Home Search Collections Journals About Contact us My IOPscience

Modelling heat transfer during hot rolling of steel strip

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

1995 Modelling Simul. Mater. Sci. Eng. 3 437

(http://iopscience.iop.org/0965-0393/3/4/002)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 210.212.58.168
The article was downloaded on 15/08/2010 at 18:42

Please note that terms and conditions apply.

Modelling Simul. Mater. Sei. Eng. 3 (1995) 437453. Printed in the L K

Modelling heat transfer during hot rolling of steel strip

Rafael Col&
Facultad de Ingenieria Mednica y Elktica, Universidad Aut6noma de Nuevo M u ,
AP 149-F, 66451 San Nicolss de 10s Gar% NL. Mexico

Received 28 July 1994, accepted for publication 21 March 1995

Abstract The heat transfer conditions encountered d-g hot rolling of steel ship are modelled;
it is assumed thaI the strip loses heat by radiation and mnveciion to k atmosphere, conduction
to the work rolls, forced convection and boiling of the water employed in descaling, whereas
it gains heal during deformation. Conduction within the stock is calculated by mens of a
two-dimensional explicit finite-difference algorithm. The simulation is conducted assuming the
existence of an isolating oxide layer, which controls the conduction to the work rolls during
the reduction. Results from the simulations compare well with data obtained during actual
production in an indushial mill.

1. Introduction
The study of hot rolling of steel strip has been subjected to renewed scrutiny in recent
years due to the need for better control over microstructure and mechanical properties.
Relationships between process variables (like reduction, temperature and speed per pass,
interpass time, etc). and chemical composition and metallurgical phenomena occurring in
the steel (such as grain growth, precipitation, recovery and recrystallization) have to be
drawn in order to achieve the more stringent requirements from consumers.
Efforts to predict the thermal and microstructural evolution of steel strip during hot
rolling have been based on mathematical models which involve a number of assumptions
made to simplify such a complex process [I-IO]. Early attempts in modelling [l, 21 were
concentrated in describing the heat transfer phenomena occurring during rolling, whilst more
recent ones [3-101 include the evolution of microstructural features which result from the
thermal history of the process. One of the early studies [l] showed that the thermal history
of a piece depends on a number of different factors which have to be evaluated if sensible
predictions are sought: among these factors, the contribution from radiation and convection
to the atmosphere, conduction to the work rolls and adiabatic heating due to deformation
are of paramount importance.
The aim of this work is to present and discuss the way into which different heat transfer
phenomena, occuning during hot rolling of steel strip in a six-stand continuous mill, are
incorporated into a two-dimensional finite-difference model. Predictions by the model are
then compared with actual measurements obtained during rolling strips of different thickness,
width and chemical composition.

2. Modelling

Rolling of ship is a three-dimensional heat transfer problem, but since the length of the piece
is much greater than either its width or thickness, modelling of this particular process can

0965-0393/95/040437+17$19.50 @ 1995 IOP Publishing Ltd 431

438 R Col&

(a) Conduction.
(b) Radiation and convection.
( c ) Uater cooling.
(d) Rdiabatic heating.
-
Rolling direction

Figure 1. Heat transfer mechanisms in hot rolling.

be reduced to the phenomena which take place along the smaller dimensions. A schematic
diagram, showing the phenomena which are to be taken into account in modelling the
thermal evolution, is presented in figure 1; the phenomena which are considered are the
following.
(i) Heat losses due to radiation and convection to the atmosphere.
(ii) Heat conduction to the work roll.
(iii) Heat losses due to boiling and forced convection of water employed in descaling
the surface of the strip.
(iv) Adiabatic heating produced by deformation.
(v) Heat conduction within the stock.
The two-dimensional explicit finite-difference model described elsewhere [Z,3,111 is
used in this work. The cross-sectional area of the strip is divided into small elements
of equal volume in order to approximate the equations for conduction; due to geometry,
computations are carried out only in one quadrant after assuming equal conditions on top and
bottom, as well as on lateral surfaces, figure 2.This assumption was made after prelimimy
calculations considering scale growth on top and reradiation from tables on the bottom
showed boundary conditions to be similar on both surfaces. The subscripts of the heat flow
(Q) in this last figure represent heat flowing in different directions, i.e. Ql and Q3 are the
through-thickness heat flow from and into a given node, Qz and Q4. are the equivalent
ones, but through the width of the sfrip; QSand Qs represent the heat flow as result of the
boundary conditions at the surface (either top or bottom) and the lateral surface.
A schematic diagrain of the continuous-strip mill (which produces around one million
tonnes of steel strip per year) used to validate the model is shown in figure 3. Such validation
was made with the aid of the readings from the two pyrometers, located at the entry and
exit of the mill. Descaling of top and bottom surfaces is produced by high-pressure water
jets, As the thickness of the stock is reduced in height (by any of the six stands) its
speed is increased, and, in order to synchronize the rotating velocities of the work rolls
the electromechanical devices known as loopers are installed between stands. It is worth
mentioning that in an industrial mill, like the one simulated in this work, the word ‘stand’
refers to the equipment employed to carry out the deformation, whereas ‘pass’ is used when
the reduction in height is being given.
 Heat transfer during hot rolling of steel strip 439

E
I ZERO HEAT FLOU
g-..-
3
D
-I
LL CI
3
I-
a
w J
I Q
LT
0 w
LT 03
w
N

' 05 05 05
j
SURFACE

Figure Z Grid used to compute the heat Row [Z].

PE
PS
Entry PUrOIIOIPr.
Exit purommer.
Metal flnu
Fl...F6
Ll...LS
- Stands.
Loopsr..
DE Onscaling.
OR Din& rolls.

Figure 3. Diagram of the siu-stand continuous miU.

The thermal evolutions at the different positions of the cross-sectional area, figure 2, are
computed assuming that the boundary conditions, and therefore the heat transfer coefficients,
change as the stock moves through the continuous mill, figure 3. For instance, simulation
of the front end of a strip involves the computation of the cooling due to convection and
radiation (i) between passes and from the entry pyrometer to the impingement of the water
jets, where phenomenon (iii) has to be taken into account; conduction to the work rolls (iv)
is accounted for during the time strip and roll are in contact, which depends on the rolling
conditions within the pass.
Numerical solution of the equation for conduction:

where x and y are the width and thickness coordinates, T the temperature, f the time, U the
temperature dependent thermal diffusivity and Qefrepresents the amount of heat gained
by the material during deformation. The finitedifference method employed for solving
equation (1) is explicit, since the future temperature of a given node is given in terms of
the current temperatures of the nodes and its neighbours, and its solution depends on the
correct choice of the computational time interval ( a t ) [2-4,12,13]. If this condition is not
440 R Colds

satisfied an accumulation and amplification error would influence the results. The condition
for stability employed in the present model is that the temperature gradient reaches no
further than the inside edge at the corner element, which yields 131

6t =
-
dy'(n i)dx
+
2na(dx d y )

where n is the number of elements into which half the thickness was divided, dx and dy are
the sizes of the comer element as they are measured along the width and thickness of the
strip.
The cooling rate at the surface of the strip increases as it undergoes descaling
[2,3,11,14] or deformation [2,3], due to the higher transfer coefficients which result from
forced convection and boiling of water and conduction to work rolls respectively, and, in
order to avoid mathematical instabilities, the number of elements through the thickness is
increased threefold.
Another feature incorporated in the model is the adiabatic increase in temperature
produced by deformation, phenomenon (iv) and Qkf in equation (l), which is calculated
assuming that the deformation is uniform through the arc of contact, by [15]

where AT is the increase of temperature, p the density and c the specific heat of the
steel. The integral represents the energy required to deform the material from €0 to cf. The
method employed to compute this increase is by dividing the arc of contact into 25 angular
segments, in order to cany out five iterations per pass; the stress-strain curve 1161 for each
iteration is then calculated assuming that both temperature and strain rate are constant, since
the temperature and strain rate changes will only be accounted for once the time interval is
finished.
Heat flow during air cooling (ha) is assumed to be given by

h, = -6746 + 21.2Ts + 4.76 x lO-'(T, + 273)' (4)

where the heat flow is expressed in W m-', and is the surface temperature, given in "C.
It should be mentioned that this last equation was drawn from experimental trials [2], in
which the emissivity of the steel was found to be equal to 0.84, and has proven its validity
on different occasions 12-5,9-11,13,17, IS]. The reason to neglect the ambient temperature
: when T.. =- To, and the error introduced is no
(To) in the radiation term is that T t >> T
more than 0.8% even at temperatures as low as 750°C [19].
It is considered that the water from the jets employed in descaling the surface of the
steel produces a heat flow (h,) given by

hw = HI(Ts - Tw) (5)

where Tw is the temperature of the water and HIis a heat transfer coefficient associated
with radiation and convection through a steam layer [17,18], Heat extracted during actual
impingement of the jets is calculated by equation (3, but with a higher heat transfer
coefficient (Hz), which is assumed to be due to convection and boiling of water [17,18].
It is worth mentioning that different portions of the strip are subjected to either the higher
heat extraction rate or to both rates, since the front end of the strip will only be subjected
 Heat transfer during hot rolling of steel strip 441

to the higher rate, since the water will only puddle in the region between the point of
impingement of the jets and the pinch rolls used to feed the strip into the mill, see figure 3.
The assumption that the cooling is symmetric is sWI valid, since the water employed to
descale the bottom surface of the ship also cools down the rolls which transport the strip.
The model incorporates the growth of the scale assuming the parabolic rate given by
[20-22]

Ax = kP& (6)
where Ax is the scale thickness at the time t, k, is the parabolic growth coefficient, which
depends on the type of oxide and temperature. The use of adequate values of kp [21,22]
yields a very rapid initial growth on bare surfaces which decreases as the oxide builds
up. The proportion of the different species within the layer is kept within reported ranges
[21-23].
It is also assumed that the scale deforms to the same extent as the strip during rolling
[24], since it is made mainly of wustite and magnetite which are fairly plastic at the
temperatures involved in hot rolling [25,26]. This is important, since the temperature
which should be used in (4) and (5) to calculate the heat flow is that at the surface of the
oxide layer, rather than the one of the metal. Heal flow through the oxide film (ho) can be
calculated by 131

where k, and d, are respectively the thermal conductivity and the thickness of the oxide
layer, and T, and To are the temperatures at the surface of the strip and oxide respectively.
The oxide is assumed to have zero heat capacity.
The heat extracted by the work rolls during the contact with the strip is calculated
assuming conduction through the oxide layer; in this case, the energy per unit of area
flowing from the metal to the work roll (Ar) will be given by [27,28]

where E is the temperature at the surface of the roll, p is related to the resistance offered
by the oxide layer to conduct the heat and is given by

where a, and cr, are the thermal diffusives for the strip and roll, and k,, ko and k, are the
e
thermal conductivities for strip, oxide and roll. in equation (8), which depends on p and
the time of contact between roll and strip (At) is given by:

Heat transfer during the actual rolling pass is calculated by dividing the roll bite angle (e),
given by
442 R Colds

...
.....
....
....
....
.....
.. .. ..
,.....
. .. .. ... I . . .

....
.,. ... .... ,,..
i : : :
I

.. .. ., ..
i i : :

.,. ,... .... ....

. .
. .
_________
0.5.h.
J._____________
;
i-;-i-:-:-:
; : :
_____ .
0.5.h'
_ _ _ _??
_ I - .q
;h_i_ _-.-.
..-.
_ .

Rolling direction

Figure 4. Diagram of the roll bite.

where Ah = ho - h f is the draft, ho and hr are the entry and exit thickness of the strip,
and R the radius of the work roll, into five angular sectors, figure 4; At in equation (10)
is calculated by dividing the corresponding angular sector by the speed of the roll. The
boundary conditions at the roll and s ~ surfaces,
p given by equation (S), are applied in the
grid shown in figure 2, in order to calculate the cooling of the strip. or in a similar one, for
accounting for the heat into the roll.
The model assumes that the boundary conditions at the lateral surface of the strip are
those for air, except for the time between the pinch rolls and descaler, where it is assumed
that cooling will be due to water falling from the top surface.

3. Results

Figures 5 and 6 show, respectively, the thermal evolution for the front and back ends of
an outgoing 1.96 mm thick and 0.94 m wide strip produced from a 27.9 mm thick transfer
bar. The points indicated in both figures correspond to actual measurements obtained by the
entry and exit pyrometers of the industrial mill, see figure 3, the continuous lines correspond
to the positions within the cross-section of the stock as they are indicated in the insets. Both
figures show how the surface of the strip cools down as a result of the water employed
in descaling and due to contact with the work rolls; conduction is fast enough to reduce
the temperature differences thmugh the thickness, but not along the width of the strip.
Figure 7 shows the thermal evolution for the front end of an outgoing 6.35 mm thick and
0.97 m wide strip produced from a 27.9 mm transfer bar. Measurements obtained during
actual production are indicated, as with the previous figures. The lower overall cooling
rate observed in the thicker strip results from the difference in the reduction schedule for
different final thickness, which also affects the rolling time.
Figure 8 shows the simulation of the oxide layer evolution for the conditionsencountered
by the front end of the 1.96 mm thick strip, see figure 5. The initial growth from zero comes
from the assumption that descaling removes the whole layer; the six reductions in thickness
shown in this figure are equivalent to those given to the steel [24]. The figure shows the
growth of the three different species (wustite, magnetite and haematite), which are calculated
from data reported elsewhere [21,22]. Considering that the scale is made exclusively of
 Heat transfer during hot rolling of steel strip 443

1050

1000

950

900

850

800

750 m nearursnmts

700
0 5 10 15 20 25 30
Time (4
Figure 5. Temperahlrr evolution for the front end of 1.96 mm thick strip

700
650
t' I
0 5 10 15 20 25 30
Time (€4
Figure 6. Temperahlre evolution for the back end of 1.96 m Lhick seip

wustite is justified after observing that the predictions by the present model do not differ
from those made with a much more complex model in which conduction through the three
different layers is computed [29].
Thermal properties of strip, work roll and oxide layer (wustite), table 1 [13,27-321 were
used to calculate the heat transfer and thermal gradients which result from the strip and work
roll contact. Figures 9 and 10 show the heat flow to the work roll (hr), as calculated by
444 R Colds

1050

1000

950

!j 900

'
4
$ 850

& 800

750
1

Figure 7. Temperahlre evolution for the front end of 6.35 mm thick strip.

20.0
C0"lrp
15.0

10.0 -
5.0-
Y

j
0.0
0.0 5.0 10.0 15.0 20.0 25.0 30.0
20.0
Edge

15.0-
H
10.0 -
5.0 -
0.0

Figure 8. Evolution of ule scale layer during rolling: the thicknesses of the h e oxide species
are plotted.

equation (8) in each of the six passes, with the temperame at the surface of the strip and
the oxide thickness shown in figures 5 and 7. It is worth noticing that the heat flow in each
pass can be extrapolated to the same initial flow at zero reduction in height, i.e. that at entry
 Hear transfer during hot roiling of steel strip 445
Table 1. Thermophysid properties of the different materials [ 13.28-321

Thermal conductivity Density Specific heal

gV m-j K-j) (kg m-3) (I kg-' K-l)
Austenite 16.5 t 0.011 x T 8050 - 0.5 x T 587.8 + 0.068 x T
Wustite 3.2 7750 775
Work roll 18.46 7400 470

point, figure 10, to a value equal to 14.8 Mw m-', which agrees with previously reported
results [4,33-361.

12.0.

"a 0.0.
2
v
8.0.
d
6.0 .

7.U
0.0 5.0 10.0 15.0 20.0 25.0 30.0
Tbjckness (mm)
Figure 9. Heat flow to the work rolls as a function of the thickness of the stock

The temperature distribution within the cross-sectional area can be obtained at any point
of the simulation; for instance, figure 11 shows the temperature distribution within the front
end of the strip as it exits the descaler. It is assumed that the heat lost by the strip during the
contact time with the roll is gained by the latter: figures 12 to 15 show the corresponding
temperature distributions within the strip, figures 12 and 14, and within the work roll,
figures 13 and 15. Figures 12 and 13 show the distribution after the strip leaves the first
stand, whereas figures 14 and 15 correspond to the exit side of the last stand.

4. Discussion

The variation of heat flow shown in figures 9 and 10 results from the decrease of the
temperature difference at the oxide-work roll interphase as straining progresses, figure 16.
It can be deduced from equation (8) that the rate of heat flowing from the strip to the work
rolls increases as the thickness of the oxide layer decreases, but this rate also decreases as
the temperature difference between both surfaces in contact diminishes [27-29]. It should
be mentioned that the data shown in figure 16 were calculated after assuming that the work
446 R Colds
16.0
Pass
* 1
.2 -
.

6.0 -

4.0
0.0 2.5 5.0 7.5 10.0 12.5 15.0
Ah 1"(
Figure 10. Heat Eow to the work rolls in each of the six passes

Figure 11. Temperature distribution within the front end of the stock as it exits fhe descaler,

rolls were at 30°C. and with the thermophysical properties of the different materials as
shown in table 1.
The effect of the oxide layer has been either ignored by some authors [34,37,38], or
incorporated as a unique node made exclusively of wustite [14,10,11]. The approach
followed in this particular work is the latter one (although the thickness of the layer is
calculated as it is formed by the three oxide species), especially after finding that modelling
 Hear transfer during hot rolling of steel strip 447

0.0

.-
C
c

Figure 12. Tempemure distribution within the front end of the stock as it exits the hnt stand.

Figure 13.,Tempature distribution within the work roll as the stock exits the fist stand

the oxide layer as a unique node agrees with the results obtained with a much more complex
model [29], which divides the scale into 40 nodes, of which 36 have the properties of wustite,
three of magnetite and one of haematite.
448 R Colds

Figure 14. Tempemme distribution within the front end of the stock as it exits the last stand.

Figure 15. Temperature distribution within the work roll as the stock exils lhe last stand,

Although the deformation withii the roll gap is highly heterogeneous,the present model
assumes that it is uniform, since the strain gradient becomes important only when the
projected arc of contact divided by the average thickness of the strip (Llh,) is less than
 Heat transfer during hot rolling of steel strip 449

650
600

300 1
250
0.0
' 2.5 5.0 7.5 10.0 12.5
I
15.0
Lul (")
Figure 16. Temperaturedifferenceat the oxide-work roll interface as a function of the reduction
in height in each of the six passes.

one 1431. This ratio is calculated by

L l R h h
-=
h, {hohr
and, as can be seen in table 2, it is always higher than one.

Table X Conditions for descding and rolling for the strip shown in figure 5. h, thickness in the
delivery side. To temperature in the entry side. Td temperature in the delivery side. A T , T, - Td.
U, average speed. At. contact time. L/hm. protected arc of contact over mean Ihickness.

h T, T* AT U AI
Stage (mm) ('C) (k) ("C) (ms-I) L/hm (s)
Descaling 27.9 987 699 288 1.00 - 1.0 x 10-1
1st stand 129 954 858 96 0.69 3.10 8.5 x
2nd sland 7.6 942 878 64 1.28 3.53 2.8 x 10-2
3rd stand 4.5 922 873 49 2.19 4.57 1.2 x 10-2
4th stand 3.2 905 871 34 3.41 4.56 5.0 IO-'
5th stand 2.4 890 863 27 4.78 4.89 2.8 x IO-'
6thstand 2.0 877 859 18 6.24 4.38 1.6 x IO-'

The model was validated using actual measurements from the production line, as
mentioned when figures 5 to 7 were first introduced. Figure 17 shows the correlation
between the temperatures predicted by the model and those measured in the line, and, as
can be seen, all the data fall within *5% (the maximum temperature difference between
any measurement and its prediction was 23°C). The predictions shown in this last figure
are those for the middle of the top surface at three different positions along the strip (front,
middle and back), since the pyrometers are fixed to the central line of the mill. It is worth
450 R Colds

Figure 17. Comparison W e e n predicted and measured temperatures.

mentioning that the data shown in figure 15 were obtained during the production of strips
ranging from 0.63 to 1.09 m in width and 1.88 to 6.35 mm in thickness.
The magnitude of the depth of the surface cooling produced by the contact with the
work rolls, figures 12 and 14, depends on the time of contact, which decreases as the speed
of the strip increases, table 2 and figure 18. This is due to the fact that the variation of
the heat flow to the work rolls, figures 9 and 10, remains within the same range in all the
stands. It is worth noticing the increase in temperature of the work rolls, figures 13 and 15,
which should be controlled since the magnitude of the thermal gradients developed in the
work rolls might affect the dimensional quality shape of the strip [39,40].
Once the thermal behaviour of the process is well known, it is possible to expand the
model towards the prediction of microstructural evolution [9,10,41], in order to predict
the strength of the material being rolled 1161. The model assumes sticking friction during
rolling [42], in order to predict the separation forces in the six stands, figure 19, and the
pressure to which the work roll is subjected in the arc of contact, figure 20. With this
information, and that of the thermal distortion of the roll, it is possible to predict the profile
and shape evolution of the ship as it is being rolled [40].

5. Conclusions

A mathematical model used for simulating hot rolling of low-carbon steel strip was
developed. The model takes into account the cooling produced by convection and radiation
to the atmosphere, conduction to the work rolls and forced convection and boiling of the
water employed in descahg; the adiabatic heating due to deformation is the only heating
effect taken into account. It is assumed that the strip is covered by an isolating oxide layer
with the properties of wustite.
The correlation between measurements conducted on a production mill and-the
predictions by the model is good enough to validate the assumptions made about the heat
transfer mechanisms and coefficients.
 110.0
-

90.0 -

c 70.0 -
c
a 50.0 -

30.0 - +
+
10.0

Figure 19. Comparison between predicted and measured separation forces in lhe 1st Sir s t a d s .

Thermal modelling should be the first step in the developing of a full hot-rolling
simulation program able to predict strength, profile and shape of the strip.
452 R Colds

8.0

Figure 20. pressure exerted by the steel as it is deformed in the last stand,

Acknowledgments

The author acknowledges the support given by CONACYT, and the facilities provided by
Hysla, SA de CV.

References

[I] Hollander F 1970 Mathematical Models for Metallurgicnl Process Deselopment (London: Lmn and Steel
Institute) p 48
[2] W i n g R A 1976 PhL3 Thesis UNverrity of Sheffield
131 Leduc L A 1980 PhD Thesis University of Sheffield
141 Sellm C M 1985 Marer. Sci. Technoi. 1325
[SI Sellan C M 1990 Mdhemnlicol Modding @Hot Rolling of Steel ed S Yue (Montreal: CUI) p 1
161 Hollander F 1990 Mathematical Modelling of Hot Rolling of Steel ed S Yue (Montreal: C M ) p I9
[7] Choquet P, Fabr6gue P, Giusti J, Chamont B, Pemt I N and Blanchet F 1990 Malhemticd Modelling of
Hot Rolling ofsteel ed S Yue (Montreal: CIM) p 34
181 Hodgson P D and Gibbs R K 1990 Mathematical Modeltinx of Hot Rolling of Steel ed S Yue (Montreal:
CIM) p 76
191 Hinojosa M, Ortiz U. Leduc Land Col& R 1993 Mater. Sci. F o r m 1P-115 323
[IO] Col& R 1993 EDP Congress ed J P Hager (Warrendale. PA TMS-AIME) p 671
[I 11 Col& R and Leduc L A 1993 Mechnnical Working mdSteel Pmcesshg XXY i.Warrendae, P A ISS-AIME)
P 13
[I21 Camahan B. Luther H A and W a c s I O 1969 AppliedNwnerieal Milhods (Landon: Wiley)
[I31 Sellm C M and Whileman I A 1981 Mer. Technol. 8 IO
[I41 Sigalla A 1957 J. Imn Slecl h t . 186 90
rl9 Col& R and Sellam C M 1987 J. Tesl. EvaL 15 342
i14 Col& R 1990 Mater. Forum 14 253
.~
1171 Col& R and Sellars C M 1988Accelerated Coolina o f f i l l e d Steel ed G E Ruddle and A F Crawlev mew
~

York Pergamon) p 121

 Heat transfer during hot rolling of steel strip 453

[IS] Col& R 1994 Mmfacturing Science and Engineering (PED 68.2) vol 2, ed W E Alzheimer (New York:
ASMB) p 61 1
[I91 Buckley G W, Lewis M and Maddison R T 1978 Met, Tcchnol. 5 228
[20] Wagner C 1951 Atom Mwemem (New York ASM6)
[21] Paldassi J 1957 Rev. Mdld 54 2
[22] Lorang F 1961 Rev. Univ. Mines 17514
[23] Darken L S and Guny R W 1946 J. Am. Chem Sac. 68 798
p4] Wtsmotie S, Mignon 1, Economopoulos M and Thomas G 1973 C. R M. 36 35
p5] Crouch A G 1972 J. A m Cerom. Soc. 55 558
[26] Mitchell T E, Voss D A and Buther E P 1982 J. Marer. Sci.'l7 1825
. . Perez A. Fuentes R and Col& R 1993 34th Mechanical Warkiw and Steel Pmcessin~
1271 . Canf . (Warrendale.
P A ISS-AIME) p 191
. Guemem M P, P&ez A and Col& R 1994 7thlnt. a m p . on Transport Phenomena in Manufocturinn Pmesser
1281
.
(Universidad Nacianal Aut6nama de Mexico) (Mexico: Pacific Center of ' T h e d - i u i d s Engineering)
p IO8
[29] Torres M and Col& R 1994 Manufacturing Science and Engineering (PED 68-2) vol 2, ed W E Alzheimer
(New York: ASME) p 577
[30] Samonov G V (ed) 1973 The Oxide Handbook (New York Plenum)
[31] Taylor R. Fowler C M and Rolls R 1980 lnt. J. Thernwphys. 1 225
[32] Slowik J, Borchard G. Kohler C. Jeschar R and Scholz R 1990 Steel Res. 7 302
[33] Stevens P G, Ivens K P and Harper P 1971 J. Iran Steel lmf 209 1
[34] Samarasekera I 1990 Mathematical Modelling of Hot Rolling ofsteel ed S Yue (Montreal: C M )p 148
[35] Anjan Kumar M K and Mehrotra S P 1992 Iran Making Steel Making 19 55
1361 Sellars C M and Kawai R 1993 Modeling of Metal Rolling Processes (London: Institute of Materials) p 648
1371 De"& C and Samarasekera I 1986 Iran Making Steel Making 6 311
[38] Hafgen H. Zauhar G, Birnstock F and Bathelt J 19- 4th lnt Steel Rolling Canj: Science ond Tcehnology of
Fk#Rolling (Deauville: IRSID) 8.2.1.
1391 Sheppard T and Roberts J M 1973 lnt. Merull. Rev. IS 1
[40] Zambrano P C and Col& R I993 Modelling of Meml Rolling Processer (Landon: Institute of Materials)
P 502
[41] Col& R 1994 Adwnces in Hat Deformation Tarures and Microstructures ed J J Jonas, T R Bieler and K I
Bow- (Warrendale, P A TMS-AIME) p 63
[42] Sims R B 1954 P m . lmr. Mech. Eng. 168 191
1431 Beynon I H and Sellars C M 1992 Madelling of Plnsrlc Deformtion and Its Engineering Applications ed
S I Anderson et01 (Ris0: Ris0 National Laboratory) p 13