O Bus Kovic Nevenka 1989

AN ABSTRACT OF THE THESIS OF
Nevenka Obiftkovid for the degree of Doctor of Philosophy
in Chemical Engineering presented on September 28, 1988.
Title: Heat Transfer Between Moving Beds of Solids and a
Vertical Tube
Abstract approved:
Redacted for privacy
Dr:James G: Knudsen
Heat transfer to moving packed beds of solids has been studied
extensively but predictions are usually specific to the experimenter's
own equipment. In order to obtain a more general predictive equation,
the heat transfer coefficient was measured for a single vertical tube
immersed in a moving packed bed of glass beads, sand, or copper in air
at atmospheric pressure and compared to measurements performed by
independent researchers with various experimental configurations.
High speed photography was used to observe the motion of the
particles as they flowed past the heated wall. The particle separation
from the wall could not be determined. There was little interchange of
the wall particles with the particles in the bulk beyond the wall
layer. Likewise, the particles in the layer adjacent to the wall
showed very little rotational or cross-flow motion.
A numerical method, based on the unsteady state conduction
equation, was developed to predict the heat transfer coefficient
between the wall and the flowing bed of solid particles. This method
could be used to predict the experimental heat transfer coefficients
obtained provided the correct value of the particle separation from
the wall was used in the numerical solution. This separation cannot be
predicted a priori so that the numerical method is, therefore, not
suitable for predicting the heat transfer coefficient.
The experimental data showed good agreement with the analytic
solution obtained by Mickley and Fairbanks (1) at long contact times.
At short contact times, the data do not agree with the Mickley and
Fairbanks solution but the heat transfer coefficient tends to level
off at an asymptotic value.
An empirical correlation was obtained by which the contact time
at which the data depart from the Mickley and Fairbanks solution may
be predicted. The following equation, was found to be valid for the
data of the present work and for data of other researchers who used
different geometries in their work.
tcr
0.3622 + 9.691 L.
dpps
This equation may be used in conjunction with the Mickley and
Fairbanks solution to predict the heat transfer coefficient for a
given flowing packed bed over a wide range of contact times.

Heat Transfer Between Moving Beds of Solids
and a Vertical Tube
by
Nevenka S. Obutkovid
A THESIS
submitted to
Oregon State University
in partial fulfilment of
the requirements for the
degree of
Doctor of Philosophy
Completed Sept. 28, 1988
Commencement June 1989

Approved:

o Chemical Engineering in charge of major
Chairman of Chemic'alfEngineering Department

Dean of Gradua5 School ci
Date thesis is presented Sept. 28, 1988
Typed by Kevin Greek for Nevenka Oblrtkovie

ACKNOWLEDGEMENT
I would like to express my thanks to my major
professor Dr. James G. Knudsen for all the understanding
and financial support at times when this job seemed to
be a neverending story; also for reading and correcting

the manuscript and the final draft.
Special thanks are extended to my dear professor
Dr. Octave Levenspiel not only for his insistent
questions on music, art, geography, history and life in
general but also for the lost backgammon games, jokes
and free lunches that anyone could get if he shows up
in front of his office around 12pm (students only).
Acknowledgments are also extended to my professors and
chairmen Dr. Charles Wicks and Dr. Robert Mrazek and Dr.
Jim Fredericks for enormous understanding and support in
the completion of this thesis.
My thanks also go to Nick Wannenmacher for his
invaluable help in experimental research whenever
necessary and to Jordana Chambers expert in solving all

my administrative problems.
I wish to express gratitude to my friend and
"computer-wiz" Kevin Greek or as I call him Mr. Blasius

who was unselfishly available whenever the word
"computer" was mentioned; also for correcting and typing

the manuscript.
Special thanks go to my mother Vera and dear friend
Dr. Vojislav Bogic for their everlasting support
understanding and suggestions.
I also want to thank the rest of faculty members
for their help and encouragement during the course of
this research.
Thanks also to all my "beanery" friends for being

there when necessary.
Finally, my greatest appreciation goes to my very
dear friend Hubert called Mici for all the lengthy hours
of listening to my complaints about this work and his

criticism that I did not always accept.
NOMENCLATURE
Symbol Description Units
Arod surface area of the heated

section available for heat
transfer. m2
Cps heat capacity of the solids. J/kgK
D heated rod diameter.

dp particle diameter.
dpi diameter of particles of

size "i".
fD (fW) friction coefficient between

a tube and dry (wet) solids.
h heat transfer coefficient. W /m2K
hi local heat transfer

coefficient. W /m2K
kg thermal conductivity of gas. W/mK
ke effective thermal
conductivity of solids-gas
emulsion. W/mK
kgs reduced thermal conductivity

of gas. W/mK
L heated length parallel to

the flow path of solids. m
LC heated length across the flow

path of solids
(characteristic length).
n number of sides of polygon.
hd
Nud Nusselt number based on
kg particle radius (dp).
h D
Nu') Nusselt number based on
kg heated rod diameter (D).
p - power input to the heater.
Cpsdppplis
Ped = Peclet number based on
kg particle diameter (dp).
CpsDpbUs
Pep = Peclet number based on
kg heated rod diameter (D).
rp particle radius.
Rc contact resistance (due to

gas film). m2K/W
s,x distance from heated wall.
T
g
temperature of the gas.
Ts bulk temperature of the
solids.
Tw steady-state average surface

temperature of the rod.
t - residence time.
tcr critical or separation time.
tmax - time where maximum heat

transfer occurs.
Us linear solids velocity. m/s
Xi - mass fraction of particles of

size "i".
GREEK SYMBOLS
oe effective thermal
diffusivity (emulsion). m2/s
as effective thermal
diffusivity (solids). m2/s
ratio of gap width to

particle radius.
porosity of packed bed.
60 void fraction for wall

separation distance greater
than 0.7d P.
Er void fraction for wall

separation distance less than
0.7d
P.
accomodation coefficient.
A mean free path of gas
particle.
Pb, Pe bulk density of solids. kg/m3
Ps density of solids. kg/m3
a thickness of the gas layer.
TD, (TO effective shear stress for

dry (wet) solids. N/m3
TABLE OF CONTENTS
Chapter Page
I. INTRODUCTION 1
II. SELECTED PREVIOUS WORK 3
III. EXPERIMENTAL EQUIPMENT 8
III.A. Solids 8
III.B. Heater rod 11
III.C. Test section and solids transport

system 13
III.D. Particle motion and porosity

measurement equipment 15
III.E. Experimental procedure 18
III.E.1. Heat transfer coefficient

measurement 18
III.E.2. Velocity measurements 20
III.E.3. Particle motion and porosity

measurement 21
IV. HEAT TRANSFER RESULTS 24
IV.A. Present study 24
IV.B. Data comparison with other researchers 29
V. MODEL ASSUMPTIONS 37
V.A. The role of the gas phase 38
V.B. Contact resistance 40
V.C. Particle shape 44
V.D. Flow and porosity distribution 45
V.D.1. Flow properties 46
V.D.2. Static properties 49

Chapter Page
V.E. Summary 54
VI. HEAT TRANSFER COEFFICIENT MODELS 56
VI.A. Analytic models 56
VI.B. Numerical model 61
VI.C. Empirical model 66
VII. CONCLUSIONS 77
VILA. Summary of Findings 77
VII.A. Furthure Work 79
BIBLIOGRAPHY 80
APPENDICES
A. Analytic heat transfer coefficient for rod 84
B. Heat transfer coefficient for two region

semi-infinite slab 91
C. Heat transfer from wall to particle through

interstitial fluid 95
D. Finite difference calculation of heat transfer

coefficient to packed beds 102
E. Experimental data 161
F. Statistical analysis 170
G. Particle size distribution 171

LIST OF FIGURES
Figure Page
(III.B.1) Heated stainless steel tube. 12
(III.C.1) General layout of moving bed

apparatus. 14
(III.D.1) Transparent test section used with

high speed camera. 17
(IV.A.1) Comparison between heat transfer

coefficients as a function of the
linear solids velocity for 0.18mm
and 0.8mm diameter glass beads. 25

coefficients as a function of
contact time with the heated surface
for 0.044mm, 0.18mm, and 0.8mm
diameter glass beads. 26

contact time with the heated surface
for 0.11mm and 0.8mm diameter sand. 27

coefficient as a function of contact
time with the heated surface for
0.044mm, 0.18mm, 0.8mm glass beads
and copper of 0.21mm in diameter. 28
(IV.B.1) Comparison between heat transfer

linear solids velocity for Colakyan's
(10) data and the present study for
0.11mm, 0.27mm and 0.8mm diameter
sand. 30

linear solids velocity for horizontal
finned (11) and vertical tube for
0.11mm and 0.8mm diameter sand. 31

coefficients as a function of
contact time for Colakyan's (10)
data and this study for 0.11m,
0.27mm, and 0.8mm diameter sand. 32
(IV.8.4) Comparison between heat transfer
coefficients as a function of contact
time for the author's (11) data and
the present study for 0.11mm and
0.8mm diameter sand. 33

Ernst's (9), Schlunder's (13), and
the present data. 35

Sullivan's (3) data and the present
study. 36
(V.A.1) Comparison between heat transfer

glass in air and vacuum conditions
(reproduced from Wunschmann (13)). 39
(V.D.2.1) Particles in orthorhombic array. 51
(V.D.2.2) Scheme for porosity calculation

by penetration s into orthorhombic
array of spheres. 53
(VI.A.1) Heat transfer coefficient comparison

by contact time to rod and slab
analytic models and experimental
data. 59
(VI.B.1) Heat transfer coefficient comparison

as a function of contact time with the
heated surface for the M&F (1) model
and the numerical solution for 0.8mm
(VI.B.2) Heat transfer coefficient comparison

as a function of contact time with the
heated surface for the M&F (1) model
and the numerical solution for 0.18mm
(VI.B.3) Heat transfer coefficient as a
function of contact time with the
heated surface for M&F (1) model and
the numerical solution by gap width
for 0.18mm diameter glass beads. 65
(VI.C.1) Comparison between heat transfer

time with the heated surface for the
present data and the M&F (1) model. 68
(VI.C.2) Empirical prediction of tcr by Eq.

(VI.C.1) and Ernst's (9) data. 72

(VI.C.1) and Desai's (21) data for
glass. 73

(VI.C.1) and Desai's (21) data for
copper. 74
(VI.C.5) Empirical prediction of t by Eq.

I.C.2) for
(VI.C.1) and tmax by Eq. (VI.C.2)
data of the present study. 75
(VI.C.6) Comparison between heat transfer

coefficients by contact time with the
heated surface for Ernst's (9),
Wunschmann's (12), Harakas (2) and
the present data. 76
(B.1) Representation of composite slab. 92
(C.1) Representation of separation

distance between a spherical
particle and wall. 96
(D.1) Finite difference map to model

wall-to-bed heat transfer. 107
(D.2) Example of cell noding scheme. 108
(D.3) Finite difference scheme at wall to

obtain axial average heat transfer
coefficient. 120
(D.4) Error introduced by modeling a

semi-infinite slab as a finite
slab. 124
(D.5) Accurary of heat transfer solution
to a semi-infinite slab on Fod
value. 128
LIST OF TABLES
Table 2s19.2.
(III.A.1) Properties of Materials Used and

Operating Conditions. 9
(III.A.2) Properties of Solids. 10
(E.1) Table of Experimental Conditions and

Heat Transfer Coefficients for Glass
Beads. 161

Heat Transfer Coefficients for Sand. 163

Heat Transfer Coefficients for
Copper. 165

Heat Transfer Coefficients for
Falling Solids. 166
Regression Analysis of Model. 170
List of All Data used to Fit the

Model. 170
(G.1) Size Distributions of Particles 171

Heat Transfer Between Moving Beds of Solids
and a Vertical Tube
I. INTRODUCTION
The most important aspect of heat transfer
information for any kind of heat exchanger is the
average heat transfer coefficient to an immersed heated

surface. This quantity is usually obtained by measuring
the power dissipation from the surface to the

surrounding media. Reliable data and correlations for
the average heat transfer coefficient have great
commercial importance, therefore an increase in heat
transfer can nearly proportionally reduce the cost of

the exchanger.
Ever since gas-solid systems have been recognized
for their excellent heat transfer characteristic,
considerable research has been conducted in this field
especially for fluidized beds. However, less data is
available for moving packed beds of solids than for
fluidized beds. Existing predictions have generally
been compared against the researcher's own data instead
of equally considering data from the literature.
Consequently, there has been little study to explain
deviations between data by various experimental setups.
The knowledge of these heat transfer variations over a
wide range of particle sizes and velocities would be of

great interest when designing or selecting materials for
2
solid-gas heat exchangers.
For this purpose, this study presents heat transfer
coefficient measurements to granular media flowing by
gravity past a vertical heated tube and compares it to
the limited number of data available in the literature.
The comparison shows that the heat transfer coefficient
to vertically oriented heaters may be predicted
graphically and a unique empirical solution to predict
the heat transfer coefficient is proposed.

3
II. SELECTED PREVIOUS WORK
A considerable amount of work has been done on the
flow patterns as well as heat transfer behavior of
granular media in systems such as fluidized beds, packed
beds, moving beds, and suspension. In spite of many
attempts, the mechanism of convective heat transfer
to/from surfaces immersed in flowing solids is not yet
well understood. Related past research is briefly
reviewed in this chapter.
Mickley and Fairbanks (1) were among the first to
obtain data for the instantaneous heat transfer
coefficient in fluidized beds and provided an
analytical solution for the proposed version of the
"packet" theory. The packet (i.e. moving particles with
entrained gas) is assumed to be in contact with the
heated surface for a short time and then is replaced by
a new packet from the bulk of the bed. They considered
the packet to be homogeneous with uniform thermal
properties and derived an expression for the local
instantaneous heat transfer coefficient hi,
hi
at
Their results indicated that heat transfer resistance
exists in the layers of particles closest to the wall,

4
but the independent effect of particle size is not
included.
Harakas and Beatty (2) considered the heat transfer
from a flat plate immersed in a rotating bed of granular
material (the flow of particles was parallel to the
plate). They varied particle size as well as
interstitial gas and observed an increase in average
heat transfer coefficient by either increasing
interstitial gas conductivity or decreasing particle

diameter. Their experimental findings are presented in
Chapter IV. Fine grained materials were considered a
one-component continuum and effective properties of this
continuum were used to predict a mean heat transfer

coefficient. However, great discrepancies in the model
predictions occurred as the particle size increased.
This result is expected since the thermal gradient which
forms within the "packet" in a short residence time does
not extend beyond one or two particle diameters from the
surface.
The convective heat transfer from a vertical flat
plate to various granular materials was investigated by
Sullivan and Sabersky (3). To explain their
observations analytically, they defined an idealized
model called the "discrete particle model". In this
model it was assumed that the conductance at the wall
was the same as that existing between adjacent rows of

5
particles and that the conductance was inversely
proportional to the average thickness of the gas film
between particles and the heated plate.
Studying the mechanism of heat transfer between a
fluidized bed and immersed surfaces Baskakov (4)
introduced an additional contact resistance Rc, in
series with the thermal resistance of the packets, Re,
as defined by Mickley and Fairbanks (1). The contact
resistance, Rc, is assumed to represent the additional
thermal resistance of the layer next to the wall due to
increased porosity.
Denloye and Botterill (5) studied the heat transfer
in flowing packed beds. They obtained data by varying
the particle sizes of different interstitial gases.
They concluded that the heat transfer coefficient
increases with increasing gas thermal conductivity,
decreasing particle size, and particle residence time.
Spelt (6) investigated the heat transfer to a
granular material flowing in an inclined chute. He
found that the heat transfer coefficient increased to a
maximum and that decreased as the velocity of the solids
across the plate increased. The phenomena is believed
to occur due to changes in the packing density of the

flowing media.
Russian researchers Donskov (7) and Kurockin (8)
also experimented with flowing packed beds. Donskov (7)

6
investigated the heat transfer coefficient by varying
the tube diameter and pitch of the tubes in a tube bank.
The results showed that the heat transfer coefficient

increases with a decrease in the ratio between pitch and
tube diameter. Kurockin (8) investigated the heat
transfer coefficient between dry and wet material, and a

single tube. He concluded that heat transfer increased
as the velocities of the granular material increased,
and he correlated his results with the following

equation:
Nu') = 0.22 Pep j [710.33

dp tfw rw
Ernst Rudolf (9) concluded that for moving beds of
solids and vertical or horizontal tubes at "short"
contact times the heat transfer coefficient is only
important within 0.1mm to 1mm away from the heated

surface. He also found that the heat transfer
coefficient at short contact times exhibits a maximum
value which is independent of the physical properties of
the particles but depends on particle size.
Colakyan's (10) work on heat transfer from a single
tube to various granular materials flowing in a
direction perpendicular to the cylinder axis provided an
empirical value for the contact resistance Rc first
introduced by Baskakov (4). The resistance was

estimated to be
6.7 x 10-5 dp2

Rc
k
g
Previous work preformed by the author (11) upon the
heat transfer coefficient between a horizontal finned
tube and moving beds of solids led to the following
empirical correlation:
0.3 dp 0.33
Nud = 2.37 Ped(3-25
Lc
g
The equation predicts the heat transfer coefficients for
plain and finned tubes within 20%, but is restricted to
tubes with fin heights equal to or smaller than that
used for this research.
It was observed that the heat transfer coefficient
is greatest for particles with small diameters as
compared to the large diameter particles of the same
substance. The heat transfer coefficient also depends
on the velocity of the particle flow over the heated
surface as explained later.
Other research concerning flow of solid material,
porosity distribution, gas properties, surface
influence, etc. is discussed elsewhere in the related

sections.
8
III. EXPERIMENTAL EQUIPMENT
An experimental test system was developed to
determine the heat transfer coefficient for granular

materials flowing by gravity past a vertical heated
tube. Particles having various solids properties and
sizes were selected for measurement as a function of
transport velocity. A physical description of each
component is reviewed here.
III.A. Solids
In the present work only granular materials are
considered; porous materials in which a solid medium has
a closed gaseous space were not considered. Four types
of solids, sand, glass beads, copper, and iron were used
as the granular media. Properties of the materials used
in the experiments as well as the range of operating
conditions are given in Table (III.A.1). Table (III.A.2)
lists all the properties of the four different solids
used in the experiments. Effective properties were
calculated using Eq. A.2 in Appendix A.
Microscopic photographs of these particles
revealed the glass beads had the most smooth and regular
shape distribution. In contrast, the iron solids have a
9
Table (III.A.1). Properties of Materials Used and

Operating Conditions.
Material Range of Operating Conditions

Particle Diameter, dp 0.044 0.8 mm
Solids Density, ps = 2500 - 8900 kg/m3
Bulk Density, Pb = 1500 5360 kg/m3
Solids Thermal
Conductivity, ks = 0.8 384 W/mK
Solids Heat
Capacity, Cps = 380 1080 J/kgK
Linear Solids
Velocity, Us = 0.004 0.3 m/s
Table (III.A.2). Properties of Solids.
Material d ks Ps ps E Pb ke aex108 asx108

(mm) (W/mK)(kg/m3)(J/kgK) (W/mK)(kg/m ) (m2/s) (m2/s)
Sand 0.11 0.80 2700 780 0.488 1280 0.14 14.0 0.38
Sand 0.80 0.80 2700 780 0.436 1410 0.17 15.5 0.38
Sand 1.2 0.80 2700 780 0.424 1440 0.18 16.2 0.38
Glass 0.044 1.04 2500 1080 0.426 1435 0.26 16.6 0.53
Glass 0.18 1.04 2500 1080 0.400 1500 0.28 17.3 0.53
Glass 0.80 1.04 2500 1080 0.400 1500 0.29 17.7 0.53
Copper 0.21 384 8950 383 0.419 5200 0.46 22.4 11000
11
wide size distribution and a flaky texture. When
pressed together, the iron would cling in clumps.
The mean diameter, dp, of each solid was
calculated according to the following formula:
d
P
= (III.A.1)
Xi
L dpi
where
xi = mass fraction of particles of size "i".
dpi = diameter of particles of size "i".
III.B. Heater Rod
The instrument illustrated in Figure (III.B.1)
consists of a stainless steel tube 0.0127m 0.D., 0.4572m
long with a 600W (115V) heating element 0.1524m long
imbedded in it. Temperature at the inside surface of
the stainless steel rod was measured with a single
imbedded T-type thermocouple positioned along the axial
center of the heater element. Power to the heater was
regulated by a 115V 5amp variac and measured with a
Jewell electric wattmeter to 5% tolerance. A
calibration procedure was performed to determine the
metal resistance (k/x) between the thermocouple and the

rod outside surface. Corrected temperatures were then
used for calculations of the heat transfer coefficient.
A table for heat transfer coefficients using corrected

STAINLESS STEEL HEATER ROD (0.4572m long)
T-type thermocouple
/1.1114111.11111.111MI.
moo. ----111111ftum
/,--"Iiii1111111111=0.
power cord
cartridge heater(0 1524m long)
Figure (III.B.1). Heated stainless steel tube.

13
temperatures is presented in Appendix E.
The surface of the heated rod, in which the
heating element is installed, maintains a constant
temperature axially over a 0.1524m length. Some
experiments were conducted to verify this condition and
temperatures along this length were observed within 0.5F
deviation.
III.C. Test Section and Solids Transport System
The heater rod was centered within a 0.0381m I.D.
tube and its position was maintained by three retaining
screws at either end of this 0.4572m long tube. The
tube was made of clear plexiglass to permit visual
examination of granular flow about the heater rod. A
small hole in this vertical test section near the center
of the heated section of the rod allowed a thermocouple
to penetrate and measure solid temperature near the tube
wall during the experiment. A cone centered at the base
of the heater rod and inside the bottom opening of the
test section maintained an equal radial distribution of
solids flow about the rod length and its height could be
adjusted to vary the flow rate of solids leaving the
bottom of the test section.
A ten gallon supply hopper acted as a solids
reservoir and emptied its contents through a 2in I.D.
pipe which was kept capped until the experiment was

14
HOPPER
FUNNEL
POSITIONING SCREWS
TRANSPARENT TEST TUBE
HEATED TUBE (0 4572m long)
CARTRIDGE HEATER
(0 1524 m long)
POSITIONING SCREWS
INVERTED FUNNEL
(solids flow conrtoller)
THERMOCOUPLE CONECTORS
POWER CORD
Figure (III.C.1). General layout of moving bed

apparatus.
15
ready to begin. Solids were fed through a funnel into
the test section and collected into a bucket below. To
recirculate the flow, solids were manually transferred

into the supply hopper. The depth of the supply hopper
outlet pipe into the funnel was kept constant throughout
experiments with all materials and the height of solids
in the hopper was maintained at the same level as much
as possible. Maintaining these positions constant was
essential to assure constant flow rate. The funnel
outlet size limited the rate of mass flow into the test
section and was largely responsible for the maximum
transport velocities achieved (Figure III.C.1).
Two thermocouples, one to measure ambient air, the
other to measure solids leaving the funnel, were
employed to determine static two phase (gas-air)

temperatures. These two thermocouples, including the
heater rod thermocouple, were wired through a selector
switch to a 0.1F digital thermocouple reader.
III.D. Particle Motion and Porosity
Measurement Equipment
In order to record motion of particles which slide
against the surface of the heating rod, an apparatus
was devised whereby a high speed motion picture camera

could be focused upon a small portion of the rod. A
plexiglass tube with an outer radius identical to the

16
heater rod was constructed within a test section to
enable photography of particles through the tube wall
under conditions closely resembling particle contact
with the heater (Figure III.D.1). A small vertical
section of both this tube and the test section wall was
removed and a tunnel between the two openings was
installed. A Hycam model 41-0004 rotating prism camera
was focused through the tunnel upon the outside edge of
the tube and the subject was illuminated from 3 feet by
a 1000W tungsten-halogen spotlight. To assure focus and
alignment, the test section was firmly clamped to the
equipment support column and the camera was positioned
on a level surface over a rubber mat. Framing speeds
may be measured directly from the film margin since
timing marks are exposed on the film during camera
operation by an internal time dot generator. The motion
was recorded at speeds to 2000 frames per second on a
high speed 16mm film, Kodak RAR 2484, developed for 8min
in D19 developer at 20C to a sensitivity of 1250 ASA.
Film spools of approximately 40ft were used for each
run. When imaged with a Sopelem macro zoom lens set for
highest magnification (250mm f/5), a view field of
3mm x 5mm was obtained, the latter dimension closely
approximating the arc length of rod surface in view.
Due to this high magnification, an extremely short depth
of field ensured only those particles within 1mm of the

17
CROSS SECTION
OUTER TRANSPARENT TUBE
INNER TRANSPARENT TUBE
FOCUSING WALL
camera view
........
Figure (III.D.1). Transparent test section used with

high speed camera.
18
tube wall could be imaged.
Spherical particles of glass (0.8mm) and copper

(mixed sizes of 0.8 to 0.2mm) were filmed for porosity
measurements for moving packed beds. The glass beads,
in an attempt to obtain a better photographic image,
were colored with blue paint.
III.E. Experimental Procedure
III.E.1 Heat Transfer Coefficient Measurement
This section provides details of the steps
performed in the experiment, experimental equipment
operation constraints or criteria chosen to select a
particular range of operation, and calculations required
to obtain a measured value of the heat transfer
coefficient as a function of solids velocity. The
experimental procedure required the heat transfer
coefficient to be measured under steady state

temperature conditions. Once the cap of the supply
hopper was removed and the flow rate through the test
section was adjusted, power to the heating rod was

applied. Steady state was reached when the temperature
variation at the surface of the rod was negligible with
time (a deviation of not more than 0.1 to 0.2F over a

few minutes). When the rod temperature stabilized, the
time required for solids leaving the test section to
19
collect in a beaker for mass survey was measured and the
thermocouple readings and power were recorded.
Measurements were repeated several times for a single
solids type, flow setting, and power level.
Precautions regarding power levels, and therefore
operating temperatures of the rod, were taken. First,
the rod surface temperature could not exceed 300C to
avoid errors introduced by T-type thermocouples which
are not designed to operate at these higher

temperatures. Second, by providing the maximum possible
temperature difference Tw-Ts the highest precision was
ensured. In all experiments this difference is at least
50C. It was also observed that at power levels below
75W significant deviation existed in heat transfer
coefficient calculations from rod temperatures compared
to those calculations at all higher power levels.
Therefore, experimental power levels were maintained
within a range of 75 to 200W.
Solids depth in the supply hopper influenced flow
velocity and therefore rod temperature changed a small
percentage as the hopper was replenished. Maintaining
an equal level within the hopper as well as performing
multiple measurements reduced what affect this variation
might have.
In a large number of test runs under different
conditions of solids velocity, the test section was

20
rotated with respect to the rod to confirm that the rod

thermocouple gave identical readings when placed in
different radial positions.
The heat transfer coefficient h is calculated for
each run using the following relationship.
Arod(Tw Ts)
where
P = power input to the heater (W)
Arod = surface area of the heated section
available for heat transfer (0.00608048m2).
Ts = bulk temperature of the solids (K).
Tw = steady state average surface temperature of
the rod (K).
Flow velocity was calculated from the time
required for solids discharged from the test section to

fill a survey beaker and the mass which accumulated.
III.E.2. Velocity Measurements
The velocity was measured in two different ways:

1. The amount of solids discharged from the test
section in a given time was collected and
weighed. Knowing the properties of the
material and the limiting cross sectional area

available for flow, the velocity was estimated
21
using the following relationship
(mass flow rate)

us
(bulk density) (flow area).
2. The velocities were estimated by visual
observation of the particles near the
transparent wall by measuring the time taken
for the particles to travel a given distance.
This method assumes that solids are flowing in
plug flow.
For data analysis, velocities measured by the first
method were used. Flow velocities at the wall were also
recorded for comparison by high speed photography.
111.E.3. Particle Motion and Porosity Measurement
At the high magnification required to photograph
particle movement, alignment and stability of the camera
and test section was critical. Film changes were
therefore performed without moving the camera position
and both alignment and focus were verified prior loading
each film. Also, the solids were allowed to flow
through the section before switching on the spotlight
and camera and precautions were taken to isolate
movement by the operator from the equipment while the
exposure was taking place. Solids velocity was
regulated by the height of an inverted funnel centered
at the bottom of the test section and a different

22
velocity was set for each run. A solids velocity
measurement by sampling the mass and duration of solids
accumulated from the test section occurred

simultaneously.
The photographic experiment was exposure limited
because of the recording speed required. Therefore, the
lens aperture was opened fully throughout the experiment
so the inherent camera aperture of f/3 was the only
restriction on the intensity of exposure. Due to the
narrow opening of the view tunnel into the test section,
spotlight placement had an effect on the intensity and
clarity of the image of the particles viewed through the

lens. Some experimentation with spotlight placement was
therefore necessary for initial runs.
Film processing occurred on site to minimize delay

for analysis. To analyze the films, a 16mm film editor
was used to view particle motion and velocity whereas an
enlarger displayed individual frames for porosity and

particle separation. A film of a static packed bed for
each particle type was used for comparison. Particle
speed was computed by tracking a vertically moving
particle throughout the height of the frame image,
counting the consecutive frames, and calculating the
corresponding elapsed time using the timing marks

exposed on the film. Individual frames selected for
appearance of distinct particles were projected upon a

23
sheet of paper where the measurements were recorded. At
least 10 to 20 distinct frames were sampled from each
film to measure particle separation and interpret
particle number per frame touching the rod wall.

24
IV. HEAT TRANSFER RESULTS
The following section provides a graphical
presentation of the experimental results for heat
transfer from a vertical bare tube and moving beds of
solids. A graphical comparison between results of the
present study and some available data from the
literature for various experimental setups accumulated
from either tables or digitized from original figures is
also included.
IV.A. Present Study
Figure (IV.A.1) presents the heat transfer
coefficient as a function of the linear solids velocity
for glass beads of 0.8mm, 0.18mm and 0.044mm in
diameter. As expected from the results of the previous
workers, the heat transfer coefficient increases with a
corresponding increase of the solids velocity. The heat
transfer coefficient can be observed to level off with
increasing velocity for all three sizes and the maximum
value for the heat transfer coefficient decreases as
particle size increases. When data are plotted on a
log-log scale all three particle diameters follow a
straight line with a slope of 0.4 and only the heat

transfer coefficient for glass beads of 0.8mm diameter
700
o glass dp= 0.044mm
600 glass dp=0.18mm
A glass dp =0.8mm
500
(NJ
400
300 A
_c
200
100
0
0.000 0.100 0.200 0.300 0.400
Velocity (m/s)
Figure (IV.A.1). Comparison between heat transfer coefficients as a function
of the linear solids velocity for 0.18mm and 0.8mm diameter U,
glass beads.
1000
S
flp
.
.11)N
heater length L=0.1524m

0 glass dp=0.044mm
100 glass dp-=0.18mm
6, glass dp =0.8mm
064,
')
50 f 1 !fill F I I 111111 1 f- I 1 flfli 1 I I
0.010 0.100 1.000 10.000

Time (s)
Figure (IV.A.2 ). Comparison between heat transfer coefficients as a function
of contact time with the heated surface for 0.044mm,
0.18mm, and 0.8mm diameter glass beads.
E
sand dp=0.8mm
100 0 sand dp=0.11mm
50
0.100 1.000 10.000
Time (s)
Figure (IV.A.3). Comparison between heat transfer coefficients as a function
of the contact time with the heated surface for 0.11mm and
0.8mm diameter sand.
1000
S A
4
011)
6,0
heater length L=0.1524m
o glass dp=0.044mm
glass dp=0.18mm
100 6, glass dp =0.8mm
A copper dp=-0.21mm
50
0.010 0.100 1.000 10.000
Time (s)
Figure (IV.A.4). Comparison between heat transfer coefficient as a function
of contact time with the heated surface for 0.044mm,
0.18mm, 0.8mm glass beads and copper of 0.21mm in diameter.
29
appears to level off (Figure IV.A.2). Figure (IV.A.3)
presents data for silica sand, and Figure (IV.A.4)
presents heat transfer to glass beads and copper against
time. Figure (IV.A.4) shows moderately higher values in
the heat transfer coefficient were obtained for copper
than for glass, although the conductivity of copper is
384 times higher than glass.
IV.B. Data Comparison with
Other Researchers
Heat exchangers are observed to be more efficient
when the fluid flows perpendicular to the heated
surface. This phenomenon is often explained by the
formation of a thinner laminar boundary layer which
forms when the fluid flows normal to the heated surface.
Following the same reasoning, a similar behavior for a
gas-solid system should be expected.
When the data for sand of 0.8mm diameter flowing
past horizontal and vertical heater configurations was
plotted jointly in Figures (IV.B.1 and IV.B.2), the
horizontal configuration seems to provide higher heat
transfer results. Apart from material properties,
"time" that solids spend in contact with the heated
surface is the second most important parameter when
predicting the heat transfer coefficient. Thus, Figures
(IV.B.3 and IV.B.4) present the same data in the time

450
400
350
A
300
(NI 250
E
200
0 0.8mm horiz. tube (10)
150-i 0.27mm horiz. tube (10)
A 0.8mm vert. tube
100 0.1 1 mm vert. tube
50
0
0.000 0.100 0.200 0.300
Velocity (m/s)
Figure (IV.B.1). Comparison between heat transfer coefficients as a function
of the linear solids velocity for Colakyan's (10) data and
the present study for 0.11mm, 0.27mm and 0.8mm diameter sand.
700
600 o sand dp=0.11mm vertical tube

sand dp =0.8 mm vertical tube
o sand dp=-0.11mm horiz. finned tube
500- sand dp =0.8 mm horiz. finned tube
CN1 400
E 04
O
300
200
100
0
0.000 0.100 0.200 0.300 0.400
Velocity (m/s)
Figure (1V.8.2). Comparison between heat transfer coefficients as a function
of the linear solids velocity for horizontal finned (11)
and vertical tube for 0.11mm and 0.8mm diameter sand.
A
141
E
data for sand: Do
S
A 0.11mm present study vert.tube
100 0 0.8 mm present study vert.tube
0.8mm Colakyan (10) horizontal tube
A 0.27mm Colakyan (10) horizontal tube
50
0.100 1.000 10.000
Time (s)
Figure (1V.13.3). Comparison between heat transfer coefficients as a function
of contact time for Colakyan's (10) data and the present
study for 0.11m, 0.27mm, and 0.8mm diameter sand.
o sand dp=0.11mm horiz. finned tube
1000 sand dp=0.8 mm horiz. finned tube
A sand dp=0.11mm vertic. bare tube
A sand dp =0.8 mm vertic. bare tube
100
50
0.100 1.000 10.000 100.000
Time (s)
Figure (IV.8.4). Comparison between heat transfer coefficients as a function
of contact time for the author's (11) data and the present
study for 0.11mm and 0.8mm diameter sand.
34
domain. The result was "almost" unexpected; vertically

oriented heaters showed higher heat transfer results
when plotted against velocity. As opposed to horizontal
heaters, vertical heaters without external obstructions
have an equally distributed flow of solids at the
surface without formation of gaps or stagnant zones.
Flow patterns about horizontal tubes are characterized
by higher particle-wall porosity and results in
decreased heat transfer. A detailed explanation of
solids flow is given in section V.D.1. Therefore, it is
concluded that for vertical and horizontal heaters with
equivalent characteristic length higher values of the
heat transfer coefficient are expected for vertical
configurations.
Experimental results from Ernst (9), Wunschmann
(12), and this study are given in Figure (IV.B.5). As
shown, they all follow the same line till they level
off. The plateau where the maximum heat transfer occurs
varies with different heater arrangements.
In Sullivan's (3) study of the heat transfer
coefficient characteristic of granular media flowing
past a thin vertical plate, moderately lower results
were obtained than would be expected (Figure IV.B.6).
Uncontrolled heat losses due to their experimental
apparatus or even a minor obstruction in the particle
flow may produce the results which they obtained.

O A 1.mm glass Wunschmann (12)
1000 0-3.1mm glass Wunschmann (12)
1
417
E
100 7 A 0.8mm glass present study
6e5
0.18mm glass present study
P!!t
o 0.044mm glass present study
Etpc,
o .5.7 mm sand Ernst (9)
.3.5 mm sand Ernst (9)
O .1.2 mm sand Ernst (9)
10
0.010 0.100 1.000 10.000 100.000
Time (s)
of contact time with the heated surface for Ernst's (9),
Wunschmann's (12), and the present data.
1000
4t
present study:
E
100 0-0.044mm glass
0.18mm glass
0.8mm glass
6,
Sullivan and Sabersky (3):

o 0.33mm glass
1.35mm glass
10
0.010 0.100 1.000 10.000 100.000
Time(s)
of contact time with the heated surface for Sullivan's (3)
data and the present study.
37
V. MODEL ASSUMPTIONS
The significance of heat transfer coefficients from
a design standpoint has already been emphasized.
Therefore the ability to predict this quantity in terms

of bed dimensions and properties is of invaluable
importance. Two representative theories of particle bed
heat transfer exist: (1) the bed may be assumed to
comprise of homogeneous properties of solid and gas and
modeled as a region with a single property or, (2) a
more complex mechanism of separate gas-particle heat

transfer interactions. The bed itself cannot be
considered homogeneous since heat transfer response is
different for solid and gas phases but for long bed
heating times this has been shown to be a good
approximation. The latter approach is most important as
the velocity of the bed increases since heat propagates
among too few particles from the heated surface for the
bed to be considered homogeneous. Modeling heat
transfer discretely between gas and particle is however
complicated because of gas movement and particle
separation, transitions from the heated surface, and
particle contact with the wall are not fully understood.
In this chapter an attempt to explain the
simplifying assumptions of the heat transfer mechanism

to moving beds of solids will be provided.
38
V.A. The Role of the Gas Phase
In a moving packed bed of solids, heat is
transferred from an immersed surface by particles
directly touching the surface and by interstitial gas
(air in this study) which is moving with the particles
at the same velocity past the heated surface. Since the
volumetric heat capacity of the air is quite low in
comparison to that of solids it is assumed that heat
removed by gas convection is negligible. Most of the
heat is absorbed by the solid phase. In spite of this,
the gas plays an important role as a heat transfer
medium between the heated surface and the first row of
particles as well as between the particles themselves.
Without the gas, heat removal by only particulate solids
would be very poor regardless of the solid properties as
demonstrated in some experiments for vacuum conditions
(2,12). Figure (V.A.1) shows experimental data
reproduced from Wunschmann and Schlunder (12). Some
researchers have found that the heat transfer
coefficient between the heated surface and granular
material increases with an increase of the thermal
conductivity and/or thermal capacity of the gas (Harakas
& Beatty (2)). The physical property of the
interstitial gas has a great influence on the heat

transfer coefficient.
1000
o 1 .mm glass in air at 760mmHg
-3.1mm glass in air at 760mmHg
A-3.1mm glass in air at 0.001mmHg
100r
10
A A A AA A A A A AAA
1 10 100
Time (s)
Figure (V.A.1). Comparison between heat transfer coefficients as a function
of contact time with the heated surface for glass in air
and vacuum conditions (reproduced from Wunschmann (12)).
40
V.B. Contact Resistance
This important factor must be considered when
trying to understand heat transfer between two immobile

touching solids. Baskakov (4) was first to introduce
the concept of contact resistance to account for the
discrepancies between experimental data and the Mickley
and Fairbanks model (1). Baskakov's (4) explanation was
instrumental in explaining the observed dependence of
the heat transfer coefficient on particle size (as dp
increases h decreases). The problem of determining a
resistance term for correction with the Mickley and
Fairbanks model (1) has been handled in many ways either
by defining an appropriate effective thermal
conductivity or by introducing an arbitrary set zone of
increased voidage close to the surface. Many
researchers questioned the physical existance of the gas
gap at the contact point although it appeared almost
impossible to explain or fit the experimental data

without it. Others such as SchlOnder (13) or Decker and
Glicksman (14) abandoned the use of a gas film
separating the surface from the first row of particles.
Schlunder (13) suggests that a known formula from the
kinetic theory of gases for estimation of reduced gas
conductivity should be used to define the gas

conductivity in the region where the mean free path of
41
the gas molecules is larger than the distance between
the particle and the wall, namely
kg
kgs = (V.B.1)
A (2 /)
1 2
a 1
where
kgs = reduced thermal conductivity of gas (W/mK).
kg = thermal conductivity of the gas as a
continuum (W/mK).
A = mean free path of gas molecules (m).
a = thickness of the gas layer (m).
/ = accommodation coefficient.
The phenomena of the reduced thermal conductivity
implies heat transfer is decreased as particle size
decreases since the percent of area within one mean free
path of the contact wall increases.
The contact resistance can be treated more
realistically by considering the microscopic structure
of the surface especially at contact of the particle

and heater. When viewing the solid surface with even an
optical microscope only structures above 0.1Am are

visible. Thus a highly rugged structure can exist at
the optically smooth surface. The physical properties
at the surface interface such as surface energy,
42
tension, and conductivity arise as a result not only

from the bulk atomic structure but also from somewhat
different atomic arrangement at the surface. Some
properties, such as surface energy which is inversely
proportional to the particle size, are instrumental in
explaining the phenomena of adhesion and/or
agglomeration that is most likely to occur in a bed of
fine particulate solids. Due to existing surface
forces, the impurity-free surface of a freshly formed
solid will be covered by layers of gas molecules in less
than a second. Physically adsorbed gas can be removed
from the surface either by heating the solid to its
melting point or by degassing the surface under high
vacuum. Brenner (15.) observed that oxygen adsorbtion on
iridium surfaces reduces surface free energy. Therefore
much larger friction coefficients were measured for
flow by surfaces under vacuum than in the presence of
air. In the case of nonmetals (e.g. diamond or NaC1)
the friction coefficient increases six times when
measured under vacuum. All these facts provide support
for the long disputed existence of an adsorbed gas layer
which is obviously carried along with each particle as

its own atmosphere. However the thickness and changes
that occur according to experimental conditions are
uncertain. Thus for the gas in gas-solid systems the

difference should be made between the strongly adsorbed
43
gas (air) which cannot leave the particle surface freely
and acts more like part of the solid, and the bulk gas
(air) that moves freely between the particles. When the
particle is in contact with another solid surface it is
beyond imagination that adsorbed gas can be removed only

by particle weight. It is more probable to expect the
generation of a "binding force" between the particles
and the wall as well as between particles themselves.
Also, it has been evident that surface phenomena can be

related to the forces between atoms and molecules. In
general, interatomic forces are short range but
sometimes they can act across or between interfaces

(London forces). As particle size decreases the
importance of forces like inertia or gravity flow
decrease, until the atomic forces become dominant.
Adhesion that usually occurs in a bed of fine solids is
only one of the examples where the sum of the binding
forces between the particles is higher than
environmental forces. Apart from the adsorbed gas most
solids will pick up moisture from the surrounding air,
as observed for glass beads of 0.044mm and iron
powders. The adsorption layer is not freely moveable
but particles can come closer to each other much easier
than when they are clean. Attraction forces between the
particles are usually caused by Van der Waals,
electrostatic, or magnetic forces. These forces are

44
considerable when particles are close together and they

decrease rapidly with distance. Since the adsorbed
layer is thicker around smaller particles and at the
same time the roughness peaks are lower than for large
particles, the ability of such particles to approach
each other closer and with larger surface area is

increased.
Therefore it is believed that reduction in particle
size can bring some improvement in the heat transfer
coefficient since the greater interparticle forces will
not allow separation at increased solids velocities as
in the case of large particles. On the other hand, when
particles become too small and interparticle forces are
dominant, most of the gas (air) will be adsorbed at the
solid surface while the amount of freely moving gas
(air) will decrease leading to a consequent decrease in
heat transfer. At this point, the gas conductivity
changes since there are very few gas molecules in free

motion. Therefore, lower values for the heat transfer
coefficient than for air alone are possible.
V.C. Particle Shape
Beds of particles of irregular shape or size should
increase heat transfer since surface area in contact

with the wall or adjacent particles is minimized for
uniform sized spheres. Although particle packings of
45
extremely irregular shape and size may require
individual measurements of heat transfer coefficient,
something could be said about what one would expect for

a heat transfer coefficient to a packed bed of
irregularly shaped particles if each particle closely
approximates a sphere. In fact, for randomly packed
beds of irregular particles porosity varies
proportionally with sphericity (16). Since effective
properties of the heat transfer medium are a strong
function of porosity, increased porosity at the wall is
responsible for lower heat transfer coefficients.
Provided porosity distribution from the heated wall
to the bed is understood a computer calculation to
simulate heat transfer to the most irregular bed would

then appear possible. However, in this research similar
or even lower heat transfer coefficients were measured
under flow conditions which implies packing at the wall
may become poorer with decreased sphericity.
V.D. Flow and Porosity Distribution
The degree of separation between particles either
between themselves or the heated wall and their motion
relative to the wall may change with bed velocity and

particle shape. An increased porosity at the wall can
be reasoned to lower the effective conductivity of the
flowing medium and in fact a reduced heat transfer

46
coefficient is normally observed. Also, the amount of
heat removed from the wall depends on the nearby
particle velocity and exchange of particles in layers
streaming adjacent to the wall. Bed velocity may also
affect suspension of flowing particles. Therefore, heat
transfer prediction is largely influenced by
interrelated dynamic properties of the bed and static
particle separation within.
V.D.1. Flow Properties
Hiroshi Takahasi and Hiroshi Yanai (17) studied a
flow profile and void fraction of granular solids in a
moving bed. They observed that in some cases spherical
particles flowing down the tube form two regions. In
the shear region, which extends 4-8 particle diameters
from the wall, particles were observed to flow slower
that in the main body of the core or plug flow region.
When the ratio of bed height to column diameter
approached about 2 to 3 the velocity profile became
fully developed. This pattern of velocity variation
across the test section was not observed for shallow
beds of spherical particles neither for irregularly
shaped coal, although the latter was not discussed in
their paper. In their experiments the void fraction of
the particles was found to be influenced by velocity,

and to reach an asymptotic value for higher flow rates.
47
Also, Sullivan (3) observed that the solids adjacent to
the wall slide along with approximately the same speed
as the bulk flow. He therefore concluded that no
boundary layer exists near smooth walls.
For flow studies perpendicular to a heated tube,
Sullivan (3) noticed the formation of a cavity
immediately downstream of the cylinder. This gap
appeared even for very low solid velocities but its size
increased with flow rate. In the same study a
triangularly shaped stagnant region of solids at the
upstream face of the cylinder was observed but its shape
and size was velocity independent.
Unlike a fluidized bed, particles in a packed bed
are not suspended and free to wander about but are

pinned between their neighbors. Only if a particle is
balanced near a void large enough to accommodate it will
it move to drop in. Particle propagation may then only
extend about one particle diameter at a time. The
frequency of particle interchange as well as rolling
over the heated length would be an important measure of
the degree of increased heat transfer expected.
Using the high speed camera described earlier,
motion of 0.8mm diameter glass beads at the heated
surface were investigated in detail. The fraction of
particles observed to interchange position from all
particles viewed in the area of one frame over the

48
duration between interchanges was noted for each film.
About one interchange per 160 particles flowing past the
frame was likely to occur at low velocities. At high
velocities, an alternating pattern of particle
streaming occurred as particles settled into voids left
by supporting particles. Few particles were observed to
rotate at the wall, while all others were slipping
without rotation. For two films, the focus and lighting
was set to image particles within the bed more than lmm
away from the wall. In each of these films the
particles flowed in a linear pattern and no particle
interchange was ever observed. This indicates that the
majority of interparticle motion appears to occur at the
wall where it may make the most significant change in

heat transfer. Despite this, few particles would
interchange position at the wall so it is reasonable to
assume interparticle motion has a negligible effect on
increasing heat transfer.
An investigation into the porosity change at the
rod wall with flow velocity was performed using the high
speed camera and glass beads. Analysis of the films
revealed a small change in porosity occurred up to a
velocity of 0.5m/s, the maximum velocity investigated.
The porosity change at the wall could not be modeled due
to insufficient amount of data. Since the lens' focal
length allowed particles within one millimeter of the

49
wall to appear in focus it was difficult to distinguish
a particle separation from the wall. Therefore, it was
not possible to conclude if a gap separates a particle

from the wall or how this gap changes with velocity.
Instead, particle separation between each neighbor was

measured. The particle orientation as observed did not
depart from an orthorhombic array at velocities of 0 to
0.5m/s; another indication porosity did not change

greatly. If a bulk porosity change did occur, the
solids velocity measured by weight of the discharge

solids would be suspect. However, particle velocity at
the rod surface was comparable within experimental error
to the bulk solids velocity computed by assuming
porosity remains as that of the static randomly packed

bed.
V.D.2. Static Properties
The bed porosity for perfect spheres can be taken
as 0.4 throughout the entire bed except for the first
row of particles touching the wall where voidage is
greater than that in the bulk of the bed and can be
calculated using the relation

50
x 0.378
(V.D.2.1)
1 co 0.7d P
given by Kimura and Kaneda (18) where
er = void fraction at the distance smaller than
0.7d
P.
co = void fraction at the distance larger than

0.7dp.
x = distance from the wall (m).
dp = particle diameter (m).
If particles form closely a regular array or pattern
between themselves an equation for porosity distribution

within a packed bed to one particle diameter from a
plane wall may be obtained. Random packing of spherical
particles normally form an orientation closely
approximated by cubic or orthorhombic arrays, the
orthorhombic array being most prevalent. When the
particle array is divided into its smallest decomposable
unit of a single particle and the void surrounding it
(Fig. V.D.2.1), this region appears as a sphere
inscribed in a 4 or 6 sided prism (rhombohedron).
Viewed from the contact point of a side tangent to the
sphere, the region is azimuthally symmetric. The
apparent particle area heated from the surface of one

side may then be represented two dimensionally by a
51
Figure (V.D.2.1). Particles in orthorhombic array.

52
plane intersecting the region a distance s from the side
(Fig. V.D.2.2). This allows an equivalent representation
of the temperature gradient between two and three
dimensional models of the particle-in-a-prism region.

The ratio of void to particle surface area in the
intersecting plane at distance s is
x(2(s/r)-(s/r)2)
E(s) = 1 (V.D.2.2)
n tan(w/n)
where
n = number of sides of polygon represented by
intersection the of plane to region.

r = radius of particle.
The porosity of the region based on volume is
2w
E = 1
3n tan(r /n)
and, for example, c=.395 for n=6.
Assuming heat transfer through particle contact
points is negligible, then, heat transfer may be modeled
two dimensionally over the particle-in-a-prism region
where the heated surface is located at s=0.
Measurements by researchers such as Benenati et.

al. (19) have shown the porosity distribution from the
wall into a packed bed of spheres appears as a damped

53
Figure (V.D.2.2). Scheme for porosity calculation by

penetration s into orthorhombic
array of spheres.
54
sinusoid asymptotic to a porosity of 0.4. When
integrated over distance traversed from the wall, Eq.
V.D.2.2 closely approximates their findings. For tube
to particle diameter ratios (D/dp) less than 6, particle
size has a significant influence on the bed porosity
distribution function and Eq. V.D.2.2 would no longer
hold. When this ratio is larger than 6 the porosity
distribution function drops from unity to 0.4 within 0.7
particle diameters from the wall (19). The results for
voidage at the wall were independent whether the wall
was concave or convex.
V.E. Summary
It is obvious that extensive knowledge of all the
above mentioned variables and flow behavior of two phase
systems is necessary to understand the phenomena of heat
transfer between heated surfaces and flowing granular
media. The following general assumptions may be made
regarding heat transfer to flowing packed beds of
solids:
1. Thermal properties of the interstitial gas
largely determine heat transfer throughout the
bed.
2. The assumption of a contact resistance term
between heated wall and particle is necessary

55
to model experimental data. A definite
physical explanation of the resistance is
complicated. The complication arises in
modeling the gas adsorbed at the solid surface
alone or an additional gas gap separating the
solid from the wall.
3. Bulk porosity distribution appears independent
of bed velocities common to most research with
flowing packed beds. The particle rotation
and propagation at the wall is too rare an
event to significantly enhance heat transfer.
However the velocity dependence of a gas film
which separates particles from a vertically
heated surface is unknown.
4. Porosity of a static bed as a function of
distance from the wall varies as defined by
Eq. V.D.2.2 for beds of regular spheres
provided the ratio of tube to particle
diameter is greater than 6.

56
VI. HEAT TRANSFER COEFFICIENT MODELS
This section reviews the effectiveness of several
analytic and numerical models of increasing complexity
upon prediction of the heat transfer coefficient given
the assumptions of the last section. A unique empirical
model is also presented which assumes the departure from

the Mickley and Fairbanks (1) model at low residence
times is most dependant upon a characteristic change of

flow properties.
VI.A. Analytic Models
To date, any experimental apparatus used to record
heat transfer to a flowing bed of particles measures
steady state temperatures on the heated surface and
within the bed once flow of solids has been adjusted to
a constant velocity. The heat transfer coefficient is
then calculated by the temperature difference from the
bed the heated surface finally assumes. To represent
heat transfer to a flowing bed of solids mathematically,
a moving coordinate system is typically used whereby the
point of reference travels with the particle. This
representation models the process as an unsteady state
heat conduction problem where the bed is assumed to
possess some initial temperature and heat is applied

from the surface for a time assumed equivalent to that
57
required by the particle to traverse the length of the
heated section in the experimental system. The heat
transfer coefficient is then calculated for the amount
of heat added to the bed and the temperature of the wall
relative to that of the bed initial temperature.
In any experimental system, heat transfer to the
bed begins when the gas-solid mixture flows past the
heated boundary and continues for the length of the
heater. What is measured therefore is a heat transfer
coefficient averaged over that length. For comparison,
the coefficient predicted by the moving coordinate
system corresponds to a single time or point along the
heated length and therefore this instantaneous value
must be averaged over the time of flow past this length.
The problem of any simple analytic treatment of
heat transfer to flowing beds which does not account for
contact resistance, such as the Mickley and Fairbanks
(1) equation, is that as the simulation time decreases
to zero the heat transfer coefficient modeled approaches
infinity when actually a maximum value is observed.
Nevertheless, the equation is a reasonable model to be
used at large residence times. A comparable model
applied to rod geometry as opposed to slab geometry was
derived and appears in Appendix A. Models applied to
both geometries are comparable at low residence times,
however the rod model heat transfer coefficient flattens

58
out for higher time and decreasing radii. Figure
(VI.A.1) shows only a small deviation from either
geometry model is observed for 0.8mm and 0.18mm glass
particles about 30sec. A greater magnitude of heat
transfer appears by both analytic models when compared
to experimental data due to the effective conductivity

calculated by Eq. A.2
The simplest way of accounting for contact
resistance analytically is to separate the treatment
into a gas gap region and a solid region in order to
model unsteady state conduction to a particle adjacent
to the wall. A second analytic model was then developed
based on a derivation for a composite semi-infinite slab
and appears in Appendix B. Unfortunately, because
particle size and curvature is not accounted in the
method it would not be expected to fit data for
different particle sizes.
Once the composite slab model was modified to treat
a hemispherical interface at the wall, the model became
appropriate to predict the value of the heat transfer
coefficient which occurs during heating of the gap. The
derivation, which appears in Appendix C, is useful until
the heat transfer coefficient becomes equivalent to that
of the Mickley and Fairbanks (1) equation based on
effective properties. This is an inherent limitation
since only half a particle from the wall could be

rod model
slab model (M&F)
.147 mm glass H&B (2)
O
1000 7:
v
E
0 .044 mm glass present study
A 0 .8 m m glass present study
1.0mm glass Wunschmann (12)
3.1 mm glass Wunschmann (12)
.5.7 mrn sand Ernst (9)
- .3.5 mm sand Ernst (9)
O .1.2 mm sand Ernst (9)
10 1 1111111 I 1 1111+11 I I I I 11}1-1 1 1 1 111111 1 1 1 111111 1 I 1
0.001 0.010 0.100 1.000 10.000 100.000

Time (s)
Figure (VI.A.1). Comparison between heat transfer coefficients as a function
of contact time with the heated surface for rod and slab
analytic models and experimental data.
60
modeled whereas heat propagation from one to three
particle diameters into the bed is necessary before the

solution assumes the behavior of the Mickley and
Fairbanks equation (1).
The maximum heat transfer coefficient due to gas
gap heating predicted by the above model appears as a
time independent value particularly for particle
diameters less than a millimeter. This maximum is not
influenced by solid properties and is given
approximately by
kg w
h
r I 6+2 6
Arctan [i
6 2
1
where
kg = gas conductivity (W /m2 -K).
r = particle radius (m).
6 = ratio of gap width to particle radius.
Denloye (20) assumed conduction into the bed could
be modeled by a high porosity region nearest the wall
and a second region which assumes the bulk porosity of
the bed. As opposed to the above model, with this
treatment he was able to approximate the heat transfer
coefficient extending to a time where agreement with the
Mickley and Fairbanks equation (1) is possible using
appropriate effective properties for each region.
However, the multiple use of effective conductivity

61
correlations which increases the subjectivity of the

heat transfer coefficient prediction is to be
discouraged when a more direct method to determine the
coefficient is possible numerically.
VI.B. Numerical Model
To be able to calculate the heat transfer
coefficient to packed beds without use of effective
conductivity correlations and to explore the effects of
different contact resistance terms and gas-particle
orientations, a finite difference method was developed
to model two-dimensional heat conduction in slab and
cylindrical geometries with spatially dependent

properties. Detailed explanation of the equations,
assumptions, input preparation, as well as verification
of the method is given in Appendix D.
Input to the program was prepared to follow the
porosity distribution by distance into the bed given by
Eq. V.D.2.2. It was discovered a separation between
particles themselves needn't be represented to fit the
data once an appropriate gas gap separated the first
particle from the wall. Figures (VI.B.1 and VI.B.2)
show the comparison between calculated and experimental
data for glass beads of 0.8mm and 0.18mm diameter

respectively. For this research, an air gap of 21Am for
0.8mm diameter and a gap of 15Am for 0.18mm diameter

1000
computer fit for glass
dp=0.8mm and gap
of 21 microns
M&F (1) model
glass dp=0.8mm
\
1 00 1 1 i I I I I il 1-4-1-4---1-1--44- 1 1-4- I , 1 1 1 1 I I I 1 1 1 1 1
0.010 0.100 1.000 10.000 100.000

Time (s)
Figure (VI.B.1). Heat transfer coefficient comparison as a function of
contact time with the heated surface for the M&F (1) model
and the numerical solution for 0.8mm diameter glass beads.
1000
computer fit for glass
dp=0.18mm and gap
of 15microns
M&F (1) model
N A
E
I'
glass dp=0.18mm
100 f I1 11 1 1 I f 41141 1 1 f 1 4111 4 t' 1 I 1 414
0.010 0.100 1.000 10.000 100.000

Time (s)
Figure (VI.B.2). Heat transfer coefficient comparison as a function of
contact time with the heated surface for the M&F (1) model
and the numerical solution for 0.18mm diameter glass beads.
64
glass was required to fit the experimental findings.
Introduction of a greater contact resistance appears
necessary for smaller particles since a proportionally
thicker gas gap to particle size ratio is needed to fit

the data.
As mentioned before, each particle may possess an
adsorbed gas layer at its surface and so modeling the
solid by a direct contact with the wall would be

improper. In order to demonstrate the necessity of
contact resistance, then, the number of nodes
representing each particle diameter was consecutively
increased for several cases so that the area of contact

with the wall decreased. The solution appeared to
converge gradually with decreasing contact area but an
increase in the computed heat transfer coefficient
continued beyond the data with a slope of 0.2 on a log-
log plot as opposed to 0.5 characteristic of the Mickley
and Fairbanks (1) solution. The analytic solution
applied from Appendix C without a gas gap confirmed this

deviation. The importance of a gas gap, however, was
observed only for short contact times since for longer
contact times the numerical solution approached the same
value with different particle sizes, various gas gaps,
and, although poorly, for no gap at all (Figure VI.B.3).
To demonstrate the importance of the concept of

reduced gas conductivity as opposed to a gas gap, the
1000
E
M&F(1) model
3mic. gap
21mic. gap
25mic. gap
100 I 1 I 1 1 I 1 I
0.010 0.100 1.000 10.000

Time (s)
Figure (VI.B.3). Heat transfer coefficient as a function of contact time
with the heated surface for M&F (1) model and the numerical
solution by gap width for 0.18mm diameter glass beads.
rn
C11
66
conductivity calculated from Eq. V.B.1 was applied in
the vicinity of the contact point by defining a narrow
region separating the particle from the wall with
reduced gas conductivity. The effect on the solution,
however, was insignificant in comparison to
implementation of a gas gap of equal width. The
additional complexity of modeling reduced gas conduction
was therefore judged unnecessary.
VI.C. Empirical Model
The major deficiency of the present analytic and
numerical methods as well as solutions from other
authors is that no changes in particle separation from
the wall influenced by mass flow rate are assumed. If a
change in flow properties is responsible for existence
of a boundary layer of fluid stagnant at the wall, a
maximum heat transfer coefficient is obtained when this
fluid film is all that is heated. The importance of the
gas gap is negligible once heating extends far into the
bed. It is therefore possible to fit data by the
present calculational methods assuming a constant width
gas gap, or, in the experimental sense, the fluid film
at the wall reaches a constant width independent of
velocity before heating extends no further than the film

itself. Unfortunately, judging by the conflicting
results often obtained between independent researchers,

67
a single criteria does not seem to hold for all
experimental setups, and cannot be generally used.
Experimental data show good agreement with the
analytic solution for heat transfer by Mickley and
Fairbanks (1) at long contact times although the
observed slope is somewhat different than predicted by
their equation (Figure VI.C.1). Great discrepancy is
observed for short contact times where experimental data
level off. To know this point of separation is
important so that the maximum heat transfer coefficient
may be predicted.
In order to obtain some general correlation for the
heat transfer coefficient, the existing correlations
and experimental data were examined for vertically
oriented heating elements provided by Ernst (9), Desai
(21), Sullivan (3), and this present study. The
experimental equipment used by Wunschmann (12) and
Harakas and Beatty (2) was studied. Intuitively one
would expect that moving bed coefficients be rather
similar when compared in the same system of reference.
The difference in heat transfer coefficients observed
by various researchers when compared at identical
contact times (see Chapter IV) appeared to be influenced
from the fluid dynamic behavior specific for each given
arrangement.
The choice of parameters that were considered

1000
44.
heater length L-------0.1524m

o glass dp=0.044mm
glass dp=0.18mm
100 A glass dp=0.8mm
Mickley 8c Fairbank's(1) for glass
50
0.010 0.100 1.000 10.000
Time (s)
Figure (VI.C.1 ). Comparison between heat transfer coefficients as a function
of contact time with the heated surface for the present
data and the M&F (1) model.
69
important and their combination brought interesting
results. It is still not possible to offer the complete
analytical solution to this problem, however, and only a
simple general technique is proposed for vertically
oriented heaters (rods or plates) when gravity flow of

solids is not obstructed by either heated or supporting
parts. The simple mass flux equation developed predicts
empirically the departure point from the Mickley and
Fairbanks (1) equation
tcr
0.3622 + 9.691 L (VI.C.1)
d pp s
where
L = heated length (m).
d = particle diameter (m).
ps = solid density (kg/m3).
tcr = critical or separation time (s).
A graphical approximation of the heat transfer
coefficient about the inflection point is simple.
1. Calculate the critical time from equation
V.C.1.
2. Using this time calculate the maximum heat
transfer coefficient from the average Mickley
and Fairbanks (M&F) (1) model
keCpspb
hmax = 2]
Irtcr
70
3. Draw a horizontal line at hmax till it
intercepts the M&F line.
The heat transfer coefficient curve is expected to
follow the M&F line and gradually depart to the value of
hmax, the curvature about the inflection point
increasing with particle radius. A correlation for the
time the curve assumes the hmax value is given by
tmax
0.29087 + 0.0492 ln(L) (VI.C.2)
dp p s
The curve for the heat transfer coefficient appears
to level off at tmax, but further extrapolation is
impossible. It is expected that the heat transfer
coefficient decreases at velocities above those now
measured due to an increase in porosity when solids flow

by gravity. For particle sizes below 0.044mm or
smaller, problems with solids flow will occur because of
high interparticle forces. The heat transfer
coefficient is therefore expected to drop and none of
the existing models can predict its value.
A demonstration of the graphical method with
experimental data of Ernst (9), Desai (21), and this
study are given respectively in Figures (VI.C.2 through
VI.C.5) of this chapter.
The results by Harakas and Beatty (2) could not be
fitted to the correlations since in their experimental

71
system solids did not flow by gravity. The maximum
value of heat transfer coefficient did not occur,
however their data follow the expected trend at lower
contact times as presented in Figure (VI.C.6).

v 0.5-0.7mm sand
v 0.3-0.5mm sand
1000 o 0.1-0.2mm sand
M&F(1) model
7- + predicted by eq.
(VI.C.1)
V
(-N
100 I 1 liti I ; I I 11111 I I I I 11111 I 4 14411

0.010 0.100 1.000 10.000 100.000
Time (s)
Figure (VI.C.2). Empirical prediction of tcr by Eq. (VI.C.I) and Ernst's (9)
data.
o 0.8mm sand
v 0.16mm sand
1000 M&F(1) model
predicted by Eq.
(VI.C.1)
100 1 1 1 111111 I 1 1 111411 1 1 411111 1 i 11111

0.010 0.100 1.000 10.000 100.000
Time (s)
Figure (VI.C.3) Empirical prediction of tcr by Eq. (VI.C.1) and Desai's
(21) data for glass.
0.625mm copper
0-0.15mm copper
M&F(1) model
1000 -r predicted by Eq. (VI.C1)
818
00
0* 8-0
T 0
N
E E
E
0
0
100 I I 114Ill I I !III I I 114111 I i 14111

0.010 0.100 1.000 10.000 100.000
Time(s)
Figure (VI.C.4). Empirical prediction of tcr by Eq. (VI.C.I) and Desai's
(21) data for copper.
1000
Tmax
A
N .
E
E
AA
E E
_c co E
glass dp=0.18mm ,--- co
100 A glass dp=0.8mm 0
L.
L_
0
L_
AA
0 0
50 I I limo I I 1 1 I 1 I I
0.010 0.100 1.000 10.000

Time (s)
Figure (VI.C.5). Empirical prediction of tcr by Eq. (VI.C.1) and tmax by Eq.
(V1.C.2) for the present study.
.147 mm glass H&B (2)
O
1000 V
V
411%
O .044 mm glass present study

A 0.8mm glass present study
t A 1.0mm glass Wunschmann (12)
0 3.1 mm glass Wunschmann (12)
v .5.7 mm sand Ernst (9)
.3.5 mm sand Ernst (9)
0 .1 .2 mm sand Ernst (9)
10
0.001 0.010 0.100 1.000 10.000 100.000
Time (s)
Figure (VI.C.6). Comparison between heat transfer coefficients by contact
time with the heated surface for Ernst's (9), Wunschmann's
(12), Harakas (2) and the present data.
77
VII. CONCLUSIONS
VII.A. Summary of Findings
The heat transfer coefficient between a single
vertical tube and a moving bed of solids in air at
atmospheric pressure was measured and compared to
measurements performed by independent researchers
characteristic of a number of different experimental

configurations. The gas and solid properties, contact
time with the heated surface, and experimental
configuration were observed to have the greatest
influence on the heat transfer coefficient. The
distinct improvement of heat transfer from vertically
oriented tubes as compared to horizontal tubes could
also be explained.
Simple models of heat transfer to packed beds such
as the Mickley and Fairbanks (1) model assume the bed
may be modeled with effective properties. As particle
size increases and residence time decreases, however,
the measured heat transfer coefficient gradually assumes
a maximum value, contrary to the model. More accurate
predictions of the heat transfer coefficient assume a
contact resistance exists between the heated surface and

the bed. Analytic and numerical models applied in this
research confirm this resistance must exist.
A static gas adsorption layer on the particle

78
surface and a flow dependent particle separation or gas
gap from the wall may both contribute to an observed
contact resistance. High speed photographic studies of

flowing packed beds performed by the author show an
insignificant porosity change occurs up to bed
velocities of 0.5m/s but a measure of gas gap was
impossible. Since inter-particle separation and
movement appeared rare, heat transfer at high velocities
is most influenced by changes in the wall porosity
distribution and thickness of the gas gap layer.
Computer simulation of heat transfer to a packed bed
showed a steady width gas gap of 21Am for 0.8mm and 15Am
for 0.18mm glass beads was capable of fitting
experimental data.
On the assumption that the heat transfer
coefficient is flow dependent, an empirical correlation
was fitted and predicts the time of departure from the
Mickley and Fairbanks (1) equation independent of heat
transfer properties of the bed.
tcr
0.3622 + 9.691 L (VI.C.1)
dp p s
This correlation is the only of its kind which can
predict the departure point from the Mickley and
Fairbanks (1) equation to data obtained by the
researchers evaluated here for various heated lengths.

79
VII.B. Further work
To demonstrate the existence of a velocity
dependent gas film about the boundary to a flowing
packed bed a more refined experiment using high speed
photography should be attempted. Since the distance
from the wall the particles could remain in focus was
much greater than the gas gap expected to fit data by
computer solution, a fiber optics plate should be used
to couple the image from the surface of a flowing packed

bed. The extremely short range of focus characteristic
from the optical surface of a fiber optics plate may
enable photographic observation of a gas film.
The behavior of the heat transfer coefficient
requires further study due the shortage of measurements
once a maximum value occurs. Using particle sizes on
the order of one half millimeter or more and high
conductivity solids such as copper would ensure an
observation of early departure from the Mickley and
Fairbanks (1) equation. The residence time may also be
decreased by increasing the flow rate of the particle
bed and decreasing the heated length of the surface.
This should allow an opportunity to collect more data to
verify theories regarding heat transfer to flowing
packed beds for velocities greater than the Mickley and
Fairbanks (1) equation applies.

80
BIBLIOGRAPHY
1. Mickley, H. S. and D. F. Fairbanks. "Mechanisms of

heat transfer to fluidized beds", AIChE J., 1 374
(1955).
2. Harakas, N. K. and K. O. Beatty, Jr. "Moving bed

heat transfer: effect of interstitial gas with fine
particles," Chem. Eng. Prog. Svmo. Ser., 59, No.
41, 122-128 (1963).
3. Sullivan, W. N. and R. H. Sabersky. "Heat transfer

to flowing granular media," In J. Heat and Mass
Transfer, 18 97 (1964).
4. Baskakov, A. P. "The mechanism of heat transfer

between a fluidized bed and a surface," Int. Chem.
Eng., 4 320 (1964).
5. Denloye, A. O. O. and J. S. M. Botterill. "Heat

transfer in flowing packed beds," Chem. Eng. Sci.,
32 461 (1977).
6. Spelt, Jan Karel. An Experimental Study of Heat

Transfer to Flowing Granular Media, M. S. Thesis,
California Institute of Technology, Pasadena, CA
(1981).
7. Donskov, S. V. Teploenergetika, In Russian, 11

(1958).
8. Kurochkin, Yu P. Inzhenerno-Fizicheskifi Zhuranal,

In Russian, 4 76 (158).
9. Rudolf, Enrst. "Warmeilbergang and Warmeaustaushern

im Moving Bed," Chemi. Ing. Techn, in German, 32
17 (Jahrel 1960).
10. Colakyan, Manuk. Moving Bed Heat Transfer and

Fluidized Flutriation, PhD Thesis, Oregon State
University (1984).
11. Obutkovid, Nevenka. Heat Transfer Between Moving

Beds of Solids and a Transverse Finned Tube, M. S.
Thesis, Oregon State University (1985).
12. Wunschmann J. and E. U. Schlunder. "Heat transfer

from heated surfaces to spherical packings," Int.
Chem. Eng., 20, No. 4, 555 (October 1980).
81
13. Schlunder, E. U. "Heat transfer to moving packings

at short contact times," Int. Chem. Eng., 20, No.
4, 551 (October 1980).
14. Decker, N. and L. R. Glicksmann. "Heat transfer in

large particle fluidized beds," Int. J. Heat Mass
Transfer, 26, No. 9, 1307 (1983).
15. Brenner, S. S. Surface Sci., 2 496 (1964).
16. Brown, G. G. Unit Operations, J. Wiley, New York,

130 (1966).
17. Takashi, Hiroshi and Hiroshi Yanai. "Flow profile

and void fraction of granular solids in a moving
bed," Powder Technology, 7 205 (1973).
18. Kimura, M. and T. Kaneda. Chem. Eng. Sci, 23 1147

(1968).
19. Benenati, R. F. and C. B. Brosilow. "Void fraction

distribution in beds of spheres," AIChE Journal, 8,
No. 3, 359 (1962).
20. Denloye, A. O. O. PhD Thesis, University of

Birmingham (1976).
21. Desai, M. PhD Thesis, University of Birmingham

(1970).
22. Cumberland, D. J. and R. J. Crawford. The Packing

of Particles, Elsevier Science Publishers B. V.
(1987).
23. Earle, Kenneth H. Kinetic Theory of Gases, McGraw-

Hill Book Company (1938).
24. Patton, J. S., Sabersky R. H., and C. E. Brennen.

"Convective heat transfer to rapidly flowing
granular materials," Int. J. Heat Mass Transfer, 29
1263-1269 (1986).
25. Overbeek, J. T. and M. J. Sparnaay. "London-Van der

Waals attraction between macroscopic objects,"
Faraday Soci-Discussions, 18 12-23 (1954).
26. Derjaguin, B. V. and I. I. Abrikossova. "Direct
measurements of molecular attraction of solids," J.
Phys. Chem. Solids, 5 1-10 (1958).
27. Argo, W. B. and J. N. Smith. "Heat transfer in
packed beds," Chem. Eng. Progress, 49 443 (1953).
82
28. Delvosall, C. and J. Vanderschuren. "Gas to

particle and particle to particle heat transfer in
fluidized beds of large particles," Chem. Eng.
Sci., 40 769-779 (1985).
29. Baddour, R. F. and C. Y. Yoon. "Local radial

conductivity and the wall effect in packed beds,"
Chemical Engineering Progress Svmposition Series
No. 32, 57 35.
30. Borland, D. and D. Geldart. "Electrostatic charging

in gas fluidized beds," Powder Technology, 5 289
(1971/72).
31. Bagnold, R. A. "Experiments on a gravity-free

dispersion of large solid spheres in a Newtonian
fluid under shear," Proc. Roy. Soc. Edin. A, A225
49 (1954).
32. Gabor, John D. "Wall to bed heat transfer in

fluidized and packed beds," Chemical Engineering
Progress Svmposition Series No. 105, 66 76 (1970).
33. Haughey, B. P. and G. S. C. Beveridge. "Local

property variations in a randomly packed bed of
equal-sized spheres," Chem. Eng. Sci., 22 715
(1967).
34. Savage, Stuart B. "Gravity flow of cohesionless

granular materials in chutes and channels," J.
Fluid. Mech., 92 53-96 (1979).
35. Levamax, Grummer Milton. "Heat transfer to gases

through packed tubes," Ind. Eng. Chem., 40 (1948).
36. Brinn, M. S. et al. "Heat transfer to granular
materials," Industrial and Eng. Chemistry, 40 1050
(June 1948).
37. Chandran, R. and Chen J. C. "Influence of the wall

on transient conduction into packed media," AIChE
Journal, 31 168 (1985).
38. Couderc, J. P., H. Angelino, M. Enjalbert et C.

Guiglion, "Echanges themiques en fluidisation
gazeuse. II--Etude mathematique a l'aide d'un
modele simple," In French, Chem. Eng. Sci., 22 99-
107 (1967).
83
39. Mickley, H. S., D. F. Hawthorn, and D. F.

Fairbanks. Chemical Engineering Progress
Symposition Series No. 32, 57 51 (1961).
APPENDICES
84
Appendix A
Analytic Heat Transfer Coefficient for Rod
Given a heated rod of radius a surrounded by a bed
of effective thermal diffusivity a and conductivity k,
the heat flux at the surface of the rod is given by
4kV e-au du
ar2 10 u(J02(ua) + Y02(ua))
where V = T(a,t) T. (1,2).
Since the heat transfer coefficient is given by
q"/V we have
2,
4k 0 e-r^ iva dz
h(t) = (A.1)
ax2 Jo z(J02(z) y02(z))
where
Foa = Fourier number based on rod radius a.
= at/a2.
When averaged over the interval of 0 to Foa the
heat transfer coefficient is
r 2
4k 0 (1 e-"a z) dz
h(t)
aFoar2 1 0 z3(J02(z) + Y02(z))
The indefinite integral above has been solved here
using a 15 point Gauss-Legendre numerical integration
method summed over progressively larger segments of the
integral range until the contribution of the last

segment does not change the final sum by an error of
85
10-6. Answers using this method were comparable within
10% to approximate values tabulated by Jaeger and Clarke
(3).
The following program creates a file of heat
transfer coefficient values for the cylinder using the
numerical method described above and for the Mickley and
Fairbanks model for five decades of time values.
Material properties are read from a file supplied as
input and the heat transfer coefficients are output to a
file called GAUSS.OUT. Rod radius A and average gas
temperature of the bed TGAVE are entered in a data
statement in the program. Effective conductivities of
the bed are calculated using Kunii's equation (4).
ke 1 E
(A.2)
kg kg
+
ks
where
E = porosity of bed.
0 = 0.2 e2.
86
References
1. Jaeger, J. C. "Heat flow in the region bounded

internally by a circular cylinder," Proc. Roy. Soc.
Edin. A., 61 223 (1942).
2. Carslaw, H. S. and J. C. Jaeger. Conduction of Heat

in Solids, Clarendon Press, 334 (1959).
3. Jaeger, J. C. and Martha Clarke. "A short table of
e-xu 2 du
," Proc. Roy. Soc. Edin. A.,
u(J02(u) y02(u))
61 229 (1942).
4. Yagi, S. and D. Kunii. "Studies on effective

thermal conductivities in packed beds," AIChE J.,
3 373 (1957).
87
GAUSST.FOR
Print file "gausst.for" 88
PROGRAM HPLOT
C CONSTRUCTS A PLOT OF AVERAGE HEAT TRANSFER COEFFICIENT VS TIME BY M&F AND
C CYLINDER MODELS
REAL KE,KS,KG
CHARACTER TITL*75,FNAME*30
COMMON FDA
C ROD RADIUS, A (m)
C AVERAGE GAS TEMPERATURE, TGAVE (C)
DATA A,TGAVE /.00635,40./
PI=ACOS(-1.)
C
PRINT *,'Enter property file name'
READ(5,'(A)')FNAME
C
C READ MATERIAL PROPERTIES
OPEN(7,FILE=FNAME)
READ(7,,(A),) TITL
READ(7,*) ROS,ROE,DP,KS,CP
CLOSE(7)
C
C CALCULATE EFFECTIVE ALFA & K
EPS=1.-ROE/ROS
PHI=0.197*EPS**1.904
CPRL=CP*ROE/PI
KG=(TGAVE+273.)*.0000741+.004038
KE=KG*(1.-EPS)/(KG/KS+PHI)
ALFA= KE /(CP *ROE)
C
C WRITE TIME, AVE CYLINDER H, M&F H TO FILE GAUSS.OUT
WRITE(6,15)
OPEN(8,FILE=1GAUSS.OUT')
DO 10 1=1,55
IF(I.LT.10) T=.001*I
IF(I.GE.10.AND.I.LT.20) T=.01+7
IF(I.GE.20.AND.I.LT.30) T=.144
IF(I.GE.30.AND.I.LT.40) T =1 +T
IF(I.GE.40.AND.I.LT.50) T=10+7
IF(I.GE.50) T=100+T
FOA=ALFA/A**2*T
HCYL=(4*KE)/(FOA*A*P1**2)*GAUSS(0.,1E-3,15,1E-4)
HMF =2*SORT(KE*CPRL/T)
WRITE( *,20)T,HCYL,HMF
10 WRITE(8,20)T,HCYL,HMF
15 FORMAT(/' t Hcyl-ave Hmf-ave')
20 FORMAT(3F10.3)
200 END
C
FUNCTION GAUSS(A1,B1,M,EPS)
C CALCULATES INTEGRAL OF INTEGRAND FUNCTN FROM LIMITS Al TO INFINITY
C WITHIN ERROR EPS BY ITERATED M POINT GAUSSIAN INTEGRATION
C B1 IS A SEED VALUE FOR INITIAL INTEGRAL
C
C REFERENCE:
C
C CARNAHAN, BRICE, H. A. LUTHER, AND,JAMES O. WILKES. *APPLIED NUMERICAL
C METHODS*, JOHN WILEY & SONS, 1969, 101-105.
C
EXTERNAL FUNCTN
DIMENSION NPOINT(7),KEY(8),Z(24),WGHT(24)
DATA NPOINT/2,3,4,5,6,10,15/
DATA KEY/1,2,4,6,9,12,17,25/
DATA Z /0.577350269,0.0 ,0.774596669,
1 0.339981044,0.861136312,0.0 ,0.538469310,
2 0.906179846,0.238619186,0.661209387,0.932469514,
3 0.148874339,0.433395394,0.679409568,0.865063367,
89
Print file ogausst.for"
4 0.973906529,0.0 ,0.201194094,0.394151347,
5 0.570972173,0.724417731,0.848206583,0.937273392,
6 0.987992518 /
DATA WGHT / 1.0 ,0.888888889,0.555555556,
1 0.652145155,0.347854845,0.568888889,0.478628671,
2 0.236926885,0.467913935,0.360761573,0.171324493,
3 0.295524225,0.269266719,0.219086363,0.149451349,
4 0.066671344,0.202578242,0.198431485,0.186161000,
5 0.166269206,0.139570678,0.107159221,0.070366047,
6 0.030753242 /
B=B1
A=A1
DO 1 1=1,7
1 IF(M.EQ.NPOINT(I))GOTO 2
GAUSS=0.
WRITE(*,7)M
7 FORMAT('ERROR GAUSS: M IS 2,3,4,5,6,10, OR 15 NOT ',I3)
RETURN
2 N=0
SUM1=0.
3 SUM=0.
JF=KEY(I)
JL=KEY(1+1)-1
C=(B-A)/2.
D =(B +A) /2.
DO 6 J=JF,JL
IF(Z(J).E0.0)SUM=SUM+WGHT(J)*FUNCTN(D)
6 IF(Z(J).NE.0)SUM=SUM+WGHT(J)*(FUNCTN(Z(J)*C+1))+FUNCTN(-2(J)*C+D))
SUM=C*SUM
SUM1=SUM1+SUM
1F(SUM/SUM1.LT.EPS)GOTO 9
C=B
N=N+1
841*(1+SUM1/(4*N*SUM))
A=C
GOTO 3
9 GAUSS=SUM1
RETURN
END
C
FUNCTION FUNCTN(U)
C INTEGRAND FUNCTION TO BE USED BY GAUSS FOR AVERAGE H
COMMON FOA
REAL JO,J1
FUNCTN=( 1-EXP(-FOA*U**2))/(U**3*(J0(U)**2+Y0(U)**2))
RETURN
END
C
C POLYNOMIAL APPROXIMATIONS TO JO & YO BESSEL FUNCTIONS
C
C REFERENCE:
C
C ABRAMOWITZ, M. AND I. E. STEGAN, ED., *HANDBOOK OF MATHMATICAL FUNCTIONS*,
C US DEPT. OF COMMERCE, AMS 55, 1970, 369-370.
C
REAL FUNCTION J0(X)
1F(X.LE.3.0.AND.X.GE.-3.0) THEN
B=(X/3)**2
J0=1-2.24999970*B +1.2656208*B**2-.3163866*B**3+.04444790*B**4
& -.003944400*B**5+.00021000*B**6
ELSE1F(X.GE.3.0) THEN
JO=FO(X)*COS(PHIO(X))/SQRT(X)
ELSE
WRITE(*,*) 'ILLEGAL ARGUMENT CALL TO JO:',X
ENDIF
90
Print file mgausst.for*
RETURN
END
C
FUNCTION YO(X)
REAL JO
IF(X.GT.O..AND.X.LE.3.0) THEN
B=(X/3)**2
YO=ALOG(X/2)/ACOS(0.)*J0(X)+.36746691
& +.605593660*B -.74350384*B**2+.25300117*8**3-.04261214*B**4
& +.004279160*8**5-.00024846*B**6
ELSEIF(X.GE.3.0) THEN
YO=F0(X)*SIN(PHIO(X))/SORT(X)
ELSE
WRITE(*,*) 'ILLEGAL ARGUMENT CALL TO YO:',X
ENDIF
RETURN
END
C
FUNCTION FO(X)
B=3/X
FO= .797884560 -.00000077*B -.00552740*B**2-.00009512*11**3
& +.001372370*13**4-.00072805*B**5+.00014476*8**6
RETURN
END
C
FUNCTION PHIO(X)
B=3/X
PHIO=X-.78539816 -.04166397*8 -.00003954*B**2+.00262573*B**3
& -.000541250*8**4-.00029333*B**5+.00013558*8**6
RETURN
END
91
Appendix B
Heat Transfer Coefficient for Two Region
Semi-infinite Slab
Consider the composite semi-infinite slab with
constant temperature wall at x=-L and zero initial
temperature as shown in Figure (B.1). The temperature
gradient at x=-L as a function of time has been listed
by Carslaw and Jaeger (1).
aT -V c o - ln 2 alt)
/(]
[ 1 + 22 e-(nL)
8x x=-L 'wait n=1 a + 1
where a = (k2 /k1),/al /a2 and V is the temperature
difference at the wall. In terms of the heat transfer
coefficient at the wall this becomes
1(1 co
2
h(x,t) 1 + 22 cn e-n /Fol. (B.1)
'wait n=1
where
FoL = alt/L2 and
a 11
c
la + 1 j
For very small or large FoL numbers, the solution may be
simplified since
co
2 cn e-n2/F0L 0
n=1 small FoL
c/(1-c)
large FoL
92
Two region composite semiinfinite slab
T(x,0) = 0
Region Region T( L,t)= V
1 2
x L 0 co >
Figure (B.1). Representation of composite slab.

93
Then
k1
h = h1
/wait small FoL
k2
h = h2
iwa2t large FoL
So that h1 and h2 are solutions for the Mickley and
Fairbanks equation for region 1 and 2, respectfully. A
good approximation for the heat transfer coefficient for
moderate values of FOL. is given by
h = k1 /L.
When B.1 is plotted on a log-log scale against time, the
heat transfer coefficient can be identified to form
three portions: a linear portion where the coefficient
behaves only by heat transfer to region 1, a transition
region where heat transfer develops into region 2, and a
second linear portion once region 1 completes heating
and the temperature gradient in region 2 is most
significant.
94
References
1. Carslaw, H. S. and J. C. Jaeger. Conduction of Heat

in Solids, Clarendon Press, 319-322 (1959).
95
Appendix C
Heat Transfer from Wall to Particle
Through Interstitial Fluid
An approximation for heat transfer from a heated
wall to a spherical particle for low heating times may

be obtained by a special treatment of B.1 over a slab
with a hemispherical interface. Assume as usual slab
geometry and heat conduction only perpendicular to the
wall is significant. Then using B.1 integrate by length
1 over the hemispherical interface by height z over 0 to
r (Fig. C.1) to find the average
h 1(1 [1+2; cn I/:-n2(Ax+r(1-cos9))2/ait cos8d8]

/wait n=1 0
1(1 m 2
1+22 cn en (6+1-cos8) 2 /Fo r cos8d8 (C.1)
/wait n=1 0
where For = alt/r2, S = Ax/r and c is as defined in
Appendix B.
Equation C.1 cannot be integrated further
analytically, however a 6 point Gauss-Lengendre

numerical integration method obtains a sufficiently
accurate solution. Agreement between the equation C.1
and the finite difference treatment described in
Appendix D is quite good for short contact times.

96
Gas Particle
Figure (C.1). Representation of separation distance

between a spherical particle and wall.
97
The asymptotic values of h for very low or high For
numbers are the same as for the composite semi-infinite

slab. In most cases if not all the gas conductivity is
much less than that of the solid and the gas thermal
diffusivity is greater than the solid, therefore the
value of c is very close to unity. A good approximation
for h at moderate values of For may then be obtained.
Assume c = 1 then
0 72 2 2
h/hi = 1+22 e-n (6+1-cosi?) /Fo r cos0d0
n=1 0
By use of the Euler-Maclaurin summation formula we can
say
e-P2 = e-P2dp - 1/2

p =1 0 2
Then
2; If/:-n2(6+1-cos0)2/For cosOd8 =
n=1 10
2
112 %-n2(6+1-cosO)2/Fordn cos0d8 =
0 0
w/2 I iFor
2 l'e-k2 dk 1 /2J =
0 16+1-cos8 0 J
cos0d0
iwFor - 1
0 6+1-cos0
And finally
6+2
h/hi = iirFor Arctan -
6+2 6 6 2
or
98
k1 w
h = Arctan (C.2)
r 6+2 6 6 2
This accounts for the plateau value of h for moderate
values of For and is independent of the solid
properties (c=1). This equation could be used to
account for a gas gap resistance term.
As seen from heat transfer prediction by the finite
difference treatment for a particle array from the wall,
equation C.1, C.2, and the finite difference treatment
agree well at the time required for heat to propagate to
one half of the particle diameter.

99
HCMSG6.FOR
Print file "heesg6.form
100
PROGRAM HCMPS
CHARACTER*30 RAMIE
C COMPUTES THE HEAT TRANSFER COEFFICIENT TO A COMPOSITE SOLID WITH
C HEMISPHERICAL INTERFACE
COMMON N,R,ALFAT,WIDTH
DATA EPS/.0001/
PI=ACOS(-1.)
C READ INPUT PARAMETERS AND ECHO INPUT
C CK1 = CONDUCTIVITY OF GAS
C ALFA1 = THERMAL DIFFUSIVITY OF GAS
C CK2 = CONDUCTIVITY OF REGION SOLID
C ALFA2 = THERMAL DIFFUSIVITY OF SOLID
C WIDTH = WIDTH OF GAP SEPARATING SOLID FROM WALL
C D = PARTICLE DIAMETER
C FNAME = T,H OUTPUT FILE NAME
READ(5,*)CKLALFALCK2,ALFA2,WIDTH,D,FNAME
WRITE(6,45)CKLALFALCK2,ALFA2,WIDTH,D,FNAME
C PRELIMINARY CALCULATIONS
SIGMA=CK2/CK1*SORT(ALFAl/ALFA2)
Cm(SIGMA-1)/(SIGMA+1)
R=0/2
C COMPUTE & PRINT TIME DECADE VALUES OF THE HEAT TRANSFER COEFFICIENT
OPEN(8,FILE=FNAME)
DO 30 1=1,60
IF(1.LT.10) T=.00001*I
IF(I.GE.10.AND.I.LT.20) T=.0001+7
IF(I.GE.20.AND.I.LT.30) T=.001+T
IF(I.GE.50) T=1.+T
ALFAT=ALFAl*T
N=1
SUM=C*GAUSS(0.,P1/2,6)
C SUM TERMS UNTIL CONTRIBUTING TERM ADDS LESS THAN EPS TO SUMNEW
20 N=N+1
SUMNEW=SUM+C**N*GAUSS(0.,P1/2,6)
IF(ABS(SUMNEW-SUM).LE.EPS*ABS(SUM)) GOTO 25
SUM=SUKNEW
GOTO 20
25 H=CK1/SORT(Pl*ALFAT)*(1+2*SUMNEW)
WRITE(8,50) T,H
30 WRITE(*,50) T,H
CLOSE(8)
45 FORMAT('1K1 = ',1PE12.4,' ALFA1 = ',E12.4/' K2 = ',E12.4,
ALFA2 = ',E12.4/' WIDTH = ',E12.4,' DIAMETER =',E12.4/
'
S OUTPUT FILE = ',A)

'
50 FORMAT(' ,1P2E12.4)
END
FUNCTION FUNCTN(X)
C INTEGRAND TO AVERAGE HEAT TRANSFER OVER FACE OF PARTICLE
COMMON N,R,ALFAT,WIDTH
FUNCTN=EXP(-(N*(R*(1-COS(X))+WIDTH))**2/ALFAT)*COS(X)
RETURN
END
FUNCTION GAUSS(A,B,M)
C CALCULATES INTEGRAL OF INTEGRAND FUNCTN FROM LIMITS A TO B
C BY M POINT GAUSSIAN INTEGRATION
C
C REFERENCE:
C
C CARNAHAN, BRICE, N. A. LUTHER, AND JAMES O. WILKES. *APPLIED NUMERICAL
C METHODS*, JOHN WILEY & SONS, 1969, 101-105.
C
101
Print file "hoseg6.for*
EXTERNAL FUNCTN
DIMENSION NPOINT(8),KEY(9),2(40),WGHT(40)
DATA NPOINT/2,3,4,5,6,10,15,32/
DATA KEY/1,2,4,6,9,12,17,25,41/
DATA Z /0.577350269,0.0 ,0.774596669,0.339981044,
1 0.861136312,0.0 ,0.538469310,0.906179846,
2 0.238619186,0.661209387,0.932469514,0.148874339,
3 0.433395394,0.679409568,0.865063367,0.973906529,
4 0.0 ,0.201194094,0.394151347,0.570972173,
5 0.724417731,0.848206583,0.937273392,0.987992518,
6 .0483076656,.1444719615,.2392873622,.3318686022,
7 .4213512761,.5068999089,.5877157572,.6630442669,
8 .7321821187,.7944837959,.8493676137,.8963211557,
9 .9349060759,.9647622555,.9856115115,.9972638618/
DATA WGHT /1.0 ,0.888888889,0.555555556,0.652145155,
1 0.347854845,0.568888889,0.478628671,0.236926885,
2 0.467913935,0.360761573,0.171324493,0.295524225,
3 0.269266719,0.219086363,0.149451349,0.066671344,
4 0.202578242,0.198431485,0.186161000,0.162269206,
5 0.139570678,0.107159221,0.070366047,0.030753242,
6 .0965400885,.0956387200,.0938443990,.0911738786,
7 .0876520930,.0833119242,.0781938957,.0723457941,
8 .0658222227,.0586840934,.0509980592,.0428358980,
9 .0342738629,.0253920653,.0162743947,.0070186100/
DO 1 1=1,8
1 IF(M.EO.NPOINT(I))GOTO 2
WRITE(*,7)M
7 FORMAT('ERROR GAUSS: M IS 2,3,4,5,6,10,15 OR 32 NOT ',I3)
RETURN
2 GAUSS=0.
JF=KET(I)
JL=KEY(14,1)-1
SUM =O.
C=(B-A)/2.
02(B+A)/2.
DO 6 J=JF,JL
IF(Z(J).E0.0)SUM=SUM+WGHT(J)*FUNCTN(D)
6 IF(Z(J).NE.0)SUM=SUM+WGHT(J)*(FUNCTN(Z(J)*C+D)+FUNCTN(-Z(J)*C+0))
GAUSS=C*SUM
RETURN
END
102
Appendix D
Finite Difference Calculation of
Heat Transfer Coefficient to Packed Beds
D.1. Summary
The programs reviewed in this appendix have
advantages over previous moving packed bed heat transfer

programs. Similar methods of numerical analysis to
determine the heat transfer coefficient to a bed of
solids have been used by researchers since Botterill and
Williams (2) however these methods incorporate a
comparably slow relaxation method of temperature
convergence to solve the unsteady state heat transfer

equation. Unlike the earlier programs, a simpler and
apparently more accurate finite difference
representation for the temperature gradient at the
heated wall is used to calculate the heat flux. Both
programs, one each to treat slab and cylindrical
geometries, are unique in that they allow any node in
the map to assume a property (i.e. conductivity,
density, and specific heat) defined by the user. No
particular pattern of node and property arrangement

needs to be followed. Finally, the method's speed
facilitates longer simulation times and larger map
sizes.
103
D.2 Provisions
1. One of ten individually defined properties may be
selected for each node.
2. A choice of ten node dimensions for the axial and
radial directions may be separately defined.
3. The time step size and frequency of output summary
listings may be changed as the program proceeds.
4. The results of a run may be copied to a binary dump
file from which a run may be restarted at a later

time.
5. The instantaneous values of the heat transfer
coefficient may be saved to a file which may be
read by a plotting program.
6. A map size of up to 50x50 node elements may be
defined. This is limited only by the maximum array
size accommodated by the MS FORTRAN compiler.
7. The implicit alternating direction (IAD) method has
been employed with a tridiagonal Gaussian
elimination matrix solution to ensure minimum time.

104
D.3. Model Assumptions
1. Material properties may vary in two dimensions.
For the slab material properties may vary in both
vertical and horizontal directions whereas for the
cylinder materials may vary radially and axially
but not azimuthally.
2. Node sizes may vary in two dimensions.
3. Temperature at the heated wall remains constant.
4. The bed configuration is symmetric at the top and
bottom boundaries of the map so that its axial
temperature distribution has a mirror image about
these boundaries. Therefore the temperature
gradient is defined as zero here.
5. To model the right hand boundary as if it extended
infinitely, the zero temperature gradient is

defined there.
6. The transient heat conduction equation is solved
for temperatures relative to the initial
temperature of the bed and so the bed is
represented with a zero temperature difference at

105
time zero.
7. Material properties do not change with time or

temperature.
D.4. Derivation of Finite Difference Equations
A finite difference solution of the transient heat
conduction equation requires a discrete representation

of temperatures at the intersection points of a grid of
nodes, each node interior modeling a different two
dimensional position or material property. To each node
and its adjacent nodes is applied a finite difference
expression of the transient heat conduction equation and
boundary condition equations are added to treat nodes at
boundaries of the map. Using either a relaxation or
matrix solver method the equations are solved
simultaneously for all node temperatures at small time
increments beginning from an initial temperature state,
each time step chosen small enough to ensure convergence

of the solution. A heat transfer coefficient local to
each node at the "heated" boundary is calculated from
the temperature gradient and node conductivity there.
The local heat transfer coefficient is then averaged and

recorded. The development and programming of the finite
difference equations, the coefficients of the IAD

solution, and heat transfer coefficient calculation is
106
presented here.
D.4.1 Representation
Figure (D.1) illustrates the map used to represent
position and property in cylindrical or slab geometries.
The map has M-1 cells and M interior points radially or
horizontally. A node may be distinguished as a cell and
a point. Cells outside and adjacent to the zero
temperature gradient boundaries are indicated by dashed
lines and possess properties identical to their
neighbors across the boundaries. At the right hand
boundary the temperature gradient is approximated by
aT TkN TkN+1
ax AxN
This requires TkN = TkN+1 for a zero temperature
gradient. The top and bottom boundaries could be
treated similarly but a more accurate representation
would account for symmetry about the boundary and so we
treat temperatures on either side of the boundary as
equivalent. The boundary condition equations are then
1. Left: Tko
= Twall
2. Right: TkN
= TkN+1
3. Top: Tm_ij = Tm+ij
4. Bottom: Toj = T2j (D.1)
Consider the dashed region at the intersection of
four nodes as shown in Figure (D.2). At each one
quarter node region adjacent to the intersection

T11
4 TM+ IN
--...)
Top reflecting boundary ,,,.._ Ti M

Tm
T Wit
M Ia 104-4 MN
II- II II-12 W
WI II- IN
Heated wall
k = 1..M M-1 cells

M points
k
j = 1..N N cells
N points
I
32 MI- I 3N
I 214- I 2N T
T ) CI" IN+ I
Bottom reflecting boundary .-----"- --4-1

T TON
01
Figure (0.1). Finite difference map to model wall-to-bed heat transfer.

108
T
k+1,j+1
k+1
2 1
(k+1,j) (k+1,j+1) Cell

Dx
k+1
T 1 T 1
k,j- i k,j T k,j+1

k D G
Node
Dx
k
L 1
(k,j) (k,j+1)
3 T 4
k-1,j
j-1 j+1
Dr Dr
j+1
Figure (D.2). Example of cell noding scheme.

109
temperature is assumed single valued. Each cell and its
characteristic property is identified by number or by
indices k,j. The transient heat conduction equation
aT
VkVT = pCp4
will be integrated over the equivalent node volume
represented by the dashed region.
VkVT dV = pCpdV (D.2)

V at v
To simplify the analysis, the divergence theorem will be
applied to the LHS of D.2.
VkVT dV = kVTn dS (D.3)

V
where n is a unit outward normal.
D.4.2. Cylinder
The RHS of D.3 may be separated into radial and
axial components
kVTndS = kVTerdS + kVTexdS (D.4)
where er and ex are radial and axial unit vectors,

respectfully. Evaluate the integral for the finite
difference terms of the LHS of D.2 using D.3 and D.4 as
substitutions
110
Tkj_1
kVTerdS k2
Arj 2
+k3
2 fi
1(1
Arj+1 2
+k4
2
Tk+lj-Tkj
kVTexdS [1(2[7frj2 w(rj-Arj/2)2]
Axk+1
+k1 [x(rj+Arj+1/2)2-irrj21]
k3[wrj2 x(rj-Arj/2)2]
Axk
+k4[x(rj+Arj+1/2)2-xrj211
After some simplification,
1 kVTerdS = x[aTkj+1 + bTkj_i (a+b)Tkj]
(D.5)
kVT.e dS w[cTk+lj dTk-lj (c+d)Tkj]
where
(klAXki.1 + k4AXk)(rj + Arj+1/2)

a
Arj+1
(k2Axic+1 + k3Axk)(rj - Arj/2)

b
Arj
111
k2Arj(rj - Arj/4) + k1Arj.1.1(rj + Arj+1/4)

c
"k+1
k3Arj(rj - Arj/4) + k4Arj.1.1(rj + Arj+1/4)

d
Axk
Evaluate now the integral term on the RHS of D.2.
pCpdV = -.(piCplAxkAArj+I(rj + Arj+1/4) +

-71
1v 2
p2Cp2AxkilArj (rj - Arj/4) +
p4Cp4Axk Arj+1(rj + Arj+1/4) +
p3Cp3Axk Arj (rj Arj/4))
= r (At/2) e (D.6)
where
e = [(p1CplAx10.1 + p4Cp4Axk)Arj4.1(rj + Arj+1/4) +
(p2Cp2Axki.1 + p3Cp3Axk)Arj (rj - Arj/4))]/At.
Combining D.5 and D.6 yields the finite difference
equation for the interior nodes of the cylinder
aT
--af(At/2)e = aTkj+1 + bTkj_i (a+b)Tkj +
cTk+ij + dTk_ij (c+d)Tkj (D.7)
The same form of equation set D.7 is used at the
map boundaries only the constant terms a, b, c, d, and e
may be simplified since cells immediately across the

boundary have like properties.
Top: k = M
a = 2k4Axm(rj + Arj41/2)/Arj.1.1
b = 2k3Axm(rj - Arj/2)/Arj
112
c = d
k3Arj(rj Arj/4) + k4Arj.1.1(rj + Arj+1/4)

d=
Axk
e = 2Axm(p4Cp4Arj+1(rj+Arj44/4)+
p3Cp3Arj (rj-Arj/4) )/At
Bottom: k = 2
a = 2k1Ax2(rj + Arj+1/2)/Arj+1
b = 2k2Ax2(rj - Arj/2)/Arj
k2Arj(rj - Arj/4) + k1Arj+1(rj + Arj+1/4)

C=
Axk+1
d = c
e = 2Ax2(piCp1Arj+1(rj+Arj+1/4)+
p2Cp2Arj (rj-Arj/4) )/At
Bottom right: k = 2, j = N
b = 2k2Ax2(rN ArN/2)/ArN
c = 2k2ArNrN/Ax2
d = c
e = 4rjp2Cp2Ax2ArN/At
Top right: k = M, j = N
b = 2k3AxM(rN ArN/2)/rN
c = 2k3ArNrN /AxM
d = c
e = 4rjp3Cp3AxmArN/At
113
Right: j = N
(k2Axki.1 + k3Axk)(rj - Arj/2)

b
Arj
c = 2k2ArNrN /Axk +1
d = 2k3ArNrN/Axk
e = 2rjArN[p2Cp2Axk.4.1 + p3Cp3Axk]/At
D.4.3. Slab
The derivation of the finite difference equations
for the slab is similar. Separate the RHS of D.3 into
vertical and horizontal components
kVTndS = kVTexdS + kVTeydS (D.8)
where ey and ex are vertical and horizontal unit
vectors, respectfully. Evaluate each term of D.8
kVTexdS = (K2AYk+1 k3AYk)

S 2Axj
Tkj+i-Tkj
(k1AYk+1 k4AYk)
2 xj+1
j kVT.e dS (klAxj+1 + k2Axj)

2AYk+1
(k4Axj+1 + k3Axj).
After simplification
114
1 kVT.exdS = [aTkj+1 + bTkj_i (a+b)Tkj]/2

S
(D.9)
1 kVTeydS = (cTk+ij + dTk_ij (c+d)Tkj]/2
where
a = (klAyk +1 k4Ayk) /Axj +1
b = (k2Ayk +1 k3AYk)/Axj
c = (klAxj.1.1 + k2AxWAYk+1
d = (k4Axj+1 + k3Axj)/Ayk.
The LHS of D.2 as applied to slab geometry is
pCpdV = (P1CP1AYk+1Axj+1 + P2CP2AYk+1Axj +

.1 V
p3Cp3AykAxj + P4CP4AYkAxj+1)/4
= (At/2)e/2 (D.10)
where
e = (P1CNAYki-lAxj+1 P2CP2AYk+1Axj
p3Cp3AykAxj P4CP4AYkAxj+1)/At
Once D.9 and D.10 are combined for the finite difference
approximation for the slab we obtain D.7. Finally, the
constant simplifications at the slab boundaries are.
Top: k = M
a = 2k4AyM /Axj +l
b = 2k3AyM /Axj
c = d
k3Axj + k4Axj+1
d
Aym
115
e = 2Aym(p4Cp4Axj+1 + p3Cp3Axj)/At
Bottom: k = 2
a = 2k1Ay2/Axj+1
b = 2k2Ay2/Axj
c = (k1Axj+1 + k2Axj)/Ay2
d = c
e = 2Ay2(p1CplAxj+1 + p2Cp2Axj)/At
Bottom right: k = 2, j = N
b = 2k2Ay2/AxN
c = 2k2AxN/AY2
d = c
e = 4p2Cp2AxNAy2/At
Top right: k = M, j = N
b = 2k3Aym/AxN
c = 2k3AxN/Ay2
d = c
e = 4p3Cp3AxNAym/At
Right: j = N
b = (k2Ayk+1 + k3Ayk)/AxN
c = 2k2AxN/Ay k+1
d = 2k3AxN/Ayk
e = 2AxN(p2Cp2Ayk+1 + p3Cp3Ayk)/At
116
D.5. Implicit Alternating Direction (IAN_
Solution Method
To solve the system of five point difference
equations defined by D.7 for nodal temperatures the IAD
algorithm is employed (1). The IAD method requires
significantly less computation time and array size than
Gaussian elimination or Gauss-Siedel methods of matrix
inversion. The method uses two difference equations to
represent temperatures in a single dimension. The
equation for each dimension applies for one half a time
step, which the method repeatedly applies for a solution
of all successive time steps. This enables tridiagonal
matrix solutions separately implicit by row (k) or
column (j). The finite difference equations for the IAD
method applied to heat conduction in a cylinder
originate from
Tkji+1 Tkj*
pCpdV =
At/2 V
kVTkj*-erdS + kVTkji.4.1-exdS
1S S
Tkj* - Tkji
pCpdV =
At/2 V
kVTkj*erdS + kVTkjiexdS
where time step subscript i indicates temperatures prior

117
a time step, 1+1 indicates temperatures following a full
time step, and * references temperatures at one half a
time step.
This becomes, after use of 0.7,
-aTkj+1*-bTkj_1*+(a+b+e)Tkj*
= [cTki.lj+dTk_ij-(c+d-e)Tkj]i (D.11a)
-aTkj.4.1*-bTkj_1*+(a+b-e)Tkj*
= [cTki.lj+dTk_ij-(c+d+e)Tkj]i+1 (D.11b).
The method first solves the map for temperatures at one

half time step (LHS of D.11a) from initial temperatures
(RHS of D.11a), then temperatures are solved at the full
time step (RHS of D.11b) from the half time step
temperature map. The coefficients of D.11 may be
generalized into the upper, lower, and diagonal terms of
a tridiagonal matrix.
UkTki4.1*-1-1-kTki_1*-1-41Tkj*
= [UjT1(4.1j+LiTk_lji-DilTkj]i (D.12a)
UkTkj+1*-akTki_1*-FplaTki*
= (D.12b)
where
Uk = -a, a + b + e
U.
J
= c, Dji = -(c + d e)
Lk = -b, Dk2 = a + b - e
L.
J
= d, Dj2 = -(c + d + e)
At the map boundaries, use of boundary equations
D.1 into D.12 requires modification of these constants.

118
4.
a b *
DJ .1 -4.
d e)
-(c
tk -b, 0kz . a *b e
c DJz ci I. e)
. (c
8ottom: ft 2
tIL
A 0ki a 'b *
UJ c 1. d, DJ 1 -(c 'A'
d e)
tk --.
_b, Okz a b e
Zi 0 D '(c * d * e)
*4
4
01.0 4.-
J .
Sotto ight.
0
k
0 00 -, b* e
U
j
tk
*4
..-
-b,
* d 0j1
D kz
.,
.
-(c *
b _
d e)
tJ 0, D. 2 . '(c I. d * e)
Top J .
ght:
U
k '0?si 00 . b *
U ' 0,
.
D.
JI
.,
-(c e)
J
tk ..... -b, 0kz . b e
li c * d 0,12 "(c 1.
d * e)
Right: j ...
Uk
' 0, k.1 b
119
Uj = c, Dj1 = -(c + d e)
Lk = -b, Dk2 = b e
Lj = d, Dj2 = -(c + d + e)
D.6. Heat Transfer Coefficient Calculation
The previous sections covered finite difference
formulation of the transient heat conduction equation,
boundary conditions representation, and coefficient
definition. This section will explain calculation of
the local and average heat transfer coefficient at the
heated wall from the temperature solution. As with the
assumptions outlined earlier, both properties and axial
or radial step size may vary in this solution.

The local heat transfer coefficient is given by hk
= q"k/Vo where Vo is the temperature difference from the
wall to the initial bed temperature and q"k is the heat
flux at the wall for cell k. Refer to Figure (D.3)
which is a representation of axial nodes and cells
adjacent to the heated wall. If
aT (Tko Tkl)
cl uk=-1(
ar Arl
where conductivity k is an average for the properties of
the cell separating Tko and Tkl, then the heat flux term
at the wall is
rk0 Tkl kk +1,lAxk +1 kk,lAxk

qui(
Ari Axk+1 Axk
120
M
hM
Dxm
M-1
Dxm_i -1I
M H
1- --1
3 h3
Dx3
3 1-
2 h2
Dx2
h1
I I
1
Dr1 Dr2
Figure (D.3). Finite difference scheme at wall to

obtain axial average heat transfer
coefficient.
121
Since half height cells lie at the top and bottom of the
map and these have a constant conductivity, the heat
flux there is
(T10 T11)
= k21
Ari
(TMO TM1)
qM = kMl
Ari
For an average h about the map height we use
M
E hk (Axk + Ax10.1)/2
have (fLh dx)/L k=1
L
where
M
L = 2 Ax
k=2
and for the symmetric top and bottom boundary cells Axi
= Ax2 and Axm = Axm+1
D.7. Verification
Both slab and cylinder geometry codes were verified
repeatedly for temperature and/or heat transfer
coefficient calculation with the following analytic

methods:
1. Homogeneous properties for slab and cylinder
with large inner radius specified.
2. Cylinder with homogeneous properties
(Eq. A.1).
122
3. Two region composite slab for slab and
cylinder with large inner radius (Eq. B.1,
C.1).
4. Steady state solutions for slab and cylinder
chosen for large time solutions and mixed
materials.
D.8. Input Selection
This section introduces constraints which when
applied to input selection ensure the numerical model
performs well under finite node size and region length
assumptions. This treatment is subjective but has
demonstrated to be effective.
D.8.1. Length of Region
The numerical solution assumes a satisfactory
result for the heat transfer coefficient may be obtained
by considering a region of finite width with the
temperature gradient set to zero at the boundary. As
shown previously, the temperature gradient at the
boundary will eventually be non-zero. The question then
arises to what accuracy would we expect the numerical
solution to have for the region length as a function of

123
time. Consider the region to have homogeneous
properties, then the temperature gradient at the heated

surface for the semi-infinite slab is
aT -V -V
ax x=0
where
L = radial length of region.
FoL = at/L2.
t = total time of solution.
The temperature gradient at the heated wall for a finite
slab with insulated boundary is
aT 2V m 2
_ E e-at[(2n+1)71./(2L)]
x x=0 L n=0
2V m 2
_ E e-at[FoL((2n+1)112) ]
L n=0
To calculate a heat flux equivalent to either model we
would like these temperature gradients equal, or as
close as possible with a relative error c so

00
. 1 _ w E e-ff[w(2n+1)/4)2]
n=0
where w = 2/71.FoL. This function has been plotted and
appears in Figure (D.4). For an error of at most 10%,
then, we would choose FoL = .33 or less. Our criteria
is then to choose an FoL so that Fol. < FoL(c) where
FoL(c) was calculated with the equation above for error

E.
1.000
0.100
0.010
0.001
0.000 0.200 0.400 0.600
FoL
Figure (D.4). Error introduced by modeling a semi-infinite slab as a finite
slab.
125
If the region is not homogeneous (as all numerical
solutions of interest for h will be) then the region's
effective properties are used and the above criteria is
still a good assumption since it is known that as the
region is heated it becomes increasingly accurate to
model it with effective properties (hence the assumption
that by heating three particle diameters from the heated
surface the region may be treated as effective).
D.8.2. Choice of Step Size and Simulation Time
The numerical treatment of the transient heat
conduction equation depends upon dividing the heat
conducting region into cells small enough such that the
temperature gradients may be reasonably approximated as
the slope of temperature points separated by a cell.
Due to the thermal diffusivity characteristic of a heat
conducting material, heat requires a finite time to
propagate. Cell size should be chosen for some length
less than the distance heat will propagate over one time
step otherwise temperature gradients will not be
calculated correctly. With regard to the temperature
gradient at the heated surface, there is some time and
cell size for which the temperature gradient cannot be
calculated correctly once the solution begins from

initial state. If the first cell is small compared to
the entire region we could assume the temperature

126
gradient as the node begins to heat is given by
aT -V
ax x=0 ixat
which is that of the semi-infinite slab. Numerically,
the temperature gradient could be represented as
aT -V
ax x Ax
however an improved numerical approximation for the
derivative is:
aT -V
ax x 2Ax
Set the analytic and numerical derivative together
as
-v -v
ixatd 2Ax
and solve
4/1. = Fod
where Fod = atd/Ax2. Since smaller Ax values lead to a
better representation for the gradient our criteria is
Fod < 4/x. This criteria should be applied to the cell
with the smallest size and largest thermal diffusivity

to determine td, the maximum time step of the solution.
A time step less than td should be chosen to ensure
accuracy of the solution. An increase in time step is
however a possibility once cells of minimum td value

127
have been heated long enough such that the temperature
difference across them has become negligible.
If the time step size is significantly less than
the minimum td value of all cells adjacent to the heated
wall then this characteristic value represents the
minimum simulation time whereafter the solution first

becomes correct. To demonstrate this criteria, the
numerical method was applied to calculate h for a semi-
infinite slab with constant material properties--a
solution which is known analytically. When arranged in
the following non-dimensional form
hAx/k = 1/(iwFod)
the Mickley and Fairbanks equation should appear as a
straight line when the LHS is plotted log-log against
Fod. Figure (D.5) shows both numerical and analytic
results agree well only when Fod is greater than 1, or
near the criteria value of 4/w. When Fod is less than
the criteria value h approaches the constant value of
k/Ax. The numerical solution may also be shown to
depart significantly from the analytic result upon this
figure when the FoL number is not satisfied.
D.8.3. Selection of Cell Number
Since porosity contour is only approximated by
position of block shaped cells, the smaller the cell
size and the larger the number of cells used to

10.000
1.000
0.010
0.001
0.01 0.10 1 .00 10.00 100.00
Fod
Figure (D.5). Accurary of heat transfer solution to a semi-infinite slab on
Fod value.
129
represent the curve should improve the approximation.
This approximation to curvature may lead to an
exaggeration of particle surface in contact with the
wall and therefore a corresponding increase in heat
transfer coefficient should be observed. The number of
cells per particle radius was increased from 5 to 10,
20, 40, 60, and 80, and the heat transfer curves were
compared. When no gap was introduced, the descent of
the heat transfer curve from a plateau value occurred at
progressively earlier times as cell number increased.
The data could be made to fit well this manner but
usually with an inordinately large map size. Once a gap
was introduced, then for as small a gap as .001dp a cell
number of 5 to 10 per particle radius produced less than
20% difference in heat transfer.
Because the curvature of boundaries separating
curved regions of unlike properties can only be
approximated by placement of box-like cells, for cells
that intersect the boundary an attempt to weight by cell
area on either side of the boundary was made. A
difference in heat transfer solutions with and without
curvature correction was 1% or less for particle
representations of 5 cells per radius. It therefore
appears the additional complexity for curvature
correction in programs by previous researchers including

finite element modeling is unjustified (2,3). For the
130
degree of particle heating required to compare
calculations of the heat transfer coefficient to data
obtained for as short as 0.1s a number greater than 10
cells per particle radius does not appear to be
required.
D.9. Program Utilities
To ease the chore of preparing an input map
representative of porosity change with distance from the
heated wall, the program MAKMAP was devised. This
program produces a map of equally dimensioned square
cells with property indices of 0 (void) and 1 (solid)
for an array of particles. The number of cells used to
represent a particle, the number of contact points to a
particle, and the number of particles extending in the
radial direction is selected on input. MAKMAP writes
its completed map to the file MAKMAP.OUT which is later
appended to an input file for the heat transfer
calculation codes.
Since the porosity by distance formulation (Eq.
V.D.1) is to be used to select a discrete number of void
or solid cells in the input map, an average porosity for

cell width Au is
w[1-Au((u-1)+Au/3)-(u-1)2]
E(u) = 1
n tan(r /n)
where an n of 6 was chosen to model orthorhombic arrays

131
of particles.
MAKMAP displays the number of cells chosen for void
and solid using the cell average porosity defined above
as well as the solid fraction apportioned to a cell when
the number of solid cells does not divide equally into a
column.
D.10. Sample Run
The following input file, 385-21.txt, was prepared
to compute the heat transfer coefficient to a packed bed
of 0.8mm glass spheres in air through a 21Am gap. This
solution applied the program for cylindrical geometry
with a rod radius of 6.35mm. The input map was
generated by the MAKMAP utility for five nodes per
particle radius and an orthorhombic configuration. For
a simulation time to lOs the criteria on Fogy (section
D.8.1) required a total radial length of 2.24mm and the
Fod criteria (section D.8.2) required a td no greater
than 20As limited by the gap and a 50ms time step
limited by the 80Am node size beyond the gap. A gradual
increase in time step from 10As to 20ms while the gap
heated was chosen to maintain stability. The output
listing is presented as 385-21.out and a plot file h185-
21 and binary restart file b185-21 was also generated.
A solution time of about 5min was required on an IBM AT.

132
D.11. Program Listings
Two versions of the heat transfer coefficient
program BLAUSIS appear here to treat cylindrical and
slab geometries for compilation on an IBM AT with MS
FORTRAN. Source files with increased array sizes to
treat a 130 x 130 node map for compilation on a high
level workstation or mainframe were used but are not
listed here. The program utility MAKMAP follows these

listings.
133
References
1. Carnahan, Brice, H. A. Luther, and James O. Wilkes.

Applied Numerical Methods, John Wiley & Sons,
452-453 (1969).
2. Botterill, J. S. M. and J. R. Williams. "The

mechanism of heat transfer to gas-fluidised beds,"
Trans. Instn. Chem. Engrs., 41 217 (1963).
3. Opaliriski, Irenenz and Andrezej Wolny.

"Discontinuity of particle contact with the surface
and heat transfer in fluidized beds," Int. J. Heat
Mass Transfer, 30 589-599 (1987).
134
385-21.TXT
Print file "385-21.txt"
135
Glass: 0.8mm diameter (21micron gap, 6 contacting particles)
4,",,h185-21',1b185-21'
100.,0.00635,6,31,1,2,2
1,80e-6
8,80e-6
9,21e-6
1,1.04,27E5
0,2.9282E-2,1.047E3
100,1E-5,100,1,5
100,1E-4, 50,1,5
100,1E-2, 20,1,5
500,2E-2, 50,1,5
1111111111111111111111
9888888888888888888888888888E188
0000000000000000000000000000000
0000111100000011110000001111000
0001111110000111111000011111100
0011111111001111111100111111110
0111111111111111111111111111111
136
385-21.0UT
Print file "385-21.out"
137
1 Glass: 0.8mm diameter (21micron gap, 6 contacting particles)
INPUT REVIEW
CENTER TEMPERATURE 1.000000E+02

INNER RADIUS 6.350000E-03
NUMBER OF AXIAL NODES 6
NUMBER OF RADIAL NODES 31
NUMBER OF AXIAL STEP CARDS READ 1
NUMBER OF RADIAL STEP CARDS READ 2

NUMBER OF PROPERTY CARDS READ 2
NUMBER OF TIME STEP CARDS READ 4
H PLOT FILE NAME h185-21
DUMP FILE NAME b185-21
AXIAL STEP CARDS

ID DX
1 8.000000E-05
RADIAL STEP CARDS

ID DR
8 8.000000E-05
9 2.100000E-05
PROPERTY CARDS
ID K RHO*CP
1 1.040000E+00 2.700000E+06
0 2.928200E-02 1.047000E+03
TIME STEP CARDS

NUMBER DT NPRT MSKP NSKP
100 1.00000E-05 100 1 5
100 1.00000E-04 50 1 5
100 1.00000E-02 20 1 5
500 2.00000E-02 50 1 5
AXIAL NODE SIZE (RIGHT=TOPMOST)

LOCATION MAP (BY ID)
RADIAL NODE SIZE

9888888888888888888888888888888
PROPERTY
0000000000000000000000000000000
0000111100000011110000001111000
0001111110000111111000011111100
0011111111001111111100111111110
0111111111111111111111111111111
TOTAL RADIAL LENGTH 2.421000E-03

138
TOTAL AXIAL LENGTH 4.000000E-04
H SOLUTION FOR 0.000000E+00 TO 9.999999E-04 SECONDS
PRINT TEMPERATURE MAP EVERY 100 1.00000E-05 SECOND TIME STEPS FOR EVERY 5 RADIAL NODE
TEMPERATURE MAP AT 1.00000E-03 SECONDS

5 10 15 20 25 30
0.163523E+01 0.291312E-02 0.446713E-05 0.126211E-080.219265E -12 0.119535E-16
0.530756E-02 0.143233E-02 0.311287E-08 0.433062E-09 0.669578E-16 0.304987E-17
0.185718E-03 0.372636E-03 0.336079E-10 0.752225E-10 0.369566E-18 0.392623E-18
0.733127E-05 0.631582E-06 0.356047E-12 0.467077E-13 0.220466E-20 0.139556E-21
0.384397E-06 0.440424E-08 0.338205E-14 0.149723E-15 0.126558E-22 0.274069E-24
0.139207E-06 0.727768E-10 0.555285E-16 0.127009E-17 0.131883E-24 0.148844E-26
HEAT FLUX
6 1.44204E+04 1.44204E+02
5 1.58051E+04 1.58051E+02
4 2.11873E+04 2.11873E+02
3 3.93284E+04 3.93284E+02
2 1.35841E+05 1.35841E+03
1 1.37648E+05 1.37648E+03
AVERAGE H @ 1.00000E-03 SECONDS: 5.76392E+02

H SOLUTION FOR 1.000001E-03 TO 1.100000E-02 SECONDS

5 10 15 20 25 30
0.182894E+01 0.540537E-02 0.421630E-04 0.119690E-06 0.899429E-09 0.284641E-11
0.942900E-01 0.317573E-02 0.147280E-05 0.694811E-07 0.221190E-10 0.173407E-11
0.256942E-01 0.106132E-02 0.288327E-06 0.229563E-07 0.345031E-11 0.635835E-12
0.615280E-02 0.244420E-04 0.415443E-07 0.429799E-09 0.403584E-12 0.103282E-13
0.164876E-02 0.213678E-05 0.449349E-08 0.304368E-10 0.358686E-13 0.583607E-15
0.913877E-03 0.404577E-06 0.749565E-09 0.473469E-11 0.498381E-14 0.749022E-16
HEAT FLUX
6 1.42125E+04 1.42125E+02
5 1.55701E+04 1.55701E+02
4 2.07618E+04 2.07618E+02
3 3.77074E+04 3.77074E+02
2 1.25058E+05 1.25058E+03
1 1.29893E+05 1.29893E+03
AVERAGE H 2 6.00000E-03 SECONDS: 5.42299E+02

5 10 15 20 25 30
0.194243E+01 0 .654737E-02 0.542806E-04 0.168729E-06 0.134212E-08 0.462154E-11
0.193550E+00 0 .400456E-02 0.381001E-05 0.100923E-06 0.761917E-10 0.288512E-11
0.871689E-01 0 .139218E-02 0.133204E-05 0.343168E-07 0.237566E-10 0.108248E-11
0.363011E-01 0 .498395E-04 0.356513E-06 0.102372E-08 0.570343E-11 0.309104E-13
0.172559E-01 0 .733603E-05 0.741354E-07 0.132179E-09 0.106355E-11 0.368455E-14
0.126157E-01 0 .234188E-05 0.233500E-07 0.380641E-10 0.303402E-12 0.102026E-14
139
Print file "385- 21.out"
HEAT FLUX
6 1.41404E+04 1.41404E+02
5 1.54730E+04 1.54730E+02
4 2.05379E+04 2.05379E+02
3 3.68349E+04 3.68349E+02
2 1.19558E+05 1.19558E+03
1 1.24556E+05 1.24556E+03

H SOLUTION FOR 1.099999E-02 TO 1.011000E+00 SECONDS

5 10 15 20 25 30
0.839442E+01 0.213273E+01 0.706441E-01 0.645534E-02 0.127650E-03 0.320485E-04
0.682873E+01 0.211021E+01 0.392352E-01 0.565404E-02 0.378581E-04 0.212877E-04
0.704141E+01 0.203370E+01 0.420421E-01 0.470824E-02 0.339188E-04 0.115880E-04
0.751188E+01 0.192834E+01 0.499976E-01 0.377332E-02 0.355434E-04 0.265703E-05
0.792600E+01 0.162199E+01 0.582741E-01 0.295621E-02 0.397742E-04 0.199767E-05
0.808665E+01 0.149063E+01 0.617622E-01 0.262738E-02 0.418705E-04 0.180481E-05
HEAT FLUX
6 1.30884E+04 1.30884E+02
5 1.39466E+04 1.39466E+02
4 1.68068E+04 1.68068E+02
3 2.62980E+04 2.62980E+02
2 9.17444E+04 9.17444E+02
1 9.50055E+04 9.50055E+02

5 10 15 20 25 30
0.143643E+02 0.603602E+01 0.544655E+00 0.142961E+00 0.586427E-02 0.103238E-02
0.129852E+02 0.607037E+01 0.461220E+00 0.141221E+00 0.374912E-02 0.945396E-03
0.132070E+02 0.615719E+01 0.487548E+00 0.136749E+00 0.399892E-02 0.801009E-03
0.136732E+02 0.631815E+01 0.541364E+00 0.131102E+00 0.463995E-02 0.625992E-03
0.140757E+02 0.561624E+01 0.591694E+00 0.111346E+00 0.528709E-02 0.531744E-03
0.142306E+02 0.530060E+01 0.612065E+00 0.102810E+00 0.555674E-02 0.495032E-03
HEAT FLUX
6 1.17613E+04 1.17613E+02
5 1.27768E+04 1.27768E+02
4 1.61496E+04 1.61496E+02
3 2.56405E+04 2.56405E+02
2 8.48541E+04 8.48541E+02
1 8.78371E+04 8.78371E+02

5 10 15 20 25 30
0.192658E+02 0.967169E+01 0.146836E+01 0.561026E+00 0.436791E-01 0.147317E-01
140
0.179964E+02 0.975233E+01 0.135292E+01 0.561731E+00 0.357365E-01 0.142605E-01

0.182098E+02 0.999791E+01 0.140437E+01 0.563035E+00 0.377593E-01 0.130657E-01
0.186498E+02 0.104637E+02 0.150454E+01 0.567931E+00 0.420459E-01 0.113504E-01
0.190263E+02 0.948774E+01 0.159610E+01 0.499322E+00 0.461113E-01 0.102599E-01
0.191705E+02 0.904117E+01 0.163279E+01 0.468805E+00 0.477660E-01 0.982647E-02
HEAT FLUX
6 1.10059E+04 1.10059E+02
5 1.19468E+04 1.19468E+02
4 1.52399E+04 1.52399E+02
3 2.45926E+04 2.45926E+02
2 7.97041E+04 7.97041E+02
1 8.24954E+04 8.24954E+02
AVERAGE H a 6.11000E-01 SECONDS: 3.56468E+02

5 10 15 20 25 30
0.233939E+02 0.129212E+02 0.268311E+01 0.122671E+01 0.147306E+00 0.698999E-01
0.222066E+02 0.130374E+02 0.254887E+01 0.123357E+01 0.131264E+00 0.685382E-01
0.224113E+02 0.134019E+02 0.262021E+01 0.125325E+01 0.136888E+00 0.646679E-01
0.228279E+02 0.141094E+02 0.275619E+01 0.129216E+01 0.148198E+00 0.587605E-01
0.231816E+02 0.129358E+02 0.287917E+01 0.115711E+01 0.158692E+00 0.549348E-01
0.233167E+02 0.123944E+02 0.292823E+01 0.109608E+01 0.162924E+00 0.534048E-01
HEAT FLUX
6 1.04433E+04 1.04433E+02
5 1.13070E+04 1.13070E+02
4 1.44284E+04 1.44284E+02
3 2.35220E+04 2.35220E+02
2 7.54326E+04 7.54326E+02
1 7.80682E+04 7.80682E+02
TEMPERATURE MAP AT 1.01100E+00 SECONDS

5 10 15 20 25 30
0.269205E+02 0.158291E+02 0.406690E+01 0.206824E+01 0.337267E+00 0.194580E+00
0.258003E+02 0.159730E+02 0.392109E+01 0.208339E+01 0.312461E+00 0.191920E+00
0.259973E+02 0.164271E+02 0.400763E+01 0.212910E+01 0.322863E+00 0.184044E+00
0.263938E+02 0.173190E+02 0.417048E+01 0.221787E+01 0.343298E+00 0.171618E+00
0.267285E+02 0.160009E+02 0.431674E+01 0.200937E+01 0.362052E+00 0.163561E+00
0.268559E+02 0.153899E+02 0.437492E+01 0.191423E+01 0.369581E+00 0.160329E+00
NEAT FLUX
6 9.96644E+03 9.96644E+01
5 1.07758E+04 1.07758E+02
4 1.37409E+04 1.37409E+02
3 2.25446E+04 2.25446E+02
2 7.18070E+04 7.18070E+02
1 7.43116E+04 7.43116E+02
AVERAGE H a 1.01100E+00 SECONDS: 3.22015E+02

H SOLUTION FOR 1.010999E+00 TO 1.101100E+01 SECONDS

141

5 10 15 20 25 30
0.389641E+02 0.267877E+02 0.115123E+02 0.741560E+01 0.265672E+01 0.215755E+01
0.380884E+02 0.270066E+02 0.113426E+02 0.748026E+01 0.258900E+01 0.214728E+01
0.382581E+02 0.276903E+02 0.114659E+02 0.767136E+01 0.262429E+01 0.211608E+01
0.385873E+02 0.290002E+02 0.116906E+02 0.801383E+01 0.269160E+01 0.206633E+01
0.388597E+02 0.273695E+02 0.118888E+02 0.748211E+01 0.275242E+01 0.203295E+01
0.389623E+02 0.266068E+02 0.119670E+02 0.723538E+01 0.277667E+01 0.201951E+01
HEAT FLUX
6 8.32472E+03 8.32472E+01
5 8.91588E+03 8.91588E+01
4 1.10315E+04 1.10315E+02
3 1.75859E+04 1.75859E+02
2 5.95000E+04 5.95000E+02
1 6.15820E+04 6.15820E+02
AVERAGE H 2 2.01100E+00 SECONDS: 2.63973E+02

5 10 15 20 25 30
0.461754E+02 0.341275E+02 0.181977E+02 0.130701E+02 0.664325E+01 0.593733E+01
0.454156E+02 0.343585E+02 0.180384E+02 0.131585E+02 0.655788E+01 0.592252E+01
0.455668E+02 0.350907E+02 0.181705E+02 0.134343E+02 0.660805E+01 0.587667E+01
0.458557E+02 0.365294E+02 0.184077E+02 0.139588E+02 0.670289E+01 0.580091E+01
0.460924E+02 0.348587E+02 0.186151E+02 0.132531E+02 0.678825E+01 0.575142E+01
0.461812E+02 0.340749E+02 0.186967E+02 0.129239E+02 0.682222E+01 0.573149E+01
HEAT FLUX
6 7.31152E+03 7.31152E+01
5 7.89236E+03 7.89236E+01
4 9.91509E+03 9.91509E+01
3 1.57515E+04 1.57515E+02
2 5.23076E+04 5.23076E+02
1 5.41314E+04 5.41314E+02

5 10 15 20 25 30
0.511833E+02 0.396173E+02 0.239617E+02 0.184888E+02 0.114184E+02 0.106307E+02
0.504964E+02 0.398468E+02 0.238147E+02 0.185860E+02 0.113290E+02 0.106140E+02
0.506343E+02 0.405729E+02 0.239460E+02 0.188936E+02 0.113852E+02 0.105622E+02
0.508956E+02 0.420162E+02 0.241801E+02 0.194982E+02 0.114912E+02 0.104751E+02
0.511085E+02 0.403882E+02 0.243840E+02 0.187274E+02 0.115864E+02 0.104188E+02
0.511882E+02 0.396234E+02 0.244640E+02 0.183671E+02 0.116242E+02 0.103962E+02
HEAT FLUX
6 6.61235E+03 6.61235E+01
5 7.14761E+03 7.14761E+01
4 9.05122E+03 9.05122E+01
3 1.45222E+04 1.45222E+02
2 4.73704E+04 4.73704E+02
1 4.90177E+04 4.90177E+02

5 10 15 20 25 30
0.550495E+02 0.440793E+02 0.290579E+02 0.235920E+02 0.164337E+02 0.156318E+02
142
0.544175E+02 0.443010E+02 0.289215E+02 0.236907E+02 0.163457E+02 0.156146E+02

0.545448E+02 0.450008E+02 0.290482E+02 0.240034E+02 0.164031E+02 0.155616E+02
0.547850E+02 0.463990E+02 0.292732E+02 0.246282E+02 0.165110E+02 0.154717E+02
0.549802E +02 0.448436E+02 0.294688E+02 0.238504E+02 0.166080E+02 0.154139E+02
0.550531E+02 0.441125E+02 0.295454E+02 0.234864E+02 0.166465E+02 0.153906E+02
HEAT FLUX
6 6.08847E+03 6.08847E+01
5 6.57637E+03 6.57637E+01
4 8.34536E+03 8.34536E+01
3 1.35163E+04 1.35163E+02
2 4.35852E+04 4.35852E+02
1 4.50984E+04 4.50984E+02

5 10 15 20 25 30
0.582683E+02 0.479229E+02 0.336729E+02 0.283819E+02 0.214066E+02 0.206232E+02
0.576817E+02 0.481340E+02 0.335457E+02 0.284784E+02 0.213221E+02 0.206064E+02
0.578000E+02 0.487995E+02 0.336663E+02 0.287834E+02 0.213783E+02 0.205545E+02
0.580229E+02 0.501321E+02 0.338799E+02 0.293982E+02 0.214838E+02 0.204660E+02
0.582036E+02 0.486598E+02 0.340654E+02 0.286415E+02 0.215785E+02 0.204093E+02
0.582711E+02 0.479675E+02 0.341381E+02 0.282873E+02 0.216162E+02 0.203865E+02
HEAT FLUX H
6 5.65474E+03 5.65474E+01
5 6.10378E+03 6.10378E+01
4 7.74608E+03 7.74608E+01
3 1.26274E+04 1.26274E+02
2 4.04476E+04 4.04476E+02
1 4.18506E+04 4.18506E+02
AVERAGE H 2 6. 01099E+00 SECONDS: 1.81355E+02

5 10 15 20 25 30
0.610848E+02 0 .513571E+02 0.379165E+02 0.328738E+02 0.262035E+02 0.254535E+02
0.605378E+02 0 .515567E+02 0.377977E+02 0.329662E+02 0.261234E+02 0.254373E+02
0.606483E+02 0 .521851E+02 0.379116E+02 0.332581E+02 0.261772E+02 0.253876E+02
0.608560E+02 0 .534451E+02 0.381132E+02 0.338489E+02 0.262783E+02 0.253026E+02
0.610243E+02 0 .520581E+02 0.382881E+02 0.331259E+02 0.263690E+02 0.252482E+02
0.610872E+02 0 .514058E+02 0.383566E+02 0.327874E+02 0.264051E+02 0.252263E+02
HEAT FLUX H
6 5.27496E+03 5.27496E+01
5 5.69189E+03 5.69189E+01
4 7.22031E+03 7.22031E+01
3 1.18170E+04 1.18170E+02
2 3.77096E+04 3.77096E+02
1 3.90169E+04 3.90169E+02
AVERAGE H 2 7. 01099E+00 SECONDS: 1.69169E+02

5 10 15 20 25 30
0.636252E+02 0.544923E+02 0.418530E+02 0.370E49E+02 0.307673E+02 0.300564E+02
0.631139E+02 0.546802E+02 0.417418E+02 0.371726E+02 0.306917E+02 0.300411E+02
0.632172E+02 0.552715E+02 0.418489E+02 0.374491E+02 0.307427E+02 0.299939E+02
0.634113E+02 0.564579E+02 0.420386E+02 0.380102E+02 0.308385E+02 0.299133E+02
0.635686E+02 0.551544E+02 0.422031E+02 0.373256E+02 0.309245E+02 0.298616E+02
0.636272E+02 0.545412E+02 0.422675E+02 0.370051E+02 0.309587E+02 0.298408E+02
143
HEAT FLUX
6 4.93126E+03 4.93126E+01
5 5.32038E+03 5.32038E+01
4 6.74636E+03 6.74636E+01
3 1.10672E+04 1.10672E+02
2 3.52439E+04 3.52439E+02
1 3.64653E+04 3.64653E+02
AVERAGE M a 8.01099E+00 SECONDS: 1.58152E+02

5 10 15 20 25 30
0.659567E+02 0.573892E+02 0.455220E+02 0.410324E+02 0.350784E+02 0.344082E+02
0.654782E+02 0.575656E+02 0.454179E+02 0.411151E+02 0.350073E+02 0.343937E+02
0.655749E+02 0.581210E+02 0.455185E+02 0.413757E+02 0.350554E+02 0.343493E+02
0.657565E+02 0.592357E+02 0.456966E+02 0.419052E+02 0.351457E+02 0.342732E+02
0.659037E+02 0.580122E+02 0.458510E+02 0.412602E+02 0.352268E+02 0.342244E+02
0.659585E+02 0.574367E+02 0.459115E+02 0.409582E+02 0.352591E+02 0.342048E+02
HEAT FLUX
6 4.61579E+03 4.61579E+01
5 4.97985E+03 4.97985E+01
4 6.31271E+03 6.31271E+01
3 1.03692E+04 1.03692E+02
2 3.29828E+04 3.29828E+02
1 3.41258E+04 3.41258E+02
AVERAGE H a 9.01102E+00 SECONDS: 1.48031E+02

5 10 15 20 - 25 30
0.681174E+02 0.600835E+02 0.489507E+02 0.447326E+02 0.391357E+02 0.385057E+02
0.676692E+02 0.602491E+02 0.488531E+02 0.448103E+02 0.390690E+02 0.384920E+02
0.677598E+02 0.607703E+02 0.489476E+02 0.450553E+02 0.391143E+02 0.384503E+02
0.679299E+02 0.618164E+02 0.491147E+02 0.455533E+02 0.391992E+02 0.383787E+02
0.680676E+02 0.606688E+02 0.492595E+02 0.449471E+02 0.392754E+02 0.383325E+02
0.681190E+02 0.601289E+02 0.493163E+02 0.446632E+02 0.393057E+02 0.383144E+02
HEAT FLUX
6 4.32272E+03 4.32272E+01
5 4.66393E+03 4.66393E+01
4 5.91110E+03 5.91110E+01
3 9.71654E+03 9.71654E+01
2 3.08885E+04 3.08885E+02
1 3.19588E+04 3.19588E+02
AVERAGE M a 1.00110E+01 SECONDS: 1.38642E+02

5 10 15 20 25 30
0.701302E+02 0 .625986E+02 0.521592E+02 0.482006E+02 0.429469E+02 0.423554E+02
0.697103E+02 0 .627539E+02 0.520678E+02 0.482736E+02 0.428843E+02 0.423425E+02
0.697952E+02 0 .632426E+02 0.521563E+02 0.485036E+02 0.429268E+02 0.423033E+02
0.699545E+02 0 .642237E+02 0.523130E+02 0.489713E+02 0.430065E+02 0.422361E+02
0.700836E+02 0 .631477E+02 0.524488E+02 0.484022E+02 0.430780E+02 0.421931E+02
0.701317E+02 0 .626416E+02 0.525020E+02 0.481357E+02 0.431065E+02 0.421758E+02
HEAT FLUX
6 4.05065E+03 4.05065E+01
5 4.36997E+03 4.36997E+01
4 5.53763E+03 5.53763E+01
144
3 9.10601E+03 9.10601E+01
2 2.89379E+04 2.89379E+02
1 2.99406E+04 2.99406E+02
AVERAGE H Q 1.10111E+01 SECONDS: 1.29894E+02
Fortran STOP
145
HSLB4PC.FOR
Print file "hslio4pc.for" 146
PROGRAM BLAUSIS
C CALCULATES TEMPERATURES AND HEAT TRANSFER COEFFICIENTS FOR TWO DIMENSIONAL SLAB
C VARIABLE NODE SIZE AND PROPERTIES ALLOWED
C PC VERSION
REAL LJ,LK
DIMENSION T(50,50),ITYPE(50,50),DK1(50,50),DJ2(50,50),
TS(50,50),C2P(50),H(50),IXSZ(50),IYSZ(50),
SIXDX(10),DY(10),DX(10),NSTEP(10),CK(10),ROCP(10),DT(10),NPRT(10),
SNSKP(10),MSKP(10),INDX(10),IYDX(10)
COMMON UK(50,50),LK(50,50),DK2(50,50),UJ(50,50),
SLJ(50,50),DJ1(50,50),F,M,N
CHARACTER TITLE*80,HNAME*40,DNAME*40,RNAME*40
C INPUT PREPARATION
C RECORD 1:
C TITLE - 80 CHARACTER PROBLEM DESCRIPTION
C RECORD 2:
C NTM - NUMBER OF TIME STEP CARDS TO BE READ
C RNAME - RESTART FILE NAME. IF PRESENT READ ONLY TIME STEP CARDS AFTER THIS CARD
C HNAME - H PLOT FILE NAME. IF PRESENT WRITE T,HAVE(T) TO HNAME.
C DNAME - DUMP FILE NAME. IF PRESENT WRITE DUMP FILE TO DNAME.
C RECORD 3:
C CENT - TEMPERATURE @ X=0
C A - NOT USED. PRESENT FOR COMPATIBILITY OF CYLINDER INPUT FILES
C M - TOTAL NUMBER OF VERTICAL NODE ENDPOINTS (=TOTAL NODES - 1)
C N - TOTAL NUMBER OF HORIZONTAL NODES
C NYSZ - NUMBER OF VERT NODE SIZE CARDS TO BE READ
C NXSZ - NUMBER OF HORIZ NODE SIZE CARDS TO BE READ
C NPROP - NUMBER OF PROPERTY CARDS TO BE READ
C READ NYSZ VERT NODE SIZE CARDS:
C IYDX - INDEX OF NODE SIZE (0-9)
C DY - NODE SIZE
C READ NXSZ HORIZ NODE SIZE CARDS:
C IXDX - INDEX OF NODE SIZE (0-9)
C DX - NODE SIZE
C READ NPROP PROPERTY CARDS:
C INDX - INDEX OF PROPERTY (0-9)
C CK - THERMAL CONDUCTIVITY
C ROCP - DENSITY*SPECIFIC HEAT
C READ NTM TIME STEP CARDS:
C NSTEP - TOTAL NUMBER OF TIME STEPS ACTIVE FOR THIS CARD
C DT - TIME STEP SIZE
C NPRT - NUMBER OF TIME STEPS A TEMPERATURE/H COEFF SUMMARY IS PRINTED
C MSKP - PRINT FOR EACH MSKP VERT NODES
C NSKP - PRINT FOR EACH NSKP HORIZ NODES
C READ VERT NODE SIZE MAP FOR M-1 NODES FROM TOP MOST POSITION (1 CARD)
C READ HORIZ NODE SIZE MAP FOR N NODES FROM RIGHT MOST POSITION (1 CARD)
C READ PROPERTY LOCATION MAP (M-I ROWS WITH N COLUMNS)
C END OF INPUT
C READ DATA
READ(5,1) TITLE
WRITE(6,2)TITLE
READ(5,*) NTM,RNAME,HNAME,DNAME
IF(RNAME.E0.")THEN
READ(5,*) CENT,A,M,N,NYSZ,NXSZ,NPROP
WRITE(6,5)CENT, M,N,NYSZ,NXSZ,NPROP,NTM,HNAME,DNAME
C READ AXIAL STEP SIZE CARDS
DO 31 I=1,NYSZ
READ(5,*) IYDX(I),DY(IYDX(I)+1)
31 WRITE(6,7)IYDX(I),DY(IYDX(I)+1)
WRITE(6,32)
C READ RADIAL STEP SIZE CARDS
DO 34 I=1,NXSZ
READ(5,*) IXDX(I),DX(IXDX(I)+1)
34 WRITE(6,7)IXDX(I),DX(IXDX(I)+1)
WRITE(6,36)
147
Print file uhslb4pc.for"
C READ PROPERTY CARDS

DO 3 I=1,NPROP
READ(5,*) INDX(I),CK(INDX(I)+1),ROCP(INDX(I)+1)
3 WRITE(6,7)INDX(I),CK(INDX(I)+1),ROCP(INDX(I)+1)
C READ TIME STEP CARDS
WRITE(6,8)
DO 4 I=1,NTM
READ(5,*) NSTEP(I),DT(I),NPRT(I),MSKP(I),NSKP(I)
4 WRITE(6,9)NSTEP(I),DT(I),NPRT(I),MSKP(I),NSKP(I)
C READ VERTICAL STEP SIZE LOCATION MAP
WRITE(6,37)
CALL WRTHD(M)
READ(5,20) (IYSZ(K),K=M,2,-1)
WRITE(6,21)(IYSZ(K),K=M,2,-1)
C READ RADIAL STEP SIZE LOCATION MAP
WRITE(6,38)
CALL WRTHD(N)
READ(5,20) (IXSZ(J),J=1,N)
WRITE(6,21)(IXSZ(J),J=1,N)
C READ PROPERTY LOCATION MAP
WRITE(6,39)
CALL WRTHD(N)
DO 10 K=M,2,-1
READ(5,20) (ITYPE(K,J),J=1,N)
10 WRITE(6,21)(ITYPE(K,J),J*1,N)
C ADJUST SIZE AND TYPE ID TO AVOID ZERO ARRAY INDEX
DO 11 K=2,M
IYSZ(K)= IYSZ(K) +1
DO 11 J=1,N
11 ITYPE(K,J)=ITYPE(K,J)+1
DO 41 J =1,N
41 IXSZ(J)=IXSZ(J)+1
THE =O.
ELSE
C PERFORM RESTART FROM FILE RNAME
OPEN(8,FILE= RNAME,FORM= 'UNFORMATTED')
READ(8)CENT,A,M,N,NXSZ,NYSZ,NPROP,IXDX,IYDX,DX,DY,CK,ROCP,INDX,
S ITYPE,IXSZ,IYSZ,T,TS,TME
CLOSE(8)
WRITE(6,18) RNAME
WRITE(6,5)CENT,M,N,NYSZ,NXSZ,NPROP,NTM,HNAME,DNAME
DO 42 I=1,NYSZ
42 WRITE(6,7)IYDX(I),DY(IYDX(I)+1)
DO 43 I=1,NXSZ
WRITE(6,36)
DO 13 I=1,NPROP
13 WRITE(6,7)INDX(I),CK(INDX(I)+1),ROCP(INDX(I)+1)
WRITE(6,8)
DO 14 I=1,NTM
READ(5,*) NSTEP(I),DT(I),NPRT(I),MSKP(I)
14 WRITE(6,9)NSTEP(I),DT(I),NPRT(I),MSKP(I)
WRITE(6,37)
CALL WRTHD(M)
WRITE(6,21)(IYSZ(K)-1,K=M,2,-1)
WRITE(6,38)
CALL WRTHD(N)
WRITE(6,21)(IXSZ(J)-1,J=1,N)
WRITE(6,39)
CALL WRTHD(N)
DO 16 K=M,2,-1
16 WRITE(6,21)(ITYPE(K,J)-1,J=1,N)
ENDIF
F=CENT
C SUM INTERVALS TO FIND TOTAL LENGTH
148
Print file "hslb4pc.for*
YLEN=0
DO 17 K=2,M
17 YLEN=YLEN+DY(IYSZ(K))
XLEN=0
DO 19 J=1,N
19 XLEN=XLEN+DX(IXSZ(J))
IF(HNAME.NE.") OPEN(7,FILE=HNAME)
DO 200 L=1,NTM
WRITE(6,23)XLEN,YLEN,TME,DT(L)*NSTEP(L)+TME
WRITE(6,24)NPRT(L),DT(L),NSKP(L),MSKP(L)
C PREPARE TRIDIAGONAL COEFFICIENT MATRIXES
C TREAT WALL AS INSULATED BOUNDARY
C CENTER CONDITIONS FOR CONSTANT TEMPERATURE
DO 40 J=1,N
DO 40 Kz1,M
IF(K.EQ.M)THEN
C TREAT TOP CONDITIONS
IF(J.E0JOTHEN
C TREAT TOP WALL
B=2*CK(ITYPE(M,N))*DY(IYSZ(M))/DX(IXSZ(N))
C=2*CK(ITYPE(M,N))*DX(IXSZ(N))/DY(IYSZ(M))
D=C
E=4*ROCP(ITYPE(M,N))*DX(IXSZ(N))*DY(IYSZ(M))/DT(L)
UJ (M,N)=0
LJ (M,N)=C+0
DJUM,N)=-(C+D-E)
DJ2(M,N)=-(C+D+E)
UK (M,N)=0
LK (M,N)=-B
DK1(M,N)=B+E
DK2(M,N)=B-E
ELSE
C TREAT TOP INTERIOR AND CENTER
A=2*CK(ITYPE(M,J+1))*DY(IYSZ(M))/DX(IXSZ(J+1))
I1=2*CK(ITYPE(M,J ))*DY(IYSZ(M)) /DX(IXSZ(J ))
O=(,CK(ITYPE(M ,J ))*DX(IXSZ(J ))+
CK(ITYPE(M ,J+1))*DX(IXSZ(J+1)))/DY(IYSZ(M ))
D=C
E=2*DY(IYSZ(M))*
(ROCP(ITYPE(M 01+1))*DX(IXSZ(J+1))+
S ROCP(ITYPE(M ,J ))*DX(IXSZ(J )))/DT(L)
UJ (M,J)=0
LJ (M,J)=C+D
DJ1(M,J)=-(C+D-E)
DJ2(M,J)=-(C+D+E)
UK (K,J)=-A
LK (K,J)=-B
DK1(K,J)=A+B+E
DK2(K,J)=A+B-E
ENDIF
ELSEIF(K.E0.1)THEN
C TREAT BOTTOM CONDITIONS
IF(J.EG.N)THEN
C TREAT BOTTOM WALL
B=2*CK(ITYPE(2,10)*DY(IYSZ(2))/DX(IXSZ(N))
O=2*CK(ITYPE(2,N))*DX(IXSZ(N))/DY(IYSZ(2))
D=C
E=4*ROCP(ITYPE(2,N))*DX(IXSZ(N))*DY(IYSZ(2))/DT(L)
UJ (1,J)=C+D
LJ (1,J)=0
DJ1(1,J)=-(C+D-E)
DJ2(1,J)=-(C+D+E)
UK (K,N)=0
LK (K,N)=-B
DK1(K,N)=B+E
149
Print file "Inlb4pc.for"
DK2(K,N)=B-E
ELSE
A=2*CK(ITYPE(K+1,J+1))*DY(IYSZ(2))/DX(IXSZ(J+1))
B=2*CK(ITYPE(K+1,J ))*DY(IYSZ(2))/DX(IXSZ(J ))
C=(CK(ITYPE(K+1,J ))*DX(IXSZ(J ))+
CK(ITYPE(K+1,J+1))*DX(IXSZ(J+1)))/DY(IYSZ(2 ))
D=C
E=2*DY(IYSZ(2 ))*
(ROCP(ITYPE(K+1,J+1))*DX(IXSZ(J+1))+
S ROCP(ITYPE(K+1,J ))*DX(IXSZ(J )))/DT(L)
UJ (1,J)=C+0
LJ (1,J)=0
DJ1(1,J)=-(C+0-E)
0J2(1,J)=-(C+0+E)
UK (K,J)=-A
LK (K,J)=-B
DK1(K,J)=A+11+E
DK2(K,J)=A+B-E
ENDIF
ELSE
C TREAT VERTICAL INTERIOR CONDITIONS
IF(J.E0.11)THEN
C TREAT VERTICAL INTERIOR WALL
B=(CK(ITYPE(K+1,J ))*DY(IYSZ(K+1))+
CK(ITYPE(K ,J ))*DY(IYSZ(K )))/DX(IXSZ(J ))
C=2*CK(ITYPE(K+1,N))*DX(IXSZ(N))/DY(IYSZ(K+1))
D=2*CK(ITYPE(K ,N))*DX(IXSZ(N))/DY(IYSZ(K ))
E=2*DX(IXSZ(N))*
$ (ROCP(ITYPE(K+1,N))*DY(IYSZ(K+1))+
ROCP(ITYPE(K ,N))*DY(IYSZ(K )))/DT(L)
UJ (K,J)=C
LJ (K,J)=0
DJ1(K,J)= -(C+D-E)
DJ2(K,J)=-(C+D+E)
UK (K,N)=0
LK (K,N)=-B
DK1(K,N)=B+E
DK2(K,N)=11-E
ELSE
C TREAT VERTICAL INTERIOR RADIAL INTERIOR AND CENTER
A=(CK(ITYPE(K+1,J+1))*DY(IYSZ(K+1))+
CK(ITYPE(K ,J+1))*DY(IYSZ(K )))/DX(IXSZ(J+1))
B=(CK(ITYPE(K+1,J ))*DY(IYSZ(K+1))+
CK(ITYPE(K ,J ))*DY(IYSZ(K )))/DX(IXSZ(J ))
C=(CK(ITYPE(K+1,J ))*DX(IXSZ(J ))+
CK(ITYPE(K+1,J+1))*DX(IXSZ(J+1)))/DY(IYSZ(K+1))
D=(CK(ITYPE(K ,J ))*DX(IXSZ(J ))+
CK(ITYPE(K ,J+1))*DX(IXSZ(J+1)))/DY(IYSZ(K ))
E=((ROCP(ITYPE(K+1,J+1))*DY(IYSZ(K+1))+
ROCP(ITYPE(K ,J+1))*DY(IYSZ(K )))*DX(1XSZ(J+1))+
(ROCP(ITYPE(K+1,J ))*DY(IYSZ(K+1))+
ROCP(ITYPE(K ,J ))*DY(IYSZ(K )))*DX(IXSZ(J )))/DT(L)
UJ (K,J)=C
LJ (K,J)=0
DJ1(K,J)=-(C+D-E)
DJ2(K,J)=-(C+D+E)
UK (K,J)=-A
LK (K,J)=-B
DK1(K,J)=A+B+E
DK2(K,J)=A+B-E
ENDIF
ENDIF
40 CONTINUE
C BEGIN TIME STEPS
DO 200 I=1,NSTEP(L)
150
Print file "hslb4pc.for"
TME=TME+DT(L)
CALL IADSLV(M,N,UK,DK1,LK,UJ,DJ2,1.J,T,TS)
IF(AMOD(REAL(I),REAL(NPRT(L))).E0.0)THEN
WRITE(6,25)TME,(J,J=NSKP(L),N,NSKP(L))
DO 140 KmM,MSKP(L),-MSKP(L)
140 WRITE(6,26)(T(K,J),J=NSKP(L),11,NSKP(L))
C CALCULATE HEAT FLUX LOCAL TO X=0 AND H, AVERAGE N (NAVE)
DO 150 K=1,M
IF(K.EQ.1)THEN
02P(1)=(CENT-T(K,1))/DX(IXSZ(1))*CK(ITYPE(2,1))
H(1) 412P(1)/CENT
NAVE =N(1).DY(IYSZ(2))/2
ELSEIF(K.ED.M)THEN
02P(M)=(CENT-T(K,1))/DX(IXSZ(1))*CK(ITYPE(M,1))
N(N) =02P(N)/CENT
HAVE =KAVE+H(M)*DY(IYSZ(M))/2
ELSE
Cl2P(K)=(CENT-T(K,1))*(CK(ITYPE(K+1,1))*DY(IYSZ(K+1))
+CK(ITYPE(K ,1))*DY(IYSZ(K )))/
(DX(IXSZ(1))*(DY(IYSZ(K+1))+DY(IYSZ(K))))
H(K) =02P(K)/CENT
HAVE =HAVE+H(K)*(DY(IYSZ(K))+DY(IYSZ(K+1)))/2
ENDIF
150 CONTINUE
HAVE=HAVE/YLEN
C WRITE HEAT FLUX, N, HAVE AND WRITE TO FILE HNAME FOR PLOTTING
WRITE(6,27)
WRITE(6,28)(K,02P(K),H(K),K=M,MSKP(L),-MSKP(L))
WRITE(6,29)TME,HAVE
IF(HNAME.NE.' ')WRITE(7,30)TME,HAVE
ENDIF
200 CONTINUE
C DUMP FINAL RESULTS TO RESTART FILE
IF(DNAME.NE.' ')THEN
OPEN(8,FILE=DNAME,FORMs'UNFORMATTED')
WRITE(8)CENT,A,M,11,NXSZ,NYSZ,NPROP,IXDX,IYDX,DX,DY,CK,ROCP,INDX,
S ITYPE,IXSZ,IYSZ,T,TS,TME
CLOSE(B)
ENDIF
CLOSE(7)
STOP
1 FORMAT(A)
2 FORMAT('1',A)
5 FORMAT(' INPUT REVIEW'//' CENTER TEMPERATURE',T35,1PE12.6 /' NUMBER
SOF AXIAL NODES',T35,I3/1 NUMBER OF RADIAL NODES,,T35,13/1 NUMBER 0
SF AXIAL STEP CARDS READ1,735,I2/' NUMBER OF RADIAL STEP CARDS READ
S',T35,12/' NUMBER OF PROPERTY CARDS READ',T35,I2 /' NUMBER OF TIME
SSTEP CARDS READ,,T35,12/1 H PLOT FILE NAME',T35,AP DUMP FILE NAME
S',T35,A//' AXIAL STEP CARDS'/' ID',T13,'DX')
32 FORMAT(//' HORIZ STEP CARDS'/' ID1,713,'DY')
36 FORMAT(//' PROPERTY CARDS,/, 101,T13,,K1,T25,,RIWCP1)
7 FORMAT(' ,l2,2(3XOPE12.6))
8 FORMAT(//' TIME STEP CARDS'/' NUMBERI,T15,,DP,T25,,NPRT,,T32,1MSK
SP',T38,11ISKP')
37 FORMAT(//' AXIAL NODE SIZE (RIGHT=TOPMOST)')
38 FORMAT(//' RADIAL NODE SIZE')
39 FORMAT(//' PROPERTY')
9 FORMAT(' ,16,2XOPE12.5,3(2X,14))
18 FORMAT(/' RESTART DATA FROM FILE ',A)
20 FORMAT(130I1)
21 FORMAT(' ,130I1)
23 FORMAT(//' TOTAL RADIAL LENGTH',T30,1PE12.6 /' TOTAL VERTICAL LENGT
SIP,T30,E12.6,//' H SOLUTION FOR ',E12.6,' TO ',E12.6,' SECONDS'//)
24 FORMAT(' PRINT TEMPERATURE MAP EVERY 1,14,1XOPE12.5,1 SECOND TIME
S STEPS FOR EVERY ',I3,' HORIZ NODES AND ',I3,' VERT NODES' /)
151
Print file "helb4pc.for*
25 FORMAT(' TEMPERATURE MAP AT ',1PE12.5,' SECONDS1/10(5X,I3,5X))

26 FORMAT(' ',10E13.6)
27 FORMAT(/6X,'HEAT FLUX',8X,'H')
28 FORMAT(' ',I3,1P2E12.5)
29 FORMAT(/' AVERAGE H 2 ',1PE12.5,' SECONDS: ',E12.5//)
30 FORMAT(' ',1P2E12.5)
END
FUNCTION RHS(ITIM,T,K,J)
C RETURNS THE VALUE OF THE RIGHT HAND SIDE OF THE IAD EQUATIONS
C ITIME - HALF (1) /FULL (2) TIME STEP (IN)
C T - TEMPERATURE ARRAY (IN)
C K,J - COLUMN, ROW INDICES (IN)
COMMON UK(50,50),LK(50,50),DK2(50,50),UJ(50,50),
SLJ(50,50),DJ1(50,50),F,M,N
DIMENSION T(50,50)
REAL LK,LJ
C BE CERTAIN UJ(N,J) AND UK(K,N) ARE 0 TO TREAT TOP CONDITIONS CORRECTLY
IF(ITIM.E0.1)THEN
C COMPUTE RHS AT HALF TIME STEP
IF(J.E0.1)THEN
IF(K.EQ.I)THEN
RHS=UJ(1,1)*T(2,1)+DJ1(1,1)*T(1,1)-LK(1,1)*F
ELSE
RHS=UJ(K,1)*T(K+1,1)+LJ(K,1)*T(K-1,1)+DJUK,1)*T(K,1)-
$ LK(K,1)*F
ENDIF
ELSE
RHSUJ(K,J)*T(K+1,J)+LJ(K,J)*T(K-1,J)+DJ1(K,J)*T(K,J)
ENDIF
ELSEIF(ITIM.E0.2)THEN
C COMPUTE RHS AT FULL TIME STEP
IF(J.EQ.1)THEN
RNS=UK(K,1)*T(K,2)+LK(K,1)*F+0K2(K,1)*T(K,1)
ELSE
RHS=UK(K,J)*T(K,J+1)+LK(K,J)*T(K,J-1)+DK2(K,J)*T(K,J)
ENDIF
ENDIF
RETURN
END
SUBROUTINE IADSLV(M,N,UK,DK,LK,UJ,DJ,LJ,T,TS)
C SOLVES THE IAD EQUATIONS USING GAUSSIAN ELIMINATION FOR A FULL TIME STEP
C M - NUMBER OF COLUMNS (IN)
C N - NUMBER OF ROWS (IN)
C UK,DK,LK,UJ,DJ,LJ - COEFFICIENTS FOR IAD EQUATIONS (IN)
C T - FULL TIME STEP TEMPERATURE ARRAY (OUT, IN)
C TS - HALF TIME STEP TEMPERATURE ARRAY (OUT,IN)
REAL LK(50,50),LJ(50,50)
DIMENSION DK(50,50),DJ(50,50),UK(50,50),UJ(50,50),
ST(50,50),B(50),TS(50,50)
EXTERNAL RHS
C COMPUTE TEMPERATURES AT END OF HALF TIME STEP IMPLICIT BY COLUMNS
DO 70 K=1,M
B(1)=DK(K,1)
TS(K,1)=RHS(1,T,K,1)/11(1)
C FORWARD GUASSIAN ELIMINATION
DO 65 J=2,N
B(J)=DK(K,J)-LK(K,J)*UK(K,J-1)/B(J-1)
65 TS(K,J)=(RHS(1,T,K,J)-LK(K,J)*TS(K,J-1))/B(J)
C BACK SOLUTION
DO 70 J=W-1,1,-1
70 TS(K,J)=TS(K,J)-UK(K,J)*TS(K,J+1)/B(J)
C COMPUTE TEMPERATURES AT END OF FULL TIME STEP IMPLICIT BY ROWS
DO 130 J :1,N
152
Print file "tislb4pc.for"
B(1)43.1(1,0
T(1,J)=RHS(2,TS,1,J)/11(1)
DO 110 K=2,M
B(K)=DJ(K,J)-LJ(K,J)*UJ(K-1,0/B(K-1)
110 T(K,J)=(RHS(2,TS,K,J)-LJ(K,J)*T(K-1,J))/B(K)
C BACK SOLUTION
DO 130 K=M-1,1,-1
130 T(K,J)=T(K,J)-UJ(K,J)*T(K+1,J)/B(K)
RETURN
END
SUBROUTINE WRTHD(N)
C WRITE MAP POSITION HEADING
CHARACTER HEAD*130
HEAD =' '
L=1
DO 6 1=1,11/10
HEAD=HEAD(1:L)//1-
6 L=I*10+1
WRITE(6,12)HEAD
12 FORMAT(' LOCATION MAP (BY ID)' / /A)
RETURN
END
153
HCYL9PC.FOR
Print file *hcyl9pc.for*
154
PROGRAM BLAUSIS
C CALCULATES TEMPERATURES AND HEAT TRANSFER COEFFICIENTS FOR TWO DIMENSIONAL CYLINDER
C VARIABLE NODE SIZE AND PROPERTIES ALLOWED
C PC VERSION
REAL LJ,LK
DIMENSION T(50,50),ITYPE(50,50),DK1(50,50),DJ2(50,50),
$ TS(50,50),Q2P(50),H(50),IRSZ(50),IXSZ(50),
SIRDX(10),DX(10),DR(10),NSTEP(10),CK(10),ROCP(10),DT(10),NPRT(10),
SNSKP(10),MSKP(10),INDX(10),IXDX(10)
COMMON UK(50,50),LK(50,50),DK2(50,50),UJ(50,50),
SLJ(50,50),DJ1(50,50),F,M,11
CHARACTER TITLE*80,HNAME*40,DNAME*40,RNAME*40
C INPUT PREPARATION
C RECORD 1:
C TITLE - 80 CHARACTER PROBLEM DESCRIPTION
C RECORD 2:
C NTM - NUMBER OF TIME STEP CARDS TO BE READ
C RNAME - RESTART FILE NAME. IF PRESENT READ ONLY TIME STEP CARDS AFTER THIS CARD
C HNAME - H PLOT FILE NAME. IF PRESENT WRITE T,HAVE(T) TO HNAME.
C DNAME - DUMP FILE NAME. IF PRESENT WRITE DUMP FILE TO DNAME.
C RECORD 3:
C CENT - TEMPERATURE 2 R=0
C A - INNER RADIUS OF CYLINDER
C M - TOTAL NUMBER OF AXIAL NODE ENDPOINTS (=TOTAL NODES - 1)

C N - TOTAL NUMBER OF RADIAL NODES
C NXSZ - NUMBER OF AXIAL NODE SIZE CARDS TO BE READ
C NRSZ - NUMBER OF RADIAL NODE SIZE CARDS TO BE READ
C NPROP - NUMBER OF PROPERTY CARDS TO BE READ
C READ NXSZ AXIAL NODE SIZE CARDS:
C IXDX - INDEX OF NODE SIZE (0-9)
C DY - NODE SIZE
C READ NRSZ RADIAL NODE SIZE CARDS:
C IRDX - INDEX OF NODE SIZE (0-9)
C DX - NODE SIZE
C READ NPROP PROPERTY CARDS:
C INDX - INDEX OF PROPERTY (0-9)
C CK - THERMAL CONDUCTIVITY
C ROCP - DENSITY*SPECIFIC HEAT
C READ NTM TIME STEP CARDS:
C NSTEP - TOTAL NUMBER OF TIME STEPS ACTIVE FOR THIS CARD
C DT - TIME STEP SIZE
C NPRT - NUMBER OF TIME STEPS A TEMPERATURE/N COEFF SUMMARY IS PRINTED
C MSKP - PRINT FOR EACH NSKP AXIAL NODES
C NSKP - PRINT FOR EACH NSKP RADIAL NODES
C READ AXIAL NODE SIZE MAP FOR M-1 NODES FROM TOP MOST POSITION (1 CARD)
C READ RADIAL NODE SIZE MAP FOR N NODES FROM RIGHT MOST POSITION (1 CARD)
C READ PROPERTY LOCATION MAP (N-1 ROWS WITH N COLUMNS)
C END OF INPUT
C READ DATA
READ(5,1) TITLE
WRITE(6,2)TITLE
READ(5,*) NTM,RNAME,HNAME,DNAME
IF(RNAME.E0.")THEN
READ(5,*) CENT,A,M,N,NXSZ,NRSZ,NPROP
WRITE(6,5)CENT,A,M,N,NXSZ,NRSZ,NPROP,NTM,NNAME,DNAME
C READ AXIAL STEP SIZE CARDS
DO 31 I=1,NXSZ
READ(5,*) IXDX(I),DX(IXDX(I)+1)
WRITE(6,32)
C READ RADIAL STEP SIZE CARDS
DO 34 I=1,NRSZ
READ(5,*) IRDX(I),DR(IRDX(I)+1)
34 WRITE(6,7)IRDX(I),DR(IRDX(I)+1)
WRITE(6,36)
155
Print file "hcyl9pc.for"
C READ PROPERTY CARDS

DO 3 I=1,NPROP
READ(5,*) INDX(I),CK(INDX(I)+1),ROCP(INDX(I)+1)
3 WRITE(6,7)INDX(1),CK(INDX(1)+1),ROCP(INDX(1)+1)
C READ TIME STEP CARDS
WRITE(6,8)
DO 4 I=1,NTM
4 WRITE(6,9)NSTEP(I),DT(I),NPRT(I),MSKP(I),NSKP(I)
C READ AXIAL STEP SIZE LOCATION MAP
WRITE(6,37)
CALL WRTHD(M)
READ(5,20) (IXSZ(K),K=M,2,-1)
WRITE(6,21)(IXSZ(K),KM,2,-1)
C READ RADIAL STEP SIZE LOCATION MAP
WRITE(6,38)
CALL WRTHD(N)
READ(5,20) (IRSZ(J),J=1,N)
WRITE(6,21)(IRSZ(J),J=1,N)
C READ PROPERTY LOCATION MAP
WRITE(6,39)
CALL WRTHD(N)
DO 10 K=M,2,-1
READ(5,20) (ITYPE(K,J),J=1,N)
10 WRITE(6,21)(ITYPE(K,J),J=1,N)
C ADJUST SIZE AND TYPE ID TO AVOID ZERO ARRAY INDEX
DO 11 K=2,M
IXSZ(K)=IXSZ(K)+1
DO 11 J=1,N
11 ITYPE(K,J)=ITYPE(K,J)+1
DO 41 J=1,N
41 IRSZ(J)=IRSZ(J)+1
TME=0.
ELSE
C PERFORM RESTART FROM FILE RNAME
OPEN(8,FILE=RNAME,FORWUNFORMATTED')
READ(8)CENT,A,M,N,NXSZ,NRSZ,NPROP,IXDX,IRDX,DX,DR,CK,RCCP,INDX,
ITYPE,IXSZ,IRSZ,T,TS,TME
CLOSE(8)
WRITE(6,18) RNAME
WRITE(6,5)CENT,A,M,N,NXSZ,NRSZ,NPROP,NTM,HNAME,DNANE
DO 42 I=1,NXSZ
DO 43 I=1,NRSZ
43 WRITE(6,7)IRDX(1),DR(IRDX(1)+1)
WRITE(6,36)
DO 13 1=1,NPROP
13 WRITE(6,7)INDX(1),CK(INDX(1)+1),ROCP(INDX(1)+1)
WRITE(6,8)
DO 14 I=1,NTM
14 wRITE(6,9)NSTEPOLDT(1),NPRT(1),MsKPOLNSKP(I)
WRITE(6,37)
CALL WRTHD(M)
WRITE(6,21)(IXSZ(K)-1,K=M,2,-1)
WRITE(6,38)
CALL WRTHD(N)
WRITE(6,21)(IRSZ(J)-1,J=1,N)
WRITE(6,39)
CALL WRTHD(N)
DO 16 K=M,2,-1
16 WRITE(6,21)(ITYPE(K,J)-1,J=1,N)
END IF
F=CENT
C SUM INTERVALS TO FIND TOTAL LENGTH
156
Print file "heyi9pc.form
XLEN=0
DO 17 K=2,M
17 XLEN=XLEN+DX(IXSZ(K))
RLEN=0
DO 19 J=1,N
19 RLEN=RLEN+DR(IRSZ(J))
IF(HNAME.NE.") OPEN(7,FILE=HNAME)
DO 200 L=1,NTM
WRITE(6,23)RLEN,XLEN,TME,DT(L)*NSTEP(L)+TME
WRITE(6,24)NPRT(L),DT(L),NSKP(L),MSKP(L)
C PREPARE TRIDIAGONAL COEFFICIENT MATRIXES
C TREAT WALL AS INSULATED BOUNDARY
C CENTER CONDITIONS FOR CONSTANT TEMPERATURE
R=A
DO 40 J=1,N
R=R+DR(IRSZ(J))
DO 40 K=1,M
IF(K.EQ.M)THEN
C TREAT TOP CONDITIONS
IF(J.EQ.N)THEN
C TREAT TOP WALL
B=2*CK(ITYPE(M,N))*DX(IXSZ(M))*(R-DR(IRSZ(N))/2)/DR(IRSZ(N))
C=2*CK(ITYPE(M,N))*DR(IRSZ(N))* R/DX(IXSZ(M))
D=C
E=4*R*DR(IRSZ(N))*DX(IXSZ(M))*ROCP(ITYPE(M,N))/DT(L)
UJ (M,N)=0
LJ (M,N)=C+0
DJUM,N)=-(C+D-E)
DJ2(M,N)=-(C+D+E)
UK (M,N)=0
LK (M,N)=-B
DK1(M,N)=B+E
DK2(M,N)=B-E
ELSE
C TREAT TOP INTERIOR AND CENTER
A=2*CK(ITYPE(M,J+1))*DX(IXSZ(M))*(R+DR(IRSZ(J+1))/2)/
DR(IRSZ(J+1))
B=2*CK(ITYPE(M,J ))*DX(IXSZ(M))*(R-DR(IRSZ(J ))/2)/
$ DR(IRSZ(J ))
C=(CK(ITYPE(M ,J ))*DR(IRSZ(J ))*(R-DR(IRSZ(J ))/4)+
CK(ITYPE(M ,J+1))*DR(IRSZ(J+1))*(R+DR(IRSZ(J+1))/4))/
DX(IXSZ(M ))
D=C
E=2*DX(IXSZ(M))*
(ROCP(ITYPE(M ,J+1))*DR(IRSZ(J+1))*(R+DR(IRSZ(J+1))/4)+
ROCP(ITYPE(M ,J ))*DR(IRSZ(J ))*(R-DR(IRSZ(J ))/4))
S /DT(L)
UJ (M,J)=0
LJ (M,J)=C+0
DJUM,J)=-(C+0-E)
DJ2(M,J)=-(C+D+E)
UK (K,J)=-A
LK (K,J)=-B
DKUK,J)=A+8+E
DK2(K,J)=A+B-E
ENDIF
ELSEIF(K.E0.1)THEN
C TREAT BOTTOM CONDITIONS
IF(J.EQ.N)THEN
C TREAT BOTTOM WALL
B=2*CK(ITYPE(2,N))*DX(IXSZ(2))*(R-DR(IRSZ(N))/2)/DR(IRSZ(N))
C=2*CK(ITYPE(2,N))*DR(IRSZ(N))* R/DX(IXSZ(2))
D=C
E=4*R*DR(IRSZ(N))*DX(IXSZ(2))*ROCP(ITYPE(2,N))/DT(L)
UJ (1,J)=C+D
157
Print file *hcyl9pc.for*
LJ (1,0=0
DJ1(1,J)=-(C+D-E)
DJ2(1,J)=-(C+D+E)
UK (K,N)=0
LK (K,N)*-B
DK1(K,N)=B+E
DK2(K,N)=B-E
ELSE
A=2*CK(ITYPE(K+1,J+1))*DX(IXSZ(2))*(R+DR(IRSZ(J+1))/2)/
S DR(IRSZ(J+1))
B=2*CK(ITYPE(K+1,J ))*DX(IXSZ(2))*(R-DR(IRSZ(J ))/2)/
DR(IRSZ(J ))
C =(CK(ITYPE(K +1,J ))* DR(IRSZ(J ))*(R-DR(IRSZ(J ))/4)+
CK(ITYPE(K+1,J+1))*DR(IRSZ(J+1))*(R+DR(IRSZ(J+1))/4))/
DX(IXSZ(2 ))
D=C
E=2*DX(IXSZ(2 ))*
(ROCP(ITYPE(K+1,J+1))*DR(IRSZ(J+1))*(R+DR(IRSZ(J+1))/4)+
ROCP(ITYPE(K+1,J ))*DR(IRSZ(J ))*(R-DR(IRSZ(J ))/4))
/DT(L)
UJ (1,J)=C+D
LJ (1,0=0
DJ1(1,J)=-(C+D-E)
DJ2(1,J)=-(C+D+E)
UK (K,J)=-A
LK (K,J)=-B
DK1(K,J)=A+B+E
DK2(K,J)=A+B-E
ENDIF
ELSE
C TREAT AXIAL INTERIOR CONDITIONS
IF(J.EQ.N)THEN
C TREAT AXIAL INTERIOR WALL
B=(CK(ITYPE(K+1,J ))*DX(IXSZ(K+1))+
CK(ITYPE(K ,J )) *DX(IXSZ(K )))*(R-DR(IRSZ(J ))/2)/
DR(IRSZ(J ))
C=2*CK(ITYPE(K+1,N))*DR(IRSZ(N))*R/DX(IXSZ(K+1))
D=2*CK(ITYPE(K ,N))*DR(IRSZ(N))*R/DX(IXSZ(K ))
E=2*R*DR(IRSZ(N))*(ROCP(ITYPE(K+1,N))*DX(IXSZ(K+1))+
ROCP(ITYPE(K ,N))*DX(IXSZ(K )))/DT(L)
UJ (K,O=C
LJ (K,J)=0
DJ1(K,J)=-(C+D-E)
DJ2(K,J)=-(C+D+E)
UK (K,N)=0
LK (K,N)=-B
DK1(K,N)=B+E
DK2(K,N)=11-E
ELSE
C TREAT AXIAL INTERIOR RADIAL INTERIOR AND CENTER
A=(CK(ITYPE(K+1,J+1))*DX(IXSZ(K+1))+
S CK(ITYPE(K ,J+1))*DX(IXSZ(K )))*(R+DR(IRSZ(J+1))/2)/
$ DR(IRSZ(J+1))
B=(CK(ITYPE(K+1,J ))*DX(IXSZ(K+1))+
CK(ITYPE(K ,J ))*DX(IXSZ(K )))*(R-DR(IRSZ(J ))/2)/
S DR(IRSZ(J ))
Cm(CK(ITYPE(K+1,J ))*DR(IRSZ(J ))*(R-DR(IRSZ(J ))/4)+
CK(ITYPE(K+1,J+1))*DR(IRSZ(J+1))*(R+DR(IRSZ(J+1))/4))/
S DX(IXSZ(K+1))
D=(CK(ITYPE(K ,J ))*DR(IRSZ(J ))*(R-DR(IRSZ(J ))/4)+
CK(ITYPE(K ,J+1))*DR(IRSZ(J+1))*(R+DR(IRSZ(J+1))/4))/
DX(IXSZ(K ))
E=((ROCP(ITYPE(K+1,J+1))*DX(IXSZ(K+1))+
ROCP(ITYPE(K ,J+1))*DX(IXSZ(K )))*
S DR(IRSZ(J+1))*(R+DR(IRSZ(J+1))/4)+
158
Print file *hcyt9pc.for"
S (ROCP(ITYPE(K+1,J ))*DX(IXSZ(K+1))
+ROCP(ITYPE(K ,J ))*DX(IXSZ(K )))*
$ DR(IRSZ(J ))*(R-DR(IRSZ(J ))/4))/DT(L)
UJ (K,J)C
LJ (K,J)"D
DJUK,J)=-(C+D-E)
DJ2(K,J)=-(C+0+E)
UK (K,J)=-A
LK (K,J)=-B
DK1(K,J)=A+8+E
DK2(K,J)=A+B-E
ENDIF
ENDIF
40 CONTINUE
C BEGIN TIME STEPS
DO 200 181,NSTEP(L)
TME=TME+DT(L)
CALL IADSLV(M,N,UK,DK1,LK,UJ,DJ2,LJ,T,TS)
IF(AMOD(REAL(I),REAL(NPRT(L))).E0.0)THEN
WRITE(6,25)TME,(J,J=NSKP(L),N,NSKP(L))
DO 140 K=M,MSKP(L),-MSKP(L)
140 WRITE(6,26)(T(K,J),J=NSKP(L),N,NSKP(L))
C CALCULATE HEAT FLUX LOCAL TO R=0 AND H, AVERAGE H (HAVE)
DO 150 K=1,M
IF(K.EQ.1)THEN
02P(1)*(CENT-T(K,1))/DR(IRSZ(1))*CK(ITYPE(2,1))
H(1) =02P(1)/CENT
NAVE =H(1)*DX(IXS2(2))/2
ELSEIF(K.E0.M)THEN
02P(M)=( CENT-T(K,1))/DR(IRSZ(1))*CK(ITYPE(M,1))
H(M) =02P(M)/CENT
HAVE *HAVE+H(M)*DX(IXSZ(M))/2
ELSE
02P(K)=(CENT-T(K,1))*(CK(ITYPE(K+1,1))*DX(IXSZ(K+1))
S +CK(ITYPE(K ,1))*DX(IXSZ(K )))/
(DR(IRS2(1))*(DX(IXSZ(01))+DX(IXSZ(K))))
N(K) =02P(K)/CENT
HAVE =HAVE+N(K)*(DX(IXSZ(K))+DX(IXSZ(K+1)))/2
ENDIF
150 CONTINUE
HAVE=HAVE/XLEN
C WRITE HEAT FLUX, H, HAVE AND WRITE TO FILE NNAME FOR PLOTTING
WRITE(6,27)
WRITE(6,28)(K,02P(K),H(K),K=M,MSKP(L),-MSKP(L))
WRITE(6,29)TME,HAVE
IF(HNAME.NE.")WRITE(7,30)TME,HAVE
ENDIF
200 CONTINUE
C DUMP FINAL RESULTS TO RESTART FILE
IF(DNAME.NE.' ')THEN
OPEN(8,FILE=DNAME,FORWUNFORMATTED')
WRITE(8)CENT,A,M,N,NXSZ,NRSZ,NPROP,IXDX,IRDX,DX,DR,CK,ROCP,INDX,
ITYPE,IXSZ,IRSZ,T,TS,TME
CLOSE(8)
ENDIF
CLOSE(7)
STOP
1 FORMAT(A)
2 FORMAT('1',A)
5 FORMAT(' INPUT REVIEW'//' CENTER TEMPERATURE',T35,1PE12.6 /' INNER
SRADIUSI,T35,E12.6/1 NUMBER OF AXIAL NODES',T35,13 /' NUMBER OF RADI
SAL NODES',115,13/1 NUMBER OF AXIAL STEP CARDS READ',T35,12/' RUNE
SR OF RADIAL STEP CARDS READ',T35,I2 /' NUMBER OF PROPERTY CARDS REA
SD',135,12/' NUMBER OF TIME STEP CARDS READ',T35,I2/' H PLOT FILE N
159
Print file ohcyl9pc.for*
$AME',T35,A /' DUMP FILE NAME',T35,A //' AXIAL STEP CARDS' /' ID',T13,
SW')
32 FORMAT(//' RADIAL STEP CARDS'/' ID',T13,'DR')
36 FORMAT(//' PROPERTY CARDS' /' ID,,T13,,K1,T25,0RHO*CP1)
7 FORMAT(' 1,12,2(3XOPE12.6))
8 FORMAT(//' TIME STEP CARDS'/' NUMBER ',T15,'DT',T25,'NPRT',T32,'MSK
$P,,T38,41SKPI)
37 FORMAT(//' AXIAL NODE SIZE (RIGHT = TOPMOST)')
38 FORMAT(//' RADIAL NODE SIZE')
39 FORMAT(//' PROPERTY')
9 FORMAT(' 1,16,2XOPE12.5,3(2X,14))
18 FORMAT(/' RESTART DATA FROM FILE ',A)
20 FORMAT(130I1)
21 FORMAT(' ,13011)
23 FORMAT(//' TOTAL RADIAL LENGTH',T30,1PE12.6/1 TOTAL AXIAL LENGTH',
S T30,E12.6,//' H SOLUTION FOR ',E12.6,' TO ',E12.6,' SECONDS' //)
24 FORMAT(' PRINT TEMPERATURE MAP EVERY 1,14,1XOPE12.5,1 SECOND TIME
$ STEPS FOR EVERY ',I3,' RADIAL NODES AND ',13,' AXIAL NODES'/)
25 FORMAT(' TEMPERATURE MAP AT ',1PE12.5,' SECONDSI/10(5X,13,5X))
26 FORMAT(' ',10E13.6)
27 FORMAT(/6X,'HEAT FLUX',8X,'H')
28 FORMAT(' ',/3,1P2E12.5)
29 FORMAT(/' AVERAGE H a ',1PE12.5,' SECONDS: 1,E12.5//)
30 FORMAT(' ',1P2E12.5)
END
FUNCTION RHS(ITIM,T,K,J)
C RETURNS THE VALUE OF THE RIGHT HAND SIDE OF THE IAD EQUATIONS
C ITIME - HALF (1) /FULL (2) TIME STEP (IN)
C T - TEMPERATURE ARRAY (IN)
C K,J - COLUMN, ROW INDICES (IN)
COMMON UK(50,50),LK(50,50),DK2(50,50),UJ(50,50),
$1.J(50,50),DJ1(50,50),FAN
DIMENSION T(50,50)
REAL LK,LJ
C BE CERTAIN UJ(M,J) AND UK(K,N) ARE 0 TO TREAT TOP CONDITIONS CORRECTLY
IF(ITIM.EQ.1)THEN
C COMPUTE RHS AT HALF TIME STEP
IF(J.EQ.1)THEN
IF(K.EQ.1)THEN
RHS=UJ(1,1)*T(2,1)+DJ1(1,1)*T(1,1)-LK(1,1)*F
ELSE
RHS=UJ(K,1)*T(K+1,1)+LJ(K,1)*T(K-1,1)+DJ1(K,1)*T(K,1)-
$ LK(K,1)*F
ENDIF
ELSE
RHS=UJ(K,J)*T(K+1,J)+1.J(K,J)*T(K-1,J)+DJ1(K,J)*T(K,J)
ENDIF
ELSEIF(ITIM.EQ.2)THEN
C COMPUTE RHS AT FULL TIME STEP
IF(J.EQ.1)THEN
RHS=UK(K,1)*T(K,2)+LK(K,1)*F+DK2(K,1)*T(K,1)
ELSE
RHS=UK(K,J)*T(K,J+1)+LK(K,J)*T(K,J-1)+DK2(K,J)*T(K,J)
ENDIF
ENDIF
RETURN
END
SUBROUTINE IADSLV(M,N,UK,DK,LK,UJ,DJ,LJ,T,TS)
C SOLVES THE IAD EQUATIONS USING GAUSSIAN ELIMINATION FOR A FULL TIME STEP
C M - NUMBER OF COLUMNS (IN)
C N '- NUMBER OF ROWS (IN)
C UK,DK,LK,UJ,DJ,LJ - COEFFICIENTS FOR IAD EQUATIONS (IN)
C T - FULL TIME STEP TEMPERATURE ARRAY (OUT,IN)
160
Print file Nlicyl9pc.for"
C TS - HALF TIME STEP TEMPERATURE ARRAY (OUT,IN)

REAL LK(50,50),LJ(50,50)
DIMENSION DK(50,50),DJ(50,50),UK(50,50),UJ(50,50),
ST(50,50),B(50),TS(50,50)
EXTERNAL RNS
C COMPUTE TEMPERATURES AT END OF HALF TIME STEP IMPLICIT BY COLUMNS
DO 70 Ka1,M
B(1)=0K(K,1)
TS(K,1)=RHS(1,T,K,1)/B(1)
DO 65 J=2,N
B(J)=DK(K,J)-LK(K,J)*UK(K,J-1)/B(J-1)
65 TS(K,J)a(RNS(1,T,K,J)-LK(K,J)*TS(K,J-1))/B(J)
C BACK SOLUTION
DO 70 JaN-1,1,-1
70 TS(K,J)=TS(K,J)-UK(K,J)*TS(K,J+1)/B(J)
C COMPUTE TEMPERATURES AT END OF FULL TIME STEP IMPLICIT BY ROWS
DO 130 Ja1,N
B(1)=DJ(1,J)
T(1,J)=RHS(2,TS,1,J)/B(1)
DO 110 K=2,M
B(K)=DJ(K,J)-LJ(K,WUJ(K-1,J)/B(K-1)
110 T(K,J)=(RNS(2,TS,K,J)-LJ(K,J)*T(K-1,J))/B(K)
C BACK SOLUTION
DO 130 K=M-1,1,-1
130 T(K,J)=T(K,J)-UJ(K,J)*T(K+1,J)/B(K)
RETURN
END
SUBROUTINE WRIND(N)
C WRITE MAP POSITION HEADING
CHARACTER HEAD*130
HEAD ='
La1
DO
::A:=I211/
NEA6(11!)//1. 8
6 LaI*1041
WRITE(6,12)HEAD
12 FORMAT(' LOCATION MAP (BY ID)'//A)
RETURN
END
Appendix E
Experimental Data
Table E.1. Table of Experimental Conditions and Heat Transfer

Coefficients for Glass Beads.
ks = 1.04 W/m-K
Cps = 1080 J/kg-K
Ps = 2500 kg/m3
pb - 1500 kg/m3 dp - 0.00018 m

Linear Mass Heat
Solids Flow Tw Ts Contact Transfer % Error
Power Velocity Rate Time Coeffivient
(W) (m/s) (kg/s) (C) (C) (s) (W/m4-K)
147. .03860 .0599 127.5 28.2 3.948 243.4 4.4

203. .03860 .0599 156.7 25.0 3.948 253.5 3.2
97. .03860 .0599 97.6 30.2 3.948 236.7 6.6
98. .07730 .1199 80.3 31.2 1.972 328.4 7.1
150. .07730 .1199 107.6 31.9 1.972 325.9 4.7
200. .07730 .1199 132.6 33.7 1.972 332.3 3.5
200. .10840 .1682 121.3 35.1 1.406 381.4 3.7
148. .10840 .1682 98.9 35.7 1.406 385.0 5.0
99. .10840 .1682 76.5 35.7 1.406 399.5 7.5
104. .00700 .0109 167.4 30.8 21.771 125.2 5.5
106. .01200 .0186 148.2 33.3 12.700 151.7 5.6
107. .02300 .0357 124.9 34.0 6.626 193.5 5.8
107. .03200 .0496 113.6 34.2 4.762 221.5 5.9
107. .05200 .0807 99.3 34.4 2.931 271.3 6.2
106. .07500 .1163 90.2 34.6 2.032 313.3 6.5
107. .09300 .1443 85.0 33.5 1.639 341.5 6.6
106. .11950 .1854 80.0 34.9 1.275 386.2 6.9
106. .16650 .2583 74.5 34.5 .915 435.8 7.2
106. .26000 .4033 67.2 33.6 .586 519.5 7.7
105. .24700 .3832 67.8 33.9 .617 506.4 7.7
105. .29000 .4499 65.9 33.9 .526 539.9 7.9
pb = 1500 kg/m3 dp = 0.0008 m
Linear Mass Heat

Solids Flow Tw Is Contact Transfer % Error
Power Velocity Rate Time Coefficient
(W) (m/s) (kg/s) (C) (C) (s) (W/m4-K)
100. .02570 .0399 119.8 28.0 5.930 179.2 6.1

148. .02130 .0330 170.6 29.3 7.155 172.2 4.1
146. .00487 .0076 248.1 29.9 31.294 110.0 3.9
100. .00590 .0092 182.8 30.3 25.831 107.9 5.7
100. .04530 .0703 111.7 32.0 3.364 206.4 6.3
148. .04380 .0679 145.3 33.1 3.479 216.9 4.3
147. .07730 .1199 129.4 33.9 1.972 253.1 4.5
98. .07700 .1194 102.3 33.9 1.979 235.7 6.6
98. .14200 .2203 93.9 33.6 1.073 267.4 6.8
149. .14320 .2221 119.4 33.6 1.064 285.6 4.5
149. .25300 .3925 116.9 34.2 .602 296.3 4.6
101. .23800 .3692 92.3 34.3 .640 286.6 6.7
98. .23500 .3645 90.8 34.4 .649 284.4 6.9
pb = 1435 kg/m3 dp = 0.000044 m
Linear Mass Heat

Power Velocity Rate Time Coeffqient
(W) (m/s) (kg/s) (C) (C) (s) (W/m4-K)
93. .00794 .0118 153.2 29.1 19.194 122.6 6.2

148. .00750 .0111 220.2 29.3 20.320 127.6 3.9
149. .00980 .0145 204.3 30.8 15.551 141.2 3.9
100. .00945 .0140 161.3 32.2 16.127 127.4 5.8
100. .01705 .0253 126.8 34.3 8.938 177.8 6.1
100. .02250 .0334 116.3 34.4 6.773 200.9 6.2
100. .07830 .1162 91.9 33.5 1.946 282.0 6.7
100. .04450 .0660 96.3 34.3 3.425 265.4 6.6
118. .00448 .0066 216.1 27.1 34.018 102.7 4.8
74. .00560 .0083 148.2 27.2 27.214 100.6 7.6
76. .01410 .0209 110.7 31.7 10.809 157.2 7.9
101. .01460 .0217 132.9 31.9 10.438 164.5 5.9
101. .01660 .0246 128.8 33.5 9.181 174.4 6.0
101. .02330 .0346 116.0 34.0 6.541 202.7 6.2
101. .03500 .0519 102.3 34.1 4.354 243.6 6.4
102. .04520 .0671 96.0 34.2 3.372 271.5 6.5
102. .07100 .1054 83.9 33.5 2.146 332.8 6.9
Coefficients for Sand.
ks - 0.80 W/m-K
Cps = 780 J/kg-K
ps - 2700 kg/m3
pb - 1280 kg/m3 dp - 0.00011 m

Linear Mass Heat
(W) (m/s) (kg/s) (C) (C) (s) (W/m4-K)
100. .01860 .0246 123.0 25.7 8.194 169.1 6.0

126. .01740 .0230 149.9 27.3 8.759 169.0 4.8
126. .04140 .0548 123.2 28.6 3.681 219.0 5.0
101. .04140 .0548 103.1 29.3 3.681 225.3 6.3
101. .08180 .1083 90.6 30.1 1.863 274.7 6.6
125. .08180 .1083 103.8 30.1 1.863 279.0 5.4
126. .15200 .2012 93.8 30.7 1.003 327.2 5.6
100. .15200 .2012 81.1 30.6 1.003 325.5 7.0
99. .22700 .3005 76.0 30.7 .671 359.2 7.3
128. .22300 .2952 88.7 30.9 .683 362.9 5.7
128. .22700 .3005 91.0 31.6 .671 353.1 5.6
147. .03370 .0446 144.8 26.3 4.522 204.1 4.2
147. .06150 .0814 122.5 30.2 2.478 261.9 4.5
147. .08570 .1134 116.4 30.7 1.778 282.0 4.6
146. .02115 .0280 162.0 27.3 7.206 178.2 4.2
100. .02115 .0280 117.8 29.3 7.206 185.9 6.1
98. .05430 .0719 95.8 30.8 2.807 248.0 6.6
150. .05380 .0712 127.0 30.9 2.833 256.7 4.4
101. .05380 .0712 101.2 31.9 2.833 239.8 6.4
50. .05380 .0712 67.5 32.6 2.833 235.9 12.9
50. .08200 .1085 63.6 33.3 1.859 271.1 13.3
100. .08200 .1085 92.8 33.5 1.859 277.6 6.7
148. .08200 .1085 122.8 34.3 1.859 274.1 4.5
Pb - 1400 kg/m3 d
P
= 0.0008 m
Linear Mass Heat
(W) (m/s) (kg/s) (C) (C) (s) (W/m4-K)
100. .01930 .0281 121.1 27.1 7.896 174.8 6.1

125. .01920 .0280 145.4 29.0 7.937 176.6 4.9
100. .02430 .0354 121.3 31.1 6.272 182.3 6.1
125. .02920 .0426 137.1 30.5 5.219 192.8 4.9
100. .06040 .0881 103.0 31.3 2.523 229.5 6.4
151. .06080 .0887 133.6 31.4 2.507 243.0 4.3
151. .09900 .1444 126.4 32.0 1.539 263.2 4.4
101. .09840 .1435 99.5 32.1 1.549 246.2 6.4
100. .18300 .2668 92.8 31.6 .833 268.9 6.6
148. .18400 .2683 120.3 31.9 .828 274.5 4.5
148. .16100 .2348 122.5 32.0 .947 268.9 4.5
99. .16500 .2406 95.6 32.3 .924 257.3 6.6
148. .01500 .0219 189.7 32.0 10.160 154.4 4.0
147. .01270 .0185 184.2 27.7 12.000 154.5 4.0
147. .02400 .0350 160.3 33.6 6.350 190.8 4.2
145. .05700 .0831 138.1 30.0 2.674 220.5 4.4
148. .06600 .0962 139.9 31.6 2.309 224.7 4.3
50. .02170 .0316 83.4 32.8 7.023 162.4 12.0
100. .02170 .0316 129.0 32.9 7.023 171.2 6.0
151. .02170 .0316 170.3 33.1 7.023 181.0 4.0
146. .05150 .0751 139.8 30.3 2.959 219.4 4.3
100. .05150 .0751 109.5 32.0 2.959 212.1 6.3
50. .05150 .0751 71.6 32.7 2.959 211.2 12.6
pb - 1440 kg/m3 dp = 0.0012 m
Linear Mass Heat

(W) (m/s) (kg/s) (C) (C) (s) (W/m4-K)
100. .06500 .0968 105.3 32.8 2.345 226.9 6.4

100. .06500 .0968 152.3 30.3 2.345 134.8 5.8
150. .01740 .0259 120.3 28.5 8.759 268.7 4.4
149. .02500 .0372 114.8 27.9 6.096 282.0 4.5
96. .04130 .0615 138.4 27.9 3.690 142.9 6.1
150. .05740 .0855 132.0 29.8 2.655 241.4 4.3
100. .05740 .0855 97.8 28.9 2.655 238.9 6.5
100. .06200 .0923 101.2 29.6 2.458 229.8 6.4
Coefficients for Copper.
ks = 384 W/m-K
Cps = 386 J/kg-K
ps - 8950 kg/m?,
Pb = 5200 kg /m3
d = 0.00021 m
Linear Mass Heat

Power Velocity Rate Time Coeffivient
(W) (m/s) (kg/s) (C) (C) (s) (W/m4-K)
153. .10700 .5931 84.2 28.7 1.424 453.5 5.1

100. .10700 .5931 65.6 29.4 1.424 455.1 7.8
200. .10700 .5931 104.4 29.8 1.424 440.7 3.8
198. .03770 .2090 136.7 30.1 4.042 305.3 3.5
150. .03770 .2090 111.4 31.9 4.042 310.3 4.6
100. .01840 .1020 113.7 32.5 8.283 202.4 6.2
150. .01840 .1020 158.1 32.3 8.283 196.1 4.1
200. .01840 .1020 185.0 32.3 8.283 215.3 3.2
168
The % Error term of Tables E.1 to E.3 was derived
using the following relations:
h
A AT
dh dP dA dAT
+ +
h P A AT
where AT = Tw - Ts. After application of measurement
precision,
dh 5 .5 1
= __
h P A AT
dh
% Error = 100 *
h
169
Table E.4. Table of Experimental Conditions and Heat

Transfer Coefficients for Falling Solids.
Linear Heat
Particle Solids Transfer
Diameter Velocity Coefficient Material
(m) (m/s) (W/m-K)
0.11 0.52 98.72 Sand

0.11 0.52 112.75 Sand
0.11 0.52 98.0 Sand
0.11 0.52 103.93 Sand
0.8 1.84 68. Sand
0.8 1.84 64.3 Sand
0.8 1.84 54. Sand
1.2 1.89 57. Sand
1.2 1.89 50. Sand
0.214 1.65 89. Copper
0.214 1.65 107. Copper
170
Appendix F
Statistical analysis
Table F.1. Regression Analysis of Model.
to
Linear model: = a + bL
dp p s
Standard
Parameter Estimate Error Value
Intercept (a) 0.362193 0.017465 20.7375

Slope (b) 9.69099 0.254237 38.118
Analysis of Variance
Source Sum of Squares Df Mean Square F-ratio

Model 3.0933 1 3.0933 1452.98
Error 0.017031 8 0.0021289
Total (Corr.) 3.110348 9
Correlation Coefficient = 0.997258

Stnd. Error of Est. = 0.046140
R-squared = 99.45%
Table F.2. List of All Data used to Fit the Model.
Heated Particle tcr Material Author

Length Diameter
(m) (m)
(m211,4)
0.005 0.0002 0.432 sand Ernst

0.005 0.0004 0.421 sand Ernst
0.005 0.0006 0.439 sand Ernst
0.0116 0.00016 0.519 sand Desai
0.0116 0.0008 0.472 sand Desai
0.0116 0.00015 0.461 copper Desai
0.0116 0.00017 0.407 copper Desai
0.0116 0.00063 0.449 copper Desai
0.1524 0.00018 1.776 glass This study
0.1524 0.0008 1.905 glass This study
171
Appendx G
Particle Size Distribution
Table G.1. Size Distributions of Particles
Silica Sand
dp = 0.11mm dp = 0.8mm
Tyler Mesh %Weight Tyler Mesh %Weight
-42 + 60 2.18 -20 + 24 35
-60 + 80 5.5 -24 + 28 21.37
-80 + 100 9.57 -28 + 35 21.16
-100 + 115 17.83 -35 + 48 15.21
-115 + 150 19.2 -48 7.26
-150 + 170 16.37

dp = 1.2mm
-170 + 200 13.4 Tyler Mesh %Weight
-200 + 250 6.77 -10 + 14 67.84
-250 + 270 5.25 -14 + 20 31.25
-270 + 325 2.5 -20 + 24 0.693
-325 1.43 -24 0.218
Copper
dp = 0.214mm
-20 + 24 0.3 -80 + 100 11.70
-24 + 28 0.78 -100 + 115 3.35
-28 + 32 5.24 -115 + 150 3.78
-32 + 48 43.3 -150 + 170 4.98
-48 + 60 1.75 -170 7.26
-60 + 80 17.52
172
Glass
dp = 0.8mm dp = 0 18mm
-14 + 20 23.94 -45 + 60 0.5
-20 + 24 72.2 -60 + 70 38.8
-24 + 28 3.05 -70 + 80 49.7
-28 + 32 0.39 -80 + 100 6.0
-32 + 35 0.2 -100 + 120 2.0
-35 + 40 0.17 -120 + 140 3.0
-40 0.05

O Bus Kovic Nevenka 1989

Uploaded by

Copyright:

Available Formats

O Bus Kovic Nevenka 1989

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

O Bus Kovic Nevenka 1989

Uploaded by

Copyright:

Available Formats

AN ABSTRACT OF THE THESIS OF

Nevenka Obiftkovid for the degree of Doctor of Philosophy

in Chemical Engineering presented on September 28, 1988.

Title: Heat Transfer Between Moving Beds of Solids and a

Heat transfer to moving packed beds of solids has been studied

extensively but predictions are usually specific to the experimenter's

own equipment. In order to obtain a more general predictive equation,

immersed in a moving packed bed of glass beads, sand, or copper in air

at atmospheric pressure and compared to measurements performed by

independent researchers with various experimental configurations.

High speed photography was used to observe the motion of the

layer. Likewise, the particles in the layer adjacent to the wall

showed very little rotational or cross-flow motion.

A numerical method, based on the unsteady state conduction

equation, was developed to predict the heat transfer coefficient

could be used to predict the experimental heat transfer coefficients

obtained provided the correct value of the particle separation from

suitable for predicting the heat transfer coefficient.

The experimental data showed good agreement with the analytic

solution obtained by Mickley and Fairbanks (1) at long contact times.

Fairbanks solution but the heat transfer coefficient tends to level

off at an asymptotic value.

An empirical correlation was obtained by which the contact time

be predicted. The following equation, was found to be valid for the

different geometries in their work.

This equation may be used in conjunction with the Mickley and

Fairbanks solution to predict the heat transfer coefficient for a

given flowing packed bed over a wide range of contact times.

Oregon State University

Completed Sept. 28, 1988

Commencement June 1989

Redacted for privacy

Redacted for privacy

Chairman of Chemic'alfEngineering Department

Redacted for privacy

Date thesis is presented Sept. 28, 1988

Typed by Kevin Greek for Nevenka Oblrtkovie

I would like to express my thanks to my major

professor Dr. James G. Knudsen for all the understanding

and financial support at times when this job seemed to

be a neverending story; also for reading and correcting

Special thanks are extended to my dear professor

Dr. Octave Levenspiel not only for his insistent

questions on music, art, geography, history and life in

general but also for the lost backgammon games, jokes

and free lunches that anyone could get if he shows up

in front of his office around 12pm (students only).

Acknowledgments are also extended to my professors and

Jim Fredericks for enormous understanding and support in

the completion of this thesis.

My thanks also go to Nick Wannenmacher for his

invaluable help in experimental research whenever

necessary and to Jordana Chambers expert in solving all

I wish to express gratitude to my friend and

"computer-wiz" Kevin Greek or as I call him Mr. Blasius

"computer" was mentioned; also for correcting and typing

Dr. Vojislav Bogic for their everlasting support

understanding and suggestions.