental Procedure 133
5.8 Outline of Experimental
Rolling Programme 135
6

CHAPTER 6. Results 139


6.1 Calculation of the Load
and Torque Functions- 139
6.2 The Slip Line Field Theory 142
6.3 Other Results 146
6.4 Discussion 146
CHAPTER 7. The Effect of Roll :Quenching Lon
the Material within the Roll Gap 164
7.1 A Simpli*kd Model of Transient
Heat Conduction during the Hot

Rolling of Steel 164
7.2 The Time Spent within the Arc

of Contact 167
7.3 The Overall Average Temperature
Drop 169
7.4 The Effect on the Yield Stress,
kp 171
7.5 The Correction to k 177
G
7.6 Discussion 178
CHAPTER 8 ASimplified_Theory of Hot Rolling
incorporating Mixed Friction Conditions
and a Varying Inhomogeneity Factor
in the Regions of Coulomb Friction 186
8.1 Solution of the Differential
Equation of the Friction Hill,
in the Zones of Dynamic Friction 186
8.2 Calculation of the Total Roll
Separating Forae_ and Total
Torque 189
8.3 The Load and Torque Functions 191
CHAPTER 9. Discussion 203
9.1 Future Work 231
7


CHAPTER 10. Conclusions 233

REFERENCES 235
APPENDIX I 239
APPENDIX TT 242
8

FIGURES

Figure No.
1 (a)+(b) Elemental slice in the roll gap. 22
2 Distribution of normal and shear
stresses in the roll gap, given
by the theories shown. 26
3 (a) Pressure distribution for smooth
rolls. ( = 0.14) (from Orowan) 28
3 (b) Pressure distribution for rough
rolls. ( = 0.4) (from Orowan) 28
4 Nomenclature for compression between
non-parallel platens. 31
5 Illustrating calculation of
horizontal force in roll gap. 32
6 Orowan's inhomogeneity factor, w,
plotted as a function of a,

i.e. 4Ad_s 34
7 (a) Pressure distribution for smooth
rolls. ( =. 0.14) (from Orowan) 36
7 (b) Pressure distribution for rough
rolls. ( 0.4) (from Orowan) 36
8 (a) Simplified pressure distribution
due to Orowan & Pascoe. 38
8 (b) Relationship between n and 2R
suggested by Orowan & Pascoe. 38
9 Exit semi-fields for increasing
reduction. (constant R/t2)
(Crane & Alexander) 50
10 Three entry semi-fields for
R/t2 = 74.7, showing that the
'critical reduction' corresponds
to the 5/741° field type. 52
11 Distributioncfradial roll and shear
stresses in the roll gap. (Slip
line field theory.) 53
12 (a) Simplified 'slip line field' proposed
by Green & Wallace for the shear
plane theory. 55
9

Figure yo.
12 (b) Regions of plastic flow suggested
by °rowan's plasticine experiments. 55
13 Effect of roll flattening on the
position of the exit plane. 57
14 Effect of temperature and carbon
content (%) on the yield stress of
carbon steels. (from Tselikov) 59
15 Variation of strain rate in the roll
gaPo 64
16 Effect of rolling speed on the
coefficient of friction. 72
17 Variation of N with coefficient
of friction. 79
18 Relationship between r and R/t
2
for 5/7-r s.l.fo solution and that
incorporating a single cirular
discontinuity. 81
19 Relationship between the load
function and L/tm predicted by
sol.f. theory. 83
20 Relationship between the torque
function and L/tm predicted by
sol.f. theory. 84
21 Load functions for forging predicted
by s.l.f. theory. 86
22 Sims, theoretical load function
plotted against L/tm o 88
23 Sims' theoretical torque function
plotted against L/tm. 89
24 Orowan Pascoe's load functions. 91
25 Orowan & Pascoe's torque functions. 92
26 Alexander & Ford's load functions
for various reductions. 93
27 Alexander & Ford's torque functions
for various reductions. 94
28 Green a Wallace's load and torque
functions. 96
29 El-Kalay a Sparling's load function,
for AU.. 003 97

1.0

Figure No.
30 El-Malay Sparling's load functions
for /= 0.4 98
31 El-Kalay Sparling's torque
functions for ixk= 0.3 99
32 El-Kalay & Sparling's torque
functions for UL.= 0.4 100
33 Load cell circuit. (schematic) 107
34 Calibration of load cells on
Denison 50 ton testing machine 109
35 Calibration of load cell No. 1
on u.v. recorder. 110
36 Calibration of load cell No. ?
on u.v. recorder 111
37 Calibration of torductors.
Mill spring calibration. 115
39 The furnace cycle. 117
40 Furnace A. 121
41 Furnace B. 122
42 Effect of the duration of the normalizing
treat:tient on the grain size and hardness
of 0.30 in stock. 127
43 Typical load and. torque traces. 134
44 Measured (m) and corrected (in') lever
arm ratios for Series E (900°C) plotted
against L/tm. 141
45 Rolling geometries which were considered
to correspond most closely to the two
types of s.l.f. solution investigated. 143
46 Experimental and theoretical (s.l.f.)
load functions. (Series A) 144
47 Experimental and theoretical (s.l.f.)
torque functions. (Series A) 145
48 Experimental load functions. (900°C) 147
49 (1000°C) 148
50 (1100°C) 149
51 torque (900°C) 150
52 (1000°C) 151
53 (1100°C) 152
FiaureNo.
54 Experimental load functions indicating
the effect of reduction, r.
(Series 8, 1000°C) 153
55 Experimental torque functions
indicating the effect of reduction,
r. (Series B, 1000°C) 154
56 Corrected experimental lever arm
ratios. (900°r) 155
57 Corrected experimental lever arm
ratios. (10000C) 156
58 Corrected experimental lever arm :
ratios. (1100°C) 157
59 Lever arm ratios, various reductions.
(1000°C) 152
GO Load function plotted against
L/t (1000°C). 159
2
61 Load function plotted against
2L/(t1 + t7). (1000°C) 160
62 Relationship between 0/0 and 1 for
various values ofd\
i t-t/ 165
G3 From which the numerical integration
of I may be obtained. 172
64 The value of the integral, I, plotted
as a function of the reduction, r. 173
65 Effect of temperature on yield stress
of low carbon steel. 174
66 Effect of r, the reduction, on the
value of dkP'/d0. 175
67 Temperature corrected load functions,
(900°C) 180
68 Temperature corrected load functions,
C)
(1000° 181
rr, Temperature
3:) corrected load functions,
(1100°c) 182
70 Temperature corrected torque functions,
(900°C) 183
71 Temperature corrected torque functions,
(1000°C) 184
1.2

Fiqpre No,
72 Temperature corrected torque functions
(1.100°C) 185
73 The function F 2(s/k) 190
74 s/k distribution predicted by the
present work for R/t2 = 50, A = 0.3 192
75 Comparison of present method with that
of El-Kalay Sparling R/t2 = 100,
Al.= 0.25 193
76 Load functions predicted by present
method ( /UL= 0.2). 197
77 Torque functions predicted by present
method ( ~vt = 0.2). 198
73 Load functions predicted by present
method ( pk.= 0.3). 199
79 Torque functions predicted by present
method (/0--= 0.3). 200
80 Lever arm ratios predicted by present
method (,A,L= 0.2). 201
81 Lever arm ratios predicted by present
method ( JA,A.,= 0.3). 202
32 Comparison of theoretical and
experimental load functions. (Series B,
1000°C, r = 10). 205
83 Comparison of theoretical and
experimental load functions. (Series B,
1000°C, r = 205), 206
84 Comparison of theoretical and
experimental load functions. (Series B,
1000oC, r = 3W). 207
85 Comparison of theoretical and
experimental load functions. (Series B,
1000°C, r = 4M). 208
86 Comparison of theoretical and
experimental torque functions.
(Series B, 1000°C, r = 10;;')- 209
37 Comparison of theoretical and
experimental torque functions.
(Series B, 1000°C, r = 205). 210
13

Figure No
CC Comparison of theoretical and
experimental torque functions.
(Series ' 1000°C, r = 30) 211
89 Comparison of theoretical and
experimental torque functions.
(Series 13, 1000°C, r = 40;:) 212
90 Lever arm ratios predicted by slip
line field theory. 216
91 Helmi E, Alexander's experimental
load functions. (1000°C) 220
92 Effect of working in two phase region. 222

93 Possible yield stress distribution
through stock in the roll gap during
the hot rolling of steel. 230
Al Radial roll hardness profile. 241
14

TABLES AND PLATES

Table i(a and b). Showing the Effect of Specimen


Condition. 132

Table II. Geometries Investigated in


Series A (1000°C). 136

Table III. Geometries Investigated in


Series B (1000°C). 138

Table IV. Geometries Investigated in


Series E (900°C) 2,-rid Series
F (100°C). 138

Table V. The Effect of Rolling Speed


on the Load Function. 163

Table VI. The Intergral, I. 176

Table VII, Load and Torque Functions


predicted by present method

for .= 0.2 194

Table VIII. Load and Torque Functions


predicted by present method
for 0.3 195

Plate I. Experimental Set-up. 110

Plate II. Experimental Set-up.


(Furnace in rolling position). 119
15

NOTATION

theoretical function of /A

theoretical function of s/k


F2
G torque per unit width - both rolls
-1
H Bland and Ford's function 2jR \ tan
-E.--2 2
L length of the deformed arc of contact (L and
L are used to represent the length of the
undeformed and deformed arc of contact in
those cases where it is necessary to
distinguish between the two)
ii mill modulus

N mill speed (r.p.m.) and El-Kalay and Sparling's


inhomogeneity factor
P roll separating force per unit width
R deformed roll radius, (R and R are used to
represent the undeformed and deformed roll
radius in those cases where it is necessary
to distinguish between the two)
Y yield stress in compression
a A,vs/k
forging platen width (Hill)
c roll flattening constant
deflection on the u.vo recorder
f horizontal force in the roll gap
h heat transfer coefficient
k yield stress in shear
lever arm ratio
n position of the neutral plane in the arc of
contact
p horizontal pressure in the roll gap
vertical pressure in the roll gap
r fractional reduction in thickness (t1-t2)/t1
s normal roll pressure
t thickness of material at any point in the
deformation zone
16

v horizontal velocity in the roll gap


w inhomogeneity factor (from Orowan)
x, y roll gap co-ordinates

OK total angle of contact and thermal diffusivity


draft (t--t2 )
temperature
strain rate ( X mean strain rate)
coefficient of friction
./4A"
shear stresses and time
0 angular co-ordinate in the roll gap

Subscripts
av average
crit critical
tot total
m mean
relates to load considerations
G relates to torque considerations
o relates to exit slip/stick changeover point
i relates to entry slip/stick changeover point
n relates to neutral plane
1 relates to entry conditions
2 relates to exit conditions

Superscripts

indicates that the appropriate correction


has been made
relates to conditions between exit plane
and neutral point
relates to conditions between entry plane
and neutral point
17

CHAPTER 1
Introduction

Until about fifty years ago, the rolling of


metals was a prime example of an industrial process in
which the practical knowledge far outweighed the theoretical
understanding. Since the early man-, water- and steam-
powered mills, its development progressed by the method of
trial and error; failures of machinery pointed to the
weaknesses and defects of design that required modification.
However, at that time, the advent of new materials and the
demand for new products provided the stimulus required to
initiate the research which, it was hoped, would produce
some understanding of the rolling process.
The first aim then, as it still is, was to provide
a method of predicting accurately the forces involved in
rolling, and of estimating the effect on these forces of
the various parameters involved. This knowledge was of
the utmost importance to the producer, who wished to provide
a product as cheaply and efficiently as possible, and to-
the mill designer, who needed to estimate the stresses in
his machine and provide it with adequate power to carry
out the operation.
Since then much effort has been put into rolling
mill research. However, a majority of this effort has
been directed towards the subject of cold rolling. Since
this is essentially a finishing process, far greater
accuracy is required and much work has been devoted to
gauge control, to produce uniformity of thickness both
across the strip and along its length. Other problems
primarily concerned with cold rolling include the effect
of tension, both front and back, and the question of
lubrication.
Thus, partly out of necessity, hot rolling has
been relatively neglected. This neglect has been due
also to the difficulties associated with the experimental
work in this field and; to a certain extent, to the lack
of understanding of many aspects of the hot rolling process.
1.8

Recent trends towards higher outputs, lower rolling


costs and more stringent tolerances on the gauge of hot
rolled products have necessitated a certain amount of
automation in this process and, of late, hot rolling has
been given more consideration both from the theoretical
and the experimental point of view. But the primary aim,
that of producing a reliable theory of hot rolling from
which accurate predictions of roll force and torque can
be obtained, has not been achieved with great success and
consequently neither has the associated understanding of
the process.
In view of the complexity of the problem of section
rolling, only the simpler case of hot flat rolling under
plane strain conditions has been considered in the present
work.
CHAPTER 2
TheThfgry_of Rolling

2.1 Introduction
Any theoretical expression used in the prediction of
rolling load and torque must include factors to take account
of the following variables:
a) The geometry of rolling. This can be represented
in various ways, but the terms used must include s
in some way, the roll radius and the thicknesses
of the specimen at entry and exit. Most of
the theories describe geometry in terms of some
dimensionless function of the ratios R/ and r.
t2

b) The effect of friction.


c) The degree of inhomogeneity of deformation
in the process.
d) A measure of the resistance to deformation
exhibited by the material being worked.
Even considering the relatively simple case of flat
rolling without spread, the early workers in this field
found it necessary to use a number of rather drastic
approximations and simplifications. Some of these are still
used today while others have since been abandoned to provide
more exact solutions.

2.2 Early Assumptions


1) The rolled stock was assumed to consist of
thin vertical sections perpendicular to the direction of
rolling with no shear stress acting between neighbouring
elements. The plastic deformation of these segments was
assumed to be homogeneous compression. That is, the vertical
and horizontal pressures were assumed constant within any
one element and thus, initially plane cross-sections
remained plane during the process.
2) The coefficient of friction was assumed
constant throughout the arc of contact and most workers
assumed that the stock was slipping on the rolls,
20

i.eo dynamic friction was operative. Thus, the sliding


traction at the roll/stock interface was expressed simply
as the product of the normal roll pressure and the
coefficient of friction.
3) It was assumed that the yield stress of the
material remained constant throughout the process. This
is not true, since in cold rolling the material will work
harden as it passes through the roll gap and in the hot
rolling process, the yield stress is a complex function
of:
a) the rate of work hardening
b) the rate of thermal softening
c) the rate of working and
d) the temperature,
all of which vary through the arc of contact.
4) The Maxwell (Huber - von Mises) yield criterion
was assumed. Under plane strain conditions, i.eo rolling
with no lateral spread, the yield stress in compression,
Y, is equal to 1.15.1 - , where YQ is the yield stress obtained

in uniaxial compression.
5) The elastic deformations of the strip were
considered negligible in comparison to the plastic strain.
6) All theories assumed a circular arc of contact;
any elastic deformation of the rolls was assumed simply to
increase the effective roll radius.
In addition to these assumptions all authors
were forced to resort to mathematical approximations, of
one type or another, in order to solve their relevant
equations and simplify the method for general rolling mill
use.

2.3 The Differential Equation of the Friction Hill


This is obtained by developing the equation
representing the horizontal equilibrium of forces in the
roll gap.
However, it is necessary first to consider the
frictional stresses within the arc of contact. In the
case of rolling with no lateral spread, constancy of
volume requires
21

where v19 v2 and v are the horizontal velocities


of the strip at entry, exit and the point in the roll gap
where the material thickness is equal to t , respectively.
It should be noted that the value v will represent an average
through the section t, since the horizontal velocity of the
material is not constant over a vertical plane.

The horizontal component of the roll velocity will


have some intermediate value between v and v7, so the
1
roll surface moves faster than the strip in the plane of
entry and slower in the exit plane. Only at the neutral
plane will that component of the velocity of the roll
surface and the average stock velocity be equal. Thus,
the frictional forces act inwards in the arc of contact
from both entry and exit planes as shown in fig. 1(a).
This also illustrates an elemental slice of material
in the roll gap and the forces acting on this are shown in
fig. 1(b)0 If f is the total horizontal force, per unit
width of the stock, acting in a vertical plane at a distance
x from the plane of exit, s is the normal roll pressure
and r, the sliding traction at the roll/stock interface,
then from the equilibrium of the element
f + 2s. ox .sin 0 + 2 't x °cos 0 (f + ,f) = 0

cos 0 cos 0
(1)
if the forces are taken as positive when compressive and
acting in the direction of increasing x. Having regard to
this assumption, it can be seen by inspection that the
positive sign applies between the exit plane and the neutral
plane, while the negative sign applies between the neutral
plane and the plane of entry.

Rearrangement of equation (1) gives:


c!), f = 2sotan 0 + 2t

6x
or the limit as Sx tends to zero

df = 2s.tan 0 s 21:

dx (2)
22

fig_1(a)+(b) Elemental slice in the roll _g_a_a.

(a)
bcp

frictional
forces on
strip

X It

Plane .------
of exit
plane 1\
of entry

s. 6 x
cosh
-Lox
cos y

(b) f +Of

T.6 x
cos cp
s.ax
cos y
23

Using the assumptions 1 and 2 given in section 2.2


1) f = pt where p is the horizontal
pressure acting on a vertical
element.
2) "C -r-AA s
then:
d(pt) 2s(tan 0 + IA)

dx (3)

s may be replaced by the vertical pressure q,


s. dx .cos 0 i ;"dx .sin 0 = q.dx
cos 0 cos 0
i.e.
q = s(1 T hAtan 0) (4)

Assuming Maxwell's yield criterion under plane strain


conditions,
then:

1 -;2 =Y=2k

- 1 and [r'2 are the two principal stresses


where ev
and k is the shear yield stress.
In the elemental slic.in the roll gap the two principal
stresses are p and q

.°. q p = 2k (5)

Combining equations 3, 4 and 5 gives a differential


equation in q and x only, since both 0 and t can be
expressed in terms of x.

2•4 _TPLSJKTJIDries
Siebel(1) was one of the first workers to
recognise the importance of friction in the rolling
process but expressions similar to equations 2 and 3 were
(2) and have since
apparently deduced first by von Karman
become known by that name.
Siebel's "Theory of the Friction Hill" explains
generally the effects of the various parameters on the
24

specific roll pressure, but his method represents only a


rough approximation. He obtained a solution to the '
differential equation by using small angle approximations,
(i.e. cos 0 = 1 and sin 0 = tan 0 = 0), and, for the
purposes of calculation, he assumed the vertical pressure,
q, was constant, through the arc of contact, and equal to
2k. The result was a considerable underestimate of the
roll pressure, since the accumulative effect of pressure on
the friction hill was neglected.
Von Karman replaced s directly by q in equation 3
and, neglecting the ,,tan 0 term of equation 4, produced,
in its final form, the expression:

d(pt/2) = q(tan 0 + ,u.)


dx
while p and q were interrelated by considering the Maxwell
yield criterion as has been shown. To obtain the total roll
force, this solution involves tedious numerical integration
and other workers developed methods which simplified this.
Tririks(3)
assumed the arc of contact to be parabolic
and, thus, prepared two sets of curves which enable the
peak pressure and the average roll pressure to be found
directly, from the reduction, and the ratio of the tangent
of the friction angle to the tangent of the angle of contact.
(4)
Tselikov obtained his solution by restricting
its application to the se al contact angles associated
with the rolling of thin strip and, thus, was able to assume
0 to be constant and equal to half the contact angle. This
method also incorporated the effect of front and back tension,
and considered the possibility of a zone in the centre of
the arc of contact where slipping friction was not operative.
In this region Tselikov assumed the velocity distribution
to be parabolic and obtained a solution by applying Newtoris
law of motion for viscous fluids. He attempted to
substantiate his method by comparing its predictions with
the experimental data published by Skefko Ball Bearing
Company Limited (5) and Rudbakh and Severdenko(6)
Nadai(7) investigated the effect on the specific roll
25

pressure distribution of various assumptions regarding


the frictional drag, He considered three hypotheses:
1)
2) "r. is constant
3) 11-, is proportional to the relative velocity
of the strip and roll surfaces,
and obtained a solution representing each case.
Ekelund(8) assumed Coulomb friction over the whole
entry region, while on the exit side he considered there
to be no relative movement between the strip and the roll
surface. His method is semi-empirical since it contains
a large number of approximations, often quite arbitrary,
which were selected to obtain agreement with the results of
certain experiments on the hot rolling of steel.
Fig. 2 shows the distribution in the roll gap of the
parameters q and '5, as calculated by Tselikov by the methods
due to himself, Siebel, von Karman and Nadai (case 3), and
given by El Waziri(9)
The assumptions of homogeneous compression i.e. plane
ructions rcmaining plane, complete Coulomb friction conditions,
i.e. 1. . fk,ks, and constant yield stress, all contributed
to the general inability of the early theories to predict
accurately the rolling forces. In certain cases, where
agreement seems to be obtained, the success may be attributed
to the fact that relatively unkown parameters such as the
coefficient of friction and the yield stress, were adjusted
to give good agreement. For instance, Tselikov found it
necessary to put /A.= 0.6, to obtain agreement between his
method and the results of Rudbakh and Severdenko, who were
considering the rolling of dry cold steel, whereas Pomp
and Lueg(10) suggest a value of about 0.1 for these
conditions.
The methods due to Trinks and Tselikov may be adapted
to take into account a varying yield stress through the arc
of contact, and also a variation of p„, but this necessitates
tedious point to point summation of von Karman ,s equation.
Mathematical approximations also introduced a certain
amount of error in the early theories, and for this reason
26


0 0.5 1.0
Fraction of the arc of contact

von Korman

Nadal

fig. 2 Distribution of normal and shear stresses in the


roll yap, given by the theories shown.
27

Orowan proposed his "Homogeneous Graphical Method" of


solution(11)

2.5 Orowan's Theory of Rolling


This theory enabled the roll pressure distribution
to be obtained by a graphical method, with varying yield
stress and coefficient of friction. The only assumptions
made were those of homogeneous deformation and dynamic friction:
no mathematical approximations were incorporated. By use
of this method, Orowan showed the effect of the earlier
approximations and by comparing the predicted pressure
(12)
distributions with those measured by Siebel and Lueg
obtained some idea of the effect of the two assumptions
he had used.
Fig. 3(a) shows that the homogeneous graphical
method is reasonably accurate when pis relatively small,
but at higher values of fig. 3(b), even this grossly
over-estimates the radial roll pressure. In the light of
this, Orowan abandoned the assumptions of plane sections
remaining plane and of Coulomb friction.
In the theories discussed so far, the frictional
stress 't has always been put equal to AAs; however, the
maximum value that T can ever achieve is k, the yield stress
in shear since unloading by plastic shear at, and below, the
surface of the stock prevents values greater than this from
being attained.
Thus Orowan assumed that dynamic friction occurred
in the roll gap only where Aks<k. If ' exceeds the shear
yield stress of the material, any velocity difference between
the stock and the roll surface is accammodoie_ci mainly by
internal shear within the stock. Orowan termed this case,
'sticking' friction since he suggested the stock surface
would 'stick' to the rolls and would be extruded from the
roll gap by plastic deformation. Unfortunately, the term,
'sticking' friction is a misnomer, since even in the zones
where q7= k there isEtill relative movement between the
stock and roll surfaces. Only at the neutral point, is
there no velocity difference. Of course, Ekelund had tried
to take this into account but his assumptions had been
28

fig.3a. Pressure

distribution for

smooth rolls. (la =014)

(from Orowan)"

1 2 3 4 5 6
t.p-deg.
Lueg's measurement - - - - - von Korman

— Homogeneous - Siebel
Graphical

300

fig.3b. Pressure

distribution for

rough rolls.( p, =0.4) 200

(from Orowan)

kg/mm2

100

0
0 1 2 3 4 5 6
p-deg.
29

incorrect, since, if 'sticking' occurs, it appears first


at the neutral point, and the region expands, with increasing
coefficient of friction, towards the entry and exit planes.
Nadai had also realized that gs cannot be greater than k.
To take into account the inhomogeneity of deformation
in the rolling process, Orowan used, as his starting-point,
the solution obtained by Prandtl(13) for the compression
of a plastic body between perfectly rough parallel platens.
Prandtl obtained the following expressions for the vertical
pressure, s, the horizontal pressure, p, and the shear stress,
T. in the vertical or horizontal planes:

s = C + 2k.x
t (6)

2 2 1-
P = C + 2k.x - 2k(1 - 4y it ) 2
t (7)

2k .y

t (8)

where C is a constant
t is the distance between the platens
and x and y are the rectangular co-ordinates
with their origin at the centre of the slab.
Thus the expression:
2 2 11-
P= s 2k(1 - 4y /t )2 (9)

relates the vertical and horizontal pressures.


However, since the roll surfaces are neither plane
nor parallel, Orowan found it necessary to resort to the
extension of Prandtl's method, due to Nadai(14), for
compression between inclined platens. Nadai replaced the
rectangular co-ordinates used by Prandtl, with polar
co-ordinates, r, G, and obtained the expressions:
2 /02 )2 (10)
p s 2k(1 -

and L 2k .9
20
30

where s represents the, azimuthal pressure on a surface


element parallel to the radius vector r, in a direction
perpendicular to r and p the radial pressure acting on a
surface perpendicular to r, in the direction of r, as shown
in fig. 4.
Equations 6 - 11 due to Prandtl and Nadai refer
to the case where T.= k, i.e. 'sticking' friction. Orowan
obtained the following expressions for the case where
= <k
2psoy
t (12)

2 2 -1 (1.3)
P = s - 2k(1-4(ps/k) (y/t) )2

for parallel platens and

-r =
(14)
0
2 2 -71-
P= s 2k(1-(tWk)(9/0) )2 (1.5)

for inclined platens.


Having found equation 15, he continued by relating this to
the rolling process to find the horizontal force, f.
Fig. 5 shows an arc AB in the roll gap. Orowan
assumed that the stress distribution over AB was described
by equations 14 and 15.
The length of any element on AB is rod@ and
r = t/2.sin 0, thus, considering a unit width of material,
the area of that element, dA, is
t dG
2.sin 0
f is the horizontal component of the force which acts across
the surface AB. This consists of two parts, one representing
a contribution from the radial stress, p, and one from the
shear stress, L 7
The first is:
p.cos G.dA
31

Sou
..... 4.•....(1)
.....,
•S,
os.
So

fiq. 4. Nomenclature for compression between

non- parallel platens.


32

fig.5• Illustrating calculation of horizontal force in

roll qap.
33

while that due to the shear stress is


Tosin Q.dA
Hence, f + f_

where fp .1 pocos G.dA


1
j-0


and fT =f .sin GodA

Orowan showed that

is tkw

where

4)

2 2
W — — 2 1 cos Q (14 (16)
(1 - a (Q/0) )2
sin 0
0

and

a = /A's
k

Fig. 6 shows w plotted as a function of a for


0 = 0 and 3011 , this shows that w is relatively independent
of 0 for all ranges covered by practical rolling. For
the shear stress component, Orowan found

f - t

and combining these obtained the expression



is tkw + t f.ts (17)
')I-ca-net; I

He then considered the two cases of slipping friction


and 'sticking' friction. He suggested that where slipping
34

0
0 0.5
a

fig.6. Orowan's inhomoaeneity factor, w,

plotted as a function of a, i.e. p.s :._


---1-c
friction occurs the shear stress contribution could be
neglected so that the approximate formula:

f = t(s - kw) (18)

could be used with little loss of accuracy. In the sticking


friction case a = 1, and w = 7(/2, therefore:

f = t is - k(if/2 T 1/2(1/0 - 1/tan 0))! (19)

By the use of these two expressions together with


equation 2 modified to give 0 as the independent variable,

i.e.

df = 2R,s,sin 0 + 2R, ",cos 0, (20)


d0

Orowan obtained a complete solution to the differential


equation of the friction hill. The pressure distribution
was then found by step-by-step calculation through the roll
gap. He demonstrated the superiority of the method to great
effect by comparing it with those pressure distributions
measured by Siebel and Lueg, as shown in figs. 7(a) and (b).

Simplified Theories of Hot Rolling


Until Orowan's work was published no distinction
had been made between hot and cold rolling; however
Orowan suggested that generally the coefficient of friction
was high enough in hot rolling, when no lubrication was
used, to assume that 'sticking' friction occurred over
practically the whole arc of contact. Since that time,
complete sticking friction has been adopted as a criterion
of hot rolling by most workers in this field, whereas theories
of cold rolling have alsways assumed that slipping friction
applies, i.e. =yAs.
Orowan was aware of complications of his technique
for everyday use, and suggested that its accuracy was higher
than that obtainable in the experimental determination of
36

t q 7a. Pressure

distribution for

smooth rolls. (µ :0.14)

(from Orowan)

1 2 3 4 5 6
deg.
Lueg s °rowan's method
measurement

100 I I I

kg imm2

fia.7b. Pressure

distribution for 50

rough rolls. ( 117-0.4)

(from Orowan)

0 1

0 1 2 3 4 5 6
- deg.
37

the physical quantities involved, i.e. yield stress and


coefficient of friction. Hence, he suggested that the
simplified method of Orowan and Pascoe(15) was adequate
for most purposes. These workers suggested that the specific
pressure distribution could be represented by a series of
straight lines so that the roll pressure diagram would thus
consist of a rectangular plinth, representing the deformation
of the material, surmounted by a simple triangular friction
hill, as shown in fig. 8(a). The apex of this diagram
marked the position of the neutral plane. In effect, these
investigators replaced the curved roll surface with flat
parallel platens in the region of the exit plane. When
'sticking' friction is assumed, i.e. 17= k, equation 2
at the exit plane (0 = 0) reduces to

df 2k
dx
or
dp 2k
dx t2 (21)

Assuming s = q, then

s - p = Tr.k (sticking friction)


2 (22)

i.e. ds = dp

and equation 21 becomes:

ds 2k (23)
dx t2

i.e. tan 0‹ (fig. 8(a)) = 2k


t2

By analogy with the forging case, t was assumed constant


and equal to t2, thus integration of equation 23 gives:
3B

fig.8a.. Simplified pressure

distribution due to

Orowan & Pascoe.

2R

fiq.8b. Relationship between n and 6 suggested by


/ 2
Orowan and Pascoe.
39

s = s exit 2k.x
t2

At the exit plane p = 0, in the absence of tension, then


from equation 22, s = 71-/2k


Thus: k(AL ± 22-C)
2 t2

Orowan and Pascoe replaced 1T/2 by 1.6 and used the expression:

= 1.6 2x
(24)
k t2

between the exit plane and the neutral point. The position
of the latter was obtained by Orowan from his more exact
theory and is :,mown in fig. 8(b) related to the parameter
(t1 t2)/2R; it was suggested however that n may be put
equal to 0.5L without much loss of accuracy.
Using equation 20, a form of von Karman's equation
due to Orowan Sims(16) derived an expression for
d(s/k)/d0 which could be directly integrated. He assumed
0 was small so:
sin 0 = 0
cos 0 = 1
and thus
df = 2Rs 0 t 2Rk
d0
when = k
Sims then substituted a shortened form of equation 19,
which neglects the shear stress contribution to f.

i.e.
f = t(s - Tr/2k)
Then:
d (t(s -- 1172k)) = 2Rs 0 2Rk (25)
40

If k remains constant over the arc of contact

dO (t ( s/k Tr/2) ) = 2R 0 s + R
(26)

Now
t2 + 2R(1 cos 0)

or
2
t = t2 + RO

since when 0 is small

- cos 0 02
9

Thus
dt = 2R0
dO

and equation 26 becomes:

R R
(tz RtV) L (s/i:
riz'p

or rearranging::

(27)
R T t R

i4 t2 4 43-1- Z + R.4)2

This gives:

2
1,1r % G,(5, ) ( t 2\(11
z 128)

as a solution from the plane exit to the neutral plane, and


41

(3, 1)2 ) „)
L -
C>)— 1-r; 11; tn. -I- 2 \
(P7
it-21 \ t2

as a solution for the entry part of the roll gap, where o(


is the total angle of contact. The position of the neutral
plane, On, is obtained by equating equations 28 and 29.
Alexander and Ford(17) modified the method of
Orowan and Pascoe. They assumed the slope of the friction
hill on the entry side could be put equal to 2k/t17 while
maintaining the value of 2k for the slope at the exit
/t2
side, as suggested by Orowan and Pascoe.

Thus, while

k + '1( (30)

The position of the neutral point does not appear


explicitly in the method but the inference is that

t2
n
.L
(31)
t1 + t2

2.7 Roll Force and Torque


The total roll force may be written:

q d+ (32)

or
q dx. (33)

42

These expressions become:


a
( 'Pa (34)

'111
r Xn
= S ci;c (35)
JO )
if the difference between s and q is neglected, and the
exit and entry solutions are considered separately.
In the theories which incorporated varying yield
stress and/or varying coefficient of friction through the
roll gap, integration of equations 34 and 35 could only
be effected by tedious summation of the area enclosed by the
pressure distribution. This was also the case when expressions
obtained for s or s/k could not be integrated analytically.
If the yield stress of the material is assumed constant,
or put equal to a mean yield stress, k, equations 34 and
35 become:

4)" ,\
P ts + (36)
0
and

Px. >14) d 4- (37)

In some cases workers have obtained expressions


for and N) which could be directly integrated and,
thus, have produced an overall expression for P, the total
roll force.
Tselikov gave the following expression for P:
43

(38)
kL)
((r-t - ea)(g

+4A 04/2
This expression was further simplified by Tselikov and is
given by Wusatowski(18) as:

kL 4-2r (39)
r 8' ( r

The semi-empirical method of Ekelund produced the


equation:

kLt2 + 3 VA - .1+ (ti - t) (40)

( , t- s

while those due to Orowan and Pascoe, Sims, and Alexander


and Ford give, respectively:


rt. 1-4)
(41)
k L

-)
-rr +col_r - 5 t 1(111 ( L.t.1
1 -4-1 I I(.,j

42)

.)) (43)

The.total torque required to carry out the operation


may be obtained by integrating the shear stresses at the
roll/stock interface.
0(

2 R. c Jci) (44)
•ci), (1)
44

N.B. The factor 2 is included to obtain the total torque


for both rolls.
Naturally, ris put equal to Ats or k depending
on the assumed friction mode. In the case of complete sticking
friction equation 44 reduces to

2
G = 4R -k(A/2 - On) (45)

if k is assumed constant.
If complete dynamic friction is assumed then

<DA

.(1 4
2R s" S4 ci (46)
10

Both of these methods are subject to a certain amount


of error. They both require accurate knowledge of the position
of the neutral point. The torque calculated from equation
46 relates directly to that factor in square brackets,
which is generally the difference of two large quantities
of similar magnitude, so that a small error in the estimation
of either s4" and s- will have a relatively large effect on
the accuracy of G.
Roll torque is often calculated as the moment of
the vertical pressure with respect to the roll axes.
In this case:
2 qxdx (47)

which approximates to
ac
2
2R sOd0 (48)
0
This method neglects the contribution of the
horizontal component of the force acting on the roll
surface, but is not subject to the sources of error
discussed in the first method. The total horizontal force
vanishes in the absence of tension, but the lever arms of
the horizontal forces with respect to the roll axes are
greater on the entry side than on the exit side, so the
moments of the positive and negative regions do not cancel.
45

For contact angles greater than a few degrees,


Orowan suggests that this method may give a value 10% or
20% lower than that calculated by the first method. On the
other hand, he points out that a small error in the calculation
of the neutral point may cause considerable errors in the
torque calculated from equations 45 or 46, and suggests
it is probably safer to use the second method if the angle
of contact is not excessively large. Of the theories
discussed so far, expressions for overall torque have been
given only by Orowan and Pascoe, Sims and Alexander and Ford.
Sims using the first method discussed, obtained an expression
identical to equation 45, while Alexander and Ford, using
the second technique, obtained the following expression:

G = kL2(71(/ + 2 / 2r) L 4 (49)


2 '3(2 --) 1 0 2)

A similar equation may be obtained for the method


due to Orowan and Pascoe.

i.e. G = kL2(1.6 + 2/3n(n+1)L/t ) (50)


2

206 Othor Theories of Hot Rolling


Until now, only those theories deriving directly
or indirectly from the differential equation of the friction
hill have been considered. The slip line field theory and
the shear plane theory, the latter being considered as a
greatly simplified version of the former, offer alternative
routes to the prediction of roll force and torque.

2.8.1 The Slip Line Field Theory-


A slip line field consists of two sets of mutually
orthogonal curves which indicate the instantaneous orientation
of the maximum shear stresses at any point within the
deformation zone of a particular working process. The two
families of curves are designated o< and p according to
some appropiate convention.
The state of stress at any particular point within
Li 6

the deforming zone is made up of two components,


1) a hydrostatic pressure,
and 2) the deviatoric component.
In the slip line field theory the material is assumed to
be rigid-perfectly plastic, i.e. it is assumed to have an
infinite Young's modulus and zero work-hardening coefficient,
thus the deviatoric component will always be equal to the
shear yield stress of the material.
The Hencky equations(19) as the stress-equilibrium
equations expressed in curvi-linear co-ordinates are known,
may be written:

p 2k0 = constant along an o.. line (51)


p + 21(0 = constant along a p line (52)

where p is the instantaneous value of the hydrostatic component


of stress and 0 is the angle made by the corresponding slip
line with some suitably defined direction of reference. Use
of these equations enables the state of stress to be calculated
at any point in a deforming zone provided that:
1) some boundary condition is known that
may be used as a starting point,
and 2) the particular field for that zone is
known.
If the boundary conditions involve stresses only,
the problem is statically determinate and the solution
presents little difficulty; however, for statically
indeterminate problems, where both stress and velocity
boundary conditions exist, solutions for both stresses and
velocities mur.,t be considered together.
In the case of plane strain, where the strain in
one co-ordinate direction is zero, compatibility requires
that the strains in the other two directions be related
thus:

dE + dE 2 = 0
47

It follows from this that the rate of extension along


any slip line is zero, Expressions analogous to those of
Hencky were obtained for the velocities by Geiringer(20):
du -•- vd0 = 0 along an of line (53)
dv ud0 = 0 along a p line (54)
where u and v are the velocities along the of and p lines,
respectively.
Slip line fields are extremely difficult to calculate
by analytical techniques. However, the geometrical method
due to Prager (21) which gives a convenient representation
of the solution to any problem, has allowed the solution
of problems with mixed boundary conditions, to be obtained,
sometimes without much difficulty, and this method is most
commonly used at present for producing slip line field
solutions.
The graphical construction of a solution by this
method takes the form of three diagrams: the physical plane,
the stress plane and the hodograph, representing the slip
line field, the state of stress of points on the slip lines
in the field and the instantaneous velocity of those points,
respectively.
Prager showed that in moving along a slip line, the
Hencky equilibrium equations require that Mohr's circle
ofstress roll, without slipping, along the tangents given
by 1-k, thus, each slip line in the physical plane is
represented by a cycloid in the stress plane. Corresponding
elements of the cycloid and the slip line are at right angles.
The Geiringer equations require the existence of the pattern
of lines constituting the hodograph, and their correspondence
with the slip lines in the physical plane. The hodograph
consists of an orthogonal network of lines having elements
normal to their corresponding element in the slip line field,
and the magnitude and direction of the velocity of any point
in the physical plane may be obtained by joining that point
to the origin of the hodograph, thus producing the related
vector of velocity.
The solution by Prager's method is graphical in
nature and consists of obtaining a slip line field with a
48

satisfactory stress plane and hodograph, which meet all


the requirements of the boundary conditions.
Alexander(22) was the first to obtain a solution
to the problem of hot rolling. Using the assumption of
complete sticking friction in which the slip lines meet
the roll/strip interface either tangentially or normally,
Alexander considered the process as compression between
inclined curved rough platens, in which material is extruded
towards both ends of the arc of contact. As the rolls are
rotating, there is an additional rotational velocity super-
imposed on the 'extrusion' process, which must be taken
into account when considering the corresponding hodograph.
As a starting point, and in analogy with the
compression of a plastic slab between rough parallel platens,
a solution which had been obtained previously by Hill(23) 7
Alexander assumed a centred fan of slip lines around the
point of contact between the rolls and stock at the point
of entry. However, in order to satisfy the velocity boundary
conditions, it was necessary to use a slightly curved centred
fan, so that the nearest slip line to the roll surface
became tangential to it, thus defining a thin slice of rigid
material in that vicinity. There was a large velocity
discontinuity along the boundary of this zone, resulting
in a band of intense shearing.
On the exit side, it was necessary to use a curved
slip line to meet the roll surface tangentially.
Alexander and Ford (17) using limit analysis, proved
7

that the proposed solution was satisfactory, by showing


that the rate of plastic work was everywhere positive and
the yield criterion was not violated in the rigid zones.
The solution was for a single geometry of rolling, in
which the R/t value was 86.7 and the reduction was 331%.
2
To extend the method and eliminate some of the
difficulties of the complete solution, Ford and Alexander(24)
assumed a straight slip line in the entry plane, which meets
the stock centre line at 450 , and a circular arc slip line
in the exit region, which meets the centre line at 450
and the roll surface tangentially. They obtained solutions
49

for three more geometries.


Crane and Alexander(25) proposed slip line field
solutions for a range of geometrical configurations. The
method used by these workers was first to construct the
exit semi-fields since the situation at the roll surface
here is less complicated than at the entry side, and
provided that the reduction is not too great, the roll
surface in contact with the exit semi-field can be
represented by a straight line tangential to the roll
surface at exit; an assumption which, in addition to
simplifying the stress plane, makes the exit semi-field
R
and semi-hodograph independent of the /t ratio. Thus,
once a particular type of exit semi-field 2has been
determined, it will suit a whole range of reductions as
the ratio R/t is allowed to vary.
2
The simplest form of exit semi-field considered
by those workers consisted of a single circular arc crossing
the stock centre line at 45° and meeting the roll surface
in the exit plane tangentially as shown in fig. 9(a).
As the size of the plastic field increased additional slip
lines were added, figs. 9(b) and (c). 710 was chosen as
the angular increment turned through by the various segments
of the slip line, since this value was,
a) small enough for reasonable accuracy,
b) not too small to involve too much
labour during construction of the field
and c) a simple factor of 45°, the total angle
turned through by the exit slip line.
The exit semi-fields were thus designated according to
the number of slip lines, e.g. 3/71° and 11/71° .
Semi-fields and exit semi-hodographs were obtained
for the solutions incorporating the single circular
discontinuity, and for the types 3/71°, 5/71°, 7/71° , 9/73°
and 11/71°, while the complete solution was found for each
R
type using an /t ratio of 74.7.
2
The principle of this method was to produce the
exit semi-fields 7 then , choosing a suitable value of the
R/t2
50

- A
to roll
centre

(a)
).`

(b)

(c)

4.04


/ •

fig.9 . Exit semi-fields for increasing reduction.

(constant R/ ) (Crane & Alexander)


t2
51.

ratio, to complete the solution to find the corresponding


value of reduction for that particular field type. It
was not possible, generally, to decide initially on a
particular geometry and then determine the appropriate
field.
Crane and Alexander found there existed a critical
reduction for each R/
t ratio in which the entry
.2
discontinuity just briefly touches the roll surface before
carrying on to define the central rigid zone. At lower
levels of reduction, the discontinuity leaving the roll
surface at entry does not rejoin it again, while at
higher reductions, there are two distinct rigid zones, as
shown in fig. 10(a), (b) and (c)(25)(26) The critical
reduction was found to be close to the semi-field type
5/7:1°, so this type lends itself to rapid completion with
little loss of accuracy, without the necessity of drawing
the hodograph. The type containing the single circular
discontinuity is also relatively simple to complete since
the exit semi-hodograph degenerates to a point.
Calculation of the specific roll pressure may be
effected by use of the Hencky equations and the total roll
force and torque is obtained by numerical integration.
The torque is found either by integrating the shear stresses
(k) at the roll/stock interface which is subject to the
errors discussed in section 2.7 or more easily by obtaining
the moment of the vertical pressure with respect to the
roll axes. Fig. 11 shows a completed field, together
with the distribution of s and 'C in the arc of contact.

2.8.2 The Shear Plane Theory


This theory, based on the observations of
localized deformation in the rolling process, was proposed
by Green and Wallace(27). It is derived from the limit
analysis technique of Westwood and Wallace(28) , who
considered the equilibrium of quasi-rigid blocks of material
on the boundaries of which the shear stress was limited
to k, the shear yield stress.
By assuming a plane surface for the rolls, Green
and Wallace obtained expressions for total roll force and
52

(a) 3/72

It.

(b) 5/71/2°

°
(c) 6171/2

fig.10. Three entry semi-fields for R/t2 ::: 74.7, showing

that the 'critical reduction" corresponds to

the 5/71,2° field type. ,


53

7
6
5 .
4-
§/k 3

2
1
I I I I I I t I I
0
1 0.5 0
fraction of the arc of contact

I I I
k °

fig.11. Distribution of radial roll and shear stresses in

the roll aap. (Slip line field theory.)


54

torque by considering the simple 'slip line field' shown


in fig. 12(a). This may be compared with the mode of
deformation suggested by Orowan from his plasticine
experiments shown in fig. 12(b), or with a slip line
field solution (fig. 11).
The expressions for load and torque may be written:
P = (55)
G = kL20Qp (56)

+t 2L
where ip 2 + (57)
2L t 4- t
1 2

2.9 The Effect of Roll Flattening


In the previous review of the rolling theories,
it was assumed that the contact surfaces of the rolls did
not deform elastically and the roll radius R has been used
throughout. However, this assumption is only justified
in certain cases when the roll pressure is small, and, in
general, the deformed roll radius R . must be considered.
The only practicable method of allowing for roll
flattening is that due to Hitchcock (29) who replaced the
actual roll pressure distribution by an elliptical one
giving the same total roll force. He considered the
deformed arc of contact to remain circular, an assumption
that °rowan (11) showed to be of doubtful. validity, and
produced an expression:

R 1 2c P
(58)
P. (t1-t2)

where c is a constant determined by the physical properties


of the roll material.
The values:
c = 1.67 x 1014
0_ in:/
ton
c = 1.91 x 10 - i
n /ton
c = 3.5 x 10-4 in2/ton
55

it

fig.12 a . Simplified `slip line field' proposed by Green

& Wallace for the shear plane theory.

(1)- plastic

- rigid

fig.12 b. Regions of plastic flow suggested by °rowan's

plasticine experiments.
56

have been quoted for steel, white iron (chill cast) and
(16)(27)(42)
grey iron rolls, respectively (For the present
work, in which indefinite chill iron rolls were used, a
constant was calculated according to the method given in
Appendix I).
Since the deformed roll radius R' cannot be
determined directly when predicting total roll force and
torque, these parameters must be determined usually by an
iterative technique, in which an initial, approximate,
value of load is determined using the undeformed roll
radius, R. However, Gupta and Ford (30) and Buxton and
Sutton(31) have obtained expressions incorporating the
deformed roll radius whereby the total roll force can be
calculated directly.
Equations for calculating roll torque based on
the summation of the shear stresses along the roll/stock
interface should incorporate the factor RR since the lever
arm of these forces about the roll centres is R, not the
deformed roll radius, R'. The increased roll radius will
have the effect of increasing the length of the arc of
contact, (fig. 13) so the torque obtained by the method
of moments must be calculated using L' where
t
L (R'(t1 - t2))2
However, it should be noted that a consequence
of using the deformed roll radius is that the exit plane
moves outside the line joining the roll centres; thus,
that portion of the roll pressure distribution acting there,
produces a negative contribution to the torque, a fact
pointed out by Ford (32) and Hill(23) , who produced the
expression:

:2 R P.L
2R sOd0 - (59)
(
\ R

which takes this into account.


Undeformed
, Roll
\Surface


Deformed
Roll
Surface

L
L

fiq.13. Effect of roll flattening on the position of the

exit plane.
58

2..10 The Yield Stressj_nHotRolling,


The hot rolling process is usually carried out at
temperatures near or above the recrystallization temperature
of the material being worked. For this reason it is generally
assumed that the thermally activated processes of softening
exactly counterbalance any work hardening that occurs, and
the assumption of a constant yield stress throughout the
arc of contact would seem to be not unreasonable. However,
the strain rate, which greatly affects the yield stress
at high temperatures, does vary in the deformation zone and
most workers have been forced to use a mean value of yield
stress, k, corresponding to the average rate of compression.
This mean yield stress k is a function of reduction,
strain rate, temperature and material composition, so any
independent method of measuring k must take account of all
these parameters.
Tselikov produced his own values for k in the
form of a graph as shown in fig.14, but this only considers
the effect of temperature and composition, however, Trinks
produced graphs taking into account composition, temperature
and compression rate for use with his method, derived from
that of Tselikov. Ekelund obtained an empirical equation
for use with his method, which relates l7 to all the factors
mentioned above:

(60)

and

ko (14 — 0.018)(1.4 + C + Mn + 0.3Cr)kg/mm2


(61)

where
C, Mn and Cr are the carbon, manganese
and chromium contents of the steel
expressed in %
59

25

20

15
Yield

Stress
10

kgimm2

0
600 700 800 900 1000 1100 1200
Temperature 00

fig.14. Effect of temperature and carbon

content (%) on the yield stress of carbon

steels. (from Tselikov)


60

is the coefficient of plasticity of the


Q
rolled stock (starting from 800°C),
n = 0.01 (14 - oole) kg sec/mm2
G is the rolling temperature (starting from
700°C),
and 1-5m is the mean rolling speed mm/sec. (This
formula is valid for speeds up to
7 m/sec).
Attempts have been made to find V, by working from
experimental values obtained on actual rolling mill tests;
however this method presupposes the accuracy of the theory
used, Notable examples of this method are those due to
Ford and Alexander(24) and Gupta and Ford(30) who used the
experimental results of Wallquist(75) and Emicke and Lucas(34)7
and Wallquist(33) respectively, and obtained graphs of the
mean yield stress as a function of mean strain rate,
temperature and composition only.
To obtain a measure of k, independently, at the
sort of strain rates encountered in the rolling process
it is necessary to use some form of high speed compression
testing such as cam plastometry or drop-hammer techniques
or hot torsion testing to provide high enough values of
working rate. Of these techniques, cam plastometry offers
the best method since by suitable design of the cams, the
compression may be effected at a constant strain rate.
Cook(35) Alder and Phillips(36) and Arnold and
(37)
Parker have used this technique to obtain yield stress
measurements, however only Cook considered ferrous materials.
He carried out a comprehensive series of experiments using
a cam plastometer and provided a series of true stress-
strain curves for various steels in compression at
temperatures of 900°C, 1000°C, 1100°C and 1200°C and strain
rate values of 1.5, 8, 40 and 100 sec -1.

The mean yield stress for load calculations is


defined as:

k kdO (61)
61

This represents an average based on the area under the


k = f(0) curve. If the same value were used for the torque
calculations it would represent an over-estimation, since
the sum of the moments of k in the regions where k<kp
is not equal to the sum of the moments in the region where
k>kp owing to the changing lever arm. For this reason,
the mean yield stress for torque calculations is defined
according to

2
11- - 0(2 (62)
-

This represents an average based on the moment of


the k = f(0) curve about the roll axes, and may be written:

kale

(63)

where is the engineering strain ( t ).


1
These functions were expressed in terms of Y* and a the
uniaxial compressive yield stress and natural uniaxial strain,
and values of kp and k s were obtained for the materials
investigated by Cook by numerical integration techniques
from the expression

(64)
P—e) cis2

(65)

where
62

2.11 The Mean Strain Rate

This parameter is defined by the expression:

-
fl
L I L X Ck.
(66)
J0

where X at any point is

1 dt
t dt

Various expressions have been obtained for X

Orowan and Pascoe(15)

tfn
4Rt. 4-4r
1.
1- 4-3r (67)

Sims(16)

64 ) (68)
(
7 1 r
Ford and Alexander ( 24 )

3 (69)
- r) 5

where v = 2 RN cos O = 2i RN
n n

Equations 67 and 69 are very similar since (2 - r)2 =


4 - 4r when r is small, but generally that method due to
Ford and Alexander is preferred, in spite of the fact
that it is based on the position of the neutral point as
defined by equation 31. The method due to Sims is thought
to be less accurate since it assumes that the angular
velocity of the material dO/clz is always equal to the
63

peripheral roll speed; this is only true at the neutral


plane.
Fig. 15 shows the form of the variation of the strain
rate in the roll gap, due to the method of Ford and Alexander,
for various reductions.

2.12 Discussion of the Theories of Hot Rolling


The assumptions of homogeneous deformation and
complete dynamic friction used by the 'pre-Orowan' workers,
was shown to be incorrect by Orowan, both by his work on
Plasticine and the excellent comparison of his more exact
theory with the pressure distribution of Siebel and Lueg.
Only in cold rolling are those assumptions close enough
to the truth for the methods to predict loads and torques
(38)
with reasonable accuracy, and some experimental results
have shown that th- theories due to Siebel, von Karman,
Trinks, Tselikov and Nadai are able to cover adequately
most cases of cold rolling. Ekelundts theory gives good
agreement with experimental results obtained in cold rolling(38)
but Stewartson(3" has shown this not to be the case for
hot rolling. Orowan pointed out that the success of
Ekelund's theory in hot rolling was due, in some cases,
to the fact that the yield stress values were arbitarily
determined to suit the experimental results.
The assumption of the plane strain conditions has
been shown to be acceptable when the width to thickness
( 15) (40)
ratio of the material is greater than about 10 or 12
Ford has shown for cold rolling, that the error involved in
employing a mean value for the yield stress is less than
2% when calculating roll force.
Orowan's method is widely accepted as the most
accurate mathematical treatment of the rolling problem,
and recently Alexander (41) has shown how the modern
computer may be utilized to obtain rapid results from a
method, which had previously been laborious and time
(41) also shows how the simplifying
consuming. This work
assumptions of the later theories can greatly affect the
value of the predicted torque while leaving the load
relatively unchanged.
V LM

1.0

0.5

0.5

x
1
/L

fich15 Variation of strain rate in the roll gap.

Oki is value at entry plane.)


65

Stewartson(39) showed that the simple method of


()rowan and Pascoe did not predict roll force with great
accuracy, but that worker assumed the position of the
neutral point to be equal to 0.5L, the value suggested by
Orowan and Pascoe as a means of simplifying their method.

Sims theory, a simplified version of Orowan's general


theory, has been widely accepted in industry, since its
inclusion with the Cook data(35) in the BISRA simplified
method (42) of determining roll force and torque. The
theory is based on small angle approximations and can be
assumed only to apply under such conditions. Sims compared
his theory with the theory of Orowan and Pascoe, and with
some measurements obtained by S.K.F. in Sweden on the hot
rolling of steel, and with some of his own experimental
work on the rolling of lead at room temperature, a quasi--
hot rolling process. Comparison has been made also
between Sims' theory and experimental measurements taken
on a production mill by Stewartson(39). All these
comparisons have been extremely favourable.
The BISRA method was critically examined by Sims
and Wright(43) using six different production mills, a
slabbing mill, a blooming mill, a plate roughing mill, a
plate finishing mill, a 2-high strip mill and a 4--high
strip mill. The experimental results were found to diverge
R
seriously from the calculated ones when was less than
t2
25, and those workers concluded that this may be due either
t
to a change in deformation mode when 1/L ,;> 2 or to the
small angle approximations of the method.

Buxton and Sutton(31) produced experimental results


that indicated that the error of the theoretical predictions
due to the BISRA method increased with increasing rolling
temperature. This, they attributed to the inadequacy of
the yield stress curves obtained by the cam plastometer
tests. They derived new parameters for the basic rolling
model by multiple regression techniques, but even this
gave differences of up to 40% between the predicted and
measured loads.
Green and Wallace compared their shear plane theory
with the theory of Sims and with the experimental results
66

of Wallquist, and suggested that it gave a more satisfactory


indication of the influence of the R/t ratio on load and
2
torque then did the Sims analysis.
Alexander and Ford first compared their simplified
method (17) with the theoretical predictions due to the Sims
method and the slip line field(24) and showed that it gave
excellent agreement with those methods. The only experimental
verification of the method to date is that of Helmi and
Alexander(44). They showed that the difference between the
method and that due to Sims is very small and found that
Sims' theory was able to predict roll force and torque
accurately when R/t2 > 10, while for R/t2 10 that
theory underestimates roll torque. The theory of Alexander
and Ford was shown to give reasonably accurate prediction
of roll force and torque over the whole range of geometric
variables considered, including the cases where Rit < 10.
2
They found generally that Alexander and Ford's theory
tended to overestimate torque while Sims' tended to
underestimate it. These findings are quite interesting
since the method of Alexander and Ford has no theoretical
basis. The assumption that the slope of the entry portion
2k /t is somewhat
of the friction hill can be put equal to 1
dubious, since the process in this region cannot be likened
2k/t
to that of forging nor does equation 2 degenerate to
unless 0 = 0.
The slip line field theory has the advantage that
it can predict the mode of deformation as well as the roll
pressure distribution and roll force and torque. Crane
and Lampkin(45) showed that under conditions of very
high friction, aluminium strip deformed in a manner closely
similar to that predicted by the corresponding slip line
field.
The form of the pressure distribution predicted
by that method differs widely from that measured by either
Siebel and Lueg(12) or Smith et al(46) 1 having a relatively
high pressure near the entry plane (see fig.11). However,
MacGregor and Palme(47) have produced experimental evidence
of the existence of a zone of high pressure in region of
the entry plane but the experiments were performed on
67

the cold rolling of square bars, where although the


friction conditions may have been high enough for the
attainment of 'sticking' friction, the process cannot be
considered even to approximate to plane strain conditions.
Smith et al showed that the projection of the pressure
pin beyond the roll surface can give rise to apparent high
pressures in the region of the entry plane.
The assumption of a rigid - perfectly plastic material
necessary for solution of the slip line field is not far
removed from the assumptions used in the other theories
of hot rolling, where the yield stress is assumed constant
in the deformation zone, and elastic strains are neglected,
so there is no reason why this method should not be as
good as or even better than other theories at predicting
roll force and torque.

68

CHAPTER 3
The Role of Friction in Hot Rolling

3.1 Introduction
Since °rowan's work, the simpler theories of hot
rolling, have, almost without exception, assumed that
complete 'sticking' is attained throughout the arc of
contact allowing T , the frictional traction at the
interface, to be put equal to k, the shear yield stress
of the material. The advantage of this assumption is that
the unknown factor /4. does not appear in the theoretical
method.
It is interesting to consider what is the value
of the coefficient of friction that'corresponds to this
assumption. For complete sticking friction to be attained

x:s k

i.e.
ikA:
(s/k.)
at all points in the deformation zone.

The theories deriving from von Karman's equation


all agree in predicting that the parameter (s/k) is least
at the entry and exit planes where the influence of the
friction hill is absent, and those methods incorporating
the assumption of inhomogeneous deformation suggest that
this value is about 1T/2. This means that the coefficient
of friction is assumed to be equal to or greater than
2/1 i.e. 0.6366. The roll pressure distributions
derived from the slip line field solutions suggest that
(s/k) is a minimum at the exit plane, where it has a
value of 1.378(24)(25). This corresponds to a value of
/44,;?, 0.726. It seems unlikely that values as high as
this could be maintained under practical rolling conditions
and, generally, independent measurements of nth, lend
support to this idea.
69

3.2_ The Measurement of Friction in Rolling


Methods which have been used for determining an
average value of friction during the rolling process
include the following techniques:
1) maximum gripping angle (Tafel(48) 2 Ekelund(8)),
2) maximum forward slip (Siebel and Fangmeier(49) 9
Pomp and Lueg(10))
3) maximum back tension (Pavlov(50) ),
4) roll force and torque for slipping (Whitton
and Ford (51) )7

5) determination of the position of the neutral


(40)
plane (Capus and Cockcroft ),
6) simultaneous measurement of radial roll
pressure and frictional drag at the roll/
(52)
stock interface (Van Rooyen and Backofen
Khayyat and Lancaster(53) ),
7) comparison of theoretical and experimental
(46) )
roll pressure distributions (Smith et al
and/or total roll force and torque (Dahl,
(54) ).
Wildschutz and Langer
Unfortunately, each of these methods has some
disadvantage in its use. Techniques 2 — 4 modify the
rolling process to such an extent that the conditions
cannot be considered typical of the normal process. Method
1 is effected at the instant the workpiece is entering the
roll gap so that the frictional conditions are characteristic
of an interface between two elastic surfaces whereas the
typical rolling situation is one in which one surface is
elastic while the other is plastic. Methods 5 and 7
require the use of theories of rolling which incorporate
various unverified assumptions.
When considering the hot rolling of steel further
drawbacks are encountered. The use of normal and inclined
pressure pins will affect the apparent surface roughness
of the roll in the vicinity of the pins and since the holes,
in which they are located, must provide sufficient clearance,
it is likely that any oxide scale present will penetrate
70

these spaces under the extreme pressure in the roll gap.


In addition there may be a departure from plane strain
conditions in the vicinity of the pressure pins. In any
case, most of these techniques would alter the scaled
surface in attempting to measure friction. Method 5,
due to Capus and Cockcroft, in which measurements are
taken from the small surface scratches on the material
caused by minute asperities on the roll surface, would
be exceedingly difficult to carry out because of the
elevated temperatures and oxidizing conditions that are
encountered in the hot rolling of ferrous materials.
Having regard to the disadvantages of the above
techniques and considering that most of them can give
only an overall average value of /A, (only method 6 can
give the variation of A along the arc of contact), other
workers have had recourse to the compression test, thus
avoiding the complexities of the rolling process itself.
Notable examples are due to Siebel and Pomp(55) who used
the coned compression plate technique, and Male and
Cockcroft(56) 7 who devised a procedure, using annular
test pieces, in which the coefficient of friction was
related to the changes in the internal and external
diameters, before and after testing.

3.3 The Value of/,t,in the Hot Rolling of Steel


- --
This parameter has been shown to vary with rolling
(48)(8)
speed, roll composition, surface finish, and temperature
The effect of temperature may be due to resulting changes
in the physical properties of the oxide scale present.
Owing to the experimental problems involved in
measuring the coefficient of friction in the hot rolling
of steel, very little data has been published on this topic.
Tafel(48) obtained the values 0.62 and 0.44 for
rough and smooth rolls, respectively, at 1200°C and
rolling speeds of 0 to 6 ft/sec., while for speeds
greater than about 9 ft/sec. the value for the smooth
rolls dropped to about 0.2.
Ekelund(8) provided the following empirical
formulae relating to the temperature of rolling, G°C.
71

1) For cast iron and rough steel rolls,


AL = 1.05 - 0.0005 0
2) For chilled and smooth steel rolls,
= 0.8(1.05-0.0005 0)
3) For ground steel rolls (from Siebel),
A4 = 0.55(1.05 - 0.0005 0)
Bachtinov(57) proposed a modification to Ekelund's
formulae to allow for the influence of rolling speed;

. a k1(1.05 - 0.0005 0)

where a depends on roll quality (see above)


and k1 is a function of rolling speed as
shown in fig. 16 for speeds up to 22 m/sec.
1
.(8)
WusatowskIs quotes the following values of
At. calculated from the maximum angle of bite when hot
rolling steels in commercial practice.

Smooth ground rolls, A = 0.212 - 0.268


Plate mill rolls, = 0.268 - 0.404
Continuous mills, = 0.509 - 0.577
The work of Male and Cockcroft(56) suggests that
the coefficient of friction may be as high as 0.55 at
700°C but drops to about 0.38 in the region of 1000°C,
while Dahl et al(54) using method 7 to determine this
parameter, propose the values 0.2, 0.3 and 0.4 for
temperatures of 900°C, 1000°C and 1100°C, respectively.

3.4 The Validity of the Assumption of Complete


Sticking Friction
Although the values of it,t, quoted in the previous
section are somewhat inconsistent, the general level of
those values is rather lower than that required by most
of the hot rolling theories for complete sticking friction
to be attained. Crane and Alexander(58) have provided
experimental evidence showing that sticking friction is
attained in parts of the roll gap, for the cold rolling
1.0

fig.16. Effect of rolling speed

on the coefficient of friction.

0.5 ( }A= a k1(1.05- 0.0005e), where

a is a function of roll

quality. (Bachtinov))

10 20
Rolling Speed m/sec
73

of aluminium under conditions of high friction. But,


generally, the roll pressure distribution is considered
to consist of three separate zones; a central region of
'sticking' friction where itsk, bounded by two regions
of slipping friction in which ,..}.s k. The region of
sticking friction may be completely absent under some
conditions, and first appears at the neutral point, expanding
towards the entry and exit planes with increasing coefficient
R
of friction, reduction and /t2 ratio. The slipping regions
originate at the entry and exit planes, and extend inwards
in the arc of contact up to the points at which sticking
friction supervenes.
Thus, in those theories deriving from von Karman,s
equation, the effect of assuming sticking friction in those
regions where fins. will be to overestimate the gradient
of the dimensionless pressure distribution (s/k), on the
exit side, and, correspondingly, to underestimate the
gradient on the entry side. The overall effect is thus
to overestimate the total roll force, and torque, and these
theories indicate that the entry section slip zone will
have a major influence on the torque, since in this region
the associated lever arms are greatest. However, this
may not be the case with the slip line field theory since
the limiting value of s/k at the entry plane is 2.57
which corresponds to "A?).0.389 for the attainment of
sticking friction, a value which is not unrealistic.

For this reason it is thought that the assumption


of full sticking friction will not detract greatly from
the applicability of this method unless .< 00389. It
may be noted that Alexander's slip line field solution
for forging between parallel platens under conditions of
mixed friction(59) indicates also that complete sticking
.JD.389, owing to the fact that a
friction occurs when /-4..
centred fan is used in this solution at both ends of the
deforming platen and 2.57, the limiting value of (s/k),
is characteristic of this form of field.
74

3.5 A Simplified Rolling Theory with Mixed


Friction Conditions
Although Orowan recognized the possibility of mixed
friction conditions in rolling, it is only recently that
a further attempt has been made to incorporate these into
a theory of rolling.
El-Kalay and Sparling(6°) proposed that the pressure
distribution in the outer regions of the roll gap, i.ee in
the zones of Coulomb friction, be described by an equation
of the type developed by Bland and Ford (61) for cold
rolling and that the central region in which t= k, be
represented by an equation of the form due to Sims.
The conjunction of these two theories is
particularly apt since they employ similar assumptions
and methods of solution of von Karman's equation.
Using that form of Orowant adaptation of von Karman's
equation relating to the regions of dynamic friction,

df = 2Rs(sin0 jicos 0)
dO

Bland and Ford, assuming homogeneous compression


and neglecting the differences between the vertical pressure,
q, and the radial roll pressure, s,

i.e. f = pt t(s - 2k)

produced the expression

d (tk ("/k - 2)) = 2Rs(sin 0 + ,cos 0)


dO (70)

The left hand side of this expression may be written

tked (s/k) + (s/k -2).d (tk)


dO d0
75

and Bland and Ford assumed that generally:

tked (s/k - 2)0d (tk)


dO dO

Thus equation 70 reduces to

tk.d ( 5/k) = 2sR(sin 0 tpcos 0) (71)


df
By making the usual small angle approximations,
those workers obtained the following expression which
can be integrated directly:

d/dO (s/k) = 2R(0 1,k)


(°/k) t2 + RO2 (72)

giving

Lrt (s," )-/k y,f1


Cort.,t,
(c)
(73)

This may be written:

H ) (74)
k R

where
H = 2(R/t tan-1 (R/t 2

2
c may be determined from the boundary conditions at the
exit and entry planes, and in the absence of front and
back tension equation 74 becomes:

( S/k) +
= 2.t/t2.exp(pH) exit side (75)

t
(s/k)- = 2. /t .exp(,,A(H1 - H))
entry side (76)
(H is the value of H, when 0 = of )
76

El-Kalay and Sparling pointed out that the Bland


and Ford and Sims Formulae give values of 2 and T/2
respectively for the limiting values of (s/k) at entry
and exit and suggested that the value for mixed slipping
and sticking conditions would be expected to lie between
these two extremes, and they derived this (s/k) ratio,
which they called N, by considering Orowan's inhomogeneity
factor, w.

Orowan showed that:


ci)
0 2
2 (1 --a2 ( /0) ) 2 cos 0 d0
sin 0 0
where a =jus
k
and that w was almost independent of 0. El-Kalay and
Sparling assumed w and a to vary in an elliptical manner,

iveo w2 + a2 = a
2 2
A 3

Now, a = 0, w = 2 i.e. A = 2
and a = 1, w = 7r/2 i.e. 1 = 1 - (11/4)2
_
2
B
-,e_ 0
then w + a2 [1 - (0741 = 1
4

or 2(s/k)2) (77)
w = 2(1 - 0.38315

At entry and exit, in the absence of front and back


tension, ( s/k) equals w, or -

2 2
(-/k)(1,2) = 2(1 - 0.38315/4 (7k) (172))2

Therefore:
(Sk)

2 N (7e)
/ (1,2) =

-2-
1 + 105326 ,x
77

Thus, El..-Kalay and Sparling proposed that the regions


of Coulomb friction could be described by the equation:

(s/k)+ = Not.exp(iuH) exit (79)

t2

(s/k)- = Not.exp(j_k(Iii H) entry (80)

while in those zones where ps.,1>k, the equations due to


Sims could be used,

i.e.
`., ius,71
÷ &Y tI

where the subscripts o and i refer to the positions in


the arc of contact of the exit slip/stick interchange
point and the entry slip/stick interchange point,
respectively. The angles at which sticking is attained,
00 and Oi were obtained by solution of the equations:

1 = N.to .exp(/LH0 ) (83)

t2

1 = N.ti .exp()A,(Hi - Hi)) (84)


t,

El-Kalay and Sparling obtained expressions for


total roll force and torque from -
78
ot,
( s
kp.R ( /k)d0
Jo
and

21-C R 2 ( s/k)0d0
O

The inhomogeneity factor N is shown in fig. 17


as a function of , A the coefficient of friction, together
(23), who, by
with similar relationship obtained by Hill
extending Prandtl's approximate solution for compression
(13) for the case when
between rough parallel platens
sticking friction is not attained, derived the following
expression relating p. and the minimum value of s/k at the
ends of the platens, i.e. N,

1- 4- sin AkN
N = (1 -/(A. 2N2 )2
(85)
IA, DT

The degree of disparity between these two relationships


is perhaps a measure of the combined effects of the
modifications to Prandtl's method due to Nadai and Orowan
and the assumption, that w and a are related in an elliptical
manner, made by El-Kalay and Sparling.
This theory was the first attempt to provide a
simple method of incorporating mixed friction conditions.
Unfortunately it represents an over simplification since
the method assumes that w, the inhomogeneity factor, is
constant throughout the regions of Coulomb friction. The
expression derived by El-Kalay and Sparling however, i.e.

= 2(1 - 0.38315 a2)

shows that w is function of a. It is, therefore, a


function of s/k as well as ofp,. This means that w must
decrease steadily from both entry and exit planes, at
which points it is equal to El-Kalay and Sparling's
derived N, until sticking friction supervenes when the
value /r/2 is attained, and thereafter remains constant.
2.0

1:9

(Hill)
N 1.8
(EI-Kalay & Sparling.)

t7

1.6
2 1

0 0.1 0.2 03 04 0.5 0.6 2-*
M

fig.17. Variation of N with coefficient of friction. (N is the limiting value

of ( s/k) at the entry and exit planes.)


80

CHAPTER 4
Interpretation and Comparison
. _ of Hot
. _
Rolling Theories

4.1 The Slip Line Field Theory


Given that the assumptions made in the theory of
the slip line field are not unrealistic, then this method
would appear to be the soundest means available at present
for calculating roll force and torque in hot flat rolling.
The primary disadvantage of the theory is that it cannot
be considered general enough for everyday usage. Solutions
exist only for certain geometries and Crane and Alexander(25)
point out that their method does not allow the solution
of a pre-determined geometry. Only after completion of a
given field can its geometry be ascertained. Thus, if the
slip line field solutions are to be generally useful it
is necessary for the predictions of the individual fields
to be incorporated into generalized equations for load and
torque covering the whole range of practical rolling geometries.
Crane and Alexander showed that two types of field
may be determined more easily than the others. These are
the type in which the exit semi--field consists of a single
circular discontinuity and the type designated 5/7-1s
°. A
number of solutions for each of these types of field,
determined over a range of reductions and R /t ratios,
2
yielded the relationships, between those two parameters,
shown in fig. 18. Over the range of interest each of these
curves can be represented closely by an equation of the
form:
(R/t )
0K, = a constant

This suggests that any group of solutions, all of the same


type, may be represented by this form of equation, and
it may be concluded that some function of (R/t2.r) will
provide a suitable parameter to describe the geometry of
rolling, since this product is found adequate to describe
the type of field and, therefore, the mode of deformation.
It may be shown that ( R/t2.r)2 is equal to L/(t1t2)2 or
60

50

40

r °/0 30 _1.

co

20

10 •
R,.r = 5.66 /7 1°)

"t 2 0
o
I I . 1

50 100 150 200

t2

fig. 18. Relationship between r and R/t2 for 5/71° s.l.f. solution and

that incorporating a single circular discontinuity.


82

L
/tm' where tm is the c:eometrie moan thickness of the
strip within the arc of contact, and L/tm represents the
aspect ratio of the deformation zone, within which any
particular type of field must be constructed. Thus, it
does not seem unreasonable that the different types of
field are characterized by a particular value for
R
( /t2 .r)2 or L/t
L
This ratio /tm may be compared directly with
similar ratios invoked by the simple theories of Orowan
and Pascoe(15) and Alexander and Ford(17), namely L/t2
and L/t1+t2 * In both these cases, the dimensionless
parameter -/kL
P was assumed to be a function of that
corresponding ratio, only. Since /kL is also similar
to Sims function Qp, it was thought useful to determine
whether any convenient relationship existed between the
compound parameters, /kL, as predicted from the slip line
fields, and L/tm. The values of the function /kL were
determined for all the fields given by Crane and Alexander
and plotted against L/tm, as shown in fig. 19. A linear
relationship is obtained. However, the solutions given
by Crane and Alexander were limited to a single R/t2
ratio of 74.7, and, therefore, it may be Questioned that
the relationship obtained is completely general, or whether
it is characteristic only of this particular R/t2 ratio.
Fortunately, some partial slip line field solutions given
by Ford and Alexander are random with respect to rolling
geometry and R/t2 ratio. These are represented by the
triangles in fig. 19 and provide sufficient confirmation
that the relationship is completely general.
The torque parameter chosen to plot against the
deformation zone aspect ratio L/tm was G/kL 2. Fig. 20
shows the result obtained. Again, a linear relationship
is produced.
The equations of the lines in figs. 19 and 20 are:

P 1.31 + 0.53 L (86)


kL
Limit of the
H rolling solutions

P.
kL

o Crane and Alexander

..0 a Ford and Alexander


.0'
.0.

OM,

t I
1 2 3 4 5 6 7 8 9 10
L
..-'tm
fiq. 19. Relationship between the load function and L/tm predicted by s.l.f. theory.
H
Limit of the

rolling solutions

'‘L2
G A
0

o Crane and Alexander


.0"
..••• A Ford and Alexander
00'

I I t

1 2 3 4 5 6
L/
I tm

fig. 20. Relationship between the torque function and Litm predicted by s.l.f theory.
85

G 2
/kL = 1.59 + 0.437 L (87)

(tit2)2

These are very similar to that given by Hill(23)


for the load required to forge between perfectly rough
platens,

b
i0e0 P = kb(1.5 + 0.5 /t)

the aspect ratio for forging, b/t, being regarded as


analogous to L/(t1 t)- the aspect ratio for rolling
based on the geometric mean thickness.
Hill provided slip line field solutions for a
whole range of forging geometries as shown in fig. 21.
However, there is no solution to the rolling problem when
L
/tm < 2.35, the value corresponding to the simplest type
of field with a single circular exit slip line. But,
it may be speculated, that if such solutions were available
they would also follow the Hill forging relationship in
which there exists a minimum (p/k) value at bit = 1, and
for h/t<1, P/k rises again to a limiting value of 5.142
at t/h = 8.75.
For completeness, the solutions due to Alexander
for mixed friction conditions(59) are included in fig. 21,
to give some indication of the effect of coefficients of
friction less than 0.389. It may be surmised that similar
results would he obtained if solutions were found for
rolling with mixed friction conditions. Although Alexander
produced only two solutions for each value of,
corresponding to h/t = 3 and 7, a straight line has been
drawn using these points, to indicate the probable
relationship over the whole range of b/t.
The single relationship between L/tm and /kL and
G/kL2 implies that all fields of the same type, as defined
by Crane and Alexander will have identical values of the
load and torque functions /kL and G /kL-
2, as well as of L/t
m
It is interesting to compare the predictions of
other well known theories of hot rolling when plotted in
p = 0.389
0.3

0.225

0.15

0.10

0.05

1 /OM
0

Hill
-e—e- Alexander

I I I I I I I I
0
0 1 2 3 4 5 6 7 8 9 10
b/t

fig. 21. Load functions for forging predicted by s.l.f. theory. (various K)
87

a similar manner to that of figs. 19 and 20.

4.2 Other Theories


The theory of Sims(16) gives the relationship
shown in fig. 22 and suggests that the parameter
L
/(t1t2)2 is not entirely adequate as a sole measure of
rolling geometry, since a family of curves is required
to describe the results for different reductions. However,
the lines are very close and are also so close to the
slip line field line that there would seem to be little
reason for preferring the more complex form of the Sims
equation. The torque function due to Sims, shown in
fig. 23, is also very close to the slip line field line.
In fact, the lines corresponding to 10% and 20% reduction
are almost coincident, with the slip line field line at
low values L/t . However, at higher values of L/tm it is
the lines for the higher reductions that become increasingly
closer to the slip line field predictions.
(15) gives the
The theory of Orowan and Pascoe
equation for roll load as -

L
P = kL(1.6 + n. /t2

while that for torque may be found by calculating the


moment of area of the roll pressure diagram about the
plane of exit, and can be written:

G = kL'(1o6 + 2 n(n + 1)L/t2)


3
These two expressions can be rewritten as
L
functions of /(t.t )2 7

i.e. = kL(1.6 +
"Tx --- (88)
(1 - r)2 (t1 t2)2

and 2 . 2 .,n(n + 1) L
G = kL (1.6 + / (89)
(1 - r): • (t 1t2)2
Oa 111•1 1110k .11•11.
S.L.F. Theory
P •-4

r Sims' Theory
kL
•-• r :: 10010
-Y.

20°/a

300/0

40%

I I 1 1 I I I I

1 2 3 4 5 ytrn6 7 8 9 10

fig.22. Sims' theoretical load function plotted against L/t m • (for various reductions)
GI<L2 Em• WM WIN MO
S.L.F. Theory

Sims' Theory

r :-. 10%

20%

30%
/ 4010

I I I I I I I I

1 2 3 4 5 6 7 8 9 10
1_4
m

fig. 23. Sims' theoretical torque function plotted against L/t.rn:.


90

Taking into account the value of n, which is given


(t ,
- t /2R, the
by °rowan and Pascoe, as a function of
load and torque functions due to those workers may be
calculated and are shown in figs. 24 and 25. The level
of these curves is significantly higher than that of the
slip line field and they are much more strongly dependent
on reduction.
As indicated, above, the simplified theory of
Alexander and Ford(17) invokes the use of the ratio
/t1 + t2, that is , the ratio of the length of the arc
of contact to the arithmetic mean thickness. This can
be seen more easily by rewriting that expression in the
form:

0.5L )
kL(71-/2 + (90)
+ 2

The difference between the geometric and arithmetic


mean thickness is negligible unless r > c. 35%, so the
difference between the slip line field load function
predictions and those duo to Alexander and Ford) as shown
in fig. 2E% is due mostly to the different intercept and
slope predicted by the two methods.
The ex73ression for torglle due to Alexander and Ford
may be written

kij- 0-172 + 2(1 - r)2(3 - 2r). L ) (91)


3 (2 r) (t1t 2

and this is plotted with the slip line field in fig. 27.

These last two figures give a good indication as


to why Alexander and Ford obtained such good agreement
between their method and the slip line field theory, in
spite of the rather dubious assumptions on which their
method is based.

The relationships for the load and torque functions


(27)
obtained from the shear plane theory of Green and Wallace
1

X/

NM.

Pzki.

•---e I": 10 °/0


x---x 200/0
o—o 30`lo

40%

1 2 3 4 5 6 7 8 9 10
L/
m

lig .24. Orowan & Pascoe's load functions. ( incorporating n. as shown in fig. 8b .)
S.L.F. -

L2
3

12--.---.--0 r :: 10°l0

x x 200/0

0 o 30%

n 11 40G/o

1 1 1 1 i I I I t
1 2 3 4 5 6 7 8 9 10
L/trn

fig. 25 ()rowan & Pascoe's torque functions. (incorporating n, as shown in fig. BID.)
d

5-
2

1 2 3 4 5 1.41.1 6 7 8 9 10

fia. 26. Alexander & Ford's load functions for various reductions. (see inset)
7

0
0 1 2 3 4 5 6 7 8 9 10
L/tm

fiq. 27. Alexander & Ford's torque functions for various reductions.
95

are identical and are shown in fig. 2D Although they


predict values much higher than the other theories it is
interesting to note that there is a minimum at L 1t )- = 1,
2
which, because of its similarity to fig. 21, lends support,
perhaps, to the idea that the choice of an appropriate
parameter, for aspect ratio, makes it possible to draw a
close analogy between forging and rolling.
The load functions due to the method of El-Kalay
and Sparling(60) for coefficients of friction of 0.3 and
0.4 are shown in figs. 29 and 30 for various reductions.
Despite the fact that the theory invokes regions of slipping
friction, the curves are surprisingly high, being close
to or even higher than the slip line field relationship
in which complete sticking friction is assumed. This
circumstance appears to be due to the use of a constant
value of inhomogeneity factor throughout the regions of
Coulomb friction. The relationships of the torque functions
due to El-Kalay and Sparling, shown in figs 31 and 32, are
also high, which those workers attribute to the fact that
the position of the neutral angle predicted by their method
is somewhat smaller than that given by the Sims method.
To sum up, it can be said that of the theories
discussed in this section, only the slip line field theory
predicts that the load and torque parameters are functions
of L/tm only, the other theories indicate that,

G L
/kL /kL2 = f(r, /tm ) (92)

with varying degrees of reduction dependency, with the


exception of the Alexander and Ford theory which invokes
the use of ratio of the length of the arc of contact
to the arithmetic mean stock thickness.
In addition to these two theories incorporating
the aspect ratio of the deformation zone, there are others
which do likewise.
96

7
G&W
o---® r=10°/° - S.L.F: (P)
6
S.L.F. (G)

4
kL

Gj 3
kL

0
0 1 2 3 4 5
tm

fig.28. Green & Wallace's load and torque functions


P4L

0-,----0 r = 10%

X x 20%

0-0 30%

S.L.F. ti____.4 40%

1 2 3 4 5 vtm6 7

fig. 29. El-Kalay & Sparling's load functions, for 1,..t = 0.3
200/0

300/0

S.L.F. 40%

I I I I I I I
1 2 3 1. 5 Li,

6 7 8 9 10
i ‘rri

fiq.30. El-Kalay & Spading's load functions for 1.t = 0.4


G l<2

0
0

200/0

0-0 30010

S.L.F. ti 40`10

! 1 1 1 i 1 I 1 1
1 2 3 4 56
Li, 7 8 9 10
' Lm

f i q . 31. El- Kat* & Sparlinq's torque functions for p z• 0.3


6

G72
3

0
0 1 2 3 4 56
L .4 . 7 8 9 10
m

fig. 32. El - Kalay & Sparling's torque functions for u


l z 0.4
101

Some Other Theories incorporating the Aspect


Ratio of the Deformation Zone
Unfortunately, while these theories all express
the load function in terms of L/taverage, none of them
consider roll torque and some of them give no indication
as to how the value of t.average was defined.
The Golovin-Tiagunov method given by Wusatowski(18)
expresses the load function in the following way

L -
(93)
2kL
av

(62)
Geleji proposed the formula:

P = 1+ cp L i 41
(94)
2kL tav

where vr is the peripheral roll veclocity


(m/sec) and c and n are coefficients determined
from the results of test measurements.

This author obtained this expression from the results


of the investigations of Puppe, Siebel and Emicke and
Lucas, and gives the following expressions for c, where
tav is the arithmetic mean thickness:

2 _ L
0.25.$.2 L7t-av- c 17( Pc, (Iv) -- 29.85( /L.av) + 18.3

2
1$: L/tav 3 c= 0.8( /taV ) - /09(Lic ) + 9.6
av
2
< 13 c = 0.013(L/tav) -- 0.29(L/tav )+ 2086
3H/tavN

Geleji suggests that for practical purposes


n = 4.
102

Zaroschchinskii(G3) also proposed that the load


function was a function of the ratio of the length of the
arc of contact to mean stock thickness,
i.e.
= f(L/t )
av

and suggested that this function may pass through a


minimum at L/tav = 1 and /kL = 2. However, he continues
by stating that /kL can never be equal to 2, since this
implies the absence of external friction, (homogeneous
deformation was assumed, but it has been shown that this
is only possible whenp„-,0) and suggests that the minimum
value of /kL (>2) will occur at some value of L/tav(>1)
determined by the stock width to thickness ratio.
Muzalevskii(64) considered the effect of L/t
av
in the range it <1, and provided experimental evidence
av
that showed /kL increasing as L/tav approached zero.
He deduced the expressionz

(95)
/2kL = 1 + 0.1( te-v /L --1)1.143

which lends further support to the earlier suggestion,


that if slip line field solutions were obtained in this
region they would follow the Hill forging relationship.
(65)
Stylbo obtained empirical polynomial expressions for
P/kL as a function of L which suggested P/kL was a
L /tav'
minimum at /tav = 0.25,
Ekelund's expression may be written

1 + B(t1

2kL t1 + t- (96)

where A = 1.6 and B = 1.2 in the orginal. However,


more recently Jedlicka(66) has provided experimental
evidence to show that, generally, the average value for
B is zero, and the expression reduces to

1 + C L
(97)
2kL t 4- t2
a
103

where 0O36 for ingot and billet


rolling,
and C 0.5 for heavy plate rolling,
thus, relating the load function to the ratio of the
length of the arc of contact to the arithmetic mean
thickness, only.
104

CHAPTER 5
Experimental

.1 The Rolling Mill

The experimental mill used for these investigations


was a single stand, two-high unit manufactured by Sir James
Farmer Norton and Co., having indefinite chill cast iron
rolls, of 10 inch diameter and 12 inch barrel length, fitted
with roller bearings. It was driven by a variable speed
D.C. motor, rated at 60 hp, with a maximum speed of
1500 r.p.m., through a 501 worm reduction gear to produce
rolling speeds up to c. 75 ft/min.
The top roll and drive shaft were supported against
the load cells and screw-down gear by hydraulic means and
the maximum daylight between the rolls was about 1.25 in.
The roll gap could be adjusted either by means of an
overhead screw-down assembly driven by an electric motor
or by the use of a large hand wheel. The setting was
indicated by two pointers, one on each side of the mill,
connected to the vanadium bronze pressure pads, reading
against scales, graduated in 1/100 mmo fixed to the
rotating screws of the screw-down gear.
The surface and core hardnesses of the rolls were
61/62° and 45° Shore C, respectively, with a hardness
gradient of approximately 10° Shore C/inch between these
two figures. Surface finish was kept approximately constant
during the experiments by regular abrading with 240 grade
silicon carbide emery paper in a direction perpendicular
to that of rolling to maintain appropriately high friction
conditions. Before each rolling trial the roll surfaces
were cleaned and degreased with a solution containing
equal quantities of carbon tetrachloride, acetone and
ethylene trichloride.

5.1.1 The Limitations of the mill


During the planning of the rolling programme, it
was necessary to consider the limitations of the mill.
105

The maximum roll separating force of about 80 - 90


tons was limited by the small cross sectional area of
the roll necks, resulting from the use of roller bearings.
The maximum available roll torque Was about
50 tons in. sii:ce, at values greater than this, there was
an unacceptably high risk that the mill would stall.

5.2 Mill Instrumentation

The mill was fully instrumented to measure separately


the roll separating force, on each side of the mill, and
the torque on each roll during a pass.

5.2.1 The Load Cells


-.----------
The rolling mill was fitted with two 75 ton Elliott
load cells, of the normal strain gauge bridge type, one
positioned in each mill housing window between the top
roll chock and a vanadium bronze pressure pad under the
screw-down mechanism. These cells were connected to a
dial indicator reading 0 - 75 tons in 0.25 ton divisions
by way of a switch network such that the total load or the
load on either of the load cells could be observed.

5.2.2 The Torgue Tranducers


The torque was measured by two A.S.E.A. magnetic
'torductors', one situated on each roll drive shaft. This
form of torque measuring device surrounds the shaft but
at no point comes in contact wit'.-1 it, thus eliminating
the problems of lead-off wires and slip rings encountered
with the more usual strain gauge type. The torque values
could be obtained from two indicator dials, -0000 lb ft/

0/4-0000 lb ft in 500 lb divisions.

5.2.3 Load and Toroue Recording


This was carried out using a 12 in. multi channel
ultra-violet recorder. This type of recorder was chosen
since it is capable of giving simultaneous continuous
recordings of the signals from both load cells and
torductors, and has the added advantage of an excellent
response time. However, use of this instrument required
106

modifications to be made to the existing set-up.

The .[,.C1. signal, provided for the Elliott indicator


system, was not suitable for the ultra-violet recorder and
an alternative preset constant voltage D.C. power supply
was installed. This was connected via a rotary stud switch,
so that the systems could be interchanged simultaneously
with one switching action as shown in fig. 33.
The load cell output was nominally 1 my per volt
input at full load, i.e. 75 tons, thus the use of an input
voltage of 10 v produced a maximum output of 10 my per load
cell. However, because of the load restrictions on the
mill, the maximum output never rose much above c. 6 my
per load cell. This relatively low output, coupled with
the fact that the impedance matching of the load cells
and ultra-violet recorder galvanometers was poor, leading
to further signal loss, meant that the current provided
was too small to produce an adequate deflection on the
recorder. To overcome this, a D.C. amplification network
was used to produce a signal of suitable magnitude. (Fig. 33)
The amplifiers used had a variable gain control,
and this was utilized to give a suitably large deflection
on the recorder, irrespective of the actual load.
The interns] circuit of the 'torductorl monitoring
system was modified during installation to give a suitable
deflection on the recorder with the galvanometers used.
In this case it was possible to obtain simultaneously
recordings on the ultra-violet instrument and readings
from the indicator dials,

5.3 y Calibration Techniques


Eefore any rolling could be commenced, it was
essential that the measuring instruments were calibrated,
and that the mill spring was investigated fully.

q 2 1 Calibration of the Load Measurinc Circuits


As the calibration of a load cell is relatively
sensitive to the length of the input and output leads,
it was essential that the final calibration be carried
107

u. v.
recorder
d.c. amplifier

fig.33. Load cell circuit. (schematic)


108

out in situ, on the mill, in a manner identical to that


used under actual rolling conditions.

The two load cells were first calibrated separately


on a Denison 50 ton testing machine, which was regularly
checked and recalibrateci by those manufacturers. The
constant voltage D0C pack was used as a source of power,
pre--set to the same voltage as was to be used in the normal
load measuring operation, and the output was measured on
a Cambridge potentiometer. From these results, a graph
was plotted of load vs output foreach cell. (Fig. 34)
The two cells were then positioned, one on top
of the other, under a single screw-down on the mill, and
a load was applied by use of the large handwhe.l. This
load was, measured using the pre-set D.C. voltage supply
and the Cambridge potentiometer together with the calibration
curves (-Y,tained on the Denison machine. As the load on
each cell was identical, this provided successively a check
on each rapective output..
The cells were then connected into the recording
circuit in a manner identical to that used in the main
experimental prgramme. Deflections relating to that load
were measured on the ultra-violet recorder, for the x200
and. x500 click stop amplification factor settings on the
D.C. amplifiers. (1::hile the amplification factors were
not necessarily accurately equal to x200 and x500, it was
reasonable to assume that they were constant, once the
amplifiers had warmed up, after being switched on. In
practice, at least i hour was allowed for this.) The cells
were then disconnected and, as a further check, signals,
equal in magnitude to the load cell outputs, were in
at the same points, to ensure that these gave the same
deflections.
The whole procedure was repreated over a range
of loads and calillration diagrams were drawn up (figs. 35
and 36). &ach of these shows two lines, corresponding
respectively to the x200 and x500 click stop amplification
settings, which were found to give suitably large deflections,
and, hence, adequate accuracy, for most rolling conditions.
10 20 30
Load tons

fig.34. Calibration of load cells on Denison 50 ton testing machine.


50

40..
x200 fig.35.
Calibration of

30 load cell No.1

Load on u.v. recorder.

tons
20

x500

10

I I I
0
0 10 20 30
Deflection cm
r
50

40
x 200 fig. 36.

Calibration of

30 ,- load cell No. 2

Load on u.v. recorder.

tons
20 ...

x500

10 ... IIS

0 a
0 10 20 30
Deflection cm
112

5.3,0:2Clibration of the_TorqueNeuring_ Circuits

The major portion of this calibration was carried


out by the manufacturers during installaton of the mill.
This was performed by rotating the roll drive shaft between
the motor and the reducer_ gear while the torque on the
fixed drive shaft in the region of the torductors was
measured by the use of a proving ring on the end of a 3 ft
arm. In this way ti,e actual torque on the shaft was related
to a signal produced at the torductor control box, and
a calibration chart was dr a wn up, as shown in fig. 37.
A certain proportion of this signal was used to
produce the deflection on the ultra-violet recorder
galvanometers. Pressing a button in the measuring apparatus
injected into the circuit a pre-set constant signal
corresponding to a particular torque which was found from
the calibration curve. This produced a deflection on the
ultra-violet recorder, which in turn related to that torque
associated with the pre-set injection signal. Thus,
calibration on the ultra-violet recorder could be effected
as often as thought necessary, by simply actuating the
injection circuit in the torductcr control box and measuring
the deflection produced. Fig. 37 shows the calibration
torque values,
The deflection obtained during rolling was measured,
and the torque was calculated by a simple proportioning
method, as the deflection was linearly dependent on the
actual torque.

5.3.3 Vill Spring Calibration


During rolling the structural parts of the mill
deform elastically. The effect of these elastic deformations
is to increase the gap between the rolls when the mill
is under load. The roll gap of a loaded mill is termed
active, whilst that of an unloaded mill is termed passive.
ThE, difference between the "wc, knol:Tn as the will spring,
is directly related to the total roll separating force
by a constant termed the mill modulus, H, given by:

g g p P t

11

10

7 Calibration Voltage

Voltage (V) 6

at torductor 5

control box 4

2 Calibration
Torques
1

0
0 1 2 3 4 5
-3
Torque lb ft x10

fig. 37. Calibration of torductors.


114

where

ga is the active gap size

gp is the passive gap size


and P is tee total mill load.
t
Once the mill modulus is known, it is possible
to calculate the passive roll gap required to give any
required reduction, assuming that there is some suitable
method to hand for predicting the expected total roll force.
The mill modulus was determined by cold rolling
a large number of pieces and determining in each case the
active and passive roll gaps and the corresponding load.
The active gap was assumed to be the same as the thickness
of the material as it emerged from the gap, and the passive
gap was measured by the use of thickness gauges, which were
made for that purpose.

Both of these methods of measurement are subject


to slight inaccuracies, the first neglects any elastic
recovery of the stock, while with the sec=d,there must
be a finite amount of clearance for insertion and withdrawal
of the gauges. Further, in order to predict the mill spring,
it is first necessary to predict the expected load, Pt,
which is also subject to some uncertainty. In practice,
therefore, all rolling e:::periments requiring specific
reductions were conducted in triplicate and after the
first of these minor adjustment of the roll gap was made
if this was found to be necessary. The experimentally
determined calibration is shown in fig. 38 and this curve
represents the average line found from some 100 experiments.
As expected, the curve does not pass through the origin,
duo to the amount of slack inherent in the mill
structure.

504 Preheating

Two preheating furnaces were used at different


times for this work. Whilst quite different in their modes
of heating, control and power rating, their basic shapes
and dimensions livere similar. This was a simple rectangular
shape with the addition of a frontal extension, or

60

50

Mill 40
Spring

in x103 30

•••

20
•••
Mill modulus (M) = tan a

10

10 20 30 40 50 60 70 60 . 90 100
Total Mill Load tons

fig.38. Mill spring calibration.


116

'nose', which contained the extended hearth plate in a


short tunnel. To keep temperature losses and oxida tion
problems to a minimum, the material was introduced directly
from thr. furnace chamber into the roll gap. To do this,
the furnaces were specially designed to run on rails in
front of the mill. During the charging, heating and soaking
periods, the furnace was kept away from the mill, to prevent
the heat from affecting either the mill housings or the
roll bearings, On completion of the soaking period, the
furnace was run up to the mill, which, with its entry table
removed, allowed the 'noscl t<Th be positioned between the
housings so that the furnace chamber exit was immediately
in front of, and almost in contact with, the roll gap. The
workpiece was pushed from the furnace chamber into the
roll gap with the aid of a simple pushing mechanism operated
from the rear of the furnace.
The furnace cycle is shown in fig. 39, and plates
I and II show the furnace in the charging, heating and
soaking and the rolling position, respectively, together
with the general experimental set-up.

5.4.1 The I'reheatino Furnaces


The furnace used for the early part of the work,
furnace A, was designed and built by the author.
It was based on an open-ended fireclay muffle,
of rectangular cross section, having internal dimensions
19- x 5.75" x 3.75" and a wall thickness of 0.€25". The
muffle v= covered with two layers of alumina cement
beL:wecn which 17 gauge Kanthal Al heating element wire
was wound.
The winding:s consisted of three sep7.rate 6:c zones,
to obtain a greater control of temperatuue uniformity.
Each zone was rated at about 1.7 kw and was controlled
by a s,,,,arate 10 amp Eurotherm Thyristor temperature
controller (0 - 1200°C). Three temperature sensing
thermocouples (Pt / Pt - 13% Rh) were positioned against
the muffle wall, one at the centre of each zone. Each
thermocouple was protected by a silica sheath and all three
117

CHARGING

I I

C•)

HEATING & SOAKING

C.)

//
ROLLING

u.

fic.39. The furnace cycle.


Plate I Experimental set-up,
CO
r--
goo
urnace
ionger

Plate II Experimental set-up. (furnace in rolling position)


120

were supported in a single aluminous porcelain tube.


Further details of furnace A may be found in fig. 40.
The maximum operating temperature was 1200°
C, this
being determined by

1) the range of the controllers,


2) the maximum working temperature of
Kanthal Al wire, which is about 1350°C.
(This allowed for the temperature drop
through the muffle wall).
In spite of being slightly under-powered, this
furnace was used for nearly all the work at 1000°C, before
its eventual failure and the installation of furnace B.
Furnace B, was built by Efco Royce Ltd., and was
basically a silicon carbide element box furnace, but
provided with the 'nose' already described, as shown in
fig. 41 and plate I. This furnace was also divided into
three zones, each zone consisting of 4 elements in series/
parallel connection rated atil kw and run from a separate
phase. However, in this case, there was a single Eurotherm
controller, which controlled the three separate zones,
through three 60 amp thyristor stacks. A single sensing
thermocouple, (Pt / Pt - 13% Rh), positioned in the centre
of the centre of the furnace chamber, was used. This
furnace was capable of operating at temperatures up to
1400°C and was encased in welded steel sheet to minimize
gas leakage.

5.4.2 Temperature Distributions during Reheating


----------.
Since the material to be rolled entered the roll
gap directly from the furnace chamber, it could be assumed
that temperature losses during transfer were negligible.
By the same token, however, such variations of temperature
as occurred within the furnace would persist within the
roll gap. It was, therefore, essential to check temperature
uniformity in the furnace and relate the actual temperature
of the billet there to that temperature indicated by the
furnace controller.
121

`MI. 28 ' insutating bricks

\ \ \
gas inlet TTTTTTTT74777-7nTiffip 1 / / .1 I I I I I I

\
•••-nik. WIZ as

pusher 1/2 /////7/11////(////4///////i/


wi ding

Syndanyo casing with


Scale I
4"
steel framework.

fig.40. Furnace A.
122

TIC

gas inlet
elements
0 0 0 0 0 0 \\\
pusher 7/l
_I MI/ //// ///// ///// ///

O 0 0

Scale r 6"

/
7

fia.41. Furnace
123

The first of the uniformity tests was carried out


in furnace A, employing a special billet, having the
dimensions 11" x 4.375" x 2", containing 16 chromel-alumel
thermocouples, placed so as to measure temperatures along
its length, across the width and through the material. The
block was positioned centrally in the furnace, heated to
a nominal 1000°C and soaked, the temperatures of the
thermocouples being recorded continuously on a 16 point
chart recorder. Results showed that the overall temperature
of the billet was subject to a certain amount of fluctuation,
the period of which, suggested that this cycling was related
to the functioning of the controllers. This effect was
caused possibly by the relatively large time lag associated
with the thickness of the muffle wall, which intervened
between the sensing thermocouples and the furnace windings,
and could have been eliminated to some extent by suitable
adjustment of the proportioning band on the controllers.
However, this was not considered worthwhile, since the.results
showed also that the temperatures of the thermocouples,
at any one time, were distributed over a range of some
30°C.
It was originally intended to carry out the rolling
tests on specimens of about 12" in length. However,
considering the poor temperature uniformity obtained, it
was decided to decrease the length of the specimens to
six inches. An incidental advantage was the resultant
doubling of the number of available test pieces.
A further test was carried out using a billet equal
in size to that of the largest piece used in the main rolling
programme, i.e. 0.625" x 5" x 6".
The thermocouples were inserted to a depth of
0.3125" at 0.5" intervals along the length of the billet,
and heating and soaking was carried out over a considerable
period of time. The temperatures, which were recorded
on a multi-point recorder, showed that if the furnace
controllers were set to 1200°C and were controlling at
that temperature, by suitable adjustment of the proportioning
band, the temperature at all points along the length of
124

the billet could be brought within the range 1196°C


+ 2°C. Similar results were obtained at 1000° C, where
the range produced was 995°C + 2°C.
This represented a considerable improvement on
the previous results and this was thought to be due to
the fact that the 6" billet could be positioned entirely
in the central zone of the furnace, whereas the 12" billet
passed through the two winding gaps between the three zones.
In addition, it should be noted that no measurements were
taken for the temperature variation through the thickness
of the material in the second test, whereas in the first
test the overall temperature fluctuations would have given
rise to 'heat fronts' passing successively through the
thickness of the stock, producing poor temperature uniformity
in that direction and contributing to the overall 30°C
range obtained. The situation was exacerbated also, in
the first case, by the fact that the controllers did not
function all at the same time.
Tests on the second furnace showed a uniformity
of + 5-C in the billet and indicated that to obtain mean
temperatures of 900°C, 1000°C and 1100 t, the controller
had to be set to 890°C, 986°C and 1080°C, respectively.

5.5The Composition
_ and Condition of the
ExRprimental
---Steel
The steel used for all the experiments described
was obtained from the United Steel Cos. Ltd., now part of
the British Steel Corporation, Special Steels Division,
and came from a single cast of steel having the following
composition:
0.14% C 0.46% Mn 0.025% S 0.01% P (+ Fe)
This may be compared with the composition of the
low carbon steel employed by Cook(35)
0.15% C 0.60% Mn 0.034% S 0.025% P 0.012%. Si
They are seen to be similar and were considered close
enough for the data provided by Cook and McCrum(42) to
be used in the present work.
125

The material was supplied in various thicknesses,


these being, nominally, 0.07, 0.10, 0.20, 0.30, 0,40, 0.50
and 0.625 in. The width was in all cases within the range
4.5 - 5 in,, so that only in the case of the thickest
pieces could plane strain conditions be questioned, when
the aspect ratio was about 8 to 1. The thicknesses were
R
chosen to give a realistic range of /t2 values (10 to 150)
for various reductions. The bars were cut into rolling
specimens 6" in length.

5 „
5 1 Grain Size Control
As the different sizes of stock, provided, had
undergone different thermo-mechanical treatments, their
grain sizes varied accordingly. Although the yield stress
of f.c.c. structures such as austenite, with which the
present work is associated, has been shown not to be
affected very greatly by grain size(67) 7 it was decided
to heat treat the specimens to standardize this parameter,
and also to produce structures of similar homogeneity and
degree of isotropy in all cases, and to eliminate completely
any retained cold-worked structure within the specimens.
The grain sizes measured at room temperature were
those of the ferrite present, whilst rolling was carried
out with the material in the austenitic condition. However,
it was thought that if all material entering the furnace,
prior to rolling, had the same ferrite grain size, it would
be reasonable to assume that this would give rise to a .
reasonably constant austenite grain size for all sizes of
stock, since the nucleation of austenite would generally
occur at grain boundaries and corners, which were
approximately equal in number in all cases.
A sample from each billet size was used to determine
its grain size using the intercept method. Three samples
were then taken from each size, heated at 900°C fb, 1, 2
and 3 hours, respectively, and air cooled. Grain size
measurements were carried out on these specimens, together
with measurements of hardness.
126

The grain size and hardness were plotted against


the duration of the treatment for each size, typical
graphs being shown in fig. 42. A suitable grain size was
chosen (0.019 mmo diameter) and all remaining stock was
treated to give this grain size.
(N.B. Hardness is not strictly applicable as
a measure of grain size in this case, as the fineness of
the small amount of pearlite present varies with cooling
rate, and, hence, with the specimen size and situation.)

5.5.2 Control of Oxidation during Preheating


It is established (60) that the presence of oxide
scale on the surface of steel, during hot rolling, has
a marked effect on the frictional conditions existino, within
the arc of contact. This effect was shown by El—Kalay and
Sparling, who related the load and torque functions to the
thickness of scale present.
In the present work, all specimens were given a
standard preheating time of 30 minutes, which was known to
be long enough to ensure that the billet had attained the
temperature required, but too short for much grain growth
to occur. In this time, heating in air produced appreciable
oxidation of the specimens and it was thought necessary
to use some form of controlled atmosphere.
The first rolling trials, carried out on the mill,
were performed using a forming gas controlled atmosphere
of the composition 75 " 2, 25% H2. However, severe
fluctuations ( 20) of the load and torque traces
indicated that those parameters were subject to considerable
variation during the passage of a single specimen through
the roll gap. Some possible causes of this effect were
considered to be
1) variations in the stock thickness
entering the roll gap,
2) variation of temperature along the
length of the specimen,
127

0.025

0.020
Grain size
Dim* mm.
0.015

0.010

0 1 2 3
Hours Treatment (900 °C)

fig. 42. Effect of the duration of the normalizing


treatment on the grain size and hardness

of 0.30 in. stock,


128

3) variaLion of frictional conditions


caused by inconsistencies of any
oxide scale, present, on the billet
surface,
4) variation of the extent of roll
flattening caused by variations
in the structure of the rolls,

and 5) roll eccentricity.


Of these, 3 was considered to be the most likely, since:

1) calculation showed that the sort of


variations of entry stock thickness
found in practice, could not give
rise to fluctuations of the order
obtained,
2) temperature uniformity tests showed
that any changes of temperature along
the length of the specimen could not
match the observed periodicity or
magnitude of the fluctuations,
4) if this possibility, alone, caused
the variations, then some recurrence
in shape of the fluctuations would
have been expected. However, none
was observed.
5) See 4).
Closer inspection of the furnace chamber revealed
that at. the flow rates of forming gas used, the hydrogen,
therein, was burning at the gas inlet in the furnace,
combining with oxygen leaking into the chamber and
producing 'water vapours, which at 10000C is highly
corrosive.
Oxygen-free nitrogen was tried next as the
controlled atmosphere, and this appeared to be a little
better; the fluctuations were reduced but far from
eliminated. It was decided to attempt a quantative
assessment of these two gases.
129

Some cylindrical test pieces, 1.25° in diameter


and about 0.4" thick, were made from the experiment steel,
and nickel plated around the circumference and on one
of the faces, the second plane face being left bare to
the furnace atmosphere. These were carefully weighed,
inserted in the furnace and held at 1000°C for 30 minutes,
after which time, they were quenched and reweighed, together
with any scale, which had come away on quenching. The
increase in weight was used as a measure of the amount of
'combined oxygen'. Prior to insertion of the specimens,
the furnace was thoroughly flushed through with the
atmosphere being used and during the tests a significant
flow rate-of the gas was maintained. However, oxidation
took place in all cases.
Atmosphere Weight Increase (gm)
75% N27 25% H2 0.57
0051
corer'. 0.25
0.1G
0.21
The increase in weight in the O.P.N. atmosphere was
approximately of that obtained using the forming gas.
Thus, although the oxidation could be reduced, it was not
possible to eliminate it entirely.
A further test in a closed tube furnace, with no
leaks, showed little difference between 0. .N., forming
gas and argon. All produced only a slight colouration
on the specimens, which were heated at 1000°C for 30 minutes,
then quenched. Thus, it would seem that oxidation occurred,
because of an ingress of air through the gaps in the
furnace casing; attempts to correct this by maintaining a
positive pressure of gas within the furnace required
prohibitively high flow rates of gas.
It was decided to restrict the flow of gas over
the top surface of the specimen by placing on it a stainless
steel cover plate. The under-surface of the specimen was
130

naturally protected by virtue of being in contact with


the floor of the furnace. Good results were obtained using
this method; little if any, oxide could be detected on
the test pieces after rolling and quenching, the surface
having a metallic appearance with a slight blue-ish tinge,
normally associated with low levels of oxidation, as
opposed to the reddish-brown matt surface produced under
heavily oxidizing conditions. Roll force and torque traces
indicated that, while this method did not give completely
oxide--free conditions, it apparently produced a consistent,
reproducible surface on the specimen, as the fluctuations
of those traces were all but eliminated.
It is interesting to note that El-Kalay and SparlincT
suggest that fluctuations in the roll separating force
will occur as the amount, and quality, of the scale changes,
and that frictional variables should be controlled at the
level of maximum reproducibility, rather than trying to
eliminate completely the formation of scale.
During experimental rolling, the protective plate
was removed from the test piece immediately prior to pushing
the furnace forward into the mill housings.

5,5.3 General Physical Condition and Surface


Finish
It would be anticipated that the best experimental
results would be obtained on specimens machined to a high
standard of dimensional accuracy and shape. To deal with
all the specimens in this way would have been, however,
laborious, time-consuming and costly and it was thought,
therefore, desirable to check on the importance of this
sort of consideration,
AczoPainsly a short rolling programme was planned
using two types of specimen. Some test pieces were selected
and machined flat to exact dimensions, and then surface
ground, while other pieces had only the excess surface
oxide, which had originated from the normalizing treatment,
removed by sand/grit blasting. The rolling geometries of
the machined and as received specimens were similar, and
131

the results are shown in Tables 1(a) and I(b).


These show that, although the general level of
the results is very high, there were no significant
differences between the machined specimens and those rolled
in the as received condition, and that all results for
a given size exhibited no more scatter than is normally
obtained in work of this kind.

Thus, it appeared that the specimens could be used


in the as received condition with no detrimental effect
on the results.

5.504 Measurement and Accuracy of Specimen


Dimensions

The thickness and width of the specimens were


carefully noted before each experiment. The width was
readily measured with the use of a steel rule, in spite of
slight barreling effects, and was generally very consistent,
any variations being less than + 0.75.
The thickness was measured using a micrometer
screw gauge. The figures given in the tabulated results,
represent an averacje of nine measurements taken on the
specimens. These were carried out on both edges and in
the centre of strip, in three positions, corresponding to
the two ends and the middle of the specimen. On no single
specimen was the variation of gauge found to be worse than
+ 1%.
Measurement after rolling was carried out in similar
manner, but generally the dimensional consistency was better
than + 1-7 at this stage.

5.6 Sioecimen Preparation


To prepare the specimens for rolling, the original
lengths of steel were sawn into 6 in. pieces, any burrs
were removed by filing and the scale produced in the
normalizing treatment was removed completely by sand/grit
blasting.
132

Table I(a) - Load


t1 Average
Condition P/Pc P/Pc Deviation
Nominal

machined 1.227 - 1.37%


machined 0,2 1.244 1,244 -
as received 1.240 - 0.32%
as received 1.265 + 1.69%
machined 1.127 -- 4.33%
machined 0.3 1.242 + 5.43%
1.170
as received 1.173 - 0.42%
as received 1.169 - 0.76,!
machined 1.131 - 0,59%
machined 0.5 1.202 1.188 + 1.16%
as received 1.198 + 0.83%
as received 1.172 - 1.37%

Table I(b) - Torque


t1 Average
Condition GVGc G ,/Gc Deviation
Nominal

machined 1.138 -0.18%


machined 0.2 1.163 + 2.02%
1.140
as received 1.120 - 1.75%
as received 1.130 - 0.=:
machined 10053 - 0.09%
machined 0.3 1.130 1.054 + 7.21.
as received 1.009 - 4.27%
as received J-023 - 3.04%
machined 1.212 + 0.92%
machined 0.5 1.284 + 6.91%
1.201
as received 1.161 - 3.33%
as received 1.147 - 4.50%

Note. Pc and Gc are theoretically calculated values.


(Equations 86 and 87)
Table I(a and b). Showing the Effect of Specimen Condition.
133

5.7 EAperimental Procedure


The prepared specimen was measured and charged
into the furnace. The stainless steel cover plate was
positioned on top of the test billet, the furnace was
closed and the gas flow rate increased threefold to flush
the system and alleviate air leakages durina the charging
process. After about 3 minutes, the O.F.N. flowrate was
reduced to its normal rate of 0.5 cu.ft/min. During the
30 minute heating and soaking period, the rolls were abraded
and cleaned, and degreased, thoroughly; the roll gap was
set or checked, as was the mill speed setting.
At the end of the soaking period, the furnace was
opened, the cover plate removed, and the mill was started.
The furnace was pushed forward into its delivering position
and by means of the push rod, the specimen was introduced
into the roll gap, directly from the furnace chamber. The
specimen was collected on the exit table and water-quenched
irmediately. When the specimen was cold, it was remeasured.
The ultra-violet recorder was started immediately before
the specimen entered the roll gap and stopped when it had
left, by means of a remote control unit, situated on the
furnace and operated by the experimenter. The recorder
speed was set to 12.5 cm./sec. under normal operating
conditions.
The magnitude of the total roll separating force
and total roll torque was ascertained by summing the two
components of each of these parameters. These, in turn,
were obtained by measuring the average deflections, produced
on the ultra-violet recorder chart paper and refering to
the appropriate calibration technique.
Fig. 43 shows typical load and torque traces obtained
during a single pass aid in each case the average deflection
is shown. This was defined by the expression:

"00
8au•gpole .1*
r,„,
134

20

-r b

10
Deflection

d average ( 1-4) •

Time ---->

fiq.43. Typical load and torque traces.


135

where d represents the deflection,


the time, and the subscripts
a and b refer to those times
indicated in fig. 43, which were chosen so as to exclude
the rise and fall portions of the traces. In most cases
the deflection was sufficiently constant for the average
value to be obtained with adequate accuracy by eye, without
recourse to a planimeter.
It is interesting to note that the torque deflections
are complementary, indicating that, whilst there are
variations in the torque on either roll during a pass, the
total torque remains effectively constant.

5.8 Outline of Exj2er.P-al Rol line Programme


Series A:
The theoretical considerations discussed in section
4.1 predict that all solutions of a particular slip line
field type should have identical values of the load and
torque functions, /kL and G/kL2, as well as of the
deformation zone aspect ratio, L/(tit2)2. A short series
of rolling experiments was planned, in which the geometries
R /t2,
, were calculated
of rolling, defined in terms of r and
to lie as nearly as possible on the lines corresponding to
the circular, and 5/722°, types of slip line field solution
as shown in fig. 18.
The reductions given to each billet size and the
corresponding R/t2 ratios are shown in Table II. Each
geometry was carried out in triplicate.
Series B:
This consisted of a comprehensive series of rolling
geometries, as shown in Table III, carried out at 10000C,
in triplicate.
Series C:
At this stage in the experimental programme furnace
B was brought into operation. In series C, some of the
geometries corresponding to 30% reduction, which had been
investigated in series B, were repeated in duplicate at 1000°C.
136

Single circular discontinuity exit semi field type

t1 inches
' r % R/t 2
1

Nominal

0.5 36 16
0.4 31 18
0.3 25 22
0.2 18 29
001 10 54
0.07 S 75

5/71° exit semi field type

....
t1 inches
r R/t 2
Nominal

0.3 40 27
0.2 32 35
0.1 19 60
0.07 14 81

Table _
11. Geometries Investigated in
o
Series A (1000 C).
137

Series D:
Those geometries, investigated in series E and
C, provided very few results corresponding to L/(tit2)2
> 5. In series D, heavy reductions were carried out on
relatively thin stock, producing results for L/(t1t2)2 = 7.
Series E and F:
Owing to the limited numbers of available specimens,
series E and F were carried out to investigate the effect
of the factor L/(t1t2)2 only, on the load and torque functions
at 900°C and 1100°C, respectively. The geometries
investigated, which were carried out in quintuplicate, are
shown in Table IV.
Series G:
See section 6..
13P)

Nominal t1 (in) Reduc. tions Investigated (%)

0.625 10 20 30
00500 10 20 30
0.400 10 20 30 40
00300 10 20 30 40
0.200 10 20 30 40
0.100 10 20 30 40
0.070 1.0 20 30 40

These geometries could not be carried out since the


high torques associated with them caused the mill to
stall.
o
Table III. Geometries Investicated in Series B (1000 0

Nominal t1 (in) Approx. Reduction Approx.


L/(t1t2)2

00625 10 1.0
0.500 20 1.6
0.400 30 205
0.300 40 3.5
0.200 40 4.1
0.100 40 6.5
0.070 40 706

Table IV. Geometries Investigated in Series E (900°C)


and Seri .....
F (1100°C)
139

CHAPTER 6
Results

The results obtained in the experimental rolling


programme are tabulated fully in Appendix II, together
with all the associated calculated parameters.

6. 1 Calculation of the Load


- and Torque
. Functions

= —
The experimental load function, /kL, was
calculated simply by dividing the measured load, P,
relating to one inch width of the material, by kp ,
obtained appropriately from the data of Cook and McCrum,
and L, the length of the deformed arc of contact.
A similar procedure was adopted initially for
obtaining experimental values of the torque function,
G/kL2 using the appropriate shear yield stress, kG.
However, inspection of the lever arm values, m, associated
with the measured load and torque values, indicated that
the latter parameter was of considerably smaller magnitude
than had been anticipated.
The lever arm is defined

m
G
2.15L

and is a measure of the position the centroid of the roll


pressure distribution, related to the length of the arc
of contact. This parameter has been used extensively
in the theoretical rolling field, since it is very sensitive
to small changes in the shape of the roll pressure
distribution. However, by the same token, it is subject
to a relatively large amount of experimental scatter,
since, by virtue of its definition it embraces any
inconsistencies in both the load and the torque parameters.
Experimentally determined values of the lever
arm were, in some cases, as low as 0.280. However, it
is difficult to visualize a value less than 0.333, in
the absence of tension, since this would correspond to
a right triangular roll pressure distribution, in which
the neutral point and the exit plane are coincident.
140

As values of the lever arm as low as this cannot correspond


to a real situation, it was considered necessary to
correct the measured torque values, to take account of
the fact that, due to roll flattening part of the roll
pressure diagram lies outside the line joining the roll
centres. The method used has been given by Hill and
relates directly to equation 59 in section 2.9. The
quantity considered as the experimental torque is given by,

G° Gmeasured + PL(R'
-- - R)
-----
R'

which is directly comparable with the values predicted


by the theory.
The effect of this correction on the lever arm
ratios is shown in fig. 44, in which m and m , , defined
by the expression,

_ G*
2PL

t-)2 for those results


are plotted as a function of '/(t12
obtained at 900ct.
The lever arm ratio, m is clearly a measure
of the position of the centroid of the experimental roll
pressure distribution, s, that is, any value of m* includes
the effect of the yield stress varying in the Cefo=tion
zone on the overall pressure distribution. On the other
hand, those theories, assuming a constant mean yield stress
in the arc of contact, predict a theoretical lever arm
based on the dimensionless (s/k) distribution. It can
be seen that the comparable lever arm ratio, based on
s
the experimental dimensionless ( /k) distribution, may
be determined by the expression,

mo /krT,2
/k PL

i.ee the ratio of the experimental torque function


to twice the experimental load function.
0.5

\•
m* t - 4
• . •

N •
s•-• -
• - •
0.4

Cb
0

0.3 o

0 1 2 3 4 5 6 7 8
L/trn

fig. 44. Measured (m) and corrected (m.


) lever arm ratios for Series E (900°C)

plotted against Li
t m•
142

6.2_ The Slip Line Field Theory

The rolling of a series of specific geometries,


all corresponding to a particular type of slip line field
solution is difficult because of the accuracy required
in matching both the reduction and the /t2 ratio to
a particular deformation mode. It can be seen that,
whereas normally it is required to predict the total roll
separating force to take account of the mill spring only,
in this case it was necessary also to predetermine the
effect of the total mill load on the deformed roll radius.
The situation was further complicated by the fact that
the ratios r and R/t, are closely interrelated for a
given type of slip line field solution.
However, some of the geometries rolled in
Series A were considered close enough to the required
deformation mode to justify comparison of the theoretical
and experimental load and torque functions. Fig. 45
shows the successful experimental geometries together
with the theoretical relationships of the 5/7--(3 slip
line field solution and the solution incorporating the
single circular discontinuity. Figs. 46 and 47 show the
experimental load and torque functions, respectively,
obtained for those geometries given in fig. 45. These
are plotted as a function of "/t,, together with the
theoretical relationships, for the two types of solution.
Fig. 46 shows that the load function is approximately
constant for the two types of deformation mode investigated,
over the experimental range of ' R /t0 ratio. However, it
can be seen that the value of the load function is not
identical to that predicted by the slip line field theory.
This would suggest that, whilst that theory may not be
P
strictly accurate in its predictions, plots of /kT,
L
against /(Lit2) will exhibit very little reduction
dependency. Similarly, fig. 47 indicates that any degree
of reduction dependency exhibited by the expermental torque
functions is only very slight.
60

50

40

r % 30

20

10

0
0 50 100 R/t2 150

fig.45. Rolling geometries which were considered to correspond most closely to

the two types of s.l.f. solution investigated. ( )


5 / 711 (0) 0

0 0
•kL Q-
ro- 8 8 0

0
10 20 30 40 50 60 70 80 90 100
R42

fici.46. Experimental and theoretical (s.l.f.) load functions. (Series A)


6 r

0
4 0

5/71/1 ( 0 ) 0 0
G* 0 0 0-
2 3 0
k L --B
G Er g
08
0 o....
'o

0 I I I t I I I I
10 20 30 40 50 60 70 80 90 100
R42

fig.47. Experimental and theoretical (s.l.f.) torque functions. (Series ,A)


146


6.3 Other Results
With the exception of the results shown in figs.
46 and 47, all other results ::salysec] in a siilar way
to that discussed in Chapter 4.
Accordingly, plots were obtained for the
(3 * 2
experimental load and torque, /k L and /kGL ' as a
function of L/(t1t2)2 for temperatures of 900, 1000 and
11000C. These re shown in figs. 40 - 53. Figs. 54
and 55 show the results of Series B replotted, distinguishing
between the four separate reductions investigated, i.e.
10, 20, 30 and 40%.
The lever arm ratios, mo are plotted against
j/(t1 t2 )- 7 for the three rolling temperatures investigated,
in figs. 56 - 58 and fig. 59 shows the reduction dependency
of this parameter exhibited by the results of Series D.
Since the theories of Orowan and Pascoe and
Alexander and Ford conflict in respect of the geometrical
parameter employed in the determination of the load and
torque functions viz: Litt in the former and 2L/(t+t2)
in the latter, it is interesting to investigate the effect
of these parameters. Figs. 60 and 61 show /kL plotted
as a function Lit2 and 2L/(t1 +t2 )' respectively, for the
results of Series B, at 10000C; these may be compared
directly with fig. 54.

6.4 Discussion
As in usual in this type of work, there is a
fair amount of experimental scatter. Figs. 54 and 55,
drawn to show the existence, or otherwise, of any
reduction dependency, over and above the effect of the
parameter, L/(t1t2)2 , show that any dependency, if present
is submerged in the general scatter of the points.
Noreover, comparison of figs. 54, 60 and 61 shows that
there is little difference in the reduction dependency
exhibited by the load function when plotted against any
of the parameters L/(tit2)2, L /t2 or 2L/(t +t
1 2 ) 7 in
spite of the fact that each of these is related quite
differently to the reduction, rf
so

••

PL

...J

S.L.F

1 2 3 4 5 Lt 6 7 8 9 10
tm

fiq.48. Experimental load functions. (9000C)


S.L.F.

1 2 3 4 5 L/ 6 7 8 9 10
tm

tia.49. Experimental load functions. (10000 C)


8 a

S.L.F.

7 to

.."
P<L
P 5

•0 •

4
Ss /--.--
co -•
--)
...''''
3 ir
4!
...........--"-.".
6 ...."'
••
2 I I i J I I 1 1 _J
0 1 2 3 4 5 LLm6 7 8 9 10

fiq. 50. Experimental load functions. (1100°C)


7 I I I I I I I I I

0%•
6
• ----"----
5
,--
4
,
G*,
-'' 2
k L
G 3 ...
4
.
/
...."
B
2 ...."

1
S. L.F.

L I J I I I I I I
0
0 1 2 3 4 5 L. m 6 7 8 9 10
--zt

fiq.51. Experimental torque functions (900° C)


0
0
0 0
00

0
0
...---
0

e
.....---"
O

.13--4.--'-
---• e
e
0

S.L.F.

1 I 1 It 1 1 I t 1

1 2 3 4 5 L/t m 6 7 8 9 10

fiq. 52. Experimental torque functions. (1000° C


8
o•

7

••

••
••


3 • ED

$ •

2
S.L.F.

1 -t

0 1 2 3 4 5 6 7 8 9 10
m

fiq.53. Experimental torque functions. (1100° C)


A
A

...„.„,...-70.

P / a6-
rx °
:
kpL

• 10% reduction
x 20%
S.L.F. o 30%
A 40%
I I I t 1 I I I

1 2 3 4 5 L 6 7 8 9 10
4rn

fig.54. Experimental load functions indicating the effect of

reduction, r . (Series 8 1000°C)


A
0
O 1e,

,x-""
•__.. .f. ..... --- x 4 0

...„ --r.• . cbt4,


co,
I 5, z 0 .

• 10% reduction
x 20%
S.L.F. 0 30%
A 40%

2 3 4 6 7 8 9 10

fig.55. Experimental torque functions indicating the effect of

reduction, r . (Series B, 10000 C)


0.6

0.5

0.4


0 1 3 4 5 Lz 6 7 8 9 10
't
m

fici.56. Corrected experimental lever arm ratios. (900 °C)


0.6



• 0
• •\
• •\
• •\
• • •
• • •
‘ • • • \
\ • • • •• •
\ • • • • • ,•
‘ • • # • ....
0.5 _ ‘•\ • •0 40# • , . , .
\ 6 4 '-o —
— s•-• — r
\ • • • e •
• * 410 il• 0 . . Co 0
• 0 •
• • •
: •
•. 00.
\ • • o •0
. ..•0•00
\

.-.. %
.... -5..... • ) I

0.4 ISM

I I

0 1 3 45 ivt
6
7 8 9 10
m

fig. 57. Corrected experimental lever arm ratios. (1000 ° C)


0.6


0.5



•so a

• •
°.2 a)

0.4

0 1 3 4 5 L 7 8 9 10
/t 6
m

tici.58. Corrected experimental lever arm ratios (1100 0 C)


0.6
o 10% reduction
x 20%

o 30%


x
• e
A 40%

X
N
•• •
• N
XX
\ 0 • X
0.5 x.
\c• eo x — -A
>0o 0-
\ • o A
\ • '° o x 0
x 0
x 0
eP X
A
• 0
A
"-A
0
• ••••• •••• •

0.13


0 1 3 4 56 7 8 9 10
71-trn

f i 0.59. Lever arm ratios, various reductions. (1000° C)


7 1

A •'
A
6

5 A

0 0
••• 0

4
0

•k L
P 3
0 •

2 -- tettg r<

• 10% reduction
1 x 20%
O 30%
A 40%

0
0 1 2 3 4 5 6 7 8 9 10
/t
2

fiq.60. Load function plotted against tYf, (1000°C)


2
A

/A

/00

00
•/
,x
X X
..•••

cb"
O
8 •
— —el

• r = 10% reduction
X r = 20%
• r = 30%
A r =40%

1 2 3 4 5 6 7 8 9 10
2
t2)
fig.61. Load function plotted against 2L.. (1000°C)
(ti+t2)'
161

L
(t1t2)2 t2

2L R V.2.(r(1-r))--)I
t1--1-t 2 t2 (2-r)

An important point, which can be seen from the


overall picture presented by the experimentally determined
plots of P/kL against L/(t1t2)2 , is that any line drawn
through the points must show a pronounced degree of
curvature, a phenomenon which has no counter-part in
any of the theories discussed so far. Further, although
the general level of most of the experimental points at
900°C and 1000°C is consistent with the theoretical
predictions, the measured values became much higher, when
L
/(t1t2)2 is greater than about 5, and at 1100°C, the
measured values are very high indeed.
Tt was considered that the reason for this
resided in the fact that, with the experimental mill
used, the attainment of high values of the aspect ratio,
Ptit2)2 , necessitated the use of very thin specimens,
with a correspondingly small heat capacity. Consequently,
whilst the billet temperature was uniform and accurately
known at the moment it entered the roll gap, the conduction
of heat away from the strip by the relatively large mass
of the roll was sufficient to ensure that the average
temperature of the strip, within the deformation zone
at any instant, was significantly lower than nominal.
This had the effect of raising the yield stress of the
strip and, hence, the roll load.
162

To test this effect, a short series of experiments


(Series G) was carried out at 1000ct, in which test pieces
of similar starting thicknesses were rolled with similar
heavy reductions at various rolling speeds, thus varying
the time spent by any part of the specimen within the
roll gap. The effect of this on the load function is
shown in Table V. These results indicate the necessity
to correct for the effect of roll quenching.

Had the geometries of rolling been identical


for all those specimens in Series G, corrections could
possibly have been made, on a purely empirical basis.
However, as the rolling geometries were subject to
variation, the differences measured in the experimental
load functions were a result of both the temperature
effect and the difference in the geometry of rolling.
It was decided, therefore, to consider the quenching
effect from a theoretical standpoint.
16?)

Run No. Mill Speed r.p.m.


kL

G1 30 6.669 8.082
G2 25 6.703 8.016
G3 20 6.772 7.243
G4 15 7.675 7.591
G5 10 8.494 7.453
G6 5 9.821 7.223

Table V. The effect of rolling speed


on the load function.
164

CHAPTER 7
The Effect of Roll 'Quenching' on the
Material within the Roll Gap

7 01 A Sig ip ivied Model of Transient Heat_Coplaction


during the Hot Rolling of Steel
The following assumptions were made:
IL) Heat conduction, in the rolls, takes place
in the radial direction only.
2) There is no thermal contact resistance at
the roll/stock interface, i.e. the heat
transfer coefficient is assumed to be infinite.
3) The thermal properties of the roll and the
strip are identical.
Assumptions 2 and 3 correspond in effect to the
single assumption that the system made up of the top roll,
the bottom roll and the strip is continuous. It is then
possible to apply the well known solution relating to an
infinite body, in which the region -14:x4:71 is initially
at a constant temperature Os and the regions Ixt
are initially at zero temperature. Thus, in this case,
En is equal to the difference between the nominal rolling
s
temperature and the temperature of the rolls, i.e. room
temperature.
The temperature distribution at any time
,equation(68):
is then described by the

erf 1-x erf l+x


12.- °s (98)
247iA 2

0
wherect is the thermal diffusivity. Fig. 62 shows /Gs
x
plotted as a function of /1 for various values of
40re\ /1.
Top Roll Strip Bottom Roll

1.0

e
0.5
''°/1 S

0
x=0 X +

fig. 62. Relationship between € ves and xA for various values of ZIT.
I—
166

The average temperature of the material within


the limits -1<x<1 at any time 1t is defined by the
expression.

+1
( (99)
1 0.dx
21
J 1

which can be shown to be

0 s
av
= -
©s •1 ierfc [0) - ierfc 1 1(100)
1

Thus, the average temperature drop through the section,


21, after a time

(101)
ay C9 c- i c(:\
1 0.5642 ierfc
1

In the case of rolling it is necessary to employ


a mean value for the thickness, 21, as this is a function
of the position in the arc of contact, which in turn is
related to V. Bradley et al(69) suggest (t1t2)' for
this quantity and since the geometric mean thickness has
been found appropriate in the earlier sections of this
work it is again employed.

Thus
VOP

tS0 Os ,2.J77F\ 0.564-2 ierfc(t t 2)2 (102)


av i
(t1t2 )2 2 p k

Bradley et al., working on similar problems,were


unable to make assumption 3 above, since they were
considering the hot rolling of aluminium. However, if
their corresponding solution is modified for rolls and
billet of identical thermal properties, the following
equation is produced
167

A Gs.2. [0.564d
E1V

(t1 t2 )2

This agrees well pith the expression above, since


(b.1t2)T /2170 C7\ is negligible in many cases
the term ierfc
of rolling. The difference between the two equations
arises from the fact that Bradley et alo assumed both
the rolls and the strip to be semi-infinite.

7.2 The Time spent within the Arc of Contact


In order to apply equation 1OT, it is necessary
to obtain some suitable value for
The horizontal velocity of the material in the
roll gap is

dx - vntn cos 0 n - vn t n
dr(

where v is the peripheral speed of the rolls.


n
The time, 't y, taken to reach any point, x, in
the arc of contact may be obtained by suitable integration

/ Tx x
d rt: t .dx
v t
n n
J

2 x2/R then
Assuming t = t2 + RO or t = t2 +

(`fix
t + x2
dT= - 2 dx

vntn
J 0

i.e. - 3 I
t-x - 1 t,x + X
(103)
vntn 3R L
168

The total time, T tot, spent by any particle within the


roll gap is found by putting the upper limit, x, equal
to zero.

i.e.
1 .L. t2 + L2
t tot ( 104 )
vntn 3R

Now
L2 = (t1 - t2)
R

0 o

tot

L t1 + 2t,
vntn 3 3

According to Ford and Alexander ((4)

vntn 1 = vntn.r
(105)
L t2 L t2

1 r o t1 + 2t2
Z tot
A 3t2

or
1. o r [3 -
r tot (1.06)
X 3 1 r
169

However, at any instant, only the material in


the exit plane will have been subjected to the quenching
effect of the rolls for the whole of the time tot
and the temperature drop, Ai0av , given by equation 102,
if ft tot is substituted, refers only to that material
in the exit plane.

7.3 The Overall Average Temperature Droj


The temperature drop averaged out over the whole
deformation zone will be

L
av .dx (107)
Gcrit

L 0

or

Gcrit
68av od x (108)
L

where

6 Gay = 2.G,. 0.5642 - ierfc (t1t2)2


(t1t2)2 2 ,pkirA_

As a first approximation, the term


(t
ierfc 1t 2 /2,10crk may be neglected. (The effect of this
will be discussed in section 7.6).
In which case

8 Gay 1.1284 Os jot -c


(109)
t t
1 2

and 1
1.1284 Gs
Qcrit JTA d(1 (110)

t1t22 ,
170

Now

or

set 2 (110a)

Substituting

tyl t_ 1.72

from equation 105 and working in terms of (x/Iii)

2 3
Ix ) (110b)

Therefore

(110c)
t.12.8ki Os I fx

where

and (3- 21•)


3 , _r

z
and C, r
3 —r)

171.

The integral I was solved numerically for reductions


of 5, 10, 20, 30, 40 and 50% (fig. 63), and plotted against
r (fig. 64).
The expression

I = (0.67 - 0.17r)1 r

was found to fit the relationship almost exactly.


Thus

6 0 crit = 1.1234 © s .(0,67 ---0.17r) ( ys (112)


1-r

Table VI shows the values of I, obtained a) from equation


011) and b) from the numerical integration, as a function
of r, the reduction. Gk. , the thermal diffusivity of steel,
varies with temperature. Fitzgerald and Sheridan(70) show
c< increasing with temperature over the range 800°C - 1100°C
from about 0.0058 in2 sec-1 to 0.0080 in2 sec-1. An average
value of 0.0069 in sec-1 was used for all the present results.
The overall average temperature drop AGcrit was
then used to calculate a corresponding increase in the
yield stress.

7.A„ The Effect on. the Yield Stress kp


Examination of the yield stress values given by
Cook and McCrum(42) shows that the rate of change of yield
stress with respect to temperature, i.e. dkp/dG, is relatively
insensitive to temperature and strain rate; a typical plot
being shown in fig. 65. However, it is strongly affected
by reduction, as shown in fig. 66.
By a suitable logarithmic plot, the relationship

dkP -0.0152 r0.28


(113)
d0
was found to describe the effect adequately, over the
172

0.9

0.8

0.7

0.6

__0.5

- 0.4

___ O. 3

0.2

0.1

x% 1.0

0 0.5

ficj.63. From which the numerical integration of I

may be obtained.
173

0.7 r ►

0.6

0.5

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5
r

fig. 64. The value of the integral, I, plotted as a

function of the reduction, r.


6.0

fig. 65. Effect of

temperature on yield

5.0 stress of low carbon

kP steel.

shear (From data of Cook


yield
stress & McCrum)

t.s.i. 4.0

3.0
900 1000 1100
Temperature °C
fig,66.Effect of r, the
0.010
reduction, on the value

of d kl="
d8
d lc
P
de

0.005

0
0 10 20 30 4-0 50
re)/0
176

r (a) I (b)

5 0.15176 0.15155
10 0.21767 0.21810
20 0.31000 0.31735
30 0.40523 0.40505
40 0.49151 0.49105
50 0.50500 0.58520

Table VI. The Integral , I.


177

temperature range 900°C - 1100°C.


Thus, the roll quenching effect on kp was
allowed for by making a simple adjustment to the value
according to the expression

kP (‘ dk
P
rolling temp crit P (114)
dP

crit is, of course , negative, so the overall effect


is an increase in the yield stress.

19 The Corrc,ction to k

The temperature correction for kG, the yield stress


for torque calculations, is complicated by the fact that
useofA8crit as previously obtained will overestimate the
effect of the temperature drop in the vicinity of the entry
plane, where the associated lever arms are greatest. Thus,
employment of an equation such as,

kG kG
rolling temp Af'4crit

is not possible.
However, since the roll quenching effect will affect
only the yield stress, and possibly, to a far lesser extent,
the coefficient of friction, the distribution of the
dimensionless parameter s/k, in the roll gap, will not be
significantly altered, so corrections to kG were carried
out according to the expression

1 1
kG = kP .kG
(115)
kP

178

Discussion
It is necessary first to consider the effect of

neglecting the term ierfc(t1t2)2/2 in the calculation
of agcrit°
This may be effected by calculating the difference
in AGav at the exit plane, which, in turn, relates to the
relative magnitude of the factors 0.5642 and ierfc(tit2)472f rTot
in equation (102). In the present series of experiments,
the most severe case of roll quenching was obtained during
the rolling of specimen- G6. Because of this, and since the
Series G represents quite a wide range of the higher
temperature losses, it is sufficient to examine this series
only.
The following list shows the value of the factor
---
ierfc(tat2)2/2TxT\ together with the ratio of this value
tot'
to 0.5642, expressed as a percentage, for each of the
experiments in Series G.

(t t,
Run ierfc -1 '7 - (approx.) % of 0.5642
. -2 ,STEtIot
G1 0.0006 0011
G2 0.0010 0.18
G3 0.0025 0.44
G4 0.0058 1.03
G5 0.0127 2.25
G6 0.0503 8.92

Thus, in the worst possible case, CAgav is


affected by about 9% at the exit plane. Clearly, in the
region of the entry plane, neglect of the ierfc term will
have no effect at all, so the overall effect on A
will be considerably less than 9%, in the worst case. If
the error in 6,0av varies linearly through the roll gap,
then the effect on 0Gcrit for the worst case, i.e. G6 i
will be only about 4.5%. Since the heat losses in specimen
179

G6 were considerably greater than those of any other specimen,


it can be seen that neglecting the ierfc(t1t2)2/24;TF\
term will have very little effect on the overall average
temperature drop in the roll gap for the present results.
Corrected plots of P/kpL and G*,/kGL2against
L/(t1t2)2 are shown in figs. 67 - 72 for the temperatures
900, 1000 and 1100°C. The general level of the results is
significantly lower than any of the theoretical relationships
previously considered and a line drawn through the results
still exhibits a significant curvature. Considering how
these results may be explained, the effects of friction and
Orowan's inhomogeneity factor w, must be considered.
P
kPL

.."

S.L.F.

I I I I I I 1 L I

-)
1 2 3 4 5 L..4 6 7 8 9 10
m

fi a . 67. Temperature corrected load functions, (900°C)


0
08
A•
k L 0
0
P •0 ••
0

fig.68. Temperature corrected load functions (1000° C )


S.L.F.

I I I I I I i I 1

LAni 7 8 9 10
1 2 3 4 56

fig.69. Temperature corrected load functions. (1100° C)


7

4 00

0
0 1 2 3 4 5 7 8 9 10
L 6
/tm

fiq.70. Temperature corrected torque functions. (9000 C)


• 0

•• •
•0.
0

•0 • •
• • ••
co •
• • •
• 0 •P 0
9
4 •
• is
c °
V 4 I •
7 r

S.L.F.


1 2 3 4 5 Lz 6 7 8 9 10
tm
0
fiq.71. Temperature corrected torque, functions. (1000 C)
0

0
00
7
tc •

1 2 3 4 5 L4rn 6 7 8 9 10

fig. 72 Temperature • corrected torque functions (1100 °C)


186

CHAPTER 8
A Simplified Theory of_HotRollinia,
incorporating Mixed Friction Conditions
and a Varying Inhomogeneity Factor
in the Regions of Coulomb Friction

It was shown, in section 3.5, that the theory of


El Malay and Sparling(60) neglects the variation of the
inhomogeneity factor, w, in the regions of Coulomb friction
in the arc of contact; a fact, which, it was considered,
may affect significantly the theoretically derived load
and torque functions.

8.1 Solution of the Differential Equation of the Friction


Hiller in the_Zonssof Dynamic Friction
(2) may
The equilibrium equation due to von Karman
be expressed:

df = 2Rs(sin 0 + tA, cos 0),

(see equation 20)

in the regions of slipping friction, where = ikUs.


Using the shortened form of Orowan's expression relating
f and s:

f = t(s - kw),

together with the expression due to El Malay and Sparling


relating the inhomogeneity factor, w, to the ratio of the
shear stress at the roll/stock interface to shear yield stress
of the material,

2 1-
W 2(1 - 0.38315 ( kks/k) )2

the solution in the regions of dynamic friction may be


obtained as follows:
187

1) Assume sin 0 = 0 and cos 0 = 1,

df = 2Rs(0
do
Substituting for f

d tk (s 2Rs(0 IA) (116)


d0

(f - w) d (tk) + tk d s 2Rs(0 (117)


-d-
(1 95 93 Y -- w

2) Assume tk d (s - w ) >> (:1s - w) d (tk)


dO k dO (118)

as did Bland and Ford(61).

Then d (s 2R s (0 (119)
dO t k P-)

or
d(s/k) [ 1 dw = 2R o s (0 + AL) (120)
d(s/k7 T k -

Now
dw= - 1.2 5326...P-2 (s/k) (121)
Cr:CF/17)

d(s/k), c 1 02.76G3,0.2 (s/sq 2R . s (0 1-/U,)


1
dO s.. 1- 0.38315)2R1 L k (122)

Divide through by s/1(

d(s/k) I 1 0.7663p.).2 . 2R (0 ti)


dO (s/k) 11 - 0.38315)U (123)
kr
188

Substituting t = t2 + RO2 7 then

11-"0 (V) dO d(s/k) 0.7663)a2 dO


(s/k) dO „t2 (s\
- 0.38315)
k)

2R(0 +j) dO
(124)
t2 A- RO2

Integrating,

—1
in (.9 + .1p6 s in f0.38315 s
/0.38315 i \k

(t2 R02 7117-1 tan-1 I-71 0 C (125)


R t2 t2

i.e.
in (s/k) 1.238/A, sin-1 (0.619Its/k)

in cst) ltan FR-10 C (126)


2

The constant of integration, C, may be found by consideration


of the boundary conditions in a manner similar to that u s ed
by Bland and Ford, since it is known that s _ 2
(11-1.5326,A. )2
at the entry and exit planes.
The resultant equations for the slipping regions are,

exit slipping zone (127)


189

F1(0 = c0 t
/4(H1 e 1-1)
entry slipping zone
ti
(128)

where
F1 (s) ..expi1.238 fr1/4.sin-1( 0.619pos\ (129)
k k 1J

C0
2 exp 1.238/ -1/ 1.5326,A,
2
2 -1-
(1+1.5326p.. )2 L Asin .1+1.5326/X

(130)

and
2 I R . I tan-1 . 1 0 (131)
t2

and C0 may be combined, i.e.

FIR) = F2 k
co (132)

Fig. 73 shows F2(s/k) plotted against (s/k) for a range of


coefficients of friction,/i.

8.2 Calculation of the Total Roll ,7.p


..parating Force
and the Total Torque
The total load/unit width may be obtained by
summing the above equations for the zones of dynamic
friction and the expressions due to Sims for the regions
where ps = k, i.e.

/ = +7r In t IT\ 0--tan


t
04Sims
" 2 o 2
190

1.t= 0.2

3
4 5
S/k

fig.73. The function F2—


( s/k)•

191

s In t + 2 R tan -tan -iR 01


()Sims /A Vt. )
2 -

where the subscripts o and i refer to the position in the


arc of contact of
1) the exit slipPstick° interchange point
and 2) the entry slipPstick' interchange point,
respectively.

The torque was obtained by calculating the moment


of the pressure distribution about the roll axes,
oc
thus
(s/k)d0
0

1

and G = R2
2k (s/k) 0d0
G 0

Unfortunately, the function F2(s/k) is not easily


solved for s/k and direct integration is not possible. It
was therefore found necessary to plot a series of dimensionless
pressure, s/k, distributions to determine the load and torque
functions, P/kL and G/kL 2. Fig. 74 shows a typical set of
completed s/k distributions, corresponding to R/t2 = 50 and
0.3, for reductions of 10, 20, 30 and 40%.
Fig. 75 shows a typical exit solution by this method,
together with the solution for the same conditions (R/t2 = 100,
m = 0.25) obtained by the method of El Kalay and Sperling.
It is seen that slipping friction occurs over a larger
fraction of the arc of contact with the result that the
load and torque values are correspondingly somewhat smaller.

00 3 Tha Load and Torcue Functions


The load and torque functions, P/kL and G/kL
obtained by the present - ciethod are tabulated as a function
of P./t2 and r in Tables VII and VIII , for fit= 0.2 and
i!
192

t
I
I
f

"7 ~--------------~--------------~~--~

~ stick
k ----------
slip
3

1
r = 100/0

o
o 0.05 0.1

fig.74. S/k distributions predicted ty the present work

for R/
/t -- 50 ,
2

\-
193

5 I

I
I /
I Present
I
I Work
I
I
I
I

4
Stick

Slip
-j~-
.8)1 I';:::1
~I
VJC:
qj"1
6>1
;:;1
-):2/
3 0:,1
1
I

S/, I
k

o
o 0.01 <{)0 lf>0 0.02 . 0.03
E.K.+S.

f;g.75. Comparison of present method with that

O_f_E_I-_K_a_l....;:,.a.>L-y_&_S-Lp"-'a_r_-'-'-lin~g=-----=--{ ~2 =100, J.l = 0.25)


....
194

P/kL

r
10% 20% 30% 40%
17

10 2.0fl5 ).111 2.08 2,026


20 2.177 2.245 2,264 2.247
50 2.359 2.530 2,646 2.71 2
100 2,594 2,907 3.151 3.291
150 2,793 3.215 3.615 3,946

G/kL2

r
10 20% 30% 40%

10 2.060 2.075 2,040 1.967


20 2.159 2.207 2,195 2.156
50 2.332 2.473 2.546 2.564
100 2.5513 2.030 3.010 3.100
150 2.756 3.105 3.437 3.657

Table VII. Load anc Torcrue Functions predicted


present method
195

ID/kL

\fir
10 30Y 40%
t2

10 2.112 2.180 2.207 2.195


20 2.238 2.3G7 2.447 2.498
50 2.494 2.797 3.042 3.242
100 2.057 3.391 3.846 4.186
150 3.14.4 3.097 4.426 4.910

G/kL2

r
RN, 10r: 20% 30% 40%
-2

10 2.094 2.140 2.144 '.110


20 2.16 2.316 2.362 2.374
50 2.468 2.729 2.912 3.039
100 2.820 3.293 3.677 3.905
150 3.105 3.791 4.214 4.571

50-)01e VTI1. LoydPPd ToTTIP Functions predicted


present method for Ak. = 0.3
196

0.3, respectively, whilst figs. 76 — 79 show these


functions plotted against L/(t1t2)2 , and figs. 80 and 81
show the corresponding lever arm predictions.
r=10%

20%
0----0 30%

40%


1 2 3 4 5 L 6 7 8 9 10
tm

fiq.76. Load functions predicted by present method ( :0.2)


7 1

r= 10%
2
• x x 20%
0-0 30%
1
40%

0
0 1 2 3 4 5 L 6 7 8 9 10
Vt
m

fiq.77. Torque functions predicted by present method (µ =0.2)


P/
kL

20%
0 30%

40%

1 2 3 4 5 L 6 7 8 9 10
/tin

fiq.78. Load functions predicted by present method. ( =03)


0----- 0 r =10%

x x 20%

0-0 30%

S.L.F. 4, A 40%

I 1 I I I I I I I

1 2 3 4 5 L / 6 8 9 10
't
m

tig.79. Torque functions predicted by present method ( µ :: 0.3 )


0.6

0.5
m X-
0

r=10%
0.4 20%
30%
40%

0 1 2 3 4 5 L/ 6 7 8 9 10
tm

fig.80. Lever arm ratios predicted 12y present method. (p. = 0.2)
0.6

0.5 ----e •
40
x x
m
A A A

0 r= 10%
0.4 20%
o---o 30%
40%

0 1 2 3 4 Litm6 7 8 9 10

f id .81. Lever arm ratios predicted by present method. (µ =0.3)


203

CHAPTER 9

Discussion

The theoretical method described in the previous


chapter is seen to predict load and torque functions which,
when plotted against the parameter L/(t t )2 and when
1 2
the coefficent of friction is low, exhibit a significant
degree of curvature. The general level of the predictions
is also considerably lower than that of any of the theories
using the assumption of full 'sticking' friction, and also
lower than that given by the method of El Kalay and Sparling,
employing mixed frictional conditions. The proposed method
1.
suggests, also that the parameter L/(t1t2)2 is not completely
adequate to describe the geometry of rolling, contrary to
what is suggested by slip line field theory, although the
degree of reduction dependency is not very great. Since
the experimental scatter is of about the same order of
magnitude as the predicted reduction dependency, it is
possible to neglect the latter in the first instance; in
any case, the effect of reduction was only investigated at
10000C.
Comparison of the experimental results with the
theoretical load and torque predictions indicates that
reasonably good agreement may be obtained when suitable
coefficients of friction are chosen. The apparent values
of J, determined in the present work are therefore

Temperature
, AAG
900°C 0.3 0.2-0.3
1000°C 0.2-0.3 0.2-0.3
1100°C 004 0.3-0.4

where frkp and /1,,A,G correspond to the coefficients of


friction for load and torque agreement, respectively.
Naturally, for complete compatibility the values of
ittiL and AA G should be identical, but, as can be seen,
p

the difference is not very great and use of the /gyp value,
which is possibly more representative owing to the greater
204

degree of reproducibility in the load results 1.0. less


scatter, would not introduce very large errors.

The values stated above give reasonably good


agreement over the whole range of rolling geometries
investigated the load results at 900°C fit the predictions
for AA= 0.3 almost exactly. Values of the load and torque
functions for 1,A.= 0.4 were not calculated since it was
considered that these would fall very near to the predictions
obtained by the full 'sticking' friction theories. This
conclusion. was reached by noting the trend of the functions
with increasing frt. and by considering the face that .1k has
very little effect on the predictions of the slip line field
theory in the range AA, = 0.389 - 0.72. Thus, the value
of 0.4 was assigned to the results at 1100°C, since those
results are very close to the predicted s.l.f. line.
The value 0.2 - 0.3 suggested for the results at
1000°C represents only a rough approximation since some
of them appear to justify values ofiv>0.3 whereas others
suggest p‹'0.2.
However, the maximum angle of bite used in the
geometries investigated was ,-ram- 1 0.2 so values of A less
,
than this seem unlikely. However, they cannot be excluded
since the physical action of 'biting may not be representative
of frictional conditions inside the arc of contact, so that
it is possible to envisage conditions in which the overall
coefficient of friction is less than that corresponding
to the angle of bite.
A check on the reduction dependency may be effected
by considering the experimental load and torque functions
obtained in Series B at 1000°C. (Figs. 82-89)
This shows that the results obtained for r = 10
arc by far the worst, being much lower than the corresponding
theoretical values of the load and torque functions. This
may be due to the effect of the roll quenching correction
or alternatively may be an effect related to the assumption
of a constant strain rate in the arc of contact. Both of
these points will be discussed in due course. The results
( µ=0.3)- , •••••••...,

0
2 0 C I ( µ=0.2 ) et
a

P71
L
1

I ; I t I I I I
0
10 . 20 30 40 50 60 70 80 90 100
R
t2

ficj.E32. Comparison of theoretical and experimental load functions. (Series B, 1000°C.r.710°/.2)


x
( 11=0.3),
X X

x x X
(P•=02)
Xx

kPL

10 20 30 Lig 50 60 70 80 90 100 110



/2

fig.83. Comparison of theoretical and experimental load functions (Series B, 1000°C, r:20010)
8
0

00
00
00
ca)

10 20 30 40 50 60 70 80 90 100 120 140

L2

fig.84. Comparison of theoretical and experimental load functions. (Series B, 1000° C, r=30°10)
._.

kL

PL

J I I I I I I I I I I I I I I
20 30 ' 40 50 60 70 80 90 100 150 200
Fk2

fiq.85. Comparison of theoretical and experimental load functions (Series B, 1000°C, r:-.40010)
( P•=0.3)- ,
0

e
e • 0
0
0
0
%
N(µ=0.2) 0
0


1 I i I I I I I I

10 20 30 40 50 60 70 80 90 100
R
/t2

fig.86. Comparison of theoretical and experimental torque functions(Series B, 1000° C, r=10°10)


]
I 1 1 1 t I I f I I

( IA= 0.3).,,,,
x x
G <1_2 3 x
(11=0.2)
_=1 X
G./1 2 2
G
kL

0 I I I I I i I III'
10 20 . 30 40 50 60 70 80 90 100 110
R
"1 2

fig.87. Comparison of theoretical and experimental torque functions (Series B, 1000 C, r:20%)
I I I I 1 I I 1 I 1 I 1 1 -'T

1,

0
0

0
0
00 ([1=0.2)
---

I I I I I I I I I i I It,
0
10 20 30 40 50 60 70 80 90 100 120 140
R
4.

fici.88. Comparison of theoretical and experimental torque functions (Series B, 1000° C, rz.30%)
5

0
20 30 40 50 60 70 80 90 100 120 150 200
R
/2

fig.89. Comparison of theoretical and experimental torque functions (Series B, 1000 C, r.:40%)
213

for r = 20, 30 and 40°: are in fair agreement with the


predicted values for the corresponding vL-luo of lA
i.e. 0.2 - 0.3.

It is, perhaps, unfortunate that the coefficient


of friction should again enter into the theoretical method
as an adjustable parameter to give agreement with experimental
results. But, it is considered that none of the available
methods of measuring that parameter is particularly suited
to the problem of hot rolling, and the apparent values
obtained in the present work arc, generally, not unrealistic.

The need for correction to the measured torque


values in the present work has been discussed from a
theoretical standpoint in chapter 6, and most of the
experimentally determined roll pressure distributions
have indicated that a significant portion of the pressure
distribution may lie beyond the line of the roll centres
(12)(46)(47)(53) . It is surprising, therefore, that the
need for this correction has not been realised by other
experimenters, working on similar problems. However, it
is clear that the effect is most important when the aspect
ratio of the deformation zone is large. Now, since most
workers are content to consider reduction values in the
range 10 --- 40%, the range of L/(tit2)', generally investigated
is determined, largely by the range of R/t2 ratios examined.
Generally, in the past, investigators working on
experimental mills with relatively large rolls, i.e.
10 in diameter, have worked with relatively thick
specimens(60) , while those working with smaller rolls(A4)
have been forced to use thinner stock. This has restricted
greatly the range of R/t2 ratio investigated and, therefore,
/(t1t2)2. On commercial mills , in those cases where
R/t2 is large, i.e. the last few stands of the hot-.
finishing train, the reduction is often quite small, (10 -
20%), and, again, the effect may not be apparent. However,
the work of Sims and Wright(43), carried out on industrial
mills, does show lever arms as low as 0.25 for R/t2 values
of about 100, which indicates that the effect may be
relatively important in those circumstances.
214

The need for torque correction is greater at


lower temperatures since the rolling load, the degree
of roll flattening and, thus, the amount of the roll
pressure distribution which produces the negative torque
contribution, are all correspondingly increased. This
point is illustrated by the results of and Alexander(44)
which show an apnarent temperature effect on the lever arm
ratio in that the lower the temperature, the lower the
lever arm for similar geometries, whereas the lever arm
ratios predicted by the new theoretical method are
relatively independent of itx and, therefore, of temperature.
Figs. 80 and 81 show the predicted lever arm decreasing
from the apparent maximum of 0.5, at L/(t t )2 = 0, to
1 2
about 0.4G, for L/(tt2))= 9, for both ikk= 0.2 and. 0.3,
and the present corrected results show negligible difference
in the results obtained at 900 and 1100°C for similar rolling
geometries.
In the same way that the minimum value of the
lever arm corresponding to a realistic roll pressure
distribution, has been reasoned to be 0.333, so, presumably,
there exists a maximum within the bounds of credibility.
This can he taken as 0.66, which corresonds to a right
triangular pressure di s tribution with the neutral plane
and the entry plan_ coincident. However, with the exception
of the slip line fiolS theory and the theory of °rowan
and L'ascoe, no theory predicts a value for m erreter than
0.5, which corresponds to a pressure distribution completely
symmetrical about the neutral plane at 0.5 L. (Using
moment technieue for torque prediction.) The excetions
exist because, firstlY, some of the slip line field
solutions predict a region of high pressure near the entry
plane (this occurs when the rigid zone extends over the
whole L) and, secondly, the positions of the neutral plane
suggested by Crowan (7,n(: Pascoe arc, in the main, greater
than 0.5. In comon with other published experimental
(13
work the present results show measured lever
arm ratios that are greater than0.5, these results being
characteristic of rolling geometries with. a low L/(t t )2
1 2
ratio, such as may be obtained when r is small, say 10 — 20
215

Helmi and Alexander obtained similar results and suggested


because of this, that their roll pressure distributions
wore different from those predicted by all rolling theories.
While this not necessarily true, since both the slip line
field theory and the theory of Orowan and Pascoe can predict
values of m >0.5, neither of these theories is particularly
helpful in explaining why m should be so high. Fig. 90
shows the predictions for the slip lino field theory. As
the rolling solutions have only been obtained for
L/tm 2.35 at which point m = 0.51, it is necessary to
extrapolate into the region L/t <2.35 to predict lever
arm ratios much in excess of 0.5. This cannot be carried
out with much confidence; the theory of Crowan and Pascoe
will only predict high values of m, by assuming correspondingly
high values for the position of the neutral plane, for
which there is no experimental evidence. It would seem,
therefore, that the explanation of lever arm ratios greater
than 0.5 resides elsewhere.

The most important point appears to be that this


phenomenon manifests itself in the geometries of rolling
in which the reduction is small i.e. r<20`;.... Consideration
of the variation of the strain rate through the arc of
contact (fig. 15) shows that for r<25., the maximum
value occurs at the entry plane, whereas for r>25%,
the strain rate passes through a maximum value within the
roll gap. In fact, differentiation of the relevant
equation of Ford and Alexander (24)
shows that "max. occurs
at a position

r o L from the exit plane.


3r
When r is small, the strain rate is correspondingly small
and this produces conditions most conducive to the exact
balancing of the rates of work hardening and thermal softening,
thus it is possible that, for small values of r on an
experimental mill, i.e. relatively low rolling speeds,
the yield stress distribution in the roll gap is greatly
dependent on the strain rate variation. This suggests that
0.6

• / Limit of the


• rolling solutions

A
5'

0.5

0 Crane and Alexander


0.4
A Ford and Alexander

0 1 2 3 4 5 L/ 6 7 8 9 10
m

fig.90. Lever arm ratios predicted by slip line field theory.


217

there may be certain instances, when the yield stress


distribution is weighted towards the entry plane, in
much the same way as is the strain rate distribution.
This means that k › However , the data of Cook and
G ?P- '
NcCrum do not show this. This indicates that the assumption
of a constant strain rate equal to the mean strain rate
is not always justifiable. If the real kG is higher than
that given by Cook and McCrum then the measured lever
arm ratio, mo , corresponding to the experimental dimensionless
s/k distribution, will appear to be greater than 0.5.

Another peculiarity, which can be seen from the


overall picture given by the present results, is that
the lever arm ratio appears to pass through a minimum in
the region of L/(t t
1 2 = 4 - 5. This situation could
be caused by an over--correction to the torque in the region
L
/(t1t2)2 > 5, which, in turn, may be the result of inadequacies
in Hitchcock's method when the load is large and the arc
of contact no longer remains circular. It seems more likely,
however, that the increase in mo in this range is related
to genuinely high values of G"' or G, caused by the increased
friction within the rollneck bearings which undoubtedly
is related to the total mill load, which is highest in
this range.
It cannot be over stressed that the temperature
correction calculated in chapter 7 represents a severe
simplification of the actual situation. However the real
situation is so complex that a first approximation, such
as that used, is inevitable. None of the three initial
assumptions is strictly accurate.
Clearly, some heat conduction will take place
perpendicular to the direction of rolling, along the roll
barrel. But this is an edge effect and will he of negligible
importance when the strip is of sufficient width; it is
not thought necessary to take it into account in the
present work.
Equally clearly, the thermal properties of the
roll and the strip could only be assumed identical if
218

their chemical and metallurgical compositions were so.


In the present case, since both rolls and strip were
ferrous materials it was considered reasonably accurate
to assume identical therc:ial properties as a first -
approximation. In any case, the effect of temperature
on the thermal diffusivity was not considered; an average
value was used, which per se introduces inaccuracies.
However, the greatest approximation exists in
the assumption of an infinite heat transfer coefficient.
This leads, generally, to an over-correction for the
temperature losses. That is, the correction to the
yield stress, will be generally too large. To some extent,
this over-correction will be offset by the radiation heat
losses which have not been taken into account. The fact
that some of the corrected load and torque functions are
apparently rather low? e.g. 10% reduction results at
1000°Ci may perhaps be explained by this sort of consideration.
If h, the heat transfer coefficient, is finite, it may
be expected to vary with the intimacy of contact at the
roll/stock interface, which, in turn, is related to the
ratio of the specific pressure to the shear yield stress
of the material, i.e. s/k. Thus, it may be written

h = f(s/k)

or the average value in the roll gap

hav = f((s/k)av)

But, (s/k) av is equal to the load function, P/k P

hav f(P/kL)

Thus, results corresponding to a low value of


P/kpL, by the nature of their geometry of rolling, might
be expected to have correspondingly low heat transfer
coefficients, and would, therefore, be correspondingly
over corrected by the assumption of an infinite heat
219

transfer coefficient. The situation is; then, exacerbated


for the lower results, viz„ those results at 1000°C for
r = 10%. On th other hand, high load functions relate
to high heat transfer coefficients, which might be expected
to give rise -Lc heat losses similar to those calculated by
assuming h = 00 , so the higher is the value of P/kL,
the less severe is the degree of over-correction.
However, the temperature correction cannot be
very inaccurate since, by its use, even the highest results
are brought down to about the same level as the theoretical
maximum, i.e. that corresponding to full sticking friction.

The need for this correction may be auestioned


since no other worker has found it necessary, although
the work of Smith et al(46) forced them to conclude that
the strip cools down and so becomes harder as it moves
forward in contact with the roll surface. However,
observations during the rolling tests indicated clearly
that heat losses were occuring and that in certain cases,
viz. in the case of thin strip, the roll quenching effect
was severe. In addition to this, most other work of this
nature, has been carried out either at high rolling speeds
or with material of sufficient thickness for the effect
to be negligible; generally, on a laboratory mill it seems
to happen that the quenching effect almost exactly counter
balances the incorrect assumption of full sticking friction.
(44) did not feel the need to correct
Helmi and Alexander
for the temperature drop within the roll gap, in spite
of the.fact that their experiments were carried out on
specimen sizes down to 0.07 inches. However, plotting
their results in a similar manner, to that used for the
present analysis of results, as in fig. 91, which corresponds
to 1000°C, shows a deviation from the theoretical values
3-,
over most of the /(t1t2)2 range, which suggests that .
some form of correction on a similar line to that used
in the present work, would have been applicable.
It may be noted that no attempt has been made
to incorporate the effect of the temperature rise due
to the energy of working, but it was reasoned that this
220

vs 1,12
kL
es

cv

1 2 3 4 5
tm

fici.91. Helm & Alexander's experimental load

functions. (1000 °C)


221

effect was accounted for by Cook dn the production of


his yield stress data. On the other hand, by the use
of heated platens, his measurements did not take into
account the appropriate quenching effect.
The only other doubt which may be expressed with
respect to the correction used, is the assumed effect of
the temperature drop on the shear yield stress, that is,
that dk /d0 is constant over the range of temperatures
in question. The quenching effect undoubtedly was large
enough to ensure that parts of the material within the
deformation zone ware, in certain cases, being worked
in the two phase (o1(1- ) region and evidence exists
which suggests that the gradient, dkp/dG, may be positive
in parts of this region, reverting to a negative value
in the single (-A) phase region below the the Act, thus
producing a curve of the form shown in fig. 92 over the
whole temperature range.
This effect will predominate at 900°C and 1000° C
since quenching into the two phase region (overall average
temperature drop effect) is not so likely from 1100°C. -
Thus, the assumption of a constant value of dkp/dG will
7
lead to an over estimation of kP which will produce
correspondingly low values of the load and torque functions
at 900 and 1000°C, and will require the assumption of
correspondingly low values of ,u. for agreement to be attained
with the theoretical method.
The coefficients of friction suggested at the
beginning of this chapter do show certain abnormalities,
which may be related to any or all of the aspects of the
temperature correction discussed. The most obvious
peculiarity is that, contrary to previous suggestions
AA- appears neither to increase nor decrease with temperature,
in that an apparent minimum value exists at 1000°C (nominal).
If this effect is real, then it is necessary to consider
the possible explanations.
Firstly, as suggested above, it may be related to
the temperature correction and especially concerned with
222

Ac1 Ac 3

Yield

Stress

0(

I I 1 i r

500 600 700 800 900 1000 1100


Temperature °C

fiq.92. Effect of working in two phase region.


223

the linearity of relationship between the yield stress


and temperature. Secondly, it may be speculated that
the effect was caused by the changeover to furnace B,
Which, it will be remembered, was carried out after the

experiments at 1000°C and prior to those of 900 and 1100°C.


However, those results obtained in Series C, indicate that
any difference between the two furnaces is only slight
and cannot account for the magnitude of the overall variation
in IA,.

Alternatively, the effect may be related to the


physical properties of the oxide scale, present, either
in the furnace or in the roll gap. In spite of the steps
taken to minimize the oxidation of the specimens, the
presence of some oxide scale, although generally very slight,
could not be avoided. It is generally considered that,
at the temperatures encountered in the hot rolling process,
the oxide has lubricating properties and its presence
would, therefore, decrease the coefficient of friction.
Since its fluidity will increase with temperature, iut,
cculd, ::erhaps, expected to decrease with an increase'
in temperature, as has been shown experimentally(8)(56)
Thus, the values, assigned in the present work, would
suggest that the result in need of explanation, is that
obtained at 1100° C. It may be, that, at that temperature,
the fluidity of the oxide in the furnace was sufficient
to cause it to adhere to the stainless steel cover plate,.
thus, removal of the latter, immediately prior to rolling,
ensured that the surface of the specimen was relatively
scale free, compared with the conditions at 900 and 1000°C.
It is noteworthy that some difficulty was experienced
with the removal of the cover plate at 1100° C and this
necessitated the redesigning of the cover plate so that
a wedge could be forced between it and the specimen, in
those cases where the inertia of the latter was insufficient
for the separation to be effected by tapping the cover
plate in the horizontal direction.

The temperature of the roll/stock interface in


the arc of contact may be found from equation 92 in
224

section 7.1. For an infinite heat transfer coefficient,


substituting x= 1, gives

0 s ery, 21 t t
2 s erf( "1 2)''
iG
2 FE\ 2 ,F7;7A

For most of the present work the term


erf(t1t2)2 approximates to 1, so
2

0.
interface S

Now 0s = 0
rolling (nominal)
-
0rolls since
the roll temperature is taken as the zero reference,
so refering to 0°C


0interface= , °rolling (nominal) - Orolls) + trolls

Thus
interface 0.5(9rolling (nominal) + rolls)

Assuming°Lolls is equal to room temperature


i.e. about 15°C, (the time interval between each
experiment was long enough to ensure the overall roll
temperature did not rise much above this), this means
that the interface temperature was apriroximately equal
to 458°C, 508°C and 558° C for nominal rolling temperatures
of 900, 1000 and 1100°C respectively. The iron-oxygen
equilibrium diagram(72) indicates that the oxide in equilibrium
with iron changes at 560°C from wUstite ("560°C) to
magnetite (.4.1:560°C). Thus, it is quite likely that rolling
at 900 and 1000°C was carried out in the presence of magnetite
in the roll gap, while at 11000c it is possible that the
oxide at the roll/stock interface was wilstite, and the
coefficient of friction varied accordingly.
225

Alternatively, since the exJeriments were carried


out, chronologically, in the order •- 1000 oC, 900°C, alooC,
the coefficients of friction may be related to a successive
roughening of the roll surfaces, in spite of the precautions
taken to prevent this.

In the same way that it was suggested above that


the heat transfer coefficient may be related to the
intimacy of contact between the stock and the roll at the
interface, a similar arguement may be proposed for ikk.

That is, that the overall coefficient of


friction is related to the intimacy or degree of contact
between the two surfaces.

Hence f (s/k )

and the average value

f(P/kL)
1°L'avrage

This complicates the situation since it is also


known that

P/kL = f(1.1„)

but it does suggest that those results in which the value


of /kFL is high, for any reason, will correspond to
conditions conducive to high values for la.

In the final analysis, the assignment of those


values of coefficient of friction to the experimental
results depends also on the accuracy and applicability
of the present method of incorporating mixed friction
conditions and a variable inhomogeneity factor in a
theoretical model. Since the method is closely similar
to those of Bland and Ford and Sims in its solution of
the differential equations of the friction hill, it may
be expected to posses similar predictive powers and regions
of applicability. The Sims theory is suspected of being
226

inaccurate when the R/t2 ratio is low. This is most likely


to be the effect of the small angle approximations since
for a given reduction, the total angle of contact is
inversely proportional to the square root of R/t2. Also,
when R/t2 is small, the value of L/(t/t2)2 is also likely
to be small, possibly less than 1.0, in which region, it
is probable that the load and torque functions increase
again, as suggested by the theory of Green and Wallace or
Hills slip line field solutions for forging. Thus, the
present method can be considered to apply only when the
product of ratios, R/t2 and r is greater than 1.0.

Any further inadequacies in the present theoretical


method are due to the assumption relating to the yield
stress within the arc of contact. The assumption of a
constant yield stress, made by many of the theories of hot
rolling, can only approximate to the real situation. Even
supposing the effect due to the strain rate variations
in the deformation zone is relatively unimportant, the roll
quenching will have a large effect, since this causes yield
stress variations both along the arc of contact, and •
through the thickness of the stock. The present method
assumes a constant yield only in the regions.of sticking
friction; in the regions of dynamic friction, it is assumed
only that the factor,

s w d (tk)
L

is small.
However, it was still found necessary to have
recourse to a mean yield stress, as determined from the
data of Cook and McCrum, which, per se, introduces
inaccuracies. Thus, it can be seen, the suggested values
of /,{, can be little more than semi-quantitative. Similarly,
the new theoretical method can be described as predicting
roll load and torque only with reasonable accuracy, and
the incorporation of mixed friction conditions and a
227

variable inhomogeneity factor, together with a simple


temperature correction, has only partly eliminated the failings
of the theories of hot rolling based on the more recent
simplifying assumptions.

It is considered that the present and, possibly, the


remaining failings of any theoretical rolling theory are due
to the failure to take the metallurgical factors into account.
That is, it remains a metallurgical problem to determine, as
accurately as possible, the variation of the yield stress
within the arc of contact. As this is likely to be complex,
it seems probable that use must be made of the electronic
computer to carry out the calculations required to incorporate
the yield stress data, with the rolling theory, to provide
load and torque predictions, by some suitable numerical
integration technique.
The rolling theory incorporated should be as
mathematically accurate and possess as few assumptions as
possible and it seems likely that a method similar to Orowans
exact theory would be most suitable. Recently, Alexander(41)
has suggested a method whereby the basic differential
equations of the friction hill may be solved, using a computer,
by the numerical method of integration known as the fourth
order Runge-Kutta process. Mixed frictional conditions are
incorporated but no account is taken of the variable
inhomogeneity factor in the regions dynamic friction.
Although this work is basically concerned with the problem
of cold rolling, it may be readily applied to the hot rolling
problem, and Alexander examines the effect of all the
simplifying assumptions on the load and torque predictions.
With any numerical technique, the variation of the
yield stress in the arc of contact may be taken into account
by the use of a suitable expression, such as that proposed
by Swift and suggested by Alexander, i.e.
Y (1 + B-d)n
o
where Y is the uniaxial yield stress associated with the
uniaxial equivalent strain e" and B and n are constants.
But, considering the effects of strain, strain rate and
228

temperature, it would probably be simpler to relate the


yield stress to these factors by some suitable empirical
relationship, as has been suggested by Samanta(73) or
Buxton and Sutton(31) , then the yield stress at any point
in the roll gap could be obtained by consideration of the
current values of those parameters.
However, the attainment of suitable yield stress
data is not easy. There is no means of measuring the
yield stress actually in the roll gap, so measurement must
be carried out independently. There can be no reason for
questioning any of the available Cook data which no doubt
are excellent in so far as they c:;c;. However, doubt has
already been cast on the applicability of the assumption
of a constant strain rate equal to the mean value and it
seems likely that this may have to be abandoned. Although
suitable cam design would allow a cam plastometer to deform
specimens at varying strain rates, similar to those found
in the arc of contact, the problem of translating the results
to relevant rolling mill data is complex, owing to the '
differing modes of deformation. In addition to this, the
strain rate varies through the thickness of the stock and,
again, it would probably be necessary to use some form of
mean value. Further difficulties are involved since the
yield stress at any point in the deformation zone depends
not only on the current values of the parameters strain,
strain rate, and temperature, but, no doubt, on the complete
thermomechanical history of the material at that point, both
before it enters the roll gap and, more importantly, during
its passage through the gap to that point, i.ee the
thermomechanical path by which its current state has been
reached. For this reason, also, it is thought that the
available Cook data is not strictly applicable, even if it
were used for step by step considerations of the yield
stress in the arc of contact.
Even supposing it were possible to reproduce,
independently, all the complexities of the metallurgical
situation within the roll gap, and to measure the yield
stress, there remains one further difficulty which has not
229

been considered even by the rolling theoreticians. The


variation of the yield stress along the arc of contact has
been discussed fully in the literature and the present work
but the variation of the yield stress through the section
has been neglected. This variation will arise because of
the variation in the strain rate and, of greater importance,
in the temperature through any section.
Combining the temperature profile and fig. 92, which
relates the yield stress to the temperature, produces a
yield stress distribution as shown in fig. 930 If this
represents the real situation then doubts may be cast on
the applicability of Orowan's work to the case of hot rolling,
since the inhomogeneity factor, w, is derived from the work
of Prandtl and Nadai, which assumed the yield stress was
constant through the section. It can be seen that if an
average value is taken through the section, pecularities
will arise, since while the shear stress at the interface
may be greater than this average value, i.e. sticking
friction case, the actual yield stress is likely to be larger
than the shear stress and the problem should be analysed
as a case of dynamic friction. In addition, if the theoretical
shear stress distribution is assumed, i.e. a maximum at the
roll/stock interface falling linearly to zero at the strip
centre line, then in some cases when slipping occurs at
the surface, there are regions in the material where T>k
and internal shearing must take place, as shown in fig. 93;
this will give rise to completely different modes of
deformation from any of those considered hitherto.
To sum up, although the present work indicates
that it is necessary to consider the effect of the temperature
in the roll gap, and to incorporate mixed friction conditions
and a variable inhomogeneity factor in the region of dynamic
friction in any theoretical method, it is likely that the
problem of hot rolling will not be completely solved and
load and torque will not be accurately predicted, even on
a laboratory mill, without considerable research being
carried out on the metallurgical factors affecting the
yield stress in the roll gap. In industrial conditions,

230

Theoretical T
distributions

Yield

Stress

(

roll/stock stock
interface
(t.

fig.93. Possible yield stress distibution through stock

in the roll gap during the hot rolling of

steel.
231.

where the various parameters are relatively unknown, the


prediction of load and torque is, therefore, even more
difficult and it is, perhaps, surprising that the theoretical
predictions of the present available methods are close enough
to be of any use. Indeed it may be more helpful to use
empirical relationships such as,

2
L
= 1.86 0.1 0.06 L
kPL (t1t )2 (t t )
2 1 2

which fits the results of 1000°C shown in fig. 54 with a


correlation coefficient of 0.986.

9.1 Future Work


Consideration of the present work would indicate
that future research should be directed towards the
following topics.
1) The Aquisition of Relevant Yield Stress Data.
This topic, undoubtedly, involves may problems, as
discussed earlier, but it may be that it is best approached
by carrying out further work on the cam plastometer. However,
it would seem that the use of specifically designed cams
to give varying strain rates would be of greater use than
those designed to give a constant strain rate. Herein, lies
a secondary problem, since the exact strain rate distribution
in the arc of contact relates to the position of the neutral
plane and this can be ascertained only by extensive research
into the specific roll pressure distribution.
2) The Temperature Distribution in the Deforming
Steel.
The use of any yield stress data in the prediction
of roll load and torque requires the knowledge of the
temperature distribution in the deforming stock. While the
use of theoretical and strict analytical methods including
finite heat transfer coefficients and variable thermal
diffusivities may eventually be possible via complex
computer techniques, it is essential that experimental
232

evidence, on which these may be based, be obtained as soon


as possible, This entails the measurement of temperature
distributions within the deforming material. Preliminary
tests by the present author indicate that this may be readily
effected by the use of metal-sheathed, mineral-insulated
thermocouples linked to some high speed continuous recording
device, such as a u.v. recorder.
3) Rolling Theory,
When suitable yield stress data are available, it
will be necessary to incorporate these in some theoretical
method for predicting rolling load and torque. Although
the present work has indicated the lines on which this
must be carried out, i.e. mixed friction conditions and
variable inhomogeneity factors, it is essential that some
effective means be devised whereby the variation of the yield
stress through the stock thickness may be taken into account.
4) Finally, owing to the complexity of the
situation, it has been suggested that industrial rolling
mill operators may prefer to use some predetermined empirical
relationship for predictions. If this is so, then it is
required that these relationships be determined as soon as
possible.

233

CHAPTER 10
Conclusions

1) Slip line field solutions, for hot rolling,


of a similar field type possess similar values of the
load and torque functions, P/kL and G/kL2 and similar
values 02 the deformation zone aspect ratio, L/(t1t2)2.
The linear equations,

P = 1.31 + 0.53 L
- -7X
kL (t1t2)2

G = 1.59 + 0.437 L
kL2 (t1t2)2

relate these parameters.


Other theories may similarly be expressed in
terms of. L/(t1t2)2 but, generally, predict that this
term alone is not completely adequate to describe the
geometry of rolling, and suggest that the reduction,
r, must also be taken into account, giving rise to
varying degrees of reduction dependency.
2) The measured torque should be corrected
to take into account that part of the pressure distribution
which, as a direct consequence of roll flattening, lies
beyond the line of roll centres and produces a negative
contribution to the torque.

Gmeasured + PL(R - R)
R

3) During hot rolling, it is necessary to


consider the temperature drop in the roll gap caused
by the quenching effect of the rolls and the corresponding
increase in the yield stress of the deforming material.
234

The average overall temperature drop, assuming


an infinite heat transfer coefficient and strip and rolls
of identical thermal properties, may be expressed.

1 1.
1.1284.G ./ vt 2 (0.67 - 0.17j0I Z. 2
Gcrit
tit2 5\ (;:;-

where et is the thermal diffusivity.


For the low carbon steel considered in the data
of Cook and McCrum, the change in the shear yield stress
may be obtained from the expression:

k = k a- 0.0152. 0.28
40crit.r

4) The results, so corrected, may be explained


by the development of a theoretical method incorporating
mixed friction conditions, i.e, two regions of dynamic
friction (ps<k) at the ends of the arc of contact,
separated by a zone of 'sticking' friction (p.s3;:k)
near the neutral plane, and a variable inhomogeneity
factor, w, in the regions of Coulomb friction; the value
of which may be determined from the expression:

2
w = 2(1 -- 0.38315 a - )2

where
a = /AA s
k

5) For suitable agreement with the present


experimental results, the apparent coefficient of friction
in hot rolling is in the range 0.2 - 0.4.
235

REFERENCES

1. E. Siebel: Berichte des Walzwerksauschusses,


Verein Deutscher Eisenhuttenleute, 1924, No. 37,
2. T. von Karrnan: Z. Angewandte Math. Mech., 1925,
5, 139.
3. W. Trinks: Blast Furn. Steel Plant, 1937, 25, 617,
4. A. Tselikov: Metallurg, 1939, (6), 61.
5. SKF Ball Bearing Co. Ltd: Rolling Mill Data for
Calculation and Design. Booklet, 1936,
6. V.N. Rudbakh and V.P. Severdenko: ONTI, Pi-L, 1936.
(N.B. From L.R. Underwood: The Rolling of
Metals. Vol. 1. Chapman and Hall, 1950).
7. A. Nadai: A.S.M.E. Journal of Applied Physics,
1939, June, A54-A62.
8. S. Ekelund: Steel, 1933, 93, (8), 27, and following
issues. Translated from Jernkontorets Annaler,
1927, 111, 39,
9. A.H. El Waziri: Iron Steel Eng., 1963, 40, Oct.,
73.
10. A. Pomp and W. Lueg: Mitt. Kaiser-Wilhelm-Institut
Eisenforschung, 1935, 17, 63,
11. E. Orowan: Proc. I.M.E. 1943, 150, (4), 140,
12. E. Siebel and W. Lueg: Mitt, Kaiser-Wilhelm-Institut
Eisenforschung, 1933, 15, 1
13, L. Prandtl: Z. Angewandte Math, Mech., 1923, 3, 401.
14. A. Nadai: 'Plasticity', 1931. McGraw-Hill, New York
and London.
15. E. Orowan and K.J. Pascoe: Iron & Steel Institute,
Special Report 34, 1946, 124,
16, R.B. Sims: Proc. I.M.E. 1954, 168, 191.
17. J.I% Alexandcr Fx_Ci H. Lord: I.rogress in Applied
Mechanics, Prager Anniversary volume, 1963, 191.
18. Z. Wusatowski: 'Fundamentals of Rolling', 1969.
Pergamon Press, Oxford.
19. H. Hencky: Z. Angewandte Math, Mech., 1923, 3, 241.
20. H. Geiringer: Proc. 3rd Intern, Cong. Applied
Mechanics, Stockholm. 1930, 2, 185.
236

21. W. Prager: Trans. Roy. Inst. Techn. Stockholm,


1953, (65).
22. Jori. Alexander: Proc. I.I .E., 1955, 169, 1021.
23. R. Hill: 'Plasticity', 1950. OUP.
24. B. Ford and J.E. Alexander: J. Inst. Met., 1963-64,
92, 397.
25. F.L.A. Crane and J,Ii0 Alexander: J. Inst. Yet.
1968, 96, 289.
26. F.A.A. Crane: Ph.D. Thesis. University of London,
1967,
27. J.W. Green and J.F. Wallace; J. Mech. Eng. Sci.,
1962, 4, (2), 136.
29 D. Westwood and J.F. Wallace: J. Mech. Eng. Sci.,
1960, 2, (3), 178.
29. J,H. Hitchcock: A.S.M.E. 'Roll Neck Bearings',
Research Publication, 1935.
30, S. Gupta and H. Ford: J.I.S.I., 1967, 205, 186.
31. S.A.E. Buxton and R.W. Sutton: GEC-AEI J. Sci. Techn.,
1969, 36, (1), 19.
32. H. Ford: Proc. I.M.E., 1048, 159, 115,
330 G. Wallguist: J.T.S.I., 1954, 177, 142.
0. Einicke and K.H. Lucas: Z. Metallk., 1942, 34,
25. Translation in Sheet Metal Ind., 1943,6 17.,
611
35 P.. Cook: 'Properties of Materials at High Rates
of Strain' I.M.B. Conf. (London) on High Rates
of Strain, 1957, 86.
36. J.F. Alder and V. A. Phillips: J. Inst. Yet., 1954-55,
83, 80.
37, R.R. Arnold and R.J. Parker: J. Inst. Met., 1959-60,
pf, 255.
38. H. Ford, F. Ellis and D.R. Bland: 1951,
168 57.
39. R. Stewartson: Proc. I.M.E., 1954, 168, 20
40, J.M. Capus and M.G. Cockcroft: J. Inst. Met.,
1961-62, 90, (8), 289.
41, Jail, Alexander: To be published.
237

, Cook and A.W. flcCrum: 'The Calculation of Load


42. P.I.
and Torque in Hot Flat Rolling. BISRA Publication,
1958.
43. R.B. Sims and H. Wright: J.I.S.I., 1963, 201, 261.
44. A. Helmi and J.N. Alexander: J.I.S.I., 1969, 207,
(9), 1219.
45. F.A.A. Crane and Lampkin: J. Strain Analysis,
1969, 4, Jan., 1,
46, C.L. Smith, F.H. Scott and W. Sylwestrowicz: J.I.S.I.,
1952, April, 347.
47. C.W. MacGregor and R.B. Palma: ;—.S.N.E. Journal
Basic Eng., 1959, 81, 669.
48. W. Tafel: Stahl Eisen, 1924, 44, 20 March.
49. E. Siebel and E. Fangmeier: Mitt. Kaiser-Wilhelm--
Institut Eiscnforschung, 1930, 12, 225.
50. P. Pavlov: Metallurg, 1934.
51. P.W. Whitton and H. Ford: Proc. I.M,E,, 1955, 169,
123,
52, G.T. Van Rooyan and W.A. Backofen: J.I.S.I., 1957,
106 235
F.A. Khavyat and P.R. Lancaster: J. Strain Analysis,
1969, 4 (4), 245.
54. V.W. Dahl, B. Wildshutz and J. Langer: Arch. Eisenh.,
1965, 36, (9), 633.
55. E. Siebel and A. Pomp: Mitt. Kaiser-Wilhelm-institut
Eisenforschung, 1929, 11, 73.
56. A.T. Male and M.G. Cockcroft: J. Inst. Met., 1964-65,
93, (2), 38.
57. Bachtinov: see ref. 18.
58. F.A.A. Crane and J.M. Alexander: J. Inst. Met.,
1962-63, 91, 188.
59. J.M. Alexander: J. Mech. Phys. Solids, 1955, 3, 233.
60. A.K.E.H.A, El Kalay and L.G.N. Sparling: J.I.S.I.,
1968, 206, 152.
61. D.R. Bland and H. Ford: Proc. I.M,E,, 1948, 159, 144,
62. A. Geleji: 'Calculation of Forces and Power Consumption
during Metal Working', Budapest, 1955, 2nd ed.
238

63. M.L. Zaroschchinskii: Stal' in Eng., 1970,


(2), 132.
64. O.G. Muzalevskii: Stal' in Eng., 1970, (6), 455.
65. K. Stylbo: Hutn, Listy, 1970, 25, (4), 239.
(BISITS 8601)
66. J. Jedlicka: Neue Hutte, 1970, 15, July, 408.
(BISITS 9134)
67. R. Armstrong,I, Codd, R.M. Douthwaite and N.J. Petch:
Phil. Mag., 1962, 7, (73), 45.
68. H.S. Carslaw and J.C. Jaeger: 'Conduction of Heat in
Solids', 1959, OUP. 54.
69. B.F. Bradley, W.A. Cockett and D.A. Peel: 'Mathematical
Models in Metallurgical Process Development',
Iron & Steel Institute Publication 123, 1970, 790
70. F. Fitzgerald and A.T. Sheridan: 'Mathematical Models
in Metallurgical Process Development', Iron & Steel
Institute Publication 123, 1970, 18.
71. H.W. Wagenaar: °Deformation under Hot Working
Conditions', Iron and Steel Institute Publication
108, 1968, 38.
72. L.S. Darken and R.W. Gurry: 'Physical Chemistry of
Metals', 1953, McGraw-Hill. 351.
73. S.K. Samanta: 'Deformation under Hot Working Conditions'
Iron and Steel Institute Publication 108, 1968, 122.
74. C.F. Zorowski and A.S. Weinstein: Iron Steel Eng.,
1962, June, 103.
75. G. Wallquist: Jernkontorets Annaler, 1962, 146, 873,
239

APPENDIX I

The Roll Flattening Coefficient

As indicated elsewhere (section 5.1) the rolls


used for the present work were indefinite chill cast
iron rolls, the radial hardness profile of which exhibited
the following characteristics


Surface or Barrel Hardness 61/62° Shore C

Core Hardness 45° Shore C

Hardness Gradient 10° Shore C/inch.

These are shown in fig. Al.


To obtain a flattening constant, the hardness
profile was assumed to undergo a stepwise change from
61/62° Shore C to 45° Shore C at that point on the radius
exactly halfway between the surface and the maximum core
radius, (see fig. Al) and the problem was analysed as
that of a double poured work roll according to the method
of Zorowski and Weinstein (74)
The roll flattening constant, c, a function of
V Poisson's ratio and E, the elastic modulus of the
roll material has been given (12)(15)as,

1) 1.67 x 10-4 in2/ton for steel rolls,


2) 1.91 x 10-4 in2/ton for chill cast iron rolls
and 3) 3.5 x 10-4 in 2/ton for cast iron rolls.

2) and 3) correspond directly to white and grey


cast iron, respectively, so the present rolls will have
some intermediate value.
Using the following,

a) VI core surface = 003


b) Ecore/Esurface = 0.58 (i.e. the ratio
of 2) to 3).)
240

c) core radius = 4.175 in (see fig. Al) and


d) contact angle = 5o (an average value, the
maximum used in the
present work was
approximately 11°),
. 2
a value of 2.81 x 10 In /ton was obtained and all
results were calculated using this.
241

Surface Hardness

60

Assumed
Profile

Hardness
0
Shore C

50

Core Hardness

40
0 1 2 3 4 5

Centre Surface
Roll radius (in)

fig. Al. Radial roll hardness profile.


242

APPENDIX II

Tabulated Results

The following pages give full details of all


the experimental results and the associated calculated
parameters. These are tabulated in three sets of
tables which contain respectively

a) 1) the experiment number


2) the nominal rolling temperature
3) the entry thickness - t1 (in)
4) the exit thickness - t2 (in)
5) the mill speed setting - N (r.p.m.)
6) the measured load P (tons/in width)
7) the measured torque - G (tons in/in width)
8) the measured lever arm ratio - m
9) the corrected torque - G* (tons in/in width)
10) the corrected lever arm ratio - m*

b) 1) the experiment number


2) the reduction - r
3) the effective roll radius - R (in)
4) the ratio R/t,
5) the length of the deformed arc of
contact - L (in)
-1
6) the mean strain rate ). (sec )
7) the load shear yield stress (tons/in2)
2
8) the torque shear yield stress - kG (tons/in )
9) the total time in the arc of contact
tot (msec)
10) the overall average temperature drop
A g (°C)
crit

c) 1) the experiment number and the dimensionless


groups
2) P/kpL
2
3) G/k GL
243

4) G*/k
G
5) P/k L
2
6) G*/kL

7) m

8) L/(t1t2 ) 3
9) 2L/(t 1

10) L/t2
Temp. t1
1 t2 N P G m G* ill''
Run
i/i9
1/19/3 1000 .36-66 .i$45 2200 18.96 13.330 .397 14.475 .431
2/19/3 1000 .3000 .1540 20.0 21.060 14.290 ot13 15.685 .420
3/19/3 1000 .2009 .1069 20.0 19.509. 10,280 *364 11,755 .416
1/27/3 1000 .2000 01110 2200 18.450 9.590 *3(.03 19.951 0420
1/ 1/4 1000 .2/20 .1100 2900 200150 10,790 0355 /2.310 .405
2/ 1/4 1000 .2120 .1110 2Q.0 20.380 10.800 0353 12,356 .404
3/ 1/4 1000 .3030 .1520 20.0 20.350 13,270 0361 14,565 0396
4/ 1/4 1000 03040 .1530 27106 20,239 13.440 .369 14.716 ,404
1/i /4 1000 04040 ,2802 2(1,0 10,450 7.340 .436 7.721 045;3
2/T/4 1000 .4666 02810 20(2 10.9 3 0 70710 0436 8,125 0459
1/1 4/4 1000 .4850. 0750 2200 8.080 5,440 .445 5,682 0465
2/1- 4/4 1000 .4870 .3760 2200 '7.940 5,420 .449 5,653 .469
3/18/4 1000 05000 .3775 2200 8.980 6.710 .468 6.994 ), '488
4/14/4 1000 .5009 .3775 2202 9015Q 7,1 20 .487 1.414 0507
A2 1000 .4860 .3990 2900 13.370 10,870 .423 11.393 0444
A3 1000 .4970 .3080 20.0 13.280 110690 .421 11.5q0 0460
A4 1000 0970 .7810 2200 90460 6,290 .430 6.609 .451
P5 1000 .4050 .2829 2209. _90879 6,830 042 7.171 0453
A6 1000 04050 .2822 22,0 10,200 7,160 o437 7.524 0460
A7 1000 .2990 .2350 2202 60367 3,300 .446 3,496 .472
A8 1000 .3040 .2295 2202 7.308 3,910 .427 4,149 0453
A9 1000 .3040 .2295 2000 7.230 30890 .429 4.124 0455
2.10 1000 .2190 .1920 2.08 3.470 1.190 .451 1.279 0484
A11 1000 .2190 0790 22.0 5.110 2,120 *448 2027.8 048
Al2 1000 .2190 01780 22.0 4.979 2,000 0435 20150 .468
A13 1000 .0990 09,99Q 2909 10090 0 220 065 0242 0731
A14 1000 .1010 .9929 22o0 1.890 0 410 .684 .455 0536
A15 1000, 01019 0920 20.0 1,879 098 .475 0442 .527
A7 1000 *0720 .0680 20.0 1,170 0 260 .728 .285 .799
A18 1000 .0720 .0680 2000 1.230 0 240 .637 * 268 .711

t1 t2 N G m G*
Run Temp.
1119 1000 .3020 .1880 i2:40 8.670 .416 8.633
A20 1000 .3040 .1850 P.09A 7.780 .403 8.305 .: 9:r
421 1000 03640 .186Q 8.510 .404 9,137 .434
A23 1000 .2200 .1440 g:g 10.920 5.360 .383 5.883 .420
424 1000 .2200 .1446 2.o..ii 1409,0r95 ,395 5.984 ,432
425 1000 .1010 0910 3.050 5:2 0435 .745 5018
426 1000 .1020 .0910 :969.00 2.820 ,510 .360 ,599
1127 . 1000 .1610 .6860 2n.9 48 19 .943 .386 1.177 .463
428 1000 00720 .0635 20.0 0438 .514 .472
1129 1000 00720 .0625 20,0 3.136 0587 904g: 03
A30 1000 00720 .6625 260 3,355 ,522 .327 .976 09
R2 1000 .2150 .1290 15.0 12.452 6.248 0369 9
9 .1;:f
R3 1000 02130 .1280 15.Q i2047(2 !:r3.26 .369 6: , g7
R4 1000 .0700 .0575 30.0 40450 .373 .457
1.116
5 1000 .0710 .0565 30.0 1.292 6.042 0357 1,635 :4:7 2
46 1000 .0710 .0565 35.0 0346
R7 1000 .2200 .1540 2500 ::X ,
010;" .392 41::.7. (
;, .
9 42 36
7
R8 1000 .2140 .1500 25.9 80446 3,892 .393 40234
R10 1000 .1010 .0700 25.0 8.063 2.277 0335 2.710
811 1000 .1000 .gToo 2.521 .345 3,034
812 1000 .0990 .070Q 3,121 .
...!
813 1000 '0990 .0600 .9.060 :::18169 .419
5.595
814 1000 slplo .0600 2000 J..9.ii7 :71 4,261 .322 5,287 .400
815 1000 .1010 02600 20.9 13,580 40 421 .330 5,472 .409
816 1000 .3010 .1810 10.0 11.568 7,165 7.638 .415
817 .1000 .2990 .1800 10.0 00568 7,095 :1 8
4 7,570 .413
818 1000 •2980 01790 1000 11,622 70070 .410
819 1000 03020 .2100 8.995 50137 .436
820 1000 93040 .2130 :Irll:g0 4.801 ..•M 7:,,n .426
822. 1000 .2995 .2400 25.0 20558 2.718
.405 .430

t P G m
Run Temp. t1 2 N
823 1000 •3010 0,24ia 25.0 5.696 2,774 2,936 0458
824 1000 03020 .2410 5.730 2,693 2,856 .440
825 1000 •3040 . 2730, 71.4pg 3.054 1 0 124 1,j89 .481
926 1000 •3030 "2730 3!~•Q 3.114 1'1 1 8 0478
827 1000 •3030 • 2720 3.239 1,156 1 F21
1:2 .468
R28 1000 •2220 01792 2 5.146 2,199 9446
429 1000 .2160 .1740 5.059 = 1.993 n416
930 1000 •8160 017” V00.°0 5.146 1,944 n4P6
R31 1000 •2190 01979 3n•Q 2.649 .751 2:;907) 0445
932 1000 •8200 36.0 2.811 • 783 ,848 0439
833 1000 •8210 :199:0 6 3:606 30157 ,988 .445 0480
934 1000 "3920 02340 5.0 12,646 9,299 '405 :(-11984 0426
835 1000 .3980 .2380 900 9.257 0406 9,736
9 9427
836 1000 •3980 .2380 •392 9.310 n413
937 1000 •4000 iZI. :,0714 1411 7,930 o!,61
938 1000 •3910 1500
: 7
27
7U 8.778 0424 6.048 1445
939 1000 .3920 0 19.0 1 ::7 0 :: .426 60342 •447
B40 1000 •4050 ..;71.: 25.0 3,917 4,100 0463
941 1000 •3964 .3160 5.( ) 3.558
:e.: 3,731 0457
842 '1000 •3960 03170 29.0 6.652 3.708 :::: 3.901 0456
843 1000 ►3940 03540 3.573 1.634 1,712 0523
844 1000 •4000 •3590 l''Ao, 1,600 .458
845 1000 •4040 03636: 3600 .
3''7
5:: 0:
1' 4
1 : 1,620 .485
947 1000 •1000 08o0 36.0 5,266 1.438 0467
R48 1000 •1000 eg8o5 36.0 5.350 1.271 1:V81 .421
R49 1000 •1010 .9910 36.0 2.287 .468 .530 .488
850 1000 •0990 .0890 30.0 2,236 .434 .493 0465
851 1000 00990 .0890 3(i•ci 2.532 .534 1111 0504
R52 1000 .0640 3~1►0 .270 .372 .301:
0( 0436
853 1000 :200 .0650 3600 1.904 .285 0373 .336 0439
Temp. t1 t2 N P G m G* m*
Run
R54 1000 .0730 .0660 36.0 20128 0 318 .369 .381 .442
R55 1000 .0700 .0500 30e0 8.737 20 179 .353 2,787 .452
056 1000 .0710 00500 3000 8,553 2.101 .342 2,673 .435
957 1000 .0710 .0500 3000 9,032 2,253 .345 2,888 .443
R59 1000 .0710 .0430 3.f*9 16.010 4,451 .323 6,126 .445
R60 1000 '0710 .0430 31.0 170234 4.794 .320 6.717 .449
R61 1000 .4960 .4455 30.6 3.789 1,762 .451 10840 .471
R62 1000 .4940 .4440 30.0 3.769 1,917 .495 1,996 0516
R63 1000 '4940 .4445 30.0 3.670 1.756 .472 1,833 .A92
964 1000 04950 03910 20.0 60189 4,238 .467 4,385 1463
. R65 1000 • 5020 .4626 2000 6.565 40140 .442 4.305 .460
966 1000 '5020 .4020 2000 6.022 3,940 .455 4.082 0471
$467 1000 05040 .3530 10.0 80571 6,471 .428 6,705 '443 m
R68 1000 04840 .3390 12.0 9.286 6,766 .420 7,046 ..438 4..
969 1000 .4850 .3466 16.0 8,578 6,401 . .431 6,640 .447
R70 1000 *6130 .5510 3100 3.769 20192 .514 2,262 .530
971 1000 .6190 +5570 3606 4.044 2,304 .563 2,385 .520
872 1000 96210 .5590 3.0 4.000 2,268 .500 2,347 .518
R73 1000 06200 .4960 20.0 6.756 5.012 .464 5,172 .479
R74 1000 '6170 .4940 26.0 6.289 40692 .469 40 832 .483
975 1000 .6140 .4910 2,0.0 6.600 40 994 .475 5.14 3 .490
R76 1000 v6220 .4350 10.0 10,135 80797 0442 9,091 .457
R77 1000 06130 44290 16.0 9,759 8.383 .440 8,660 .454
978 1000 '6140 .4300 1n.0 9.633 8,262 .441 8,530 .455
Cl 1000 .2170 .1520 25.6 9.276 4,190 .381 40598 .419
C2 1000 .2170 .1520 25,0 8.951 3,873 .366 4.253 .402
C3 1000 .3030 .21 20 26.0 9.56o 5,172 .392 5,537 0420
C4 1000 .3030 02120 20.0 9,565 5.124 .386 5,494 .414
C5 1000 .3990 .2790 15.0 9.657 60014 0393 6,345 .415
C6 . 1000 03990 .2790 15.0 9.249 5,832 .398 6,136 0419
Run Temp. t1
1 t2 N P G m G* m*
C7 1000 .4970 .3486 1.:)90 9,523' 6,941 .415 7,231 .432
C8 1000 .4960 .3470 1 6.
0 9,655 7,454 .439 7,752 6457
D1 1000 *0720 .0460 3(1,0 1 5.680 4,271 0326 5,927 .453
02 1000 *0715 .0455 36.0 14.677 30886 .320 5.349 .440
03 1000 00710 .0450 30.0 14.263 3,662 .311 5.048 .429
04 1000 *0700 .0450 30.0 14.494 3,895 .330 5.345 6453
05 1000 .0710 .0450 30.0 15.296 4,135 .325 50716 .449
n6 1000 0720 .0450 3(20 14.644 3,775 0367 5,211 .424
07 1000 .1020 .0600 200 15.295 4,929 0320 6.236 .405
08 1000 01010 .0555 20.0 19.564 6,994 .335 9.009 .431
n9 1000 .1010 .0530 20.0 19,707 6,857 .320 8,865 0414
G1 1000 *0710 .0420 30,0 17.713 4,658 .298 6.656 0426
82 1000 00720 00425 25..0 17.475 4,579 6295 6,514 0420 f.)
(33 1000 !0680 .0420 20.0 16.250 3,852 *283 50622 61-1 to
G4 1000 00690 .0440 15.0 17.775 4,031 .271 6.154 0414
G5 1000 .0690 .0450 1000 180763 3,958 .254 6,338 040r,
G6 1000 *0675 .0460 5
.00 19.388 3.982 .255 6,606 .423
Fl 900 .0700 .0455 20.0 18.063 4.294 .286 60497 0432
E2 900 .0695 .0450 26.0 17.275 4.090 .286 6,118 .428
E3 900 00710 .0465 20.0 1 5.260 4,468 .295 6.702 0442
F4 900 *0720 .0465 2(100 18.975 4,627 .287 7,006 0434
E5 900 .0715 .0465 25.0 18.954 4,614 0288 7,004 .438
F6 900 .1010 00636 2..0 16.088 50057 0324 60557 0420
E7 900 *1010 .0630 20.0 16.750 5,285 0324 6,904 .423
E8 900 .1010 .0640 20.0 17,163 5,215 6315 6,929 0418
E9 900 .1015 .0635 21500 16.800 5,030 .367 6.658 .407
E12 900 .1010 .0620 2000 17.150 5,369 0317 7.045 0416
Eli 900 .2150 .1275 2600 15.605 7,607 .351 8,593 0397
E12 900 .2140 .1275 2000 160368 7,956 0351 9,044 .399
E13 900 *485 ',1276 2000 15.795 7,433 0350 8,476 .399
Run Temp. t t2
1 N P G m G' m*
514 900 02170 .1280 26.6 16,555
E15 900 60277" .357 9,375 .404
e2165 .1285 2i)
:.0 16.267 7,963
E16 900 *3010 .351 9,030 .398
.1850 20,0 15.947 9.664 1419
E17 900 '3090 .1850 26.0 16.564
E18 900 9,976 .'3.1
p g 1(01:: .405
.2980 .1835 20.0 16,303 9,646 .376
E19 900 0030 .1810 10,596 .413
20.0 16.616 10,047 .373 11,004 .409
E20 900 '3090 .1820 2000 16,578
F21 . 900 10,204 '373 11.139 .407
06080 .5475 2000 5.431 2,791 .456
E22 900 46150 05500 2.938 o VI (I
2n,0 5.689 3.094 .466 3,250 .409
E23 900 '6160 .5505 26.0 5.667
F24 3.101 .467 3,255 1490
900 .6280 .5520 20.0 6.533 3,844 .466
E25 900 .6120 .5510 4,034 .489
29410 5.522 20839 .454 2,990 1478
E26 900 '5010 .4010 20,0 9.027
E27 900 5,519 0422 5,835 .446 m
!4900 .39-70 22.9 8,767 5.111 hiN
E28 900 .4990 '417 5.420 .46.2 tO
.4010 20.0 8.942 5,280 .411 5,593 .46
E29 900 .04870 .3980 20.0 8,526 5.013
E30 900 • '4840 .429 5,311 .455
.3970 25,0 8.175 4.693 .424 4.970 .449
E31 900 .3960 .2790 20,0 12.186
E32 900 70744 .404 8,274 .431
0910 .2780 201.0 11.945 7,515 .407
E33 900 .3940 8,033 .435
.2786 200 12.270 7,696 .400 8.236 0428
E34 900 03970 .2790 29..0
F.35 900 '3920 .2780
12.357 7.841 0401 8,384 .429
2n00 12,164 7,570 .400 8,105 .429
Fl lloo '6270 .5470 26,0 3.989
F2 1100 2.260 .442 2.330 '455
'6100 .5470 20,0 3.605 1,848 .450
F3 1100 '6100 .5470 1,912 .465
25.6 3,500 1,913 .479 1,973 .495
F4 1100 .6090 .5480 2f-2,0 3,496
F5 1100 1,775 .452 1.836 .468
'6100 :5470 20.0 3.666 1,906.464
F6 1100 .4980 .3945 20.0 1.970 '480
F7 5.984 3.585" .410 3,723 .426
1100 .4990 .3950 25,0 6.104 3,676
Fe 1100 .4950 .411 3.819 .427
.3940 2000 6.267 3,913 .432 4.066 .449

Temp. t1 t2 N P G m G*
Run
F9 1100 !4940 .3940 25.0 6.089 3,614 .413 3,759
rio 1100 •4890 03930 26 00 5.874 3,373 .407 3,511
Fll 1100 •3950 .2700 20.0 5,377 5.618
F12 1100 .3925 .2700 20.0 6.362 : 4
4 r3
F13 1100 •3960 02710 26.0 :313:,, 5,699 0394 :::
F14 1100 •3900 02700 5,687 .409 5,962
F15 1100 .4000 .2705 °69.00 3,7
9,252 6,117 .401 .
c.:7.
F16 1100 03150 01765 25.0 7,907 :::10:
F17 1100 • 3000 .1760 4:%;
20,0 7,351 1.
120288
27770 7,873 o397
8.284 !!
F18 1100 .3150 .1785 2d•q 7,783 .374
F19 1100 4,299O .1790 12.245 ;3 7.968 * 403
F2Q 1100 .2980 .1750 ;i1.000 12,070 7g
7 .373 7.674
F21 1100 02060 20.0
12,650 .
6,030 357 ca
n
F22 1100 .2190 .1260 2.0 00 13.016
2
6,468 0351 7:1 10-' 0
F24 1100 .2180 .1280 20.0 '063 7'067
F25 1100 .2240 41340 12.832 0 70283
F26 1100 • 0990 .0580 :(.0(30 15,450 4,W
66 :*(7) 6,104
P27 1100 * 1 000 .0590 2000 15.950 5,068 018 6,498
F28 1100 • 1000 .0600 20.0 50713 0318 .409
P29 1100 .099Q .058p 25.0 M7:I
F30 1100 .1010 2600 11::Vg 444i! .
0
P31 1100 • 0710 :Mg 20.0 16.825
1:
4,153 .285 t:M.
5,943
.411
F32 1100 .0715 .0430 4,114 .280
P33 1100 • 0715 2
29 .1(6) 16.650 4,258 .294 0417
F34 1100 • 0730 .
.9c)44;00 2000 160400 4,035 .278
16.738 4,0,4 .287
:
(:. 7
):: 3 34
F35 1100 *0680 .0425 20.0 5,898 .422

Run r R R/t2 L A kP
k tot G r .4 a •
crit
1/19/3 *4867 5.365 34.84 *8856 6.85 5,440 5.180 93.5 93.7
2/1 9/3 .4867 5.404 35.09 .8883 6.88 5,440 93.1 9306
3/19/3 04700 5.583 52.67 .7244 8.39 5,560 ::g00 72.5 121.5
1 /27/3 .4450 5.583 50.30 .7049 8.14 5,520 115.5
1/ 1/4 .4811 5.556 50.51 .7528 8.24 5,510 ::1 1:00
2/ 1/4 .4764 5.567 50,15 .7498 8.20 5,510 75.7
3/ 1/4 .4983 5,379 35.39 .9013 6,92 5,450 ::140 E
4/ 1/4 .4967 5.376 35.14 .9010 6.90 5,450 (15.7
1/10/4 .3069 5.237 18,70 .8058 4.49 40950 78.5 5302
2/1 3/4 .3079 5.246 18.67 .8098 4.49 4,950 15
1.
4:0913 00 5311
1 /1 4/4 .2268 5.206 13.88 .7568' 3.45 3,960 72.2 39.8
2/14/4 .2279 5.201 13.83 .7598 '3,45 3,960 30 c :1° 0 rs.)
3/1 4/4 .2450 5.206 13.79 *7986 3.55 4.170 3.780 76©6 43f90: cra
4/1 4/4 .2450 5.210 13,80 .7989 3.55 4,170 3.780 40.4 1-1
A2 .3642 5.212 16.87 .9605 4050 5.100 4.760 52,07
43 .3803 5.197 16.87 .9911 4,55 5,150 4.780 10.6
44 .2922 5.228 18.60 .7787 4.40 4,950 4.570 75.5 5;.:*3
45 .3037 5.225 18.53 .8017 4,45 5.000 4.600 781 52 4 8
46 .3037 5.233 18,56 *8023 4.45 5,000 4,600 52.8
47 .2140 5.280 22.47 05813 4.29 4,600 4,170 55.6
48 .2451 5.276 22.99 .6269 4.58 40 670 4,350 r:.
.
.i
49 .2451 5.273 22.97 .6268 4.58 4,670 4.350 59,3
410 .1233 5.361 27.92 .1805 3,75 3,900 ..c-5)0"8
. :5/
All .1826 5.359 29,94 .4630 4.63 4,450 4.030 34::: 65.4
412 .1826 5,349 29.88 .4626 4.62 4,450 ::::: 4204 65.4
413 *0404 5.766 60,69 .1519 3.25 2.500 2.300 )12.6
414 .0891 5.590 60,76 .2243 4.76 30 600 g:'1
415 .0891 5.584 60,69 .2242 4.75 3,600 t:90 0 igo4 89.9
417 * 0556 5.822 85.62 *1526 4.50 3,200 :::46' 99.6
418. .0556 5.864 86.24 .1532 4.52 3,200 ta 0°01)
(1 12.5 99.4
Run r R R/t2 L A kP kG ldtot aecrit
A19 .3775 5.308 28.23 .7779 5.88 5,270 4.960 77.2 7503
A20 .3914 5.287 28,58 .7932 5.97 5.350 4.950 79,6 77.1
421 .3882 5.316 28:58 .7921 5.96 5,320 4.920 78,9 76.5
423 .3455 5.404 37.53 .6408 6,61 50290 4.880 61.5 89.6
424 .3455 5,400 37.50 .6406 6.60 5.290 4.880 61.5 89.6
425 .0990 5.851 64.30 .2419 5.14 3,960 3,500 2000 91.9
426 .1078 5.720 62.86 .2508 5.29 3,850 3.550 2102 94.3
427 .1485 5.777 67.18 .2944 6.33 4.350 3.940 24.8 105.9
428 .1181 5.810 91.50 .2222 6.66 40050 3.720 15.5 12.8
429 .1319 5.928 94.84 .2373 7/q3 4.200 3.850 19.4 136.0
430 .1319 50991 95.85 .2386 7.17 4,200 3.850 19.3 12$,7
82 .4000 5.406 41.91 .6819 5.45 5,280 4.920 89.7 1.x646
83 .3991 5.412 42.28 .6783 5.47 5.290 4,920 8901 1:17,R
94 .1786 6.003 104.39 .2739 12.84 4.813 4.410 14.9 121.0
95 .2042 6.171 109.22 .2991 13.90 5.027 4.610 15,9 125.7
96 .2042 6.108 108.10 .2976 13.53 5.022 4.600 1600 126.0
87 .3000 5.370 34.87 .5953 7.60 50210 4.840 45.1 73.7
98. .2991 5.371. 35.81 .5863 7.69 5.220 4.840 44.4 75.1
810 .300 5.731 81.87 .4215 11.74 5.370 5.015 30.0 131 /7
811 .3000 5.818 83.12 .4178 11.73 50365 5.010 29.2 130.4
812 .2929 5.835 83.36 .4114 11.65 5.360 5.005 28.6 129.6
813 .3939 5.996 99.93 .4836 1 1.18 5,670 5/255 42;9 174.1
814 .4659 5.919 98.65 *4926 11.19 5.680 5/ 265 64.6 176,0
815 .4059 5.931 98.85 .4931 11.20 5,680 5.265 44.5 175,9
916 .3987 5.271 29.12 .7953 3003 40920 4.630 160.9 111.4
817 .3980 5.273 29,30 .7922 3.03 4,930 4.640 160.1 111.8
818 .3993 5.274 29.47 .7922 3005 4,930 4.640 160.2 112.4
819 .3646 5.275 25.12 .6966 5.19 5.060 4.650 6703 65.8
820 .2993 5.267 24.73 .6923 5.12 5,050 4.640 66.8 64.9
822 .1987 5.267 21.94 .5598 5.13 4.630 4.180 41.9 48,1
Run r R R/t2 L 7 1t kP kG 't.tot Le orlt .
B23 .1993 5.267 21.85 .5621 5.13 4,630 4.180 42.1 48.0
924 .2020 5.264 21.84 .9667 5.16 4,650 4.200 42.5 48.2
925 .1020 5.277 19,33 .4045 4.29 30750 3.470 24.7 34.0
826 .0990 5.292 19.38 .1984 4.23 3,710 3.440 24.2 33,8
R27 .1023 5.294 19.46 .4051 4.31 3,760 3.480 2407 34.1
928 .1937 5.336 29.81 .4790 7.10 4,760 4,280 29.4 54.2
829 .1944 5.338 30.68 .4735 7022 4,765 4,285 29.1 55,4
930 . .1991 5.336 30.85 .4790 7,31 4,760 4.320 29.5 56,0
831 .1005 5.338 27.10 .3427 5,04 3,800 3,520 20.7 4:11,2
832 .1000 5.359 27.07 .3434 5.03 3,790 3.510 20.6 42,9
R33 .1041 5.386 27.20 .3520 5,13 3,830 3.540 21.1 42.3
934 .4031 5.225 22.33 .9086 1.33 4,380 4.210 371.7 00,6
835 .4020 5.- 219 21.93 .9138 1.32 4,375 4.200 374.1 128.9
936 .4020 5.217 21.92 .9136 1.32 4,375 4.200 374.2 12eg9
937 4,3200 5.205 19.14 .8163 3,45 4,900 4.510 10792 6.7.- “,
838 .2941 ' 5.214 18.89 .7744 3.33 4,830 4.430 100.5 61.6
839 .3010 5.216 19.03 .7845 3.37 4,860 4.450 16271 62.2
840 .2123 5,216 16.35 .6698 4.56 4,630 4,190 50.8 39.6
841 .2020 5.222 16.52 .6463 4,49 4,590 4,150 48.8 39,4
842 .1995 5.237 16.52 .6432 4.46 4,580 4.130 48.4 39.1
843 .1015 5,251 14.83 .4583 3.75 3,680 3.420 28.1 280 0
844 .1025 5,258 14.65 .4643 3.74 30690 3.430 28,4 27.8
045 .1015 5,246 14.45 .4638 3.70 3,670 3.410 28.5 27.5
847 .2000 5.740 71.75 .3388 11,17 4,920 4,500 19.4 98,1
848 .1950 5.771 71.69 .3355 11,05 4,870 4.470 19.1 9700
1349 .0990 5.643 62.01 .2375 7.57 3,940 3.690 Y306 7507
850 .1010 5.628 63.24 .2372 7,72 3,970 3.680 13.6 77.4
851 .1010 5.711 64.17 .2390 7.78 3,980 3.690 13.5 7701
852 00986 5,727 89.49 .2002 9.08 4,040 3.730 11.3 98.1
953. .0972 5.764 88.68 .2009 8,98 4.010 3.700 11.2 96.5
l'
RIM r R R,/t2 L I kP G tot ,ei 0 .
crlt
854 .0959 5.854 88.70 .2024 9.92 3,980 3.680 11.1 94.7
855 .2857 6,228 124.55 .3529 16,94 5,460 5,080 19.1 "f4809
856 *2958 6.144 122.89 .3592 17094 5,480 5.120 19.8 15007
857 .2955 6.209 124.17 06/1 17,13 5,480 5.120 9.7 1500
859 03944 5,607 153.64 .4301 20.80 5,970 5,495 23.1 178.1
860 .3944 5.730 155.50 .4341 20.99 5,975 5.500 22.9 17703
961 .1028 5.209 11.71 +5154 3.35 3,658 3.406 31.9 2307
862 .1012 5.213 11.74 .5105 3.33 3,637 3,393 31.5 23.6
R63 .1002 5.208 11.72 .5078 3.31 3,622 3.3$2 31.4 23.6
R64 e2101 5.167 13.22 .7331 3.26 4,016 3.612 70.1 39,5
865 01992 5.183 12089 .7199 3.15 3,954 3.550 6804 38.2
866 .1992 5.169 12,86 .7190 '3.15 3,954 3.550 _68.5 38.2
867 .2996 5.160 14.62 .8827 1.97 4,074 3.751 174.0 65.7
968 .2996 5.180 15.28 .8667 201 4,092 3.762 170.2 67.7
869 •2990 5.166 15.19 .8655 2o00 4.092 3.758 170,4 67.5
870 .1011 5.171 9.38 .5662 2/98 3.242 3/000 15.3 21.0
871 .1002 5.183 9.31 .5669 2.95 3,233 2,984 35.2 20.7
872 .0998 5.181 9.27 .5668 2.94 3.225 2;981 35.2 20.7
873 .2000 5.153 10.39 .7994 2.83 3,920 3,523 75.5 3207
874 .1994 5.144 10.41 .7954 2.83 30917 3.521 76.2 3288
c:t75 .2003 5.151 10.49 .7960 2.85 3,923 3.526 76.2 33/0
976 .3006 5.152 11,84 .9816 1977 4,491 4.144 193.9 54.0
877 .3002 5.149 12.00 .9734 1.`7,8 4,494 4.145 1q2.3 54.6
878 .2997 5.147 11,97 .9732 1.78 4,490 4.147 192.3 54.5
Cl .2995 5.401 35.53 .5925 7.67 5,220 4.840 44.6 74,3
C2 .2995 5.387 35,44 •5917 7..66 50 220 4,640 44.7 74.3
C3 03003 5.293 24.97 .6940 5.15 50 050 4.640 6697 65.1
C4 0003 5,295 24.98 .6942 5.15 5,050 4.640 66.7 65.1
C5 0008 5.226 18.73 .7919 3.35 40 850 4.450 1n2.8 61.4
C6 .3008 5.217 18.70 .7912 3.34 40 850 4.450 162.9 61.4
L°06 Ci479 011°9 029°9 LC 6 L 220° 99°07 5475°5 606C° £13
C°I6 6°99 091°9 099'9 147°L 069° 6E0 E47 ECGIG 247047" 213
9 0 16 9 0 L9 OLT°9 0L9°9 Oviii, 8E69° G1"E47 10S°5 0/.047° 113
COST 6°I? O82'9 OLL°9 91 0 1/ 106+7° e5 0 001 9C2°9 T9ece 03
t°90 001 4? 0E2°9 S2L°9 476 0 0/ 0L80 oc.86 2472.9 47 472.C° 63
47.474,T T.07 002.9 ooe9 68.ot 6284r, 647°96 COCe9 £99C° 93
S°L471 PI+, 0 412 0 9 00..9 0961/ 698 £0 0 66 6020 9 29LE° 1.'
.
9°1.471 C.1 4, 0472°9 00..9 86 0 01 08e47. 82.86 061'9 92.E' 93
9°c4I so0E 012'9 022.'9 OV'ET 22247" EE"EST OCI°L L647C° 53
9°9LI 20 1E 522,9 OEL°9 24/cEI E5247 0 west 160 0 ). 2t.5E° 473
.9'SLI Gan 061'9 069.9 IE•Ei L91 47° 847°25I L90°L 1547C° £3
8.10 80oe 022 0 9 G2L°9 ISeCt 9CIt 0 4/PSG"( 186°9 525C° 23
I°6LI 900C 012 0 9 02L°9 05I01 291 47 0 2,°55I 22.0°L 0 0GE° T3
2°0ec 9,601 019°47 506°47 90E S2047• 9/.°E91 47E5fi. c91E° 99
9'2s2 0 0 0(7 - 026°47 GTE'S C9 0 9 9SI47° C6°681 L6I•L 9/.47E° SO
4'92 L 0 It OvI's LCS°G 17c 0 0t C8I 47 4, +70°681 866°9 C29E° 470
in
,in 0'61E 9 6 2E opE.5 seL 4 s 800,T 161 47.° 98.091 952.°9 4729E° £0
cv S°o6I 9 0 92 .9C.)7 0 51 029°S L9, LT 47C47 47• 18'981 599, 9 L6047° 20
C29I 5oC2 09S*G 810°9 6E 4,12 £14747 0 16 6 6S1 91/.0 9 590+/* 19
1°002 9,0 01/47'5 009'S 9'0°21 SC47S 0 tI°9II 47S1°9 252.47° 60
51 161 IsLi? 02's OsL e S 220 2T SEES,' t9'211 S61'9 tq547' 80
6.'') t 6").'7 062'5 004'S 2E 6 IT OCOS" 6C°001 £20 4 9 OW' LO
=0/.1 9,22 0C47 4 5 099°5 476°61 461 47 0 86°47471 4/2S*9 092.E 0 90
9 0 891 8e12 00.0 4 8 OWS tP•02 68147 0 89°4171 £89 0 9 299C° SO
9°L9I vIa oLcog 0 605 E2°61 12.0 47 6 IC°L47I 629°9 UGC° tO
C 0 691 0 0 22 06E 0 5 OWS 478 0 61 4721 47° alstI 2i75°9 299E 0 20
C°49I 6 0 I" 58E,G 029°G 9L 0 61 9C1 47° SL°47471 98G°9 9C9C° 20
0°89I 1'12 02C 05 OTe°8 44 °61 2L1 47° 47S°547I 569'9 IT9E° 10
6°C9 5*211 0920 4/ 085°47 66.1 LeLil° U0°171 29to5 4700ce 90
8°C9 5°241 0020 47 085$47 66 0 1 S9LVI eett 091"s 8662' 43
2
4TaD G i7 qcq2 0 d)E 4/E
Y a 23 a unu
Run r R R/t2 L 71 kP
k
G tOt crit r Le
E14 .4101 5.523 43.15 704.2 - 6.680 6.180 68.1 91.4
E15 .4065 5.519 42.95 .6969 7.38 6.670 6.170 67.6 9009
E16 .3854 5.386 29.12 .7904 6.00 6.500 6.010 77.6 68.6
E17 .4013 5.374 29.05 408163 6.05 6,590 6.070 81.1 69.4
E18 .3842 5.400 29.43 .7863 6.03 6,490 6.010 77.0 69.0
F19 .4026 5.383 29.74 .8104 6.13 6,600 6.060 80s4 70.6
E20 .4110 5.367 29.49 •256 6.13 6,640 6.105 82.6 70.8
E21 .0995 5.252 9.59 .5637 1.99 4.250 3.840 51.8 2211
E22 .1057 5.246 9.54 .5839 2.04 4,325 3.910 w3.8 22,4
F23 .1063 5.243 9.52 .5860 2.05 4,340 3.920 54.0 224
F24 .1210 5.242 9.50 .6312 2.17 4,500 4.080 c8.4 23,0
E25 .0997 5.254 9,54 .5661 1.99 4,250 3,840 5200 22,0
E26 .1996 5.254 13.10 .7248 3.18 5,400 4.900 6800 329
F27 .1898 5.265 13.26 .6997 3613 5,310 4.830 65.3 32,8
E28 .1964 5.256 13.11 .7177 3.16 5,360 4.880 67.2 32.3
E29 .1828 5.269 13.24 06848 3,08 5,250 4.770 63.8 32,4
F.30 .1796 5,264 13.26 .6767 3906 5,230 4.750 63.1 3204
E31 .2955 5.293 18.97 .7869 4.46 6.050 5.540 75.5 47,4
E32 .2890 5.297 19.05 .7737 4.44 6,025 54,510 74.0 47.3
E33 .2944 5.297 19.05 .7839 4.47 6,050 5.530 75.1 47.5
E34 .2972 5.294 18.98 .7904 4.47 6.065 5.560 75.9 47,5
E35 .2906 5.300 19.06 .7773 4.45 6,030 5.520 74.3 47.3
F'1 .1276 5.140 9.40 .6413 2.21 2,840 2.580 60.5 25.9
F2 .1033 5.161 9.43 .5702 2.01 2,675 2.438 5304 27.5
F3 .1033 5.156 9.43 .5699 2001 2,675 2.438 53.4. 27.5
F4 .1002 5.161 9.42 oR611 1.98 2,650 2.425 52.5 27.2
F5 .1033 5.161 9.43 .5702 2.01 2,675 2.438 53,4 27.5
F6 .2078 5.162 13.09 .7310 3.23 3,330 2.993 69.9 41.5
F7 .2084 5.165 13.08 07329 3.23 3,335 2.998 70.1 41.4
F8 .2040 5.174 13.13 07229 3.21 3,320 2.981 68.9 41.3
Run r R R/t2 L X k P kG 1:"tot (9 crit
F9 .2024 5.171 13.12 .7191 3.20 3,310 2.975 66,6 41.2
P10 .1963 5.172 13•16 •7046 3.16 3,290 2.950 67.1 41.0
Fll .3165 5.187 19.21 .8052 4,60 39 780 3.470 79.5 60.9
F12 .3121 5.202 19.27 .7982 4.58 3,775 3.463 76:4 60.6
F13 .3157 5.201 19.19 04063 4.59 3.777 3.467 79.3 60.6
F14 .3077 50206 19.2.8 .7904 4.56 3,770 3.450 7705 60.4
F15 .3237 5.201 19.23 08207 4.63 3,790 3.488 81.0 61.1
F16 .4397 5.250 29.74 .6527 6.24 4,055 3;825 88.8 90.9
F17 .4133 5.278 29,99 .8090 6:19 4,050 3.800 82.4 89.5
F18 .4333 5.253 29.43 .9468 .
6.19 4.055 3.820 67.8 8?!..1
F19 .4013 5,287 29.53 .7965 6.10 4,025 3.780 60.4 87,6
F20 .4027 5.283 29.66 .7962 '6,12 4.025 3,780 80.5 88,1
F21 .3981 5.433 43.82 .6675 7.42 4.175 3.855 65.5 114.3
F22 .4247 5.393 42.60 .7082 7.44 4,190 3.900 71.1 A15.2
F24 .4128 5.396 42.16 .6969 7.34 40 160 3.882 69.4 U2.9
F25 .4018 5.401 40.30 .6972 7.13 40 150 3.862 66.9 106,3
F26 .4141 6.059 104.46 o4984 11.56 4,515 4,180 44,3 198 08
F27 .4100 6.093 103.27 .4998 11,47 40 510 4.169 44.0 195.5
F28 .4000 6.104 101.73 .4941 11.32 4,500 4,150 43.2 01.7
F29 .4141 6.060 104.48 .4984 11.56 40 515 4.180 44.3 198,8
F30 .4158 6.056 102.65 .5043 11.47 40 515 4,182, 44.8 19605
F31 e3944 6.689 155.55 .4328 13.95 4,640 4.283 34.4 239.6
F32 .3986 6,660 154,89 .4357 13.96 40650 4.292 34.9 240.5
F33 .3986 6.642 154.46 '0,4351 13.94 4,650 4.292 34.9 24007
F34 .4110 6.536 152.00 .4428 13.92 4,660 4.318 3694 243.6
F35 .3750 6,844 161.05 .4178 14.01 40610 40256 3201 237,1
L / 1/.2-
gun P/I
cr 1—
C/ 2
L- 5ZI_ V
kr I- (ti t2)
1/19/3 3.936 3.285 3.567 3.246 2.941 .453 40117 3.899 5.747
2/19/3 4.346 3.496 3,838 3.585 3,166 ,442 4,133 3.913 5.768
3/19/3 4.841 3,745 4.283 3,821 3,381 ,442 4,975 4,735 6.834
1/27/3 4.749 3.726 4,254 3,795 3,399 ,448 4,731 4.533 6.351
1/ 1/4 4,865 3,676 4,193 3,843 3.313 .431 4.930 4,676 6,844
2/ 1/4 4,933 3.708 4,242 3,907 3.361 ,430 4,888 4,643 6,755
3/ 1/4 4,149 3.148 3.455 3.409 2.838 ,416 4,200 3,962 5.929
4/ 1/4 4.120 3.190 3,493 3,389 2.873 ,424 4.178 3,943 5.889
1/10/4 2,620 2.463 2.590 2,347 2,321 * 494 2.396 2,356 2.87R
2/10/4 2.727 2.562 2.699 2,444 2.419 ,495 2.397 2.357 2,752
1/14/4 2.696 2.631 2.748 2,452 2.560 0 510 1.775 1,760 2.018
2/1 4/4 2.639 2,601 2.713 2.400 2..467 ,514 1,776 1,761 2%021
3/1 4/4 2,697 2.784 2.901 2,456 2,643 ,538 1.835 1,820 2.115
4/14/4 2.747 2,951 3,073 2.502 2.500 '559 1.839 1.821 2,116
42 2.729 2.475 2.594 2,448 24.327 ,475 2.49 2,416 3:103
43 2,602 '2,362 2.468 2,330 2,210 ,474 2.533 2,462 3.218
44 2.439 2,270 2,385 2.191 2.143 ,489 2.332 2,297 2.771
45 2,462 2.310 2.426 2.211 2.178 ,493 2.372 2,334 2.843
46 2.543 2.418 2.541 2.283 2.252 ,500 2,374 20336 2.845
47 2,381 2.342 2,481 2.131 2.220 ,521 2.193 2.177 2.474
48 2.496 2.287 2.427 2,216 2.154 ,486 2.37 3 2.350 2.732
49 2.470 2.277 2,414 2,193 2.143 ,489 2.373 2.350 2.731
410 2,339 2.309 2,482 2,088 2,21§ '531 1.855 1.651 10982
All 2.480 2.454 2.638 2,182 2,321 '532 2.338 2.327 2.567
412 2,414 2,319 2.493 2.124 2.194 ,516 2.336 2,324. 2.564
413 2.871 4,147 40562 2,449 3,892 ,794 1.566 1,566 1.599
414 2.341 2.397 2.658 1,970 2.237 ,568 2.327 2.324 2.438
415 2.317 2,329 2,586 1,950 2,176 ,558 2,326 2.323 2.1+37
417 2.396 3.722 4,082 1.989 3,3,s19 ;852 2,181 2,180 2o244
418. 2.510 3,411 3+805 2,084 3.160 0758 2,189 2.188 2.252

it-
c/ z
/kr L C;t4 1 Rio 46- 21- (Tit:-
y
L.2.
RAvt irp 2
A19 3.044 2,722 2.912 2 . :S61
2- 0478 3.265 3,175 4,138
A20 2,868 2.498 20666 2.458 2.285 .465 3.244 4+288
A21 3.154 2.757 2.960 2,705 2.539 0469 3.233 4.258
A23 3.221 2.674 2.935 2.709 2
3;4,04
2: 9
13; 0 456 3..600 3.521 4
A24 3.193 2,731 2.988 2,685 2.513 ,468 3.599 1.520
425 3.296 3.115 3.636 2.775 0552 2.523 2,520
426 2.920 2.283 2.682 20443 ,459 2,599 i':ii!
427 3.241 2.762 3,243 2.672' 2.674 'or5: 3,148
428 2.722 2.384 2.797 2.171 2,231 3.287 3.880 l:: :ig
429 3.146 2.708 3,245 2,494 2.572 3.537 3,529
430 34,343 2.382 2.986 20652 3.547 ;27 1.
2 3.453 2,731 3.010 2,746 9.2,!013016 tn6 4 3,964 4
83 3.477 2.759 3,044 2.764 2:.F9)2:0 :::: 4.108 3,978 N) 55 721r ;69 ul
84 3.383 2,756 3,373 2.746 ,499 4.318
85 4.018 3.132 3.964 3.241 32:719387 .493 4.723 :::972 5
86 3,825 2.889 3.647 3.083 2.939 .477 4.699 ,::.);(
87 2,802 2.364 20572 2.433 20234 ,459 3.234 1;:(1):: 6
88 2.760 2.339 2.545 2,391 2.205 0461 3.221 9
810 3.562 2.556 3.042 2,817 2.406
.- ,427 4,930 .33
6iii 1
811 3.898 2.883 3.470 3.092 2 ,t45 4,915 4.99,Ei!. 5.968
812 3.908 3.086 3,685 3,108 ,472 4:941 4,868 5.877
813 5.039 3.651 4,554 3,716 2Qesiii3 ,452 6,274 6.082 8.059
814 4.793 3.335 4.138 3,517 .432 6.120 6,326 8.211
815 4,848 3.453 40274 3.558 3:M ,441 6.334 6,126 8,219
816 2.956 2.447 2,608 2.340 2,064 ,441 3.407 3.300 4
817 20962 2.437 2.600 2.344 2.057 ,439
818 2.976 2.428 2.592 2,351 30,IT = .e 61
.T
819 2,552 2.277 2.421 2.238 .Ig :44-1'4'
2.() 3.317
820 2,474 2.159 2.295 2.175 2.018 .464 :
,..41 :*..T3% -3.250
822 2.178 1.953 2.075 1.982 1.888 .476 2.075
.
2088 2.332
5 4t/ z L- AL ;V-
z /k4 L Pel t" 411i
19L C/// 2 er 4t
823 2.188 2.100 2.223 1,992 2.023 .508 2.687 2.674 2.333
824 2.175 1.997 2.118 1.979 1.927 ,487 2.160 2.087 2.351
825 2.014 1.980 2.094 1.880 1.955 .520 1.404 1.402 1.482
826 2.107 2.047 2.172 1,967 2.029 ,516 1.385 1.383 10459
827 2.127 2.024 2.152 2.009 ,506 1.411 1,409 1.489
828 2.257 2.081 2,239 ;:M53 2.022 , 496 2.403 2.389 2,676
829 2.242 1.916 2,074 2.020 1,869 ,463 2.442 2,428 20721
830 2.257 1.961 2.118 2,030 1.905 ,469 2.478 2,463 2.769
831 2.034 1,817 1.956 1.8+65 1.797 ,451 /0650 1,648 1.760
832 2.160 1.892 2,048 1,985 1•8R2 or 1.645 1,643 10734:•
833 2.342 2.253 2,435 20150 2,235 0 520 1.683 10680 1.778
834 3.178 2.676 2.818 2,357 2,090 ,443 3.000 2,903 3.883
835 3+121 2,639 2,776 2.322 2.066 0 445 2.969 20874 3.840
836 3.06 2.522 2.656 2.296 1.976 ,430 2,968 2.873 .10839
837 2.337 2.239 2,339 2.047 2.049 ,500 2.475 20429 3.001
838 2.347 20171 2..277 2,066 2.004 0455 2.357 2,322 2,805
839 2.374 2,209 2,316 2, 087 2.036 ,488 2.394 20.356 2.863
840 2.131 2,084 2.182 1,968 2.015 ,512
4 10863 1.850 2.100
841 2.128 2.052 2,152 1.966 1,969 ,506 1.827 1,816 2.045
842 2.258 2.170 2.283 2.088 2,112 ,506 1.815 1.804 2.029
843 2.119 2,275 2.384 1.999 2.250 ,563 1.227 1,225 1.295
844 2.194 2.048 2.163 2.072 2.043 ,493 1.225 1.223 10293
845 2.112 2.102 2.209 1,995 2,087 ,523 1.211 10209 1,276
847 3.159 2.784 3.229 2.654 2.713 .511 3.786 3,765 40235
848 3.275 2.527 3,003 2,756 2.527 ,459 3,739 3.'717 4.167
849 2.444 2,272 2.573 2,126 2.238 ,526 2,478 2,474 2.610
850 2.374 2,095 2.381 2.060 2.066 ,502 2,527 2,524 2.666
851 2.662 2.534 2,891 2,312 2.511 ,543 2.546 2,542 2.685
852 2.240 1,806 2.114 1,885 1,778 .472 2.970 2,966 3,129
853 2.364 1,909 2.249 1,993 1.896 .476 2+936 20 932 3.090
gif L/4
GA. /kr, L 5;
4 2/Y /L-2.
1(1-
R54 2.641 2,109 2,526 2,232 2.134 .478 2.916 2.913 3.067
855 4.534 3.444 4.404 3.520 3.419 ,486 5.965 5,882 7.058
856 4.345 3.180 4.046 3.359 3.128 ,466 6,029 5.937 7.1e4
857 4.565 3.375 4.326 3.531 3,346 .474 6.060 5.968 7.222
859 6.235 4.379 6.026 4,631 4.476 ,483 7.784 7.546 10.002
R60 6.645 4.626 6.481 4.943 4.621 ,488 7.856 7,615 10.095
561 2.010 1.947 2,034 1.912 1.935 6 506 1.097 1,095 1.158
1362 2.041 2.168 2.257 1.942 2.148 0 553 1.090 1.089 ).150
R63 1,996 2.016 2.102 1.699 2,000 ,527 1.064 1,082 1.14'2
R64 2.102 2.183 2.259 1,920 2,063 6 537 1.666 1.655 1873
865 2.285 2.250 2.340 2,093 2,143 .512 1,603 1.593 1.7Y1
866 2.118 2.147 2.224 1.940 2,437 6 525 1.600 1,591 1.783
867 2.384 2,214 2,294 2,032 1.956 q,481 2.093 2,060 2.500
868 2.616 2.395 2.493 2.224 2,118 2.140 2,106 2.557
1369 2,422 2.274 2,359 2.056 2,004 ::877 2.131 2,098 2.546
R70 2.053 '2.279 2,352 1.954 2.239 .573 .973 10023
871 2,207 2,403 2,467 2.101 2,369 .564 90
9;61 6 964 1.018
B72 2.188 2,368 2,451 2.084 2.335 0 560 .962 .961 1.014
R73 2.156 2,226 2.296 1.997 2,128 .533 1.441 1,433 1,612
974 2,019 2.106 2.169 1.870 2,009 .537 1.441 1,432 1.610
875 2,114 2,236 2,304 1,957 2.133 ,545 1.450 1.441 1.621
876 2.299 2,203 2,277 2,036 2.017 0 495 1.687 1,857 2.256
877 2.238 2,134 2,205 1.980 1,951 0 493 1.898 1,868 2.269
R78 2,205 2.104 2.172 1.951 1,922 ,493 1.894 1.864 2.263
Cl 2,999 2.466 2.706 2.602 2.348 .451 3.262 3,211 3.893
C2 2.698 2.285 2,510 2,514 2,178 0 433, 3,258 3,207 3.893
C3 2.710 2,314 2.476 2,381 2,177 $ 457 2.738 2,695 3.274
C4 20729 2.292 2,457 2.397 2.159 ,450 2.739 2,696 3.274
C5 2.514 2.155 2.274 2.214 2,002 ,452 2034 2,336 2.838
C6 2.410 2,094 2,203 2,122 1.939 6 457 2.371 2.334 2.836
st- & 21- L/2.
kurt kp L c(//462- L /(4. Wlo 41) /b
C7 2,367 2.141 2,231 2,060 1.94i .471 2,112 2,079 2.524
CS 2.399 2.299 2.390 2.087 2.079 ,498 2.118 2,085 2.532
D1 6.469 4.561 6.329 4,896 40791 ,489 7,071 9.070
D2 6.094 40214 5,800 4.596 4.375_ ,476 1
72Z: 7.074 9.095
D3 5.932 3.995 5.507 4,461 4014 ,464 7.110 9.165
n4 6.149 4.377 6,006 40-636 4.528 ,488 7,000 9.047
05 6.297 4.427 6:119 4.743 4.668 ,485 7.171 9,242
06 5.954 3.947 5.448 4.468 4,088 .457 72
7:397: 7,174 9.327
D7 5.335 3,663 4,660 3.915 3,419 ,437 6.429 6.210 3,383
D8 6.341 4,528 5,833 4.524 4,161 .460 7.162 6,844 9.706
D9 6.252 4.267 5.517 4.394 3.878 ,441 7.428 7.058 101254
G1 6.669 4,301 6,146 4,922 4,535 ,461 7,111 104508
82 6.703 4,285 6.096 4,801 4.367 .455 (
1:31 :0
0 7,745 10.433
G3 6.772 4.137 6.039 4,703 4.194 ,446 7.843 70 620 9.979
64 7,675 4.483 6.843 50157 4.598 .446 7.591 7,403 9.506
80494 . 4.658 70459 5.326 4,677 4439 7,458 7,291 9.236
(46 9,821 5.451 9.043 5,322 4.966 ,460 7.223 7,092 8.749
Fl 6,455 3.991 6,038 4,972 4,649 0468 7,375 7,208 9.148
E2 6,211 3,844 5,750 4,766 4.412 ,463 7,395 7.224 9.191
F3 6,529 4.157 6.235 5.050 4,823 ,478 7.252 7,093 8.961
E4 6.630 4.111 6.224 5.119 4.806 ,469 7.349 7.177 9.145
E5 6.679 4,168 6.327 5,165 4,893 ,474 7.322 7.156 9,079
R6 4.989 3,446 4.467 3,939 3.570 ,453 6.060 5.914 7.698
F7 5.112 3.573 4.667 4,086 30731 ,457 6.104 5,936 7.729
E8 5.3j4 3.606 4,792 4.261 3,649 ,452 6.007 5.654 7.546
E9 5.129 3.404 4.506 4,107 3,666 ,439 5,904 7.670
E10 5.137 3.515 4.613 4,090 3.672 ,449 ::n72 6,051 7.954
F11 3.372 2,561 2,893 2,905 2.493 ,629 4.190 4,051 5.442
F12 3.553 2,699 3,068 3.062 2,645 ,432 4.051 ,5,425
E13, 3.549 2.692 3,070 3.064 2.650 ,432 14:M3 4,007 5.293

/
C.*"
yr 7. 1_,/ .2. 2,i_.
61 P/
V kr L z 14 L. G "'-
KO- :
"kt L LI-1-b, LA
F.14 3.535 2.725 3.086 3.046 2,
2.659 ,437 4.207 4,064 5,477
El5 3.499 2.657 39013 2.598 ,431 4 f:( 1b7:r 31- )
E16 3.104 2,574 2,814 :716"9/ 2,569 ,453 3.35 ',,M
;
El7 3.068 2,466 2.698 2,732 .440 3.ct: 3,305 ;: ,:i
itil
E18 3.195 2,596 2.851 2,846 ,446 3,363 3.266
E19 3.107 2,756 2,762 ...i6:0 ,,6
44L;
41 3,349 4,477
E20 30(J24 2.452 20677 2,688 2.379 , ,443 c,144fifI01 3,363 :::::
E21 20267 2.287 2,408 2.314 ,531 ,Q76 1,030
E22 2.253 2,437 ::17 6: 20342 ,541 1.002
E23 2,228 ::M 2.418 2,141 ,543 1:2:
E24 2,304 2.365 2,482 2.207 ,540 1.072 1.143
F25 2.295 2.307 2,430 2,206 ,529 16:0900970
°77:5 1:04:97ri
E26 2.306 2,144 2,267 2,160 [4.0P41 ,491 1.617 1,607 1:ET
E27 20360 2.292 2.230 2.166 ,486 1.387 1,578
F28 2,324 .
01160)10 2.225 2,197 9479 10604
E29 2.371 2.241 2.374 2.243 2,246 ,501 1.555 1, 0 578
F30 2,310 2.157 2,285 2,185 ,495 1.544
F31 2,560 2.257 2.412 2.362 2,226 ,471 2.367 1::
F32 2.563 2.279 2.436 2,366 2,249 ,475 2.347 2,313
F33 2.587 20265 2.424 2.388 ,468 69 2,333 Illi
F34 2.578 2.257 2,414 2,379 ,.:3277 ,468 ..337 2,338 2.833
F35 2,595 2.270 2,430 2.396 2.243 ,468 2.355 2,320 2.796
Fl 2.190 2,130 2.196 2,018 2.023 ,501 1.095 1,092
F2 2.363 2.331 2,412 2,187 2.232 .510 ,986 1:10;4
7
F3 2.296 2,416 2,492 2.124 2.305 ,543 ,935 0987 .1.042
F4 2.351 2,325 2,405 20177 2.227 ,511 .971 .970
F5 20300 2.405 2.485 2.184 ,526 .987 ,986 1:gl.'
F6 2.458 2.242 2.328 2.195 2.078 ,473 1.649 1,638 1.853
F7 2.497 2,283 2.371 2.230 2,118 ,475
F8 2,611 2.512 2,610 2.333 2.332 ,500 1::g 1.637 :'
14t2p.. ...3,tc tro
/t2
e*
• 4/ 7 r Z
1(iL k, L rn0 L
Run V4
F9 2.558 2.349 2.443 2.286 2,183 4,478 1.630 1,620 1.8,25
F10 2.534 2,303 2.397 2.266 2,144 ,473 1,607 1,598 1.793
Fll 2,729 2.390 2,497 2.322 2.125 ,457 2.466 2,422 20982
F12 2,917 2,883 3.006 2,484 2,561 ,515 2.452 2,410 2.956
F13 2,942 2,528 2.652 2,505 2.258 .451 2.461 2.415 2.975
F14 2,950 2.639 2.766 2,515 2.355 ,469 2.436 2,395 2,927
F15 2.964 2.604 2.729 2,536 2,319 .457 2.495 2.446 3.034
F16 3.558 2,843 3.023 2.805 2,353 0 425 3,616 3.470 4.831
F17 3.743 2.956 3.166 20971 2.513 ,423 3.521 3,399 4.597
F18 3.579 2,841 3,024 2,831 2.393 0 423 3,571 3,432 4.744
F19 3.820 3.102 3.323 3,046 2,656 .435 3,443 3,333 4.450
F20 3.766 2,988 3,203 3,000 2.551 ,425 3.457 3,345 4.473
F21 4.539 3.511 3.903 3,442 2.960 0 430 4.176 4.045 5.38:3
F22 4.386 3.307 3.650 3,308 2.753 $ 416 4,263 4.106 5.621
F24 4.354 3.404 3.748 3,304 2.845 ,430 4.172 4,028 5s444
F25 4,435 3,526 3.880 3,400 2.974 $ 437 4.024 3,895 5.203
F26 6,866 4.582 5.878 4.522 30872 0 438 60577 6.349 8.593
F27 7,076 4.866 6.239 40 690 4,136 .441 6.567 60287 8,472
F28 7.067 40878 6.264 4.723 4.186 .443 6,379 6,176 6,235
F29 6,871 4.649 5.947 4,525 3.917 0433 6,578 60350 8.594
F30 6.933 4.565 5.871 4,583 3,880 0 423 5.533 6,304 8.548
F31 8.379 5.172 7.460 5,239 4.665 0 445 7.832 7.592 10.064
F32 8.311 5.050 7,294 5,188 4,553 .439 70657 70610 10.132
F33 8.230 5.241 70445 5,136 4,646 0 452 7,846 7,599 10,118
F34 7,948 4,766 60781 4.924 4.281 0 427 7.904 7,635 10.298
F35 8.691 5.404 7,941 5,472 4,999 .457 7.771 7.561 9.830

You might also like