Professional Documents
Culture Documents
Notation
e, Discrepancy between directly measured co-
efficient of friction and that computed with
equation (9).
F Resultant frictional force.
Fa Front pull force.
Fb Back pull force.
1 ‘NEUTRAL POINT
i /
Q Standardized unit, Q = p
RO Roll radius.
r
s=s
(
Percentage reduction, r = 100 I--
3 .
Instantaneous value of the arc along which the
L I integration is done,
- - L
‘4
Fig. 2. Direction of resultant pressure and of frictional
forces at maximum reduction
t
T
tf
Torque.
Instantaneous value of the thickness along
which the integration is done.
Final thickness of the strip.
Initial thickness of the strip.
ti
f Friction forces. U Circumferential velocity of the roll.
Proc Instn Mech Engrs Val 174 No 32 1960
Velocity of the strip at an intermediate thick- (3) Power supplied by the front pull on the strip is
ness.
Velocity of the strip at exit. wa= truxfU= tfVf"Xf (3) . . .
Velocity of the strip at entrance. (4) The power supplied by the rolls is 2 w R ; after the
Velocity of the strip at neutral point. frictional losses 2Wf are deducted, it becomes 2( W,- Wf).
Power introduced by the front pull. It was easier to compute ( w R - W f ) than to compute
Power deducted by the back pull. each power separately. The result is
Frictional power losses.
Power of internal deformation.
Roll power.
Instantaneous value of the angle along which
the integration is done.
Angle of contact.
Angle at which the speed of the strip is equal
--2
d3
koi+"a ti-tf t i ] ) taIl-lJ(;- 1)
Coefficient of friction. 2
Coefficient of friction computed with equa-
tion (9).
Stress.
Stress components.
Yield limit at uniaxial load at finalthickness.
Yield limit at uniaxial load at initial thickness.
Stress along R, 8, and Z directions.
Back pull stress.
Front pull stress.
Yield limit at uniaxial load, a. m
2
Time. The condition that the sum of the powers vanishes is
2(@R-@f)+wa = Wi+@b . (5)
RESULTS Inserting the values computed earlier into this equation as
The five main powers involved are computed in Appendix I followed in Appendix I gives
and the results are as follows:
(1) The internal power of deformation is
-
1I
Wi = - .2v f t f {
d3 -1
["oi+"of-"oi
ti-tf
ti
:;
~--?-("of-"oi)
it is seen I
The second of equations (1) can be obtained by integration
of equation (6) as given by Ford (5). - ["oi+"- ti-tf ti]}x
For
it is seen that
P = f J(;) X
x tan-1 J(;-1)
For
Ooj-uoi 4 a0 M o+ao;
f (J
80 t I
2 I I
STANDARDIZED UNIT, Q
Maximum reduction possible, predicted by equation (6b)
for a wide variety of conditions, is presented in Figs. 3-7.
The effects of each of the individual variables such as thick- b e = = . r=f{L' ,/$ $}
-00
z T
3 0.2
5r
4 20 =o$l.'
/ /
O ,
0'
0
a
I
0.100
p = 0.07.
I
0.200 0300
I
Ro = loin. mb = 0.
I
0100
I
0.500
80,
I I I I 1
i
Q
U
0
0. I
0.2
0.4
06
08
0.400 0500
FINAL THICKNESS, Lf-In. STANDARDIZED UNIT. Q
u p = 0.07. Ro = 10 in.
Fig. 7. The effect of simultaneous front and back pulls on maximum reduction
I*=-
T
. . . (8)
PRO
. -a0 --
d3 tf
From the measured torque and separation force, Whitton
has computed the coefficients of friction for various metals,
. . . (9)
surface conditions and lubricants, and his results are Columns 1-9 of Table 1 are data readings as recorded
presented in Table 2 of his paper. during Whitton's (2) (3) experiment. A full account of this
Proc Instn Mech Engrs VoI I74 No 32 1960
Table 1. Comparison of the CoeBcients of friction according to equations (8) and (9)
(The values for colums 1-9 are from Whitton’s (2) experimental work.)
Roll radius, Ro = 2 in.
Test Lubricant Nominal Zoefficient Initial Final Material, Back Coefficient Dis-
No. reduction, of thickness, thickness, yield pull of crepancy
r, per cent friction, ti, in. t f , in. limit, stress, friction in the
IL 00 ton/in2 oxb ton/inZ according equation
to equa- of coeffi-
tion (9), cient of
pc friction,
e,, per cent
~-
r=
100 E f
ti
1 2 4 9 I 10 I 11
Table I-continued
Test Lubricant Pass Nominal Coefficient Initial Final Material, Back :o&cient Dis-
No. No. reduction, of thickness, thickness, yield pull of crepancy
r, per cent friction, ti, in. tf, m. limit, stress, friction in the
P 00 ton/$ uxb ton/in* according equation
to equa- of coeffi-
tion (9), cient of
PC friction,
e,, per cent
t= -PC--P
u--
100 fi-t.f P
ti x 100
-4 2 3 4 5 6 7 8 9 10 11
a ‘foof
00 x AL--
2
1 2 3 4 5 6 7 8 9 I0
10 2 0.0609 0.0503 34 40 37 0.0669 0.0645 0.058
26 3 0.0506 0.0383 40 48 44 0.056 0.054 0.054
25 4 0.0438 0.0336 44 49 46.5 0.053 0.052 0.049
work is given in (2). A condensed description and tables of the maximum reduction for a wide range of the variables.
with partial data are given in (3). Column 5 in Table 1 If the range is to be extended, more graphs can be easily
of the present paper gives the coefficient of fiction as constructed.
computed by Whitton according to equation (S), while
column 10 presents the coefficient of friction as computed
CONCLUSIONS
byequation (9). The discrepancybetween the two approaches
is given in column 11. Of the long list of readings recorded The energy approach yields a criterion for maximum reduc-
by Whitton (2) (3), only a few representative cases are shown tion in cold strip rolling with front and back pull, Further-
in the present Table 1. Table 2 compares the solutions more, a procedure is suggestedfor the experimentaldetermi-
according to equation (6a), and equation (9) with Whitton’s nation of the apparent coefficient of friction, the simplicity
.measured values of the coefficient of friction. of which enables it to be applied on any standard mill.
About 6 per cent of the readings show over 25 per cent The energy approach appears promising as a simple,
discrepancy. Regarding all the difficulties encountered in short method for the determination of the torque and power
any experimental determination of coefficients of friction, requirements in operation of cold strip rolling.
it seems safe to state that the analysis is not contradicted The effect of each of the process variables on the maxi-
by the data. It is therefore concluded that the energy mum possible reduction can be studied from Figs. 3 to 7.
approach is correct. This approach results in an analytical It should be borne in mind that the rolls were assumed to be
expression for the prediction of maximum possible rigid bodies and that flattening was neglected. The graphs
reduction. are therefore not reliable for the range where tr approaches
Based on equations (6), a simple procedure is suggested zero. The following observations can be made:
here for the determination of the coefficientof friction under (1) Larger reductions are possible with higher values
actual working conditions which is a modification of of the coefficient of friction. (This phenomenon is well
Whitton’s experiment. Measurements of the torque and known qualitatively.)
separation force are not required. Almost any mill can be (2) Larger reductions are possible for thinner strips.
used. As the strip gets larger this effect diminishes.
(3) Higher reductions are feasible with bigger rolls.
Procedure This is one advantage of big-diameter rolls which should
A strip of arbitrary original thickness ti is rolled while not be overlooked (Fig. 4).
screwing down the roll until maximum bite is achieved. (4) Increasing the back-pull stress decreases the
The best choice is no applied front or back pull, because maximum reduction. With high back pull and too small
then the values uXf= uxb = 0 are known without a coefficient of friction a condition can be reached where
measurements. The yield limits uOiand yor are known or no reduction is possible at all.
can be measured. Inserting these values lnto equation (6) (5) With no back pull, and changing front pull,
the apparent coefficient of friction for the specific working different effects can be expected. Usually, increase in
conditions of the mill can now be calculated. fiont pull causes an increase in maximum possible
By inspection of equation (7) it can be seen that the reduction. However, in some combinations of the
maximum reduction r, which is a function only of the variables, an increase in front pull can cause a decrease
initial ti and final tf thicknesses, is implicitly expressible as in maximum possible reduction. This is explained as
the following function: follows:increasing the front pull decreases the separation
force. Decreased separation force with constant coefficient
of friction causes a decrease in the frictional force. If the
loss in frictional force is not fully compensated by gains
in front pull force, then the maximum reduction is
reduced (Fig. 6).
where (6) Simultaneous increase in both front and back pull
causes a decrease in maximum possible reduction, as
It is thus most convenient to use the quantity Q in the shown in Fig. 7.
graphical presentation of the effect of the process variables
on maximum reduction. For more detailed explanation of
the utilization of the standardized dimensionless terms, ACKNOWLEDGEMENTS
The author wishes to thank Dr. C. W. Phillips for his
uxf and 3,
Q, 2 2
see Appendix I (Figs. 5b, 6b guidance during the study and for assistance in preparing
-=Q -00 the report. The author is also indebted to Dr. I?. W.
d 3 d 3 Whitton for the experimental data as well as for his com-
and 76). ments during the course of the study. It was a pleasure to
It will be assumed that in any mill the actual apparent receive help from Mr. N. H. Miller who programmed the
coefficient of friction is easily determined. Figs. 3-7 have computations, and from Mr. F. J. Semer in processing the
been plotted from equation (6b) for easy determination data.
Proc Instn Mech Engi-s Vol174 No 32 1960
APPENDIX I
Fig. 1 describes the geometry of the process and the assumed
pattern of the deformations.
Equations (10) to (26) establish the relations between the angular . . . (28)
position a, the thickness t and the velocity v of the process. dS dS
The stress and strain fields
S =2Ra . . . . . (10)
sin a = t / 2 R . . . . . (11)
And the strain-rate field becomes
Therefore
S=-
sin a
ta
. . . . . (12)
t = tj+2Ro(l-~os a) . . . . (13)
ForO<a<z
2
a3 a5 a7
sin a = a--+--- All other i i j = 0
3! T! 7!+.'.
Power balance a t maximum reduction
cos a == 1 -""+---+
a4
2 ! 4! 6!
a6
. .. The internal power of deformation
With very small angles a 1 The power per unit volume (7)is
sin a M a . . . . . (14)
C O S ~M 1-z .a2
. , . . (15) . . (31)
Therefore Integrating over the volume, per unit width gives
2
aa = 2Roa . . . . . (17)
And for future reference
Therefore
From equation (28)
. . 4 (33)
From volume constancy are derived
Inserting these equations into equation (32) yields
v =L, V ? ! f X L . . (24)
1+-Ro a2 Ro t f + , Z
tf Ro
For the case discussed here, when the neutral point is at the exit
( a , = O)J
wj= U . . . . (25) .
v=wfx1=2Lfxux- 1 . .
Ro 2 + , 2
Ro
Ro g+az
(26)
. . . (50)
. . . (38) or
The power supplied by the front pull is 'Jxf-Oxb ti
wR-Wf = -tfpU
Wa = Favf = a x f t f U . . * (39)
The frictional power wf (Fig. 2) is
wf=-L
a2
O
pq(U-v)Roda . . (40)
The rolls power I$R
' is
WR= -Ja=op~aURoda
a2 . . . (41)
puevda . . (42)
. . . (44)
The form of the yield condition for a plain strain condition is
2
0.3 = UR--
4 3 O0
. . . . (45)
and therefore
. . . (53)
. . . (46) The equation for the power balance is
2(WR-@j)=@i+@b-@a . (54)
And inserting all the individually computed powers it becomes
f
2
'Jxf-'Jxb+- ('Jof-'Joi)
- d3
ti-tf Ro(g+d) . . (48)
43
Proc Insrn Mech Engrs Vol174 No 32 I960
When uof-Uoi -
< o0 z “of +ooi then It is thus seen that when uxf = uxb = 0 and O o f - U o i Q UO,
2 then the maximum reduction can be expressed as a function
of one variable only. Let this variable be called Q so that
r = f { j 3 ; -uo F}
- . . (61)
43 43”O
This dependence is shown in Figs. 5-7.
APPENDIX I1
- ‘Jxb t‘
tf -
OXb
4 3 uo
}. . (57)
(I) LARKE,
REFERENCES
E. C. 1957 ‘The Rolling of Strip, Sheet and Plate’
(Chapman and Hall, Ltd., London).
(2) WHITTON,P. W. 1953 Ph.D. Thesis, London University
‘An Investigation of Surface Friction in Rolling’.
(3) WHITTON, P. W. and FORD, H. 1955 Proc. Instn mech. Engrs,
Equation (7) can be written in the following form: Lond., vol. 169, p. 123, ‘Surface Friction and Lubrication
in Cold Strip Rolling’.
ht’ (4) VAN ROOYEN, G. T. and BACKOFEN, W. A. 1957 J . Iron St.
2 tf =- P
. . . (59)
Inst., vol. 186, p. 235, ‘Friction in Cold Rolling’.
tan-1 J($-I) J($) (5) FORD,H. 1948 Proc. Znstn mech. Engrs, Lond., vol. 159,
p. 115, ‘Research into the Deformation of Metals by
Because the maximum reduction r is a function of the left-hand Cold Rolling’.
side of the equation, it therefore follows that r is also a function of (6) BLAND,D. R. and FORD,H. 1948 Proc. Znstn mech. Engrs,
the right-hand side of that equation Lond., vol. 159, p. 162, ‘The Calculation of Roll Force
and Torque in Cold Strip Rolling with Tension’.
f U - l
(7) HOFFMAN,0. and SACHS,G. 1953 ‘Introduction to the
. . . .(60) Theory of Plasticity for Engineers’, p. 54, Equation (5.43)
.=it&)) (McGraw-Hill Book Company, Inc., New York).
Communications
Dr. J. M.Alexander, B.Sc. (Eng.) (Member), wrote to say would be 1.75. Reference to the author’s Fig. 5 showed
that he thought the author’s whole approach to the problem that for that value of Q the maximum reduction achievable
of cold strip rolling was dangerously misleading. By neg- would be almost 100 per cent. Those figures actually
lecting roll flattening he had achieved a simplified theory,. applied to rolls of 20 in. diameter and p = 0.07. Thus,
although that was a moot point, bearing in mind the number according to the author’s analysis, with 20-in.diameter rolls
of equations contained in Appendix I. The main point at and p = 0.07 the maximum reduction which could be
issue was whether the equations he had developed truly achieved to produce a steel strip of final thickness 0.016 in.
represented the behaviour of material during cold rolling. was at least 99 per cent, i.e. from an initial thickness of
It was very easy to show that they did not, as follows. 1.6 in.! In view of that, the author must certainly agree
Consider a rolling mill with 20-in.diameter steel rolls, that it would be possible to roll a steel strip from 0.030 to
rolling without tensions a strip of steel to a final thickness 0-016 in. in the mill with 20-in. diameter rolls. In fact, it
of 0.016 in., which was a typical gauge for sheet material. was not possible to roll the strip below a thickness of about
To simp@ the calculations, let it be assumed that there 0.024 in., due to roll flattening.
were no applied tensions, and that the strip had previously The author had neglected completely the fact that there
been cold-rolled so that it could be assumed to be non- was a limiting thickness in rolling, i.e. a thickness below
hardening. What would be the maximum reduction which which it was impossible to cause permanent deformation.
could be applied to that strip, to bring it to the h a l thick- When that thickness was attained, increased roll force
ness of 0.016 in. ? merely served to flatten the roll surface still farther, thus
Using the author’s analysis, the value of Q = pz/(R,/tf) increasing the length of the arc of contact and introducing
Proc Instn Mech Engrs Vol174 No 32 1960
longitudinal compressive stresses due to the increased larger was the limiting thickness and again roll flattening
frictional forces at the surfaces. Those additional longi- precluded the attainment of very large reductions. The
tudinal compressive stresses suppressed yielding of the strip author’s final three conclusions that increasing the tensions
so that it could not be reduced farther. decreased the maximum reduction possible were very
The problems of ‘maximum reduction’ and ‘limiting difficult to understand. It could be seen from equation (62)
thickness’ had been discussed exhaustively at a symposium that the higher the tensions the smaller the minimum thick-
held by the Institute of Metals at their spring meeting in ness which could be rolled, so that increased tension was
1960. The theoretical background to those two problems beneficial from that standpoint. Also, increased tensions
had been discussed by Ford and Alexander*, and an assisted plastic flow of the strip and by reducing the roll
analysis presented for determining the limiting thickness force, reduced the roll flattening, which should again be
in which both the elasticity of the rolls and of the strip had beneficial.
been allowed for. The author had neglected the elasticity There were many other features of the author’s theory
of both rolls and strip in his analysis, which rendered it which were untenable, representing as they did an un-
quite unrealistic. realistic oversimplification of the theory of rolling. Up to
The limiting thickness was given by the equation the present the theory of rolling had been developing along
more and more realistic lines, and the principle that roll-
14*22p2R(1-vv,2) 9.06pR(1--~,2) flattening should be allowed for in cold strip rolling had
h,, =
[ Es + Er
](m~-t)
been tacitly assumed by all writers. It might well be possible
. . * (62) to neglect roll-flattening in the case of hot rolling, when the
material being deformed was considerably softer than the
where R was the undeformed roll radius, E,, vr Young’s rolls, so that roll pressures were not so high. Neglect of roll
modulus and Poisson’s ratio for the rolls and E,, vs Young’s flattening in cold rolling could only lead to serious errors
modulus and Poisson’s ratio for the strip; Y was the yield in any theory, and while that might not be too disastrous
stress in uniaxial tension, m, a factor between 1 and 2 / 4 3 if it were only ‘trends’ in the process variables which had
( 2 / 4 3 for yield criterion of von Mises), and t the mean of been sought, the author claimed much more for his analysis.
the back and front tensions. He claimed,for example, to be able to predict the maximum
For steel rolls and steel strip, E, = E, = E, vr = Y, = v, reduction and to be able to determine the friction coefficient
equation (62) reduced to by applying his theory to experimental results. The co-
efficient, he would be determining, in fact, would be a
Ln= (9*06+14.22~)pR(1-v2)(m
E
Y-.f) (63) parameter farther removed from the true friction coefficient
than that determined from the experiments of Whitton and
Considering the example just quoted, with p = 0.07, Ford.
equation (63) showed that the minimum thickness which He seemed to have been developing his theory completely
could be rolled was 0.0245 in. Therefore it was seen that without reference to all the work which had been previously
the author’s theory gave a completely unrealistic result for carried out on that topic. Stonet, Hill and Longman$ and
a very important case of rolling, and that inaccuracy would Simss had all considered the problem of maximum reduc-
apply to many other important instances. tion in cold strip rolling, and developed much more
Viewed in the context of the limiting thickness (equa- realistic analyses, since they all allowed for roll flattening.
tion (63)), certain of the author’s conclusions were The author made no mention of the work of those authors;
misleading. He concluded that larger reductions were and it could only be assumed that he was unaware of it.
possible with higher values of the Coefficient of friaion, For that reason, presumably, his notation was quite different
where as it could be seen from equation (63) that the from that generally adopted; which made his analysis
smaller the coefficient of friction the smaller the limiting difficult to follow. In any case, the whole theory had been
thickness. The implication that it was desirable to have a made unnecessarily complicated. Considering Appendix I
high friction coefficient was therefore misleading, since the and Fig. 1, for example, it hardly seemed necessary to intro-
rolling-mill manager was usually more interested in duce the curved section of radius R since that concept was
achieving the smallest thickness possible, without a vast not used in the theory at all. Also, there were nine equations
expenditure of power in roll flattening. The author’s con- (from (10) to (18)) required to derive the expression
clusion that larger reductions were possible with thinner
strips was also misleading. After a certain ‘thinness’ the
effect of roll flattening was so important that it could not
be neglected and enormous expenditure of power was which was precisely the same as the equation h = ho+ R‘+2
required to achieve even small reductions. which featured in the theory of Bland and Ford, except
The author concluded that higher reductions were that R‘ was the flattened radius of the roll. That could surely
feasible with bigger rolls. The larger the roll diameter the
* FORD,H. and ALEXANDER,
J. M. 1960 J . Inst. Metals, vol. 88,
p. 193, ‘Rolling Hard Materials in Thin Gauges-Basic Con-
*
f STONE,
p. 705.
M. D. 1953 Iron Steel Engr., vol. 30 (Z),p. 61.
HILL,R. and LONGMAN, I. M. 1951 Sheer Metal Ind., vol. 28.
siderations’. s SIMS,R.B. 1954J. Iron St. Inst., vol. 178,p. 19.
Proc Instn Mech Engrs Vol174 No 32 1960
Mr. M. Bentwich (Haifa, Israel) wrote that the analysis while R was given by
presented in Appendix I was not very consistent when R Sin a+Ro cos = Ro+tf/2
judged from a purely mathematical standpoint:
Combining the two
(1) Equations (14) and (15) indicated that the functions 1-cos a
sin a and cos a were considered as power series expansions,
which were truncated to retain powers of a of order 2
x = -Ro
(2+- ____
sina
or smaller. If, then, powers of the same order were retained was obtained which yielded upon differentiation
after expanding sin a and carrying out the division in
equation (12), the following was obtained
5
S = sin a M t ( l+$) = Ro[g0+a2( 1 + 3 ) ] If a were considered to be very small or to vanish
‘maximum bite’ and ‘maximum reduction’ would be used strain compression test and hence his values of u should be
to distinguish between those two phenomena. It was, in used directly and all reference to the 2/.\/3 in equations
fact, that latter restriction which was most important from (1)-(7) should be omitted when calculating from his results.
an industrial standpoint and considerableresearch had been The author justified the assumption that uof-uoi -g
devoted to an analysis of the problem and to suggesting uo w (uOf+uoi)/2 which had been used in calculating the
methods of increasing the maximum reduction. In general values of p in Table 1, by comparing the results with three
the conditions necessary for high rolling reductions were values calculated by the full theory (Table 2). In each
the exact antithesis of those which the author postulated instance, however, a pass on a work-hardened material had
as being important for achieving maximum bite. For been chosen and it was for reductions on annealed stock
example, larger reductions on thin tinplate were possible that the greatest error was likely to arise. Since 40 per cent
only by reducing the coefficient of friction, reducing roll of the data in Table 1 were for annealed material, it was
diameter or by increasing the tensions. worth comparing the results of the two calculations under
It was difficult to justify an analysis which neglected the those conditions. That had been done in Table 4 after cor-
effect of roll flattening, particularly where large reductions recting the equations for the 2 / 4 3 factor. The discrepancies
were involved. Orowan* had shown that even for relatively resulting from the approximations were particularly marked
small reductions it was important to allow for the elastic for the annealed copper and steel and the overall agreement
deformation of the rolls and that the errors involved in with Whitton’s results was worse when the correct value
neglecting the effect increased with the reduction. The of the yield stress in plane compression was used.
theoretical deformed roll radius for some of the experiments
quoted in Table 1 had been calculated and were shown in
Table 3. Those figures, however, were for relatively small Table 4. Comparison of the coeficient of friction according
to equations (6u), (8), and (9)
was shown in Fig. 10 where the maximum bite was shown complete stalling occurred when neither back or front
as a function of the coefficient of friction for a constant tension was applied was also suggested by calculations of
ingoing gauge. The larger of the two bites was obviously forward slip*. It could be shown, using the simplifying
imaginary since slipping would occur at the lower draft. assumptions of a constant coefficient of friction, an ideal
Although the author had qualified his conclusions by plastic material and homogeneous deformation with a uni-
stating that the curves were probably unreliable when t , form pressure distribution, that forward slip reached a
approached zero he had given no indication of the limits of maximum value when the contact angle a2 was equal to the
validity. In Fig. 9 it was suggested that the theory was friction angle tan-’ p. The analysis showed that forward
slip then decreased with further reduction until a point was
reached where the forward slip became zero when the
neutral point had advanced to the exit point and stalling of
the strip took place. When that occurred the friction angle
was half of the contact angle*. A friction coefficient under
those conditions could similarly be derived from a simple
force analysis. Using the same notation as that of the author
and the previously mentioned simplifying assumptions, a
force equilibrium when stalling occurred could be estab-
r
lished such that:
/oazuR.Ro.sinada = p.~~.R~.cosada
therefore
1-cos “2
p=-
sin a2
0 “2
005 0.06 0.07 =a-
COEFFICIENT OF FRICTION,p 2 * . .
Fig. 10. Effect of coeficient of fiction on maximum bite It would be noticed that equation (64) was identical with
Ro = 10 in. that derived by considering forward slip and that it
t; = 0.4 in. depended only on the draught and the roll radius. When
Uzi = U x b = 0. account was taken of both work hardening and the friction
hill the maximum reduction would be found even greater
unreliable over the whole range which was of particular than that given by equation (64). A number of years
interest to the cold rolling industry. For a coefficient of earlier he had attempted that technique by rolling a strip
friction of 0.075 the theory was not applicable for ingoing which had been tapered in the rolling direction such that
gauges of less than 0.5 and 0.25 in. for strip rolled with the thin end had entered the roll gap first. The use of a
rolls of radius 10 and 5 in. respectively. tapered strip thus had eliminated the need for adjusting the
roll gap during rolling in order to obtain larger reductions.
Professor G. T. van Rooyen (Pretoria) wrote that the Experimentally it had, however, been found impractical
energy method for deriving at operating forces and power since the draught required had been far in excess of the
requirements for plastic deformation processes was a well- capacity of the mill available unless the friction coefficient
known technique. One of the greatest difficulties en- had been very small
countered in that approach was that of evaluatingthe energy It was, of interest, to compare the friction coefficients
losses in the process. Energy losses due to friction could derived by equation (64) and the author’s equation (7) since
generally only be found when the friction coefficient, pres- exactly the same initial assumptions had been used. The
sure distribution as well as the relative velocity of slipping results for a roll of 20 in. diameter were tabulated in
was known. Energy losses due to redundant strain (accom- Table 5.
panying non-homogeneous deformation) were even more A comparison of the friction coefficients showed agree-
difficult to calculate. Those factors in the engineering ment at certain points while they disagreed rather widely
approach were generally taken care of by means of a defor- at other points. The friction coefficients as calculated with
mation efficiency. The fact that the author did not consider equation (7) showed the rather puzzling result that once a
energy losses due to redundant strain could therefore be certain reduction had been reached, increased reductions
expected to limit the applicability of the results to such could be obtained with lower and lower fiiction coefficients
instances where the friction coefficient and the accompany- which contradicted experience and appeared to reflect on
ing reductions were small. the accuracy of equation (7).
The method suggested by the author for determining the * UNDERWOOD, L. R. 1952 ‘The Rolling of Metals’, vol. 1, pp. 22,
fkiction coefficientby rolling with increasing reduction until 23, Fig. 3 (Chapman and Hall, Ltd, London).
Proc Instn Mech Enps Vol174 No 32 I960
Table 5. Comparison of .friction coefficientscalculated Equation (6) and its reduced form equation (7) were
by equations (64)an;-( 7) used to establish the results in Fig. 3 through 7, from
which the conclusionsregarding the behaviour of the system
ti,
parameters were drawn. Nowhere in this development were
P
in. the conditions for the equilibrium of the forces acting on
per cent Equation (64: Equation (7) the strip considered. Even though the effect of the surface
___-
4.75 5 0.08 0.075 shearing forces on the strip had been neglected in establish-
4.50 10 0.12 0.109 ing the distribution of axial stress through the contact
4 20 0.16 0.102
3 40 0.24 0.205
2 60 0.30 0.232 I
1 80 0.35 0.230 // I
0.5 90 0.38 0.206
0.25 95 0.40 0.175
0.1 98 0.41 0.137
0.5 0.45 10 0.035 0.034
0.5 0.40 20 0.050 0.048
0.5 0.30 40 0.07 1 0.065
0.5 0.20 60 0.087 0.073
0.5 0.10 80 0.10 0.073
0.5 0.05 90 0.106 0.065
0.5 0.025 95 0.1 10 0,055
0.5 0.010 98 0.112 0.043
system parameters analogous to equation (6) of the paper. That indicated that for reductions greater than about
For the instance of zero back and front tension and moderate 20 per cent the discrepancy between the results could not
strain hardening, equation (66), after integration, reduced be disregarded. It was felt that even greater discrepancies
to the following relation for p in terms of w2. might be found in the case where there was front tension
but no back tension. Since the principle of equilibrium
using the author’s model for the stress distribution did not
agree with the power balance approach for large reduc-
That could be rewritten using equation (21) in the form tions, considerable doubt was cast on all results presented
in his paper for large reductions.
It was felt that that discrepancy was due primarily to
neglecting the effect of surface friction on the stress distri-
bution in the contact arc. Before the results presented by
the author could be accepted, the model used in the analysis
should give closer agreement between the power balance
For s m a l l reductions the second term in the denominator and strip-equilibrium principles than it currently did.
of equation (68) was small compared to unity and could
be neglected giving Professor P. W. Whitton, B.Sc. (Eng.), Ph.D.
(Associate Member), wrote that he had a few general com-
ments on the use which the author had made of the experi-
mental work of Ford and himself (2)(3), and also a few
That result might be compared with equation (7) of the queries relating to the analysis and comparisons presented.
paper which when expanded had the form Ford and himself (3) had pointed out in 1953 that, due to
the continuous nature of the practical friction determination
tests, it had been difficult to ensure that the strip thickness
readings corresponded exactly with the roll force, torque
and back tension, etc. For that reason ‘steady state’ experi-
ments had been run to check the p values carefully. In
addition their fiction values, used by the author, had been
. . * (70) examined by other workers in Japan* in 1957, who had
For small reduction all but the first terms of the series reported very close agreement. In the paper the dis-
expansions could be neglected resulting in crepancies between the p = T/PR method and that of the
author, as listed in column 11 of Table 1, should be read
with that independent verification in mind. The ‘Avitzur’
equations gave values which, with very few exceptions, were
Comparing equation (71) with (69) it was apparent that the high compared with the observed values. On the evidence
power balance approach based on the simplifying assump- thus available it was very much more reasonable to suppose
tions used by the author satisfied equilibrium only for small that the discrepancies were due to assumptions of the
reductions. The discrepancy between the results obtained theory, albeit the overall agreement seemed particularly
from equations (68) and (7) for larger reduction was good.
illustrated in the following examples. For R, = 2 in., The procedure proposed by the author to achieve a p
zf = 0.2 in., r = 20 per cent equation (68) yielded for the calculation was, of course, well known; it was the
coefficient of friction method of calculation that was Merent. The main dis-
advantage of the determination of maximum ‘bite’ was
p = 0,0795 simply the difficulty of determining the point at which
From Fig. 3 the value calculated by the author for the same skidding commenced in a repeatable way and variations in
conditions using equation (7) was that must reflect upon the choice of curve to be taken from
the author’s figures for maximum reduction. That was
p = 0.075 particularly important for gauges below 0.020 in. Bearing
That represented a 6 per cent discrepancy based on the in mind too, the advanced stage of knowledge in cold rolling
author’s calculated value. For Ro = 2 in., tr = 0.2 in., and the intensive research in the subject over the last
r = 80 per cent equation (68) yielded 20 years, the assumptions made in the energy approach
undoubtedly masked demonstration of true effects and did
p = 0.339
not, indeed, lead to a simpler solution. For example, the
Again from Fig. 3 the calculated value was rigid-roll assumption led to the conclusion of the author
that larger reductions were possible for thinner strips-in
= 0-225
For that larger reduction the discrepancy had increased to * YAMANOUCHI,
H. and MATSUURA, Y. 1957 Rep. Castings
Research Lab., Waseda University, Tokyo, vol. 8, p . 57,
51 per cent based on the author’s calculations. ‘Measuringthe Coefficient of Friction in Cold Strip Rollmg’.
Proc Instn Mech Engrs VOI 174 NO 32 1960
general true but not universally so-with hard strip larger It could be said, without any hesitation, that the paper
reductions were possible by reducing thicker strip, a result was particularly opportune because the study of friction
of roll deformation and its effect on load. Similarly the in rolling was receiving close attention generally throughout
u +a. research organizations and also for forming ‘new metals’
assumption Of OO was likely to be wide Of the whae it played an important role in determining roll wear,
2
mark for first passes. deformation possible, limiting thickness, etc.
Repeated mention (Dr. Alexander, Mr. Guy, Dr. Wein- when that simplified approximated stress discribmion was
stein and Dr. Zorowski) had been made that roll flattening coupled with the energy approach, the introduced error in
and the friction hill should be taken into account. Solutions the power balance was small.
by the equilibrium approach were very sensitive to stress The example of R, = 2 in., tf = 0.2 in. and r = 80 per
distribution and roll flattening. That was not the case when cent, given by Dr. Weinstein and Dr. Zommki, would
the energy approach was utilized. Of the energies computed, imply a2 = 0.635 radian. That was a big angle of contact
the ideal power of deformation, and the front and back pull and violated assumption No. 7 on p. 866. The error arose
powers were entirely independent of the assumed stress because Fig. 3 in the advance publication was incorrect.
distribution. The errors, introduced by the assumed stress The analysis with strain hardening effecr assumed linear
distribution, in roll and frictional energies were partially strain hardening. In the annealed state (first pass) the strain
compensated by each other. The frictional losses were hardening was not linear. Furthermore, it was very doubt-
computed by multiplying the velocity difference bet ween ful which value to pick for yield at zem reduction. There-
the roll and the strip by friction stress. The error in the fore, equation (6a) should not be applied to the first pass.
distribution of the friction force was proportional to the That had been the source of discrepancies pointed out by
error in the stress distribution. At the exit no error in Mr. Guy and shown in his Table 4. That was the reason
friction losses was introduced because the assumed stress why Table 2 did not include first passes.
was in agreement with the friction hill solution. Friction The discrepancy between the experimental coefficient
losses in the vicinity of the entrance accounted for most of of friction and that computed by equation (9) for the first
the friction losses. At the neutral point, both the difference pass was not higher than that for the other passes. There-
in velocities and the friction losses were zero, so that no fore, equations (9) and (6b) were recommended, even for
error was introduced there. In between, the amount of first passes, with yield value as that at half the reduction.
error was limited as long as the pressure was reasonable. Professor van Rooyen and Dr. Weinstein and Dr. Zorow-
The quantities which made the power balance were there- ski had correctly pointed out that the accuracy of the
fore not too sensitive to roll flattening. That was not to say approach improved as the reductions and coefficient of
that attempts to account for flattening and obtain better friction reduced.
accuracy should not be made. Furthermore, another smooth He wished to thank Dr. Bentwich for his derivation of
stress distribution might be chosen, not that of equation (44), the arc length S and the velocity, in a more consistent
for better agreement with the experimental data. It was approach to the approximations.
only to be said that sufficiently good results had been The solution for maximum possible reduction before
obtained with the assumed stress distribution. skidding, as presented in that work, covered a problem
Dr. Weinstein and Dr. Zorowski had checked if equili- which had not been answered previously. Its limitations
brium was maintained with his assumed stress distribution had been pointed out by several writers in the communi-
and had found increasing error as the reduction increased. cations. Also, the general agreement with Professor
Attempts to find criteria for skidding through the equili- Whitton’s data, and the consistent higher values of the com-
brium approach failed. Either because the expression for the puted coefficients of fiiction suggested that bettering of the
stress distribution was too complicated, or far-fetched from results was both desired and possible. In conclusion, the
reality. possibilities of the energy approach, in solving many of the
The stress distribution assumed in equation (44) was not rolling problems had been demonstrated. Therefore, that
intended to replace the right stress distribution. However, approach should be explored, not ignored,