Friction at Strip-Roll Interface in Cold Rolling

University of Wollongong
Research Online
University of Wollongong Thesis Collection University of Wollongong Thesis Collections
2002
Friction at strip-roll interface in cold rolling

Yinjian Liu
Recommended Citation
Liu, Yinjian, Friction at strip-roll interface in cold rolling, Doctor of Philosophy thesis, Faculty of Engineering, University of
Wollongong, 2002. http://ro.uow.edu.au/theses/1827
Research Online is the open access institutional repository for the

University of Wollongong. For further information contact the UOW
Library: [email protected]
Friction at Strip-Roll Interface
in Cold Rolling
A thesis submitted in fulfilment of the requirements for the award of the degree
DOCTOR OF PHILOSOPHY
from
by
YINGJIANLIU
BE, ME (NEU)
Mechanical Engineering
Faculty of Engineering
2002
STcmusfamily/
Certification i
CERTIFICATION
I, Yingjian Liu, declare that this thesis, submitted in fulfilment of the requirements for
the award of Doctor of Philosophy, in the Mechanical Engineering, Faculty of
Engineering at the University of Wollongong, Australia, is wholly m y o w n work unless
otherwise referenced or acknowledged. The document has not been submitted for
qualification at any other academic institution.
SAZU>
Yingjian Liu
15 August 2002
Acknowledgements ii
ACKNOWLEDGEMENTS
The author would like to express sincere gratitude to his supervisor, Prof. Kiet Tieu with
Mechanical Engineering, Faculty of Engineering at the University of Wollongong for
his guidance, support and encouragement throughout the course of this research.
The author is also deeply grateful to BHP Research and University of Wollongong for
providing him the postgraduate scholarship to make this research and thesis possible.
H e wishes to thank Dr. W . Y. D. Yuen for his support to this project.
Acknowledgement is also given to technical staff in the Faculty of Engineering,

particularly, to Mr. M . Morrillas, Mr. K. Maywald, Mr. T. Kent and Mr. J. Abbott for
their help during experiments.
Acknowledgment is also given to my colleagues, Mr. D. D. Wang, Dr. C. Lu, Dr. E. B.

Li, Mr. B. M a , and Dr. Z. Y. Jiang for their enthusiastic help in the experiment and
discussion on academic issues.
The author would like to express thanks to his darling wife H. Mao, for her love,
continuous encouragement and the waiting during the Ph.D study. H e hopes that she
will proud of this dissertation and that this will provide some satisfaction for her many
sacrifices. Finally, he is deeply grateful to his parents for their continuous
encouragement and support.
Thank you all indeed!

Abstract iii
ABSTRACT
Friction and lubrication are important in the roll bite of cold rolling process as they
influence the rolling force and other rolling process parameters. The rolling force
determines the reduction of strip being rolled, the rolling mill stretch and final stri
thickness and shape accuracy. The ability to control more accurately the rolling force i
industry is becoming increasingly important. Inaccurate understanding of friction in the
roll bite has affected the accuracy of rolling force prediction, and hence the accuracy
the final strip thickness.
Much research work has been done in modeling the pressure distribution for the rolling
process under fully hydrodynamic lubrication and mixed film lubrication. At the same
time, some experimental works have also been done in determining the value of friction
coefficient in cold rolling by using embedded pin transducer method in the work roll.
This study proposes to determine the friction coefficient and temperature in the roll b
in cold rolling by experimental methods. A sensor roll embedded with loadcells and
strain gauged pins was designed and manufactured to determine the friction coefficient
along the roll bite in cold rolling. The loadcells and strain gauges were calibrated in-
situ. The average friction coefficient was also derived from the forward slip which was
determined by Laser Doppler method and a strip marking method. The temperature over
Abstract iv
the roll bite was measured by the embedded thermocouple under different rolling
conditions.
A large number of experiments were performed to determine the friction coefficient
under different reduction, rolling speed, surface roughness, material property and
lubricants. The relationships between the rolling parameters such as rolling force,
rolling torque, temperature, lubricant, friction coefficient and surface roughness etc.
have been discussed. Empirical formulae of friction coefficient and rolling force were
given.
The measured friction coefficient was used in the calculation of rolling force and torqu
which were compared with the measured values to validate the measured friction
coefficient.
A theoretical model was developed by to consider the hydrodynamic inlet zone, plastic
work zone and hydrodynamic outlet zone. The effects of lubricant and friction
coefficient etc. in the mixed film model were discussed. The effect of the modification
to the hydrodynamic film thickness at the inlet caused by the strip elastic recovery at
entry was also considered.
The temperatures in the lubricant and at the asperity contacts were calculated,
respectively. The lubricant temperature at the roll and strip interface and lubricant m
temperature was obtained by using the energy equation. The temperature at asperity
contact was calculated by using the moving heat source theory. The temperature
calculation considers not only the plastic deformation of the bulk material, but also th
Abstract v
frictional heat at the interface. The effects of different friction coefficient, reduction and
fraction of plastic work converted to heat have been discussed. The final strip and roll
surface calculated temperature was verified by the experimental work and other author's
work.
Table of Contents vi
Table of Contents
CERTIFICATION i
ACKNOWLEDGEMENTS ii
ABSTRACT iii
TABLE OF CONTENTS vi
LIST OF FIGURES xi
LIST OF TABLES xix
NOMENCALTURE xx
PUBLICATIONS xxvi
Chapter 1 INTRODUCTION 1
1.1 General background 1
1.2 Importance of this research project and its main objectives 3
1.3 Scope of the thesis 3
1.4 Overview of this thesis 4
1.5 S u m m a r y of contributions of this thesis 5
Chapter 2 ROLLING THEORY AND LUBRICATION IN COL

ROLLING 7
2.1 Introduction 7
2.2 Rolling theory 8
2.2.1 Rolling geometry 9
2.2.2 Material resistance 9
2.2.3 Rolling force and torque 12
2.2.4 Friction coefficient and lubrication 17
Table of Contents vii
2.3 Friction coefficient measurement 21
2.3.1 Sensor roll method 22
2.3.2 Strip marking method 23
2.3.3 Laser Doppler method 24
2.4 Lubrication theory in cold rolling 25
2.4.1 Hydrodynamic lubrication 25
2.4.2 Mixed film lubrication 27
2.4.3 Effect of roughness 28
2.5 Current research on friction coefficient measurement 32
2.6 Current research on lubrication model in cold rolling 34
2.7 Problems and expectation 36
2.8 Summary 38
Chapter 3 LITERATURE SURVEY ON FRICTION

MEASUREMENT AND LUBRICATION MODEL IN
COLD ROLLING .39
3.1 Introduction 39
3.3 Hydrodynamic lubrication model in cold rolling.... 48
3.4 Mixed film lubrication model in cold rolling 52
3.5 Thermal effect in cold rolling 54
3.6 Summary 58
Chapter 4 FRICTION COEFFICIENT MEASUREMENT AND

ROLLING EXPERIMENTS 60
4.1 Introduction ...60
4.2 Sensor roll design 61
4.3 Sensor roll calibration 64
4.4 Experimental facilities 67
4.5.1 Signal record 71
4.5.2 Friction coefficient measurement 77
Table of Contents viii
I. Sensor roll method 77
II. Strip marking method 90
III. Laser Doppler method 92
4.6 Experimental results discussion 93
4.6.1 Effect of rolling parameters on rolling load 94
4.6.2 Effect of rolling parameters on forward slip 97
4.6.3 Strip surface roughness in rolling process 99
4.6.4 Temperature in cold rolling 103
4.6.5 Friction coefficient in cold rolling 109
I. Aluminium alloy.... 109
II. Carbon steel 110
4.7 Friction coefficient empirical formulae 113
I. Aluminium alloy 113
II. Carbon steel 116
4.8 Empirical formulae of rolling force 119
4.9 Summary 120
Chapter 5 VALIDATION OF FRICTION COEFFICIENT

MEASUREMENT IN COLD ROLLING 122
5.1 Introduction 122
5.2 Yield strength 123
5.3 Theoretical considerations 130
5.4 Results and discussions 136
5.4.1 Validation of friction coefficient measurement 136
5.4.2 Neutral angle 140
5.5 Summary 142
Chapter 6 MIXED FILM LUBRICATION IN METAL ROLLI

6.2 Rolling model 145
6.3 Programming 151
Table of Contents ix
6.4.1 Effect of elastic entry and exit 154
6.4.2 Effect of lubricant viscosity 155
6.4.3 Effect of friction coefficient at asperity contact 158
6.4.4 Comparison between calculated and experimental data 160
6.4.5 Oil drop method 161
6.5 Summary 164
Chapter 7 INFLUENCE OF HYDRODYNAMIC INLET AND

OUTLET ZONE IN MIXED FILM MODEL 165
7.2 Theoretical analysis 166
7.2.1 Inlet zone 167
7.2.2 Plastic work zone 168
7.2.3 Outlet zone 168
7.3 Solution process 169
7.4 Results and discussion 171
7.5 Conclusions 176
Chapter 8 THERMAL EFFECT IN MIXED FILM LUBRIC

COLD ROLLING PROCESS 177
8.2 Theoretical analysis 179
8.2.1 Inlet zone 179
8.2.2 Plastic work zone 182
8.2.3 Outlet zone .....187
8.3 Solution process 188
8.5 Conclusions • 203
Chapter 9 CONCLUSIONS AND RECOMMENDATIONS 204

9.1 Conclusions • -204
9.2 Recommendations 207
Table of Contents x
Appendix A DATA ACQUISITION PROGRAM 209
Appendix B EMPIRICAL FORMULA OF FRICTION COEFFICIENT 22
Appendix C ENERGY EQUATION 239
Appendix D MOVING HEAT SOURCE BOUNDARY CONDITION 246
Appendix E TEMPERATURE AT ASPERITIES CONTACT 251
Appendix F UNCERTAINTY ANALYSIS OF FRICTION COEFFICIENT

MEASUREMENT 253
References 256
List of Figures xi
List of Figures
Figure 2.1 Three-dimensional view of strip rolling process 8
Figure 2.2 Distribution of normal pressure and rolling force 10
Figure 2.3 Parameters of deformation zone 11
Figure 2.4 Frictional shear stress... 19
Figure 2.5 The lubrication regimes.... 21
Figure 2.6 Radial and oblique pins in roll bite 22
Figure 2.7 Marks on the strip 23
Figure 2.8 Laser Doppler method 24
Figure 2.9 Asperity contact in the mixed lubrication regime 29
Figure 2.10 Longitudinal, isotropic and transverse surface roughness 29
Figure 4.1 Sensor roll overview 62
Figure 4.2 Sensor roll section view 62
Figure 4.3 Side view of sensor roll and pins .....63
Figure 4.4 Front view of sensor roll and pins .63
Figure 4.5 Calibration process illustration 64
Figure 4.6 Radial strain gauge calibration 65
Figure 4.7 Oblique strain gauge calibration 65
Figure 4.8 Radial loadcell calibration 66
Figure 4.9 Oblique loadcell calibration 66

List of Figures xii
Figure 4.10 Hille 100 rolling mill 67
Figure 4.11 Data acquisition system 68
Figure 4.12 Data acquisition screen 70
Figure 4.13 Typical rolling force record 72
Figure 4.14 Typical rolling torque record 72
Figure 4.15 Strip thickness within segment part 73
Figure 4.16 Rolling force over roll bite 73
Figure 4.17 Rolling torque over roll bite 74
Figure 4.18 Rolling force recorded by ordinary roll 74
Figure 4.19 Thermocouple in sensor roll 75
Figure 4.20 Temperature record 75
Figure 4.21 Pin signal test for carbon steel 76
Figure 4.22 Pin signal test for aluminium alloy 77
Figure 4.23 Force distribution 78
Figure 4.24 Force distribution s=29.48% (lubricated) 78
Figure 4.25 Force distribution s= 17.71% (dry) 79
Figure 4.26 Pressure distribution e=29.48% (lubricated) 79
Figure 4.27 Friction coefficient s=29.48% (lubricated) 80
Figure 4.28 Pressure peak in traditional rolling theory 80
Figure 4.29 Pressure distribution 8=29.48% (lubricated) 81
Figure 4.30 Friction coefficient e=17.71% (dry) 81
Figure 4.31 Force distribution (e=36.82%, n=3rpm) 82
Figure 4.32 Friction coefficient (s=36.82%, n=3rpm) 82
Figure 4.33 Force distribution (8=37.01%, n=5rpm) 83

List of Figures xiii
Figure 4.35 Force distribution (e=37.61%, n=7rpm)..... 83
Figure 4.36 Friction coefficient (e=37.61%, n=7rpm) 84
Figure 4.37 Force distribution (s=40.45%, n=15rpm).. 84
Figure 4.39 Force distribution (s=40.38%, n=30rpm) 85
Figure 4.40 Friction coefficient (s=40.38%, n=30rprn) 85
Figure 4.41 Force distribution (s=40.38%, n=50rpm)....... 85
Figure 4.43 Force distribution (e=40.12%, n=65rpm)....... 86
Figure 4.47 Force distribution (8=29.75%, n=5rpm).... 88
Figure 4.51 Force distribution (s=32.66%, n=20rpm)... 89
Figure 4.55 Friction coefficient vs rolling speed (strip marking method) 92
Figure 4.56 Rolling force (ur=0.118m/s) .94
Figure 4.57 Rolling torque (ur=0.118m/s) .94
Figure 4.58 Rolling force (aluminium alloy) 95

List of Figures xiv
Figure 4.59 Rolling torque (aluminium alloy) 95
Figure 4.60 Rolling force (carbon steel) 96
Figure 4.61 Rolling torque (carbon steel) 96
Figure 4.62 Effect of lubricant on rolling force 97
Figure 4.63 Effect of lubricant on rolling torque 97
Figure 4.64 Effect of reduction on forward slip 98
Figure 4.65 Effect of rolling speed on forward slip 98
Figure 4.66 Effect of roughness on rolling force 99
Figure 4.67 Effect of roughness on rolling torque 99
Figure 4.68 Strip surface roughness along the rolling direction 100
Figure 4.69 Strip surface roughness along transverse direction..... 100
Figure 4.70 Strip surface roughness along rolling direction 101
Figure 4.71 Strip surface roughness along transverse direction 101
Figure 4.72 Strip surface roughness along rolling direction..... 102
Figure 4.73 Effect of roughness on forward slip 102
Figure 4.74 Effect of roughness on friction coefficient 103
Figure 4.75 Strip surface temperature vs reduction 104
Figure 4.76 Strip surface temperature vs rolling speed 104
Figure 4.77 Strip surface temperature vs rolling speed 105
Figure 4.78 Roll surface temperature vs rolling speed 106
Figure 4.79 Strip surface temperature variation 106
Figure 4.80 Roll surface temperature variation 107
Figure 4.81 Roll surface temperature (8.43%) • 107
Figure 4.82 Roll surface temperature (13.82%) 108

List of Figures xy
Figure 4.86 Friction coefficient vs rolling speed (27.36%) 110
Figure 4.87 Friction coefficient vs rolling speed (38.94%) 110
Figure 4.88 Friction coefficient vs rolling speed (ground) 111
Figure 4.89 Friction coefficient vs rolling speed (pickled) Ill
Figure 4.90 Friction coefficient vs rolling speed (0.0086 Pa.S) 112
Figure 4.91 Friction coefficient vs rolling speed (0.93 Pa.S) 112
Figure 4.92 Comparison of measured and predicted friction coefficient (n=3rpm) .115
Figure 4.93 Comparison of measured and predicted friction coefficient 116
Figure 4.94 Comparison of measured and predicted friction coefficient (n=3rpm). 117
Figure 4.95 Comparison of measured and predicted friction coefficient 118
Figure 4.96 Comparison of calculated and measured rolling force 120
Figure 5.1 Test sample dimension 126
Figure 5.2 Engineering stress and strain (carbon steel) 126
Figure 5.3 True stress and strain (carbon steel) 126
Figure 5.4 Engineering stress and strain (aluminium alloy) 127
Figure 5.5 True stress and strain (aluminium alloy) 127
Figure 5.6 Relation of R and C 3 128
Figure 5.7 Validation of yield strength (carbon steel) 129
Figure 5.8 Validation of yield strength (aluminium alloy) 130
Figure 5.9 Plastic work zone in rolling 130
Figure 5.10 Rolling force • 135
Figure 5.11 Rolling torque 136
Figure 5.12 Pressure distribution for lubricated condition 137

List of Figures xvi
Figure 5.13 Pressure distribution for dry condition 138
Figure 5.14 Rolling force comparison (dry) 139
Figure 5.15 Rolling torque comparison (dry) 140
Figure 5.16 Neutral angle (aluminium alloy 6060-T5) 141
Figure 5.17 Neutral angle (carbon steel BHP-300) 141
Figure 6.1 Rolling process and sheet surface topography 145
Figure 6.2 Forward slip vs Sommerfeld number 153
Figure 6.3 Friction coefficient vs Sommerfeld number 153
Figure 6.4 Total rolling pressure comparison 154
Figure 6.5 Film pressure comparison 155
Figure 6.6 Total pressure distribution over the roll bite 156
Figure 6.7 Film pressure distribution over the roll bite 156
Figure 6.8 Film thickness distribution over the roll bite 157
Figure 6.9 Asperity contact ratio in the roll bite 157
Figure 6.10 Rolling pressure under different friction coefficient 159
Figure 6.11 Film pressure under different friction coefficient 159
Figure 6.12 Film thickness under different friction coefficient 159
Figure 6.13 Asperities contact ratio 160
Figure 6.14 Pressure comparison s=23.84% (lubricated) 161
Figure 6.15 Pressure comparison s=29.48% (lubricated) 161
Figure 6.16 Image of footprint after rolling 162
Figure 6.17 Film thickness under different lubricant viscosity 163
Figure 6.18 Film thickness under different reduction ratio 163
Figure 7.1 Schematic diagram of rolling process 167
Figure 7.2 Calculation flow chart in the inlet zone 170

List of Figures xvii
Figure 7.3 Film thickness in the inlet area 171
Figure 7.4 Film pressure in the inlet area 171
Figure 7.5 Film pressure in the outlet area 172
Figure 7.6 Film pressure in the roll bite 172
Figure 7.7 Film pressure at the edge of inlet/plastic work zone.... 173
Figure 7.8 Strip elastic deformation at entry - h u m p 174
Figure 7.9 Film pressure distribution in roll bite (ur=3.55m/s) 174
Figure 7.10 Film pressure distribution in roll bite (ur=16.47 m/s) 175
Figure 7.11 Film thickness at entry 176
Figure 8.1 Flow between moving surfaces..... 181
Figure 8.2 Time step 191
Figure 8.3 Temperature distribution in the roll bite 191
Figure 8.4 Temperature comparison at asperity contact 192
Figure 8.5 Temperature comparison 192
Figure 8.6 Pressure without thermal effect 193
Figure 8.7 Pressure with thermal effect 193
Figure 8.8 Total pressure comparison 194
Figure 8.10 Asperity contact in roll bite.... 195
Figure 8.11 Film thickness in roll bite 195
Figure 8.12 Strip surface temperature in oil valley (/=0.8) 196
Figure 8.13 Film mean temperature (/=0.8) 196
Figure 8.14 Temperature at asperity contact 197
Figure 8.15 Strip surface temperature 198
Figure 8.16 Film m e a n temperature 198

List of Figures xviii
Figure 8.19 Temperature comparison 200
Figure 8.20 Rolling force comparison 200
Figure 8.21 Rolling torque comparison 201
Figure 8.22 Temperature distribution in the roll bite 201
Figure 8.24 Temperature in the hydrodynamic inlet zone 202
Figure 8.25 Temperature in outlet zone 203
Figure C l Flow between moving surfaces 241
Figure F.l Uncertainties analysis of friction coefficient measurement 255

List of Tables xix
List of Tables
Table 4.1 W302 specification 61
Table 4.2 Aluminium alloy 6060-T5 71
Table 4.3 Aluminium alloy 5052-H34 71
Table 4.4 Carbon steel BHP-300 71
Table 4.5 Comparison of forward slip 93
Table 5.1 Experimental results. 138
Table 5.2 Rolling force and torque comparison 138
Table 6.1 Rolling force and torque under different viscosity..... 158
Table 6.2 Rolling force and torque under different friction coefficient 160
Table 6.3 Viscosity of lubricants at 40°C 162
Table 7.1 Results comparison 172
Table 7.2 Rolling force and torque 175
Table 8.1 Lubricant properties 189
Table 8.2 Properties of materials 189

Nomenclature xx
Nomenclature
a=half pitch of asperity at contact, m

^4=ratio of contact area, dimensionless
A' = instantaneous cross sectional area, m
j40=cross-seetion area, m 2
b =strip width after deformation, m

B =strip width before deformation, m
/Jj, B2 ^distance, m
c, na =friction coefficient at the asperity contact
cf,cr, cw =lubricant, roll, and strip specific heat, J/kg.K respectively
cp=specific heat, J/kg.K
C =non-dimension pressure parameter, e ap;

C = f l o w constant
C 0 ^parameter
C},C2,Ci,C4,C5, C6 =parameters
e =engineering strain
E =temperature viscosity coefficient
E ,EX,E2, E3 =parameters
Er,Es=rol\ and strip's Y o u n g modulus respectively, Pa
/ =horizontal tension force per unit width, N / m
/i, fi > fn>fn parameters

/(*',/') =heat partition function
F =non-dimension horizontal tension force, / /(ayo y1)
Fl =applied load, N
Nomenclature xxi
Fol =force acting on oblique loadcell, N
Fos =force acting on oblique strain gauge, N
Frl =force acting on radial loadcell, N
Frs =force acting on radial strain gauge, N
G =plastic thermal parameter
G , Ge2 =torque at elastic entry and exit area, N - m
Gt =total rolling torque, N - m
G,' =rolling torque neglecting elastic effect, N - m
h,ht =fllm thickness and average film thickness, m
fy, /z2=film thickness at inlet and outlet of work zone, m
fy =film thickness at point i in inlet area, i = 0,l,2, ,JV
hB, fy=film thickness at point B and C, m
i/=non-dimensional film thickness, h/(x,2/2i?)
Hx =film thickness at inlet of work zone, fy/(Xj J2K) , dimensionless
H2 =film thickness at outlet of work zone, h2 j{xx2 /2R) , dimensionless
H\ =film thickness at inlet of work zone,fyj§ , dimensionless
H, =non-dimensional average film thickness, fy / 8
k conductivity, w / m . K
fy =parameter
kf,fy,kw =lubricant, roll, and strip conductivity, w / m . K
fc^=shear yield strength, Pa
kSi, kSi =shear yield strength at entry and exit point of work zone, respectively, Pa
A:=diffusivity, m /s
Kf,Kr,Kw=lubricant, roll, and strip diffusivity, m /s
/=half pitch of asperity of surface tooth, m
L, Lx =strip length before and after deformation respectively, m
LQ = circumference of the work roll, m
L2 = original linear dimension along the loading axis of the specimen, m

Nomenclature xxii
L' = mark length left on the strip after the roll turns one revolution, m
L" = instantaneous gauge length, m
A L =change in length, m
m = interface shear factor
M =time intervals
n =roll rotation speed, rpm
N =number of intervals over roll bite
p, pa,p f=pressure, average contact pressure, and film pressure respectively, Pa
/?r=radial pin pressure, Pa
pg =oblique pin pressure, Pa
pfi =film pressure at point i in inlet zone. / = 0,1,2, ,N
P,Pa,Pf =non-dimensional pressure, average contact pressure, and film pressure,
respectively, P = piayo, Pa = pa Icryo, Pf = pf Iayo
/^-^transformed film pressure, \-e~aPf
PX,P2, P3 =non-dimensional parameters
Pe , Pe contribution of the elastic entry and exit arc to the rolling force, N
Pd, P0 =rolling mill drive and operator side load, N
Pt =total rolling force, N
Pt ,==rolling force neglecting elastic effect, N
q =frictional shear stress, Pa
qa,qf =friction stress at contact area and film valley, Pa
£}=frictional shear stress, dimensionless, Q = ql<?yo
Qa, Qf =non-dimensional friction stress at contact area and film valley,
Qa=<Ial°y^Qf=<lfl(ry«
r =half height of the surface tooth, m
f =length-to-width ratio of a representative asperity
R, R' =roll and deformed roll radius, m
S =engineering stress, Pa
5*0 =Roelands' thermoviscous parameter

Nomenclature xxiii
Sf =forward slip
?=time, s
tx, t2 =back and front tension, Pa
tex, tei =plane-strain horizontal pressure at entry and exit respectively, P a
Tx,T2 =dimensionless back and front tension respectively, Tx=txl<Jy yx, T2=t2/cry v,
T =temperature
T0 =ambient temperature, K
Tf,Tr,Tw =lubricant, roll surface, and strip surface temperature in lubricant, K
Tf,Tr,Tw =4ubricant, roll surface, and strip surface temperature in oil, dimensionless
ff=Tf/T0,Tr=Tr/T0,Tw=Tw/T0
Tm =lubricant m e a n temperature, K
Tm =lubricant m e a n temperature, dimensionless Tm = Tm /T0
Trb,Twb=bulk temperature of roll and strip
Trb,Twb=bulk temperature of roll and strip, dimensionless Trb =Trb /T0, Twb =Twb /T0
Tr ,TW =roll and strip surface temperature at the entry of plastic work zone
Tr ,TW =roll and strip surface temperature at the entry of plastic work zone,
dimensionless, F n =Tn IT0, 7 ^ =TWi IT0
Tr ,TW =roll and strip surface temperature at the exit of plastic work zone
Tr ,TW =roll and strip surface temperature at the exit of plastic work zone,
dimensionless, T,2 =Th /T0, TWi =TWi /T0
Tt, Ttt =rolling mill total torque and top spindle torque, N - m
ATS =strip surface temperature rise due to plastic work, K
ATS =strip surface temperature rise due to plastic work, dimensionless ATS = ATSIT0
AT =temperature rise, K
ATr, A7 w =temperatureriseof roll and strip at asperity contact, K
ATr, A r w =temperature rise of roll and strip at asperity contact, dimensionless,
ATr=ATr/T0,ATw=ATJT0
Nomenclature xxiv
Tr, Tw=xo\\ and strip temperature at asperity contact, dimensionless
w,v,w=speed along x, y, and z direction, m/s
ux, u2 =surface speed, m/s
ur, uw =roll and strip speed, m/s
u w =strip speed at entry of roll bite, m/s
u„2 =strip speed at exit of roll bite, m/s
Vd, F 0 =output of rolling mill drive and operator side loadcell, voltage
F0/=output of oblique loadcell, voltage
Vos =output of oblique strain gauge, voltage
Vr =output of roll speed, voltage
Vrl =output of radial loadcell, voltage
F„=output of radial strain gauge, voltage
F^sliding speed, m/s
Vt,Vtt = output of rolling mill total torque and top spindle torque, voltage
Vtc =output of thermocouple, voltage
W =strip width, m
W =strip constrained yield stress (1 + C3s)c* (1 + C5s)C6, dimensionless
x =distance, m
xx =roll bite length, m
x',y', f = d u m m y variables of x,y,t, respectively
X,X' =dimensionless variable, X = x/xx, X'-x'lxx
y =strip thickness in roll bite, m
v 0 =strip thickness in its annealed state, m
yx,y2 =strip thickness at entry and exit of roll bite, m
yn =workpiece thickness corresponding to the peak pressure
F=non-dimensional y -coordinate, y/h
Y =non-dimensional strip thickness, v/v,
Z =Roelands' pressure-viscosity parameter

a =viscosity pressure factor, Pa_1
Nomenclature xxv
/?=temperature-viscosity coefficient, 1/K
£=RMS combined surface roughness, ^Sr2 +SW2 , m
dr, Sw =RMS roll and strip surface roughness, m
s = reduction ratio, ——— x 100%
st =true strain, ln(yx I y)
fy=true strain rate, 1/s
<f> =angle from the exit plane
^=^011 bite angle
(j)XN =roll bite angle difference in inlet area
(j>n =neutral angle in roll bite
Q>x =flow factor in x direction
^wc =speed flow factor in x direction
7=fraction of plastic work converted to heat
7=lubricant viscosity, Pa.S
TJ0 =lubricant viscosity at ambient temperature, Pa.S
A =strip length elongation coefficient

ju =friction coefficient
vr, vs =roll and strip's Poisson ratio respectively
0=angle between radial pin and oblique pin, rad
Ga =asperity slope, rad
p =density, kg/m
pf,pr,pw =lubricant, roll and strip density, kg/m
<r2, crx ^vertical and horizontal compressive stresses, Pa
<jy, cry =yield strength of workpiece, Pa
Tt =shear stress along roll bite, Pa

ft
CJ, aia =heat flux and heat flux at asperity contact, w / m
£=strip thickness draft coefficient
y/ =strip width spread coefficient

Publications xxvi
Publications
1. Liu, Y. J., and Tieu, A. K , (2002). The Influence of Thermal Effects in Mixed-Film
Lubricated Cold Rolling Process, Proceedings of The Third Australasian Congress
on Applied Mechanics ( A C A M 2002), February 20-22, 2002, Sydney, Australia, pp.
321-326.
2. Tieu, A. K., Liu, Y. J., Kosasih, P. B., and Jiang, Z. Y., (2001). Effects o
Deformation and Temperature on a Mixed Film Lubrication Model, Proceedings of
2 n d World Tribology Congress ( W T C 2001), September 3-7, 2001, Vienna, Austria.
3. Liu, Y. J., Tieu, A. K, Wang, D. D., and Yuen, W. Y. D., (2001). Friction
Measurement in Cold Rolling, Journal of Materials Processing Technology, Vol.
111/1-3, pp 142-145.
4. Liu, Y. J., Tieu, A. K, and Qiu, Z. L., (2001). The Influence of Inlet and
Zone in Mixed-Film Lubricated Cold Rolling Process, Proceedings of 2 n d
International Conference on Mechanics of Structures, Materials and Systems
( M S M S 2001), February 14-16, 2001, Wollongong, Australia, pp. 321-326.
5. Tieu, A. K., Qiu, Z. L., and Liu, Y. J., (2000). Thin Film Lubrication in C
Rolling, Synopses of the International Tribology Conference, Nagasaki, Japan,
2000, pp. 239.
6. Tieu, A. K, Qiu, Z. L., DeOliveria, R., and Liu, Y. J., (1999). A Mixed Fil
Lubrication Model for Strip Rolling and Its Correlation with Experiment, 26
Leeds/Lyon Symposium on Tribology, September 1999, Leeds, England, pp. 457-
465.
Publications xxvii
7. Tieu, A. K., Li, E. B., and Liu, Y. J., (1999). A n Experimental Determination of
Friction in Cold Rolling, 26th Leeds/Lyon Symposium on Tribology, September
1999, Leeds, England, pp. 467-472.
8. Jiang, Z. Y., Tieu, A. K., and Liu, Y. J., (1999). An Analysis of the Meta
Process by a Full 3-D Rigid Plastic Finite Element Method, Proceedings of
International Symposium on Advanced Forming and Die Manufacturing
Technology (AFDM'99), Pusan, Korea, September 7-9, 1999, pp. 225-230.
9. Liu, Y. J., Tieu, A. K, Li, E. B., and Yuen, W. Y. D., (1999). Forward Sl
Friction in Cold Rolling, Proceedings of International Symposium on Advanced
Forming and Die Manufacturing Technology (AFDM'99), Pusan, Korea, September
7-9, 1999, pp. 495-500.
Chapter 1 Introduction 1
Chapter 1
Introduction
1.1 General background
The steel industry has had a long history of development. New ideas continue to
revolutionize the steel-producing process today as they did a hundred years ago. The
latest advances of making 'clean' steel, development of the continuous casting proces
for thin slabs and strip, introduction of the ingenious strip profile and shape contr
technologies in rolling mills are only few examples that illustrate the great potent
further innovations and discoveries. It is no wonder that many specialists, engineers
scientists from different countries still find the steel industry an exciting field
their creativity.
Rolling is an important metal forming process. About 70% of all metals are rolled at
least once during production. Cold rolling is used to produce the products with superior
mechanical properties, dimensional tolerances and surface finish. Significant
developments have been achieved in thickness control (gauge control) for both hot and
cold rolling in the last 30 years, w h e n significant advances of computer technology and
control theory have been made. But the nature of rolling is so complex, with m a n y basic
parameters not fully understood (e.g. the mechanism of friction and lubrication), that
further stringent quality requirements are difficult to achieve by current conventional
rolling technologies.
The widespread use of digital computers in a rolling process has raised their predict
capabilities to a level that was impossible just several years ago. It has also shown up
the deficiencies in our understanding of m a n y aspects of the rolling processes which
include the tribology at the interface. Thus, a significant amount of effort in metal
forming is currently directed at developing relevant fundamental information regarding
tribological phenomena which involves friction, lubrication and wear, and applying
them in sophisticated process models. Such models can have important benefits in
improving product quality and reducing costs and lead-time on n e w products. M a n y
areas of general interest to the tribological community in general, such as non-
Newtonian lubricant behavior and lubricant film breakdown, rough surface lubrication,
and asperity deformation, are also of great significance to those studying the tribology
of forming operations.
Currently there are two strands of researchers: those who study friction and lubricat
performance in rolling from an experimental point of view, and those w h o develop the
mathematical models by considering rolling process parameters such as lubricant,

material properties, surface roughness and thermal effect in hydrodynamic or mixed
film lubrication condition from a theoretical point of view.
The main objective of this research is to characterize the nature of friction in the r
bite and develop a thermal model for the mixed film lubrication process in cold rolling.
1.2 Importance of this research project and its main objectives
Friction and lubrication at the interfaces between two rolls and a metal strip being
plastically deformed by the rolls are one of the most important considerations in both
theory and practice of plastic working. H o w e v e r the nature offrictionand lubrication at
the strip-roll interface is not clearly understood. The traditional approach is to assume
that the frictional force in the roll bite is proportional to the normal force, with the
friction coefficient remaining constant in cold rolling. But this affects the accuracy of
the mathematical model and consequently, the thickness and shape of the strip.
The main objectives of this research are: firstly, understanding of friction variation
strip-roll interface in cold rolling; secondly, developing a cold rolling model by
considering hydrodynamic inlet zone and outlet zone as well as thermal effect under
mixed film lubrication.
1.3 Scope of the thesis
This thesis mainly concentrates on the measurement of friction coefficient in cold
rolling and the effect of various rolling parameters on friction coefficient and rolling
force. A comprehensive mathematical model taking into account material property,
surface roughness and lubricant viscosity etc. was established to solve cold rolling
problem.
1.4 Overview of this thesis
The study of friction and dynamics at the strip-roll interface is mainly composed of
measurement of friction and a roll gap model set-up. In this thesis, the friction
measurement has been carried out using a sensor roll embedded with pin loadcells. The
variable and average friction coefficients have been measured.
Firstly, a sensor roll was designed, manufactured and assembled with care. After thi
the calibration of the sensor roll was carried out in-situ several times to ensure the
results are repeatable. A high-speed data acquisition system for the sensor roll has been
developed for experiments under dry and lubricant conditions with aluminum alloys and
carbon steel. Strip marking method and Laser Doppler method were used in the
experiments to measure the forward slip which can be used to determine the friction
coefficient. Different rolling speeds, reduction and temperature were tested. Carbon
steel samples with different surface conditions such as pickling, sand blasting and
grinding were tested carefully. The 'oil drop' method was also used to measure the film
thickness for different rolling conditions. The measurement results were used to validate
the conventional rolling theory as well as the mixed film lubrication model.
A new program was developed to calculate rolling pressure distribution in the roll b
based on the Alexander theory. Based on numerous experimental results, empirical

formulae for friction coefficient and rolling force were developed, which were validated
by measured rolling load data. A mathematical model, considering thermal effect,
hydrodynamic inlet zone and outlet zone, was developed for cold rolling process. The
model error was determined from a large amount of experimental results. All the
empirical formulae and mathematical model were tested against experimental
measurements.
1.5 Summary of contributions of this thesis
The contributions in this research can be summarized as follows:
• A sensor roll embedded with two groups of sensor-loadcells and strain gauges was
designed, manufactured and assembled. A special jig was designed for the
calibration of the sensor roll. Four pins with two loadcells and two strain gauges
were calibrated several times to m a k e sure the calibration results are repeatable.
• A fast and reliable data acquisition system was developed. Four signals from the
sensor roll as well as force, torque, speed and temperature can be collected at the
same time.
• A large number of tests were carried out for aluminium alloy and carbon steel under
different rolling conditions. The friction coefficient was measured along the length
of contact. The results were compared with those from the strip marking method and
the Laser Doppler method in which forward slip was measured firstly, and then used
to calculate the average friction coefficient. Empirical expressions for variable
friction coefficient in the roll bite were established. The measured friction
coefficients were validated by different experimental techniques.

• A n 'oil drop' method was used in the experiment to measure film thickness. This test
was carried out for different type of lubricants. The measured film thickness was
used to verify the mixed film lubrication model.
• The mathematical models which includes the hydrodynamic inlet zone, plastic work
zone, and hydrodynamic outlet zone in the mixed film lubrication was developed.
• The thermal effect in mixed film lubrication was also considered w h e n the
temperature in the oil valleys and at the asperity contacts was calculated
respectively.
Chapter 2 Rolling Theory and Lubrication in Cold Rolling 7
Chapter 2
Rolling Theory and Lubrication in
Cold Rolling
2.1 Introduction
A rolling process involves contact between the workpiece and the work roll. Friction i
defined as the resisting force tangential to the interface between two bodies when, un
the action of an external force, one body moves or tends to move relative to the surfa
of the other. Practical characterization of friction in metalworking usually involves
identification of the mode of friction and/or lubrication and the level of shear stres
acting on the workpiece surface. This chapter presents a review of the basic rolling
theory, the techniques used in such characterization which involves experiments to
measure friction coefficient at strip-roll interface and mathematical model set-up i

rolling.
2.2 Rolling theory
Rolling is a process in which a piece of metal such as steel, aluminum etc. is deform
between two rotating rolls into a specific desired shape, particularly, a thinner flat strip
with a desired thickness as shown in Figure 2.1.
Figure 2.1 Three-dimensional view of strip rolling process
Rolling can be carried out at either a high temperature or an ambient temperature. The
first one is referred to as hot rolling, which involves large thickness reduction. Hot
rolling is performed at a temperature range 850-1150°C, w h e n the steel is "soft" with
low deformation resistance. The latter is referred to as cold rolling, normally producing
thin strips, which will be described in details here. Cold rolling is the final procedure in
steel strip production, so it plays an important role in quality control of products.

2.2.1 Rolling geometry
In rolling process, the metal goes through the work rolls and is deformed by the roll
pressure. During rolling, the material is deformed in three dimensions, as shown in
Figure 2.1, so that the material thickness is reduced from v, to v2, by a ratio
£ = yl jy2 , which is called draft coefficient. The material width changes from B to b
with a spread coefficient y/ = b/B. The material length increases from L to Lx, with a
elongation coefficient X = LxjL. The constant mass flow is given by
yx.B.L = y2.b.Lx (2-1)
The reduction ratio s is defined as
e=yi ^2xl00% (2-2)
In cold rolling, the strip width spread can be neglected as the width is m u c h larger than
the thickness, so the plastic deformation is mainly along the length and thickness
direction, and Eq. (2-1) can be simplified as:
yx.L = y2.Lx (2-3)
2.2.2 Material resistance
As two work rolls plastically deform the material, there exists a resistance to the
deformation of the rolled materials. The resistance of the material without tension i
usually determined as
P,
<r = -T (2-4)
where <y = resistance to deformation
Pt - rolling force
Ac = projected area of contact between roll and material.
The rolling force can be determined if the distribution of pressure px in the deformation
zone (Figure 2.2 and Figure 2.3) is known.
P = Bm.[ePx.dx = Bm.lR\Pe.dO (2-5)
where B„ = m e a n width of material
px = normal rolling pressure at distance x from the exit plane
pg - normal rolling pressure at an angle 6
Le = projected arc of contact between roll and material
a = roll bite angle
R' = deformed roll radius
Figure 2.2 Distribution of normal pressure and rolling force

\ Vi | h n ! h x
Figure 2.3 Parameters of deformation zone
Both entry and exit strip tension can reduce the rolling force. Therefore, in order to
correctly determine the resistance to deformation of the material rolled with tension,
(2-4) should be modified as follows [Ginzburg, 1989]:
a- + {Py°y^Py°y) (2-6)
A,
where (J ,& — entry and exit strip tension respectively
J3 , P — respective coefficient for entry and exit strip tension.
There are m a n y factors affecting the resistance to deformation in rolling. T h e main
parameters include:
• material chemical composition
• material metallurgical characteristics
• material temperature
• geometry of the deformation zone
3 0009 03286588 8
• friction in the deformation zone
• material work hardening prior to rolling
• strain rate of deformation.
It is not possible to derive a comprehensive analytical relationship between the
resistance to deformation and the parameters listed above. Practical solutions to the
problem are based on empirical approaches which include [Ginzburg, 1989]:
• laboratory non-rolling tests such as tension, compression and torsion tests
• tests on laboratory small-scale rolling mills, and on full scale rolling mills.
There are many methods of calculating the resistance to deformation for different
rolling conditions. They are given by Ekelund [1933], Siebel [1941], Orowan and
Pascoe [1946], Sims [1954], Ride [1960], Green and Wallace [1962], Ford and
Alexander [1963], Schultz and Smith [1965], Tselikov [1967], Wusatowski [1969],
Denton and Crane [1972], Yokoi et al [1981], Ginzburg [1985], Alexander et al.
[1987], and Swift [1940].
2.2.3 Rolling force and torque
With numerous methods to calculate resistance to deformation, there are many formulae
to calculate rolling force and torque. T o calculate the rolling force for flat products on
smooth roll barrels of equal diameter, a general equation is used, and based primarily
on the assumption [Roberts, 1965] that where deformation occurs in the roll bite, the
rolling pressure is equal to the resultant resistance to deformation of the material being
rolled. Under these circumstances, the rolling force is merely the product of the
projected area of contact and rolling pressure. Thus the rolling force is given by:
Pi=<ry-A=VyBm'Le (2-7)
In the presence of strip tension, Eq. (2-7) can be written as
P, = Bm {ay - f5yx ayi - [Syi ayi )VÂ7 (2-8)
where R'=deformed radius of work rolls
Ay = absolute reduction = (yx - v2)
The rolling torque is the sum of the torques required to drive both rolls. When
of equal diameter are used, the general equation for the rolling torque is give
Tr=2.Pra (2-9)
where Tr = rolling torque
a = lever arm as shown in Figure 2.2
The lever arm a is usually expressed as a fraction of the projected arc of con
a = m.Le = m^R'Ay (2-10)
where m = lever arm coefficient.
Defining the lever arm coefficient presents the most difficult part in the cal
the roll torque. From Eqs. (2-9) and (2-10), the lever arm coefficient is equal
m = ^— (2-11)
2.P,.Le
Formulae for force and torque in hot rolling include Sim's formula [1954], Cook
McCrum's formula [1958], Wright and Hope's formula [1975], Ford-Alexander's
formula [1963], Denton-Crane's formula [1972] and Green-Wallace's formula [1962]

In hot rolling, resistance to deformation depends mainly on the temperature of the rolled
material, strain, strain rate,frictionand its roll bite geometry. However, the principal
parameters affecting the resistance to deformation in cold rolling are material grade,
work hardening andfrictionin the roll bite. Roll flattening plays a m u c h more important
role in cold rolling due to a higher resistance to deformation. Also the effect of strip
tension becomes more significant as cold rolling is conducted with greater specific
tensions in comparison with those in hot rolling. These and some other features of cold
rolling process are usually taken into consideration to a different degree in the methods
for calculating rolling force and torque. Wusatowski [1969] proposed a method for cold
rolling by determining the resistance to deformation from the work-hardening curves.
Bland and Ford's general solution [Bland and Ford, 1948] for rolling force and torque is
based on Orowan's general theory of rolling [Orowan, 1943], considering friction,
tension and material yield stress variation in the roll bite. Graphical methods for
calculating rolling force and torque based on Bland and Ford's general solution was
proposed by Ford et al [1951], using a constant m e a n value of yield strength along the
roll gap. Based on a simplified analysis of deformation during rolling with dry slipping
friction, Stone [1953] developed a method for calculation of rolling force and torque
considering the effect of tension and flattening. Roberts [1978] derived empirical
equations for rolling force and torque calculation in temper rolling based on
experimental data. Alexander has written a Fortran program from Orowan's formulation
which solves V o n Karman's equation for rolling force and torque [Alexander et al,
1987].
According to the slab method, the brief Orowan's formula can be written as:
The horizontal force equilibrium for an element in the roll bite is given by:
%- = -2R (psinfi + q cos fly. (2-12)

dip
The deformation condition given by von Mises yield criterion under the plane st
condition can be used.
pcosfi + f/y = U5<ry (2-13)
Assuming the constrained yield stress satisfies Alexander's empirical relation [
<Ty=<Tj\ + C3e,)c*(\ + C5e,)c' (2-14)
Fleck et al. [1987, 1992] have derived an accurate model for rolling thin strip
both the plastic deformation in the strip and the elastic deformation in the ro
realistically accounted for. Dixon and Yuen [1995] extended this model for a non
constant yield stress, which occurs due to work hardening and temperature variat
the roll bite, and proposed a new approach in modelling the temper rolling proce
Pauskar et al. [1997] developed a microstructure dependent yield stress model f
accurate prediction of rolling force using the finite element method.
Following is a brief description of a simplified theory developed by Roberts [1
cold rolling process, using friction coefficient obtained in the laboratory rang
0.025 to 0.05 with cottonseed oil. With this theory, the rolling force is given
expression
[D^_+ jl)D | (p2nDPfy-s)

Pl = a.B. (2-15)
+(P
2 \ E 2Eyx(i-£)
where D = diameter of work rolls
<p = dimensionless constant of value 1.08

E- elastic modulus of the work roll
ju = effective friction coefficient in the roll bite.
In the bracket of Eq. (2-15), the first term is the length of contact for a perfectly
roll, the second term represents the lengthening of contact due to roll elastic flatt
and the third term accounts for the increased rolling force due to friction by the
mathematical conception of an increased length of contact.
In the absence of tensions and without any reduction being taken, the resistance to p
strain deformation of a material is 1.15 times the tensile yield strength cry\ namely,
cr = \.15cry (2-16)
During the rolling of most metals, work-hardening occurs, so that in the absence of s
tensions the resistance to deformation of the material increases progressively from e
to exit in the roll bite in accordance with the stress-strain curve. For the purpose
simplification, it is assumed that, for tensionless rolling, the "average" resistance
corresponds to half the reduction taken during rolling pass, i.e.
cr = lA5rjy(e/2) (2-17)
Considering the effect of tensile stresses ayi and cryi from entry and exit tensions, i
the reduction range between 20% to 50%, the resistance to deformation a in the roll
bite is given by,
a, +crv (l-s)
J
a = l.\5cryi - » , »\ (2-18)
ym (2 _£)
which is based on a simplification that an average tensile stress is the "average" te
force exerted on the strip divided by "average" thickness of the strip [Roberts, 1965
The torque exerted by each spindle without considering bearing losses is
a > (a -a >
T=DHa £ 1 + -^- + y\ yi (2-19)
(J
V a )
It is worth nothing that Eqs. [2-12, 13, 14, 15 and 16] are based on several simplified
assumptions. They can be used in the design of a new rolling mill and a determination
of rolling capability limits of existing mills, or used to determine the friction coeff
in the roll bite and the compressive yield strength of the strip under actual rolling
conditions. The accuracy of these equations can not satisfy the requirements of real tim
control. In general, the rolling force can be represented as a function of work roll
diameter, strip width, material chemical composition, metallurgical characteristics,
temperature, friction, work hardening, strain, strain rate, reduction, entry and exit
tensions, etc., which can not be determined by present rolling theory in a comprehensive
analytical equation. Therefore, in practice, the rolling force is measured by loadcells.
2.2.4 Friction coefficient and lubrication
Since the frictional force is the result of an interaction between contacting bodies at
their interface, the nature of friction cannot be understood without explaining the nat
of the interface. Modeling of the interface is one of the subjects of tribology which is
branch of science that studies friction, lubrication and wear of surfaces in relative
motion.
Friction coefficient p. is usually expressed by:
p = JL = IjL (2-20)
Pn Pn
where F = force required to move the body
P„ = normal force
r, = shear strength of the interface
pn = normal pressure.
Both rt and pn are obtained by dividing the corresponding forces F and Pn by the
apparent area of contact Ac between two bodies (Figure 2.4), i.e.:
F
*,= — (2-21)
A„
and
Pn=^r (2-22)
A„
The definition embodies Amonton's two basic friction laws: the frictional force is
proportional to normal force, and it is independent of the contact area. For a con
the interface shear stress r. must increase at the same rate as the interface pres
This relationship is valid for sliding friction, which is often referred to as Cou
friction. The condition of sliding friction is:
T", =/#./>„<*, (2-23)
where ks = shear yield strength of the interface.
When r, reaches the value of ks, it will take less energy for the material to shea
the body of the workpiece (Figure 2.4). This is described as sticking friction. The
condition of sticking is:
T,=/ii.pn>ks (2-24)
When this condition occurs, it assumes that the term friction coefficient is no longer
applicable since there is no relative sliding movement at the interface.
-—^ STICKING SLIDING

FRICTIDN X FRICTIDN
X >Tf
Figure 2.4 Frictional shear stress
The interface shear factor m is another non-dimensional quantity that has been
proposed by some researchers as an alternative to the friction coefficient. It is desc
by:
ri=m.ks (2-25)
where m = interface shear factor
The value of the interface shear factor varies from m = 0 for frictionless case to m =
for sticking friction.
Lubrication during cold rolling reduces the rolling loads and helps to obtain good
surface quality by reducing the friction coefficient, wear and staining. Lubrication an
friction phenomena in metal forming have long been studied for the fundamental reason
that they affect the working force required, product surface quality and tool life etc
has been realized that friction in the lubricated forming process derives from viscous
shear of the hydrodynamic film of the lubricant and from shearing of the real contact
with or without the boundary films [Kasuga and Yamaguchi, 1968]. Kasuga and
Yamaguchi [1968] proposed the concept of the real contact ratio and found that it is
mainly the ratio of lubricant film thickness to the combined roughness of the tool and
the workpiece. The amount of lubricant carried into the contact interface has been
calculated by Mizuno based on a hydrodynamic theory [Mizuno, 1966]. Mizuno showed
that the thickness of the lubricant film is proportional to the average velocity of the
lubricant at the inlet to the contact zone. This w a s revised by Wilson and Mahdavian
[1974], w h o considered the thermal effects due to viscous shear, and by Sutcliffe and
Johnson [1990], w h o considered the influence of the surface roughness. O n the other
hand, the viscous shear stress of the lubricant film is determined by the lubricant
properties at high pressure, the film thickness, and the relative sliding velocity. The
relative sliding velocity especially affects the lubricant properties and the lubricant film
thickness [Wang et al, 1995]. This suggests that an average velocity of the lubricant at
the contact zone inlet and the relative sliding velocity at the contact interface affects the
friction behavior of the lubricated forming process. In practice, cold strip rolling
operation runs in the mixed lubrication regime, in which a part of the total interface
pressure is provided by asperity contact at surface peaks and a part by the pressurized
lubricant in the surface valleys. Wilson and Chang [1994], Chang et al [1996], and Qiu
et al [1999] developed mathematical model for strip rolling under mixed lubrication.
In actual rolling process, there are different types of lubrication: boundary lubricati
mixed film lubrication, elastohydrodynamic lubrication ( E H D ) or plastohydrodynamic
lubrication ( P H D ) and fully hydrodynamic lubrication shown as in Figure 2.5. In most
cold rolling process, the lubrication regime is under the mixed film lubrication
condition.
Boundary
c
o
\ Mixed-film
c
a> \ Hydrodynamic
'o
\ EHD ^—~
£
o
o
Sommerfeld No.
Figure 2.5 The lubrication regimes
The Sommerfeld is expressed as —.
where
^=lubricant viscosity, Pa.S
uw =strip speed at exit, m/s
cr =yield strength of workpiece, Pa
v2 =strip thickness at exit, m
2.3 Friction coefficient measurement
The ratio of the interfacial frictional stress to normal pressure is defined as fric
coefficient. Friction at the interface between rolls and strip being plastically defo
by the rolls is one of the most important considerations in both theory and practice o
plastic working. However, the nature of friction at the strip-roll interface is not c
understood. In modern steel industry, a mathematical model of cold strip rolling mus
able to predict the pressure distribution in the roll gap, the separating force and the roll
torque with accuracy and consistency. M a n y factors influence on the accuracy of
mathematical model in cold rolling, but one of the most important factors is friction
coefficient. This is because incorrect friction coefficient value will affect the accuracy
of the mathematical model and consequently, the thickness and shape of the strip.
Hence, it is important to determine the correct value of friction coefficient in cold
rolling.
2.3.1 Sensor roll method
The sensor roll method [Liu and Tieu et al, 2001] is a direct method to measure frictio
coefficient distribution in the roll bite. T w o pins with sensors are embedded in the roll at
two different angles, one is in the radial direction and the other at a certain angle to the
radial direction, to measure the local force in the roll bite respectively, as illustrated in
Figure 2.6.
Figure 2.6 Radial and oblique pins in roll bite
The friction coefficient is calculated from the following formula [Rooyen and Backofen
1957]
M = 'tan0 (2-26)
where p = friction coefficient
pe - oblique pin pressure
pr = radial pin pressure
2.3.2 Strip marking method
Strip marking method is an indirect method to measure the forward slip which will
the average friction coefficient in the roll bite. After the roll turns one revolu
line markings on the roll left an image on the strip, as illustrated in Figure 2.7
A By
-,
/
L*
Figure 2.7 Marks on the strip (refer to Fig. 4.2)
The average forward slip over one revolution can be calculated from Eq. (2-27).
V-Ln
S,=- x!00% (2-27)
where Sf= forward slip
Ln = circumference of the roll
L'=marking length left on the strip after the roll turns one revolution
Thefrictioncoefficient can be obtained from the forward slip by the following equation
[Ford et al, 1951].
*(A
(f>x (f)A*\ (2-28)
y-i 2 Ap
2.3.3 Laser Doppler method
In the Laser Doppler method, two LDV probes [Tieu et al, 1998] are used to measur
the roll and strip speed respectively, as shown in Figure 2.8.
Figure 2.8 Laser Doppler method [Tieu et al, 1998]
The forward slip can be calculated by the following formula:
« w -ur (2-29)
Sf=~^ -xl00%
where Sf = forward slip
uw = exit speed of strip
ur = roll speed
The friction coefficient will be determined by Eq. (2-28)

2.4 Lubrication theory in cold rolling
Lubricant is used in most cold rolling operations. These lubricants are usually in the
form of fatty or minerals oil, applied either neat or in emulsion form. Lubricants
facilitate the reduction of strip by:
• reducing the rolling forces required for deformations, resulting in a lower
energy expenditure;
• production of high quality surface, resulting in a higher value added product;
• reducing roll wear and decreasing the need for frequent roll changes;
• reducing strip and roll temperatures.
Roberts [1978] stated that frictional effects in the roll bite appear to be associated
boundary or thin film lubrication at low rolling speeds-but with hydrodynamic
lubrication at high rolling speeds. Initially, it was believed that boundary lubrication
was prevalent under virtually all rolling conditions and that a constant friction
coefficient could be used to characterize it. The realization that the viscosity of the
lubricant and the rolling speed profoundly influenced thefrictionalconditions in the roll
bite, led to the belief that hydrodynamic effects were also present. But other authors also
believe that the rolling operation runs in the mixed lubrication at lower rolling speed,
and in fully hydrodynamic lubrication at higher rolling speed.
2.4.1 Hydrodynamic lubrication
There are two types of hydrodynamic lubrication:
• Elastohydrodynamic (EHD); and
• Plastohydrodynamic (PHD).
E H D lubrication recognizes that viscosity of the lubricant increases with pressure and
that the two contacting surfaces deform elastically.
PHD lubrication is an extension of EHD lubrication to the case when the workpiece
material is subject to plastic deformation.
Hydrodynamic lubrication in cold rolling processes has its advantages in the production
of metal sheet and strip. Since the thickness of the hydrodynamic film increases with
rolling speed, future high speed mills will operate in a hydrodynamic regime. A thick
lubricant film is thus developed and asperity contact and friction between the strip and
roll is significantly reduced.
In ordinary rolling process, there must be sufficient friction to ensure the workpiece t
enter the roll gap at the beginning of rolling and, once rolling is under way, to ensure
freedom from excessive slip. Continuous slip makes control of the rolling process
extremely difficult.
Making use of highly viscous oil can generate a thick hydrodynamic film. A theory used
to predict the film thickness and the friction resulting from hydrodynamic lubrication is
useful in assessing the performance under the given operating conditions, von K a r m a n
[1925] developed the earliest method to account for friction in the rolling process. In his
analysis the lubrication process was characterized by the Coulomb model which
assumes that frictional stress is proportional to interface pressure. This proportionality is
assumed to be dependent on material property of roll, workpiece and lubricant. Cheng
[1966] published a theory of plastohydrodynamic lubrication but in this theory the film
thickness calculations are based on E H D theory. Bedi and Hiller [1967-1968] attempted
to conduct analyses which integrate the mechanics of the hydrodynamic lubrication and
plasticity processes. They devised a method of calculating the film thickness by
assuming that the power dissipated in the work zone is a minimum. Avitzur and
Grossman [1972], Wilson and Walowit [1971] developed a model to calculate lubricant
film thickness. D o w et al [1975] extended Wilson and Walowit's work to take into
account the influence of temperature on the film formation process.
2.4.2 Mixed film lubrication
In cold rolling the lubrication system must be carefully designed to provide adequate
lubricant film thickness, thus avoiding problems of severe metal-to-metal contact. O n
the other hand, if the film thickness is too high, poor surface quality m a y result from
unconstrained grain deformation. In addition, the friction between strip and rolls must
be sufficient to draw the strip through the rolls and yet not so high as to cause excessive
roll separating forces. These conflicting constraints are typical of those imposed on
metal working lubrication systems and are the reasons w h y it is often more difficult to
develop a successful lubrication system for metal forming system than for conventional
elastic machine elements.
As in many metal forming operations, the requirements imposed on the lubrication
system in a cold rolling process generally require that it operates in the mixed
lubrication regime. In this regime, the surface loading is shared between the pressurized
bulk lubricant film in roughness valleys and the boundary films separating asperity
peaks. Thus, both the mechanics and chemistry of lubrication must be taken into
account in designing the lubrication system.
An analysis of the contact between the two surfaces in the mixed lubrication regime is
complicated. Friction is determined by the properties of the lubricant film and the
interaction of asperities. The thickness of a lubricant film is influenced by the shape of
the contact, the speed of the running surfaces, the contact pressure and temperature. The
asperity interaction is determined by the film thickness, combined roughness, variation
of contact pressure and shear stress, direction of roughness, surface chemistry and the
hardness of asperities.
In order to model friction in the mixed lubrication regime, it is necessary to calcula
behaviour of asperities in the roll bite in combination with lubricant flow. Models
regarding mixed lubrication in cold rolling have been developed by Tsao and Sargent
[1975], Sheu [1985], Sutcliff [1989], Chang et al [1996] and Qiu et al [1999].
2.4.3 Effects of roughness
In the mixed lubrication regime, some of the load at the interface is supported by the
asperities in contact between the roll and workpiece, and some by the pressurized fluid
in the valleys as shown in Figure 2.9. T o model this regime, a relationship between the
surface topography, geometry, kinematics and lubricant rheological properties and the
pressure generated in the valleys must be developed.

Figure 2.9 Asperity contact in the mixed lubrication regime
The surface roughness can be longitudinal, isotropic or transverse as seen in Figure
r'M r' = l r'<\
Figure 2.10 Longitudinal, isotropic and transverse surface roughness
f is the length-to-width ratio of a representative asperity [Ginzburg, 1989].
The first work on modelling surface roughness was by Tzeng and Saibel [1967] w h o
introduced stochastic concepts in the study of random surface roughness. This dealt
with one-dimensional transverse roughness only. This work was extended by
Christensen and Tonder [1971, 1972, 1973] to determine the stochastic Reynolds
equation for transverse and longitudinal roughness.
The methods by Tzeng and Saibel [1967] and Christensen and Tonder [1971, 1972,
1973] were limited to the two specific roughness structures (transverse and
longitudinal), and were difficult to extend to three-dimensional surface roughness w
the existing stochastic theory.
Patir and Cheng [1978] introduced a new method of deriving the average Reynolds
equation using a flow simulation method. This method could also be extended to the
mixed lubrication regime where the efffect of roughness is important. For a steady-o
dimensional problem, the form proposed by Patir and Cheng [1978] is reduced to:
f ti dp. ux + u2 dh, ux-u2 d(f>s

d_ I "I | s
dx
<f>: 12/7 dx 2 dx 2 dx
where ^>X\<PS= pressure and shear flow factors, respectively
2
S = R M S composite roughness, ^Sr + 5s
The flow factors incorporate the roughness effects of the material into the Reynolds
equation. However, Patir and Cheng's formulation is unsuitable for conditions of hig
fractional contact area (A>0.5) which occurs in many bulk metal forming processes
[Wilson and Chang, 1996; Wilson and Marsault, 1998]. This model also becomes
difficult to use when the mean lubricant film thickness is much smaller than the sur
roughness.
Sheu and Wilson [1983] used a simple upper-bound technique to investigate workpiece
asperity flattening or surface indentation in the presence of plane strain plastic
deformation. This lead to further work in 1988 which found that the bulk plastic flow of
the workpiece tends to reduce the effective hardness of the asperities [Wilson and Sheu,
1988; Wilson and Chang, 1996]. Wilson and Sheu [1988] also found that the rate of
asperity flattening with bulk straining was related to the spacing of the asperities and to
the difference in pressure between 'loaded' and 'unloaded' roll and strip.
Tripp [1983] extended Patir and Cheng's work [1978] to define a percolation limit. Thi
is the point where thefractionof contact area is such that no open paths exist for the
lubricant flow. H e treated the interface like a porous medium model or random network.
Sutcliff and Johnson [1990] analyzed the lubrication process in the inlet region of th
roll bite to determine the hydrodynamic build up of oil pressure. They incorporated the
roughness model of Christensen [1969-1970].
Lo [1994] combined the porous medium model and percolation theory derived by Tripp
[1983] with the flow factor method derived by Patir and Cheng [1978] to describe the
lubricant flow in the mixed lubrication regime. This method avoided some of the
problems encountered by Patir and Cheng [1978] and can be used for high fractional
contact area (A>0.5).
Wilson and Marsault [1998] compared pressure flow factors for longitudinal surfaces
with varying height distribution. They compared three c o m m o n surface roughnesses:
1. Christensen;
2. S a w tooth; and
3. Sinusoidal
Wilson and Marsault [1998] stated that the mathematical expression
0x'=3.46Ht-x (2-31)
derived by Wilson and Chang [1994] is an excellent approximation for both the
Christensen and sinusoidal surfaces under contact conditions. Wilson and Marsault
[1998] also derived the pressure and shear flow factors for materials with transverse
lays.
2.5 Current research on friction coefficient measurement
There are several ways to describe the friction coefficient which include the Amonton
laws where friction coefficient is a constant value or the statement that r. = mks.
Constant friction coefficient can be calculated from the forward slip which can be
measured from the strip marking method or the Laser Doppler method [Liu et al, 1999;
Lenard, 1992].
While both of these approaches lead to reasonable predictions of process variables th
are, strickly speaking, incorrect. Evidence exists in the literature showing the
inapplicability of either method-see for example [Rooyen and Backofen, 1957; Banerji
and Rice, 1972; Al-Salehi et al, 1973]. A limited number of cold rolling tests were
reported in these references, all showing that the ratio offrictionalshear stress to normal
stress is indeed not constant in the roll bite, confirming the suggestion by S h a w et al.
[1960] that the ratio varies along the contact zone for relatively high interfacial normal
stresses.
A number of studies were undertaken by Lim and Lenard [1984], Lenard [1991], and
H u m et al [1996] to measure the friction coefficient along the roll bite in both cold and
hot rolling by using embedded pins method. A sensor method was also used by Jeswiet
and Rice [1989], N y a h u m w a and Jeswiet [1991] and so on to measure that coefficient.
All of those works confirmed that the rjp ratio is in fact varying from entry to exit in
the roll bite.
The accuracy of measurements obtained by the embedded pins technique has been
questioned by Stephenson [1983], suggesting that a possible 5 0 % error in frictional
stresses m a y be unavoidable. It is undeniable that interrupting the contact surface,
weakening of the roll and thereby changing its m o d e of deformation, using pins of finite
thickness, coping with the metal extruded into the clearance between the pin and its
housing and accounting for frictional resistance there, must affect the readings. Another
drawback of the embedded pins technique is its unlikely use in production mills making
it impossible to substantiate directly experimental results in full-scale plant trials.
However, since no other method capable of yielding directly the values of the interfacial
stresses in the roll gap during strip rolling, the embedded pins method [Lenard, 1991]
provides a means to compare roll separating forces measured by the force transducers to
the integral of the roll pressure distribution produced by the pin technique over the
contact surface. The results indicated that the difference between the roll measured and
calculated separating forces is not larger than 1 7 % . S o m e confidence in the
experimental roll pressure distributions m a y therefore be restored.

Chapter 2 Rollim* Theory and Lubrication in Cold Rolline 34
2.6 Current research on lubrication model in cold rolling
Most modern rolling mills rely on elaborate computer-based models to develop rolling
schedules and to achieve proper gauge and flatness control. This has an important
influence on productivity, product quality and scrap reduction. The method used widely
in the industry for designing rolling mill is the traditional slab methods developed by
V o n Karman, Nadai, Orowan, Bland and Ford, Hill and Sims etc.. The accuracy of the
prediction of the pressure distribution by this slab method is affected by the presence of
inhomogeneous deformation. Accurate determination of metal flow in forming
processes under realistic conditions became possible w h e n thefinite-elementmethod
( F E M ) was introduced, and numerical models of the mixed film lubrication process in
cold strip rolling developed.
The cold rolling process usually operates either in the mixed lubrication regime for l
speed rolling process [Chang et al, 1996; Qiu et al, 1999] or in fully hydrodynamic
regime at high speed [Lin and Houng, 1991; Lugtetal, 1993; Saxena et al, 1996].
In fully hydrodynamic rolling, the surfaces in contact are completely separated by a t
film of lubricant. The models was developed by Lugt et al. [1993] to describe the
configuration of a rigid, perfectly plastic sheet rolled by a rigid work roll. The Barus'
[1893] viscosity-pressure relations have been applied. The governing equations have
been solved throughout the complete contact area, i.e. the inlet, the plastic work zone
and the outlet zone. A few years later, elastic deformation of the surfaces of rolls and
strip has been fully incorporated in the model by Lugt and Napel [1995]. Both
Roelands' [1966] and Barus' viscosity-pressure relations have been applied. The
thermal effects regarding heat development caused by plastic deformation as well as

Chapter 2 Rolling Theory and Lubrication in Cold Rolling 35_
work hardening have been included. In Lin and Houng's work [1991], a refined
computational model is developed to deal with a series of governing equations
associated with the three lubricating areas simultaneously. Accordingly, the three major
zones are not treated separately but as an integrated unit. The thermal effects on rolling
performance are included in three zones: inlet zone, plastic work zone and outlet zone.
But in practice, most cold rolling usually operates in the mixed lubrication regime,
where the film is not thick enough to completely separate the rolling surfaces and solid
contact occurs at the tops of the surface asperities. The reason is that the theoretical film
thickness is operating normally at sub-micron in industrial rolling mills. However, such
a thin film will not completely separate the rolling surfaces due to inevitable surface
roughness. Significant progress has been m a d e in the analysis of this lubrication regime
recently. Sutcliff and Johnson [1990] carried out an inlet analysis, in which the
deformation pressure is shared between the contact asperities and fluid film at the
surface valleys. F r o m the asperity crushing rate, the fractional contact area and the
average film thickness were obtained. The average film pressure was calculated by
integrating a simplifiedfirst-orderReynolds equation, with two arbitrary constants to be
determined. Wilson and Chang [Wilson and Chang, 1994, 1996] developed an
analytical model for strip rolling. The sheet surface roughness was approximated by a
longitudinal sawtooth topography and evolution of surface roughness was estimated,
with the relationship between the fractional contact area, average deformation pressure
and the film pressure derived by using an upper bound theory [Wilson and Sheu, 1988].
The film pressure pf was determined from a simplifiedfirst-orderReynolds equation
(^x(ht3/l2TjJdpf/dx)=-((ur+uJ/2)ht +C') with a flow constant C. which was
determined by a trial and error method. A similar model was also proposed by Qiu et al.
[1999], but with a more rigorous second-order Reynolds equation, hence eliminating the
need to introduce an unknown flow constant. The convergent solution can be obtained
for normal rolling speeds ( 0 < w,. < 1 8 mis).
2.7 Problems and expectation
In modern steel industry, more accurate strip gauge and flatness are requested by the
customers. It is possible to achieve this target with the development of modern
computer technology. A lot of efforts have been m a d e to develop mathematical model
in cold rolling.
Friction coefficient is one of many factors affecting the accuracy of mathematical mode
in cold rolling. In traditional cold rolling model,frictioncoefficient in the roll bite is
assumed to be constant, which strictly speaking is not correct. S o m e of previous
research works shown that friction coefficient varies from entry to exit along the roll
bite. In Rooyen & Backofen's [1957] work, several problems occurred, (i) The elastic
distortion of the roll caused forces to be impressed on the pins which amounted to about
5 - 1 0 % of the values obtained w h e n the pin did m a k e contact; (ii) N o experiment was
made to establish experimentally the starting point of entrance and exit; (iii) Friction
coefficient ratio is formulated with measurements at two different locations; (iv) The
assumption that the pin is less stiff than the insert is supported by experimental results.
However, no absoluted proof is available, (v) The pin indentation has been obtained in
experiments. It was about 7.62pm deep, (vi) A s seen from the calibration curves for the
oblique pin in the roll, hysteresis is found as the load begins to decrease from
m a x i m u m . Such situation m a y exist in the area of the neutral point. In Banerji & Rice's
[1972] work, it was found out that friction coefficient varied widely throughout the roll
gap but the location of a single pressure peak did not coincide with the location of the
neutral point. In the following work [Jeswiet & Rice, 1982; Britten & Jeswiet, 1986;
N y a h u m w a & Jeswiet, 1991], it was also found that the measured torque differs from
the one calculated from measured friction coefficient. O n e pressure peak, two pressure
peaks and three pressure peaks were also found by Al-Salehi et al [1973] for the
different rolling materials and rolling conditions. In L i m and Lenard's [1984] design,
radial and oblique pins do not enter the roll gap at the same time. In other word, both
pins are not at the same axial line, and therefore the pin results m a y not correlate with
each other.
In our new design of sensor roll with embedded pins, some problems were overcome.
At the beginning, two pressure peaks were found. But later these problems were
overcome after a number of modifications were m a d e to the sensor roll. The details will
be illustrated in chapter 4.
Since 1970's, tribology has been applied in cold rolling. Reynolds equation, surface
roughness theory combined with traditional rolling theory provides a more sophisticated
mathematical rolling model. Actually, most of rolling processes operate in mixed film
condition. The previous mixed film model [Chang et al, 1996; Qiu et al, 1999] only
considered plastic deformation area, and did not cover the effect of inlet and outlet area
as well as thermal effect throughout the whole roll bite. Actually, these factors,
especially the thermal effect have the influence on the accuracy of mathematical model.
The modified profile of the strip caused by elastic deformation in the inlet must be also
considered.
2.8 Summary
Firstly, the basic concept of rolling theory is introduced in this chapter, and then the
friction coefficient and lubrication in rolling are briefly reviewed. Since the friction has
an effect on the accuracy of mathematical model, and the product quality, it has
attracted m a n y researchers w h o have obtained advanced knowledge of friction in
rolling. Three methods to determinefrictioncoefficient are illustrated in this chapter,
with the emphasis on the sensor method to measure friction coefficient variation in the
roll bite. Existing problems in measuring friction coefficient are also discussed. A
simple suggestion to solve these problems is given, and more details will be explained
in chapter 4.
As for the consideration of lubrication theory in cold rolling, hydrodynamic and mixed
film lubrication are reviewed, and the effect of roughness is also discussed. The
deficiency of existing mixed film models are also highlighted, and the model accuracy
will be improved by considering inlet and outlet effect as well as the thermal effects in
chapters 7 & 8.
Review of the current research on friction measurement and lubrication model in cold
rolling was also carried out.

Chapter 3 Literature Survey 39
Chapter 3
Literature Survey on Friction
Measurement and Lubrication Model
in Cold Rolling
3.1 Introduction
The friction coefficient in the roll bite can be measured by the embedded pin-transducer
technique. It was originally suggested by Siebel and Lueg [1933] in the rolling process
and adapted by van Rooyen and Backofen [1957] and Al-Salehi et al [1973]. The
method has been applied to measure interfacial stresses in several bulk forming
processes. Cold and hot rolling were studied [Lim and Lenard, 1984; Karagiozis and
Lenard, 1985; Lenard and Malinowski, 1993]. The variations have been presented by
Lenard [1991] and Y o n e y a m a and Hatamura [1987, 1989]. Since the major criticism
concerns the possibility of some metal intruding into the clearance between the pins and
their housing [Stephenson, 1983], it is necessary to substantiate the results by
independent means. This substantiation has been performed successfully by H u m et al,
[1996]. A strain gauged cantilever with its tip in the contact zone and its refinements
were presented by Banerji and Rice [1972], N y a h u m w a and Jeswiet [1991] and Jeswiet
et al. [2000]. Detailed information on the distributions of interfacialfrictionalshear
stresses and die pressures m a y be obtained by these methods, but the setup and the data
acquisition were elaborate and costly.
Most rolling models rely on computer-based models to develop rolling schedules and to
achieve proper gauge and flatness control. M u c h efforts has been m a d e to study the
models in cold rolling process. The lubrication system in cold rolling production
generally requires that it operates in the mixed lubrication regime. In this regime, the
interface loading is shared between the pressurized bulk lubricant film in roughness
valleys and the boundary films separating asperity peaks. The mechanics of rough
surface lubrication in bulk forming processes such as rolling is complicated by the high
fractional contact areas, which often exceed 9 0 % in production processes. Under these
conditions the traditional methods such as the average Reynolds equation proposed by
Patir and Cheng [1978] will not work. The earliest model of mixed lubrication in cold
rolling was developed by Sargent and Tsao [1980]. The most serious deficiency of their
model was that it failed to take account of the influence of the bulk plastic deformation
of the strip on asperity deformation [Wilson and Sheu, 1988; Sutcliff, 1988; Korzekwa
et al, 1992]. Sutcliffe and Johnson [1990] and Sheu and Wilson [1994] developed
analyses, which allow for this. All these models treat relatively high speed rolling
Chapter 3 Literature Survey 4J_
conditions where the hydrodynamic pressure builds up in the inlet and transition zone
and the pressure gradients in the work zone have a negligible effect on lubricant flow.
Alternative methods better suited for metal forming at low rolling speed were provided
by Chang et al. [1996], Wilson and Chang [1994, 1996], and Qiu et al. [1999]. The
thermal effects were also considered by Wilson and Mahdavian [1974], and Lin and
Houng [1991] for fully hydrodynamic lubrication conditions. But the effects of
hydrodynamic inlet zone and outlet zone as well as thermal factor in mixed film
condition have not been considered until now.
In both rolling theory and practice, two important factors must be considered: friction
coefficient and resistance to deformation. The traditional approach tofrictionis to
assume that thefrictionalforce in the roll bite is proportional to the normal force, and
the friction coefficient is constant in the roll bite. But this will incur a loss of accuracy
in the roll gap model, and affect the thickness and shape of the strip. In order to
understand the friction mechanism in cold rolling, m a n y research efforts have been
m a d e in both experiments and theoretical modelling.
Two methods, direct and indirect methods have been used in the measurement of
friction coefficient. In the direct method, the sensors are embedded in the roll to
measure forces, which are used to determine the friction coefficient variation along the
roll bite. In the indirect method, forward slip isfirstmeasured, and then used to
calculate the friction coefficient. Another indirect method is the "inverse method" which
will be discussed in details at the end of this section.

In the direct method, the embedded pin-transducer technique in which a small pin is
fitted radial into the roll and ending flush with the roll surface, makes contact with some
type of load transducer which provides a measure of the load on the pin. It was
originally used by Siebel and Lueg [1933] for the determination of normal pressure
distribution along the arc of contact, and then adapted by van Rooyen and Backofen
[1959]. In this method, the two pin-transducers were embedded in the work roll of
(j>152.4 m m diameter to measure normal pressure and interfacial shear stresses.
Aluminum strips were rolled with no lubrication and the ratios of surface shear stress to
normal stresses were plotted for a sand blasted roll and ground roll with a roughness of
1.52 p m and 0.28 p m R M S respectively. The measurements were also repeated with
"rolling oil" in the roll gap at reductions of 5 1 % and 5 4 % . The rolling speed was 0.127
m/s. T o explore the friction coefficient at the entrance as well as the exit side of the
neutral point, tests were m a d e by rolling in both the forward and reverse direction. With
a 5 0 % reduction, the horizontal projection of the contact arc was approximately 10
times the diameter of the pins at the roll surface. Smith et al. [1952] have shown that,
with such geometry, pressure correction for the finite size of the pin is less than 2 % ,
except at the roll gap entrance, exit and the neutral point at which the point of entry
could be recognized visually as a clearly marked line on the partly rolled specimens. In
the calibration, side loads of 0kg, 4kg, 8kg and 12kg, respectively, were applied to the
radial pin. The unloading curves were coincident with each other in all cases.
Successive calibrations of the oblique and radial pin without exception were
reproducible within ± 1%. Such finding was taken to m e a n that the friction coefficient
between the pin and insert could be established with a reasonable confidence, remained
unchanged even after rolling a large number of strips. Furthermore, there were no
visible signs of metal having been extruded up into the annular space between the pins
and inserts, which would be expected to be a source of some binding action. It was
found that the elastic distortion of the roll caused rolling forces to be impressed on the
pins which amounted to about 5 - 1 0 % of the values obtained w h e n the pins did m a k e
contact. The same finding has been reported by Smith et al, [1952]. The record
obtained in grooved-strip rolling w a s similar for the two pins, so that the ratio of the
stress in the oblique pin to that in the radial pin, and resulting value of friction
coefficient p, is not likely to be greatly affected by whatever causes this response. The
results showed a significant variation of thefrictioncoefficient w h e n the sand blasted
roll was used or w h e n the ground roll was lubricated. Using the ground roll without
lubrication resulted in africtioncoefficient that was essentially constant on either side
of the neutral point. In all three cases presented by the authors, the location of zero
interfacial shear stress appeared to coincide with the location of the m a x i m u m rolling
pressure. Moreover, single and smooth pressure peaks were reported. Rabinowicz
[1965] points out the importance of removing all surface layers from the contacting
surfaces by the use of strong caustic soda solutions. Five percent N a O H solution was
used to clean both rolls and strips; further, the roll surfaces were washed and cleaned
with carbon tetrachoride. But some problems were found with Rooyen and Backofen's
method [1957]. Firstly, the friction coefficient ratio was formulated with measurements
at two different locations along rolling direction; Secondly, no proof was available for
the assumption that experimental results indicates the pin to be less stiff than the insert;
Thirdly, as seen from the calibration curve for oblique pin in the roll, hysteresis was
found w h e n the load was decreased from m a x i m u m . Such situation might exist in the
area of neutral point; Fourthly, the ratio of oblique-to-radial pins pressure, which is used
to calculate the friction coefficient, cannot be directly measured, since correction factors
must be applied to raw outputs to account for finite pin widths, and friction between the
pins and inserts. A n y errors in the corrected ratio are greatly magnified by the
calculation process; Finally, a 1 % error in the pressure ratio could produce 6 % ~ 3 0 % in
the calculatedfrictioncoefficient [Stephenson, 1983]. Despite these problems, Rooyen
and Backofen's experiments [1957] still produced some reasonable experimental results.
Al-Salehi et al. [1973] presented results obtained by rolling aluminium, copper and mi
steel strips and using technique similar to that of Rooyen and Backofen [1957]. Normal
pressure, interfacial shearing stress andfrictioncoefficient were presented for 14.17%
and 34.41%) reduction of aluminium strips; for 14.4% and 17.2% reduction of copper;
and for 7.3% reduction of mild steel. T w o rolls of <j>158.75 m m diameter were used with
average surface roughness between 0.20 and 0.37 p m C L A . The m a x i m u m rolling
speed is 1.524 m/s. In all five instances the variation of friction coefficient in the roll
bite was very pronounced. T w o pressure m a x i m a were recorded during 14.17%
reduction of aluminium and three pressure maxima were recorded during 34.41%
reduction of the same materials. A single pressure peak was evident in the other three
experiments. A very substantial correction factor was needed for all the experiments
owing to the rather large ratio of pin width to arc of contact length in the roll bite. But
this is unnecessary if the reduction or roll diameter is the same as those used by Rooyen
and Backofen [1957].
Lim and Lenard [1984] designed another sensor roll with pin-transducers embedded in
the roll without insert support. The radial and oblique pins were located at the different
location along the rolling direction. The experiments were carried out on the rolling mill
with <j>254 m m diameter. The rolls were hardened to R c = 5 6 and ground to a finish of
0.20 p m C L A . The two different kinds of aluminium alloy were used in the
experiments. The reductions are between 3.14% and 22.35%. The experimental rolling
speeds are from 0.00465 m/s to 0.1107 m/s. N o special cleaning of the rolls or the strips
was done during the testing program, in spite of the comments of Rabinowicz [1965];
the purpose was to achieve practical industrial conditions. N o lubricants were used in
any of the experiments. A n e w method to calculate the friction coefficient was used. A n
assumption was m a d e that the average coefficients between the radial and oblique pins
and their respective holes are the same. Forward and reverse rolling were performed
each time for the determination offrictioncoefficient. But no calibration details were
given. Thefinalexperimental results revealed: (1) The ratio of the shear stress to normal
pressure in the roll bite does not remain constant during cold rolling; (2) The average
friction coefficient decreases with increasing rolling speed; a finding that supports the
general observations of Rabinowicz [1965]; (3) A controversy regarding the existence
of multiple pressure maxima, the two pressure peaks were found for the experiment
with 3.14% reduction and 0.038 m/s rolling speed. But smooth rolling pressure curves
with one maxima were found in most of the experimental results.
In few years later, a new design was made by Lenard [1991], This time, the radial and
oblique pins locate the same axial line. In another word, both pins touch the strip at the
same time. The <|>250 m m diameter rolls hardened to R c = 4 8 and ground, were used in a
two-high rolling mill. The specimen of aluminium alloy and carbon steel were used in
the experiments. The ranges of rolling speed and reduction are 7.5% ~ 25.6% for 1100-
H14, 7.8% ~ 2 0 % for 5052-H34 and 8.3% ~ 11.5% for AISI 1005 respectively.
Although multiple pressure peaks were obtained in the experiments [Al-Salehi et al,
1973; Lim and Lenard, 1984] and predicted by Li and Kobayashi [1982] theoretically,
Lenard [1991] still has doubts about the multiple pressure peaks since not all the double
or triple peaks were reproducible in the experiments. Finally, the difference between the
measured force and calculated force based on the measured friction coefficient were
below 1 7 % and in most of cases, below 1 0 % .
The multiple pressure peaks were also reported by Lagergren [1997]. This was made
possible by a specially designed conical load transducer with its housing drilled into the
bottom roll. The load transmitting pin had a diameter of 1.7 m m . The pin protruded into
the model material during rolling about 50 p m , for m a x i m u m reduction of 4 5 % . The
surface of the transducer pin measuring the load was felt by hand to be flush with the
surface of the roll. The calibration of the transducers was done by accurate normal and
tangential loading. But the specimens were soft or hard wax. This is very different with
real rolling situation.
Another new design was made by Hum et al. [1996]. The two-high rolling mill had rolls
of cj)250mm diameter, hardened to R e 52, and ground to a surfacefinish0.18pm. The
roll pressures and the interfacial shear stresses were measured by four pin-transducers
combinations. The pins are 1.8mm of diameter and hardened to R e 55. In the current
set, two more pin-transducers were added. O n e of the n e w ones was in the radial
direction and w h e n it and other radial pin gave readings that are close, the likelihood of
a successful experiment increases. The n e w pin was in an oblique direction, such that
two oblique pins eliminated the need for two runs each time. The n e w equipment was
used for the study of hot rolling of aluminium strips. In each case the peak pressures
were found to be close to the exit region, not near the locations of the neutral points. In
those cases the two locations do not coincide, indicating that mathematical models using
the friction hill theory m a y not yield accurate predictions of rolling parameters.
Moreover, the rolling pressure predicted by the traditional theories, resulted in the well-
k n o w n but highly unrealistic sharp peaks, but they were not supported by the
experimental data. Reasonable friction coefficients were measured. The forward slip
and the friction coefficient were both dependent on the rolling speed. The forward slip
increases and the friction coefficient, in general, decreases with rolling speed.
The investigation of the axial pin position above the roll surface was presented by
Plancak et al [1996] in the simple upsetting of cylindrical specimen. The radial
clearance between the pin and hole is 0.04 m m . The experimental results showed that
the most realistic measured values of normal pressure were obtained for the case w h e n
the pin head protruded from the die surface prior to deformation by 0.15 m m . If the pin
head was flush with or below the die surface it gave pressure values which were too
low, whereas w h e n the pin protruded 0.25 m m above the die surface it produced
high pressure measurements.
A strain gauged cantilever with its tip in the contact zone and its refinements were
presented by Banerji and Rice [1972], N y a h u m w a and Jeswiet [1991], Jeswiet [1995],
and Jeswiet et al [2000]. Detail information on the distributions of interfacial frictional
shear stresses and die pressures m a y be obtained by these methods, but the setup and the
data acquisition were elaborate and costly. Limited results were presented. It has been
shown that the peak normal stress did not coincide with the theoretical friction hill for
some cases.
The average friction coefficient can also be measured by indirect method [Liu et al,
1999]. In this method, thefrictioncoefficient is calculated from the forward slip which
can be measured either by the strip marking method using the marks left on the strip
before and after rolling or the Laser Doppler method which measures the roll speed and
the exit speed of the strip. In Liu et al [1999] paper, the Ford's et al. formula [1951]
was used to calculate the friction coefficient. Other formula can also be used to
calculate the friction coefficient, for example: Sims's [1952] formula, Ekelund's [1933]
formula and Roberts's [1978] formula etc.. All predict the expected trend of lower
frictional resistance with increasing velocity.
The inverse method has been used by many researchers. In this technique, process
parameters are determined experimentally. A model, which calculates these parameters,
is used andfrictioncoefficient is adjusted until the measured and predicted parameters
match. The result is an effective coefficient, which m a y mask some phenomena, not
accounted for in the model. The more rigorous the model, the closer is the inferred
friction coefficient to the actual value. Moreover, increasing the number of measured
and calculated parameters to be matched also improve the predicted accuracy. This
method m a y be applied to extrusion, drawing and rolling [Evans and Avitzur, 1968].
Lin et al. [1991] used inverse calculations to infer the magnitude of the friction
coefficient during cold rolling of steel strips.
3.3 Hydrodynamic lubrication model in cold rolling
It is now believed that a hydrodynamic lubrication regime may exist in high speed
rolling. The possibility of plasto-hydrodynamic ( P H D ) lubrication in cold rolling has
prompted some of the earliest attempts at the mathematical modeling of friction in
rolling. Nadai [1939] proposed a solution based on an assumed film thickness. Cheng
[1966] used plasto-hydrodynamic theory to estimate the lubrication film thickness
carried into the roll/strip interface. Bedi and Hiller [1967-1968], Avitzur and
Grossmann [1972] applied the m i n i m u m energy method within the plastic work zone to
calculate the film thickness. However, none of these workers investigated the lubricant
flow at the inlet zone which could be a major influence on entrained film thickness.
Walowit pointed out the difference between the different zones in elasto-hydrodynamic
inlet zones and developed a better inlet zone analysis, and he was thefirstto derive the
entrained film thickness by application of the Reynolds equation to the inlet zone.
Wilson and Walowit [1971] used a simplified version of this inlet analysis in an
isothermal lubrication of strip rolling with front and back tensions. Their well-known
inletfilmthickness formula is:
lrj0aR'{u +ur)
= (3
* rt-.-**^ "1)
where TJ0 and a are the base viscosity and pressure-viscosity coefficient of the lubrican
respectively, M W ] and ur are the inlet strip speed and roll speed respectively, R'
deformed roll radius, ay the strip yield strength, s the back tension stress, and xx the
contact length.
Wilson and Walowit's analysis has been extended by Atkins [1974], Dow et al. [1975],
Wilson and M u r c h [1976] and Aggarwal and Wilson [1978] w h o treated various aspects
of thermal effect on viscosity during rolling process. All these models applied
lubrication theory which is only appropriate for the thick film regime where the m e a n
lubricant film thickness is m u c h larger than the surface roughness. Another model for
the thick film hydrodynamic lubrication of strip rolling which combines a slab plasticity
model, a hydrodynamic lubrication model and a thermal model has been developed by
Chung and Wilson [1994]. The rolling model considers m a n y effects reported by
previous researchers, such as the relationship between pressure and viscosity, and the
thermal loss of traction at high speeds described by Wilson and Murch [1976] over a
wide range of speeds and reductions. The predictions of the model are in close
agreement with experimental measurements of exit speed, roll separating force and
rolling torque in rolling for aluminium, with a mineral oil or polyphenyl ether as
lubricants.
Film thickness always plays a central role in different lubrication theories. If the fil
thickness is large, the asperities will not touch and friction will be low. O n the other
hand, if the film thickness is small, some asperities will touch andfrictionwill be high.
In order to calculate the part of thefrictiondetermined by asperity contact, the film
thickness should be calculated accurately. Since the elastic deformation of the strip at
the inlet zone can be large compared with the film thickness, the effect of this
deformation can have a large impact on the calculated film thickness. The elastic
deformation of the strip has been considered by Lugt et al. [1993]. The film thickness
considering the elastic deformation of the strip was used as follows:
h(x) = h0+ (x2 /2R') + yx- v(x') (3-2)
where h(x) and h0 are the film thickness in the inlet zone and constant film thickness
respectively; yx and y{x) are constant strip thickness and strip thickness in the inlet
zone. R' is deformed roll radius. In Lugt and Napel [1995] works, the elastic
deformation of both the strip and the rolls have fully been incorporated in the model.
Thermal effects regarding heat development caused by plastic deformation as well as
work hardening have been included. The new film thickness can be written as:
h{x) = hQ+ (x212R) + dr (x) + ds(x) (3-3)
where dr(x) and ds(x) are elastic deformation of the roll and elastic or plastic
deformation of the strip respectively.
A model has been developed by Saxena et al [1996] for cold rolling under
hydrodynamic lubrication which combines the finite-element analysis of the strip with
an analysis of the lubricant film. The viscosity of the lubricant is assumed to depend on
both the pressure and the strain rate, whilst the strip is modelled as perfectly rigid-
plastic material. The thickness of the lubricant film is assumed to vary parabolically
from inlet to exit. The film thickness in the inlet zone without tension is derived and
expressed as follows:
^3rj0a{uWi+ur)
1
tan0
The final results show that the film thickness decreases at high reduction, but increase
with R'/hx. Therefore, for high reduction or low R'I hx values, either the lubricant
viscosity or the roll velocity must be sufficiently high to maintain hydrodynamic
lubrication.
Lin and Houng [1991] applied the Reynolds equation to the three zones: inlet zone,
plastic work zone and outlet zone under the fully hydrodynamic lubrication for high
speed rolling. Thermal effect is also considered in the analysis.

3.4 Mixed film lubrication model in cold rolling
In metal-forming processes, different regimes of lubrication can occur at the tooling-
workpiece interface. These regimes can be characterized by the thickness of the
lubricant film relative to the surface roughness and by the fraction of the interface load
carried by the contact of roughness peaks or asperities. Wilson [1978, 1979] has
described four main regimes: thick-film, thin-film, mixed film and boundary. The thick-
film and thin-film regimes are often combined and called the full-film regime and that
terminology will be used here.
The study of hydrodynamic lubrication in metal forming has reached a level of
sophistication which allows the prediction of conditions in rolling in the full-film
regime [Sa and Wilson, 1994]. However, conventional full film lubrication, in which the
tooling is completely separated from the workpiece by a thin film of liquid lubricant
does not often occur in bulk metal forming because of the need to control surface
roughness [Wilson and Schmid, 1992]. Most process tend to operate in the mixed
regime in which part of total interface pressure is provided by asperity contact at surface
peaks and part by the pressurized lubricant in the surface valleys.
The mixed regime is difficult to model because it is necessary to handle both the
mechanics of asperity contact and lubricant flow between rough surfaces. Mixed
lubrication in metal forming introduces special challenges because the bulk plastic flow
of the workpiece tends to reduce the effective hardness of the asperities [Wilson and
Sheu, 1988]. In addition to a more involved modelling of asperity interactions, this
effect can result in high fractional contact area, which render the traditional ways of
modelling lubricant flow between rough surfaces, such as Patir and Cheng's [1978,
1979] average Reynolds equation, inapplicable.
The earliest model of mixed film lubrication in cold rolling was developed by Sargent
and Tsao [1980]. The serious deficiency of their model was that it failed to take account
of the influence of the bulk plastic deformation of the strip on asperity scale
deformation. T o date, most attempts to model the mixed lubrication of metal forming
process, such as the analysis of sheet metal forming by Wilson [1990] and the analysis
of rolling by Sutcliffe and Johnson [1990], and Sheu and Wilson [1994], treat relatively
high speed conditions where the hydrodynamic pressure builds up in the inlet/plastic
work zone and the pressure gradients in the work zone have a negligible influence on
the lubricant flow. Sutcliffe and Johnson [1990] carried out an inlet analysis, in which
the deformation pressure is shared between the contact asperities and fluid film in the
surface valleys. F r o m the asperity crushing rate, thefractionalcontact area and average
film thickness were obtained. The average film pressure was calculated by integrating a
simplified first-order Reynolds equation. However, there is ample experimental
evidence that lubricant viscosity and rolling speed influence frictional conditions at low
rolling speeds.
Wilson and Chang [1994, 1996] developed an analytical model for strip rolling under
low speed conditions where the pressure generated in the inlet zone are negligible. The
analysis shows that relatively high hydrodynamic pressure can be generated by wedge
action in the converging channels in the plastic work zone. This can occur even under
conditions where it was previously considered that hydrodynamic effects are
unimportant. The sheet surface roughness was approximated by a longitudinal sawtooth
topography and evolution of surface roughness was estimated, with the relationship
between thefractionalcontact area, average deformation pressure and the film pressure
derived by using an upper bound theory [Wilson and Sheu, 1988]. The film pressure pf
was determined from a simplified first-order Reynolds equation.
Qiu et al. [1999] developed a similar deformation model, but with a more rigorous
second-order Reynolds equation shown below.
d_ ht3 dpf ^ = _ 0 \ur+uwdht | htduw\

n
dx \2n dx 2 dx 2 dx I
There is no unknown constant in the Reynolds equation. Film pressure is solved from
the Reynolds equation through an over-relaxation method, with boundary conditions
automatically applied. In this case, the film pressure is set to zero at the entry p
the plastic work zone and the exit of the plastic work zone. The variations of the y
stress with strain are considered in the model. An efficient iteration procedure is
developed to solve the contact area, film thickness and hydrodynamic pressure. The
model is more practical with fewer assumptions, and converges quickly. It is applica
to a wider range of rolling regimes, particularly at high rolling speed.
3.5 T h e r m a l effect in cold rolling
The steel rolling process involves extremely high pressures and velocities. A large
amount of heat is generated from plastic deformation and friction causing the
temperatures of the work rolls and strip to increase. The transfer of thermal heat
contact is of equal importance to the transfer of forces and forms a significant pa
studies of tribology. It is reasonable to expect that the parameters that affect fr

forces will also affect the amount and the rate of heat transfer in the deformation zone.
The property of lubricant is greatly affected by temperature because lubricant viscosity
is very sensitive to temperature variation, so the thermal effect must be considered in
the mixed film lubrication.
Several analytical models have been developed to model the thermal behavior of the
strip rolling process. Johnson and K u d o [1960] used an upper-bound technique to
predict strip temperatures. Cerni et al. [1963] provided a transient solution for
temperatures of work roll subject to whole circumferential convective cooling and a line
heat source. Patula [1981] obtained a steady-state solution for temperatures in a rotating
roll subject to a constant surface heat input over one portion and convective cooling
over another portion of the circumference. Tseng et al. [1990] extended Cerni's work to
allow variable heat flux.
The finite difference approach has been largely used in the numerical approach. Pioneer
work was conducted by Peck et al [1954], w h o used a Lagrangian formulation in which
the cylinder is considered fixed with respect to the co-ordinate system, with the
boundary conditions rotating with it (i.e. periodic). Only radial heat transfer was
examined in this work. Recently, H u s and Evans [1990], Lenard and Pietrzyk [1990],
and Y a m a d a et al. [1991] analyzed temperatures of the hot rolling process using two-
dimensional finite element code. A two-dimensional study which accounted for the
circumferential heat flow was also carried out by Parke and Baker [1972], while
Poplawski and Seccombe [1980] reported a model taking into consideration the heat
transfer in the axial direction as well.

Y u e n [1987-1] examined the transient thermal variation of a rotating cylinder with both
radial and axial heat flows. H e also used the moving heat source formulation to examine
thermal exchanges between two semi-infinite sliding solids in contact over a finite
region. The heat flux partition to each solid was determined and the resulting
temperature fields in the solids evaluated. These studies, m a d e in the context of strip
rolling, considered boundary conditions where one solid had a bulk temperature
different from the other [Yuen, 1987-2, 1987-3, 1988, and 1993]. The thermal boundary
layer in the subsurface of a rotating cylinder subject to surface heat flux was also
examined [Yuen, 1994].
Wilson and Mahdavian [1974] developed one-dimensional incompressible Reynolds
equation which takes into account viscosity variations across the lubricant film
thickness due to energy dissipation within the film. Only the conduction m o d e is
considered. Another attempt was also m a d e to develop a more sophisticated
mathematical model of metal forming processes by taking convective heat transfer into
account, along with conduction [Bhatt and Sengupta, 1996]. The generalized energy
equation was reduced into a simple form for conduction and convection modes of heat
transfer across a hydrodynamic incompressible fluid film. The solution of this energy
equation with appropriate boundary conditions led to an expression for temperature
distribution across the lubricant film in which some of the terms can be grouped in a
non-dimensional number called Peclect number. Wilson et al [1989] devised a unique
method to speed up the surface temperature calculation. B y using both analytical
solution and finite difference scheme in the formulations, they were able to calculate
directly the surface temperatures. The model considered only convection parallel to the
roll-strip interface and conduction normal to this direction. While the work of Wilson et
al. focused on calculating surface temperatures, their idea demonstrates a potential in
significantly reducing computational time since the formulation are analytical-based,
and explicit. Chang [1998] extended their idea to calculate interior temperatures of the
materials.
Lugt and Napel [1995] developed a model for simulating hydrodynamic lubrication in
cold rolling, in which the thermal effects regarding heat development caused by plastic
deformation as well as work hardening have been applied. It was assumed that the
deformation energy is entirely transformed into heat. In addition, friction-induced heat
generation, and the conduction of heat into the rolls were neglected. Hence, the
temperature of the strip was determined by the equilibrium between the deformation
energy and the heat generation as follow:
^ = _^L_.^ (3-6)
dx ypcp dx
But the accuracy of the model will be reduced because of the assumptions and the
neglect of friction-induced heat.
Lin and Houng [1991] applied the energy equation for the inlet zone, plastic work zone
and the outlet zone to calculate strip and rolls temperature under the fully hydrodynami
condition. This method neglected the interior temperatures of the strip and the rolls. I
efficiently calculated the strip and roll surface temperature as well as the temperature
the lubricant for fully hydrodynamic condition.
Much work has been done in the study on the thermal contact phenomena between two
solid bodies in contact. Because of the surface roughness, it is generally believed that
the actual contact area between two contact solids is only a portion of the nominal
contact surface, even with good surface finish. The temperature at the lubricant contact
can be calculated by using Lin and Houng's [1991] method. Qiu and Cheng [1998]
provided a method to calculate the temperature at the solids contacts. The heat flux in
the contact area was used and the temperature from a moving heat source can be
calculated by the method outlined by Carslaw and Jaeger [1990].
The heat flux distribution to the two contact surfaces depends on the properties of thei
materials and the history of contact surface temperature. This distribution changes with
the position of the heat sources and time. A simulation program has been developed by
Qiu and Cheng [1998] to calculate the temperature distribution of three-dimensional
rough surfaces in sliding contact under mixed lubrication.
3.6 Summary
Friction and lubrication play an important role in cold rolling. Many efforts have been
made by the researchers to try to understand the mechanism offrictionand lubrication
in the roll bite. The embedded pin technique and the sensor roll methods were used to
measure friction coefficient point by point in the roll bite. The average friction
coefficient in the roll bite can also be measured by using the strip marking method and
Laser Doppler method. The measured results were validated by the experimental rolling
force and torque. The author will measure friction coefficient in the roll bite by using
embedded pin technique.
The mixed film lubrication model and the fully hydrodynamic model have been
developed for m a n y years. In the practical rolling process, the mixed film model play a
dominant role. Several researchers have developed mathematical models under the
mixed film condition in the cold rolling. The author will extend their works to consider
the effects of hydrodynamic inlet zone and outlet zone as well as the thermal effect in
the lubricant and at the asperity contacts interface.

Chapter 4 Friction Coefficient Measurement and Rolling Experiment 60
Chapter 4
Friction Coefficient Measurement
and Rolling Experiments
4.1 Introduction
The ratio of the interfacial frictional stress to normal pressure is defined as the
coefficient. There are several ways to describe that coefficient. These include the
Amonton's laws and the statement that xi = mks where rt is thefrictionalstress, ks
the shear yield strength, and m a constant multiplier between zero and unity. While
both of these approaches lead to a reasonable prediction of process variables they are,
strictly speaking, incorrect for rolling process. A number of cold rolling tests were
reported in references [Rooyen and Backofen, 1957], [Banerji and Rice, 1972], and [Al-
Salehi et al, 1973], all showing that the ratio of frictional stress to normal stress is
indeed not constant in the roll bite, confirming the suggestion by S h a w et al. [1960] that
for relatively high interfacial normal stress that ratio varies along the contact zone.
The objectives of the present study are to understand the relationship between the
friction coefficient and the process parameters and h o w the friction coefficient varies in
the roll bite. Friction coefficient can be measured by either direct or indirect methods.
Direct method refers to the sensor roll method, and indirect method includes strip
marking method and the Laser Doppler method as discussed below.
4.2 Sensor roll design
The sensor roll is designed with 225 mm diameter, 254 mm barrel length, and 900 mm
total length. The roll is hardened to 48 ~ 50 H R c , and the hardness of its nitrided surface
is 65 to 70 H R c within a depth of 0.2 ~ 0.3 m m . After grinding, the roll surface average
roughness is 0.36 p m along the rolling direction, 0.66 p m along the axial direction. The
roll is cut into the two parts along the axial direction as shown in Figure 4.1 in order to
embed sensors into the roll body. The separated two parts are held tight together by
eight bolts and two keyways. O n the separating surface of the roll body, there are four
pins to guide the segment part into position properly, and two keyways on either side of
separating surface of roll body to stop the segment movement along the rolling
direction. The roll is m a d e from Bohler W 3 0 2 grade material. The W 3 0 2 specification
is as follows:
Table 4.1 W 3 0 2 specification
Element C Si Mn Cr Mo V
% 0.39 1.0 0.4 5.1 1.3 1.0

Segment Pa
%,
Drive side
Roll body
Operation side
Figure 4.1 Sensor roll overview
Roll body
S e g m e n t par"
-Radial pir
Figure 4.2 Sensor roll section view
Four pin-transducers are embedded in the sensor roll. T w o of them are the A L D - W - 1 0
load washers which are compact temperature-compensated load transducers with a
working range 453.6 kgf, and the other two are the temperature-compensated full-brid
strain gauges glued to the surface of the pins. Two pins sit directly on the load wa
respectively, another two pins with strain gauges are supported by the ball bearings
pins have the same material property as the sensor roll when in contact with the stri
during rolling. A thermocouple is also designed to measure the temperature of roll

surface. All electrical wires go through interior hole along the sensor roll axis and come
out at the end of the roll operator's side where the wires are connected to an amplifier
box via a slip ring, and the signals are then transmitted to the data acquisition system.
Figure 4.3 Side view of sensor roll and pins
Figure 4.4 Front view of sensor roll and pins
All pin diameters are 2 m m , and the m a x i m u m clearance between the pin and hole is
about 0.027 m m . T w o pins are radial pins along the radial direction of roll; another two
pins are oblique pins inclined at an angle of 25 shown in Figure 4.2. Figures 4.3 & 4.4
are the images of side view and front view of sensor roll and pins position. After the
sensor roll is assembled, it is sealed by using silastic, and then sent for grinding.
4.3 S e n s o r roll calibration
Before the experiment starts, the pin-transducers must be calibrated. W h e n the sensor
roll assembly is completed, the pins are normally 0.5 mm above the roll surface so that
force can be applied directly on the top of the pin. Because the roll surface is curved
special equipment called "calibration jig" is designed and used in the calibration (Fi
4.5).
call bearing
Figure 4.5 Calibration process illustration
During the rolling process, rolling force is direct to the centre of the roll. So the
force is also applied towards the roll center. The force acting on the pins can be
measured by a transducer as shown in Figure 4.5 on the top of radial and oblique pins,
and the original applied load is from known weights. The signals from radial and
oblique pins can be obtained each time when a weight is applied. The loading and
unloading signals are recorded. The sensor roll will be ground after calibration, an
pins' surfaces are totally flush with the roll surface. After grinding, a pair of ro
roll-sensor roll with 225.60 mm diameter and bottom roll with 227.78 mm diameter, ar
installed in the housing of two-high rolling mill to perform rolling test. Due to th
mismatch of manufactures, the diameter of top and bottom rolls are slightly differen
The calibration curves for radial strain gauge, oblique strain gauge, radial loadcel
oblique loadcell are repeatable and shown in Figures 4.6, 4.7, 4.8, and 4.9. The upp
line represents the loading, and the lower line represents the unloading case.
Radial strain gauge

3000 j- —l
_ 2000 -
2
0)
o
£ 1000 -
0 -
0 1 2 3
Output (v)
Figure 4.6 Radial strain gauge calibration
Oblique strain gauge

3000 -r
_ 2000
z
O
O
5 1000 -
u.
0-
0 0.2 0.4 0.6
Output (v)
Figure 4.7 Oblique strain gauge calibration

Radial loadcell
3000
0.1 0.2 0.3 0.4 0.5

Output (v)
Figure 4.8 Radial loadcell calibration
Oblique loadcell
4000 i
3000
g 2000
1000
0.5 1 1.5
Output (v)
Figure 4.9 Oblique loadcell calibration
The average regression equation for the above curves can be obtained as follows:
For radial strain gauge pin:
F =952.2F (4-1)
For oblique strain gauge pin:
F = 4493. O F (4-2)
For radial loadcell:
Frl = 5930.0F W (4-3)
For oblique loadcell:
Fol = 2 6 6 4 . 9 0 3 ^ (4-4)
The hysteresis exists during unloading for Figure 4.6, 4.7, 4.8 & 4.9, and hysteresis of
the oblique transducers are more serious than radial transducers. These phenomena are
caused by the friction between the pin and the roll body. Until now, the information
regarding the calibration for pin-transducers measurement technique are limited. Only
Rooyen and Backofen [1957] and Al-Salehi et al. [1973] mentioned the calibration
results in their publications.lt was not discussed at all in Lim and Lenard [1984] and
Lenard [1991].
4.4 Experimental facilities
A Hille 100 two-high experimental rolling mill (Figure 4.10) with rolls 225 mm
diameter and 254 m m length, driven by a variable speed motor of 75 hp, was used. The
maximum rolling force, torque and speed are 1500 kN, 13 kN-m, and 70 rpm or 0.8 m/s,
respectively. The roll gap can be set by mechanical screwdown system and two
Figure 4.10 Hille 100 rolling mill

The two loadcells to measure the rolling force on the operator side and drive side are
located on each side of the mill housing. The total torque can be measured by the
brushes on the torque shaft located before main gearbox. After the gearbox, the two
spindles driving the two work rolls are equipped with strain gauge bridges to measure
the torque of top and bottom rolls.
Operator side loadcell Dperator side loadcell
Drive side loadcell Drive side loadcell
Hille 100
Total torque reading Amplifier Total torque reading
Top torque reading Top torque reading
box Data
rolling
Radial straingauge Radial straingauge
acquisition
nill
Oblique straingauge Dblique straingauge
system
Radial loadcell Radial loadcell
Dblique loadcell Dblique loadcell
Speed
Temperature
Figure 4.11 Data acquisition system
During the experiments, eight signals are recorded by a computer through an amplifier
box, and speed and temperature are directly connected to data acquisition system
(Figure 4.11). The m a x i m u m sampling rate is 250 k/s. Every recorded figure in the
computer is the m e a n value of 30 readings, so it will take 1.8 m s for each acquired
figure in the total 15 channels (another five channels: time record, entry and exit
thickness, entry and exit roll gap). The acquisition time 1.8 m s is calculated as:
30 x 15 x 1000 / 250,000 = 1.8. For the rolling condition with 3 7 % reduction and 70 rpm
(0.824 m/s) maximum rolling speed, the rolling time over roll bite will be 25 ms, so it
can pick up 14 sampling points in the roll bite (25 /1.8 » 1 4 ) . M o r e sampling points can
be recorded at low rolling speeds. So this system is fast enough to record experimental
data in the roll bite. If more recording points are required, the above 30 readings can be
reduced to 20 or 10. The data acquisition system is written by C + + code in Lab
W i n d o w s environment. The program is listed in Appendix A , and the data acquisition
screen is shown in Figure 4.12. The calibration results for rolling mill loadcells, torque,
speed, and thermocouple are shown as following:
Rolling mill drive side loadcell: Pd = 433.12Vd (4-5)
Rolling mill operator side loadcell: P0=A19.1SV0 (4-6)
Rolling mill total torque: Tt = 11.892Fr (4-7)
Rolling mill top torque: Ttt = 11.013Vtt (4-8)
Roll speed: ur=\0.543Vr (4-9)
Thermocouple: T = 65.50Vtc (4-10)
The uncertainty of friction coefficient measurement has been corrected in Appendix F.
From Eq. (F-14) in Appendix F, it can be seen that highfrictioncoefficient value and
small uncertainties of radial & oblique pin can reduce the uncertainty of friction
coefficient value, and vice versa. In Figure F.l from Appendix F, it has been shown that
the friction uncertainty of 4 5 % of the contact length from entry and 6.8% from exit are
less than 3 0 % , and 4 7 . 7 % of the contact length in the middle are higher than 3 0 % .
Around the neutral point, the uncertainty values become m u c h larger due to small
friction coefficient values close to zero.

<D 10 ^
en en CO
o ra
16 ^ 1 3 :
• <o •: •3*-;
a *t i ID \
4 • O ID
at
• 3 i o
I ro ! »— 4)
I••> j ' s==-
JJ ~ 1 c1 <
N
to en j 3 = LO
ro
1 "8 CM • CJ :
CD j <D = c
a :• J=£ : <D
<D CD ^ U) 4 r
1
> ** o
Q
<•
CD
<> Uu
01 ..
Oi ^:
Q 3 -!_
O
r
<TJ CO CD
~ 13 CD
C C") CD 1
2 o 4 •
<0 ]
n
SBSnto CD
| | en 4 •
ID
ro
O
to
C3
a
ro o
CO o :
rn a
CO
• * » • •
o
4 • 4 •
J£
Pi)
a _
ro ?s
o o in w
CO
73 CO
(0 CO Q_ CM ._
O o "- CO O
_l • * f r
o
0) CO o Q> CO O
m CD
(M o O 4•
< 4 • 4 •
m
CTi
ro « ro ^
.£ co en to CO
ro —
J5
co CM
CD
a
1o
UD
1 si s' LU
o o
OT :
* - ' -g-
GC 4 • 4 •
J ) _tJ,
O CJ -i- ro rp
-o co en co co
ro O 0)
CD CD
Strip Th
CD i 1 i
8.00-
ro CO O u->
ro LD CD o o o o
4 • 4 • D Ul 4 • ^ •
o o CM
CD
Figure 4.12 Data acquisition screen

Aluminium alloy 5052-H34, 6060-T5, and carbon steel BHP-300 were used in the
experiments. The lubricants include Rolkleen 485 used at a B H P tin mill, mineral oil
A W S 10, A W S 100, and A L P H A SP 1000. Tables 4.2, 4.3, and 4.4 show the chemical
composition for the three test materials. The experiments were carried out under dry and
lubricated conditions in cold rolling. The specimen dimension is 3x 100x (1000-2000)
mm.
Table 4.2 Aluminium alloy 6060-T5
Element
Si Fe Cu Mn Mg Cr Zn Ti Al
Content 0.40 0.15 0.8 0.04
(%) 0.7 0.15 0.25 0.15 Rest
0.80 0.40 1.2 0.35
Table 4.3 A l u m i m u m alloy 5052-H34
Element
Si Fe Cu Mn Mg Cr Zn Ti Al
Content 2.2 0.15
(%) 0.25 0.4 0.1 0.10 0.10 ... Rest
2.8 0.35
Table 4.4 Carbon steel BHP-300
Element
C Si Mn P S
Content(%) 0.25 0.50 1.60 0.040 0.040
4.5.1 Signal record
In order to measure the friction coefficient, the sensor roll was cut into two pa
shown in Figure 4.1, and then it was joined together by bolts and keyways. So it is
rigid solid roll any more. This will result in a difference of rolling force and tor
recordings from those of a normal roll. The signals for carbon steel under 32.12%
reduction and 3 rpm rolling speed are shown as follows:
800
2" 600 V^L

4>
O 400
o>
I 200 •Total force
•Drive side force
Operator side force
10 15 20 25 30 35 40 45
Time (s)
Figure 4.13 Typical rolling force record
10
î»yfc<"'A"" •••ml
10 15 20 25 30 35 40
Time (s)
Figure 4.14 Typical rolling torque record
In Figures 4.13 & 4.14, it can be seen that there is a 'W pattern signal during rol
This is caused by the segment part where point A (Figure 4.2) starts to touch the s
the 'W pattern begins, and when the point B touches the strip (rolling direction is
anticlockwise), the 'W pattern is completed. Such phenomena could come from the
sensor roll radius tolerance up to 60pm from dial indicator after the segment part
fastened to the sensor roll body and ground together as one unit. In normal rolling
process, strip thickness varies within a small margin. But due to roll radius toleranc
caused by the segment part, the strip thickness within the segment part is changed as
shown in Figure 4.15. In Figure 4.15, position 1 and 10 are point A and B in Figure 4.
respectively. A variation of strip thickness results in the change of force and torque
recording, and a *W pattern is formed. The elapsed rolling time of 'W pattern in
Figures 4.13 and 4.14 is the same as those in Figure 4.15
2.75
— Sample 1
Sample 2
Sample 3
— Sample 4
3 4 5 6 7 8 9 10
Position on segment part
Figure 4.15 Strip thickness within segment part
The curve length of segment part is 175.35 m m , but the roll bite length for the case in
Figure 4.13 & 4.14 is just 10.42 mm. Comparing with the arc length of segment part,
the roll bite length is much smaller. Strip thickness does not change much within the
roll bite, so the rolling force and torque over the roll bite remain nearly constant
in Figure 4.16 & 4.17.

800
600
o
400
r\s\r~< n r ft n rn n rr n nr* r* r w r*r* wr* n nr< n rin n
O)
c - 4 — Total force
f 200
£ -•— Drive side force
Operator side force
Bitry Roll bite Beit
Figure 4.16 Rolling force over roll bite

10
8
?
±iC
6
V
3
cr
o^
«J
4
CD
C 2
fi
0
Bitry Roll bite Exit
Figure 4.17 Rolling torque over roll bite
Therefore, as rolling force and torque are nearly constant in the roll bite, the
measurement of friction coefficient is not affected. If the ordinary roll without spl
used in the rolling experiment, the rolling force and torque signals will remain
approximate constant during rolling as shown in Figure 4.18. So, it can be seen that
'W pattern signal is caused by the segment part of the sensor roll. But this will not
affect friction measurement due to constant force and torque in the roll bite.
Figure 4.18 Rolling force recorded by ordinary roll
The roll surface temperature in the roll bite is measured by using a thermocouple
embedded in the sensor roll. The design detail is shown Figure 4.19. The wire is
soldered on the top of nut, and the nut is threaded into the sensor roll. The following
Figure 4.20 is the temperature recording under 26.94% reduction, 5 rpm rolling speed
for aluminium alloy 6060-T5.
A portion of sensor rol

Segment part
Figure 4.19 Thermocouple in sensor roll
o<^
O
"jiT 2 4 -
•
3
* *
CB
k.
fl>
^"^lllMHllii
i" I P m ^ p
a. 23-
E gi^MiJ
at
I- t 9 14 19 24
22-
i Time(s)
Figure 4.20 Temperature record
The temperature peaks are formed when the thermocouple's pin touches the strip
surface shown in Figure 4.20. The time over the temperature peak is equal to the ro
time over roll bite. After a large number of tests, the measured temperature was fo
not very accurate because of the junction of the thermocouple being pushed below ro
surface.
Rooyen and Backofen [1957] investigated the pin binding by using a specimen with two
longitudinal grooves machined along its length while the system was under load in th
roll gap. It was found that the elastic distortion of the roll caused forces to be i
on the pins which amounted to about 5-10% of the values obtained when the pins did
make contact. The same finding has been reported by Smith et al. [1952]. So, Rooyen
and Backofen [1957] concluded that "the record obtained in groove-strip rolling was
similar for the two pins, so that the ratio of the stress in the oblique pin to that
radial pin, and consequently the experimentally determined value //, is not likely t
greatly affected by whatever causes this response". Tests with longitudinal grooves
similar to Rooyen and Backofen [1957] for carbon steel and aluminium alloy were
repeated and results were shown in Figures 4.21 & 4.22. In Figures 4.21 & 4.22, it c
be seen that the measured force (0-30 N) in signal test is less than 1% comparing wi
total measured force (2500 N) in real rolling experiment for the same reduction. So
are confident about experimental results which the error is less than with 5-10% of
Rooyen and Backofen's [1957].
—•— Radial loadcell —•— Oblique loadcell

Radial straingauge — * — Oblique straingauge
9.9 9.95 10 10.05 10.1

Time (s)
Figure 4.21 Pin signal test for carbon steel

Figure 4.22 Pin signal test for aluminium alloy
4.5.2 Friction coefficient measurement
Friction coefficient can be measured by the sensor roll method, strip marking method,
and Laser Doppler method. These three methods will be illustrated as follows:
I. Sensor roll method
The determination of the friction coefficient in the roll gap is done by an analysis
equilibrium of forces acting on the radial and oblique pins (Figure 2.6). Here, the
equation-Eq. (2-26) in Rooyen and Backofen [1957] is used in calculating the friction
coefficient.
Double and multiple pressure peaks have been measured [Al-Salehi et al, 1973; Lim
and Lenard, 1984; Lagergren, 1997] and predicted [Li and Kobayashi, 1982, Lenard,
1981]. But other authors [Banerji and Rice, 1972; Jeswiet and Rice, 1982; Britten &
Jeswiet, 1986; Nyahumwa and Jeswiet, 1991] did not observe the existence of multiple
pressure peaks. Among hundreds of rolling test in this laboratory, no multiple press
peak were found for radial pins. But two pressure peaks were discovered for the obliq
pin at the beginning of the experiments as shown in Figure 4.23. This was due to the
segment movement along the rolling direction. After this phenomenon was found, the
sensor roll was modified in which the two keyways were laid along axial direction of
the roll.
1500
1000
u
£ 500
0
8.35 8.4 8.45 8.5 8.55 8.6
Time (s)
Figure 4.23 Force distribution
Firstly, the experiment was carried out for aluminium alloy 5052-H34 under lubricated
condition. The lubricant Rolkleen 485A was used in the experiment. The force
distribution from the radial and oblique pins are shown in Figures 4.24 & 4.25 for
29.5% and 17.7% reduction, respectively. The experiments for Figures 4.24 & 4.25
were carried out under lubricated and dry condition, respectively.
5052-H34, ^=3.08111111, y2=2.172mm, u =115.4mm/s,

s=29.48%, Lubrication
1500 -r- -i
g-1000
8
£ 500
0
20.05 20.1 20.2
Time (s)
Figure 4.24 Force distribution 8=29.48% (lubricated)

5052-H34, y 1 =3.094mm, y 2 =2.546mm, u =114.2mm/s,

s=17.71%, Dry
1500
9.65 9.7 9.75 9.8

Time (s)
Figure 4.25 Force distribution 8=17.71% (dry)
After analysis of the measured data, the pressure in the roll bite shown in Figures 4.2
& 4.29 are obtained. From the Figures 4.26 & 4.29, the pressure peak in the roll bite a
predicted by the standard rolling theory cannot be detected. In traditional rolling th
the calculated pressure peak is formed under different friction coefficient as shown in
Figure 4.28. But in Figures 4.26 and 4.29, no obvious pressure peak is found. This coul
be due to the effect of the pin diameter (2 mm) compared with the roll bite contact
length (10.1 mm for lubricated condition, 7.9 mm for dry condition) and the effect of
friction variation along the roll bite.
5052-H34, y ^ O S m m , y 2 =2.172mm, u =115.4m m/s,

8=29.48%, Lubrication
400
£ 300
| 200
w
w
0)
£ 100
Entry Roll bite Exit
Figure 4.26 Pressure distribution s=29.48% (lubricated)

5052-H34, y ^ . O B m m , y2=2.172mm, ur=115.4mm/s,

£=29.48%, Lubrication
0.2
» 0
^ = 4 ^
-0.4
Figure 4.27 Friction coefficient s = 2 9 . 4 8 % (lubricated)
Figure 4.28 Pressure peak in traditional rolling theory [Ginzburg, 1989]
Eq. (2-26) rewritten as below is used to calculate friction coefficient shown in Figures
4.27 & 4.30. From Figures 4.26, 4.27, 4.29, and 4.30, it can be seen that the location of
the single pressure peak may not coincide with the location of neutral point and that the
friction coefficient varies widely throughout the roll bite.
( \
M 'tgO (4-11)
Pr
5052-H34, y 1 =3.094mm, y2=2.546mm, u =114.2mm/s,

s=17.71% Dry
600
ra
a. 400 *—m-
3
Ifl
in 200
0) Radial pin
a.
Oblique pin
Figure 4.29 Pressure distribution 8 = 1 7 . 7 1 % (dry)
5052-H34, y1=3.094m m , y 2 =2.546m m , u=114.2m m/s,

s=17.71%, Dry
0.4-,
.1 0.2
o
<*-
a>
° 0
c
o
1 -0.2 -i
LL
-0.4) Bitry Roll bite Exit
Figure 4.30 Friction coefficient 8 = 1 7 . 7 1 % (dry)
Normally, the friction coefficient should be positive in the roll bite. But in the Figures
4.27 & 4.30, this value becomes negative in some part of roll bite. This means that
friction direction on the strip surface along the roll bite is changed to opposite of the
rolling direction. This is also applied to the following graphs of friction coefficient in
this chapter.
T h e tests w e r e carried out for aluminium alloy 6060-T5 under the lubricated condition.
The rolling speed was set at 3 rpm, 5 rpm, 7 rpm, 15 rpm, 30 rpm, 50 rpm, and 65 rpm
for the same reduction around 38%. The force signal distribution from loadcells and
strain gauges and friction coefficient for those rolling conditions are shown in Figur
4.31-4.44. The friction coefficient is obtained by using Eq. (4-11). The average fricti
coefficient p for each case is also shown in the following graphs. This value is
obtained by summing all the absolute friction coefficient values in the roll bite, and
divided by the number of value.
(1) Aluminium alloy 6065-T5, 36.82% reduction, 3 rpm rolling speed
Aluminium alloy 6065-T5, s =36.82%, n=3rpm

1500
1000 illMli—
A^ V
0)
o \
500 %
w — • — Radial pin
Jf —m— Oblique pin
l M R / % • • ••
47.6 47.8 48 48.2
Time (s)
Figure 4.31 Force distribution (e=36.82%, n=3rpm)
Aluminium alloy 6065-T5, 8=36.82% n=3rpm, ji =0.1374
0.5
c
o
***
"0
**•••••
c ••'
••••
o ••••• • •••
„•••••
1
Li. -0.5
Figure 4.32 Friction coefficient (e=36.82%, n=3rpm)

Aluminium alloy 6065-T5, s =37.01%, n=5rpm
Figure 4.33 Force distribution (s=37.01%, n=5rpm)
Aluminium alloy 6065-T5, e =37.01%, n=5rpm, ^ =0.1372

0.5 -,
a
o
E «*+"*
0) •••••
o ^ • • • ^ • • '••••••••••*
oo
-0.5
Figure 4.34 Friction coefficient (s=37.01%, n=5rpm)
Aluminium alloy 6065-T5, e=37.61% n=7rpm

1200
1000
2- 800
g 600
i£ 400
200
57.39 57.44 57.49 57.54 57.59

Time (s)

Aluminium alloy 6065-T5,e =37.61%, n=7rpm, ^=0.1226

0.5
c
"5 • <•
• •
8
c
,*••• *"**
0
-0.5
Figure 4.36 Friction coefficient ( e = 3 7 . 6 1 % , n = 7 r p m )
(4) A l u m i n i u m alloy 6 0 6 5 - T 5 , 4 0 . 4 5 % reduction, 15 r p m rolling speed
Aluminium alloy 6065-T5, 8=40.45%, n=15rpm

1200
1000
2" 800
3 600
£ 400
200
0 •*-* «*-*-*
7.97 7.99 8.01 8.03 8.05 8.07 8.09
Time (s)
Figure 4.37 Force distribution ( s = 4 0 . 4 5 % , n = 1 5 r p m )
Aluminium alloy 6065-T5, 8=40.45% n=15rpm, ^ =0.0939

0.5
o
u
o
c
o
-0.5
Figure 4.38 Friction coefficient ( s = 4 0 . 4 5 % , n = 1 5 r p m )

Aluminium alloy 6065-T5, e=40.38%, n=30rpm

1000
800
W\, 600
8
o 400
u.
200
0
11.79 11.81 11.83 11.85
Time (s)
Figure 4.39 Force distribution (8=40.38%, n=30rpm)
Aluminium alloy6065-T5, e =40.38%, n=30rpm, n =0.0865
4-*
C
»
"o •
•
E 0 -
0) • *
4
o
o
c
o -0.5-
^5
Time (s)
o
u.
Aluminium alloy 6065-T5, e=40.21%, n=50rpm

1000
800
% 600
S
O 400
LL
200
19.898 19.908 19.918 19.928

Time (s)

Aluminium alloy 6065-T5, e =40.21% n=50rpm, ti =0.079
c
Friction coefficient
t•
• •
o
Jt
Aluminium alloy 6065-T5, s ^40.12%, n=65rpm
-•— Radial pin

-•— Oblique pin
-
T 1
9.435 9.44 9.445 9.45 9.455 9.46 9.465

Time (s)
Aluminium alloy 6065-T5, e =40.12%, n=65rpm, ii=0.0757

0.5
c
o
o
0)
8
c
o
-0.5

The experiments were also carried out for carbon steel BHP-300 under lubricated
condition. The test rolling speeds were given as 3 rpm, 5 rpm, 7 rpm, 15rpm, 20 rpm,
and 30 rpm for around 31% reduction. The higher speed was not tested for lubricated
carbon steel because the difficulty of biting the strip into the roll gap at the star
rolling process. The force signal distribution and friction coefficient for those rol
conditions are shown in Figures 4.45-4.54. The friction coefficient was obtained by
using Eq. (4-11). The average friction coefficient is also shown in the figures. The
calculated average value was obtained the same way with aluminium alloy.
(1) Carbon steel BHP-300, 31.16% reduction, 3 rpm rolling speed
Carbon steel BHP-300, s=31.16%, n=3rpm
25.1 25.2 25.3 25.4 25.5 25.6 25.7

Time (s)
Carbon steel BHP-300, 8=31.16%, n=3rpm, n =0.0876
0.5
"5
£ •• + ••••*
o • ••
o • •• • • •• •••
• •
-0.5

Carbon steel BHP-300,G =29.75%, n=5rpm

PSOO -,
2000 -
1500 -
8
0 1000 -
LL
500-
0^
33.5 33.6 33.7 33.8
Time (s)
Carbon steel BHP-300,e =29.75%, n=5rpm, n =0.0863

0.5
o
£ ..•^•••••••^^
<L> ••••••"
o •••••••
o <>•••••••<••••••
c
o
o -0.5
LL
Carbon steel BHP-300,e =30.63%, n=7rpm
47.9 47.95 48 48.05 48.1 48.15

Time (s)

Carbon steel BHP-300, s = 3 0 . 6 3 % n=7rpm, LI =0.0856
o
oi
Friction coefficient
•
, •••'
o
.••••••••••••••***
en
6
Figure 4.50 Friction coefficient (e=30.63%, n = 7 r p m )
(4) Carbon steel B H P - 3 0 0 , 3 2 . 6 6 % reduction, 2 0 r p m rolling speed
Carbon steel BHP-300, £ =32.66% n=20rpm

2500
2000
\W 1500
5 1000
LL
500
0 •-•-*•
39.5 39.52 39.54 39.56 39.58 39.6
Time (s)
Figure 4.51 Force distribution (s=32.66%, n = 2 0 r p m )
Carbon steel BHP-300, s =32.66% n=20rpm, jx =0.0723

0.5
0)
o
£ -+ *-
<V "• •"
8
c
o
ts-0.5
Figure 4.52 Friction coefficient (e=32.66%, n = 2 0 r p m )

Carbon steel BHP-300, e =31.67%, n=30rpm

2500
2000
S 1500
8
5 1000
u.
500
0» W » 1 » »
16.27 16.29 16.31 16.33 16.35
Time (s)
Figure 4.53 Force distribution (s=31.67%, n=30rpm)
Carbon steel BHP-300,e =31.67% n=30rpm, u. =0.0676

0.5
0)
o
£
8 °
c
o
is
-0.5
From the above experimental figures for aluminium alloy and carbon steel, it can be
seen that the friction coefficient in the roll bite is not constant, and no obvious p
peak was observed over the roll bite.
II. Strip marking method
The strip marking method has already been explained in Chapter 2. Eq. (2-28) can be
used to calculate the mean value of friction coefficient in the roll bite. That equat
be rewritten again here as:

S
R*(A AIV (4-12)
f =
yi 2 Ap
After transformation, Eq. (4-12) can be rewritten as [Lenard, 1992]:
y\-yi (4-13)
P
2~jR'(yx-y2) - A^SfR'y7
The deformed roll radius is calculated by the following equation [Hitchcock, 1935]:
C P
R'^R (4-14)
1 + W{y
- x-y° 2)'
v
where
16(1 -v2)
C0 = (4-15)
7TE,
The forward slip Sf in Eq. (4-13) is determined by the strip marking method [Liu et al,
1999] in the experiment. Eq. (2-27) to determine forward slip is shown below as:
5_iL^lxioo% (4-16)
f
Ln
L0 is the circumference of the roll, and 706.5 m m for <j)225 m m roll diameter. L is
mark length left on the strip after the roll turns one revolution. The following gra
show the relationship between rolling speed and the average friction coefficient
obtained from the strip marking method.
The experiment was carried out for aluminium alloy 6060-T5 under the lubricated
condition with 11.65%o reduction shown in Figure 4.55. It can be seen that the frict
coefficient decreases as the rolling speed increases. This confirms the theoretical
prediction.
Aluminium alloy, 6060-T5, e =11.65%, lubricated

0.145
0) 0.125
o
o 0.105
o
c
0.085
o
LL
u
0.065
0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.55 Friction coefficient vs rolling speed (strip marking method)
III. Laser Doppler method
Basically, this method is the same as the marking method. The difference is forwa
measurement. Eq. (2-29) is used to calculate the forward slip. It is shown as:
uw -ur
(4-17)
Sr= — -xl00%
u.
The roll and strip exit speed ur,uWi can be measured by two L D V probes [Tieu et al,
1998] installed on the rolling mill as shown in Figure 2.8. After the forward sl
measured, the friction coefficient can be calculated by using Eq. (4-13).
The following table 4.5 shows the forward slip measurement results by using the
marking method and Laser Doppler method for aluminium alloy under lubricated
condition. The error is about 10%. The friction coefficient listed in the table i
calculated from marking method. The error probably comes from measurement system
The two methods are useful tool to determine the forward slip, and the average fr
coefficient in the laboratory rolling mill. But the strip marking method is diffi
in the industrial rolling mill because of surface quality of the product. The Laser
Doppler method has been adopted by some mill manufacturer to measure strip speed,
but the water sprays can affect the accuracy of the laser probe, especially in hot rolling.
Table 4.5 Comparison of forward slip
Entry thickness Exit thickness Reduction Marked distance

Mmm) v2(mm) £% L'(mm)
Case 1 3.081 2.895 6.04 176.5
Case 2 3.084 2.738 11.22 177.5
Case 3 3.075 2.541 17.37 178.5
Case 4 3.080 2.388 22.47 179.5
Roll speed Exit speed Forward slip Forward slip Error Friction coefficient
Kr(m/s) uw (m/s) LDV(%) M M (%) % p
Case 1 0.11543 0.11623 0.74 0.65 12.16 0.127
Case 2 0.11520 0.11643 1.07 1.23 13.01 0.133
Case 3 0.11510 0.11693 1.60 1.80 11.11 0.154
Case 4 0.11413 0.11670 2.24 2.37 5.49 0.177
4.6 Experimental results discussion
Rolling process is a complicated dynamic and physical process. In the plastic wor
zone, many factors such as rolling force, torque, temperature, reduction, forward slip,
friction coefficient, lubrication, and material yield strength etc. affect each other. So it is
important to know the relationship between them. One of these factors, friction
coefficient is the most sensitive and it is not known accurately. So the emphasis will be
on the relationship between friction coefficient and other factors. Other relation will
also be mentioned.
4.6.1 Effect of rolling parameters on rolling load
Normally, the rolling force and torque will increase as the reduction increases. T
statement has been proved by the rolling theory and experiments. Here, the experime
made for aluminium alloy 6060-T5 with 0.118 m/s rolling speed had proved this
statement again. The experiments were made under both dry and lubricated condition
as shown in Figures 4.56 & 4.57. It can also be seen that rolling force and torque
dry condition are higher than those values under lubricated condition.
Aluminium alloy6060-T5, u =0.118m/s

800
• Dry condition
• Lubricated condition •
600-4
•
5 400
• •
O) • •
| 200
s.
1 1 1 1
12 22 32 42
Reduction (e%)
Figure 4.56 Rolling force (ur=0.118m/s)
Aluminium alloy 6060-T5, ur=0.118m/s

Q
0-
• Dry condition
i* 6-
• Lubricated condition •
± •
m • •
3 4-
O" • •
0
O) 2 V
c
• •
1 n
! 12 22 32 42
Reduction (e%)
Figure 4.57 Rolling torque (ur=0.118m/s)
The following graphs are plots of rolling force and torque against rolling speed under
different reductions. The rolling force and torque increasing with the reduction r
can be seen in Figures 4.58 & 4.59 at 38% reduction. When the speed is greater tha
0.24 m/s, the rolling force remains stable. At a lower reduction, the rolling force does
not change m u c h over the speed range. But for 3 8 % reduction at lower speed, the
rolling force drops dramatically in Figure 4.58. This could be caused by friction
decreasing as speed increases. The rolling torque increases under all of reduction. But
after 0.24 m/s rolling speed, the rolling torque stays nearly constant as shown in Figure
4.59.
Aluminium alloy6060-T5
500
400
X ••• • 38%
0) • • • •
300 • • • • • • • •
o
k. • 27%
o
O) 200 XXX 18%
c X X X X
x11%
100
1
0
0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.58 Rolling force (aluminium alloy)
Alum inium alloy 6060-T5

6-
• • •
± •••
•
• 38%
a. 4-
• •
© • •
• 27%
3
k.
18%
2-
sO) X X x X X11%
c x xx
1 0-
0.2
r ,
0.4 0.6
,
0.8
Rolling speed (m/s)
Figure 4.59 Rolling torque (aluminium alloy)
In Figure 4.60, the rolling force increases slightly with rolling speed rising for the
reduction of 7 % , 1 1 % , and 1 8 % , but the rolling force drops a little bit af a lower speed
and then remains nearly constant at higher speed for 2 8 % reduction. The rolling torque
increases when rolling speed rises for all reduction, but for 28% reduction, the rolling
torque drops after 0.35 m/s speed in Figure 4.61. Because of the difficulty at roll bi
28% reduction at higher rolling speed under lubricated rolling condition, there are no
experimental points above 0.47 m/s speed.
Carbon steel BHP-300, lubricated

1000 i
g. 800
xxx x x X x
• 7%
g 600
• 11%
••• • • • • 18%
f> 400
c ••• • • • • 28%
200
1 1 "1 1 ' T
0
0 0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.60 Rolling force (carbon steel)

10
8
I xxx
X X A X
• 7%
a) 6
3 • 11%
O"
18%
o 4 •
• •
O) • X 28%
•••
c ••• • •
= 2
o I
0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.61 Rolling torque (carbon steel)
The effect of various lubricant viscosity on the rolling force and torque is also
considered. The lubricants used in the experiment are Rolkleen 485 which is being used
at a local cold mill, mineral oil AWS 10, AWS 100, and ALPHA SP 1000, respectively.
The different lubricants have little effect on the rolling force and torque at 28.75%
reduction shown in Figures 4.62 and 4.63. The rolling torque increases slightly as
rolling speedrisesas shown in Figure 4.63.
Carbon steel BHP-300, e =28.75%, lubricated

800
• • *
2: 750 * * *
o
700 • 0.0086 Pa.S
CO • 0.058 Pa.S
c
=S 650 0.089 Pa.S
&
• 0.93 Pa.S
600 i i _,
0.05 0.1 0.15 0.2
Rolling speed (m/s)
Figure 4.62 Effect of lubricant on rolling force
Carbon steel BHP-300, e=28.75% lubricated
1
± 7 •
* A
• u • 0.0086 Pa.S
o-
o • 0.058 Pa.S
OI 0.089 Pa.S
0.93 Pa.S
§.
1 1 •~r
0.05 0.1 0.15 0.2

Rolling speed (m/s)
Figure 4.63 Effect of lubricant on rolling torque
4.6.2 Effect of rolling parameters on forward slip
The forward slip is an active and sensitive parameter in rolling process. In the
calculation of average friction coefficient from the Laser Doppler method and the strip
marking method, forward slip is an important factor. It is necessary to know the
relationship between forward slip and other rolling parameters. Normally, forward slip
increases with reduction rising, and the forward slip under dry rolling condition is
higher than that value under lubricated rolling condition shown in Figure 4.64. This is
caused by the fact that friction coefficient under dry condition is higher than that value
under lubricated condition.
Aluminium alloy 6060-T5, u r =0.118m/s

12
• Dry
10
• Lubricated •
8
•
in 6
•o w •
•
k 4 4
n m
2 • B
£ •
0 •
12 22 32 42
Reduction (e%)
Figure 4.64 Effect of reduction on forward slip
Figure 4.65 shows that the forward slip decreases as rolling speed increases for most
reduction, and the higher the reduction is, the higher the forward slip at low speeds.
Well, for some reduction e.g. 1 8 % and 2 7 % , the forward slip increases slightly as
rolling speed rises at lower speed range (<0.2 m/s).
Aluminium alloy 6060-T5, lubricated

8
••.
6
• 38%
• 27%
4
5 • 18%
km
CO • t 11%
* 2 -XXx X- -x-
s.
0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.65 Effect of rolling speed on forward slip

4.6.3 Strip surface roughness in rolling process
In order to study the effect of surface roughness on the rolling process, some of steel
strips were processed either by sand-blasting or pickling to achieve various surface
roughnesses. Actually, it is difficult to control the material's surface roughness durin
processing. The roughness of as-supplied steel strip is around (1.0-1.55) pm, sand-
blasting (4.92-6.06) pm, pickled (1.36-1.90) pm, and grinding (0.2-0.34) pm before
rolling. Carbon steel BHP-300 was used in the experiments. The results in Figures 4.66
and 4.67 show that the rolling force and torque increase as the surface roughness rises.
As the friction at the interface increases with surface roughness rising, rolling load a
increases due to a higher surface friction. When rolling speed increases, the rolling fo
and torque increase slightly.

Carbon steel BHP-300, e = 1 0 % Lubricated
380
=-360
340 -•— Ra: 5.1-6.75um

-•—Ra: 3.46-5.45um
& Ra: 0.90-2.05um
-•— Ra: 1.77-2.98um
320
10 20 30 40 50 60
Speed (rpm)
Figure 4.66 Effect of roughness on rolling force

Carbon steel BHP-300, e = 1 0 % Lubricated
~
E
±
a.
o
3
-•— Ra: 5.1-6.75um
? 1 -•— Ra: 3.46-5.45um
Ra: 0.90-2.05um
-•— Ra: 1.77-2.98um
0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.67 Effect of roughness on rolling torque

The roughness could change after rolling because roll and strip surface roughness
transfer to each other during rolling. The roll surface roughness gradually increases as
rolling progress proceeds. In Figures 4.68 and 4.70, the strip surface roughness along
the rolling direction increases after rolling because the roll surface roughness (0.68
is higher than the strip surface roughness prior to rolling. However, the roughness in t
transverse direction does not change significantly, as shown in Figures 4.69 and 4.71.
Carbon steel BHP-300 u r =0.118m/s, lubricated, rolling

direction
0.8
0.6
• before rolling
2 0.4
• after rolling
3
0.2
& ~\*_*-V
10 20 30 40
Reduction (E%)
Figure 4.68 Strip surface roughness along the rolling direction
Carbon steel BHP-300, u r =0.118m/s, lubricated,

transverse
E 0.8
K 0.6
w
w 0.4
<u
O) -•— before rolling
c
| 0.2
-•— after rolling
10 20 30 40
Reduction (s%)
Figure 4.69 Strip surface roughness along transverse direction

Carbon steel BHP-300, 8=19%, lubricated, rolling

direction
0.8
| 0.6
-•— before rolling
% 0.4 -•— after rolling
oi
| 0.2
0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.70 Strip surface roughness along rolling direction
Carbon steel BHP-300, s = 1 9 % lubricated, transverse

-1
~ 0.8- -#"*-._•. -•• *• A _ -•

E
^
10
0.6 ^ 7*^^\Ax~<
10
V
c 0.4- — • — before rolling
£
OI — • — after rolling
I 0-2-
u (- . ... | . , ,
() 0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.71 Strip surface roughness along transverse direction
In Figure 4.68, the roughness along rolling direction increases with reduction rising
after rolling, and the roughness begins to drop after 25% reduction. The roughness
along transverse direction increases with reduction rising after rolling shown in Figu
4.69. The main trend of roughness along rolling direction drops with speed increasing i
Figure 4.70, and the roughness along transverse direction drops at higher rolling speed
as shown in Figure 4.71. It can be seen in Figure 4.72 that the higher the reduction is
the higher the strip surface roughness, and the roughness increases at a lower rolling
speed for 28% reduction.

Carbon steel BHP-300, lubricated, rolling direction

X
X X X X
x><
I 2 • 7%
10 • • 11%
in • • * • 1
ai • 18%
ot 1 tB •
X 28%
0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.72 Strip surface roughness along rolling direction
During experimental process, it is also found that the rougher the surface roughness is,
the higher forward slip for the same rolling speed, and the forward slip decreases when
rolling speed increases as shown in Figure 4.73. And the same trend also applies for the
relationship between friction coefficient and rolling speed in Figure 4.74. It should be
noted that the surface roughness was obtained by commercial methods such as grinding,
pickled, and sand blasting.
Carbon Steel BHP-300,e =11.0%, lubricated
j. 1.5
to
•o
-•— Ra:5.1-6.75um
1_
W
-•—Ra:3.46-5.45um
<t
-H-^jp Ra:1.77-2.98um
k.
o - X — Ra:0.90-2.05urrj
0.5
0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.73 Effect of roughness on forward slip

Carbon Steel BHP-300, e=11.0%, lubricated
0) m%
- Ra:5.1-6.75um
•
o -Ra:3.46-5.45um
u Ra:1.77-2.98um
c *—*- * — - ^ Ra:0.90-2.05um
"--^-* X
o
0 0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.74 Effect of roughness on friction coefficient
4.6.4 Temperature in cold rolling
The temperature rising of strip and roll in cold rolling is mainly due to the plastic
deformation work and the friction at strip and roll interface. Detail of how the pla
work and friction affect the temperature of strip and roll will be explained in Chap
from a theoretical point of view. In this section, a series of experiments were carr
to measure strip and roll surface temperatures during rolling, and the roll surface
temperature along the roll bite was also measured by a thermocouple embedded in the
sensor roll. Strip and roll surface temperatures were measured by a hand held thermal
meter as soon as the rolling mill stops. The factors effect on strip and roll temper
will be discussed.
In Figure 4.75, the strip surface temperature under dry and lubricated condition
increases with reduction increasing, and the temperature under dry condition is high
than that under lubricated condition. Such phenomena are due to the fact that frictio
under dry condition is higher than the friction under lubricated condition when other
rolling parameters are the same.
Aluminium alloy 6060-T5, uP=0.118m/s

RO
• Before rolling
5* • Dry condition •
^ 60- Lubricated condition
k.
3 •
(0 •
5 40- a
o. •
E ? _ • • • • • • • • •
4)
s
20
Q.
V)
0- ! 12 22 32 42
Reduction (s%)
Figure 4.75 Strip surface temperature vs reduction
Aluminium alloy 6060-T5, e = 1 8 %

60
•
•
,-. 50 •
O •
•
*r 40
ro 30
• •
«3 •:•• • • • •
Q. 20 • •
E • Before rolling
0) 10 • Dry condition
Lubricated condition
,
0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.76 Strip surface temperature vs rolling speed
The strip surface temperature also increases as rolling speed rises for both dry
lubricated rolling conditions, and the temperature under the dry rolling conditi
higher than the value under lubricated rolling condition as shown in Figure 4.76
rolling speed increases, deformation strain rate will increase too. The heat gen
deforming body can be expressed as [Kobayashi et al, 1989]:
r = r<Âa (4-18)
where y is heat generation-rate, <r,y stress, stj strain rate, and y represents the fraction
of mechanical energy transformed into heat. From Eq. (4-18), it can be seen that more
heat is generated with increasing rolling speed. Finally, the strip temperature rise with
increasing rolling speed too.
In Figure 4.77, the strip surface temperature rises as rolling speed and reduction
increases, and roll surface temperature is just raised slightly at higher rolling speed a
reduction as shown in Figure 4.78. The heat energy is mainly produced by plastic
deformation of the strip during rolling. But at roll surface, the heat energy is absorbed
by the roll itself. The roll surface temperature is lower than the strip surface temperat
at the same rolling speed and reduction as it can be seen in Figures 4.77 and 4.78. When
plastic deformation increases, more heat is generated, and the temperature will rises for
both roll and strip surface.

80
a "
o 60
a) o
a
re cu a
o *~
pt-
5
>_
• • 0 0
i 40 X
• X x
a.
to E
a)
io re
•>z a
20 ML t t $ • *
• 3efore rolling a 38% 27% • 18%X11%
r I
0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.77 Strip surface temperature vs rolling speed


40
o 30
8^ ji*. T *
*
T
surl
ratu
J? 2. 20
— fl)
•*• 10
0)
• Before rolling a 3 8 % 2 7 % * 1 8 % x 11 %
0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.78 Roll surface temperature vs rolling speed
The roll and strip surface temperature variation for aluminium alloy 6060-T5 and
carbon steel BHP-300 have been compared as shown in Figures 4.79 and 4.80 at
constant rolling speed 0.118m/s. The roll and strip surface temperature of carbon steel
are always higher than the temperature of aluminium alloy under the same plastic
deformation. This is because the yield strength of steel is higher than that of aluminiu
alloy for the same reduction, and heat in cold rolling is mainly from plastic deformati
so more heat will be generated during steel rolling process. The experiments were
carried out at 0.118 m/s rolling speed under different reduction for carbon steel and
aluminium alloy.
Strip temperature variation, lubricated, ur=0.118m/s

50
o
*d
40
ra
k.
£
> 30
<l> rr-
"
3 °O
re
k.
20
0)
Q.
E 10 • Aluminium alloy 6060-T5
0) • Carbon steel BHP-300
II *
12 22 32 42 52
Reduction (s%)
Figure 4.79 Strip surface temperature variation

Roll temperature variation, lubricated, ur=0.118m/s

5 -,
c • Aluminium alloy
o 6060-T5 |
re
=k. 4 - • Carbon steel
re BHP-300 •
>a> ~> 3 -
k. o •
£~ 2
re
k.
e
- m.
<u
E
2- 1 -. - • • *
'-* —r- , , ,
12 22 32 42 52
Reduction (8%)
Figure 4.80 Roll surface temperature variation
Roll surface temperature during rolling process is also measured by a thermocouple
embedded in sensor roll. The different reductions under 0.118m/s rolling speed and
condition were tested. From Figures 4.81 to 4.85, it can be seen that roll surface
temperature changes over the whole strip length. When the thermocouple pin embedded
in the sensor roll touches the roll bite, the temperature rises sharply to a peak va
after thermocouple pin leaves the roll bite, the temperature drops slowly, and the
temperature rises again when the pin enters the roll bite the second time. This
experimental work will be used to validate the theoretical calculation in Chapter 8
compare with Jeswiet and Zhou's work [1991].
Aluminium alloy 6060-T5, 8=8.43% uf=0.118m/s, dry

25
o 24.5
^
ii
k.
3
*d 24
k.
re
0)
Q.
E
Q»
23 5
h-
23
10 12 14 16 18
Time (s)
Figure 4.81 Roll surface temperature (8.43%)

Aluminium alloy 6060-T5, e=13.82%, ur=0.118m/s, dry

27
iC 26
to 25
k.
0)
24
<D
WW
23
11 13 15 17 19
Time (s)
Aluminium alloy 6060-T5, e =19.16%, ur=0.118m/s, dry

30
22
10 12 14 16 18 20
Time (s)
Aluminium alloy 6060-T5, 8=24.60% ur=0.118m/s, dry

31
O 29
0)
k.
I 27
k-
Q> Y"*«Nwpmi imnman.
a.
| 25
23
15 17 19 _ 21 23 25
Time (s)

Aluminium alloy 6060-T5, e =31.98%, ur=0.118m/s, dry
4.6.5 Friction coefficient in cold rolling
There are three methods (Sensor roll method, Strip marking method, Laser Doppler
method) described in previous section 4.5.2 to measure friction coefficient in cold
rolling. The first one is used to measure friction coefficient point by point over roll
and other two methods are used to determine the average friction coefficient from the
measurement of forward slip. The strip marking method is used to measure the average
friction coefficient for most of the experiments in this thesis. Aluminium alloy and
carbon steel were used in the experiment to measure friction coefficient. The mean
value of friction coefficient from the sensor roll method will be calculated from the
values over the roll bite, and compared with the average friction coefficient from the
marking method.
L Aluminium alloy
Aluminium alloy 6060-T5 w a s used in the experiment to measure friction coefficient.
Both friction coefficient values from sensor roll method and strip marking method are
shown in Figures 4.86 and 4.87. It can be seen that friction coefficient value decrease
with rolling speed, and the friction coefficient values from the two methods are close
each other and drops quickly at low rolling speed range in Figure 4.87.
Aluminium alloy 6060-T5, 8=27.36% lubricated

0.115
g 0.105 + • sensor roll -

o • marked method
% 0.095 • • *
o t
o
§ 0.085 4 * •
: -
§ 0.075
0.065 -I • - T • - T " " " 1 " 1

0.2 0.4 0.6 0.8
Rolling speed (m/s)
Figure 4.86 Friction coefficient vs rolling speed (27.36%)
Aluminium alloy 6060-T5, 8=38.94%, lubricated

0.145
•• • sensor roll method
•a
S 0.125 • a marked method
o
E
o 0.105
o *
*
c
o » 1
tj 0.085
0.065 r 1 1 • — T
0.2 0.4 0.6 0.8

Rolling speed (m/s)
Figure 4.87 Friction coefficient vs rolling speed (38.94%)
II. Carbon steel
The steel strips with ground surfaces roughness (0.25-0.34) p m are used in the
experiment. A group of experiments were carried out for 0.058 Pa.S viscosity lubricant
under 22.51% reduction as shown in Figure 4.88. Another group of tests were
performed with pickled strip using the same viscosity lubricant and (1.36-1.90) p m
surface roughness under 30.80% reduction as shown in Figure 4.89. It can be seen from
these two figures that the friction coefficient measured from the sensor roll and strip
marking methods decreases as rolling speed increases.

Carbon steel BHP-300, e =22.51%, grinding, 0.058 Pa.S
0.105
• sensor roll method
c 0.095
• •
a marked method
0) • a •
o a
E 0.085
* *
8
c
o
£ 0.075 ,. , ,.,
0.065
0.1 0.2 0.3 0.4
Rolling speed (m/s)
Figure 4.88 Friction coefficient vs rolling speed (ground)
Carbon steel BHP-300, 8=30.80% pickled, 0.058 Pa.S

n 11^
•
c 0.105 -
.2
| 0.095- * ; • • •
o 4
•
o
c 0.085 -
o • sensor roll method
£ 0.075 - • marked method
c) 0.1 0.2 0.3 0.4 05

Rolling speed (m/s)
Figure 4.89 Friction coefficient vs rolling speed (pickled)
The following Figures 4.90 and 4.91 show the friction coefficient with different
lubricant viscosity. The experiments were carried out with ground carbon steel strip
under 30.64% and 30.97% reduction respectively. Two different oil viscosity's 0.0086
Pa.S and 0.93 Pa.S were used in the experiments. The same trend of friction coefficient
decreasing with higher rolling speed was observed again, the higher the lubricant
viscosity is, the lower the friction coefficient value.

Carbon steel BHP-300,e =30.64% grinding, 0.0086 Pa.S

0.145
• sensor roll
0.125 •
a marked method —
0)
u •
0)
0.105 +
8c a •
•
0
0.085
s
0.065 1 1 1
0.05 0.1 0.15 0.2

Rolling speed (m/s)
Figure 4.90 Friction coefficient vs rolling speed (0.0086 Pa.S)
Carbon steel BHP-300, 8=30.97%, grinding, 0.93 Pa.S

0.105
•
0.095
o *
*
g 0.085
o
*j 0.075 4 • sensor roll method

a marked method
0.065 _ . \ ,
0.05 0.1 0.15 0.2
Rolling speed (m/s)
Figure 4.91 Friction coefficient vs rolling speed (0.93 Pa.S)
It can be seen from the above figures that friction coefficient value decreases when
rolling speed increases for all experiments. The average friction coefficient value from
the sensor roll is close to that from the strip marking method, this means that friction
coefficient value obtained from the sensor roll is reliable as confirmed in the next
chapter against theoretical calculations.

4.7 Friction coefficient empirical formulae
It has been found out that friction coefficient is not constant over the roll bite. This
quite different from the conventional rolling theory thatfrictioncoefficient is assumed a
given constant value in the calculation. So it would be valuable if the empirical
formulae of friction coefficient over the roll bite can be obtained. In order to carry out
this work, a statistical software-JMP is used to calculate that formula. After trial and
error, it was found that friction coefficient variation over the roll bite satisfies the fifth
order polynomial. If the formula is used in the calculation, the absolute value should be
given. The final result for aluminium alloy and carbon steel is listed as below
respectively. Y represents the friction coefficient value, and X represents the
dimensionless roll bite length. X is equal to 0 at the roll bite entry point. The detail
calculations of empirical formulae offrictioncoefficient are shown in Appendix B.
I. Aluminium alloy
The formulae of friction coefficient distribution over roll bite were obtained to base on
the Figures 4.32, 4.34, 4.36, 4.38, 4.40, 4.42, and 4.44. The detail expression of each
formula will be written one by one as follows:
(1) Aluminium alloy 6060-T5, 36.82% reduction, and 3 rpm rolling speed under
lubricated condition. The lubricant is Rolkleen 485.
/i = -0.311 + 1.1196X-5.7511X2 +15.073X3 -16.2347X4 +6.3158X5 (4-19)
p = -0.1912 -0.5218X + 4.8515X2 -14.2263X3 +17.5845X4 -7.3474X5(4-20)

p = -0.2297 -0.0559X + 3.9758X2 -13.5358X3 + 17.5897X4 -7.5600X5(4-21)
p = -0.2425 + 0.4576X + 0.6304X2 -3.8617X3 +5.7588X4 -2.6251X5 (4-22)
p = -0.4215 + 2.6328X-9.0353X2 + 15.4036X3 -11.7617X4 + 3.2821X5 (4-23)
p = -0.401 + 1.5333X-2.384X2 +1.3227X3 (4-24)
(7) Aluminium alloy 6060-T5, 40.12%) reduction, and 65 rpm rolling speed under
p = -0.129 -0.0135X + 0.2205X2 (4-25)
The fourth and fifth order in Eq. (4-24), the third, fourth, and fifth order in Eq. (
are found to be negligible. This is caused that a few data points were acquired in the roll
bite because the rolling speed was too fast.
The comparison of measured and predicted data for Eq. (4-19) is shown in Figure 4.92
and the rest of them are in Appendix B. It can be seen that both data are identical.
(Y By X
0,2
0,1 -
-0,0 -
-0.1 -
-0.2 -
-0,3
,0 ,1 1.1
Poly nomial F it degree=5
(Poly nomial Fit degree=5 j

Y =-0,311 +1,11955 XQ 5.75113 X-2 + 15.073 X-3 0 16.2347 X-4 + 6.31578 X-5
">
(Summary of Fit J
RSquare 0,987241
R Square Adj 0,984963
Root Mean Square E rror 0,017159
Mean of R esponse +108662
Observations (or S u m W gts) 34
)
(Analysis cf V jriance
Source DF
ZD
S u m of Squares Mean Square F Ratio
Model 5 0.63789339 0.127579 433,3124
Error 28 0,00824394 0,000294 Prob>F
C Total 33 0.64613734 C0001
V
/
(Parameter Estimates J
Term Estimate Std Error t R atio Prob)J]t|
Intercept -9.31099 0,023364 -13,31 C00Q1
X 1.1195493 0,434905 2,57 0,0156
M -5.751129 2,525276 -2,28 0,0306
M 15,072977 6.117701 2,46 0,0202
M -16,23474 6,513324 -2.49 0,0189
X-5 6.3157849 2,519172 2.51 0,0183
Figure 4.92 Comparison of measured and predicted friction coefficient (n=3rpm)
Eqs. (4-19)~(4-25) were obtained under nearly same reduction and different rolling
speeds. So if rolling speed factor is also considered, the empirical formula of friction
coefficient for aluminium alloy under 3 9 % reduction can be given as:

p = -0.2705 + 0.7360X- 1.0703X 2 + 0.0251X 3 +1.8481X 4 - 1.1335X5 - 0.0005/wr
(4-26)
An example with 7rpm rolling speed was chosen to compare measured friction
coefficient and predicted friction coefficient from Eq. (4-26). The graph is shown
Figure 4.93. It can be seen that two group data are close.
Aluminium alloy, 8=37.61%, ur=0.0824m/s
-•— measured data

-•— predicted data
Roll bite (((./(j),)
Figure 4.93 Comparison of measured and predictedfrictioncoefficient
II. Carbon steel
The formulae will be given for the friction coefficient distribution over roll bi
Figures 4.46, 4.48, 4.50, 4.52, and 4.54. The detail expression of each formula w
written one by one as follows:
(1) Carbon steel BHP-300, 31.16% reduction, and 3 rpm rolling speed under
lubricated condition. The lubricant is a mineral oil ALPHASP 1000.
p = -0.3302 + 1.9983X-7.3978X2 +14.3123X3 -13.0121X4 +4.5385X5(4-27)

/w = -0.1795-0.0278X + 1.0225X 2 -1.5978X 3 +1.3248X 4 -0.4793X 5 (4-28)
[Y By X ]
1.2
Poly nomial Fi jree=5
(Poly nomial F it degree=5 j

Y =-0.3302 + 1.99831 X 0 7.39784 X-2 + 14.3123 X-3 0 13.0121 X~4 +4.53848 X~5
\
[Summary of Fit J
RSquare 0.981178
R Square Adj 0,977087
Root Mean Square Error 0.013956
Mean of Response -0.05748
)
[k no ly s isof Vo riance
Source DF
ZD
Model 5 0.23352181 0.046704 239.7996
Error 23 0.00447957 0.000195 Prob>F
C Total 28 0.23800139 C0001
\
(Porameier Estimates J
Term Estimate Std Error t R atio Prob>|t|
Intercept -0.330193 0.021697 -15.22 C0001
X 1.9983138 0.39808 5.02 C0001
X-2 ^.397841 2.286138 -3.24 0.0037
X-3 14.312254 5.495509 2.60 0.0159
X-4 -13.01212 5.816332 -2.24 0.0353
X-5 4.538476 2.238753 2.03 0.0544
J
Figure 4.94 Comparison of measured and predicted friction coefficient (n=3rpm)
lubricated condition. The lubricant is a mineral oil A L P H A S P 1000.

p = -0.1651 + 0.3139X-0.1365X 2 -0.5024JT 3 +1.1614JT4 -0.5689X 5 (4-29)
p = -0.1897- 0.0109X + 2.6734X2 -6.7318X3 +6.2728Z4 -1.871 IX5 (4-30)
p = -0.2381 + 0.7299X- 1.5901X2 +3.8199X 3 -5.1937X 4 +2.5868X 5 (4-31)
The calculation detail for Eq. (4-27) is shown in Figure 4.94. The calculations f
rest equations are list in Appendix B. The similar empirical formula of friction
coefficient considering rolling speed factor for carbon steel under 31% reduction
obtained as:
p = -0.2088 + 0.6007Z-1.0857X2 +1.860X3 -1.8894Z4 +0.841 \X5 -0.0009/wr
(4-32)
The measured friction coefficient for 20 rpm rolling speed and predicted friction
coefficient from Eq. (4-32) are shown in Figure 4.95.

Carbon steel, E =32.66%, ur=0.2355m/s
0.2 -i
£ 0.1 - A•
0)
o
t 0- ^ > ^
) 0.2 {^====5*=s"'"a<3 0.8
8 «
1
1 •°-
'•&
o
iE -0.2 j
-0.3 - Roll bite (<j>/<j>,)
Figure 4.95 Comparison of measured and predicted friction coefficient

4.8 Empirical formula of rolling force
In cold rolling process, there are a lot of factors to affect the total rolling force, s
material yield strength, reduction, strip temperature, strip surface roughness, lubricant
viscosity, rolling speed, and friction coefficient. In order to conclude the relationship
between rolling force and these rolling parameters, m a n y experimental data obtained
from laboratory rolling mill were used to calculate the empirical formula. The friction
coefficient is average value calculated from the strip marking method. Finally, the
empirical formula covering eight rolling parameters for carbon steel under lubricated
condition at 0.8 m/s m a x i m u m rolling speed is given by:
P. = 9.6712E10.O-;0148 .£-29151 J26955 .J"00217 .n'00002 ufM21 .p039S0 (4-33)
where
Pt total rolling force, N
o~y material yield strength, P a
£ reduction, %
T strip temperature, °C
8 roughness, m
n viscosity, Pa.S
ur roll speed, m/s
p averagefrictioncoefficient
In order to validate Eq. (4-33), the experimental data was used to calculate total rollin
force from Eq. (4-33), and then compared with measured force. The comparison result
is shown in Figure 4.96. It can be seen that the calculated and measured data are close to
each other.

1000 -j 1
800 ^
600
• r
o •
5 •
r 400 »
ra ?no -
• measured data
• calculated data
0 -I —r— — r - - —, , 1
0 2 4 6 8 10
Test no.
Figure 4.96 Comparison of calculated and measured rolling force
Eq. (4-33) is obtained from experimental data in the laboratory. All strip used in t
experiment is 100 mm wide. Eq. (4-33) is useful to predict the friction coefficient
inverse method (section 3.2).
4.9 Summary
A sensor roll has been designed, manufactured, and used in the experiment. A large
number of experiments were carried out to study friction coefficient and other rolli
parameters such as roughness and lubricant viscosity. The following conclusions can
made as follows:
(1) The Laser Doppler method, strip marking method, and sensor roll method have
been used to determine friction coefficient in cold rolling. The Laser Doppler
method and marking method can only measure the mean value of the friction
coefficient, but the sensor roll technique can determine friction coefficient
variation in the roll bite.
(2) The friction coefficient in the roll bite is not constant

Chapter 4 Friction Coefficient Measurement and Rolling Experiment 121_
(3) N o obvious pressure peak was found over the roll bite, and sometime the location
of the single pressure peak does not precisely coincide with the location of the
neutral point.
(4) The roll surface temperature along the roll bite was measured.
(5) N o multiple pressure peaks described by other authors were found in the
experiment.
(6) The rolling force and torque under dry condition are higher than the case under
lubricated condition.
(7) W h e n the rolling speed increases the rolling force and torque increases slightly.
(8) The roll surface roughness and strip surface roughness can be transferred to each
other along the rolling direction during rolling. But no significant change in
transverse direction for strip was observed.
(9) The effect of different lubricant property on the rolling force and torque is small.
(10) The forward slip increases as the reduction increases, and the forward slip under
dry condition is higher than under lubricated condition. The forward slip
decreases at the higher rolling speeds. The higher strip surface roughness is, the
higher forward slip.
(11) The finished roughness of strip surface increases with increasing reduction.
(12) The strip surface temperature increases with increasing reduction and rolling
speed.
(13) The friction coefficient decreases with increasing rolling speed for most of rolling
experiment, and the average friction coefficient values from sensor method and
marking method are close to each other.
(14) Empirical formulae of friction coefficient for certain rolling condition were
obtained. T h e rolling speed factor was also included.
(15) Empirical formulae of rolling force were also obtained.

Chapter 5 Validation ofFriction Coefficient Measurement in Cold Rolling 122
Chapter 5
Validation of Friction Coefficient
Measurement in Cold Rolling
5.1 Introduction
In both rolling theory and practice, two important factors must be considered: friction
coefficient and deformation resistance yield strength. The traditional approach to
friction is to assume that the friction force in the roll bite is proportional to the normal
force, with the friction coefficient remaining constant in the roll bite. But this will incur
a loss of accuracy in the roll gap model, and affect the thickness and shape of the strip.
In order to understand the friction mechanism in cold rolling, m a n y research efforts
have been m a d e in both experimental and theoretical modeling [Rooyen and Backofen,
1957; L i m and Lenard, 1984; Chang et al, 1996; Qiu et al, 1999].
Direct and indirect methods have been used in the measurement of friction coefficient.
In the direct method, sensors are embedded in the roll to measure forces, which are used
to determine the friction coefficient. In the indirect method, forward slip is first
measured, and then used to calculated thefrictioncoefficient as [Ford and Bland, 1951].
In chapter 4, the friction coefficient has been determined by using the direct method-
sensor roll and the indirect methods-strip marking method and Laser Doppler method.
In this chapter, the measured friction coefficient will be used in the traditional rolling
theory-Orowan's equation [Orowan, 1943] to calculate rolling force and torque which
will be compared with measured values to confirm the validity of the measured friction
coefficient. Tests will be carried out to determine the yield strength of experimental
specimens which are needed in the calculation of rolling force and torque.
5.2 Yield strength
In the theoretical calculation, the material's yield strength should be determined before
any calculation. A mathematical expression used by Alexander et al [1987] in the
calculation was:
o-y=ayo(l + C3£t)c<(l + C5£t)c<- (5-1)
where ay is the yield strength associated with the equivalent true strain £t and true
strain rate £t. In plane strain compression the yield strength 2ks =2<ry /V3 = l.l5cry
and the effective true strain £t is 2/V31n(v0 / v), the strain rate £t is
1.15x(2ww ,y2tan^)/v2. v0 is the initial thickness of strip and v is the strip thickness
in the roll bite, and at the exit, y is equal to y 2 , uw is the strip speed at exit, and (/> is
the angle from exit plane. In cold rolling, y0 is the thickness at which the strip was la
in its annealed state. In the experiment, all test material were in annealed state, so y
equal to yx in this thesis. But the effect of strain rate on the yield strength can be
neglected in cold rolling, and in particular, the tests on the Hille 100 rolling mill with
rolling speed ur <l m/ s .So Eq. (5-1) can be written as [Swift, 1940]:
cry=CTyo(l + C3£t)C> (5-2)
Three different materials, aluminum alloys 5052-H34 & 6060-T5, carbon steel BHP-
300 were used in the rolling experiment. A n expression of aluminium alloy 5052-H34
for the material's resistance to deformation similar to that by Karagiozis and Lenard
[1985] was used in the calculation as shown below.
ay =199.60(1 + 201.8fr)0097 (5-3)
But the yield strengths for aluminium alloy 6060-T5 and carbon steel BHP-300 were
not known. Therefore, the test was carried out to obtain expressions of yield strength for
these two materials.
A test was carried out on the tensile machine to test engineering stress S and
engineering strain e [Storer et al, 1997]. The engineering stress is expressed as:
S = Fl/A0 (5-4)
where F, is an applied load, and A0 the original cross-sectional area of sample.
The engineering strain can be written as:
e = AL/L2 (5-5)
where A L is the change in length, and L2 the original linear dimension along the
loading axis of the specimen.
The true stress is the instantaneous normal stress, calculated on the basis of the
instantaneous cross sectional area A' ; that is, a• = F, I A'. If no necking has occurred,
the true stress is related to the engineering stress as follows:
cry=S(l + e) (5-6)
The true strain is the natural logarithm of the ratio of instantaneous gauge length L" t
the original gauge length L2, that is,
£t=\n(L"IL2) (5-7)
or
*, =ln(l + e) (5-8)
The test sample and its dimension are shown in Figure 5.1.
The test result of engineering stress against engineering strain for carbon steel BHP-3
is shown in Figure 5.2. The true stress against true strain is shown in Figure 5.3. Similar
graphs for aluminium alloy 6060-T5 are shown in Figures 5.4 and 5.5 respectively.
200
50 50
%
y TOT
Figure 5.1 Test sample dimension
BHP-300
400
c/)
to
10 300
 400 J
IB
in
in /
a> 200 -
»-.
in
a
3 0-
c) 0.1 0.2 0.3 04
True strain, ct
Figure 5.3 True stress and strain (carbon steel)

Aluminium alloy, 6060-T5

300
co
5 10 15 20
Engineering strain, e %
Figure 5.4 Engineering stress and strain (aluminium alloy)
Aluminium alloy, 6060-T5

300
0.05 0.1 0.15 0.2

True strain, st
Figure 5.5 True stress and strain (aluminium alloy)
In Eq. (5-2), there are three unknown parameters ayg, C3, and C4 which have to be
determined. Eq. (5-2) can be transformed info the natural logarithm expression as
follows:
In<r = In aVny<>+ C4 ln(l + C3st) (5-9)
First, in order to solve the three unknown parameters for carbon steel, n points relating
to the true stress and the true strain will be chosen in the plastic deformation ar
Figure 5-3, e.g. (<x„,^ ), 0,2,^2), , (°V.>**.)• C* is

&wen a known series
values (Cl,C2, ,C 3 m ), and then the number of parameters

[{CT\O , o-2o, , ayo); (C4, C2, , C4m)] can be solved from Eq. (5-9) related to a
known series values {C\,Cl, ,C3m). A series of (crj,,cr2, ,ay) expressions can
be obtained. In order to determine the correct value of C3, a group of residuals can
calculated as follows:
*1=S«-*»)2 (5-10)
R2
=lL«-VyX C5"11)
i=\
i=\
After the calculation, a group of residual value (R\R2, ,Rm) can be obtained.
Therefore, the graph of residual R versus C3 can be drawn as shown in Figure 5.6.
C3
Figure 5.6 Relation of R and C 3

W h e n the residual R reaches the minimum value in Figure 5.6, the correct value of C 3
is obtained, and the corresponding ayo and C4 are also obtained by using Eq. (5-9).
Finally, the expression of yield strength can be calculated. The result for carbon ste
BHP-300 is as follows:
0.3449
<r =238.40(1+ 23.87*,) (5-13)
The same procedure can be applied to the data in Figure 5.5 for aluminium alloy 6060-
T5. Thefinalresult of yield strength expression for aluminium alloy 6060-T5 can be
wntten as:
a= 155.34(1 + 26.57*,)0.3056 (5-14)
In Figures 5.7 and 5.8, it has been shown that the calculated yield strength values (true
stress in thefigures)are close to the measured values in the plastic deformation part. So
it confirms that Eqs. (5-13) and (5-14) are correct and reliable.
Carbon steel BHP-300

600-
500 -
0.
in d. •
in 400-
01
ki
a
*d
in 300 - . ' • Experimental data
41
3 a Calculated data
SM
H 200- - i • i "• i
(3 0.1 0.2 0.3 04

True strain
Figure 5.7 Validation of yield strength (carbon steel)

Aluminium alloy 6060-T5

300
a. 250
in
2 200 a »
in
ai
150 • Experimental data —
a Calculated data
100
0.05 0.1 0.15
True strain
Figure 5.8 Validation of yield strength (aluminium alloy)
5.3 Theoretical considerations
Neutral j Elastic
plane ! |^| x , c
zone
Figure 5.9 Plastic work zone in rolling

Chapter 5 Validation of Friction Coefficient Measurement in Cold Rolling 131
Consideration of the equilibrium of an elemental slice of material in the plastic work
zone described in Figure 5.9 leads to the basic linearfirst-orderdifferential equation
first put forward by von K a r m a n [1925]. With some minor changes in the notation used
by von K a r m a n and subsequent workers [Orowan, 1943], an equation for the force
equilibrium of an elemental slab can be written as
d[y(p - 2ks + T, tan 0)] / d(/> = 2R' (p sin ) (5-15)
where $ is the .angle of the elemental section considered as shown in Figure 5.9, y the
local strip thickness, p the local normal pressure on the deformed roll surface, ks the
shear yield strength at the considered section, zj the surface shear stress at the
elemental section, R the original roll radius and R' the radius of deformed arc of
contact (assumed circular). In Eq. (5-15) the upper algebraic signs refer to the exit side
of the neutral plane, while the lower signs apply to the entry side.
From the geometry of the deformed arc of contact, the variation of strip thickness y is
expressed by
y = y2 +2R'(l- cos (/)) (5-16)
In cold rolling, the friction at strip/roll interface is normally considered as slipping
condition. So frictional stress is proportional to the normal pressure:
T,=MP (5"17)
Substituting Eq. (5-17) into Eq. (5-15) leads to the differential equation [Alexander,
1972].
T7 = S,(0/> + *2(0 (5"18)

where
2R% \
gi(^) = ±//sec0 + secc^ /(l + //tan^) (5-19)
V y )
2R' d(2k )
*,(*) = K sJ (5-20)
y 2kssm(f>+ /(l + //tan^)
In all these equations the uppermost of any pair of algebraic signs refers to the exit side,
the lower to the entry side.
The horizontal and vertical compressive stresses on the strip in the plastic work zone are
related by the von Mises yield criterion:
o-2-ax= 2k. (5-21)
and the vertical compressive stress relates to the rolling pressure and roll/strip surface
shear stress as:
cr2 = p + rt tan <f> (5-22)
and substituting the above value of cr2 into Eq. (5-21) gives the horizontal stress at any
section as
ax = p - 2ks + zt tan < (5-23)
Thus, at the entry of plastic work zone, when </> = </>lt 2ks = 2kh, and ax = -t^ , the roll
pressure at the entry point is given by the expression
/>i=2£. -t. -rtan (5-24)
and at the exit, ^ = 0,
Pi=2ks, s 2 -*~e2 (5-25)

On the assumption that the deformed arc of contact is circular and neglecting the e
effect, the rolling force and torque can be written as [Alexander, 1972]:
Pt'=WR'£pcos(^-0.5^)d</> + WR' f tt sim>-0.5^)d(f>- |" r, sin(^-0.5^)d<j>
(5-26)
Gt' = WE(R'-R) § p sin(c^ - 0.5^ )dtp + WR (' [R' r. - (R'-R)Tt cos(ctf - 0.5^ )W
- WR! £ [R' r, - (R-RX cos(^ - 0.5^ )]d<f>
(5-27)
Actually, the elastic zones at entry and exit have an effect on rolling force and t
Ford et al. [1951] developed the following equations to consider these effects.
(1-v )y, R'(2ks -te)

K s ,yx
P =w — . — — (5-28)
4 \yx-y2 Es
2-
P W Ml Vs)
,= lf E~ ^-A (5-2")
where P , P are the contributions of the entry and exit elastic zone to the rolling force
respectively, W strip width.
and
tei=tx-2pPei/(yxW) (5-30)
tH=t2-2pPH/(y2W) (5-31)
The tension te and te themselves depend on Pe and P , so Eqs. (5-28) and (5-29)
have to be solved by iteration. The contribution of the elastic zones to the torqu
given by the following equations:
Gei=pX'Pei (5-32)
Gei=-pR'Pe7 (5-33)
Thus, the total value of the rolling force is given by
Pt = Pt'+Pei + Pe2 (5-34)
and the total value of the rolling torque per roll by
Gt=Gt'+Gei+Ge2 (5-35)
The deformed roll radius including elastic effect was given by [Alexander, 1972] as
shown below
(
C P
R'=R 1 + ôrt (5-36)
\
WqA + Ae2+Al+J\)
where
& = yl-y7 (5-37)
Ae = 0 - v 2 ) ( 2 * S-L)y2/Es (5-38)
At=vs(l + vs)(y2t2-yxtx) (5-39)
16(1 -v2)
C„ = (5-40)
KE.
Since both P and P, are relatively small, they can be neglected especially for the
entry force Pe [Alexander, 1972].
In the following calculation, the constant friction coefficients measured by the strip
marking method and Laser Doppler method are used in Eq. (5-17) to calculate rolling
force and torque, and then compared with the measured rolling force and torque. The
variable friction coefficient measured by the sensor roll method is also substituted into
Eq. (5-17) point by point to calculate rolling force and torque in order to verify the
validity of sensor roll measurement. T o do this, the program calculating rolling force
and torque [Alexander, 1972] is modified so that variable friction coefficient values
along the roll bite can be considered rather than a constant friction coefficient value.
The fourth order Runge-Kutta numerical approxiamte method is used to solve Eq. (5-
18). T h e roll bite length will be divided a number of divisions. The following figures
show the relation between the division number and the rolling force and torque value.
From Figures 5.10 and 5.11, it can be seen that the calculated roll force and torque
values reach a stable situation after about 15 roll bite divisions. In the calculation, 21
divisions in the roll bite were set, although more division's number can also be applied
as needed.
800
^ 600
-*
U
o 400
OI
c 200
1 0
0 10 20 30 40 50
Division number in roll bite
Figure 5.10 Rolling force

E 4 / —
± /
3
m f
3 /
5 2 /
D) /
1
0J , , , ,
0 10 20 30 40 50
Division number in roll bite
Figure 5.11 Rolling torque
5.4 Results and discussion
The Orowan's equation will be used to calculate rolling force and torque bas
constantfrictioncoefficient determined from the strip marking method and the variable
friction coefficient measured by the sensor roll method.
5.4.1 Validation of friction coefficient measurement
Firstly, the comparison between the measured and calculated pressure distrib
the roll bite are shown in Figures 5.12 and 5.13. The measured radial pressu
two experiments in these figures were obtained under lubricated and dry cond
aluminium alloy 5052-H34. In Figures 5.12 and 5.13, "measured" means pressur
roll bite measured by radial pin in sensor roll; "strip marking method and L
Doppler" mean the average friction coefficient measured by the strip marking
and the Laser Doppler method respectively used in the Orowan's equation; "se
means the variable friction coefficient measured by radial and oblique pins
the sensor roll used in the Orowan's equation. It can be seen that the curves in the
figures are close to each other except near the exit zone. In bothfigures,the measured
pressure by the radial pin drops more sharply at the exit zone. Such phenomenon was
probably caused by the pin protrusion slightly below roll surface or due to a minor
movement of the segment. So this would result in the measured pressure drop at exit as
shown in Figures 5.12 and 5.13. The measured pressure from the sensor roll does not
have an obvious pressure peak as obtained from the calculation. W h e n the friction
coefficient increases, so does the pressure. In these two graphs, the calculated pressure
derived from the strip marking method is closer to the measured values. It is also
convenient to use the strip marking method to measure the forward slip and determine
the average friction coefficient. However, if it is necessary to determine the friction
coefficient distribution along the roll bite, the sensor roll method is a better tool.
Figure 5.12 Pressure distribution for lubricated condition

Figure 5.13 Pressure distribution for dry condition
The experimental parameters and the results are shown in Table 5.1. For all of these
cases, the friction coefficient value under dry condition is higher than that under
lubricated condition. The friction coefficient from the sensor roll method is the highes
followed by the Laser Doppler method, with the strip marking method giving the lowest
friction values.
Table 5.1 Experimental results
Item yi(mm) y2(mm) 8% u r (mm/s) u., Sensor roll u. Marking u,Laser

No. method method Doppler
method
1 3.080 2.172 29.48 115.4 0.0808 0.0658 0.0725
lub.
2 3.094 2.546 17.71 114.2 0.1314 0.1027 0.114
dry
Table 5.2 Rolling force and torque comparison
Rolling force(kN) Rolling torque(kN-m)

Test
Sensor Marking Laser Measured Sensor Marking Laser Measure
Roll Method Doppler Value Roll Method Doppler Value
No.
Method Method Method Method
1 192.51 170.75 180.70 179.00 1.577 1.358 1.438 1.492
lub.
2 183.01 174.97 176.15 160.00 1.272 1.128 1.223 1.232
dry
In Table 5.2, the calculated rolling force and torque from the experimentally derived
frictions coefficient from the three methods do not differ significantly with the
measured value. As a matter of fact, the maximum difference in force is 14%, and
torque 16%. The Laser Doppler and the strip marking method can only yield a mean
value of friction coefficient, but the sensor roll method can determine the variation of
friction coefficient in the roll bite.
The rolling force and torque comparisons are shown in Figures 5.14 and 5.15. From the
two graphs, it can be seen that the measured rolling load and torque validate the
pressures measured by the sensor roll. Constant friction coefficient was measured by the
strip marking method, and the variable friction coefficient was measured by the sensor
roll method.
Dry condition
^00 \
*-" 900
0)
o a •
$. •
ft
1
c moJ
m
& • Measured value
a Constant friction coefficient
Variable friction coefficient
0- i i
10 20 30
Reduction rate (e%)
Figure 5.14 Rolling force comparison (dry)

Dry condition
o
• IVIeasured value
Rolling torque (kN-m)
a Constant friction coefficient

C
Variable friction coefficient

•
•
p\J
, . *
-*
L> "
3
0 10 20 30
Reduction rate (c%)
Figure 5.15 Rolling torque comparison (dry)
5.4.2 Neutral angle
In the rolling process, the strip surface speed is slower than the roll speed in fr
neutral point so the friction at strip surface follows to the rolling direction. Bu
direction of friction at the strip surface is just opposite to the rolling directio
neutral point. At the neutral point, the strip surface speed is equal to the roll s
speed. Thus, the friction and friction coefficient value should be zero at that poi
shown in Figures 4.32, 4.34, 4.36, 4.38, 4.40, 4.42 and 4.44 for aluminium alloy 60
T5 as well as Figures 4.46, 4.48, 4.50, 4.52, and 4.54 for carbon steel BHP-300. Fr
the above figures, the position of the neutral point can be determined, so that the
angle can be calculated from the roll bite angle. The calculated neutral angle from
Orowan's theory (section 5.3) can be determined when the rolling pressure reaches t
maximum value. A comparison from sensor roll experimental determination and
theoretical calculation are shown in Figures 5.16 and 5.17. In theoretical calculat
friction coefficient obtained from the strip marking method was used.
Aluminium alloy 6060-T5,e =38.94%, lubricated
0)
0)
*- 15 ^r
l o
OI
<D
1
l
OI • sensor roll determination
c a theoretical calculation
ra
2 0.5
3 0.2 0.4 0.6 0.8
Rolling speed (m/s)
o
Figure 5.16 Neutral angle (aluminium alloy 6060-T5)
Carbon steel BHP-300, s =31.27%, lubricated

1.5 -
"ST
0>
OI
m • •
1l
£
OI
• 0.5-
ra
3 • sensor roll determination!
a theoretical calculation
n
0 -) 0.1 0.2 0.3 04
( Rolling speed (m/s)
Figure 5.17 Neutral angle (carbon steel BHP-300)
In Figure 5.16, it can be seen that the neutral angle decreases when rolling speed
increases for both sensor roll determination and theoretical calculation. But
6.8%~21.1% errors exist between the experimental determination and the calculated
value. The error can come from the sensor roll measurement itself in the location of
neutral point, or from the assumption made in the theory. In the theoretical calculation
a constant friction coefficient from the strip marking method is used to determine the
rolling force and torque. The error could occur due to the measurement of the mark
length left on the strip after the roll turns one revolution, which is used to calculat
forward slip. In Figure 5.17, the neutral angle decreases slightly as rolling speed
increases. The error is less than 12.31% which is smaller than the case with aluminium
alloy.
5.5 Summary
The experimental rolling force and torque were compared with the calculated results
based on constant friction coefficient and variablefrictioncoefficient. The computer
program to solve the Orowan's equation also considers variablefrictioncoefficient in
the calculation of rolling force and torque. The determination of thefrictioncoefficient
using a sensor roll has been validated by the experimental results. The m a x i m u m
difference in force is 1 4 % , and torque 1 6 % . These values were found to be reasonable
when compared with theoretical calculations.
The friction coefficient curve measured by sensor roll can be used to determine the
position of neutral point over the roll bite. It has been found that the calculated neutral
angle is reasonable close to experimental values. 6.8%~21.1% errors exist between the
experimental determination and the calculated value for aluminium alloy. For carbon
steel, this error is less than 12.31%.

Chapter 6 Mixed Film Lubrication in Metal Rolling 144
pressure is shared between the contact asperities and fluid film at the surface valleys.
From the asperity crushing rate, thefractionalcontact area and average film thickness
were obtained. The average film pressure was calculated by integrating a simplified
first-order Reynolds equation, with two arbitrary constants to be determined. Wilson &
Chang [1994,1996] developed an analytical model for strip rolling. The sheet surface
roughness was approximated by a longitudinal sawtooth topography and evolution of
surface roughness was estimated, with the relationship between the fractional contact
area, average deformation pressure and the film pressure derived from an upper bound
theory [Wilson & Sheu, 1988].
Qiu et al. [1999] has developed a model similar to that proposed by Wilson and Chang
[1994 and 1996], but with a more rigorous second-order average Reynolds equation,
hence eliminating the need to introduce an u n k n o w n flow constant. The film pressure is
solved from the average Reynolds equation using an over-relaxation method, with an
appropriate boundary conditions applied. Convergent solution can be obtained for
normal rolling speeds (0<ur <lSm/s). The model is then extended to incorporate
variable yield stress characteristics of the workpiece in the roll bite to allow for work-
hardening effect. In this paper, the deformed roll radius is considered, but the effect of
the elastic entry and exit zone is not included. In this chapter, this effect of elastic entry
and exit will be discussed. The "oil drop" experiment was carried out to measure the
lubricant film thickness after rolling. The measured lubricant film thickness will be
compared with the calculated data. The hydrodynamic effect at entry and exit of the
mixed film lubrication model will be considered in Chapter 7.

6.2 Rolling model
Figure 6.1 shows the geometry of a rolling process under mixed-film lubrication, and
the models are governed by the following dimensionless equations [Qiu et al, 1999].
Figure 6.1 Rolling process and sheet surface topography
Horizontal Force Equilibrium. The dimensionless horizontal force equilibrium for an
element in the roll bite (Figure 6-1) is given by:
— = 2R\Psm(j>±QGOS(l>)lyx (6-1)
d(j)
In all these equations the uppermost of any pair of algebraic signs refers to the exit
the lower to the entry side of neutral plane.
Deformation Condition. Under plane strain condition, the von Mises yield criterion
can be used:
pcos<f>-f/y = 2ks (6-2)
but
* . - ^ , («)
thus
pcos^-f/y = l.\5ary (6-4)
Assume the constrained yield stress satisfies Alexander's empirical relation [198
*yô-yWy (6-5)
where W is the dimensionless constrained yield strength given by
Wy=(l + C3 £t )Q (1 + C5 £t )c« (6-6)
with £t and £t being the true strain and strain rate:
where
*,=-iln^- (6-7)
V3 v
^=1.15x(2ww2v2tan^)/v2 (6-8)
substitute Eq. (6-5) into Eq. (6-4), and expresses in dimensionless form
P = (1.15Wy + Fyx I y) I cos <j> (6-9)
Frictional Stress. If the solid contact area ratio is A, the frictional stress between two
roughness can be obtained in dimensionless form:
Q = AQa+Q-A)Qf (6-10)
where the dimensionless frictional shear stress at the contact area [Chang et
is:
Qa=cWy/2sign(uw-ur) (6-11)
The dimensionless shear stress at the fluid valley is:
*' 2ff dtf, ^ h,
The dimensionless contact pressure Pa at asperity contact and the film pressu
the valley should satisfy:
P = APa + (1 - A)Pf (6-13)
Contact Area and Film Thickness. For longitudinal roughness on the workpiece
surface, Chang et al [1996] proposed a formulation of the contact area from
bound analysis:
dA 2<f>R'
(6-14)
d<f> 0a[2l(l-A) + yE']
where
AW-(P-Pf)f2
E<= ^J: UU. (6-15)
(P-Pf)f
fx = -0MA2 + 0.3A5A + 0.515 (6-16)
f2 =l/[2.57l-A-A\n(l-A)] (6-17)
For the saw-tooth roughness as shown in Figure 6-1, the RMS roughness before
has been determined to be

5=\- \\h-r\2dz=rl43 (6-18)
The average film thickness after flattening is:
H, =htld = 43(\-A)' (6-19)
Lubrication Equation. The lubricant pressure for rough surface can be written in a
form similar to those proposed by Patir and Cheng [1979]:
8_ ph^Pj} ph] dp, ^ diOûph,) d(Ouzuph,)

<J>,x + — <D. Y2r] dz dz
dx Yin dx dx (6-20)
pd(Omht) htdp
+ + o.l
dt ~dT
At steady state condition, assuming the lubricant is isoviscous and incompressible with
constant density, and neglecting the axial flow, the average Reynolds equation can be
simplified to:
d_f hf dPf^ = _rT) \ur+uw dK | ht duw

(6-21)
O,
dx Yin dx ) "1 2 ~ dx 2 dx
From the metal flow continuity condition:
uw=uv2ly = ur{\ + Sf)y2ly (6-22)
gives:
duw ur(l + Sf)y2 dy

(6-23)
dx y dx
From the geometry, the workpiece thickness along the arc of contact can be written as:
y = y2+R'0: (6-24)
thus
dy
= 2R'</> (6-25)
d<f>
Under the isothermal condition, lubricant viscosity m a y b e expressed as:
apt (6-26)
n = r?oe
For the saw-tooth roughness [Chang et al, 1996]
h43Ht for Ht<43

(6-27)
2
ll + 3#; Ht>S
cD = 1 (6-28)
ux
Substituting Eqs. (6-22), (6-23), (6-25), and (6-26) into Eq. (6-21) gives:
d_ 3dP^ 6R'n0aur dH, 2{\ + Sf)y2HtR'f

Q>XH, l + (l + Sf) (6-29)
d</> d(f> y y
V
Elastic entry a n d exit. A t the entry and exit of the roll bite, extra rolling force will be
produced because of the strip elastic recovery. The incremental rolling force will affect
the total rolling force and deformed roll radius, and hence the final calculated results.
The calculation of rolling force at elastic entry and exit has been discussed in Chapter 5,
so Eqs. (5-28) ~ (5-31) and Eqs. (5-36) ~ (5-40) can be used in the calculations.
Boundary Conditions. Considering the elastic deformation regions
F(0) = te2 y21 ayo yx and F(fa) = tei yx I cryo yx, but the rolling force at elastic entry and
exit are small. So from Eqs. (5-30) and (5-31), t »tx and te% *t2, the following
boundary conditions apply to the plastic region:

(i) F(0) = T2.y2/yx (6-30)
(ii) F(<f>x) = Tx (6-31)
when substituted in Eq. (6-9), yields
(iii) P(O) = \.\5Wy(0)-T2 (6-32)
(iv) />(* ) = [1 • 1 SfV y{<f>x) - Tx ] / cos A (6-33)
From Chang et al [1996]:
(v) P / (0) = 0 (6-34)
(vi) Hh = {H\+Sf IAS (6-35)
(vii) Ax = (V3 - H\ ) / 2V3 (6-36)
(viii) Pf{<t>x)^P{(l>x)-Axlf2{Ax) (6-37)
where
</>x=cos-l[\-{yx-y2)l{2R')-\ (6-38)
37oa(i/r+^) (6_39)
H\=hd/S (6-40)
and
uWi=uf(\ + S,)y2/yi (6-41)
In the above, the expression for hx in Eq. (6-39) was derived by solving the fu
Reynolds equation at the inlet zone [Saxena et al, 1996]. In the fullfilmtheory, the film
pressure at the inlet plane is assumed to be equal to the deformation pressure
(Pf(fa) = P(fa))- This is generally invalid for the mixed film lubrication, where the
boundary condition should be determined from Eq. (6-37). For the current work,
however, Eq. (6-39) has been used. The value of hx will be refined further with an
additional iterative loop in the future.
6.3 Programming
Qiu et al. [1999] has made a program to solve Eqs. (6-1) to (6-29) numerically. For Eq.
(6-1), a fourth-order Runge-Kutta integration method, starting from both boundaries
(double shooting), is used. Equation (6-14) is also solved by the Runge-Kutta method,
starting at ^. The speed of the workpiece uw needs to be determined to calculate the
frictional stress Q. This requires the forward slip Sf to be established [see Eq. (6-22)
An initial approximation of the forward slip can be made from Ford et al [1951]:
Sf *£fl-*-Y (6-42)
4v 2 ^ 2pj
When the deformation pressure and workpiece thickness profiles have been calculated,
the forward slip can be determined from:
Sf=yJy2-\ (6-43)
where yn is the workpiece thickness corresponding to the peak pressure (at the neutral
position).
Eq. (6-29) is discretized by using a central-difference scheme, and then solved by an
over-relaxation method until a convergence criterion is satisfied.

In this chapter, Qiu's program [Qiu et al., 1999] has been modified to consider the
effect of elastic entry and exit. The rolling separation force is obtained by integrating
the pressure profile plus the rolling force at the elastic entry and exit zone, and the
deformed roll radius (assuming the roll bite geometry is deformed into a circular
profile) can be determined by Eq. (5-36).
To start the iteration, the following initial values have been used:
(i) Rolling pressure set to the constrained yield stress, P = Wy,
(ii) Film pressure Pf = Q.95P,
(iii) Contact area ^4 = 0.8, and
(iv) R'=R
6.4 Results and discussions
In the lubrication regime, there are several conditions: boundary, mixed film, EHD, and
hydrodynamic lubrication conditions as described in Figure 2.5. In the cold rolling
process, lubricant is used to reduce wear and to achieve a better surface quality. On the
other hand, there must be sufficient friction to draw the strip into the roll bite. So it
required that the cold rolling process runs in the mixed film lubrication regime. In order
to identify the rolling condition, the relationship between the Sommerfeld number and
friction coefficient and forward slip is used. The Sommerfeld number is defined as
[Avitzur, 1989]:
TJU
Sommerfeld n u m b e r = — — (6-44)
^ 2
where
7=lubricant viscosity, Pa.S
uW2 =strip speed at exit, m/s
ay =yield strength of workpiece, Pa
v2 =strip thickness at exit, m
The forward slip obtained from the strip marking method (section 2.3.2) against
Sommerfeld number and the friction coefficient against Sommerfeld number by using
experimental data in Hille 100 rolling mill are shown in Figures 6.2 and 6.3
respectively. Comparing Figures 6.2 and 6.3 with Figure 2.5, it can be seen that the
rolling experiment operates under the mixed film lubrication condition.

0.000B-00 2.000E-08 4.000E-08 6.000E-08

Sommerfeld number
Figure 6.2 Forward slip vs Sommerfeld number

0.12
0.000E+00 2.000E-08 4.000E-08 6.000E-08

Sommerfeld number
Figure 6.3 Friction coefficient vs Sommerfeld number

The calculated total pressure, film pressure, film thickness, and frictional shear stress
etc. under the different rolling speed, yield strength, tensions, and surface roughness
conditions, have been discussed by Qiu et al. [1999]. The calculated total rolling force
has also been compared with industrial data. The following discussion will focus on the
effect of elastic entry and exit, lubricant property, and friction coefficient value at
asperity contact. The calculated film thickness at exit will be validated by the oil drop
experimental data.
6.4.1 Effect of elastic entry and exit
Although the rolling force at elastic entry and exit is small compared with the total
rolling force, it still has an effect on the rolling parameters such as rolling force and
deformed roll radius, hence affect the film pressure and thickness over the roll bite. The
experimental data of 31.67% reduction and 0.353 m/s rolling speed from carbon steel
was used to calculate total rolling pressure, film pressure, and film thickness with and
without consideration of elastic entry and exit. The comparisons are made in Figures 6.4
and 6.5.
Carbon steel BHP-300, e=31.67%, ur=0.353m/s,

lubricated R
n .^
l/l
m i0)
3
n
/y
tfl c
in 0
0) Ul ^ ^ \ .
Q. c
^^j**1^
ra E w ith elastic entry and exit

*d
o
1- - " ^
1 0.8 0.6 0.4 0.2 0

Roll bite (<()/()),)
Figure 6.4 Total rolling pressure comparison

Carbon steel BHP-300, E = 3 1 . 6 7 % u =0.353m/s,

lubricated
with elastic
entry and exit jr >.
entry and exit

Si
0) </>
E £
ir 2.
—*^— 1 — — — - , 1 , >*
0.8 0.6 0.4 0.2

Roll bite ((})/((. j)
Figure 6.5 Film pressure comparison
From Figures 6.4 and 6.5, it can be seen that the total rolling pressure and film pressure
have not changed much under two situations-with elastic entry & exit and without
elastic entry & exit. The difference of total rolling force and torque are 1.38% and
0.23%, respectively for 3 mm strip thickness at entry. When the rolled strip is getting
thinner (0.5-1.5 mm) in cold rolling, the elastic zones may play a more important role.
The rolling force in the elastic area could reach 10% of total rolling force. So the elastic
entry and exit zone should be considered in the analysis of cold rolling to achieve more
accurate results.
6.4.2 Effect of lubricant viscosity
In the mixed film regime, a part of the roll bite is asperity contact, and the lubricant
separates the remaining part of the roll bite. The property of lubricant has an effect on
the film thickness, furthermore the film pressure and total pressure. The following
graphs show how the different viscosity affects the total rolling pressure, film pressure,
film thickness, and the contact area ratio. In Figure 6.6, it can be seen that the total
rolling pressure distribution increases as the viscosity of lubricant decreases. This

phenomena can be explained from Figures 6.8 and 6.9. W h e n the viscosity decreases,
the film thickness is reduced too, so the asperity contact area ratio increases (when th
fully asperity contact happens, A = l). Consequently, the total pressure increases with
more asperity contact.
Carbon steel BHP-300, e =31.67%, ur=0.353m/s,

lubricated
* fl DOtt Pn ^
• n n^ft P a ^
n *~.
in 0.116 F^S
0) in
3 0)
— * — 0 232 Pa S
* fMfi/l Pn °.
10 c
10 o
01 10
p—
Q c
ra a
E
o •o
0.8 0.6 0.4 0.2

Roll bite [t)/«),)
Figure 6.6 Total pressure distribution over the roll bite
Carbon steel BHP-300, E =31.67%, ur=0.353m/s,

lubricated
a n nnft Pa <^
• n n^ft Pa 9
0.116 Pa.S
w n 'r^ P-I =;
11 * n AM Pn °.
a> to
*S
E E
iT S i * ^ ^ * ^ * : 1 1 1 ^ * f
0.8 0.6 0.4 0.2

Roll bite (/<!>,)
Figure 6.7 Film pressure distribution over the roll bite

Carbon steel BHP-300, e=31.67%, ur=0.353m/s,

lubricated
0.5
0.008 F>a.S
3f 10
0.058 Pa.S 0.4
10 10 0.116 Pa.S
in <1> 0.3
V c 0.232 Pa.S
c o 0.464 Pa.S
o 10 0.2
f c
+- (11
EJ IB«***»^)IM*XXXXX)I0I« 0.1
if E
rrr;:7-""7";;;;;7"-"r?rr>Tmm o
0.8 0.6 0.4 0.2
Roll bite (<j>/<t>,)
Figure 6.8 Film thickness distribution over the roll bite
Carbon steel, BHP-300, e =31.67%, u =0.353m/s,

lubricated
ID
Ql (0
0.9
m 10
*d
<u
o c 0.8
ra o
c 10
O c 0.008 F^.S 0.7
•*-ai
O E 0.058 Pa.S
o
o •a 0.116 Pa.S 0.6
& < 0.232 Pa.S
0.464 Pa.S 0.5
0.8 0.6 0.4 0.2
Roll bite {if/i,,)
Figure 6.9 Asperity contact ratio in the roll bite
In the inlet area, the film thickness and asperity contact area changed dramatically as
shown in Figures 6.8 and 6.9. As viscosity decreases, the film thickness decreases and
the asperity contact area increases. In Figure 6.7, it can also be seen that the film
pressure distribution in the inlet area of roll bite drops as the viscosity decreases. This is
probably caused by the asperity contact area increasing significantly with a decreasing
viscosity in the inlet area shown in Figure 6.9. As the asperity contact area rises, the
pressure in asperity contact area increases too, so the film pressure drops. Further away
from the inlet area, the film thickness and asperity contact area reach a steady-state
value comparing with the inlet area as shown in Figures 6.8 and 6.9. It shows that the
thinner the film thickness (lower viscosity) is, the higher the film pressures. At exit, film
pressure decreases more sharply with lower viscosity. Total rolling force and torque
comparison for different viscosity at 3 1 . 6 7 % reduction and 0.353m/s rolling speed are
shown in Table 6.1.
Table 6.1 Rolling force and torque under different viscosity
Viscosity 0.008 Pa.S 0.058 Pa.S 0.116Pa.S 0.232 Pa.S 0.464 Pa.S
Force (kN) 641.3 624.4 614.1 601.7 588.1
Torque (kN-m) 5.408 5.341 4.721 4.851 4.878
It can be seen that the rolling force and torque decreases with increasing viscosity. So
the viscosity variation has really affected the rolling force and torque. Because the
viscosity is affected by temperature, the comments m a d e on the effect of lubricant
viscosity in this section also apply to the temperature effect. However, the 'thermal
effect' issue will be discussed in chapter 8.
6.4.3 Effect of friction coefficient at asperity contact
In Eq. (6-11), the value of friction coefficient c at the asperities contact must be given
before calculation. In Qiu's work [Qiu et al, 1999], the value of 0.2 is assumed under
some calculation case. But the different value c could result in very different calculated
results as shown in the following graphs. In Figure 6.10, the total rolling pressure, and
the neutral angle increases as friction coefficient rises. The film pressure also increases
when friction coefficient value increases in Figure 6.11. The film thickness and
apserities contact ratio A does not change m u c h with the different value of friction
coefficient at asperities contact shown in Figures 6.12 and 6.13.

C a r b o n steel BHP-300, e=31.67%, u r =0.353m/s,
lubricated
-•—0.05
••—0.10
0.15
0.8 0.6 0.4 0.2

Roll bite (<(./((,,)
Figure 6.10 Rolling pressure u n d e r different friction coefficient

lubricated
a- IZ
<D (0
n
*5
E E
iT S
0.8 0.6 0.4 0.2
Roll bite (+/$,)
Figure 6.11 Film pressure under different friction coefficient
Carbon steel BHP-300, E = 3 1 . 6 7 % , ur=0.353m/s,

lubricated
0.5
in in
in o
« c
f .2
o w
£ at
E J
0.8 0.6 0.4 0.2
Roll bite (<t>/(j),)
Figure 6.12 Film thickness under different friction coefficient


lubricated
ai <
2 o
« w
.2 o
£
0.6 0.4 0.2
Roll bite (<|)/(j),)
Figure 6.13 Asperities contact ratio
The rolling force and torque under the different friction coefficient are shown in Table
6.2
Table 6.2 Rolling force and torque under differentfrictioncoefficient
Friction coeff. 0.05 0.1 0.15 0.2 0.25
Force (kN) 415.1 481.0 549.3 624.4 719.6
Torque (kN-m) 3.584 4.099 4.538 5.341 5.860
It can be seen that both rolling force and torque increases as friction coefficient value
rises. So it is important that the correct friction coefficient value is given before th
calculation. This value can be determined by using either the "strip marking method" or
the "Laser Doppler method" as described in Chapter 4.
6.4.4 Comparison between calculated and experimental data
The comparison between the measured and calculated pressure distribution are shown in
Figures 6.14 & 6.15 for lubricated rolling condition. The experimental material is
aluminium alloy 5052-H34 at the 115.4 mm/s rolling speed, and 23.84% & 29.48%
reduction ratio respectively. The pressure from the mixed film lubrication model shown
above compares well with the measured radial pressure from the sensor roll except
towards the exit. This confirms again that the sensor roll method is reliable to meas
friction coefficient. The measured data in the graphs are from radial pin recording.
Aluminium alloy 5052-H34, y 1 =3.092mm, y 2 =2.355mm,

u =115.4mm/s, e =23.84%, lubricated
500
ra
3
10
in -•— .Measured data
s. -•— Marked method
-h—Sensor roll
a.
-•— Mixed film model
Figure 6.14 Pressure comparison s=23.84% (lubricated)
Aluminium alloy 5052-H34, y=3.08mm, y 2 =2.172mm,

u =115.4m m/s, 8=29.48%, lubricated
500
~ 400
ra
CL
~ 300
0>
io 200
IO
0)
-•— .Measured data
100 - A — Strip marking method
Sensor roll method
-•— Mxed film
0
Figure 6.15 Pressure comparison e=29.48% (lubricated)
6.4.5 Oil drop method
The oil drop method was used to determine the oilfilmthickness in the rolling process.
The test involved dropping a known quantity of oil on the strip surface at the begin
Carbon steel BHP-300, s = 2 9 %

1.2 • •
E
10 0.8
a
10 a a
0) 0.6 •
o •
A •
Z 0.4 • 0.0086 PaS '
I 0.2 a 0.089 PaS
0.93 PaS
I I — 1
5 10 15 20
Rolling speed (rpm)
Figure 6.17 Film thickness under different lubricant viscosity
The test was also made under the different lubricant properties while the other rolling
conditions are same. In Figure 6.17, it can be seen that the measured film thickness
increases as viscosity and rolling speed increase. The calculated film thickness from
mixed film model was compared with the measured film thickness shown in Figure
6.18. It can be seen that both data decreases with increasing reduction ratio, and the
calculated results are fairly close to the measured data.
Carbon steel BHP-300, ur=0.118m/s
• Experimental data
E 0.8 + a Mixed film model
*
in 0.6 a
mu
<
c •
o 0.4 w
£
•#-*
a
E 0.2
iZ
0 ' •' i •' "•, i '• •• r i
10 15 20 25
Reduction ratio (%)
Figure 6.18 Film thickness under different reduction ratio

6.5 Summary
A rolling model under mixed film lubrication has been developed by Qiu et al [1999]
A rigorous average Reynolds equation is used to calculate the hydrodynamic pressure
and film thickness. The variation of the yield strength with strain is considered in the
model, which makes the model more practical.
The elastic entry and exit zone in the roll bite should be considered in the analy
especially when the strip thickness is getting thinner. In the Qiu's model [Qiu et al,
1999], the viscosity is given a constant value. In the above analysis, it can be seen that
the different viscosity value can really affect thefilmpressure and thickness etc.. So the
thermal effect on the viscosity will be discussed later. The friction coefficient value
must be carefully chosen in the calculation because the different friction coefficient can
affect the results significantly.
The total pressure distribution calculated by mixed film model coincides well with
measured value. The measured film thickness from "oil drop method" is close to the
calculated data, thus validate the mixed film model. The rolling speed, reduction ratio,
and viscosity property can affect thefilmthickness.

Chapter 7 Influence of Hydrodynamic Met and Outlet Zones in Mixed Film Model 165
Chapter 7
Influence of Hydrodynamic Inlet an
Outlet Zones in Mixed Film Model
7.1 Introduction
In practice, a mixed film regime prevails in the roll bite, where metal to metal co
occurs at the surface asperities, and the surface valleys arefilledwith oil. Significant
progress has been made in the analysis of this lubrication regime recently [Sutcliff and
Johnson, 1990; Wilson and Chang, 1994; Chang et al, 1996; Wilson and Sheu, 1988].
Qiu et al. [1999] developed a similar mixed film model, but with a more rigorous
second-order Reynolds equation. There is no unknown constant in the Reynolds
equation. The film pressure is solved from the Reynolds equation through an over-
relaxation method, with boundary conditions applied automatically. In the calculation,

Chapter 7 Influence of Hydrodynamic Inlet and Outlet Zones in Mixed Film Model 166
the film pressure is set to zero at the edge of entry and exit of the plastic work zone.
Other researchers [Lugt et al, 1993; Lugt and Napel, 1995; Lin and Houng, 1991] have
developed a model while considers the effect of the hydrodynamic inlet and outlet
zones. The film pressure was set equal to the material's yield strength at the entry edge
of plastic work zone, and zero at the starting point in the inlet zone. The same boundary
condition was also applied in the outlet zone, material's yield strength at the exit of
plastic work zone, and zero at the end point of the hydrodynamic outlet zone. But the
research work to account for the effect of inlet and outlet zones under mixed film
conditions is limited. A s part of the asperity contact exists over the roll bite in the mixed
film regime, the film pressure at the end of the plastic work zone is no longer the same
as the material's yield strength.
This chapter will discuss a model similar to that as proposed by Qiu et al, [1999],
taking into account the hydrodynamic effect of inlet and outlet zones under the mixed
film rolling condition.
7.2 Theoretical analysis
The overall lubricating area is divided into three major zones: inlet zone, plastic wor
zone and outlet zone. First, full hydrodynamic lubrication prevails at the inlet zone.
Secondly, mixed-film lubrication occurs in the plastic work zone where the roll and
strip surfaces are separated by lubricant in some areas and have asperity contacts in
other areas. Finally, the full hydrodynamic lubrication takes place in the outlet zone.
The whole rolling process is shown in Figure 7.1. Assumptions in the analysis are as
follows:
(1) The deformed roll radius is considered, and the roll remains circular after plastic
deformation;
(2) Isothermal conditions prevail;
(3) The density of the lubricant is constant;
(4) The Reynolds equation for steady incompressible flow applies;
(5) The lubricant is Newtonian with a viscosity n at pressure pf given by n = n0e°Pf,
where rj0 and a are viscosity at ambient temperature and viscosity pressure factor,
respectively.
In
Lubricant
Uwi OJ
T
1 /
Figure 7.1 Schematic diagram of rolling process
7.2.1 Inlet zone
In the inlet zone , the lubricant is drawn into the space between the workpiece and the
roll by hydrodynamic wedge action. The first-order Reynolds equation is used to

analyse the film pressure and the film thickness. The film thickness in the inlet zone
be written according to the geometry:
2 _ 2
h = hB+- ^- (7-1)
2R
Thefirst-orderReynolds equation can be written as:
dp f (x2-x2)/2R'
- £« f
î
6n0.e .ur. \+{+sfy± (7-2)
dx yx ]hB + (x2-xl)2R<]
The boundary conditions relating to Eq. (7-2) in the inlet zone are:
[x = 0, h = hB at point B
(7-3)
i x = Bx, pf = 0 at point A
where Bx is the distance at the starting point of the inlet zone
7.2.2 Plastic w o r k zone
The equations of plastic work zone in Chapter 6 will be used to calculate the total
pressure, film pressure, film thickness, and asperity contact ratio etc.. The same
boundary conditions described in Chapter 6 will be used except the film pressures at t
entry and exit of the plastic work zone. This film pressure at the end points of the pl
work zone will be calculated at the inlet zone as well as in the plastic work zone.
7.2.3 Outlet zone
In the outlet zone , the first-order Reynolds equation is also used to analyse the fil
pressure and the film thickness. The film thickness in the outlet zone can be written
according to the geometry:

h= h +
c ^ (7"4)
Thefirst-orderReynolds equation can be written as:
d
Pf * <Vt I t, „ M x2l2R[ 6n0.e^.ur.[ + ^+ S,JL *'2 , (7-5)
f
d^ (hc + x /2R')
The boundary conditions relating to Eq. (7-5) in the outlet zone are:
\x = 0,h = hc at point C
(7-6)
|x = B2,pf = 0 at point D
where B2 is the distance at the end point of the outlet zone
7.3 Solution process
A program has been written in the FORTRAN code to incorporate hydrodynamic film
zones with mixed film model in the plastic work zone by Qiu et al. [1999]. First, the
film pressure and thickness are solved in the inlet area according to the prescribed
boundary conditions. The solution of the film pressure at the boundary of the
inlet/plastic work area is used as new boundary conditions for the plastic working are
The known film thickness at the boundary of inlet/plastic work area described by Eq.
(6-35) is also used as a boundary condition. The calculated value of the film pressure
and thickness at the boundary of plastic/outlet area is set as the boundary conditions
the outlet zone. The horizontal distance will be defined as negative after the exit po
plastic work zone. The calculation flow chart for inlet zone is shown in Figure 7.2.
START
Divide the inlet zone

into N divisions. The
point is from 0 to N
i
Set k,
0w=kA
I
hN=h0+-±(R'<f>XN)
K
I
Set initial value at point N :
PJN =0,h = hN
*T
Solve first-order Reynolds equation
in inlet zone from point N to point
0 to get film pressure:
PfN>PfN-\> >PfO
ir
Set initial value at point 0:
Pf =Pfo>h = K
1r
Solve first-order Reynolds equation
in inlet zone from point 0 to point
N to get film pressure:
P ro ->P a ' iP fN
No
Figure 7.2 Calculation flow chart in the inlet zone

7.4 Results and discussion
The film thickness in the inlet area decreases sharply along rolling direction as shown in
Figure 7.3. The film pressure increases from zero to a certain value at the transition
between the inlet and plastic deformation area (Figure 7.4). The trend of the film
pressure in outlet area is shown in Figure 7.5. It can be seen that the film pressure
decreases to zero in the outlet zone. When the inlet and outlet zone are considered in
calculation, the film pressure will be affected comparing with the case of non-
consideration of inlet and outlet zone (Figure 7.6). The results of rolling force, torq
and film thickness between the two situations, with or without the hydrodynamic inlet
and outlet zones, and the measured values in the experiments are shown in Table 7.1.
Carbon steel BHP-300, e =28.52%, u =0.883m/s
20
15
in
0)
c 10
o
1.006 1.004 1.002

Inlet (<|>/<|t,)
Figure 7.3 Film thickness in the inlet area
C a r b o n steel BHP-300, e =28.52%, u r =0.883m/s

0.0006
ti
-v 0.0004
Q.
 0.06
3
10
in
Oi
0.04
a
E 0.02
0
0.00B-00 -1.00E-05 -2.00E-05 -3.00E-05 -4.00E-05
Outlet (<(,/<{,,)
Figure 7.5 Film pressure in the outlet area
Carbon steel BHP-300,e =28.52%, ur=0.883m/s
to
a.
to
i_
3
W
in
Q)
Q.
E
1.1 0.9 0.7 0.5 0.3 0.1 -0.1

Roll bite (()>/<(>,)
Figure 7.6 Film pressure in the roll bite
Table 7.1 Results comparison
N o t considering Considering inlet Measured value

inlet & outlet
area & outlet area
Rolling force(kN) 755.4 764.9 772.0
Rolling torque(kN-m) 7.249 7.315 7.378
Film thickness(pm) 0.835 0.855 0.873

From the Table 7.1, it can be seen that the calculated results are close to the measured
values, when the hydrodynamic inlet and outlet zones are considered.
The value offilmpressure at inlet/plastic work zone will be changed as rolling speed
increases. This can be seen from Figure 7.7. The same material aluminium alloy 1100-
To in Qiu et al. [1999] will be used for the calculation. Its yield strength is:
0.097
a =199.60(1 + 201.8^)' (7-7)
Aluminium alloy 1100-To,e =10.10%
2 3 4
Rolling speed (m/s)
Figure 7.7 Film pressure at the edge of inlet/plastic work zone
In Figure 7.7, the value of film pressure at the end point between the inlet and the
plastic work zone increases when rolling speed rises. This value can reach Q.22xayo at
the rolling speed of 4.55 m/s and 10.10% reduction. So, if the film pressure at bound
of inlet is still set to zero at higher rolling speed, more error will be incurred. H
effect of inlet and outlet zone, especially the inlet zone should be considered in th
analysis of mixed film model.
Because of the strip's elastic recovery at the entry of the roll bite, a h u m p on the strip
surface can be resulted as shown in Figure 7.8. From an elastic-plastic finite elemen
Chapter 7 Influence ofHydrodynamic Inlet and Outlet Zones in Mixed Film Model 174
analysis by Jiang et al. [2001], the calculated m a x i m u m h u m p value for t w o cases
(3.55m/s and 16.47m/s rolling speed with zero tension for a tandem cold mill) are 1.6
pm and 1.1 pm, respectively. Comparing with the surface roughness of the roll & strip
and the film thickness at roll bite entry, the hump should be considered. After the e
hump is considered, the hydrodynamic film thickness at roll bite entry will become
thinner. Combining with the examples in Jiang et al. [2001], the comparison of film
pressure and total pressure was determined with and without hump consideration. In
Figures 7.9 and 7.10, it can be found that the film pressure distribution is slightly
different under two different conditions-no hump and with hump.
exaggerated
hunp
Figure 7.8 Strip elastic deformation at entry - h u m p
u =3.55 m/s
2.5
% 2
to
1.5
in 1
in
v — • — w i t h hump
Q. 0.5
E —«—without hump
• 1 1— 1 0
1.1 0.9 0.7 0.5 0.3 0.1 -0.1
Roll bite (<j>/<|),)
Figure 7.9 Film pressure distribution in the roll bite (ur=3.55m/s)

u=16.47m/s
2.5
1 2
to
|
1.5
v
3
10 1
in
0)
l_ • with hump
Q. 0.5
— • — w ithout h u m p
E
» 1 ,— i 1 • I — *
1-1 0.9 0.7 0.5 0.3 0.1 -0.1

Roll bite (<j>/<|> j)
Figure 7.10 Film pressure distribution in the roll bite (ur= 16.47 m/s)
The rolling force and torque are shown in Table 7.2. At 3.55 m/s rolling speed, the
errors of rolling force and torque are 0.11 % and 0.92 % respectively between the
condition of no hump and with hump. At 16.47 m/s rolling speed, the 2.39 % and 4.62
% errors are produced for rolling force and torque. It can be seen that the error i
as rolling speed increases. At high rolling speed, the elastic deformation at entry
be considered in the calculation.
Table 7.2 Rolling force and torque
Rolling Rolling force (kN/m) Rolling torque (kN-rn/m)
speed (m/s) no h u m p with h u m p no h u m p With h u m p
Ur=3.55 936.6 937.0 108.29 108.11
Ur=16.47 15580 15952 269.97 282.43
The hydrodynamic film thickness at entry and exit has also been affected by the hump.
One case at 16.47m/s rolling speed is given. In Figure 7.11, it can be seen that the
thickness with a hump is lower than the film thickness without hump at hydrodynamic
entry. T h efilmthickness under both conditions decreases towards rolling direction.
Hydrodynamic entry, ur=16.47m/s

20
-•—without hump
10 <0 -with hump 15
io m
oj a>
10
-1
o —
IE
C =5
1.0015 1.001 1.0005
Entry z o n e (<|>/<|>,)
Figure 7.11 Film thickness at entry
It will be useful if the strip elastic recovery at exit is also considered in the calc
of the film thickness and pressure in the outlet zone beyond the exit plane. But this is
beyond the scope of this thesis.
7.5 Conclusions
The influence of inlet and outlet zone on the model has been discussed in the mixed film
lubrication condition. F r o m the results, w e can see that the inlet and outlet zone affect
the film pressure and film thickness as well as the total rolling force and torque. At high
rolling speed, the inlet zone can have a significant effect on the rolling parameter. T h e
effect of the h u m p of the elastic zone at entry is also discussed. So the hydrodynamic
inlet and outlet zones should be considered in the mixed film lubrication model.
Chapter 8 Thermal Effect in Mixed Film Lubricated Cold Rolling Process 177
Chapter 8
Thermal Effect in Mixed Film
Lubricated Cold Rolling Process
8.1 Introduction
Cold rolling process involves high pressures and high speeds. A large amount of heat i
generated from plastic deformation and interface friction causing the temperatures of
the work rolls and strip to increase. The transfer of thermal heat at the contact is as
important as the transfer of forces and forms a significant part of the study of tribology
in cold rolling. It is reasonable to expect that the parameters that affect frictional forces
in rolling will also affect the amount and the rate of heat transfer in the deformation
zone.
Wilson and Mahdavian [1974] developed a thermal Reynolds equation to take into
account viscosity variations across the lubricant film thickness due to energy dissipation
within the film, but the dominant m o d e of heat transfer in the lubricant is conduction.
Bhatt and Sengupta [1996] developed a similar thermal Reynolds equation but took into
account the effect of both conduction and convection on heat transfer in the lubricant
film. Lugt and Napel [1995] considered thermal effect in a model which assumed that
the deformation energy is entirely transformed into heat, and friction-induced heat
generation, and the conduction of heat into the rolls are neglected. Roelands viscosity-
pressure-temperature equation [1966] was used. Lin and H o u n g [1991] analyzed the
thermal effect in the three zones, namely inlet & outlet, and plastic work zone under the
fully hydrodynamic lubrication. Conduction is assumed to be the only m o d e of heat
transfer in the three zones and the energy equation was used in these zones to calculate
the temperature variations. All of works above were carried out for the fully
hydrodynamic lubrication.
Liu and Tieu [2001] develop a new model to consider the effect of inlet and outlet zone
under mixed film lubrication. The results are shown that both zones, especially the inlet
zone have an influence on the model accuracy. But the thermal effect was not
considered in it.
The analysis of the thermal effects in the mixed film lubrication is more complicated.
Because the generated heat comes not only from the plastic deformation, but also from
the frictional shear stress at the asperity contact as well as the viscous shearing of
pressurized lubricant in the surface valleys. This chapter considers the thermal effects in
the three zones, inlet, outlet and plastic work zones under the mixed film lubrication
regime.
8.2 Theoretical analysis
The overall lubricating area is divided into three major zones: inlet zone, plastic work
zone and outlet zone. First, the fully hydrodynamic lubrication prevails at the inlet zone.
Secondly, mixed-film lubrication operates in the plastic work zone where the roll and
strip surfaces are separated by lubricant in some area and asperity contacts in the
remaining area. Finally, the fully hydrodynamic lubrication takes place at the outlet
zone. The whole rolling process is shown in Figure 7.1. Assumptions in the analysis are
made as follows:
(1) The roll are elastic and circular, and deformed roll radius is considered;
(2) Conduction is assumed to be the only m o d e of heat transfer in the three zones;
(3) The density of the lubricant is constant;
(4) The Reynolds equation for steady incompressible flow applies;
(5) The lubricant is Newtonian with a viscosity n at pressure pf, and the average
temperature Tm, given by n = n0 exp[apf - f3(Tm -T0)]\
(6) The lubricating flow is laminar and the inertia forces are neglected;
(7) The yield strength of material is not affected by the temperature in cold rolling.
8.2.1 Inlet zone
At the inlet zone, the film thickness can be written the same as Eq. (7-1). The first-or
Reynolds equation involving temperature effect can be written as

dp f
= -6urrj0 expfcp, - fi(fm - TQ)J + (l + S, )(y2 /yx)] (*' * i ) / 2 * ' (8.1}
dx
Energy Equation. The general form of energy equation combined with Reynolds
equation can be written as follows (see Appendix C):
d2Tf
2
= Ex (E22 Y2/H2 + 2E2E3 Y/H + E2) (8-2)
dY
where
Ex=(-Px/C)exp[-E{Tm-l)]
-E2=6[l + Sf )(l - e)+ lJ/7 -Hx) (8-3)
E3=(l + Sf)(l-£)-l
The boundary conditions of the inlet zone are:
T=Y, T=\; Tf=\ at X = co (8-4)
The 'moving' boundary conditions for the roll and strip surfaces are given as (see
Appendix D):
dT dX'
Tr=Tf(X,l/2)=P3[(yH) f 7=1/2 1 (8-5)
8Y (X'-X)/2
fw = Tf (X -1/2)= S ^ x [ (- l/H)dTf /dY

y=-i/2 X - TT7T + 1
w /v ;J 2 A,v /; (X'-Xj/2
[(l+^Xi-^
(8-6)
Refer to Appendix C for definition of Px, P2, and P3.

Y = l/2
Y =0
strip surface
Y=-l/2
Tf-Tw
Figure 8.1 Flow between moving surfaces
Thefluidtemperature Tf can be obtained directly by integrating Eq. (8-2) yielding
T,=El'EI Y^+ML r Ely2 (8-7)

2
12 H 3 H 2
where (refer to Figure 8.1)
7 = 1/2, Tf=Tr
(8-8)
7 = -1/2, T>=T;
Apply Eq. (8-8) to Eq. (8-7)
£2^3 1
C, = — [Tr-Tw] (8-9)
1 J
Ex 12 H
î £,2 i E:
c2=-^[^+rJ- 2
(8-10)
2EX 192 7/ 8
The roll and strip temperatures are obtained by applying the boundary conditions Eq.
(8-8) to Eq. (8-7)

(8-11)
12 H H 2 2'
El (-1/2)4 | E2E3 (-1/2)3

f T?2
| El J__Q +C
TW=EX (8-12)
12 ' H2 3 H 2 '22 2 2
Thefluidmean temperature is:
— rV2 — El
2 (8-13)
T=y TfdY = E, 960H + -5-
24 + C
m J_1/2 / 1
Thefilmthickness at the inlet edge of work zone is given by the isothermal expression
[Saxena et al, 1996 and Qiu et al, 1999]:
HB =(H\+Sf IAJ3 (8-14)
1 (8-15)
s
3n0a(ur+uwJ
*.= (8-16)
8.2.2 Plastic work zone
The plastic work zone under the mixedfilmlubrication is governed by Eqs. (6-1) ~ (6-
54) [Qiu etal, 1999],
Temperature Calculation. In the plastic work zone, the rolling process is under mixed
film lubricated condition. So the temperature in the lubricant and at the
contact will be calculated separately. The energy equations are used to c

temperature of lubricant, roll surface and strip surface.
Temperature in lubricant.
Energy Equation. In the plastic work zone, the Eq. (8-2) and Eq. (8-3) can be
expressed as follows (see Appendix C):
d2Tf
T = Ex (E22 Y2/H2 + 2E2E3 Y/H + E2) (8-17)
dY
where
Ex=(-Px/C)cxp[-E(T1-l)]
E2=6[(l + Sf)(l-£)/Y + l].(H-Hx) (8-18)
E3='l + Sf)(l-£)/Y-\
In the rolling process, if part of the plastic work is retained in the strip, the temperature
of the lubricant at strip surface can be written in dimensionless form:
T=T„+iG)nY (8-19)
where
G = ay/(cwpwT0) (8-20)
y is afractionof plastic work converted to heat
The 'moving' boundary conditions for the lubricant at the roll surfaces are
Appendix D):
xj 1 > dT( dXx (8-21)

T,=PA 1/2
+ Tr
H) dH H (X'-x)
Tw expression is the same as Eq. (8-19). Eqs. (8-7) ~ (8-13) can still be used here.
Temperature at asperities contact.
The temperature at asperities contact can be calculated as follows:
One-dimensional temperature increment equation at asperities contact can be considered
in moving heat source analysis proposed by Carslaw and Jaeger [1959] (see Appendix
E):
,_ coa(x',f).dx'df (x-xj '

dl = — - — - r — r — x exp (8-22)
AK(t-f)
27ik(t-f)
The heat flux generation at asperity contact area is:
®a = PaPaK (8-23)
The heat generated is conducted into the two contact surfaces according to a partition
factor f(x',f), which represents the fraction of heat flux conducted into the roll sur
with a velocity ur. The fraction of heat flux conducted into the strip surface with a
velocity uw, then becomes [l-/(x',?')]. The temperature rise formulation for the
surfaces acted on by a single heat source at time f is extended as follows.
The temperature rise on the roll surface can be determined from:
[{x-x')-ur(t-f)f
(8-24)
2mr \t -1)
The temperatureriseon the strip surface is given by:
,rr, r* ,, . . 0) „ (x' , fid*'At [(x-x')-uw(t-f)]2}

(8-25)
27U\w{t-f) 4Kw((-t')
The surface temperature rise produced by all the frictional heat sources can be obtained
by integrating the above differential temperature equations over the contact heat
area from time t'= 0 to time t'-t. The total surface temperature rise in the fixe
coordinate system at time t over the roll bite xx for roll and strip surfaces is:
Mr(x,t)= \\dTr(x',f) (8-26)

0 0
ATw(x,t)= )\dTw(x\t') (8-27)

0 0
In the rolling process, another important heat source is the plastic deformation in the
plastic work zone. If only part of plastic work is transformed into heat, the te
rise at asperity contact caused by plastic work can be expressed as follows:
AT=jGlnY (8-28)
In the present model, the simplifying assumption is made that allfrictionalheat enters
the surfaces and no heat loss takes place over the surfaces. Assume heat flux
cva(x',f)and the partition factor f{x\f) are the average values and invariant at
subregions during small time intervals. The following discretized equations will
for solving Eqs. (8-26) and (8-27). The contact time over the roll bite will be
into M small time intervals. The roll bite area is divided into N intervals.
W-l N [(x-x<)-ur(t-f)]2
^y-TLfi^n^^-^ AKr(t-f)
Ax'At' (8-29)
M-\ N ' l(x-x')-uw(t-t')]2

(
ATW(x,/)= Y T [ 1 -/(*',f)] ^ * ' ' ° • exp fAx'Ar*(8-30)
4Kw(t-f)
In order to calculate the temperature of asperity contact surfaces, the heat partition
factors f(x',t') must be solved by assuming that the surface temperatures at the asperity
contacts are equal because there is no temperature jump across the interface. This yield:
Tri +ATr(x,t) = TWi +ATw(x,t)+ATs (8-31)
In dimensionless expression
frb + ATr(x,t) = fwb + ATw(x,t)+ ATS (8-32)
If the local frictional heat flux generated at a station i is coai, the model assumes tha
the heat flow into the roll is f(x',t')a)ai while the heat flow into the strip is
[1 - f(x',t')]a)ai. If there were no plastic heating, the partition factor would be betwe
zero and one. As plastic heating is increased, the strip becomes hotter, more heat is
transferred from the strip to the rolls, and the partition coefficient increases. Under
conditions where plastic heating of the strip is of dominant important, the partition
coefficient may be much larger than unity.
If it is assumed that the time is zero at the entry point of roll bite, the time for roll
surface at a position x is:
f' = — (8-33)
ur
But the time for strip surface at position x is:
v, v2 1-fl-X) 3
f= ^ X + (8-34)
(l + Sf)ur y2
X = - (8-35)
xx
From Eq. (8-32), the heat partition factors f(x',t') can be solved, together with
temperatures at the asperity contacts.
Tr=Tn+ATr(x,t) (8-36)
Tw=TWi+ATw(x,t)+ATs (8-37)
The dimensionless expressions are:
fr=frb+ATr{x,t) (8-38)
Tw=Twb+ATw{x,t)+ATs (8-39)
8.2.3 Outlet zone
In the outlet zone , the first-order Reynolds equation is used to analyze the fi
and the film thickness. The film thickness equations in the outlet zone are the
Eq. (7-4). The first-order Reynolds equation can be written similar to Eq. (7-5).
Energy Equation. In the outlet zone, an equation similar to Eq. (8-2) and Eq. (8
be written (see Appendix C):
d2Tf
2
= Ex ($1 Y2/H2 + 2E2E3 Y/H + E2) (8-40)
dY
where
Ex=(-Px/C)exp[-E(Tm-l)]
E2=6(Sf+2).(H-H2) (8-41)
I E 3 = Sf
The 'moving' boundary conditions for the roll and strip surfaces are given as (see
Appendix D):
/ dX'
+ T„ (8-42)
, ^ dH r=j (X'-X) 1 / 2
lW/
T =
(i+sffyJ
/2
I-
*2
H dH
dX'
1/2
?=-\ (X'-X)
+ r„, (8-43)
Eqs. (8-7) ~ (8-13) can be used here.
8.3 Solution process
A program has been written in FORTRAN code to solve hydrodynamic and mixed film
zone throughout the whole roll bite. First, the film pressure, thickness, and temperat
are solved in the inlet area according to the prescribed boundary conditions [Liu and
Tieu, 2001]. The calculated value of film pressure, thickness, and temperature at the
transition point of inlet/plastic work area provide the boundary conditions for the pl
working area. Subsequently, the value of the film pressure, thickness, and temperature
at the transition point of plastic/outlet area is set as the boundary conditions for t
outlet zone. The temperature boundary conditions in the inlet zone are known. The
variation of material's yield strength given in Chapter 5 is also used in the calculat
8.4 Results and discussions
The calculations were carried out for various conditions, and calculated results were
compared with the experimental data. The comparisons were also made: (i) with and
without thermal effect, and (ii) with and without hydrodynamic inlet & outlet zone
consideration.
Calculation Parameters. The following results have been obtained by using the above
technique over a range of conditions. The conditions of lubricant properties and the
properties of materials used in the analysis are shown in Tables 8-1 and 8-2.
Table 8-1 Lubricant properties
Diffusivity, Kf (m2/s) 8.59 XlO" 8
Conductivity, (W/m.K) 0.145
Specific heat, (J/kg.K) 1909
Density, (kg/m3) 884.1
Base viscosity, (Pa-S) 0.058
Viscosity pressure factor, (Pa"') 6.5 xlO" 9
Table 8-2 Properties of materials
Strip
Items Roll Carbon steel Aluminium alloy
Diffusivity, (m2/s) 14.1 xlO" 6 11.6X10"6 73 xlO"6
Conductivity, (W/m.K) 48.9 41.0 177
Specific heat, (J/kg.K) 443 434 875
Density, (kg/m3) 7836 8131 2770
In Eq. (C-9) of Appendix C, the temperature-viscosity coefficient j3 must be
determined before the calculation. In the cold rolling process, the viscosity of oil is not
only affected by the pressure, but also affected by the temperature. B y using a physical-
chemical background, Roelands [1966] gave the viscosity expression considering both
pressure and temperature factors as:
Pf ^r0-138^
n = n0 exp^ [ln(770) + 9.67] 1 + (8-44)
196.2x10' r-ns
The values of z and S0 in Eq. (8-44) were given by Houpert [1985]. After the
mathematical transformation, a formula calculating temperature-viscosity coefficient
was also given by Houpert [1985].
lCTo-138)
So = (8-45)
ln(770) + 9.67
But Eq. (8-45) was derived for low pressures, and is not suitable for high pressure
rolling process. So it is better to combine Eq. (8-44) with Eq. (C-9) to calculate
temperature-viscosity coefficient /?:
P = apf-[\n{n,) + 9.61] 1 + Pf
f
Wr - i 3 8
0
+1 /(r-r 0 )(8-46)
196.2xlO r-138
Time Step. Before the calculation, the time step in the roll bite must be chosen. Figure
8.2 shows that the maximum temperature rise reaches a steady state with the increase
time steps in the roll bite. With small time step, the results of temperature can be
different from the steady state value. In the present study, 100 time steps are
recommended to reach steady state. Although more accurate results can be obtained
with more time steps and division number N, it will take longer time for computer to
complete the calculation, with no further significant improvement of the results.

Carbon steel, e =28.52%, ur=0.883mls

0.2
u
3
1° _ 0.15
Q) O J
0 — •
E Y- • • ' •• • g
o> < 0.1
E »
3 .«2 — • — N=10
E *- 0.05 — • — N=50
"5 N=100
I I I I
50 100 150 200 250

Time step
Figure 8.2 Time step
Calculated temperature. Figure 8.3 shows the lubricant m e a n temperature, the strip
surface temperature in lubricant, and the temperature at asperity contact in the roll
under 26.05% reduction and 0.068 m/s rolling speed.
Aluminium alloy, 8=26.05%, u=0.068m/s

0.06
-strip surface temperature in oil valley
-• lubricant mean temperature
temperature at asperity contact
0.8 0.6 0.4 0.2

Roll bite M>A|>1)
Figure 8.3 Temperature distribution in the roll bite
In Figure 8.3, the temperature increases from entry to exit in the roll bite. The
temperature at the asperity contact is higher than the strip surface temperature in oil
valley at the entry area of the roll bite. This is because the sliding speed reaches th
maximum at the entry area of the roll bite, so the frictional heat is higher than the
work heat which just starts in the same area. But the plastic work plays a dominant rol
when moving towards the exit of roll bite, and the strip surface temperature in the o
valley is higher than the temperature at asperity contact after a certain time.
In the Jeswiet and Zhou's work [1992], the temperature in the roll bite was measured
point by point under dry rolling condition for aluminium. Figure 8.4 shows the
comparison between the calculated result and experimental results which are close to
each other.
Figure 8.4 Temperature comparison at asperity contact
A n experimental work measuring temperature point by point over roll bite was also
carried out by the author under dry rolling condition at 31.98% reduction and 0.118m/
rolling speed. The calculated temperature at asperity contact was compared with
measured value in the roll bite shown in Figure 8.5.

Aluminium alloy, £=31.98%, ur=0.118m/s
0.03
0.02
a —
2 t
* i-
Q. 0.01
E
0
1 0.8 0.6 0.4 0.2 0
Roll bite {i)It),)
Figure 8.5 Temperature comparison

It is found that the values are close each other. This proves that the temperature
measuring technique by using embedded thermocouples is reliable.
Pressure calculation. Figures 8.6 & 8.7 show the pressure calculation results without
and with thermal effect for the same rolling condition.
Carbon steel, s=28.52%, u r =0.868m/s

-3
%
to ^ ^ ' ^ \
-2
%
^^?'
to
s'
&. - 1
3
U> .' total pressure
(0 \
V
film pressure
V
Q.
•>
-0
1 0.8 0.6 0.4 0.2 (
Roll bite (<|>/<|>1)
Figure 8.6 Pressure without thermal effect
Carbon steel, E =28.52%, u=0.868m/s
9.
to
^ ^ ^
^
-O-
- \
L.
3 / total pressure j "•
10
(0 • film pressure
•
1 • 1 1
0.8 0.6 0.4 0.2

Roll bite (4>/4>l)
Figure 8.7 Pressure with thermal effect
The total pressure comparison is shown in Figure 8.8. The total pressure with thermal
effect is lower than the value without thermal effect. The peak pressure position shi
towards exit w h e n the thermal effect is considered.

Chapter 8 Thermal Effect in Mixed Film Lubricated Cold Rolling Process 19 A
Carbon steel, 8 =28.52%, ur=0.868m/s
s.
to
CD
i-
3
10
(0
CD
k.
Q. • no thermal effect
o • with thermal effect
0.8 0.6 0.4 0.2

Roll bite {t)/«f i)
Figure 8.8 Total pressure comparison
Carbon steel, e =28.52%, ur=0.868m/s
s.
to
<D
3
10
10
CD
i—
Q.
E
iZ
0.8 0.6 0.4 0.2
Roll bite (<j>/<H)
When thermal effect is considered, the film pressure's peak drops sharply compared
with non-thermal effect consideration as shown in Figure 8.9. Such phenomenon can be
explained by the following two graphs. In Figure 8.10, the asperity contact with thermal
effect is smaller than non-thermal effect. So the film thickness with thermal
consideration is higher than without thermal effect as shown in Figure 8.11. This will
lead that the film pressure will quickly build up at the entry of the roll bite. The sam
reason applies to the exit of roll bite. So the film pressure under thermal effect is al
higher at exit. Due to thermal effect with lower lubricant viscosity, film pressure's pe
drops compared with the case of no-thermal consideration.

Carbon steel, £=25.82%, ur=0.868m/s

1
Asperity contact area, 0.8
- 0.6
'-*'
A
- 0.4
- 0.2
w ith thermal effect
-0
0.8 0.6 0.4 0.2 C)
Roll brte ((((/col)
Figure 8.10 Asperity contact in roll bite
Carbon steel, 8=28.52%, u =0.868m/s
v
-0.4
xl w ith thermal effect
in I K
in - 0.3
0)
c \ ^ "' * - . . _ _
- 0.2
o
\ ^ • • - - - - . _
E -- 0.1
ii \-^_
.. n
1 0.8 0.6 0.4 0.2 0
Roll bite (4>/<t>D
Figure 8.11 Film thickness in roll bite
The effect of plastic deformation. The following graphs show the effect of plastic
deformation on the temperature of strip surface in the oil valley, the lubricant mean
temperature, and the temperature at asperity contact.
In Figure 8.12, the strip surface temperature in oil valley increases from entry to ex
the roll bite for all reduction. At the same time, the strip surface temperature also
increases as reduction ratio rises. The reason is that more heat is generated when the
plastic work reduction is increased, and the reduction increases from entry to exit over
the roll bite.
C a r b o n steel, u r =0.868m/s
0.15
-•—28.52%
CD -•—20.09%
8-2 15.04%
-*— 8.97%
t z. o
3 5 t
m ra t- 0.05
*-*-* X X X Y,
0.8 0.6 0.4 0.2

Roll bite (4>/CP>1)
Figure 8.12 Strip surface temperature in oil valley (y =0.8)
C a r b o n steel, u =0.868m/s
-•—28.52%
0) HK-20.09%
in 15.04%
ra CD -T" -X—8.97%
(j) i- o
ra i—
0.8 0.6 0.4 0.2

Roll bite (<t»/(|)i)
Figure 8.13 Film mean temperature (^=0.8)
Figure 8.13 shows the film mean temperature at different reduction ratio. At a high
reduction ratio, the fluid in the work zone also rises to a high temperature. The h
generated is transported out of the strip material during the plastic deformation
raise the fluid temperature. The film mean temperature also increases with reductio
from the entry to exit in the roll bite. But the film mean temperature begins to dr
closer to the exit.
At the asperity contact, the heat not only from the friction at the interface, but also from
the plastic deformation. When the reduction ratio increases, the temperature at aspe
contact increases too shown in Figure 8.14.
Carbon steel, ur=0.868m/s
d-l
ro
0)
10 o
(0
i_
0)
c ^^
i» o o
-i o
>1 t
^-i
1-
ra *d
0)a> "w"
a a
10
E
CD
H
^ , -, , , \ o
1 0.8 0.6 0.4 0.2 0
Roll bite {t,/t,i)
Figure 8.14 Temperature at asperity contact
Heat partition coefficient y. In cold rolling, most of the generated heat is produce
by plastic work. But it is difficult to say how much plastic work is transferred int
Some authors [Lugt and Napel 1995; Lin and Houng 1991] assumed that the entire
plastic work is transferred into heat, thus can result in a loss of accuracy of the
So a parameter y called fraction of plastic work converted to heat is assigned. This
parameter value is between one and zero. The following graphs will show the effect o
different y on the temperature for the rolling condition of 28.52% reduction and 0.8
m/s rolling speed. The temperature of strip surface, lubricant, and the temperature
asperity rise with the fraction of plastic work converted to heat shown in Figures
8.16, and 8.17.

C a r b o n steel,e =28.52%, u r =0.868m/s
0.8 0.6 0.4 0.2

Roll bite (<t>/<j>\)
Figure 8.15 Strip surface temperature
C a r b o n steel, 8=28.52%, u r =0.868m/s
0.8 0.6 0.4 0.2

Roll bite (<j>/(|)i)
Figure 8.16 Film mean temperature
C a r b o n steel,e = 2 8 . 5 2 % u r =0.868m/s
0.15
0.05
0.8 0.6 0.4 0.2

Roll bite {if/t)))

The effect of friction coefficient at asperity contact. In the theoretical calculation, the
friction coefficient at the asperity contacts must be given. The different value
coefficient can give the different temperature calculation. Because the lubrican
temperature and strip surface temperature are solved from the energy equation, s
temperature are not affected by friction coefficient value. But the temperature
solid asperity contact is affected. Normally, the higher the friction coefficien
the higher the temperature of asperity contact (Figure 8.18).
Carbon steel,e =28.52% ur=0.868m/s

0.15
ro
 H
ro <
0) (1) •"—'
I- 0.05
a. Q.10
E
CD ro
1-
1 0.8 0.6 0.4 0.2

Roll bite (<h/d>i)
Experimental data. The experiments were carried out to measure the strip surface
temperature by using a contact thermometer immediately after the strip comes ou
roll bite. Figure 8.19 shows the experimental and calculated results under the r
speed of 0.118m/s. The two results are very close each other.
Chapter 8 Thermal Effect in Mixed Film lubricated Cold Rolling Process 200
Carbon steel, u =0.118m/s
Q.
• experimental data
35 • calculated data
10 20 30 40
Reduction ratio (%)
Figure 8.19 Temperature comparison
The measured rolling force and torque were compared with the calculated rolling force
and torque as shown in Figures 8.20 and 8.21. It can be seen that the calculated value
are close to the measured ones. In the laboratory, the strip surface temperature rise
reach 50 C in cold rolling experiment. Such a temperature variation can result in near
ten times lower lubricant viscosity. From the experimental results shown in Figure 4.6
with a temperature rise of 40°C, it can be seen that the difference in average rolling
force is 1.52% for the viscosity values of 0.0086 Pa.S and 0.089 Pa.S. From the
theoretical calculation in Section 6.4.2, Figure 6.6, the difference in rolling force
1.62% for the two above viscosities. In the rolling process, the rolling force is main
from material's plastic deformation at high reduction. So the rolling force and torque
will not change largely due to viscosity's variation caused by temperature's increase.
Carbon steel, ur=0.118m/s

800
•
•
600 •
•
•
8 + •
£ 400 •
•
& 200 • • experimental data—
• calculated data
. ! , 1
10 20 30 40
Reduction ratio (%)
Figure 8.20 Rolling force comparison

Carbon steel, u r =0.118m/s
•
E •
± •
CD •
3
O1 •
•
at
• experimental data
c
• calculated data
1 - • ~ — , !
1
10 20 30 40
Rteduction ratio (%)
Figure 8.21 Rolling torque comparison
Industrial data. The industrial data in Qiu et al. [1999] were used to calculate strip
temperature, lubricant mean temperature, and temperature at asperity contact. The film
pressure with and without thermal effect consideration was also compared. The
temperature distribution is shown in Figure 8.22. It can be seen that the strip
temperature, lubricant mean temperature, and temperature at the asperity contact
increase from roll bite entry to exit. At entry area, the temperature at the asperity
is higher than the other two temperatures. This is because more frictional heat is
generated due to higher sliding speed at high rolling speed.
Steel, 8=29.2%, u=3.96m/s

0.15
-•—strip surfacetemperature
in oil valley 0.05
-•—lubricant mean temperature
roll and strip temperature

at asperity contact
0.8 0.6 0.4 0.2

Roll bite (<()/o> 1)
Figure 8.22 Temperature distribution in the roll bite

In this case, the lubricant temperature is higher due to high rolling speed. S o the
lubricant viscosity becomes less. The film pressure with thermal effect is lower than
thermal effect as shown in Figure 8.23.
Steel, s=29.2%, u =3.96mAs

-4
- " no uiermai eirect
w ith thermal effect
to - 3
o^
"ST
3 -2
10
10
CD - 1
o.
E •/
tl -0
1 0.8 0.6 0.4 0.2 ()
Roll bite ((),/<(> l)
Temperature in inlet and outlet zone. The conditions of the inlet and outlet zones are
different from the plastic work zone. They are under the fully hydrodynamic condition
so the energy equation has been applied to solve for the temperature. The roll surfac
temperature, strip surface temperature, and lubricant mean temperature in inlet and
outlet zone do not changed as shown in Figures 8.24 & 8.25.
Carbon steel,s =28.52%,u=0.868m/s

1.2
-r\ r\ rv r\ n n n n n rv n n r^ n n n r* n 1n n
o
H 0.8
CD
i_
0.6
3
ro
i_ -•—roll surface temperature 0.4
a -•—strip surface temperature
a. 0.2
E lubricant mean temperature
CD 0
1.006 1.004 1.002
Inlet (<b/ibl)
Figure 8.24 Temperature in the hydrodynamic inlet zone

Carbon steel, E =28.52%, u=0.868m/s

1.2
F
1
o
E
CD
0.8
3
-4-«
0.6
ro
k.
ai 0.4
a — • — roll surface temperature
E
CD 0.2 - — • — strip surface temperature
lubricant mean temperature
i i ' i
0.00001 0.00002 0.00003 0.00004

Outlet (4>/<j> 1)
Figure 8.25 Temperature in outlet zone
8.5 Conclusions
The influence of thermal effect in inlet, outlet and plastic work zones have been
discussed in the mixed film lubrication condition. From the results, it can be seen that
the thermal effects has a significant influence on the viscosity of lubricant, and simil
on the film pressure and film thickness as well as the rolling force and torque. The
calculated results have been verified by the experiments. So the thermal effects in the
mixed film lubrication must be considered in a model of mixed film lubrication.

Chapter 9 Conclusions and Recommendations
Chapter 9
Conclusions and Recommendations
9.1 Conclusions
The main conclusions of this thesis can be summarized as:
(1) The sensor roll has been designed, manufactured and calibrated. The pin
transducer embedded in the the sensor roll can be used to measure the friction
coefficient along the roll bite. The temperature in the roll bite can also be
measured by the thermocouple in the sensor roll.
(2) The Laser Doppler method, strip marking method, and sensor roll method have
been used to determine friction coefficient in cold rolling. The Laser Doppler
method and strip marking method are used to measure forward slip, which is used
to calculate the average value of the friction coefficient, but the sensor roll
technique can determine the friction coefficient variation in the roll bite.
Chapter 9 Conclusions and Recommendations 205
(3) T h efrictioncoefficient in the roll bite is not constant. N o obvious pressure peak
w a s found over the roll bite, and sometime the location of the single pressure peak
does not precisely coincide with the location of the neutral point. N o multiple
pressure peaks described by other authors were found in the experiment.
(4) The friction coefficient decreases with increasing rolling speed for most of rolling
experiment, and the average friction coefficient values from sensor method and
marking method are close to each other. Empirical formulae of friction coefficient
and rolling force have been derived. A rolling speed factor has also been
considered in the empirical formula offrictioncoefficient.
(5) The rolling force and torque under dry condition are higher than this value under
lubricated condition. W h e n the rolling speed increases the rolling force and torque
increases slightly. T h e effect of different lubricant property on the total rolling
force and torque is small.
(6) The roll surface temperature along the roll bite was measured by thermocouple.
The strip surface temperature increases with increasing reduction and rolling
speed.
(7) The roll surface roughness and strip surface roughness can be transferred to each
other along the rolling direction during rolling. But no significant change in
transverse direction for strip w a s observed. The finished roughness of strip surface
increases with increasing reduction.
(8) The forward slip increases as the reduction increases, and the forward slip under
dry condition is higher than under lubricated condition. The forward slip
decreases as rolling speed increases. The higher strip surface roughness is, the
higher the forward slip is.

(9) The determination offrictioncoefficient using sensor roll has been validated by
the experimental results. These values were found to be reasonable w h e n
comparing them with the theoretical calculation. The friction coefficient curve
measured by sensor roll can be used to determine the neutral point over the roll
bite. It has been found that the calculated neutral point is close to experimental
value. The test to determine material's yield strength was also carried out. The
uncertainty analysis offrictioncoefficient measurement has been carried out.
(10) The elastic entry and exit zone in the roll bite should be considered in the analysi
especially w h e n the strip thickness is becoming thinner. The total pressure
distribution calculated by mixed film model coincides well with measured value.
The measured film thickness from 'oil drop method' is close to calculated data.
The rolling speed, reduction ratio, and viscosity property can affect the film
thickness.
(11) The effect of hydrodynamic inlet and outlet zone has been discussed in the mixed
film lubrication condition. The effect of elastic h u m p of strip at entry w a s also
discussed. So the hydrodynamic inlet and outlet zone should be considered in the
mixed film lubrication model.
(12) The influence of thermal effect in inlet, outlet and plastic work zones has been
discussed in the mixed film lubrication condition. F r o m the results, it can be seen
that the thermal effects affected the property of lubricant, hence, affected the film
pressure and film thickness as well as the total rolling force and torque. The
calculated results have verified by the experiments. So the thermal effects in the
mixed film lubrication should be considered, particularly for high speed rolling.
9.2 Recommendations
Friction and lubrication play an important role in cold rolling process. The fully
understanding of the nature of these factors will improve the model accuracy and
benefit industrial rolling mill. Therefore, the following suggestions are recommended
for future research in this area.
(1) In the measurement of friction coefficient and temperature along the roll bite by
using embedded pins in the roll, thefrictionbetween the roll and the pin as well as
the pin protrusion below roll surface are a major concerns. Because the friction
between roll and pin exists, the hysteresis happens during calibration process. It is
better to modify sensor roll design to overcome these concerns or avoid hard
specimen and higher reduction in the experiment.
(2) Increasing data acquisition speed to obtain more sampling points over the roll bite
for high rolling speed.
(3) Try to apply sensor roll in hot rolling experiment to measure friction coefficient
along the roll bite.
(4) Studying the temperature variation along the roll bite in hot rolling by using
thermocouple embedded the roll.
(5) Carrying out material yield strength test to consider the strain rate for
experimental specimen.
(6) Consider the random pattern of roll and surface roughness and apply in the mixed
film model. Extend surface roughness study from one dimension to two or three
dimension.
Chapter 9 Conclusions and Recommendations ÔQ
(7) Determine analytical expression of hydrodynamic film thickness at entry by
considering film thickness as well as the effect of strip elastic deformation at entry
and elastic recovery at exit.
(8) Apply the thermal model to industrial rolling mill to achieve better control of str
thickness and shape.

Appendix A Data Acquisition Program 209
Appendix A
Data Acquisition Program
#include <cvirte.h> /* Needed if linking in external compiler; harmless otherwise */

#include <userint.h>
#include "sensorm.h"
#include <formatio.h>
#include <dataacq.h>
#include <ansi_c.h>
#include "nidaqex.h"
#define T O T A L _ R O W 3000
#define T O T A L _ C O L U M N 15
#define NUMCHANELS 13
#define N U M S A M P L E S 30
extern RecallFunction(void);
static int daqpanel;
int TotalSaveNumber=0;
//short Chanels[]={l,3,4,5,6,7,8,9,10,l 1,12,13};

short Chanels[]={l,3,4,5,6,7,14,9,10,l 1,12,13,15};
short GainsD={l, 1,5,5,5,5,-1,10,20,5,2,20,5};
//short Gains[]={l,l,5,5,-l,-l,-l,5,20,5,5,10};
short p i B u f f e r | > n J M _ C H A N E L S * N U M _ S A M P L E S * 2 ] = {0}; // double-buffer for input
short p i H a l f B u f f e r [ N U M _ C H A N E L S * N U M _ S A M P L E S ] = {0}; // half buffer for input
//float BufferForSave|>njM_SAMPLES][NUM_CHANELS+2]={0.0};
//float BufferForShow[NUM_SAMPLES][NUM_CHANELS+2]={0.0};
float BufferForSave[NUM_CHANELS+2];
int BufferForShow_Flag=0;
float thicknessscalel;
float thicknessshiftl;
float thicknessscale2;
float rlscale;
float rlshift;
float thicknessshift2;
float rsscale;
float alscalel;
float alscale2;
float asscale2;
float alshift;
float rsshift;
float asscalel;
float loadcellscale 1;
float loadcellshiftl;
float asshift;
float loadcellscale2;
float loadcellshift2;
float torquescale;
float torqueshift;
//float samplerate=0.00005;
float samplerate;
char FileName[50J;
char LogFileName[320];
//Variable statements for Radial Loadcell and Strain Gauge, Angle Loadcell and Strain Gauge
//LoadCell (roll force), Strip Thickness, and Torque
float RLSG[2],ALSG[2],LC[2],ST[2],T;
float InitTime,ElapsedTime;
FILE* GlobalFp;
int Start_Flag=0;
int Stop_Flag=0;
intFileClose_Flag=l;
int StartSave_Flag=0;
int FileOpenFlag=0;
int BoardADInitializeO
{
unsigned long ulDevType; //Board type
short shDevType; //Borad type
f64 dSampRate;
il6 iSampUnits = 0;
il6 iScanUnits= 1;
il6 iSampTB;
il6iScanTB;
ul6uSampInt;
ul6uScanInt;
ul6 uNumChans=NUM_CHANELS;
il6iRetVal = 0;
//Return value for Error Handler
il6iDBmodeON=l;
//Doulbe-buffered O n
u32 ulCount = 2 * N U M _ C H A N E L S * N U M _ S A M P L E S ;
Get_DAQ_Device_Info (1, ND_DEVICE_TYPE_CODE, fculDevType);

shDevType=ulDevType;
Init_DA_Brds (1, &shDevType);
AI_Configure(l, 1, 1, 10, 1,0);

AI_Configure (1, 3, 1, 10, 1, 0);
AI_Configure(l,4, 1,10,1,0);
AI_Configure (1, 5,1,10,1, 0);
AI_Configure (1, 6, 1, 10, 0, 0);
AI_Configure (1, 7, 1, 10, 0, 0);
// AI_Configure (1, 8, 1, 10, 0, 0);
11
AI_Configure (1, 14, 1, 10, 1, 0);
AI_Configure (1, 14, 1, 10, 0, 0);
AI_Configure (1, 15, 1, 10, 1, 0);
AI_Configure(l,9, 1, 10, 1,0);
AI_Configure (1, 10, 1, 10, 1, 0);
AI_Configure (1, 11, 1, 10,1, 0);
AI_Configure (1, 12, 1, 10, 1, 0);
AI_Configure (1, 13, 1, 10, 1, 0);
DAQJDB_Config(l,l);
D A Q _ R a t e (250000.0, iSampUnits, &iSampTB, &uSampInt);
//dSampRate=0.00005;
dSampRate=0.00006;
DAQ_Rate(dSampRate, iScanUnits, &iScanTB, &uScanInt);
SCAN_Setup (1, 13, Chanels, Gains);
//SCAN_Setup (1, 13, Chanels, Gains);
//SCAN_Semp(l,NUM_CHANELS,Chanels,Gains);
S C A N S t a r t (1, piBuffer, ulCount, iSampTB, uSampInt,iScanTB, uScanlnt);
return 0;
}
void ReadDaqParameterO
{
FILE* fp;
fpôpenC'cAMiuWparameter.wdd'V'r");
fscanf(fp, "%f%f%f ,&thicknessscalel,&thicknessshiftl,&samplerate);

fscanf(fp, "%f%f%f',&thicknessscale2,&rlscale,&rlshift);
fscanf(fp, "%f%f%f',&thicknessshift2,&rsscale,«fealscalel);
fscanf(fp, "%P/of%f',&alscale2,&asscale2,&alshift);
fscanf(fp, "%f%f%f',&rsshift,&asscale 1 ,&loadcellscale 1);
fscanf(fp, " % f % f % f ',&loadcellshiftl ,&asshift,&loadcellscale2);
fscanf(fp, "%f%f%f',&loadcellshiff2,&torquescale,&torqueshift);
fclose(fp);
}
void WriteDaqParameterO
{
FILE* fp;
fp=fopen("c:\\liu\\parameter.wdd","w");
fprintf(fp, "%10.3f%10.3f%8.4fAn'',thicknessscalel,thicknessshiftl,samplerate);
fprintf(fp, "%10.3f%10.3f%10.3f\n",thicknessscale2,rlscale,rlshift);
fprintf(fp, "%10.3f%10.3f%10.3f\n",thicknessshift2,rsscale,alscalel);
fprintf(fp, "%10.3f%10.3f%10.3f\n",alscale2,asscale2,alshift);
fprintf(fp, "%10.3f%10.3f%10.3f\n",rsshift,asscalel,loadcellscalel);
fprintf(fp, "%10.3f%10.3f%10.3f\n",loadcellshiftl,asshift,loadcellscale2);
fprintf(fp, "%10.3f%10.3f%10.3foi",loadcellshift2,torquescale,torqueshift);
fclose(fp);
}
void SetDaqParameter()
{
SetCtrlVal (daqpanel, D A Q P A N E L _ T O R Q U E S H I F T , torqueshift);
SetCtrlVal (daqpanel, D A Q P A N E L _ T O R Q U E S C A L E , torquescale);
SetCtrlVal (daqpanel, D A Q P A N E L _ L O A D C E L L S H I F T 2 , loadcellshift2);
SetCtrlVal (daqpanel, DAQPANELL0ADCELLSCALE2, loadcellscale2);

SetCtrlVal (daqpanel, D A Q P A N E L A S S H I F T , asshift);
SetCtrlVal (daqpanel, D A Q P A N E L L 0 A D C E L L S H I F T 1 , loadcellshiftl);
SetCtrlVal (daqpanel, DAQPANELJLOADCELLSCALE1, loadcellscalel);
SetCtrlVal (daqpanel, DAQPANEL_ASSCALE1, asscalel);
SetCtrlVal (daqpanel, DAQPANEL_RSSHIFT, rsshift);
SetCtrlVal (daqpanel, DAQPANEL_ALSHIFT, alshift);
SetCtrlVal (daqpanel, D A Q P A N E L A S S C A L E 2 , asscale2);
SetCtrlVal (daqpanel, DAQPANEL_ALSCALE2, alscale2);
SetCtrlVal (daqpanel, DAQPANEL_ALSCALE1, alscalel);
SetCtrlVal (daqpanel, D A Q P A N E L R S SCALE, rsscale);
SetCtrlVal (daqpanel, DAQPANEL_THICKNESSSHIFT2, thicknessshift2);
SetCtrlVal (daqpanel, D A Q P A N E L R L S H I F T , rlshift);
SetCtrlVal (daqpanel, D A Q P A N E L R L S C A L E , rlscale);
SetCtrlVal (daqpanel, DAQPANEL_THICKNESSSCALE2, thicknessscale2);
SetCtrlVal (daqpanel, DAQPANEL_SAMPLE_RATE, samplerate);
SetCtrlVal (daqpanel, DAQPANELTHICKNESSSHIFT1, thicknessshiftl);
SetCtrlVal (daqpanel, DAQPANELTH1CKNESSSCALE1, thicknessscalel);
int _stdcall WinMain (HINSTANCE hlnstance, HINSTANCE hPrevInstance,

LPSTR IpszCmdLine, int nCmdShow)
{
if (InitCVIRTE (hlnstance, 0, 0) = 0) /* Needed if linking in external compiler; harmless
otherwise */
return -1; /* out of memory */
if ((daqpanel = LoadPanel (0, "sensorm.uir", DAQPANEL)) < 0)
return-1;
ReadDaqParameterO;
SetDaqParameterO;
BoardADInitialize();
// SetCtrlAttribute (daqpanel, DAQPANELSAVE, ATTRVISIBLE, 1);

// SetCtrlAttribute (daqpanel, D A Q P A N E L R E C A L L , ATTRVISIBLE, 1);
SetCtrlAttribute (daqpanel, D A Q P A N E L D A T A F I L E , ATTRVISIBLE, 0);
DisplayPanel (daqpanel);
RunUserlnterface ();
return 0;
int CVICALLBACK TorqueShiftF (int panel, int control, int event,

void *callbackData, int eventDatal, int eventData2)
{
switch (event) {
case E V E N T _ C O M M I T :
break;
case E V E N T _ V A L _ C H A N G E D :
GetCtrlVal (daqpanel, DAQPANELJTORQUESHIFT, &torqueshift);
break;
}
return 0;
}
int CVICALLBACK TorqueScaleF (int panel, int control, int event,
switch (event) {
case EVENTCOMMIT:
break;
GetCtrlVal (daqpanel, D A Q P A N E L T O R Q U E S C A L E , &torquescale);
break;
}
return 0;
}
int CVICALLBACK LoadCellShiftF2 (int panel, int control, int event,

void *callbackData, int eventDatal, int eventData2)'
{
switch (event) {
case E V E N T C O M M I T :
break;
case E V E N T V A L C H A N G E D :
GetCtrlVal (daqpanel, DAQPANEL_LOADCELLSHIFT2, &loadcellshift2);
break;
}
return 0;
}
int CVICALLBACK LoadCellScaleF2 (int panel, int control, int event,
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANELLOADCELLSCALE2,
&loadcellscale2);
break;
}
return 0;
}
int CVICALLBACK ASShiftF (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, D A Q P A N E L A S S H I F T , &asshift);
break;
}
return 0;
}
int CVICALLBACK LoadCellShiftFl (int panel, int control, int event,
switch (event) {
break;
case E V E N T J V A L C H A N G E D :
GetCtrlVal (daqpanel, D A Q P A N E L L O A D C E L L S H I F T 1 , &loadcellshiftl);
break;
}
return 0;
}
int CVICALLBACK LoadCellScaleFl (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANELLOADCELLSCALE1,
&loadcellscalel);
break;
}
return 0;
}
int CVICALLBACK ASScaleFl (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, D A Q P A N E L A S S C A L E 1 , &asscalel);
break;
}
return 0;
}
int CVICALLBACK RSShiftF (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANELJRSSHIFT, &rsshift);
break;
}
return 0;
}
int C V I C A L L B A C K ALShiftF (int panel, int control, int event,

Appendix A Data Acquisition Program
{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANEL_ALSHIFT, &alshift);
break;
}
return 0;
}
int CVICALLBACK ASScaleF2 (int panel, int control, int event,

switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANEL_ASSCALE2, &asscale2);
break;
}
return 0;
}
int CVICALLBACK ALScaleF2 (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, D A Q P A N E L A L S C A L E 2 , &alscale2);
break;
}
return 0;
}
int CVICALLBACK ALScaleFl (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANEL_ALSCALE1, &alscalel);
break;
}
return 0;
}
int CVICALLBACK RSScaleF (int panel, int control, int event,

A Data Acquisition Program 216
{
switch (event) {
break;
GetCtrlVal (daqpanel, D A Q P A N E L R S S C A L E , &rsscale);
break;
}
return 0;
}
int CVICALLBACK ThicknessShift2 (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANEL_THICKNESSSHIFT2, &thicknessshift2);
break;
}
return 0;
}
int CVICALLBACK RLShiftF (int panel, int control, int event,

{
switch (event) {
break;
case E V E N T V A L _ C H A N G E D :
GetCtrlVal (daqpanel, DAQPANEL_RLSHTFT, &rlshift);
break;
}
return 0;
}
int CVICALLBACK RLScaleF (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANEL_RLSCALE, &rlscale);
break;
}
return 0;
}
int CVICALLBACK ThicknessScaleF2 (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANELTHICKNESSSCALE2,
&thicknessscale2);
break;
}
return 0;
}
int CVICALLBACK SampleRateF (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANEL_SAMPLE_RATE, &samplerate);
if(samplerate<0.001) samplerate=0.001;
if(samplerate>1.0) samplerate=1.0;
SetCtrlAttribute (daqpanel, D A Q P A N E L D A Q T I M E R A T T R I N T E R V A L ,
samplerate);
break;
}
return 0;
}
int CVICALLBACK ThicknessShiftl (int panel, int control, int event,

{
switch (event) {
break;
GetCtrlVal (daqpanel, DAQPANELJTHICKNESSSHIFTl, &thicknessshiftl);
break;
}
return 0;
}
int CVICALLBACK ThicknessScaleFl (int panel, int control, int event,

{
switch (event) {
case EVENT_COMMIT:
break;
GetCtrlVal (daqpanel, DAQPANELJTHICKNESSSCALEl,
&thicknessscalel);
break;
}
return 0;
}
int CVICALLBACK Stop (hit panel, int control, int event,

switch (event) {
Stop_Flag=l;
Start_Flag=0;
SetCtrlAttribute (daqpanel, D A Q P A N E L _ D A Q S T A R T , ATTRVISIBLE, 1);
SetCtrlAttribute (daqpanel, DAQPANEL_DAQQUIT, ATTRVISIBLE, 1);
SetCtrlAttribute (daqpanel, DAQPANEL_DAQSTOP, ATTRVISIBLE, 0);
SetCtrlAttribute (daqpanel, DAQPANELDAQTIMER ATTRENABLED,

0);
if((StartSave_Flag==l & & FileClose_Flag=0 & & BufferForSave[5]<50.0) ||
Stop_Flag=l || TotalSaveNumber>=TOTAL_ROW*TOTAL_COLUMN) {
FileClose_Flag=l;
FileOpenFlag=0;
StartSave_Flag=0;
fclose(GlobalFp);
}
// SetCtrlAttribute (daqpanel, D A Q P A N E L S G N L T I M E R A T T R E N A B L E D ,
0);
break;
break;
}
return 0;
}
int CVICALLBACK SaveToFile (int panel, int control, int event,

{
intij;
char *filename;
FILE* fp;
switch (event) {
break;
break;
}
return 0;
}
/*
int CVICALLBACK SaveToFile (int panel, int control, int event,
{
intij;
char *filename;
FILE* fp;
switch (event) {
filename=FileName;
fp=fopen(filename,"w");
for(i=0;i<NUM_SAMPLES;i++) {
for(j=0y <NUM_CHANELS+2;j++) {
fprintf(fp,"%f\t",BufferForSave[i]0]);
}
fprintf(fp,'V');
}
fclose(fp);
break;
break;
}
return 0;
int CVICALLBACK Start (int panel, int control, int event,

{
switch (event) {
Start_Flag=l;
Stop_Flag=0;
ClearStripChart (daqpanel, DAQPANELTHICKNESS);

ClearStripChart (daqpanel, D A Q P A N E L R A D I A L ) ;
ClearStripChart (daqpanel, D A Q P A N E L A N G L E ) ;
ClearStripChart (daqpanel, D A Q P A N E L R O L L F O R C E ) ;
ClearStripChart (daqpanel, D A Q P A N E L _ T O R Q U E ) ;
SetCtrlAttribute (daqpanel, D A Q P A N E L D A Q S T A R T , ATTRVISIBLE, 0);

// SetCtrlAttribute (daqpanel, DAQPANEL_SAVE, ATTR_VISIBLE, 1);
// SetCtrlAttribute (daqpanel, D A Q P A N E L R E C A L L , ATTRVISIBLE, 1);
SetCtrlAttribute (daqpanel, D A Q P A N E L D A Q Q U I T , ATTRVISIBLE, 0);
SetCtrlAttribute (daqpanel, D A Q P A N E L D A Q S T O P , ATTRVISIBLE, 1);
InitTime = Timer () + 0.10;

if(samplerate<0.005)samplerate=0.005;
if(samplerate> 1.0) samplerate=l .0;
SetCtrlAttribute (daqpanel, DAQPANELDAQTIMER,
ATTR_INTERVAL,samplerate);
SetCtrlAttribute (daqpanel, DAQPANEL_DAQTIMER, A T T R E N A B L E D ,
i);
break;
break;
}
return 0;
}
int CVICALLBACK shutdown (int panel, int control, int event,

{
switch (event) {
SetCtrlAttribute (daqpanel, D A Q P A N E L D A Q T I M E R , A T T R E N A B L E D , 0);
DAQ_Clear(l);
DAQ_DB_Config(l,0);
QuitUserlnterface (0);
break;
break;
}
return 0;
}
int CVICALLBACK DaqReiniToDefault (int panel, int control, int event,

{
switch (event) {
ReadDaqParameter();
SetDaqParameterQ;
break;
break;
}
return 0;
}
int CVICALLBACK DaqMakeDefault (int panel, int control, int event,

{
switch (event) {
WriteDaqParameterO;
break;
break;
}
return 0;
}
int CVICALLBACK FileNamelnput (int panel, int control, int event,

{
switch (event) {
case EVENTCOMMIT:
break;
GetCtrlVal (daqpanel, DAQPANEL_DATAFILE, &LogFileName);
break;
}
return 0;
}
int DataAquire(il6 iNumPoints)

{
// static unsigned long iNumDaq=0;
// short int *iPtr;
// double tl;
intij;
float SThicknessEntry, SThicknessExit, SForceDrive, SForceOperation;
float SLVDTDrive, SLVDTOperation, SSpeed, STorque,Temperarure;
float SRadialLoad, SAngleLoad, SRadialStrain, SAngleStrain;
float TempReadings[NUM_CHANELS];
char* filename;
struct tm *ptr;
timet It;
ElapsedTime = Timer() - InitTime;
if(BufferForShow_Flag>=NUM_SAMPLES) {
BufferForShow_Flag=0;
for(i=0;i<NUM_SAMPLES;i++) {
for(j=0;j<NUM_CHANELS+2;j++) {
BufferForSave[i] [j]=BufferForShow[i] [j];
}
}
}
S C A N D e m u x (piHalfBuffer, iNumPoints , N U M C H A N E L S , 0);
if (iMeasureMode == 0) {
iPtr = piHalfBuffer;
fH[0] = fShift_H + fScale_H0 * daqfilter(iPtr, N U M _ S A M P ) ;
iPtr += N U M S A M P ;
fH[l] = fShiftJi + fScaleJiO * daqfilter(iPtr, N U M S A M P ) ;
}
else {
iPtr = piHalfBuffer;
fH[0] = fShift_H + fNorm_H + fScale_H0 * (0.5 - daqfilter(iPtr, N U M _ S A M P ) ) ;
iPtr += NUM_SAMP;
fH[l] = fShift_h + fNorm_h + fScale_h0 * (0.5- daqfilter(iPtr, N U M S A M P ) ) ;
}
for(i=0;i<NUM_CHANELS;i++) {
TempReadings[i]=0;
for(j=0;j<NUM_SAMPLESy++) {
TempReadings[i]=TempReadings[i]+piHalfBuffer[i*NUM_SAMPLES+j];
}
TempReadings[i]=TempReadings[i]/NUM_SAMPLES;
}
/*
prmtf('Uicknessshiftl=%f,thicknessscalel=%f,TempReadmgs[0]=%fm^thicknessshiftl,thickne
ssscalel,TempReadings[0]);
prmtf("micknessshift2=%f,thicknessscale2=%f,TempReadh^^
TempReadings[ 1 ]);
printf("loadcellshiftl=%f,loadcellscalel=%f,TempReadings[3]=%f\n",loadcellshiftl,loadcellscalel,Temp
Readings[3]*2.0/4095);
prmtf('1oadcellshift2=%f,loadcellscale2=%f,TempReadmgs[2]=%fn",loadcellshift2,loadcellscale2,Temp
Readings[2]*2.0/4095);
printf("TempReadings[4]=%f\n",TempReadings[4]);
printf("SLVDTOperation=%f\n",TempReadings[5]);
printf("SSpeed=%f\n",TempReadings[6]);
prmtf("torqueshifr=%f,torquescale=%f,TempReadmgs[7]=%fn",torqueshift,torquescale,TempR
eadings[7]);
prmtf(''rlshift=%f,rlscale=%f,TempReadmgs[8]=%f\n",rlshift,rlscale,TempReadings[8]);
prmtf("alshifr=%f,alscalel=%f,alscale2=%f,TempReadmgs[9]=%fo",alshift,alscalel,alscale2,T
empReadings[9]);
prmtf('Ysshifr=%f,rsscale=%f,TempReadmgs[10]=%fui",rsshift,rsscale,TempReadings[10]);
printf(" asshift=%f,asscale 1 =%f,asscale2=%f,TempReadings[ 11 ]=%fxn" ,asshift,asscale 1 ,asscale2
,TempReadings[ 11 ]);
*/
SThiclmessEntry==thicknessshiftl+thicknessscalel*TempReadings[0]*10.0/4095;
SThicknessExit=îcknessshift2+thicknessscale2*TempReadngs[l]*10.0/4095;
SForceDrive=abs(loadcellshiftl+loadcellscalel*TempReadings[3]*2.0/4095);
SForceOperation=abs(loadcellshift2+loadcellscale2*TempReadings[2]*2.0/4095);
SLVDTDrive=0.0+2.32194*TempReadings[4] *2.0/4095;
SLVDTOperation=0.0+11.58762*TempReadings[5]*2.0/4095;
SSpeed=0+10.543*TempReadings[6]*20.0/4095;
STorque=0+torquescale*TempReadings[7]* 1.0/4095;
//SRadialLoad=rlshift+rlscale*TempReadings[8]*0.5/4095;
SRadialLoad=0+5930.0*TempReadings[8]*0.5/4095;
//SAngleLoad=alshift+alscalel*TempReadings[9]*2.0/4095+alscale2*TempReadings[9]*2.0/40
95*TempReadings[9]*2.0/4095;
SAngleLoad=0+693.0*TempReadings[9]*2.0/4095;
//SRadialStrain=rsshift+rsscale*TempReadings[10]*2.0/4095;
SRadialStrain=0+952.2*TempReadings[10]*5.0/4095;
//SAngleSfrain=asshift+asscalel*TempReadings[ll]/4095+asscale2*TempReadings[ll]/4095*
TempReadings[l l]/4095;
SAngleStrain=0+4493.0*TempReadings[ 11 ] *0.5/4095;
Temperature=65.50*TempReadings[12]*2.0/4095;
// printf("%f, %f\n",TempReadings[12],Temperature);
// shortChanels[]={ 1,3,4,5,6,7,8,9,10,11,12,13};
// shortGains[]={l, 1,5,5,-1,-1,0.5,5,20,5,5,10};
// Gain*Coefficient=10.0v
// printf("%f%f\n",TempReadings[6]*20.0/4095,SSpeed);
BufferForSave[0]=ElapsedTime;
BufferForSave[ 1 ]=SRadialLoad;
BufferForSave[2]=SAngleLoad;
BufferForSave[3]=SRadialStrain;
BufferForSave[4]=SAngleStrain;
BufferForSave[5]=SForceDrive+SForceOperation;
BufferForSave[6]=SForceDrive;
BufferForSave[7]=SForceOperation;
BufferForSave[8]=STorque;
Appendix A Data Acquisition Prngrnm 223
BufferForSave[9]=SThicknessEntry;
BufferForSave[ 10]=SThicknessExit;
BufferForSave[l l]=SLVDTDrive;
BufferForSave[ 12]=SLVDTOperation;
BufferForSave[ 13]=SSpeed;
BufferForSave[ 14]=Temperature;
if(Start_Flag=l && BufferForSave[5]>=50 0) {

// printf("%f\n", BufferForSave[5]);
StartSave_Flag=l;
if(FileOpenFlag==0){
lt=time(NULL);
ptr=localtime(&lt);
filename=asctimeOj>tr);
j=0;
for(i=0;i<24;i++){
if(isalnum(filename[i])) {
FileNameO]=filename[i];
j=i+i;
}
}
FileNamelj]-.';
FileName[j]='d';
j=j+i;
FileName[j]='a';
FileNameOl-t';
j=j+i;
FileNameO]='\0';
GlobalFp=fopen(FileName,"w");
FileOpenFlag=l;
FileClose_Flag=0;
TotalSaveNumber=0;
}
for(i=0;i<NUM_CHANELS+2;i++) {
fprintf(GlobalFp,"%9.3f\t",BufferForSave[i]);
}
fprintf(GlobalFp,"\n");
TotalSaveNumber++;
}
if((StartSave_Flag==l && FileClose_Flag==0 && BufferForSave[5]<50.0) || Stop_Flag==l ||
TotalSaveNumber>=TOTAL_ROW*TOTAL_COLUMN){
FileClose_Flag=l;
FileOpenFlag=0;
StartSave_Flag=0;
fclose(GlobalFp);
// SetCtrlAttribute (daqpanel, D A Q P A N E L S A V E , ATTR_VISIBLE, 0);
// SetCtrlAttribute (daqpanel, D A Q P A N E L _ R E C A L L , ATTR_VISIBLE, 1);
}
RLSG[0]=SRadialLoad;
RLSG[l]=SRadialStrain;
ALSG[0]=SAngleLoad;
ALSG[l]=SAngleStrain;
LC[0]=SForceDrive;
LC [ 1 ]=SForceOperation;
ST[0]=SThicknessEntry;
ST[l]=SThicknessExit;
T=STorque;
BufferForShow_Flag=BufferForShow_Flag+1;
return(O);
}
int DaqDisplay(void)
{
// double dV;
// dV= fVO;
PlotStripChart (daqpanel, DAQPANELRADIAL, RLSG, 2, 0, 0, VALFLOAT);
// SetCtrlVal (iHandleDaqPanel, DAQPANELTHICKENTRY, fH[0]);

// SetCtrlVal (iHandleDaqPanel, D A Q P A N E L T H I C K E X I T , fH[l]);
PlotStripChart (daqpanel, DAQPANELANGLE, ALSG, 2, 0, 0, VALFLOAT);
// SetCtrlVal (iHandleDaqPanel, DAQPANELFORCEDR fFc[0]);

// SetCtrlVal (iHandleDaqPanel, D A Q P A N E L F O R C E O P , fFc[ 1 ]);
PlotStripChart (daqpanel, DAQPANELROLLFORCE, LC, 2, 0, 0, VALFLOAT);

PlotStripChart (daqpanel, D A Q P A N E L T H I C K N E S S , ST, 2, 0, 0, V A L F L O A T ) ;
// SetCtrlVal (iHandleDaqPanel, DAQPANELPSTDR, fPst[0] );

// SetCtrlVal (iHandleDaqPanel, D A Q P A N E L P S T O P , fPst[ 1 ]);
PlotStripChartPoint (daqpanel, DAQPANELTORQUE, T);
// SetCtrlVal (iHandleDaqPanel, DAQPANELSPEEDROLL, fVO);
return(O);
}
int CVICALLBACK DAQTimer (int panel, int control, int event,

{
switch (event) {
static unsigned long ulPtsTfr;

static short iHalfReady;
static short iDAQstopped;
case EVENTTIMERTICK:
do{
DAQDBHalfReady (1, &iHalfReady, &iDAQstopped);
} while (iHalfReady = 0);
DAQ_DB_Transfer (1, piHalfBuffer, &ulPtsTfr, &iDAQstopped);
DataAquire(NUM_CHANELS*NUM_SAMPLES);
DaqDisplay();
break;
}
return 0;
}
int CVICALLBACK RecallHistory (int panel, int control, int event,

switch (event) {
if(Start_Flag=l && Stop_Flag=0)

MessagePopup ("Operation Guide","PleasefirstStop sampling data, then Recall history data!");
else {
strcpyO-ogFileName,"");
FileSelectPopup ("c:\\liu", "*.dat", "*.dat", "Recall History Data",
V A L L O A D B U T T O N , 0, 0, 1, 1,
LogFileName);
if(strlen(LogFileName)=0)
MessagePopup ("Operation Guide","PleasefirstSelect afilename, then recall history data!");
else RecallFunction();
}
break;
case EVENT_VAL_CHANGED:
break;
}
return 0;
}
Appendix B Empirical Formula of Friction Coefficient
Appendix B
Empirical Formula of Friction
Coefficient
Appendix B Empirical Formula of Friction Coefficient 227
Aluminium alloy
F = -0.311 + 1.1196X-5.7511X2 +15.073Z3 -16.2347X4 +6.3158X5 (B-l)
fr By X ]
Y =-0.311 41.11955 XO 5.75113 X~2 4 15.073 X-3 A 16.2347 X-4 4 6.31578 X~5
[Summery of Fit J
RSquore 0.987241
RSquare Adj 0.984963
Root Meon Square Error 0.017159
Mean of R esponse -0.08662
Observations (or S u m W ats) 34
)
IA naly sis of V ariance
Source DF
3
Model 5 0.63789339 0.127579 433.3124
Error 28 0.00824394 0.000294 Prob>F
C Total 33 0.64613734 C0001
(Parameter Es imates J
Term Estimate Std Error t R alio Prob>|t|
Intercept 43.31099 0.023364 -13.31 <.0001
X 1.1195493 0.434905 2.57 0.0156
X-2 -5.751129 2.525276 -2.28 0.0306
X~3 15.072977 6.117701 2.46 0.0202
XA -16.23474 6.513324 -2.49 0.0189
X-5 6.3157849 2.519172 2.51 0.0183
Y = -0.1912 -0.5218X + 4.8515X2 -14.2263X3 + 17.5845X4 -7.3474X5 (B-2)
sy* J
0.15
>
Poly nomial F it deqree=5
[PoIy nomiaI Fit d eq re e=5

Y =-0.1912 0 0.52179 X 4 4.85153 X-2 0 14.2263 X-3 4 17.5845 X-4 0 7.34737 X-5
( S u m m a r y of Fit J
RSquore 0.989694
R Square Adi 0.988338
Observations (or S u m W qts) 44
>
(A no Iy s is fc Va riance
Source DF
ZD
S u m of Squares M e a n Square F Ratio
Model 5 0.54254929 0.108510 729.8553
Error 38 0.00564958 0.000149 P ro b >F
C Total 43 0.54819887 C0001
V. )
(P a ra m e te rEstimates J
Term Estimate Std Error t Ratio Probst)
Intercept -0.191197 . 0.013638 ^4.02 <,0001
X -0.52179 0.258716 -2.02 0.0508
X-2 4.851532 1.524333 3.18 0.0029
X-3 -14.22629 3.730978 -3.81 0.0005
XM 17.584474 4.003541 4.39 <.0001
X-5 -7.347368 1.558419 -4.71 <.0001
Y = -0.2297 -0.0559^ + 3.9758X2 -13.5358Z3 +17.5897X4 -7.5600X5(B-3)
(V By X )
0.2
0.1 -
-0.0 -
-0.1 -
-0.2 -
-0.3
.2
Poly nomial F it deqree=5
[Polynomial Fit degree=5 J

Y =-0.2297 0 0.05592 X 4 3.97582 X~2 0 13.5358 X-3 4 17.5897 X 4 0 7.56003 X-5
*\
(Summary of Fit j
R Square 0.995866
R Square Ad j 0.994832
Observations (or S u m W qts) 26
f
\
[Analysis cf Variance
S o u re e DF
ZJ
Model 5 0.39844035 0.079688 963.5832
Error 20 0.00165399 0.000083 Prob>F
C Total 25 0.40009435 C0001
>
\
(P a ra m e te rEstimates j
intercept -0.229688 0.015582 -14.74 C0001
X -0.055915 0.282655 -0.20 0.8452
M 3.9758197 1.609358 2.47 0.0226
X-3 -13.53576 3.845194 -3.52 0.0022
X-4 17.589737 4.050862 4.34 0.0003
X-5 -7.560033 1.553342 4.87 <.0001
/
(4) Aluminium alloy 6060-T5, 4 0 . 4 5 % reduction, and 15 rpm rolling speed under
Y = -0.2425 + 0.AS16X + 0.6304X2 -3.8617Z3 +5.7588X4 -2.6251X5 (B-4)
(V By X ]
0.15
0.10 -
0.05 -
-G.00 -
-0.05 -
>-
-0.10 -
-0.15
-0.20
-0.25
.0 .6 1.2
Poly nomial Fit deqree=5
Polynomial Fit deqree=5

Y =-0.2425 40.45762 X 4 0.6304 X~2 0 3.86168 X-3 4 5.75881 X 4 0 2.62505 X-5
^
(Summary of Fit J
RSquare 0.99338
R oot Mean Square Error 0.010629
0 bservations (or S u m W gts) 14
[Analysis of Varionce J
Source DF S u m of Squares Mean Square F Ratio
Model 5 0.13561555 0.027123 240.0849
Error 0.00090378 0.000113 Prob>F
C Total 0.13651934 <.0001
\
(Parameter Estimates j
Intercept -0.242531 0.036677 -6.61 0.0002
X 0.4576204 0.604296 0.76 0.4706
M 0.6304 3.207605 0.20 0.8491
X~3 3.86168 7.291458 -0.53 0.6108
X-4 5.7588132 7.396074 0.78 0.4586
X-5 -2.625051 2.750663 -0.95 0.3679
Y = -0.4215 + 2.6328X - 9.0353JT2 + 15.4036JT3 -11.7617X 4 + 3.2821X5 (B-5)
[V By X "j
0.15
0.10 -
0.05 -
-0.00 -
>-
-0.05 -
-0.10 -
-0.15 -
-0.20 IT
.00 .25 .50 .75 1.00 .25
Poly nomial Fit degree=5
[Poly nomial Fit degree=5

Y =-0.4215 4 2.63277 X u 9.03526 X~2 4 15.4036 X-3 0 11.7617X4 43.28212 X-5
[Summary of Fit I
RSquare
RSquare Adj
Root Mean Square E rror
Observ ations (or S u m W gfs)
(Analysis (f Variance J
0.010598 9
Model 5 0.05298979
Error 0 0.00000000 0 Prob>F
9
C Total 5 0.05298979
(Paramete Estmates J
Term Estimate Std Error t Ratio Pre b>|t|
9 9 9
Intercept -0.421475
9 9 9
X 2.63277
X-2 43.035265 9 n 9
9 9 9
X-3 15.40359
i 9 9
X4 -11.76174
9 1 9
X-5 3.28212
Y = -0.401 +1.5333X - 2.3SAX2 +1.3227X 3 (B-6)
My X )
(Polynomial Fit degree=5

Y =+3.401 4 1.53333 X u 2.384 X-2 4-1.32267 X-3 4 0 X 4 4 0 X - 5
\
(Summary of Fit J
RSquare 1
R Square Adj ?
Root Mean Square Error 7

Mean of R esponse 4],0435
)
[Analysis of Variance
Model 3 0.02416900 0.008056 7
Error 0 0.00000000 Prob>F
C Total 3 0.02416900
\
(Parameter Estimates
Term
D Estimate Std E rror t Rat io Prob>|t|
Intercept Biased -0.401 7 7 7
X Biased 1.5333333 7 7 9
X-2 Biased -2.384 7 7 9
X'3 Biased 1.3226667 7 9 7

X4 Zeroed 0 9 7 ?
X-5 Zeroed 0 7 7 7
/
Y = -0.129 - 0.0135Z + 0.2205X 2 (B-7)
CY By X )
0.10
0.05 -
-0.00 -
>
-0.05 -
-0.10 -
-0.15
.25 .50
-r .00 1.25
.75
X

Y =-0.129 u 0.0135 X 4 0.2205 X-2 4 0 X-3 4 0 X 4 4 0 X - 5
(Summary of Fit J
RSquare 1
RSquare Adj 7
Root Mean Square Error 7
Mean of R esponse 4].02367
0 bserv ations (or S u m W gts) 3
V
J
Analysis of Variance j
Model 2 0.01788467 0.008942 9
Error 0 0.00000000 Prob>F

C Total 2 0.01788467
Parameter Estimates )
Term Estimate Std E rror t Ratio Prob>|t|
9 9 7
Intercept Biased -0.129
X Biased -0.0135 7 7 7
M Biased 0.2205 7 7 9
X-3 Zeroed 0 7 7 7
X4 Zeroed 0 7 9 7
Zeroed 0 9 9 7
Carbon steel
(1) Carbon steel B H P - 3 0 0 , 3 1 . 1 6 % reduction, and 3 r p m rolling speed under
Y = -0.3302 + 1.9983X - 7.3978X2 + 14.3123Z3 - 13.0121X4 + 4.5385Z5 (B-8)
(^ By X ]
0.15
>-
1.2
Poly nomiolFit degree=5

Y =-0.3302 4 1.99831 X 0 7.39784 X-2 4 14.3123 X-3 0 13.0121 X 4 44.53848 X-5
>
(Summary of Fit j
RSquare 0.981178
Observ ations (or S u m W gts) 29
)
[Analysis cf Variance
Source DF
ZD
Model 5 0.23352181 0.046704 239.7996
Error 23 0.00447957 0.000195 Prob>F
C Total 28 0.23800139 <,0001
[Parameter Estimates J
Intercept -0.330193 0.021697 -15.22 <.0001
X 1.9983138 0.39808 5.02 C0001
X~2 -7.397841 2.286138 -3.24 0.0037
X-3 14.312254 5.495509 2.60 0.0159
X4 -13.01212 5.816332 -2.24 0.0353
X-5 4.538476 2.238753 2.03 0.0544
T = -0.1795-0.0278X + 1.0225X2 -1.5978Z3 + 1.3248X4 -0.4793X5 (B-9)
fr By X ]
0.10
0.05 -
-0.00 -
-0.05 -
>
-0.20
~r i — r \ i
.7 .8 .9 1.0 1.1
i
.3 .4 .5
(Poly nomial Fit degree=5

Y =-0.1795 0 0.02777 X 4 1.02247 X-2 u 1.59781 X-3 4 1.32482X4 0 0.47933 X-5
(Summary of Fit J
RSquare 0.987984
Observ ations (or S u m W gts) 40
[Analysis of Variance
Model 5 0.25381231 0.050762 559.1081
Error 34 0.00308692 0.000091 P ro b >F
C Total 39 0.25689923 C0001
[Parameter Es imates
Term Estimate Std Error 1 R atio Prob>|t|
Intercept -0.17945 0.011435 -15.69 <.0001
X -0.027775 0.215539 -0.13 0.8982
X-2 1.0224683 1.263635 0.81 04241
X-3 -1.597812 3.082086 -0.52 0.6075
X4 1.3248233 3.29842 0.40 0.6905
X-5 -0.479328 1.28114 -0.37 0.7106
F = -0.1651 + 0.3139X-0.1365X2-0.5024Z3+1.1614X4-0.5689Z5(B-10)
[Y By X
0.15
0.10 -
0.05 -
-0.00 -
>
-0.05 -
-0.10 -
-0.15
-0.20
r 1.2
1.0
[poly nomial Fit degree=5

Y =-0.1651 40.3139 X 0 0.13654 X-2 u 0.50241 X-3 4 1.16144 X 4 0 0.56891 X-5
"^
(Summary of F it J
RSquare 0.989984
R Squore Ad j 0.987897
R oot Mean Square E rror 0.008169
0 bservations (or S u m W gts) 30
)
\
[A no Iy s isof V aria nee
Source DF
D
Sum of Squares Mean Square F Ratio
Model 5 0.15829851 0.031660 4744386
Eiror 24 0.00160154 0.000067 Prob>F
C Total 29 0.15990005 C0001
[Parameter Estimates j
Term Estimate Std Error t R atio Prob>|+|
Intercept -0.165092 0.012336 -13.38 C0001
X 0.3139017 0.227072 1.38 0.1796
X-2 -0.136537 1.30731 -0.10 0.9177
X-3 -0.502411 3.148096 -0.16 0.8745
X4 1.1614361 3.336349 0.35 0.7308
X-5 -0.568906 1.285592 -0.44 0.6621
Y = -0.1897-0.0109X + 2.6734X 2 -6.7318X 3 +6.2728X 4 -1.8717X 5 (B-11)
(V By X )
(poly nomial F it degree=5 J

Y =-0.1897 0 0.01093 X 4 2.67341 X-2 u 6.73178 X-3 4 6.27279X4 0 1.87166 X-5
[Summary of Fit J
RSquare 0.995291
Root Mean Square E rror 0.009499
[Analysis of Variance J
Source DF S u m o f Squares MeanSquare F R atio
Model 5 0.07628392 0.015257 169.0749
Error 4 0.00036095 0.000090 Prob>F
C Total 9 0.07664487 <.0001
Intercept -0.189746 0.057886 -3.28 0.0306
X -0.010932 0.876371 -0.01 0.9906
X-2 2.6734119 4.383797 0.61 0.5749
X-3 -6.731778 9.551613 -0.70 0.5198
X4 6.2727899 9.382803 0.67 0.5404
X-5 4.871664 3.401297 -0.55 0.6114
lubricated condition. The lubricant is a mineral oil A L P H A S P 1000.
Y = -0.2381 + 0.7299JT-1.590IX2 +3.8199X3 -5.1937X4 +2.5867X5(B-12)
fr By X "]
0.15
0.10
0.05
-0.00 -
-0.05 -
-0.10 -
-0.15 -
-0.20
.00 .25 .50 .75 1.00 1.25
Poly ncmial Fit degree=5
[Polyncmial Fit degree=5 J

Y =-02381 4-0.72992X a 1.59G06 X-243.81987 X ^ u 5.19371 X 4 42.58568 & §
[Surrmary of Fit j
RSqucre 0.999277
Root Mean Squae Error 0.00421
Mean of Respx>nse -0.03317
Observations (crSumW ate) 8
\
[Analysis of Variarce
Source DF S u m of Squares Mean Squae F Ratio
Model 5 0.05160639 0.010321 552.8476
Error 2 O.O0DO3734 0.000019 PrcteF
C Total 7 0.05164423 0.0018
V
\
Term Estimate Std Error t Ratio ProMtl
1 ntercqot 40.238055 0.044538 -6.34 0.0333
X 0.7299225 0.626466 1.17 0.3641
X-2 4.590035 2.978262 43.53 0.6468
X-3 3.8198655 625656 0.61 0.6036
X4 -519371 5.979073 -0.87 0.4766
X5 2.586677 2120742 1.22 0,3469
)
x C Energy Equation
Appendix C
Energy Equation
Inlet zone
The general form of energy equation [Incropera and Dewitt, 1990] can be written as
follows:
D[CT) , x T dp Dp ^
vp (C-l)
p ' = v(kAT) .-A^- + 0
H
Dt p dT Dt
Extend Eq.(C-l)
dT dT dT) Jd2T d2T d2T) T dpi dp dp dp

+ o
P°i U + V r- W-
j y6x
2+ 2+
dy 2
dz ) p'dT\" dx d}
dx dy dz
(C-2)
where
2
fdv)2 (dw\ du dv \ fdu
A.. dw
A„V (dv dwy
0 = 77 (du) + 2 +2 + — + — + — + — + dz dy
•+ • (C-3)
ydx; ydz j
dy dx dz dx
In the rolling process, the dominant modes of heat transfer are conduction normal to the
surfaces and convection parallel to the surfaces. So Eqs. (C-2) and (C-3) can
rewritten as:
Appendix C Energy Equation 240
dT . d2T T dp dp (du)
(C-4)
pc u — = k— .—.w — + 77 —
P
dx dz2 p dT dx \dz)
If the density is not changed with the temperature, and the convection along rolling
direction is neglected and the only significant mode of heat transfe
normal to the surfaces, so Eq. (C-4) can be reduced to (for the lubri
d2 Tf
n (du) (C-5)
~a7 •f\dzj
The equation of force equilibrium in lubricant may be written:
dpf __ dx
(C-6)
dx dz
du
X = 77
(C-7)
dz
Substitute Eq. (C-7) into Eq. (C-6), and then integrating
dp, du r (C-8)
z—- = 7 — + /n
dx dz
where
77 = n0 exp[opf - p(Tm -T0)] (C-9)
Substitute Eq. (C-9) into Eq. (C-8)
(C-10)
'''' /n = "* expfe7/ " /^(rm ~ r o ) M
dx dz
Integrating Eq. (C-10)

U = Ur
z = h/2
z =0
z = -h/2-
U-Uw
Figure Cl Flow between moving surfaces
The boundary conditions for Eq. (C-l 1) (shown in Figure Cl):
z = —, u = ur (ur = rolls speed)

(C-12)
i
z- —, u = uw (uw = strip speed)
Yielding:
tf_ dp^ h
8 dx 2 , f (C-13)
U = <~ J ^
n0Qxp[apf-/3(Tm-T0)]
h2 dpf h
8 dx (C-14)
u...n =exp[op -/3(T + /:22
0 f m -T0)]
Subtract Eq. (C-14) from Eq. (C-13)
uw~ur (C-15)
fxx=n0 Qxp[apf -fi{Tm - r 0 ) ] . — -
Substitute fn into Eq. (C-10)

du z dpf uw~ur (C-16)
~dz rj dx h
Substitute (C-16) into Eq. (C-5)
d2Tf n (z dpf uw-u (C-17)

2
Ik TJ dx h
The first-order thermal Reynolds equation can be expressed as:
dpf , ( \h-hx (C-18)

-T- = -6Ww + ur)—-3—
dx h
Substitute Eq. (C-18) into Eq. (C-17), and replace z by y
d2Tf JL -6(uw+ur) h-hx y_]__ uw -u, (C-19)

kf\ h h' h h
In the inlet zone

\uw+ur=ur[(l + Sf)y2/yx+\]
(C-20)
\uw-ur=ur[(\ + Sf)y2/yx-\]
yx and y2 are strip thickness at entry and exit of roll bite, respectively
Substitute Eq. (C-20) into Eq. (C-19)
d2Tf ^ ^2/yl+^^-hsf)yJyî]
nur +S ^
kf \
Dimensionless items:
- T
f y rr h
u ^ ff,=-4r. Px=^r,E = pYQ
2 2
X I2R kfl0
f
7V=-A T = f,
h H =- /2R ,
x T7TZ: H.=xx/2R
P2 ={2Rlx2)kf{xxl^cwurkJ2^ ={2Rlx2)k{(xxlnprcrurkrr,
(C-22)
C = exp(-o??/)
Hence
d2T f Wr .{6l\ + Sf)([-£)+l\H-Hx)xY/H-l\ + Sf)([-£)-\}.

dy2 kfT0'K ~LV~ ' f/K ' ,v ""' ' ": '" ' " /»"
(C-23)
s is reduction ratio, {yx -y2)l' yx x 100%
Finally
^ . z A ^ l i + ^Xi-^l^-Z/JxT/Tf+ll + ^Kl-^-llxexp^^-l)]
dY C
(C-24)
Eq. (C-24) can be transformed into the following expression:
d2Tf = E (E2 Y2/H2+2E E Y/H + E23) (C-25)

r x 2 2 3
dY
^(-Pêxp^fo-l)]
E2=6h + Sf)d-e)+lYdH-H.) (C-26)
E3=(\ + Sf)([-£)-l
Plastic w o r k zone
In the plastic work zone, Eq. (C-20) can be expressed as follows:
uw+ur=ur[Q + Sf)y2/y + l] (C-27)

uw-ur=ur[(l + Sf)y2/y-l]
is strip thickness in roll bite. Please refer to Figure 6.1 regarding y, yx and y2
Eq. (C-21) for the plastic work zone can be written:

ix C Energy Equation
d2Tf nu
2
(
-6[(l+ Sf)y2/y + \].^-^.^.]---[(l + Sf)y2ly-l]j (C-28)
dy'< c
/ V h h h h
Hence
d2Tf
^.{-6[{\ + Sf){\~£)IY +\$H -Hx)xY IH -[(\ + Sf){\-£)IY -\}}2 .—
dy: Vô
(C-29)
and then
d
^ = -^\6{{\ + Sf){\-£)IY+\l{H-Hx)^YIH + [{\ + Sf){\-£)IY-\-\Y
dY C
xexp[-E(Tm-l)]
(C-30)
Finally
d2Tf
2 = Ex (E2 Y2/H2 + 2E2E3 Y/H + E2) (C-31)
HY
where
Ex=(-Px/C)exp[-E(T„-l)}
E2=6[(l + Sf)(l-£)/Y + l].(H-Hx) (C-32)
E3=(\ + Sf)(\-£)IY-\
Outlet zone
In the outlet zone, Eq. (C-20) can be changed into as follows:
uw+ur=ur[(l + Sf) + l]
(C-33)
uw -ur =urSf
So, Eq. (C-21) for the outlet zone can be written:
*dyT'- k *l'2fr-6g[+ S/>l]*-*'

2
h
" ' "^
h h h
(C-34)
f \
Hence
s +2 H H xY,H s ]2 (C-35)
^--T^^-^
V'o
f ^ - ^ - f -^
dy kfTn h
and then
d2Tf P}
[6(Sf+2).(H-H2)xY/H + Sf]2.exp[-E(Tm-$] (C-36)
2
dY c
Finally
d 2T
L = EX {E\ Y2/H2 + 2E2E3 Y/H + E2) (C-37)
dY
where
Ex=(-Px/C)exp[-E(Tm-\)}
E2=6(Sf+2).(H-H2) (C-3 8)
E3 -Sf
D Moving Heat Source Boundary Condition
Appendix D
Moving Heat Source Boundary
Condition
Inlet zone
For the case of conduction in the semi-infinite solid (v > 0) with a heat additio
area per unit time of q(t) on the surface, the temperature rise is governed by [C
and Jaeger, 1959],
^_±^=0 (D-D
dy2 K dt
The boundary condition for Eq. (D-l) is:
, dT
-k —
dy
then the temperatureriseat y = y at time t is given by [Wilson et al, 1989]
l 2i
Uv\
K) !'^-De-n^^dX (D-3)
T=- ft J Q
where X is defined as the time measured backwards from time t and q is heat flux.
Appendix D Moving Heat Source Boundary Condition 247
At the surface of the strip and rolls ( y = 0), the temperature is given by
rK\*<.
T = |g(r-/i)r1/2JA + r0 (D-4)
\ft
The backward time X can be calculated as:
X = x-x (D-5)
W h e n the time X is equal to zero, it is the starting point. W h e n X is at time t, the
position of x' is at x. So
fx' = oo, X =0 (D-6)

\x'=x, X = t
Substitute Eqs. (D-5) and (D-6) into Eq. (D-4)
(K) r tt n *' Mi +T
*
(D-7)
k
At the roll surface, Eq. (D-7) can be changed into:
1 (K Y/2xr ^ dx' (D-8)

,2^0
The lubricant forms a very thin film between the rolls and strip, the conduction Eq.
(D-l) can also be applied for the lubricant. Because the heat is transferred from
surface through the lubricant to the roll surface. So the boundary conditions at r
lubricant interface is:
dTf (D-9)
x=af
dy y=-»
At the roll/lubricant interface, the expressions of qr in Eq. (D-8) and qf in Eq. (D-9)
are the same. So, substitute Eq. (D-9) into Eq. (D-8)
vV2
If*.
J(*, ar,
<&'
T = (D-10)
k\n ay 4^Wf(x'-x), 2 + To
According to the dimensionless items in Eq. (C-22) of Appendix C, the Eq. (D-10) can
be transformed into the following expression:
dTt dX' (D-ll)

x x 1/2 + 1
dY 4 ( '- )
At the strip/lubricant interface, Eq. (D-4) can be written as:
2
1 (K- ^ \q (t-X)X-l/2dX + T (D-12)
Zw= w 0
KVft
The boundary conditions on the side of strip/lubricant interface is:
dT,
(D-13)
dy y=—«
At the strip/lubricant interface, the expressions of qw in Eq. (D-12) and qf in Eq.
(D-13) are the same. If it is assumed that the speed of strip surface in the inlet zo
constant, Eq. (D-5) can be changed into:
x-x (D-14)
X=
(\ + Sf)y2/yx ur
Substitute Eqs. (D-6), (D-13), and (D-14) into Eq. (D-12)
K \l2xr ST/ dY + r 0 (D-i5)

r„, X I.)"
y=-/Uy(x'-x)
1/2
[(l + Sf)y2/yx] kw ft I fy
Appendix D Moving Heat Source Boundary Condition _249
According to the dimensionless items in Eq. (C-22) of Appendix C, Eq. (D-15) can be
changed into the following expression:
dTf dX'
Tw
[(l + Sf)(l-£)r
H dY r=-\ (X'-X)l/2
+1 (D-16)
Plastic work zone
Similar derivation in the plastic work zone can be made. The temperature formula at
roll/lubricant interface is:
T dX'
^W H dY ?={- (X'-X)1/2
+ T„ (D-17)
In the rolling process, if part of the plastic deformation work is converted into heat, the
temperature at the strip/lubricant can be expressed as follows:
r
T.
W =T + °>
Ml
mr (D-18)
C
wPw*0
Outlet zone
In the outlet zone, the boundary conditions are similar as in the inlet zone. The o
difference is the expression of X. A s the speed of the strip surface is same in the outlet
zone, X can be written as:
1 pA- J\,
X =- (D-19)
(1 + S/) ur
So, the temperature at the roll/lubricant and strip/lubricant interface can be express
follows:
T,=P3I(\_yp f dX' (D-20)

/2 + Tr_
dY >4 (x-xy
dTf dX'
T., =•
(l + Sf1/2
) l-i) — +T
1 (T-X)1/2 2 w
(D-21)
Appendix E Temperature at Asverities Contact 251
Appendix E
Temperature at Asperities Contact
The temperature increment at the point (x,y) on either of two surfaces at any time t
due to an instantaneous heat source of strength OJ(X',/,f)dx'dy'df at x',y
be calculated by the following equation given by Carslaw and Jaeger [1959].
Trr, co(x\y\f\dx'dyjdf
dT
= r / \WY~X ex
P (E-l)
2 AK(t-f)
4pep[xK(t-ff
Integrating Eq. (E-l)
a>(x\y\t') (x-x'y+(y-yj
AT-J/J- .exp
AK(t-f)
dx'dy'df (E-2)
ApcM((-tT
Because the only temperature variation along the rolling direction is considered, and
temperature variation along the width of the strip is not changed, so Eq.
transformed into:
G){x',y\Q (x-x')-V (y-yf

Ar=J- 2
df J exp
AK(t - f)
dx' [exp
AK({-f)
dy' (E-3)
ApcM(t-t<)j
and
Appendix E Temperature at Asperities Contact 252
{exp (y-y'J dy' (E-4)

_ AK(t-t'\
-f-«J
<y-yj' dy' (E-5)

exp
AK({-t')
= djAK(t-t') jexp '

rf
y'-y ' (E-6)
4K(f-f)J V V^F^J
From [Spiegel, 1968]
\e x
dx = — ^7t (E-7)
1
2
thus
f . ^
(y'-yj y-y (E-8)
exp 4K(t-f) ^K(t-t'))
J v
Substitute Eq. (E-8) into Eq. (E-6)
(y-yj' dy = Â7iK{t-t') (E-9)

exp
AK(t-f)
Substitute Eq. (E-9) into Eq. (E-3)
co(x',f) (x-x'J' (E-10)

Ar =r^^^..Jexp dx'
0 "A~/7
AK(t -1')
Differentiating Eq. (E-10), one-dimensional equation can be obtained.
2
i\
1rr, oj(x\t\dx'df (x-x')
v (E-ll)
dT = / r — x exp AK(t-f)
27ik(t-f)
Appendix F Uncertainty Analysis nf Friction Coefficient Measurement 253
Appendix F
Uncertainty Analysis of Friction
Coefficient Measurement
The result R is a given function of the independent variables xx,x2,x3, ,x„. Thus,
J J \X\ i x2, x3, , Xn ) (F-l)
Let uf be the uncertainty in the result and ux,u2,u3, ,un be the uncertainties in the
independent variables, then the uncertainty in the result is given as [Holman, 19
Beckwith and Marangoni, 1990]
f -,/• \ f sr \ f >Sf \ ( -ss \

df df df (F-2)
u, •Mi + . dx un + ydX u. + + u.
dx
V^ A l x J
j 3 K^n I
Eq. (2-26) is rewritten here
P 'tan0 (F-3)
KPr J
Consider the following:
dp. _ 1
(F-4)
dpg prtand
Appendix F Uncertainty Analysis of Friction Coefficient Measurement 254
^P _ Pe
(F-5)
dpr p2tm0
Using Eq. (F-2), w e have
-,1/
Ap =
dp Y f dp "
^Pe + Ap, (F-6)
\dPe i
dPr ^
where
Apg, Apr -discrete uncertainties of pe and pr, and
A//=the overall uncertainty of function p resulting from the individual
uncertainties of Ape and Apr
thus
1 Y ( Pe
Ap = &Pe + 2 *Pr (F-7)
pr tan# p tan0
well
1 'e Ap6
^Ve
Ap (F-8)
pr tan 0 pr tan 0 p6
(F-9)
pr tan 9 prtan& pr
From Eq. (F-3)
Pe 1
•= // + - (F-10)
pr tan 6 tan 6
Substitute Eq. (F-10) into Eqs. (F-9) & (F-8) respectively, w e have
1 4P*
4P* = /" + (F-ll)
prtand tan<9
Appendix F Uncertainty Analysis of Friction Coefficient Measurement 255
1 > 4P,
Apr = p + (F-12)
p;tan0 tan#
Substitute Eqs. (F-l 1) & (F-12) into Eq. (F-7)
Ap p+ (APA2 ' ^
(F-13)
tan6>
+
KPO , V fr J
and, finally
r
Ap
1+
1 (APA2
+
^}2 (F-14)
P plan 6 sPe , {Pr J
From Eq. (F-14), it can be seen that high friction coefficient value and small
uncertainties of radial & oblique pin can reduce the uncertainty of friction
value, and vice versa. The hysteresis in the sensor roll calibration will aff
accuracy of friction coefficient measurement. The signals from radial strain
oblique loadcell were used as radial and oblique pins signals respectively. T
uncertainties of friction coefficient measurement over the roll bite in Figur
shown in Figure F. 1.
Aluminium alloy 6065-T5,8=37.01%, n=5rpm

0.5
o 45.45% points <30% 47.73% points >30%

•<*f*°
8 o
c
w W H ^ ^ •••••
••••• t
o /
o 6.82%<30%
0.5
Figure F.l Uncertainties analysis of friction coefficient measurement

References 256
References
[Aggarwal and Wilson, 1978] B. B. Aggarwal and W . R. D. Wilson, Thermal Effects in

Hydrodynamically Lubricated Strip Rolling, Proceedings of 5th Leed-Lvon
Symposium on Tribology. Institution of Mechanical Engineers (I. Mech. E.),
1978, London.
[Alexander, 1972] J. M. Alexander, On the theory of Rolling, Proc. R. Soc. Lond.

Vol. 326, pp. 535-563.
[Alexander etal, 1987] J. M. Alexander, R. C. Brewer and G. W. Rowe,

Manufacturing Technology-Vol.2: Engineering Process, Ellis Horwood Limited,
Chicchester.
[Al-Salehi et al, 1973] F. A. R. Al-Salehi, T. C. Firbank and P. R. Lancaster, An

Experimental Determination of the Roll Pressure Distribution in Cold Rolling
International Journal of Mechanical Sciences, Vol. 15, pp. 693-710.
[Atkins, 1974] A. G. Atkins, Hydrodynamic Lubrication in Cold Rolling, Internati

Journal of Mechanical Sciences, Vol. 16, pp. 1-10.
[Avitzur and Grossman, 1972] B. Avitzur and G. Grossman, Hydrodynamic

Lubrication in Rolling of Thin Strip, A S M E Journal of Lubrication Technology,
Series B, Vol. 94 (NI), pp. 317.
[Avitzur, 1989]B. Avitzur, Boundary and Hydrodynamic Lubrication, Wear, Vol. 139,
pp. 49-76.
References__ 257
[Banerji and Rice, 1972] A. Banerji and W . B. Rice, Experimental Determination of
Normal Pressure and Friction Stress in the Rolling Gap During Cold Rolling,
Annals of the CIRP. Vol. 21, pp. 53-54.
[Barus, 1893] C Barus, Isothermals, Isopiestics and Isometrics Relative to Visco

American Journal of Science. Vol. 45, pp. 87-96.
[Beckwith and Marangoni, 1990] T. G. Beckwith and R. D. Marangoni, Mechanical

Measurement,, 4th edition, Addison-Wesley Publishing Company.
[Bedi and Hiller, 1967-1968] D. S. Bedi and M.J. Hillier, Hydrodynamic Model for
Cold Strip Rolling, Proceedings of Institution of Mechanical Engineers, Vol.
182, pp. 153.
[Bhatt and Sengupta, 1996] V. K. Bhatt and D. K. Sengupta, Analysis of Thermal

Effects in the Yield Phase of Hydrodynamic Lubricant Film in Plane Strain
Forging, A S M E : Journal of Tribology, Vol. 118, pp. 880-885.
[Bland and Ford, 1948] D. R. Bland and H. Ford, The Calculation of Roll Force an
Torque in Cold Strip Rolling with Tensions, Proceedings of Institute of
Mechanical Engineers, Vol. 159, pp. 144-153.
[Britten & Jeswiet, 1986] D. Britten & J. Jeswiet, A Sensor for Measuring Normal
Forces with Through and Transverse Friction Forces in the Roll Gap,
S M E / N A M R I Transactions, pp. 355.
[Carslawand Jaeger, 1959] H. S. Carslaw and J. C Jaeger, Conduction of Heat in

Solids, Clarendon Press, Oxford.
[Cerni etal, 1963] S. Cerni, A. S. Weinstein and C F. Zorowski, Temperatures and

Thermal Stresses in the Rolling of Metal Strip, Iron and Steel Engineer. Vol. 40
pp. 165-173.
References 258
[Chang etal, 1996] D. F. Chang, N. Marsault and W . R. D. Wilson, Lubrication of
Strip Rolling in the Low-Speed Mixed Regime, Tribology Transactions, Vol. 39,
No.2, pp. 407-415.
[Chang, 1998] D. F. Chang, An Efficient Way of Calculating Temperatures in the

Rolling Process, Journal of Manufacturing Science and Engineering, Vol. 120,
pp. 93-100.
[Cheng, 1966] H. S. Cheng, Plastohydrodynamic lubrication in Friction and

Lubrication in Metal Processing, A S M E Transactions, N e w York, pp. 69-89.
[Christensen, 1969-1970] H. Christensen, Stochastic Models for Hydrodynamic

Lubrication of Rough Surfaces, Proceedings of Institution Mechanical
Engineers, No. 84, pp. 1013-1022.
[Christensen and Tonder, 1971] H. Christensen, K. Tonder, The Hydrodynamic

Lubrication of Rough Journal Bearing Surfaces of Finite Width, A S M E Journal
of Lubrication Technology, Vol. 93, pp. 324-330.
[Christensen and Tonder, 1972] H. Christensen, K. Tonder, Waviness and Roughnes

Hydrodynamic Lubrication, Proceedings of Institution Mechanical Engineers
Tribology Group, Vol. 186, pp. 186.
[Christensen and Tonder, 1973] H. Christensen, K. Tonder, The Hydrodynamic

Lubrication of Rough Journal Bearing, A S M E Journal of Lubrication
Technology, Vol. 95, pp. 166-172.
[Chung and Wilson, 1994] Y. S. Chung and W. R. D. Wilson, Full Film Lubrication
Strip Rolling, A S M E : Journal of Tribology, Vol. 116, pp. 569-576.
[Cook and McCrum, 1958] P. M. Cook and A. W. McCrum, The Calculation of Load
and Torque in Hot Flat Rolling. BISRA.
References 259
[Denton and Crane, 1972] B. K. Denton and F. A. Crane, Roll Load for Hot rolling of
Steel Sheet and Strip, Journal of the Iron and Steel Institute, pp. 606-617.
[Dixon and Yuen, 1995] A. E. Dixon and W. Y. D. Yuen, A Computationally Fast

Method to Model Thin Strip Rolling, Proceedings of Computational Techniques
and Application Conference. Melbourne, 3-5 July, 1995.
\Dowetal, 1975] T. A.Dow, J. W. Kannel and S. S. Bupara, A Hydrodynamic

Lubrication Model for Cold Rolling Including Thermal Effects, A S M E Journal
of Lubrication Technology. Vol. 97, pp. 4-13.
[Ekelund, 1933] S. Ekelund, Analysis of Factors Influencing Rolling Pressure a

Power Consuming in the Hot Rolling of Steel, Steel, Vol. 93, pp. 27-29.
[Evans and Avitzur, 1968] W. Evans and B. Avitzur, Measurement of Friction in

Drawing, Extrusion and Rolling, A S M E Journal of Lubrication Technology,
Vol. 90, pp. 72-88.
[Fleck et al, 1987] N. A. Fleck and K. L. Johnson, Towards a New Theory of Col
Rolling Thin Foil, Institution Journal of Mechanical Sciences, Vol. 29, pp. 507-
524.
[Fleck et al, 1992] N. A. Fleck and K. L. Johnson, M. E. Mear and L. C Zhang,

Rolling of Foil, Proceedings of Institution of Mechanical Engineers, Vol. 206,
pp. 119-131.
[Ford et al, 1951] H. Ford, F. Ellis and D. R. Bland, Cold Rolling with Strip
Part I - A New Approximate Method of Calculation and a Comparison with
Other Methods, Journal of the Iron and Steel Institute. Vol. 168, pp. 57-72.
[Ford and Alexander, 1963] H. Ford and J. M. Alexander, Simplified Hot Rolling
Calculations, Journal of the Institute of Metals. Vol. 92, pp. 397-404.
References___ 260
[Ford and Bland, 1951] H.Ford and D. R. Bland, Cold Rolling with Strip Tension,
Journal of the Iron and Steel Institute. May, Part I, pp. 57-72.
[Ginzburg, 1985] V. B. Ginzburg, Basic Principles of Customized Computer Model

Cold and Hot Strip Mills, Iron and Steel Engineers, pp. 21 -35.
[Ginzburg, 1989] V. B. Ginzburg, Steel - Rolling Technology: Theory and Practic

Marcel Dekker, Inc..
[Green and Wallace, 1962] J. W. Green and J. F. Wallace, Estimation of Load and
Torque in the Hot Rolling Process, Journal of Mechanical Engineering Science,
Vol. 4, pp. 136-142.
[Hitchcock, 1935] J. Hitchcock, Roll Neck Bearings, ASME Research Publication,

Appendix.
th
[Holman, 1989] J. P. Holman, Experimental Methods for Engineers, 5 edition,

McGraw-Hill Book Company.
[Houpert, 1985] L. Houpert, New Results of Traction Force Calculations in

Elastohydrodynamic Contacts, Journal of Tribology, Vol. 107, pp. 241-248.
[Hum et al, 1996] B. Hum, H. W. Colquhoun and J. G. Lenard, Measurements of

Friction During Hot Rolling of Aluminum Strips, Journal of Materials
Processing Technology, Vol. 60, pp. 331-338.
[Hus and Evans, 1990] C T. Hus and R. W. Evans, Finite Element Analysis on the
Rolling of Steel, Advanced Technology of Plasticity. Vol. 2, pp. 587-593.
[Incropera and Dewitt, 1990] F. P. Incroperaand D. P. Dewitt, Fundamental of He

and Mass Transfer, 3rd edition, John Wiley & Sons, Inc..
[Jeswiet, 1995] J. Jeswiet, Aspect Ratio, Friction Forces and Normal Forces in Str
Referencps: 261
Rolling, Journal of Materials Processing Technology, Vol. 53, pp. 846-856.
[Jeswiet et al, 2000] J. Jeswiet, P. Wild and H. Sefton, Sensing Friction: Meth
Devices, International Workshop on Friction and Flowstress in Cutting and
Fpnnjng, Paris, 2000, pp. 111-118.
[Jeswiet and Rice, 1982] J. Jeswiet and W. B. Rice, The Design of a Sensor
for Measuring Normal Pressure and Friction Stress in the Roll Gap During
Cold Rolling, S M E / N A M R I Transactions, pp. 313.
[Jeswiet and Rice, 1989] J. Jeswiet and W. B. Rice, The Natural Zone and Temper
at the Roll/Strip Interface, Annals of the CIRP, Vol. 39, pp. 275-277.
[Jeswiet and Zhou, 1992] J. Jeswiet and S. Zhou, Temperature, Heat Flux and
Conductivity in Bar Rolling, Annals of the CIRP, Vol. 41, pp. 299-302.
[Jiang et al, 2000] Z. Y. Jiang, A. K. Tieu and C Lu, Analysis of the Elastic
Deformation Regions for Cold Strip Rolling by Finite Element Method,
Proceedings of 2 n d International Conference on Mechanics of Structures,
Materials and Systems ( M S M S 2001), Wollongong, Australia, February 14-16,
2001, pp. 351-356.
[Johnson and Kudo, 1960] W. Johnson and H. Kudo, The Use of Upper - Bound
Solutions for the Determination of Temperature Distributions in Fast Hot
Rolling and Axisymmetric Extrusion Processes, International Journal of
Mechanical Sciences, Vol. l,pp. 175-191.
[Karagiozis and Lenard, 1985] A. N. Karagiozis and J. G. Lenard, The Effect of

Material Properties on the Coefficient of Friction in Cold Rolling, Proceedi
ofEurotrib^, Lyon, pp. 1-7.
[Karman, 1925] T. Von. Karman, Zeitschrift fur angewandte Mathematik und

Mechanik, Vol. 5, pp. 139.
References 262
[Kasuga and Yamaguchi, 1968] Y. Kasuga and K. Yamaguchi, Friction and

Lubrication in the Deformation Processing of Metal, Bulletin of J S M E , Vol.
11-44, pp. 344-365.
[Korzekwa et al, 1992] D. A. Korzekwa, P. R. Dawson and W. R. D. Wilson, Surfa

Asperity Deformation During Sheet Forming, Internationl Journal of Mechanical
Sciences. Vol. 34, pp. 521-539.
[Lagergren, 1997] J. Lagergren, Measurements of Multiple Normal Pressure Peaks

Flat Rolling, Steel Research. Vol. 68, No. 7, pp. 313-325.
[Lenard, 1981] J. G. Lenard, A Theory of Flat Rolling-the Effect of a Variable

Coefficient of Friction, Proceedings of C O B E M 1981. Rio de Janeiro, Brazil,
pp. 13-24.
[Lenard, 1991] J. G. Lenard, Measurements of Friction in Cold Flat Rolling, Jo

Materials Shaping Technology. Vol. 9, No. 3, pp. 173-180.
[Lenard, 1992] J. G. Lenard, Friction and Forward Slip in Cold Strip Rolling,
Tribology Transactions, Vol. 35, pp. 423-428.
[Lenard and Malinowski, 1993] J. G. Lenard and Z. Malinowski, Measurements of

Friction During Warm Rolling of Aluminum, Journal of Materials Processing
Technology. Vol. 39, pp. 357-371.
[Lenard and Pietrzyk, 1990] J. G. Lenard and M. Pietrzyk, Thermal-Mechanical

Modelling of a Hot Strip Mill, Advanced Technology of Plasticity, Vol. 2, pp.
601-606.
[Li and Kobayashi, 1982] G. J. Li and S. Kobayashi, Rigid Plastic Finite Eleme
Analysis of Plane Strain Rolling, A S M E Journal of Engineering for Industry,
Vol. 104, pp. 55-64.
References 263
[Lim and Lenard, 1984] L. S. Lim, J. G. Lenard, Study of Friction in Cold Strip Rolli
Journal of Enpineering Materials and Technology, Vol. 106, pp. 139-146.
[Lin and Houng, 1991] J. F. Lin and S. C Houng,^ Refined Model for the Thermal
Hydrodynamic Lubrication Theory of High-Speed Rolling Part I: Effects o
Operating Conditions, Tribology International Vol. 24, pp. 151-161.
[Lin etal, 1991] J. F. Lin, T.K.Huang and C T. Hsu, Evaluation of Lubricants i

Cold Strip Rolling, Wear. Vol. 147, pp. 79-91.
[Liu et al, 1999] Y. J. Liu, A. K. Tieu, E, B, Li and W. Y. D. Yuen, Forward S

Friction in Cold Rolling, Proceedings of International Symposium on Advanced
Forming and Die Manufacturing Technology (AFDM'99), Pusan, Korea,
September 7-9, 1999, pp. 495-500.
[Liu and Tieu era/., 2001] Y. J. Liu, A. K. Tieu, D. D. Wang and W. Y. D. Yuen
Friction measurement in cold rolling, Journal of Materials Processing
Technology, Vol. Ill, pp. 142-145.
[Liu and Tieu, 2001] Y. J. Liu, A. K. Tieu, The Influence of Inlet and Outlet
Mixed-Film Lubricated Cold Rolling Process, Proceedings of 2 nd International
Conference on Mechanics of Structures, Materials and Svstems ( M S M S 2001),
Wollongong, Australia, February 14-16, 2001, pp. 321-326.
[Lo, 1994] S. W. Lo, A Study on Flow Phenomena in Mixed Lubrication Regime By

Porous Medium Model, A S M E Journal of Tribology, Vol. 116, pp. 640-647.
[Lugt et al, 1993] P. M. Lugt, A. W. Wemekamp and W. E. ten Napel, Lubrication

Cold Rolling: Elasto-Plasto-Hydrodynamic Lubrication of Smooth Surface
Wear, Vol. 106, pp. 203-214.
[Lugt and Napel, 1995] P.M. Lugt and W. E. ten Napel, The Influence of Elastic
References 264
Deformation of the Roll and the Sheet in a Hydrodynamically Lubricated Cold

Rolling Process, Transactions of the A S M E . Vol. 117, pp. 468-475.
[Mizuno, 1966] T. Mizuno, A Experimental Research on Cold Rolling, Journal of

(in Japanese), Vol. 7-66, pp. 383-389.
[Nadai, 1939] A. Nadai, The Forces Required for Rolling Steel Strip Under Tens
A S M E : Journal of Applied Mechanics. Vol. 6, pp. 54-62.
[Nyahumwa and Jeswiet, 1991] C Nyahumwa and J. Jeswiet, ,4 Friction Sensor for
Sheet-Metal Rolling, Annals of the CIRP. Vol. 40, pp.231-234.
[Orowan, 1943] E. Orowan, The Calculation of Roll Pressure in Hot and Cold Fla
Rolling, Proceedings of Institute of Mechanical Engineers. Vol. 150, No. 4, pp.
140-167.
[Orowan and Pascoe, 1946] E. Orowan and K J. Pascoe,v4 Simple Method of

Calculating Roll Pressure and Power Consumption in Flat Hot Rolling, Iron an
Steel Institute (London). No. 34, pp. 124-126.
[Parke and Baker, 1972] D . M. Parke and J. L. L. Baker, Temperature Effects o

Cooling Work Rolls, Iron and Steel Engineers, Vol. 49, pp. 83-88.
[Patir and Cheng, 1978] N. Patir and H. S. Cheng, An Average Flow Model for
Determining Effects of Three Dimensional Roughness on Partial Hydrodynamic
Lubrication, A S M E : Journal of Lubrication Technology, Vol. 100, pp. 12-17.
[Patir and Cheng, 1979] N. Patir and H. S. Cheng, An Application of Average Fl

Model to Lubrication Between Rough Sliding Surfaces, A S M E : Journal of
Lubrication Technology. Vol. 101, pp. 220-230.
[Patula, 1981] E. H. Patula, Steady-State Temperature Distribution in a Rotati

References 265
Subject to Surface Heat Fluxes and Convective Cooling, A S M E : Journal of Hea

Transfer. Vol. 103, pp. 36-41.
[Pauskarefa/., 1997] P. Pauskar et al, Microstructure Dependent Flow Stress M

for Steel Rolling, Mechanical Working and Steel Processing Conference
Proceedings 1997, Iron and Steel Soc. of A I M E , Warrendale, PA, U S A , pp. 973-
982.
[Peck etal, 1954] C F. Jr. Peck, J. M. Bonetti, F. T. Mavis, Temperature Stre

Iron Work Rolls, Iron and Steel Engineer. Vol. 31, pp. 45-57.
[Plancak et al, 1996] M. Plancak, A. N. Bramley and F. H. Osman, Some Observa

on Contact Stress Measurement by Pin Load Cell in Bulk Metal Forming,
Journal of Materials Processing Technology. Vol. 60, pp. 339-342.
[Poplawski and Seccombe, 1980] J. V. Poplawski and D. A. Seccombe, Mathematic

Modelling System for Cold Tandem Rolling Mills: Roll and Strip Thermal
Considerations, Science and Technology of Flat Rolled Products, Vol. 1, pp.
611-622.
[Qiu et al, 1999] Z. L. Qiu, W. Y. D. Yuen and A. K. Tieu, Mixed-Film Lubrica

Theory and Tension Effects on Metal Rolling Processes, A S M E Journal of
Tribilogy, Vol. 121, pp. 908-915.
[Qiu and Cheng, 1998] L. H. Qiu and H. S. Cheng, Temperatures Rise Simulation
Three-Dimensional Rough Surfaces in Mixed Lubricated Contact, A S M E :
Journal of Tribology, Vol. 120, pp. 310-318.
[Rabinowicz, 1965] J. E. Rabinowicz, Friction and Wear of Materials, Wiley, N

York.
[Ride, 1960] J. S. Ride, Analysis of Operational Factors Derived from Hot Str
Tests, AISE Yearly Proceedings, pp. 867-880.
References 266
[Roberts, 1965] W . L. Roberts,^ Simplified Cold Rolling Model Iron and Steel
Engineers, pp. 75-86.
[Roberts, 1978] W. L. Roberts, Cold Rolling of Steel. Marcel Dekker, Inc., New
and Basel.
[Roelands, 1966] C J. A. Roelands, Correlation Aspects of Viscosity - Temperat

Pressure Relationship of Lubricating Oils," Ph. D. thesis, Delft University of
Technology, The Netherlands.
[Rooyen and Backofen, 1957] G. T. V. Rooyen and W. A. Backofen, Friction in Co

Rolling, Journal of the Iron and Steel Institute. Vol. 186, pp. 235-244.
[Sa and Wilson, 1994] C Y. Sa and W. R. D. Wilson, Full Film Lubrication of St

Rolling, A S M E : Journal of Tribology. Vol. 116, pp. 569-576.
[Sargent and Tsao, 1980] L. B. Sargent and Y. H. Tsao, Surface Roughness

Considerations in Metalworking, A S L E Transactions, Vol. 23, pp. 70-76.
[Saxena et al, 1996] S. Saxena,P. M. Dixit and G. K Lai, Analysis of Cold Stri
Rolling Under Hydrodynamic Lubrication, Journal of Materials Processing
Technology, Vol. 58, pp. 256-266.
[Schultz and Smith, 1965] R. G. Schultz and A. W. Smith, Determination of a

Mathematical Model for Rolling Mill Control, AISE Yearly Proceedings, pp.
461-467.
[Shaw etal, 1960] M. C Shaw, A. Ber and P. Mamin, Friction Characteristics of

Sliding Surfaces Undergoing Sub-Plastic Flow, A S M E Journal of Basic
Engineering. Vol. 82, pp. 342-346.
[Sheu, 1985] S. Sheu, Mixed Lubrication in Bulk Metal Forming Processes, Ph.D
thesis. Northwestern University Evanston, IL.
References _^^^ 267
[Sheu and Wilson, 1983] S. Sheu and W . R. D. Wilson, Flattening of Workpiece

Surface Asperities in Metal Forming, Proceedings of N A M R C XI. (SME), pp.
172-178.
[Sheu and Wilson, 1994] S. Sheu and W. R. D. Wilson, Mixed Lubrication of Stri
Rolling, Tribology Transactions. Vol. 37, pp. 483-493.
[Siebel, 1941] E. Siebel, Calculation of Roll Force Anti -Friction Bearing for
MillsiGQrman), Schweinfurt. 1941.
[Siebel and Lueg, 1933] E. Siebel and W. Lueg, Investigations into the Distrib
Pressure at the Surface of the Materials in Contact with the Rolls, Mitt. K.
Inst. Eisenf, Vol. 15, pp. 1-14.
[Sims, 1952] R. B. Sims, The Forward Slip in Cold Strip Rolling, Sheet Metal I
Vol. 29, pp. 869-877.
[Sims, 1954] R. B. Sims, The Calculation of Roll Force and Torque in Hot Rolli
Proceedings of Institute of Mechanical Engineers, No. 168, pp. 191- 200.
[Smith etal, 1952] C. L. Smith, F. H. Scott and W. Sylwestrowicz, Pressure

Distribution Between Stock and Rolls in Hot and Cold Flat Rolling, Journal
Iron and Steel Institute, Vol. 170, pp. 347-359.
[Spiegel, 1968] M. R. Spiegel, Mathematical Handbook of Formulas and Tables,

Schaum's Outline Series, McGraw-Hill Book Company, N e w York.
[Stephenson, 1983] D. A. Stephenson, Friction in Cold Strip Rolling, Wear, Vol

293-311.
[Stone, 1953] M. D. Stone, Rolling Thin Strip, AISE Yearly Proceedings, pp. 11
128.
References 268
[Storer et al, 1997] R. A. Storer et al, Metal - Mechanical Testing; Elevated and Low-
Temperature Tests; Metallography, Annual Book of A S T M Standard, Vol.
03.01, pp. 550-556, 1997, West Conshohocken.
[Sutcliff, 1988] M. P. F. $>u\c\ffi, Surface Asperity Deformation in Metal Form

Processes, Institution Journal of Mechanical Sciences. Vol. 30, pp. 847-868.
[Sutcliff, 1989] M. P. F. Sutcliff, Friction and Lubrication in Metal Rolling,

thesis, Cambridge University. England.
[Sutcliffe and Johnson, 1990] M. P. F. Sutcliffe and K. L. Johnson, Lubricatio

Rolling in the Mixed Regime, Proceedings of Institution of Mechanical
Engineers, Part B, 204 (B4), pp. 249-261.
[Swift, 1940] H. W. Swift, Stresses and Strains in Tube Drawing, Phil. Mag., 40
[Tieu et al, 1998] A. K. Tieu, E. B. Li and W. Y. D. Yuen, Forward Slip Measure

in Cold Rolling Using Laser Doppler Velocimetry, Proceedings of 7
International Conference on Steel Rolling (STEEL ROLLING'981 1998, Chiba,
Japan, The Iron and Steel Institute of Japan, pp. 145-149.
[Tripp, 1983] J. H. Tripp, Surface Roughness Effects in Hydrodynamic Lubricati

Flow Factor Method, A S M E Journal of Lubrication Technology, Vol. 105, pp.
458-463.
[Tsao and Sargent, 1975] Y. H. Tsao and L. B. Sargent, A Mixed Lubrication Mode
Cold Rolling of Metals, A S L E Transactions, Vol. 20, pp. 55-63.
[Tselikov, 1967] A. I. Tselikov, Stress and Strain in Metal Rolling, Mir Publis
Moscow.
[Tseng etal, 1990] A. A.Tseng, S. X. Tong, S. H. Maslen, J. J. Mills, Thermal

References 269
Behavior of Aluminum Rolling, A S M E : Journal of Heat Transfer, Vol. 112, pp.

301-308.
[Tzeng and Saibel, 1967] S. T. Tzeng and E. Saibel, Surface Roughness Effect
Slider Bearing Lubrication, A S L E Transaction. Vol. 10, pp. 334.
[Wang etal, 1995] Z. Wang, K. Kondo and T. Mori, Consideration of Optimum

Conditions for Surface Smoothing Based on Lubricating Mechanisms in
Ironing Process, A S M E Journal of Engineering for Industry. Vol. 117, pp. 351-
356.
[Wilson, 1978] W. R. D. Wilson, Friction and Lubrication in Sheet Metal Form

Mechanics of Sheet Metal Forming. Plenum Press, N e w York, pp. 157-177.
[Wilson, 1979] W. R. D. Wilson, Friction and Lubrication in Bulk Metal Formi

Processes, Journal of Applied Metalforming. Vol. 1, pp. 1-19.
[Wilson et al, 1989] W. R. D. Wilson, C T. Chang and C Y. Sa, Interface

Temperatures in Cold Rolling, Journal of Materials Shaping Technology, Vol. 6
No. 4, pp. 229-240.
[Wilson, 1990] W. R. D. Wilson, Mixed Lubrication in Metal Forming, Advanced

Technology of Plasticity (Japanese Society of Technical Plasticity), Vol. 4, pp.
1667-1676.
[Wilson and Chang, 1994] W. R. D. Wilson and D. F. Chang, low Speed Mixed
Lubrication of Bulk Metal Forming Processes, Tribology in Manufacturing
Processes. K. Dohda, S. Jahanmir and W . R. D. Wilson, A S M E , N e w York, pp.
159-168.
[Wilson and Chang, 1996] W. R. D. Wilson and D. F. Chang, low Speed Mixed
Lubrication of Bulk Metal Forming Processes, A S M E Journal of Lubrication
References 270
[Wilson and Mahdavian, 1974] W . R. D. Wilson and S. M . Mahdavian,^ Thermal

Reynolds Equation and Its Application in the Analysis of Plasto-Hydrodyn
Inlet Zone, A S M E Journal of Lubriration Technology. Vol. 95, pp. 572-578.
[Wilson and Marsault, 1998] W. R. D. Wilson and N. Marsault, Partial Hydrody

Lubrication with Large Fractional Contact Areas, A S M E Journal of Tribolog
Vol. 120, pp. 16-20.
[Wilson and Murch, 1976] W. R. D. Wilson and L. E. Murch, A Refined Model fo

Hydrodynamic Lubrication of Strip Rolling, A S M E : Journal of Lubricatio
[Wilson and Schmid, 1992] W. R. D. Wilson and S. R. Schmid, Surface in Metal

Rolling, A S M E : Engineered Surfaces. PED-Vol. 62, pp. 91-100.
[Wilson and Sheu, 1988] S. Sheu and W. R. D. Wilson, Real Area of Contact an
Boundary Friction in Metal Forming, International Journal of Mechanical
Sciences, Vol. 30, pp. 475-489.
[Wilson and Walowit, 1971] W. R. D. Wilson and J. A. Walowit, An Isothermal

Hydrodynamic Lubrication Theory for Strip Rolling with Front and Back
Tension, Tribology Convention, Institution of Mechanical Engineers, London,
1971, pp. 169-172.
[Wright and Hope, 1975] H. Wright and T. Hope, Rolling of Stainless Steel in
Strip Mills, Metals Technology, December 1975, pp. 565-576.
[Wusatowski, 1969] Z. Wusatowski, Fundamentals of Rolling, Pergamon Press, O
[Yamadaef al, 1991] K Yamada, S. Ogawa and S. Hamauzu, Two-Dimensional

Thermo-Mechanical Analysis of Flat Rolling Using Rigid-Plastic Finite El
Method, Transactions of Iron and Steel Institution, Japan, Vol. 31, pp. 566-570.
271
[Yokoi etal, 1981] T. Yokoi et al, Model for Calculation of Pass Schedules for
Slabbing Mills, Tetsu-to Hagane. Vol. 67, No. 15, pp. 2356-2364.
[Yoneyama and Hatamura, 1987] T. Yoneyama and Y. Hatamura, The Development of

a Die Sensor, J S M E International. Vol. 30, pp. 670-677.
[Yoneyama and Hatamura, 1989] T. Yoneyama and Y. Hatamura, Measurement of

Actual Stress and Temperature on the Roll Surface During Rolling, J S M E
International. Vol. 32, pp. 113-117.
[Yuen, 1987-1] W. Y. D. Yuen, On the Correction of Strip Shape by Localized C

of the Work Rolls, Proceedings of 4th International Steel Rolling Conference,
Vol.2,pp.E.23.1-E.23.12.
[Yuen, 1987-2] W. Y. D. Yuen, A New Formulation of Heat Transfer between Two

Moving Bodies in Contact over a Finite Region with Different Bulk
Temperatures, Mathematical Engineering Industry, Vol. 1, pp. 1-19.
[Yuen, 1987-3] W. Y. D. Yuen, Temperatures Fields of Moving Bodies with Diffe

Bulk Temperatures in Contact over a Fixed Region, Mathematical Engineering
Industry, Vol. 1, pp. 215-234.
[Yuen, 1988] W. Y. D. Yuen, Heat Conduction in Sliding Solids, International

of Heat Mass Transfer, Vol. 31, pp. 637-646.
[Yuen, 1993] W. Y. D. Yuen, Temperatures Fields in Sliding Solids with Intern

Sources, International Journal of Heat Transfer, Vol. 36, pp. 3711-3722.
[Yuen, 1994] W. Y.D.Yuen, The Thermal Boundary Layer in a Rotating Cylinder

Subject to Prescribed Surface Heat Fluxes, International Journal of Heat Ma
Transfer. Vol. 37, No.4, pp. 605-618.

Friction at Strip-Roll Interface in Cold Rolling

Uploaded by

Copyright:

Available Formats

Friction at Strip-Roll Interface in Cold Rolling

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Friction at Strip-Roll Interface in Cold Rolling

Uploaded by

Copyright:

Available Formats

University of Wollongong

Friction at strip-roll interface in cold rolling

Research Online is the open access institutional repository for the

the award of Doctor of Philosophy, in the Mechanical Engineering, Faculty of

Engineering at the University of Wollongong, Australia, is wholly m y o w n work unless

qualification at any other academic institution.

Acknowledgement is also given to technical staff in the Faculty of Engineering,

Acknowledgment is also given to my colleagues, Mr. D. D. Wang, Dr. C. Lu, Dr. E. B.

Thank you all indeed!

industry is becoming increasingly important. Inaccurate understanding of friction in the

the final strip thickness.

A large number of experiments were performed to determine the friction coefficient

entry was also considered.

Chapter 2 ROLLING THEORY AND LUBRICATION IN COL

Chapter 3 LITERATURE SURVEY ON FRICTION

Chapter 4 FRICTION COEFFICIENT MEASUREMENT AND

Chapter 5 VALIDATION OF FRICTION COEFFICIENT

Chapter 6 MIXED FILM LUBRICATION IN METAL ROLLI

Chapter 7 INFLUENCE OF HYDRODYNAMIC INLET AND

Chapter 8 THERMAL EFFECT IN MIXED FILM LUBRIC

Chapter 9 CONCLUSIONS AND RECOMMENDATIONS 204

Appendix A DATA ACQUISITION PROGRAM 209

Appendix B EMPIRICAL FORMULA OF FRICTION COEFFICIENT 22

Appendix C ENERGY EQUATION 239

Appendix D MOVING HEAT SOURCE BOUNDARY CONDITION 246

Appendix E TEMPERATURE AT ASPERITIES CONTACT 251

Appendix F UNCERTAINTY ANALYSIS OF FRICTION COEFFICIENT

Figure 2.1 Three-dimensional view of strip rolling process 8

Figure 2.2 Distribution of normal pressure and rolling force 10

Figure 2.3 Parameters of deformation zone 11

Figure 2.4 Frictional shear stress... 19

Figure 2.5 The lubrication regimes.... 21

Figure 2.6 Radial and oblique pins in roll bite 22

Figure 2.7 Marks on the strip 23

Figure 2.8 Laser Doppler method 24

Figure 2.9 Asperity contact in the mixed lubrication regime 29

Figure 2.10 Longitudinal, isotropic and transverse surface roughness 29

Figure 4.1 Sensor roll overview 62

Figure 4.2 Sensor roll section view 62

Figure 4.3 Side view of sensor roll and pins .....63

Figure 4.4 Front view of sensor roll and pins .63

Figure 4.5 Calibration process illustration 64

Figure 4.6 Radial strain gauge calibration 65

Figure 4.7 Oblique strain gauge calibration 65

Figure 4.8 Radial loadcell calibration 66

Figure 4.9 Oblique loadcell calibration 66

Figure 4.11 Data acquisition system 68

Figure 4.12 Data acquisition screen 70

Figure 4.13 Typical rolling force record 72

Figure 4.14 Typical rolling torque record 72

Figure 4.15 Strip thickness within segment part 73

Figure 4.16 Rolling force over roll bite 73

Figure 4.17 Rolling torque over roll bite 74

Figure 4.18 Rolling force recorded by ordinary roll 74

Figure 4.19 Thermocouple in sensor roll 75

Figure 4.20 Temperature record 75

Figure 4.21 Pin signal test for carbon steel 76

Figure 4.22 Pin signal test for aluminium alloy 77

Figure 4.23 Force distribution 78