You are on page 1of 4

An Experimental Study of Boundary Conditions in Hot and Cold Flat Rolling

John G. Lenard, University of Waterloo/Canada - Submitted by S. Kalpakjian (1)


Received on January 11,1990

The significance of heat transfer during hot rolling and of the interfacial forces during warm and cold rolling at the roll/strip contact surface is
considered.The techniques used for the determination of the coefficients are given briefly and the heat transfer and frictional coefficientsare
established experimentally as functions of the process and material parameters. The dependence of both coefficients on those parameters is
discussed. Average coefficientof friction results,obtained by theembedded-pin technique,are shown to be similar to those, obtained by measuring
the forward slip. It is shown that the frictionalforces depend on the reduction, the rubbingvelocity,the temperature and the presence of a lubricant.
The heat transfer coefficient,while affected by these, is much more strongly dependent on the condition of the surface, that is, the presence of
scaling or a lubricant.

KEY WORDS hot rolling, cold rolling, heat transfer coefficient,friction

INTRODUCllON

Predictive-adaptivecomputer models of hot strip mills are designed to make They were loaded in compression resulting in an interfacial pressure of 49
decisions rcgarding mill draft schedules in real-time. The quality of these MPa.Some of the specimenswereheated to7800Cwhilethe otherswere kept
decisions depends directly on the quality and rigour of the computations of at room temperature. All were instrumented with 05 mm diameter, sheathed
thepredictive component.Apredictivetypemathematica1modelofflatrolling thermocouples. The contact surfaces were prepared with either scales,
should account for the balance of forces within the plastically flowing metal, lubricants or were kept dry. The results, presented below in Table 1 (repr-
its thermal balance and microstructuraldevelopmentand the current state of oduced from Murataet al., 1984)show that the heat transfer coefficientvaried
the deformation of the work-rolls in addition to appropriate representations from a low of 5.8 kW/m2K to a high of 465 kW/m2K which is nearly a
of the initial and boundary conditions. It is fairly straightfonvard to describe two-order of magnitude variation. It is noted that the testing temperature of
the conditions of thermal and mechanical equilibrium in a strip during the 78WC is significantly below the hot rolling range for most steels. Further, one
rolling process. Roll deformation can also be computed with reasonable may only speculate about the effectsof the relative velocity between the strip
accuracy. Having access to cooling rates and data from Jominy tests and and the roll, the random nature of scale formation or the varying roll pressure
continuous cooling curves would allow one to monitor numerically the on the coefficientof heat transfer.
dynamic metallurgical changes experienced by the rolled strip. Some of the
uncertainties of the modellingprocess are associated with the determination Further examination of the published values of a make its choice for use in
and the description of the material's resistance to deformation, referred to modelling even more difficult. Stevens et al. [2] suggests 38700 W/m2K for
above as the initial condition. Most of the difficultiesare encountered how- hot rolling with watercooled rolls. Preisendanz et al. [3] also provide values
ever,when events on the boundariesof the domain of interest are considered. within the ranges mentioned above; they give 1.2 and 2.5 MJIm2sec at 7 W C
For the mechanical component these refer to the nature of friction at the and l l W C respectively. Silvonen et al. [4] use 7oooO W/m2K in their recent
roll/strip interface and the shape of that interface. In the thermal portion the publication while Bryant and Chiu [5] employ 7000 kW/m2K. These values
heat transfer coefficientat the contact needs to be prescribed. Review of the have all been obtained on full scale production mills. In a laboratory setting
literature reveals that if modelling is considered, not enough published Harding [6] determined 2055 W/m2Kat 7WCand 5100 W/m2Kat 11WC,
information is available to choose either coefficientwith confidence. close to the values of Pietnyk and Lenard [7], also determined on a two-high
experimental mill.
The heat transfer coefficient
The coefficientof friction
Murataet al. [l] measured the heat transfercoefficientinuniaxialcompression
and reported the effects of several lubricants and scale on the magnitude of The traditional approach in modelling friction during the flat rolling process
a.The specimenswere made of carbon steel, containing 0.12% C. has been to assume that Amonton's law holds and that the frictional forces in
the roll gap are proportional to the normal forces, with the factor of propor-
- -
tioiiality the friction Coefficient remainingconstant and depending only on
the contacting materials. The constant friction factor, relating the interfacial
frictional stresses to the yield strength of the material in pure shear, has also
Lubricant
been used. In either case the choiceof the appropriatevalue of the coefficient
has been problematic. Experimental data is scarce and contradictory; com-
no scale scale( l O u ) pare, for example the results of van Rooyen and Backofen [8] and Al-Salehi
et al. [9]. Both studies used the embedded-pin technique to measure directly
Dry -
29.1 34.9 -
7.0 10.5
the interfacial normal and shear stresses at the roll/strip interface. In [8]
aluminum strips were rolled with and without lubrication, using a ground roll
Water 233 - 81.4 10.5 aswell asaroll with itssurfaceprepared bysand-blasting.Smoothrollpressure
curves were reported and the locations of zero shear stresses were found to
Oil A 233 - 465 5.8 be close to the positions of the peak pressures. In [9] aluminum, copper and
steel stripswere rolled, all without lubriixtion. The roll pressure distributions
Oil B -
7.0 105 contained several peaks. The positions of the neutral points did not often
coincidewith the locations of the maximum pressures.
oil c 333 - 465 5.8 - 9.3
The condusion is again as stated above: difficulties may be encounteredwhen
Oil D 70 - 233 5.8 - 9 3
modelling is considered and the coefficient of friction is to be chosen.
Oil D + 200/4caCO3 -
70 174 12.8 - 23.3
THE EXPERIMENTALPROGRAM
Oil D + 4O%CaCQ 12.8 - 17.4 17.4 In order to alleviate the choice of both coefficientsa systematicexperimental
program was initiated to establish their dependence on the material and
Nap03 -
12.8 17.4 process parameters. The experiments were conducted on a two-high, single
stage rolling mill, having work-rolls of 250 m m diameter containing the
Kpo3 5.8
embedded pins or 150 mm diameter for the forward slip measurements. The
Table 1: Heat Transfer Coefficients(Murata et al. [1]) mill is powered by a 25kW DC motor. Its speed is infinitely variable and is
controlled by a rheostat. Roll forces, roll torques, roll speed, entry velocity
and roll gap separation are continuously monitored by a computer-baseddata

Annak of the ClRP Vd. ;rS/l/lM 279


acquisition system. In order to measure the interfacial normal and shear -
These results along with the data given in [12) [14], a compilation of which
stresses the embedded-pin technique is followed, details of which were given is shown in Table 2, allowed an appreciation of the dependence of the heat
in [lo]. For the determination of the heat transfer coefficient thermocouples transfer coefficienton the reduction, entry thickness, roll speed, temperature
embedded in the tail end of the strips were used. The temperature mea- drop in the pass and entry temperature. In the table these parameters are
surementsallowed the estimationof the surfacetemperaturesby extrapolation -
given in columns 1 5, respectively. The time of contact is also shown, see
and these led directly to the heat transfer coefficients. The details of the column6, calculated on the basis of rigid rolls and rigid-plasticstripbehaviour.
temperature measurements were given in [111. The heat transfer coefficientat the roll/strip interface is then calculated for
each case and the appropriate values are given in the last column of Table 2.
RESULTS AND DISCUSSION The technique of computation of that coefficientis of some importanceas the
method itself may influencethe magnitude ofuin asignificantmanner. In the
?7w heuf transjwcoefficien: Low carbon steel slabs, measuring19x5oX200and present case the procedure followed is given below.
38x50~200rrrm, were rolled to different reductions, at different temperatures
and at different roll speeds. Four thermocoupleswere ernbedded in each slab, The coefficientof heat transfer (a)is usually defined in terms of the heat flu,
their locations being shown in Figure 1.
-
that is, the rate of heat flow per unit area (9)and the temperature difference
between the contacting media (7 T,,,,,- T r o t , )as

I 1.8 or zs I : 1 The heat flux is computed from


-ct
<COUPLE #2 THER>IOCOUPLE #4

THERMOCOUPLE#I THERMOCOUPLE #3 I -
with the product of the density and the specific heat taken as
p c 4.9 i %I.! / m 3K .
In theseequations theaveragestrip thicknessisdesignated byli ATstands
for the temperature drop of the strip in the pass and A t shows the time of
contact. The roll temperature is assumed to remain constant at 400C. This
assumptionis reasonable in the experimental environment considered in the
present context where the contact times are fairly short, typically in the order
of 1 second. In an industrial situation however, the temperature rise of the
roll during the pass may be significantly higher and ignoring it may lead to
cumulativeerrors. The average strip temperature during the pass is obtained
Figure 1: The embedded thermocouples from each time-temperature profile. Its definition is shown in Figure 2. The
results given in Table 2 confirm the earlier conclusions regarding the
Typical results of the time-temperatureprofilesobtained are shown in Figures dependenceof the heat transfer coefficienton the process parameters. At this
2 and 3, with the appropriate process parameters also given in each figure. point however, all that may be concluded is that avaries with the reduction,
-
%red. 'w IT(K) Sl(sec) a ( k W / m 2K )
-
19 19 2.5 221.2 599 0.65 24

5.7 19 3 110.6 679.7 0.296 24.9

18.6 19 4 80.3 714.9 0.4 11.9

10 19 4 65.2 6893 0.29 145

20 19 4 130 783 0.42 16.6

20.9 19 4 132 722 0.43 17.8

7 15 4 75 714.5 0.22 17

time. s 21.3 19 10 58 767 0.17 18.5

Figure 2 Time-temperatureprofile during hot rolling of steel 1111 7 15 10 39 752.5 0.09 20.5

20 38 4 152 744 0.59 29.1


-1
rPQrtion - 18.7 Z
10 30.4 4 78 741 0.37 20.2

3
Y 19.7 38 10 64 746 0.23 31.4
L
18.7 38 4 100 666 057 22.3
4, -
Table 2 Heat Transfer CoefficientsObtained on a Laboratory Mill

temperature, entry thickness, speed and - as Murata et al. [l] show - the

6001 I
- predictim
predictim
I
- mowemmt. 13 h t r d
p o u e m m c , I 2 ( t 5 n hlom arfcce)
I I
condition of the surface.The precise nature of the functional relationships is
not yet known. In fact, it appears that the condition of the surface, that is, the
presence of scaling or a lubricant, has a much more significanteffect on the
magnitude of the heat flux than the process parameters. At the beginning of
this study both friction and heat transfer were considered as surface phe-
Figure 3 Time-temperatureprofile during hot rolling of steel [13] nomena and they were both expected to be dependent on the same process
and material variables. A functional relationship, connecting the coefficient
280
of friction to the material's density, rubbingvelocity,the true area of contact Introduction of a lubricant changes the picture somewhat. This is observed in
and the normal load, was presented in [21]and the same approach was hoped Figure 6 where mineral oil mixed with 1% oleic acid, by volume, is used. No
to yield a predictive type relationship for the heat transfer coefficient. Evi- locationofzeroshear ispresentin theroll gap, indicatingthepossibleexistence
dently, this has not materialized and more detailed research, including the of hydrodynamic lubrication [ 17.All of these results have been obtained with
developmentof more data, is required. the embedded pins, a techniquewhich has been criticized[18]. It is undeniable
that interrupting the roll surface changes its mode of deformation, that the
?7ie coefficint offricrin: Results of experiments conducted on various alu- friction of the pins against the holes in which they are located is difficult to
minum alloys and low carbon steel have been presented in [U] [17].Pre- - describe and that the material extrudinginto the clearance between the pins
liminary conclusions regarding the dependence of the coefficient of friction and their holes need to be accounted for. As well, the method is unlikely to
on material parameters were also drawn, indicating that the coefficient is find use in an industrial situation, where roll surfaces must of course remain
strongly dependent on the rubbing speed and the reduction. Yield strength, intact. Even though the friction results given above and in [17]have been
surface anlsotropy and surface hardness also affected the frictional forces. substantiated successfully in several independent ways, other methods to
Typical data, obtained with the embedded transducers, are shown in Figure determine the average coefficient of friction in the roll gap have also been
4, all concerningunlubricated rolling of commercially pure aluminum strips tried. One of these is the non-intrusive way to monitor friction by measuring
at room temperatures [17].The roll pressure and shear stress distributionsin the forward slip of the rolled metal in the pass, defined as
the roll gap during a pass are given and it is observed that all roll pressure
curves are smooth and their maxima are located near the neutral points.

-
&tim
31.92
A 27.42 I where u.is the exit speed of the strip and u,is the surface speed of the roll.
The usual method of measurings, is to mark the roll at two places a known
0 14.22
distance apart, measure the distance between the impressionscreated by the
marks on the strip and use the above formula but replace the velocities by
their corresponding displacements.It is then possible to evaluate the coeffi-
cient of friction from the magnitude of the forward slip. Results obtained by
-
this method are given in Figures 7 9.In Figure 7 forward slip measurements
are plotted versus the reduction per pass for two aluminum alloys, one being
of the commercial purity type (1100-Hl4)while the other is a much harder
alloy, designated as6061-T6.The open trianglesandcirclesindicatethe passes
when both rolls and the strips have been cleaned with n-heptane. The full
squares and triangles and the open squares designate passes in which two
differentlubricantshavebeen ud,oneofwhich isanaphthenicand paraffinic
hydrocarbonwith 7.15 cSt viscosity and the other is a syntheticester of similar
viscosity. Both lubricants contain friction modifiers, antioxidantsand pressure
and antiwear additives. As expected, forward slip increases with reduction
when metal to metal contact is assured. Essentially no change of forward slip
is obselved when the lubricants are used, indicating that the nature of the
boundary has changed significantly.
Figure 4 Roll pressure and friction stress distributions [17]
No negativeforward slip is found however, contradictingthe earlier suspicion
The effect of raising the temperature somewhat,into the warm rolling range, of the possible existence of hydrodynamic lubrication. This aspect is being
results in a significantincrease of the coefficientof friction, see Figure 5 [lq, investigated further at present.
showing the roll pressure and friclioiistress curves obtained during rolling of
aluminum strips.

'
1- I I

-4
-940 I
.m
I
.I .'IS 4 0 '
b lo h I
40
rodions
I-iyre 2: Roll prercure and friction stress distribution:wilrni rolling 1171
Figure 7:Forward slip vs. reduction

u:
U
W
U
0

- io in h
REDUCTION ( X )
io d
Figure 8 Coefficient of friction vs. reduction
Figure 6 Roll pressure and friction stress distribution with a lubricant [lq
281
The coefficient of friction may be calculated from the values of the forward 9. A-Salehi,F.A.R., Firbank,T.C. and Lnnca.ster,P.R., 1973, An Exper-
slip. Several formulae are available in the literature for the computations, imental Determinationof the Roll Pressure Distribution in Cold Rolling,
including Roberts’ relationship [19], giving the frictional coefficient in terms Int.J.Mech.Sci.,15:693-710.
of the forward slip, roll torque, roll separating force, reduction and the roll
radius. A simpler version, given by Ekelund and quoted by Schey [20], is based 10. Karagi0zis.A.N. and LenardJG., 1985,The Effect of Material Properties
on geometry only and is of the form on the Coefficient of Friction in Cold Rolling, Prw. Eurotrib 85, iy2.15.

11. Karagiozis,A.N. and LenardJ.G., 1988, Temperature Distribution in a


Slab During Hot Rolling,J.Eng.Mat. and Techn.,llO17-21.

where @ is the roll gap angle, R is the roll radius and h is the exit thicknessof 12. Pietrzyk,M. and LenardJG., 1989,A Study of Heat Transfer During Flat
the strip. Rolling, Proc. NUMIFORM 89, Eds. Thompson,E.G.,Wood,R.D.,
Zienkiewiq0.C. and Samuelsson,A,343-348.
Results obtained using this equation are shown in Figure 8, where the coef-
ficient of friction is given on the ordinate and the reduction is plotted on the
abscissa. The surface conditions of the two alloys are also indicated on the 13. Karagiozis, AN., 1986, MASC. Thesis, University of New Brunswick,
figure. Two opposite trends are noted. The first indicates that for the harder, Fredericton, Canada.
cleaned metal increasing reduction leads to increasingfriction. For the softer
alloytheincreasefollowsaninitialdip.Anotherfactorisinfluencingtheresults 14. Pietqk,M. and LenardJ.G., 1989, A Study of Thermal-Mechanical
here; the entry thickness of the strip appears to have a significant effect. For Modellingof Hot Flat Rolling, J.Mat.Shaping Techn.,7:117-126.
-
the thicker sample the forward slip coefficient of friction relationship may
not be as accurate as for the lower entry thickness, indicating the increased 15. Lim,LS. and LcnardJ.G., 1984, Study of Friction in Cold Strip Rolling,
effect of the nonhomogeneity of the deformation. Comparing the current ASME, J.Eng.Mat. and Techn.,106:139-146.
resultsto those obtained earlier with the embedded pins allows one to observe
that the general trends of the results are similar, indicating again that if the 16. LenardJG., 1989, Synthetic Lubricants in Cold Strip Rolling, Magyar
variation of the coefficient of friction in the roll gap during rolling is wanted, Kemikusok Lapja, (in press).
thepin-techniquemaybeusedwithconfidence.Aswel1,ifanaveragefrictional
-
coefficient in the roll gap is required and simulation of flat rolling and 17. Lenard,J.G., 1989, Friction in Hot/Cold Rolling, Proc. Eurotrib 89,
comparison to measurements indicate that that is often sufficient, especially Helsinki, 2210-215.
if estimates of the roll force but not the roll torque are needed - the results
obtained from measurements of the forward slip are perfectly satisfactory. 18. Stephenson,D.A.,1983, Friction in Cold Strip Rolling, Wear, 92:293-311.

CONCLUSIONS 19. Roberts,W.L, 1978, Cold Rolling of Steel, New York,Dekker.


The heat transfer coefficient and the coefficient of friction at the roll/strip
20. Schey,J.A., 1983, Tribology in Metalworking - friction, lubrication and
contact surface during flat rolling were examined. Published values of the
wear, ASM, Metals Park, Ohio.
coefficientswere reviewed and results obtained in a laboratory setting were
presented. Both coefficientsdescribe surface phenomena and as such, they 21. Len;lrd,J.G., 1990, Measurements of Friction in Cold Flat Rolling,
were expected to be dependent on the same processand material parameters, J.Mat.Shaping Techn., (submitted).
including the reduction, rubbing velocity, temperature, real contact area and
the presence of scales or a lubricant. This was found to be true as far as the
frictionalforceswere concerned.The heat transfer coefficientat the roll/strip Acknowledgements
interface was much more dependent on the condition of the surface than on
the other parameters, however. The financial assistance of the Natural Sciences and Engineering Research
Council of Canada is gratefully acknowledged.
REFERENCES

1. Murata,K., Morise,H., MitsutsukM., Naito,H., Komatsu,T. and Shi-


da,S., 1984, Heat Transfer Between Metalsin Contact and its Application
to Protection of Rolls, Trans. ISLJ,XB-309.

2 Stevens,P.G., 1vemK.P. and Harper,PJ., 1971, Increasing Work-Roll


Life by Improved Roll Cooling Practice, J.Iron and Steel Inst., 209:l-11.

3. PreisendaqH., Schuler,P. and Koschnitzke,K., 1967, Calculation of


Temperature Variations at the Surface of Hot Ingots During Contact
with Rolls, Arch. Eisenhuttenwes., 38205-213.

4. Silvonen,A., Malinen,A. and Korhonen,AS., 1987, A Finite Element


Study of Plane Strain Hot Rolling, Scand. J. Metal., 16103-108.

5. Bryant,G.F. and Chiu,T.S.L, 1982,Simplified Roll-Temperature Model:


Convective Cooling,Metals Techn., 9478-484.

6. Harding,R.A, 1976, Ph.D. Dissertation, University of Sheffield, Shef-


field, UK.

I. Pietnyk,M. and Lenard,J.G., 1988, Experimental Substantiation of


Modelling Heat Transfer in Hot Flat Rolling, Proc. 25th Nat. Heat
Transfer Conf., Ed. H.RJacobs, 347-53.

8. vanRooyen,G., and Backofen,W.A., 1957,Frictionincold Rolling,J.Iron


and Steel Inst.. 186235-244.

282

You might also like