Microstructural Engineering Applied to

the Controlled Cooling of Steel Wire Rod:

Part I. Experimental Design and Heat Transfer
P.C. CAMPBELL, E.B. HAWBOLT, and J.K. BRIMACOMBE
The goal of this study was to develop a mathematical model which incorporates heat flow,
phase transformation kinetics, and property-structure-composition relationships to predict the
mechanical properties of steel rod being control cooled under industrial conditions. Thus, the
principles of microstructural engineering have been brought to bear on this interdisciplinary
problem by combining computer modeling with laboratory measurements of heat flow, austenite
decomposition kinetics, microstructure and mechanical properties, and industrial trials to determine heat transfer and obtain rod samples under known conditions. Owing to the length and
diversity of the study, it is reported in three p a r t s , 118'191 the first of which is concerned with the
heat flow measurements. A relatively simple and reliable technique, involving a preheated steel
rod instrumented with a thermocouple secured at its centerline, has been devised to determine
the cooling rate in different regions of the moving bed of rod loops on an operating Stelmor
line. The measured thermal response of the rod has been analyzed by two transient conduction
models (lumped and distributed parameter, respectively) to yield overall heat-transfer coefficients for radiation and convection. The adequacy of the technique has been checked by cooling
instrumented rods under well-defined, air crossflow conditions in the laboratory and comparing
measured heat-transfer coefficients to values predicted from well-established equations. The
industrial thermal measurements have permitted the characterization of a coefficient to account
for radiative interaction among adjacent rod loops near the edge and at the center of the bed.
I.

INTRODUCTION

M I C R O S T R U C T U R A L engineering is an interdisciplinary approach to the quantitative prediction of the thermal, microstructural, and mechanical property evolution
of a metal subjected to a given thermomechanical process. Recent demands on the metals industry to improve
product quality and performance, while at the same time
reducing cost, have spurred the development of this
methodology. The root of the microstructural engineering approach is imbedded in the mathematical model,
which links the basic principles of heat and mass transfer
and microstructural phenomena to the operating process.
In addition, both laboratory experiments and industrial
trials are necessary to obtain empirical and semiempirical relationships characterizing transport phenomena by which the model can be tuned to operating
variables.
In the present study, microstructural engineering has
been applied to the Stelmor cooling of steel wire rod, Ill
a process situated after the finishing stand of a rod mill
which provides controlled cooling of the steel through
the temperature range of austenite decomposition. The
process was developed to replace lead patenting, which

P.C. CAMPBELL, formerly Graduate Student, The University of

British Columbia, is with BHP Central Research Laboratories,
Wallsend, New South Wales 2287, Australia. E.B. HAWBOLT,
Professor, Department of Metals and Materials Engineering and The
Centre for Metallurgical Process Engineering, and J.K.
BRIMACOMBE, Stelco/NSERC Professor and Director, The Centre
for Metallurgical Process Engineering, are with the University of British
Columbia, Vancouver, BC V6T 1Z4, Canada.
Manuscript submitted February 14, 1990.
METALLURGICAL TRANSACTIONS A

utilized a molten lead bath to impart controlled thermal

changes and desired properties to wire rod. In the Stelmor
line, forced air is the cooling medium, but more recently, other processes have exploited water and molten
salt baths to develop desired cooling characteristics, t2,3,aJ
Nonetheless, since its development nearly 25 years ago,
the Stelmor process has become the most popular patenting technique in the world. In 1982, there were 69
mills with 153 Stelmor lines operating in 26 countries, tSJ
Global capacity for the production of wire rod through
this process has been estimated to be 21 million tonnes
per year. [2]
II.

PROCESS DESCRIPTION

In the Stelmor process, rods exiting the last stand of

the rod finishing mill travel through an intermediate zone
of water cooling boxes prior to arriving at the laying head
(Figure 1). The water boxes provide control over rod
temperature prior to continuous cooling, thus affecting
prior austenite grain size, while the high-velocity water
jets remove surface scale. At the laying head, the rod is
looped continuously into coils and placed on the line where
the chain conveyor, seen in Figure 1, pulls them through
the successive cooling zones. Air is forced up from
below the loops by a series of fans in zones to effect
control of the rate of rod cooling. For lower carbon grades,
where a maximum fraction of proeutectoid ferrite is desired, slow cooling rates, and thus, high transformation
temperatures, can be achieved. For higher carbon grades,
where a fine pearlite microstructure is desired, maximum cooling rates are employed. Typically, steel arrives
at the Stelmor laying head between 840 ~ and 940 ~
is cooled through the austenite-ferrite and austenite-pearlite
VOLUME 22A, NOVEMBER 1991--2769

IV.
B

Cc

(E
ip

PREVIOUS W O R K - - H E A T

Heat flow in the long rods processed on the Stelmor

line is essentially one-dimensional, governed by the following transient heat conduction equation:
O(OT]
kOT
-- k
+---+
Or \ O r /
r Or

D
Fig. 1--Schematic diagram of the Stelmor line: (a) delivery pipe and
water boxes, (b) laying head, (c) conveyor, (d) plenum chambers,
and (e) coil forming chamber.

transformation temperatures (770 ~ to 600 ~

and exits
the line at a temperature suitable for handling ( - 5 0 0 ~

Ill.

SCOPE AND OBJECTIVES

In the present study, a mathematical model has been

developed to enable prediction of the thermal history,
microstructural evolution, and mechanical properties of
steel rod cooled by the Stelmor process. In addition, a
series of experiments has been conducted in the laboratory at The University of British Columbia (UBC) as
well as on an operating Stelmor line at the Stelco Hilton
Works No. 2 Rod Mill in Hamilton, ON, Canada. The
experiments were performed to obtain data on heat transfer from cooling rods, microstructural evolution, and on
microstmcture-composition-property relationships in order
to augment existing data in the literature. Part I of this
three-part article includes experimental design and results pertaining directly to the heat-transfer aspects of the
project. Investigation of the microstructure evolved in
continuously cooled rod and correlations developed to
link steel composition, microstructure, and mechanical
properties are presented in Part II. uS] Formulation, validation, and predictions of the mathematical model are
given in Part III. u91
Turning specifically to Part I, it was realized at the
outset that within the timeframe of this project, heat
transfer on a Stelmor line would be too complex to predict accurately from first principles. Both the forced convection of air through the array of rod loops and the
radiative interchange among them could not be characterized easily from existing correlations. Thus, a reliable
technique was sought to measure the cooling conditions
in the relatively hostile environment of a Stelmor line.
This was accomplished by instrumenting lengths of steel
rod with thermocouples at the centerline and measuring
the thermal response. The temperature-time results were
then utilized to calculate heat-transfer coefficients as a
function of process parameters. The method was checked
first in the laboratory under well-defined air cross flow
conditions. Correlations for describing the heat-transfer
coefficients as a function of the process variables were
subsequently incorporated into the mathematical model.
It should be stressed that the instrumented rod tests in
both the laboratory and plant also provided vital steel
samples for microstructural examination and mechanical
property evaluation.
2770--VOLUME 22A, NOVEMBER 1991

TRANSFER

OT
qre = PCp

-~t

[1]

where qrR is the heat released due to the austenite decomposition reactions. This equation can be solved utilizing numerical techniques, as outlined in Part III of this
article. 09/An important aspect of the solution of Eq. [1]
is the characterization of the boundary condition at the
rod surface, which can be written as
r = ro

OT
-k-= h o v ( T s - Ta)
Or

[2]

where hov represents the overall heat-transfer coefficient. Although the Stelmor process employs forced air
to cool the steel rods, radiation from the rod surface also
contributes to the removal of heat. As a result, the overall heat-transfer coefficient must be linked to the combined effects of convection and radiation, or in equation
form:
hov = hc +

hR

[3]

Radiation heat losses from a cooling rod can be quantified by the following equation:
he = irE \ Ts

Ta /

[4]

where F is a radiation factor which accounts for the emissivity and relative geometries of the cooling body and
its surroundings while temperatures are absolute (K).
Assuming that the rod is capable of radiating unhindered
to a black body at ambient temperature, F simply reduces to e, the emissivity of the steel ( = 0 . 8 for an oxidized surfacet6]), which allows ready solution of
Eq. [4]. In an actual system, such as the Stelmor process, the value for F will depend on the geometry of the
overlapping rods, for which a simple solution is not
available in the literature. As a result, experiments, or
detailed radiative calculations, are required to determine
the radiation factor as a function of bed position and steel
temperature.
Clearly, the heat-transfer coefficient, due to convection, is the key thermal variable which must be controlled in the Stelmor process. Correlations for convective
heat transfer from cylindrical bodies in crossflow are
available in the literature for a range of cooling fluids.t7.8.9]
All of the correlations are empirical and relate the Nusselt
number (Nu) to the Reynolds number (Re) and Prandtl
number (Pr). One of the equations, as given by Kreith
and Black, E71 is
Nu = CReXpr y

[5]

where C, x, and y are constants which depend on the

magnitude of Re and on the cooling medium.
Correlations also have appeared in the literature, where
the objectives were to study heat transfer from cooling
steel rods and bars. Mehta and Geiger u~ conducted a set
METALLURGICAL TRANSACTIONS A

of experiments in a bar mill to determine the effect of

operating parameters on the cooling rates of steel bars.
Although the cooling medium employed in the mill was
water, the principles utilized for the investigation are applicable to air cooling. The thermal response of the bars,
measured in the mill, was utilized to back-calculate surface heat fluxes and heat-transfer coefficients v i a solution of the transient heat-conduction equation (Eq. [ 1]).
Stelmor cooling of rods was examined experimentally
by Hanada e t al. I~l~ T h e objective of the experiments was
to quantify the differences among cooling conditions at
various locations on the bed and for different Stelmor
deck configurations. Although details of the experimental technique were not reported fully, rod samples
200 m m in length and 5.5 or 11 m m in diameter were
instrumented with a C H R O M E L - A L U M E L * thermo-

ematical model for prediction of phase evolution during

rod cooling by Iyer e t al. tl2j The technique employed for
temperature measurement during the tests involved
threading a steel plug, instrumented with a thermocouple, into a hole to the rod centerline. An air source
supplied a constant velocity for cooling of the 10-mmdiameter rods employed in the tests. A range of cooling
rates was studied, and the thermal responses of the rods
were found to compare favorably with mathematical model
predictions.
Although these earlier studies provided encouraging
results on heat transfer under Stelmor conditions, it was
decided to confirm the findings in this work. But more
importantly, a technique for measuring cooling rates was
sought which performed equally well under laboratory
and plant conditions.

* C H R O M E L - A L U M E L is a trademark of Hoskins Manufacturing

Company, Hamburg, MI.

V.

couple inserted through a radial hole drilled to the

centerline. No detail was provided on the method utilized to anchor the thermocouple in place. In order to
simulate the Stelmor process, the instrumented rod was
either placed in bundles with other rods at various angles
of contact or was cooled as a single rod. The bundles
and single rods were preheated to a temperature typical
of Stelmor cooling, then cooled in a crossflow of air.
Results showed the effect of cooling air velocity, rod
diameter, and geometry of the cooling bundle on the average cooling rate of the instrumented rods. Additional
work on a pilot-scale Stelmor line showed the effect of
bed position and damper angles on the average rod cooling rates.
Instrumented steel rods were employed to tune a math-

In order to cool steel rod under well-defined conditions in a crossflow of forced air and thereby simulate
the Stelmor process, the apparatus shown in Figure 2
was constructed. The equipment made use of a "constant
velocity duct" (CVD) at the discharge end to provide a
uniform air velocity at the rod surface, over a length of
200 mm. A 10-hp Rootes-type compressor with a rated
capacity approaching 100 1/s supplied air to the CVD,
resulting in a peak air velocity of 22 m / s . The velocity
was controlled by bleeding air from a parallel line originating at the compressor. Five vanes were mounted inside the upper zone of the CVD to facilitate uniform air
flow through the discharge end. To increase back pressure in the system and enhance the uniformity of the
velocity distribution at the bottom of the CVD, a 60-mesh

APPARATUS AND SAMPLE PREPARATION

Fig. 2 - - Diagram showing test rod m position under constant velocity duct for typical laboratory rod cooling test.
METALLURGICAL TRANSACTIONS A

VOLUME 22A, NOVEMBER 1991--2771

screen was inserted 50-mm upstream of the air discharge. A pitot tube was employed to evaluate the distribution of air velocities at the discharge end of the duct.
An orifice plate situated upstream from the CVD was
calibrated against the average air velocity exiting the apparatus and was utilized to measure the air velocity for
each test. Owing to heatup of the compressor during its
operation, the air temperature was measured with a thermometer, and a mean temperature was recorded for each
cooling test.
The composition of all steel grades utilized in the laboratory and plant tests is presented in Table I. The steels
employed in the experiments were obtained from Stelco
Inc. in Hamilton, ON, Canada and are typical of plaincarbon material rolled in a rod mill. The steel grades can
be divided into three categories: (1) eutectoid or neareutectoid grades (steels A, B, and F in Table I),
(2) medium-carbon grades (steels C, D, G, and H), and
(3) lower carbon grades (steels E, I, and J). Steel
grades A through E were employed for the laboratory
experiments, while grades C and E through J were under
study in the plant trials. Table II summarizes the conditions investigated during the laboratory tests for which
four different rod diameters, between 8 and 15 mm, were
adopted and air velocities from the CVD ranged from 6
to 22 m/s. A summary of conditions in the plant trials
is presented in Table III; three rod diameters (7.5, 9.1,
and 15 mm) were examined in the tests, and two locations on the Stelmor deck were investigated, as indicated
in the table. In total, 68 tests were performed in the plant.
During the campaign, the conveyor speed ranged from
0.43 to 0.71 m / s , but for the majority (90 pct) of the
trials, it was 0.46 to 0.56 m/s.
For the laboratory tests, 350-mm lengths were cut from
the rod loops supplied by Stelco and straightened, with
some of the samples being machined down to smaller
diameters. As indicated in Figure 2, the centerline temperature in the rod samples was monitored by a thermocouple which was connected to a chart recorder (Kipp
and Zonen model BD 41) and data logger (John Fluke
Manufacturing Inc. model 2280). The method employed
for attaching the thermocouple to the rod is depicted in
Figure 3. As can be seen, two holes were drilled into
the centerline of the rod, one through which a 0.25-mmdiameter
mullite-sheathed,
CHROMEL-ALUMEL
thermocouple was introduced and the other through which

Table I.

Table II. Summary of Steel Grades, Rod

Diameters, and Air Velocities Studied in the Laboratory
Steel
Grade

Rod
Diameter
(mm)

A (1080)

10

B
B
B
C
C
D
E
E

15
11
8
11
8
11
11
8

(1070)
(1038)
(1037)
(1020)

Air
Velocity (m/s)
20, 18, 16, 14, 21, 20,
9, 11, 19,22
22, 16, 11, 9
22, 10, 6
22, 15, 6
22, 15, 10, 6
22, 13,6
22, 16, 12
22, 15, 13, 6
22, 12, 6

Table III. Summary of Steel Grades, Rod Diameters,

and Number of Tests Completed during Plant Trials
Steel
Grade

Rod
Diameter
(mm)

Number of
Center
Bed Tests

Number of
Edge
Bed Tests

C
E
F
F
G
H
I
J

15
15
9.1
7.5
9.1
7.5
9.1
7.5

7
7
4
5
4
6
6
5

4
1
3
4
3
2
2
5

a steel set screw was threaded to anchor the thermocouple junction. This arrangement provided an effective
means for ensuring good contact between the therrnocouple and the rod while minimizing disturbance of the
thermal response being measured.
The experimental technique followed for the laboratory tests, involving pitot tube and rod temperature response measurements, was repeated in the plant, at the
Stelco Hilton Works No. 2 Rod Mill, with minor changes.
For temperature response determination, thermocouples
were mounted at the rod centerline utilizing the same
technique, but longer rod lengths ( - 4 5 0 ram) were cut.
Heating of the samples was once again accomplished in

Chemical Analysis of Rod Samples Employed in Laboratory and Plant Experiments (in Weight Percent)

Grade
Code

Mn

Si

Cu

Ni

Cr

Mo

Cb

A
B
C
D
E
F
G
H
I
J

0.789
0.69
0.393
0.377
0.201
0.772
0.369
0.335
0.183
0.200

0.74
0.76
0.82
0.79
0.50
0.87
0.77
0.72
0.38
0.95

0.021
0.014
0.016
0.007
0.005
0.011
0.006
0.010
0.002
0.004

0.033
0.019
0.021
0.024
0.009
0.017
0.015
0.015
0.014
0.007

0.237
0.022
0.28
0.238
0.017
0.163
0.19
0.244
0.004
0.174

0.005
0.088
0.007
0.009
0.007
0.007
0.006
0.008
0.006
0.007

0.002
0.005
0.007
0.005
0.003
0.004
0.003
0.004
0.004
0.004

0.052
0.028
0.027
0.034
0.020
0.024
0.024
0.026
0.014
0.023

<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002

<0.002
<0.002
<0.027
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002

<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002
<0.002

2 7 7 2 - - V O L U M E 22A, NOVEMBER 1991

METALLURGICAL TRANSACTIONS A

Fig. 3 - - S c h e m a t i c diagram of method employed to mount thermocouples at centerline position in rods for Stelmor simulation tests.

ing, the centerline temperature was monitored by the data

logger, with a sampling frequency of 1 Hz. Upon completion of cooling, individual rods were removed from
the CVD and saved for mechanical testing and microstructural examination. Rods also were sectioned through
the thermocouple area to verify the exact location of the
hot junction. In all cases, the hot junction proved to be
at or near the centerline of the rod.
To obtain a range of cooling rates typical of Stelmor
cooling, a variety of rod diameters from 8 to 15 m m was
studied in conjunction with cooling air velocities ranging
from 5 to 22 m / s . The measured thermal response from
the tests provided data for determination of heat-transfer
coefficients at the rod surface as a function of rod temperature, diameter, and air velocity (Table II).
B. Plant Trials

a tube furnace, situated directly adjacent to the Stelmor

line. The temperature in the rod samples was monitored
by a strip chart recorder (Kipp and Zonen model BD41)
during heating. However, during cooling, a hand-held
data logger (Metrosonics Company Model DL-702),
which could be carried easily along the length of the bed,
was employed.
In order to supply rods with a uniform temperature
over a sufficient length, the tube furnace was specially
constructed, with particular attention being paid to minimizing the longitudinal thermal gradient. A 63-mm OD
quartz tube was wound with 2.4-mm-wide C H R O M E L
strip and insulated with two layers of 6.4-ram-thick
FIBERFRAX* sheet. This assembly was contained in a
*FIBERFRAX is a trademark of Standard Oil Engineered Materials
Company, Niagara Falls, NY.

shell of thermobestos insulation and encased in an aluminum tube. The furnace was 690 m m in length and
designed for a peak temperature of 900 ~ operating with
a 220 V power supply. To minimize scale formation on
the test rods, a flow of nitrogen of approximately 3 to
6 1/min was maintained through the furnace. Measurement of the axial temperature profile down the furnace
showed that the 200-mm center section was isothermal
to within ---5 ~

VI.

For the plant trials, a slightly different method was

followed. Sample heating was monitored by the strip chart
recorder only. After reaching the desired test temperature, samples were also soaked for approximately
5 minutes. However, it was difficult to maintain this soak
period in all cases, because test times were dictated by
the rolling mill schedule. In each test, an attempt was
made to match the grades and diameters of the instrumented rod with those being processed on the line. At
the appropriate time, the thermocouple leads were disconnected from the chart recorder and connected to the
hand-held data logger. Each instrumented rod was quickly
withdrawn from the furnace and placed on the Stelmor
line during normal operation. Care was taken to ensure
the instrumented rod was woven into the rod loops, thus
preventing unwanted movement as it traveled the length
of the line. The instrumented rods were placed at two
locations, one at the center of the bed, where the coils
are loosely packed, and one at the edge of the bed, where
the packing density is much higher. A schematic diagram of the two positions is shown in Figure 4. The
sampling frequency of the hand-held data logger was
1 Hz. Temperatures were recorded until the instrumented rod reached the end of the Stelmor deck, whereupon it was removed from the coils and saved for

PROCEDURE

A. Laboratory Tests
Prior to each laboratory test, the tube furnace was heated
to, and held at, the desired austenitizing temperature while
being flushed with nitrogen at a flow rate of 3 1/min.
For the medium- and high-carbon grades, an austenitizing temperature of 850 ~ was chosen, whereas for the
low-carbon grades, a temperature of 875 ~ was adopted.
Rod samples were placed in the tube furnace, and the
centerline temperature was monitored with the strip chart
recorder. Once the desired temperature had been achieved,
the samples were held for an additional 5 minutes soaking time. For the laboratory tests, the rods were then
withdrawn quickly from the furnace and placed in the
cross flow of air, as indicated in Figure 2. During coolMETALLURGICAL TRANSACTIONS A

Fig. 4 - - S c h e m a t i c diagram of relative packing density of coils at the

edge and center of the Stelmor line.
VOLUME 22A, NOVEMBER 1991

2773

subsequent mechanical testing and microstructural

evaluation.
The set point on the Stelmor blowers at the Hilton
Works is not continuously variable but is either "full on"
or "off." In an attempt to determine the air velocity distribution with the blowers set at full on, a series of pitot
tube measurements was made on the line during a down
period in the plant. With air to the Stelmor decks turned
on but in the absence of rod loops, the pitot tube was
moved to various locations on the bed in each of the four
cooling zones. For each measurement, the pitot tube was
held at the same height as the rod loops when the Stelmor
line is in operation. In total, 16 separate regions on two
Stelmor lines were investigated, yielding nearly 450 individual air velocities to map out the velocity distribution across the bed as well as along its length.

VII.

900
......
850
0
e,....

8OO

Z}

750

r'~

7O0

E
(l)
I---

650

Steel B 1070
8-ram Dia.
22 m/s Vel.
Steel E 1020
1l-ram Dia.
6 m/s Vel.

",,/'-",,

600
550
500
20

40

60

80

100

Time (s)
Fig. 6--Typical laboratory thermal responses measured in 8-mmdiameter (steel B 1070) and 1l-ram-diameter(steel E 1020) rods cooled
with air velocities of 22 and 6 m/s, respectively.

RESULTS

A. Laboratory Tests
The results of the pitot tube measurements of the velocity profile over the length of the CVD for three orifice
plate pressure drops are shown in Figure 5. The results
confirm the essentially uniform (-+ l0 pct) air velocity
over the length of the CVD.
Two typical thermal responses measured at the centerline of a high-carbon (steel B 1070) and a low-carbon
(steel E 1020) steel rod cooled during the laboratory tests
are shown in Figure 6. Recalescence due to both the
austenite-ferrite and austenite-pearlite transformations is
evident for the 1020 grade, while recalescence due only
to the austenite-pearlite transformation is apparent for the
1070 steel. Owing to the differences in rod diameter, as
well as cooling air velocity, a significant difference between average rod cooling rates exists in the two samples. Similar results were obtained from the remainder
of the laboratory tests, and a complete report can be found
elsewhere. [~31
The cooling curves were employed to determine heattransfer coefficients as a function of rod diameter and air

30
28
26
24
22

g
-~a
o
o)
>
?<

20
18
U-Tubeah=8.8 rnrn Hg

16
14
12
10

U-TubeAh=2.7 rnm Hg

8
6
4

, , , ,
3

11

13

15

17

19

Blower Outlet Position (cm)

Fig, 5 - Velocity profiles for three orifice plate pressure drops. Note
the constant velocity across the width of the duct.
2774--VOLUME 22A, NOVEMBER 1991

velocity. Two techniques were adopted: one assumed

negligible temperature gradients (lumped parameter), and
the other was based on a finite-difference technique to
back-calculate an effective heat-transfer coefficient from
the measured centerline temperature. The first method
assumes negligible internal resistance to heat flow, which
can be assessed by the magnitude of the Biot modulus, Bi:

hD
Bi = - k

[6]

where h is the heat-transfer coefficient, D is the rod diameter, and k is the thermal conductivity of the material.
In general, if Bi < 0.1, there will be a small error in
assuming negligible internal gradients for calculation of
the heat-transfer coefficients. [~4] For this condition, the
heat-transfer coefficient is given by

hov - -proC~ In
2t
LT---~-TAJ

[7]

where Ta is the ambient air temperature, To is the initial

temperature (for a given time interval), T is the temperature at time t, and ro is the rod radius. The specific heat
of the steel, Cp, was calculated at each temperature based
on the composition and phases present, with values taken
from the literature.i~s,16jT]
The second method involved the use of an iterative
scheme whereby the heat-transfer coefficient was initially guessed; then finite-difference equations were utilized to solve for the rod centerline temperature (the model
will be described in detail in Part IIIil9]), and the predicted temperature was compared with the measured value.
A difference of 0.01 ~ was taken as a limiting value
for each series of iterations. This process was repeated
for successive time steps throughout the thermal excursion of each rod sample. The advantage of this technique
is that the radial temperature gradient through the cooling rod is not ignored, particularly for large diameter
rods and high cooling rates, thus providing a more realistic
estimation
of the
overall
heat-transfer
coefficient, hov.
METALLURGICAL TRANSACTIONS A

A plot of calculated heat-transfer coefficient as a function of rod centerline temperature, employing both techniques for a cooling test on steel C (1037), is presented
in Figure 7. It may be noted that in the temperature range
from 620 ~ to 700 ~ where the austenite decomposition reactions occur, the latent heat released by the
transformations makes calculated h values meaningless.
However, it can be seen that for temperatures before and
after the transformation, little difference exists between
the heat-transfer coefficients calculated by the two methods. It is also evident that the change in the magnitude
of the overall heat-transfer coefficient as the sample cools,
~ 1 9 0 W / m 2 ~ at 800 ~ reducing to ~ 1 2 0 W / m 2 ~
at 500 ~ is due primarily to the decrease in radiative
heat transfer. Plots similar to that shown in Figure 7 were
utilized to determine the variation in overall heat-transfer
coefficient as a function of temperature for all laboratory
tests.
Comparisons between theoretically predicted and empirically calculated values for h have been made for the
laboratory data at several temperatures. Utilizing Eq. [5]
with the properties of air evaluated at the mean film temperature, as given by Kreith and Black, ~71the convective
heat-transfer coefficient was calculated from the following equation:
\ 0.466 ~

hc = 0 . 6 8 3 -

~ 1/3

[8]

\ k II

D\lx]

as a function of rod diameter and cooling air velocity.

Figure 8 shows a comparison of measured and predicted
heat-transfer coefficients for a rod temperature of
525 ~ The heat-transfer coefficients are plotted as a
function of v~176
, w h i c h according to Eq. [8],
should provide a linear relationship, owing to the fact
that over a small range of temperature (---5 ~
the radiative heat-transfer coefficient is constant and independent of rod diameter or air velocity. The predicted line
shown in Figure 8 is based on the sum of Eqs. [4] and
[8]. The agreement between the measured and predicted

~"
E

t-t--

90

8o

7-

70

o
0
03

525 ~

200
180

--

1
I
I

[] Measuredh o

190

[]

Predicted h

t~

170
r

160

~E

150

o
O

140

03
c"

t-O3
7-

[]

130
120
110
100
90
80
70

0.6

0.'8

1
V 0,466

1.2

1.4

D o.534
Fig. 8 - - M e a s u r e d heat-transfer coefficient at 525 ~ for the labodiameter) ~
The
solid line is based on the combined effects of radiation and convection

ratory tests plotted against (air velocity)~

(Eqs. [4] and [8]).

values is seen to be quite good, as would be expected,

and provides confirmation of the use of the instrumented
rod to characterize cooling under Stelmor conditions.
Similar results have been obtained at other temperatures
for the laboratory results.
Additional support for predicting the heat-transfer
coefficients, based on the combined effect of radiation
and convection according to Eqs. [4] and [8], can be
found in the literature. Average cooling rates for two rod
diameters at various air velocities reported by Hanada
et al. (lu have been converted into heat-transfer coefficients, employing the techniques outlined previously. The
results, shown in Figure 9, once again exhibit good
agreement between predicted and measured hedt-transfer
coefficients.

300

200
190
180
170
160
150
140
130
120
110
100

'~
(1)
o

210

Test C7

B._~. Measured h-5.5 mm Dia.

o Measuredh-ll mm Dia.
Predicted h

280

o Finite-Difference Method
Lumped-Parameter Method

t~

t~

t~

260
240

220

9-~
O
200

[]
O 13

mO@Om0 +

[]

[]
[]

[]

160
140

120
I--

[]

100

~o
o

Phase Transformation Range

60

60
500

540

580

620

660

700

740

780

Temperature (~
Fig. 7--Heat-transfer coefficients calculated from measured rod
centerline temperature, utilizing both a lumped-parameter and a finitedifference technique. The results are for a steel C (1037) 8-mm-diameter
rod cooled with an air velocity of 6 m / s .
METALLURGICAL TRANSACTIONS A

40

~
0.2

0.4

0.6

0.8

1
1.2
V 0.466
D 0.534

1.4

1,6

1.8

Fig. 9 - - Measured heat-transfer coefficient from Hanada e t a l . [HI plotted

against (air velocity)~ 466/(rod diameter) ~
The solid line is based on
the combined effects of radiation and convection (Eqs. [4] and [8]).
VOLUME 22A, NOVEMBER 1991--2775

B. Plant Trials

The transverse profiles of the air velocities, measured

at several locations on two Stelmor lines, are presented
in Figure 10. As can be seen, in an attempt to provide
uniform cooling across the loops, a higher air velocity
of nearly 30 m / s is applied to the edge of the bed, where
the packing density of the rods is highest, as compared
to about 18 m / s at the center, where the rod packing is
minimum.
Typical thermal responses measured at the centerline
of instrumented rods during the plant trials are shown in
Figure 11. As can be seen, they have a similar appearance to those obtained in the laboratory tests (Figure 6).
It is important to note, however, that some of the temperature responses measured in the plant exhibited erratic behavior and, therefore, were discarded. This
behavior was thought to arise from vibration and move-

Fig. 1 0 - - A v e r a g e air velocities measured on Stelmor lines 2 and 3.

The shaded area indicates ---1 standard deviation for the measurements.

850
800

750

......

', ~
',,
~
',,,, \\, / _ ~

--

Steel F 1080
7.5-rnm Dia.
Edge FFFF
Steel E 1020
15-mmDia.

700

280
260
~E

240

~,

220

D_
t-

ment of the test rods on the Stelmor line, which resulted

in poor thermocouple contact or from damaged thermocouple wires. The valid thermal responses from the plant
tests were employed to calculate the heat-transfer coefficients as a function of process variables. As was indicated earlier, the Stelmor process variables include a
difference in air velocity between the center and the edge
of the bed and the option of operating the line with or
without forced air flow. In this latter condition, radiative
cooling dominates, with the combination of forced (due
to motion of the line) and natural convection contributing in a minor way to the overall heat transfer.
Similar to the laboratory results, the calculated surface
heat-transfer coefficients for 800 ~ have been plotted
against V0"466//O 0"534 in Figure 12 for center and edge positions and three rod diameters employed in the experiments. The line in Figure 12 represents the predicted
heat-transfer coefficient based on convection only. Owing
to the difficulty in predicting the radiative component of
the heat-transfer coefficient in rods bundled on the Stelmor
line, no attempt has been made to predict an overall heattransfer coefficient from first principles, although this is
an important next step. Instead, plots such as those shown
in Figure 12, have been employed to determine the relative magnitude of hR, assuming the difference between
the average measured heat-transfer coefficient and the
predicted value based on convection is due to radiation.
A series of plots similar to Figure 12 has been produced at different temperatures from the plant thermal
data. The difference between the measured heat-transfer
coefficient and that predicted from the correlation for
convection only (Eq. [8]) has been calculated to determine the radiative component of the heat-transfer coefficient as a function of rod temperature. Figure 13 shows
the calculated radiative heat-transfer coefficient for the
center and edge of the bed as a function of temperature.
As can be seen, with decreasing temperature there is a
dramatic decrease in the magnitude of the radiative component, as would be expected. The radiative coefficient
at the center of the bed is consistently larger than that
at the edge, where greater radiative interchange occurs

E:

650

o
0

O
0

B
O

200

o :~

180
160

600

140

550

I-t~
"1-

120

500

O Center Full (800~

+ EdgeFull (800~
- - Pred. h Convective

100
80

2(3

40

6()

'

80

Time (s)

Fig. 11--Typical plant thermalresponsesmeasuredin 7.5-mm-diameter

(steel F 1080) and 15-mm-diameter(steel E 1020) rods, cooled at the
edge and center of the Stelmor bed, respectively. The "FFFF" in the
figure legend indicates that cooling air was "on" in all four cooling
zones for the tests.
2776--VOLUME 22A, NOVEMBER 1991

60
0.7

09

1.1

V 0.466
D 0.534

Fig. 1 2 - - M e a s u r e d heat-transfer coefficient at 800 ~ for the plant

tests plotted against (air v e l o c i t y )0 466 /(rod diameter) ~
The predicted line is based on convection only (Eq. [8]).
METALLURGICAL TRANSACTIONS A

80
c~ Center of Bed Tests

~.

+ Edge of Bed Tests

7O

where x represents 875-T in degrees centigrade.

Equations [10] and [11] are employed in the mathematical model to predict the radiative heat-transfer coefficient for cooling conditions obtained on the Stelmor line
but, it should be noted, apply to conditions where the
rod surface temperature is less than 875 ~

, , m

6o
o

50

~t-

4o

F--

30

T
~
._.

20

"~
rr

10

VIII.

400

500

600

700

800

Temperature (~
Fig. 13--Calculated values for radiative heat-transfer coefficient from
the plant trials plotted against temperature. Also included is the predicted value from Eq. [4].

among the more densely packed loops (Figure 4). Also

included in Figure 13 is the predicted radiative heattransfer coefficient based on Eq. [4]. A comparison of
the predicted and measured values for the radiative
heat-transfer coefficient can be utilized to construct an
effective "radiation correction factor," R, for the two locations on the Stelmor bed. This correction factor can
be utilized in the following equation to calculate hR:

Ts

TA/'

[9]

Results from the plant trials have been employed to obtain the value of R for the center (Rc) and edge (RE)
position of the bed. Each is plotted in Figure 14 as a
function of rod temperature. The equations are
Rc = 2.02x-~

-~176176176

[1 1]

o Center of Bed Tests

1

NOMENCLATURE
Bi
C

1.1

0.9

In this first part of a three-part paper on microstructural engineering applied to the controlled cooling
of steel rod, the question of heat transfer in the process
is addressed experimentally and theoretically. A series
of experiments, conducted in the laboratory as well as
on an operating Stelmor line, has been performed to
measure the thermal response of an instrumented steel
rod under controlled cooling conditions. Results from the
experiments were utilized to back-calculate the overall
heat-transfer coefficient, and a comparison of the measured values was made with empirical correlations. The
following conclusions can be drawn from the work:
1. The experimental technique employed for the laboratory and plant tests provided a reproducible means
for measuring the thermal response at the centerline
of a cooling rod.
2. Comparison between predicted and measured heattransfer coefficients for the laboratory tests showed
that reasonable estimates of hR and hc could be made
utilizing standard equations for laboratory conditions.
3. For an operating Stelmor line, heat-transfer coefficients reflected the radiative interaction among adjacent loops on the bed. By assuming the portion of
the overall heat-transfer coefficient due to convection
can be predicted reasonably by a published equation,
correlations for hR as a function of rod temperature
and position on the bed have been determined.

[10]

RE = 8.94x-~176176176

+ Edge of Bed Tests

//~

- Predicted

D
F
h
hc

0.7

0.6

0.5

hov

hR

0.4

~
rr

O.3

k
Nu
Pr
qrR

0.2

0.1

Biot number
empirical constant used in Eq. [5]
specific heat, J kg -1 ~
rod diameter, m
radiation factor in Eq. [4]
heat-transfer coefficient, W m -2 ~ t
heat-transfer coefficient due to convection,
W m -2 ~

0.8
u_
~

SUMMARY AND CONCLUSIONS

- - ' r ~ - ' ~ r ~ - - - ~ r - - ~
400
500
600

r
700

800

Temperature (~
Fig. 14--Radiation correction factor for center and edge of bed plotted against temperature. Lines are calculated from the regression
equations for both parameters (Eqs. [10] and [11]).
METALLURGICAL TRANSACTIONS A

ro
R
Rc
RE

overall heat-transfer coefficient, W m 2 ~

heat-transfer coefficient due to radiation,
W m -2 ~
thermal conductivity, W m -~ ~
Nusselt number
Prandtl number
rate of heat release during phase
transformation, W
radial position, m
rod radius, m
radiation correction factor
radiation correction factor for center of the
Stelmor line
radiation correction factor for edge of the
Stelmor line
VOLUME 22A, NOVEMBER 1991--2777

Re
t

TA
To
T~
x

Y
P
o
8

/z
v

Reynolds number
time, s
temperature, K or ~
ambient temperature, K or ~
initial temperature, K or ~
surface temperature, K or ~
empirical constant used in Eq. [5] and
symbol denoting undercooling, ~ in
Eqs. [10] and [11]
empirical constant used in Eq. [5]
density, kg m -3
Stefan-Boltzman constant,
5.6710 -8 W m -2 K - 4
emissivity
kinematic viscosity, m 2 s -1
air velocity, m s -1

ACKNOWLEDGMENTS
The authors wish to acknowledge the Natural Sciences
and Engineering Research Council of Canada for support
of research expenses. The University of British Columbia
awarded a University Fellowship, while the Cy and
Emerald Keyes Foundation provided a scholarship to P.C.
Campbell. The cooperation of Steltech and Stelco Steel
in organizing the plant trials is deeply appreciated.

REFERENCES
1. J.K. Brimacombe, E.B. Hawbolt, I.V. Samarasekera, P.C.
Campbell, and C. Devadas: in Proc. Thermec "88, I. Tamura,
ed., Iron and Steel Institute of Japan, Tokyo, pp. 783-90.

2778--VOLUME 22A, NOVEMBER 1991

2. A. Tendler: Wire J., 19-81, vol. 14, pp. 84-91.

3. J. Tominaga, K. Matsuoka, and S. Inoue: Wire J. Int., 1985,
vol. 18, pp. 62-72.
4. P. Bercy, U.G. Boel, N. Lambert, and M. Economopoulos: Metall.
Plant Technol., 1984, vol. 4, pp. 46-51.
5. A.A. Jalil: Iron Steel Eng., 1982, vol. 59, pp. 46-48.
6. G.H. Geiger and D.R. Poirier: Transport Phenomena in
Metallurgy, Addison-Wesley, Reading, MA, 1973, pp. 367-69.
7. F. Kreith and W.Z. Black: Basic Heat Transfer, Harper and Row,
New York, NY, 1980, pp. 249-51.
8. C.O. Bennett and J.E. Myers: Momentum, Heat andMass Transfer,
McGraw-Hill, New York, NY, 1974, pp. 381-83.
9. Chemical Engineer's Handbook, J.H. Perry, ed., McGraw-Hill,
New York, NY, 1969, pp. 10.13-10.15.
10. A.J. Mehta and G.H. Geiger: Mechanical Working and Steel
Processing Conf. XV, ISS-AIME, Pittsburgh, PA, 1977,
pp. 458-82.
11. Y. Hanada, K. Ueno, A. Noda, H. Kondoh, T. Sakamoto, and
K. Mine: Kawasaki Steel Technical Report, 1986, vol. 15,
pp. 50-57.
12. J. Iyer, J.K. Brimacombe, and E.B. Hawbolt: Mechanical Working
and Steel Processing Conf. XXII, ISS, Pittsburgh, PA, 1984,
pp. 47-58.
13. P.C. Campbell: Ph.D. Thesis, The University of British Columbia,
Vancouver, 1989.
14. F. Kreith: Principles of Heat Transfer, 3rd ed., Intext Educational
Publishers, New York, NY, 1973, p. 140.
15. British Iron and Steel Research Association: Physical Constants
of Some Commercial Steels at Elevated Temperatures,
Butterworth's Scientific Publications, Guildford, Surrey, United
Kingdom, 1953, pp. 3-14.
16. I. Barin, O. Knacke, and O. Kubaschewski: Thermochemical
Properties of Inorganic Substances; Supplement, Springer-Verlag,
New York, NY, 1977, pp. 245-46.
17. JANAF Thermochemical Tables, The Thermal Research
Laboratory, Dow Chemical Company, Midland, MI, 1960.
18. P.C. Campbell, E.B. Hawbolt, and J.K. Brimacombe: Metall.
Trans. A, 1991, vol. 22A, pp. 2779-90.
19. P.C. Campbell, E.B. Hawbolt, and J.K. Brimacombe: Metall.
Trans. A, 1991, vol. 22A, pp. 2791-2805.

METALLURGICAL TRANSACTIONS A