Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Journal of Materials Processing Technology 80 – 81 (1998) 239 – 244

A comparison of friction coefficients in cold rolling


J. Jeswiet1
Queen’s Uni6ersity, Kingston, Ont., Canada

Abstract

A sensor which uses strain gauges mounted on a conical cantilever sensing element, has been designed to measure the three
forces of interest at the roll/strip interface, namely the normal force, the friction force in the rolling direction and the friction force
in the direction transverse to rolling. Successful experiments have been conducted with the sensor using 1100 aluminium, with a
low aspect ratio, as the rolled product. An analysis of the experimental results reveals a ‘rotating friction vector’ which is not
parallel to either axis of symmetry. This vector can be found when the transverse friction force, which is orthogonal to the rolling
direction, is added vectorially to the friction in the rolling direction. When the rotating friction vector is considered in terms of
‘Coulomb’ friction, m, a constant, direction dependent m occurs. The m arrived at in this manner is compared to an ‘old’ model
by Ford and is found to be in agreement for this specific rolling situation. Experimental results are presented for a 24.2%
reduction. © 1998 Elsevier Science S.A. All rights reserved.

Keywords: Rolling; Friction; Strain gauges

1. Introduction 2. Rolling forces

Measuring forces directly at the metal interface in The forces which occur at the roll/metal interface are
deformation processes is a difficult problem due to the illustrated in Fig. 2. These are defined as: Fn, the
harsh environment; high pressures, high temperatures, normal force which does the actual deformation of the
contaminants and high abrasion. Because of the com- strip and is directly related to the roll separating force
plexity of the friction phenomenon and the difficulty in (RSF); Ff, the friction force in the rolling direction and
measuring it, the usual approach is to assume a model, a direct result of the input torque, Tiu,; Ft, the friction
many of which are based upon constant ‘Coulomb’ force in the transverse direction across the roll face,
friction, m, across the billet surface area which is in which provides a resistance to strip spread.
compression.
Experiments have been conducted with a specially
designed sensor, which measures three orthogonal 2.1. Rolling force sensor
forces at the billet surface; see Fig. 1 for examples of
results. Bar and sheet rolling is an example of such a Developing sensors which can measure interface
process. When combined, according to the Coulomb forces has been the goal of a series of researchers [2–6].
friction model, a varying coefficient of friction is found. Direct measurement of the normal force, Fn, has been
Also, a friction force which is directional dependent is successful, but designing a sensor which will measure
also derived and compared with an ‘older’ model by friction forces is a more difficult problem. Past efforts
Ford [1]. concentrated on the measurement of the friction force
in the rolling direction, Ff, whereas there is also a
component in the orthogonal direction which must be
taken into account. A schematic of the sensor used by
Britten and Jeswiet [4], which has been used to measure
the three friction forces, is shown in Fig. 2.
A rolling mill presents an excellent opportunity to
1
E-mail: Jeswiet@me.queensu.ca verify if a friction force sensor is working properly. The

0924-0136/98/$19.00 © 1998 Elsevier Science S.A. All rights reserved.


PII S0924-0136(98)00142-3
240 J. Jeswiet / Journal of Materials Processing Technology 80–81 (1998) 239–244

Fig. 1. An illustration of the sensor, the sensor path as data is being aquired and typical experimental results.

friction force sensor measures forces correctly if the 4. Analysis of the friction results
following conditions are met: Fn, when integrated
over whole the arc of contact, must be equal to the The friction force profiles, for Ff, can be plotted
externally measured RSF. Also, the friction force in along a path to give a picture of how it varies
the rolling direction, Ff, when integrated over the through the gap (Fig. 4). The friction force profiles
contact surface and multiplied by the torque arm, the shown are very similar to those observed by van
roll radius, must equal the input rolling torque, Tiu. Rooyen and Backofen [5]. The area under the curve
is integrated to give a single value for Ff. The integra-
tion, Ff, is multiplied by the upper work roll radius,
3. Results where the sensor is located, to give the frictional
torque: Tsensor = R Ff. When comparing the friction
torque with the spindle torque, the error at higher
3.1. Experimental conditions reductions is within 5%.; Table 2 gives an example.

Experimental runs were made with 1100-H18 alu-


minium which has a tensile yield of 152 MPa and a
nominal surface hardness of 44 Brinel. The billets had
nominal dimensions of 6×16.3 ×305 mm. The roll
surface speed was 75.4 mm s − 1 for all roll passes. All
experiments were run without lubrication.
The mill is two high with 230 mm diameter work
rolls. Roll surface roughnesses are: upper roll (con-
tains sensor), 165 mm (rolling direction); 330 mm
(transverse direction); lower roll: 171 mm (rolling di-
rection); 279 mm (transverse direction).

3.2. Experimental data

Geometric details for each billet are given in Table


1 and examples of experimental curves at 24.2% re- Fig. 2. An illustration of the sensor showing the orientation of the
duction are given in Fig. 3. forces being measured.
J. Jeswiet / Journal of Materials Processing Technology 80–81 (1998) 239–244 241

Table 1
Billet geometric data; reduction Dr=(Dh/hin)

Run hin (mm) Dh (mm) r bin (mm) Db (mm)

55 6.0 1.52 0.25 15.88 1.14


60 6.0 1.45 0.24 16.48 1.07
56 6.0 1.45 0.24 16.51 1.09
57 6.0 1.45 0.24 16.48 1.02
53 6.0 1.45 0.24 15.77 1.12
59 6.0 1.45 0.24 16.59 1.09
54 6.0 1.47 0.25 16.15 1.09
58 6.0 1.42 0.24 16.48 1.04

4.1. The friction 6ector

If the experimental friction data are considered as


vectors along the rolling and transverse directions, then
a picture can be obtained of how friction actually varies
in the roll gap, both in magnitude and direction. This
can be done by considering the friction vectors at the
same instant in the roll gap, shown at position x in Fig.
5.
Ft and Ff can be added vectorially to give an absolute
friction vector Fu, which is not parallel to the rolling
direction; Ff is the vector normally associated with
rolling. Fig. 5 shows the magnitudes of all three friction
forces, Fu, Ft and Ff, plotted on the same graph. The
magnitude of the Fu vector remains reasonably con-
stant —its direction is not shown — however, the other Fig. 4. Examples of friction curves for experimental data, Ff and Ft
two vectors change magnitude considerably in compari- and the friction vector, Fu. The abscissa is in seconds ×10 − 2.
son but do not change direction. As can be seen from
Fig. 6, Fu remains within a definite band. the end of each curve is due to the sensor engaging and
If Coulomb’s model is used, one obtains the coeffi- disengaging the billet surface. Excluding the engage-
cients mu, mf and mt, as shown in Fig. 7. Note that the ment and disengagement sections, for a specific reduc-
sharp increase which can be seen at the beginning and tion mu remains fairly constant along the sensor path or
particular value of a, the distance from the billet edge.
The friction vector, Fu, varies slightly in magnitude
as it progresses through the roll gap, however, it rotates
by 90° towards the rolling centre line as it progresses
from the entrance to the neutral zone. Its magnitude
and direction also depend upon the distance from the
centre line of rolling. Note that this is for bar rolling,
not strip rolling and that the effect observed is probably
due to the low aspect ratio. The deviation of the
friction vector from the rolling direction may not be as
great for strip rolling. The effect of lubrication is also
unknown and further tests are planned.

5. Ford’s model

Ford [1] derived a relationship which relates exter-


nally measured rolling torque (Tiu), roll separating force
Fig. 3. Experimental friction curves in the rolling direction for 24.2% (RSF) and roll radius (R) to the coefficient of friction
reduction. mFord = (T/RSF)R. Although no longer in use, a com-
242 J. Jeswiet / Journal of Materials Processing Technology 80–81 (1998) 239–244

Table 2
Integrated force and torque data for r= 0.242

Run Sensor

Ff (N) Fn (N) RSF (N) Tiu (N · m) Til (N · m) Total (N · m)

55 196 5042 35751 199 188 388


60 182 4940 42075 184 200 384
56 191 5180 39440 211 200 411
57 227 7405 41020 192 225 416
53 276 6942 37069 174 209 383
59 223 7156 39703 213 211 424
54 231 6025 26264 190 225 414
58 227 6123 39178 193 226 420
Avg 219 6102 37562 194 210 405

[Avg Ff]*[0.115]*8 =202 N · m.


[Avg Fn]*8 = 48812 N.

parison was made with this relationship because the The foregoing suggests that an analysis can be made
data was readily available. The relationship is for all which includes the angle of the absolute friction vector.
the friction acting in one direction, i.e. no neutral point. Knowing that to a reasonable approximation, mu =
However, since the neutral point is fairly close to the mFord and that mFord = (T/RSF)R, we can find Fu, where
exit and most of the torque is derived from the friction Fu = T · N/RSF · R. For directional components com-
force on the entrance side, the Ford model is compared plex numbers can be used as follows: a vector function
to our experimental results (Fig. 8, Table 3). Fb u = A · Fu eiu, which gives Fb u = A · Fu (cos u+i sin u)
The comparison shows that the values are reasonably can be assumed. The real and imaginary components
close, so that if an approximation for mu is needed are Ff and Ft, so that for the real component Fb f =
Ford’s method could be used. However, there is a A · Fu cos u and for the imaginary component Fb t =
A · Fu sin u. The possibility of using this is being
directional component u, which must be taken into
investigated further.
account (Fig. 5).

6. Conclusions

Interface friction has been measured successfully for


low aspect (billet) rolling. Several details about the
interface friction condition have emerged from the
analysis.
(1) A friction vector, Fu, is the result of adding
friction vectors, Ff and Ft,.

Fig. 6. Plots of the friction force vector, Fu, for four sensor paths at
the same percent reduction, r =24.2%. The abscissa is in seconds
Fig. 5. Derivation of Fu. × 10 − 2.
J. Jeswiet / Journal of Materials Processing Technology 80–81 (1998) 239–244 243

Table 3
A comparison of values of mu and mFord

Run no. r a (mm) mu mFord mFord

Avg

40 0.20 1.0 0.025 0.035 0.036


44 0.20 5.4 0.033 0.031 0.036
45 0.20 4.7 0.039 0.042 0.036
60 0.242 1.6 0.049 0.038 0.046
56 0.242 2.7 0.053 0.046 0.046
57 0.242 6.1 0.030 0.041 0.046

(3) Fu rotates through 90° as it moves from the


entrance to the neutral point.
(4) When analysed as Coulomb friction, a coeffi-
cient of friction, mu, which is fairly constant is found
to be very close to the value of friction calculated by
Ford, mFord.
(5) It appears that mu, although constant in magni-
tude, is direction dependent. It may be noted that the
surface roughness was higher in the transverse direc-
tion, by a factor of two, than in the rolling direction,
which may account for part of the difference in m.
The effect of lubrication is unknown because this
work is for dry rolling; however, the sensor is being
improved so that studies can be carried out with lu-
bricants in the near future.
Fig. 7. The variation of Coulomb friction.

7. Nomenclature
(2) The magnitude of the friction vector, Fu, stays
reasonably constant within upper and lower limits as
a distance from billet edge to the path of the
is passes through the roll gap.
sensor (mm).
b billet width (mm).
Ff friction force in the rolling direction (N).
Ft friction force across the face of the roll (N).
Fn friction force normal to the billet surface (N).
Fu vector resultant of Ff and Ft
hin entrance thickness of the billet (mm)
hout exit thickness of the billet (mm)
r reduction, dimensionless; or percent
Tiu upper work roll spindle torque (N · m)
Til lower work roll spindle torque (N · m)

References

[1] H. Ford, Advanced Mechanics of Materials, Longmans, Green


and Co., London, 1963.
[2] A. Banerji, W.B. Rice, Experimental determination of normal
pressure and friction stress in the roll gap during cold rolling,
Ann. CIRP 22 (1) (1972) 53.
Fig. 8. A comparison of Coulomb friction with the results obtained [3] J. Jeswiet, W.B. Rice, The design of a sensor for measuring
using Ford’s method of calculation. normal pressure and friction stress in the roll gap during cold
244 J. Jeswiet / Journal of Materials Processing Technology 80–81 (1998) 239–244

rolling, SME/NAMRI Trans. 1982, p. 313. [5] G.T. van Rooyen, W.A. Backofen, Friction in cold rolling, J. Iron
[4] D. Britten, J. Jeswiet, A sensor for measuring normal forces with Steel Inst. (1957) 235.
through and transverse friction forces in the roll gap, SME/ [6] C. Nyahumwa, J. Jeswiet, A friction sensor for sheet metal rolling,
NAMRI Trans. 1986, p. 355. C. Ann. CIRP, 40 (1) (1991) 231 – 234.

You might also like