Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.

pdf by Indian Institute of Technology Bombay user on 01 June 2024

An Experimental and Numerical
Study of Flow and Convective
Heat Transfer in a Freely Falling
J. Hruby
Curtain of Particles
R. Steeper The flow characteristics and convective heat transfer in a freely falling curtain of
spherical particles with an average diameter of 650 ixm has been studied experimen-
G. Evans tally and numerically. Both heated and unhealed particle flows have been con-
sidered. This work is part of a larger study to determine the feasibility of using par-
C. Crowe ticles to directly absorb the insolation in a solar central receiver for high temperature
applications. The particles of interest are Norton Master Beads"1' which are primarily
Sandia National Laboratories, Livermore, aluminum oxide. Measurements have been made of particle velocity in heated and
Livermore, Calif., unheated particle flows, and particle temperature and air temperature in heated par-
Washington State University, ticle flows. Comparison of the measurements with calculations has been made for
Pullman, Wash.
two particle mass flow rates at room temperature and at two initial elevated particle
temperatures. Excellent agreement between numerical and experimental results is
obtained for particle velocity in the unheated flow. For the heated particles, both
data and predictions show the same trends with regard to particle velocity, particle
temperature, and air temperature. However, the calculations of these quantities
overpredict the data. The results suggest that the drag coefficient inflows where the
particles are hot compared to the air is larger than predicted using conventional
methods to account for nonisothermal effects. The prediction of particle
temperature and air temperature attained with a drag coefficient that is larger than
the standard drag coefficient agrees well with the data.

Introduction
The flow characteristics and convective heat transfer in a in a solar receiver and directly absorb the solar insolation
freely falling curtain of particles have been investigated ex- (Martin and Vitko, 1982 and Hruby, 1986b). A conceptual
perimentally and numerically. This work is part of a larger design of a solid particle receiver is shown in Fig. I. The solid
study to determine the feasibility of using particles as direct particle receiver is being evaluated for high temperature
absorbers of the insolation in a solar central receiver. (greater than 550°C) applications of solar energy. The advan-
A solar central receiver system uses mirrors to reflect and tages of a solid particle receiver over traditional fluid-in-tube
focus sunlight onto a receiver located on top of a tower. By receivers are: (1) the particles can directly absorb solar radia-
redirecting the sunlight in this manner, the solar flux at the tion eliminating thermal fatigue limitations on tube materials,
receiver surface can be equivalent to hundreds of suns. A and (2) the particles have high volumetric heat capacities and
working medium heated in the receiver can be used to produce maintain their integrity at high temperatures. These advan-
electricity, or incorporated into products as process heat. Cur- tages, coupled with the possibility that the particles can serve
rently, the working media in solar central receivers are gases or as the storage medium, could enable the solid particle receiver
liquids which flow through tubes heated by the solar radia- to be a cost effective means of high temperature solar energy
tion. An example of such a system is the water/steam central utilization. High temperatures are attractive for fuels and
receiver system located near Barstow, California, which pro- chemicals production, industrial process heat applications,
duces electrical energy (Radosevich, 1985). Other common and Brayton cycle electricity generation.
working fluids include molten salts, molten metals, and air. In order to evaluate the performance of solid particles as the
The concept of interest here, termed the solid particle working media in a central receiver system, the behavior of
receiver, employs sand-size refractory particles which free-fall freely falling particle curtains must be understood. The air-
particle flow is expected to be dilute in the receiver. Dilute
Contributed by the Fluids Engineering Division for publication in the JOUR-
flow implies that the particle velocity is controlled by the
NAL OF FLUIDS ENGINEERING. Manuscript received by the Fluids Engineering aerodynamic drag and the gravitational force, not by particle-
Division January 28, 1987. particle collisions. This report presents experimental data and

172/Vol. 110, JUNE 1988 Transactions of the ASME

Copyright © 1988 by ASME
 numerical predictions of the aerodynamic and thermal
behavior of heated and unheated particles freely falling in in-
Particle Dispersion itially quiescent air at room temperature. Other experimental
studies of dilute freely falling particle curtains for the solid
particle receiver project include the development of a particle
velocity measurement system (Hruby and Burolla, 1984) and
an experiment in which the particles were heated with a ra-
Insulated diant flux (Burolla, Hruby, and Steele, 1984). Numerical work
Cavity
has focused on developing a model of a solid particle solar
central receiver including radiative transfer within the particle
cloud (Houf and Greif, 1985), and including the two phase

Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
flow and heat transfer of air and particles within the cavity
receiver (Evans et al., 1986a). This model also accounts for
convection and radiation from the walls of the cavity.
The numerical and experimental work performed previously
has demonstrated the importance of two-way coupling be-
tween the particles and the air. Both measured and predicted
velocities in the curtain were higher than the velocity of an
Aperture for isolated sphere in free-fall. The higher particle velocity is the
Solar Flux result of entrainment of the surrounding air by the particle
curtain, and can only be predicted with two-dimensional
modeling and momentum coupling between the air and parti-
cle phase (Evans et al., 1986b). The previous work has also
shown that coupling between momentum and energy is
necessary for correct prediction of particle temperature.
Momentum and energy coupling is important because the
particle-air heat transfer can influence both particle velocity
and particle temperature. The present experiment was de-
signed to study particle velocity in both heated and unheated
particle curtains, as well as particle temperature and air
temperature in heated flows. The influence of initial particle
Fig. 1 Conceptual design of a solid particle central receiver temperature and mass flow rate was investigated.

Nomenclature

a = Stefan-Boltzmann con
stant = 5.669E-08
CD = drag coefficient W/(m 2 -K 4 )
3
FP = drag force exerted on a Ppart
specific heat of particle, P = density, kg/m
single particle in the x- J/(kg-K) X = ratio of CD/CDstokes =
direction, N d„ = particle diameter, m or ( V X l + O.^Ref3)
N = particle number flow /urn jX dynamic viscosity,
=
rate acceleration due to kg/(m«s)
Nu = Nusselt number = hdp /k gravity, m/s 2 v = kinematic viscosity,
P = pressure, N/m 2 h = convective heat transfer m 2 /s
Pr = Prandtl number coefficient, W/(m 2 «K)
= heat transfer from a k = thermal conductivity, Subscripts
QP
particle to air = W/(m-K) eff = effective viscosity or
NukfTrdp(T~Tp), W particle number density, conductivity
Re = particle Reynolds particles/m 3 / = evaluated at the film
number = pdp lup — u 1 I\i t = time, s temperature
s/-y = gas x, v momentum At, = particle transit time g = gas phase
equation source term across a computational / = the rth trajectory
due to particles, N cell along the z'th trajec- P = particle phase
^P = gas energy equation tory, s s = surface
source term due to par- u = vertical velocity, m/s turb = turbulent quantity
ticles, kg»K/s V = horizontal velocity, m/s «, s, e, w = values evaluated at
T = temperature, K X = vertical position coor- north, south, east, or
T*g = dimensionless gas dinate, m west faces of control
temperature = y = horizontal position volume
(r,-7^/(7^-7^) coordinate, m oo = evaluated at ambient
T; = dimensionless particle Ax = vertical grid spacing, m conditions
temperature = Ay = horizontal grid spacing,
m Superscripts
( ^ - r j / ^ - r j
T„n = initial particle particle emissivity or /' = initial particle
temperature, K turbulent dissipation, conditions
T = ambient temperature, K m 2 /s 2 * = nondimensional
specific heat of air, turbulent kinetic energy, quantity
J/(kg«K) m 2 /s 2 x, y = x- or jc-direction

Journal of Fluids Engineering JUNE 1988, Vol. 110/173

 Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
Fig. 3 Funnel·shaped Inserts used In the hopper to provide a uniform
temperature particle curtain at the hopper exit

Fig. 2 Photograph 01 experimental apparatus Identifying three

assemblies: (1) hopper, (2) catch bin, and (3) temperature probe platform Fig. 4 Probes used to measure particle temperature and air
temperature In the particle curtain

The calculations and experiments were performed using provide two particle mass flow rates. For the low mass flow
Norton Master Beads™ with an average size of 650 !-tm (94.6 rate, the slit was covered with a 3.2 mm stainless steel screen.
percent between 417 and 71O!-tm and 5.4 percent between 250 Slit dimensions in this case were 64 mm by 6.4 mm (aspect
and 417 !-tm). Master Beads™ are 86 percent aluminum oxide; ratio of 10: 1) and the particle mass flow rate was about 0.02
the remaining 14 percent is comprised of near equal amounts kg/so For high mass flow rate, the screen was replaced with a
of silica, iron oxide, and titania. Master Beads™ have been single strand of 1.5 mm wire oriented longitudinally along the
identified as particles that have high solar absorptivity, do not slit. With this arrangement, and slit dimensions of 51 mm by
agglomerate at 1000°C, and have high fracture resistance 5.1 mm, the particle mass flow rate was approximately 0.04
(Hruby and Steele, 1986a). kg/so
Catch Bin. The catch bin assembly, shown in Fig. 2, con-
Description of the Experiment sisted of a funnel, an insulated receptacle, and an electronic
The experimental apparatus consisted of the following five scale platform. Falling particles were collected by the wide-
assemblies: (1) hopper, (2) catch bin, (3) temperature probe mouthed funnel and channeled to the receptacle. The catch
platform, (4) laser Doppler velocimetry (LDV) system, and (5) bin weight was successively measured by the scale to yield the
data acquisition system. These assemblies are described in this particle mass flow rate.
section.
Temperature Probe Platform. The probe platform, also
Hopper. The hopper (see Fig. 2) was a cylindrical, shown in Fig. 2, was a moveable stage mounted on a vertical
stainless steel receptacle that held 20 kg of heated Master beam next to the particle curtain. It supported the temperature
Bead™ particles. The particles were discharged through an exit probes and could be positioned at any elevation from the hop-
slit to form a steady, freely falling curtain. The hopper was per to the catch bin.
suspended from a hoist and could be positioned from a few Two types of temperature probes, shown in Fig. 4, were
centimeters to 3.5 m above the catch bin assembly. used depending upon whether the particle temperature or air
A uniform particle temperature at the hopper exit was ob- temperature was being measured. For measurement of particle
tained in the following manner. First, a convection oven was temperature, a sampling cup consisting of a small, stainless
used to uniformly heat the particles prior to loading them in steel foil cup fitted with a bare type K thermocouple junction
the hopper. Second, the hopper was ringed with band heaters was constructed. A hole was drilled in the bottom to allow a
controlled to maintain a constant interior wall temperature. steady flow of particles past the junction when the cup was
Third, two funnel-shaped inserts (see Fig. 3) were stacked just placed in the particle curtain. Several sizes and shapes were
above the exit slit to insure a uniform flow of particles within tried, the final choice being a short cylinder 10 mm high by 28
the hoppeL It was observed that without these inserts, cooler mm in diameter with a 3 mm hole in the bottom. Disturbance
particles from the top surface would funnel down the center of the particle curtain by the sampling cup was small as shown
and exit giving rise to a cool region within the curtain. Finally, in Fig. 5.
three 1.6 mm shielded thermocouple probes were fixed ver- Air temperature was measured with a rake consisting of
tically in the hopper with their tips 13 mm above the exit slit to three bare type K thermocouples. The junction and last 7 mm
record the temperature of the particles at this location. of lead wire of each thermocouple were bent at right angles.
The particle curtain was generated by a rectangular slit in When positioned in the curtain with the right angle section
the bottom plate of the hopper. The slit could be modified to parallel to the flow and the junction downstream (see Fig. 6),

174/Vo1.110, JUNE 1988 Transactio!1S of the ASME

 Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
Fig. 6 Air temperature rake In the particle curtain during data
Fig. 5 Particle temperature sampling cup In the particle curtain during collection
data collection

the thermocouples were shielded from contact with falling Experimental Procedure
particles and so measured air temperatures. The three junc-
tions were spaced 18 mm apart. The measurement of air Preparation. Master Bead™ particles were heated in a con-
temperature with thermocouple probes turned downstream vection oven at the selected temperature for a minimum of five
has been performed successfully by others (Brewster and hours prior to running an experiment. The hopper was
Seader, 1984). preheated for a minimum of one hour. The hopper
A HeNe laser mounted on the elevator stage was used for temperature was set slightly cooler than the temperature of the
visually positioning the stage with respect to reference marks convection oven to compensate for heat loss in transferrring
on the wall. A linear position transducer was attached to the particles from the oven to the hopper. After the hopper was
elevator to provide an accurate position measurement for the filled with particles and positioned, data collection began.
data acquisition system. Data Collection. Experiments were run using one of two
LDV System. Vertical particle velocities were measured routines. For the first routine, the hopper was fixed at the top
with a TSI laser Doppler velocimetry system mounted on a of its travel, about 3.5 m above the catch bin. The elevator
fixed table. The system used a Lexel4W argon ion laser tuned stage with temperature probes was moved from top to bottom
to 514.4 nm with optical components configured in a to sample at eight selected drop heights. Particle temperatures
backscatter mode. The final lens had a focal length of 2.2 m and air temperatures were measured using this routine.
allowing sufficient removal of the system from the hot particle Because the particle velocity measurement (LDV) system
curtain. This application of LDV was unconventional because could not be traversed vertically, the second routine required
of the large particle sizes, and attention to the diameter-to- moving the hopper rather than the elevator to acquire particle
fringe spacing ratio was required. A detailed discussion of the velocity as a function of height. For each run, the hopper was
LDV system can be found in Hruby and Burolla (1984). positioned to give the desired drop height between the hopper
and the fixed position of the LDV laser. The temperature
Data Acquisition System. The data required from this ex- probe platform was also fixed at the LDV height, and a repeat
periment }Vere particle temperature, air temperature, particle set of particle temperature data was gathered along with
velocity, and spread of the particle curtain, all as functions of velocity values.
fall height (the distance from the hopper discharge slit). A The step-by-step procedure of the two routines was similar.
Hewlett Packard 9816 computer, 3497A scanner, and 3456A First, the hopper and elevator were positioned. The hopper
digital voltmeter were used to acquire most of this informa- door was opened and the particle sampling cup was inserted
tion. Data collected with the computer system included and held centered in the particle curtain. About ten seconds
weights from the scale, voltages from the thermocouples, fre- was allowed for equilibration of the curtain and the various
quencies from the LDV counter, voltages from the position thermocouples; then the computer sampling program was trig-
transducer, and time from the HP 9816 internal clock. gered. Data from the digital scale, position transducer, clock,
Particle curtain dimensions were acquired by taking still plus hopper slit and sampling cup thermocouples were col-
photographs of a roughly eight inch square region of the fall- lected in about two seconds. Eight values from each ther-
ing curtain at various drop heights. Both front and edge views mocouple were taken and averages and standard deviations
of the curtain falling at high and low mass flow rates were were calculated.
filmed with the particles at room temperature (particle Depending on the routine, either air temperature or particle
temperature was not expected to affect spread). Width and velocity was measured next. For air temperatures, the ther-
depth measurements were scaled from the photographs to give mocouple rake was inserted immediately after withdrawing
approximate spread-versus-fall height data. the sampling cup. While the rake was held centered in the cur-

Journal of Fluids Engineering JUNE 1988, Vol. 110/175

tain, the computer collected eight values from each ther- switching from central to upwind when the absolute value of
mocouple. For velocity measurements, the computer collected the cell Reynolds number is greater than two.
and averaged forty values as soon as the particle sampling cup
was withdrawn and the laser aperture opened. Gas Equations. The integrated forms of the conservation
When these measurements were complete, the hopper door equations for mass, momentum, and energy with a planar
was shut and the elevator or hopper was positioned for a geometry are given by:
subsequent run. A set of runs for eight different drop heights (pu)%.Ay + (j>v)?>Ax = 0 (1)
comprised one experiment and took from five to nine minutes e
/ du \ ( du \"
to complete.
{puu-^.—)^ .Ay+ (evu-^.—^ .Ax
Three parameters were varied for the different experiments:
collection routine, initial particle temperature, and mass flow

Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
rate. As discussed above, there were two collection routines. = (Pw-Pe)'Ay + pg'Ax'Ay
Three initial particle temperatures were used: room
temperature (isothermal), 603 K, and 773 K. Finally, mass
flow rate was set at either 0.02 kg/s or 0.04 kg/s. At least two
du dv
+ /%f' •Ay + /*eff« • Ax + Si (2)
experiments were run at each permutation of these three ~~dx~ dx
parameters.
Uncertainty. Some aspects of the chosen procedure and
apparatus introduced uncertainty in the experimental results. (puv-^r~)l 'Ay+ (^-^'-|r)",A*
First, as discussed above, the particle velocity measurement
routine involved changing the overall fall height of the particle du dv
= (P,-P„)-A*-/xeff.- •Ay + Meff' -Ax + S>; (3)
curtain. This would not be expected to influence the results
unless disturbances were propagated upstream. By comparing
particle temperature measurements from the two different r / k\ dTi" r ^ /A:\ Sri"
routines, it was shown that, in fact, altering the downstream
conditions made no difference. Therefore, the only important
K - t ) - r - a d - '^ V - t)e f f ~djL '** = S> VT
length was the distance between the height of the hopper door (4)
and the elevation where the measurements were made; total where e, w, n,s indicate that the corresponding terms are to be
fall heights and conditions below the measurement location evaluated at the east, west, north, and south faces of the con-
were not important. trol volume, and Ax and Ay are the control volume dimensions
Another source of uncertainty with respect to particle in the x and y directions, respectively. The reduction in the
temperature was the use of the particle sampling cup due to its area of the gas phase on the control surface due to the
thermal mass and possible flow disturbance. Thermal mass ef- presence of the particles is negligible. The dissipation term in
fects were shown to be negligible by good agreement of hopper the energy equation is small and has been neglected. The sign
exit temperatures with sampling cup temperatures taken just of the body force term for the x component of momentum is
below the exit. Since the sampling cup disturbed only the such that the coordinate system is oriented with the x direction
downstream flow, this effect was negligible for the reasons aligned with gravity. The source terms, Spx, Spy, and SpT,
discussed above. refer to the momentum and energy added to the gaseous phase
by the particles. Additional equations and relationships for
the gas flow solution are:
Numerical Analysis
(1) Pressure is determined using the SIMPLE procedure
A model of dilute gas-particle flows with heat and mass described by Patankar (1980) which is formulated to insure
transfer has been developed by Crowe et al. (1977). The model local continuity;
(PSI-Cell, i.e., Particle Source in Cell) includes two-way mass, (2) Differential equations for turbulent kinetic energy, K,
momentum, and thermal coupling between the phases and has and dissipation, e, are solved as presented in Launder and
been applied to simulations of spray drying (Crowe, 1980), Spalding (1972);
electrostatic precipitators (Eschbach and Stock, 1979), cyclone (3) Effective viscosity and conductivity are given by
separators (Crow and Pratt, 1974), and combustion (El-
Bainhawy and Whitelaw, 1980). The PSI-Cell code was A'eff - ^ + i"turb (5)
modified for the present study by including buoyancy in the
air and temperature dependent properties of the air and the i"turb=c,<^2/e. cM = 0.09 (6)
particles. A single particle radiation model was also included
in the particle energy equation. Briefly, the PSI-Cell code con-
sists of a steady, two-dimensional planar or axisymmetric, - = J*eff- (7)
elliptic, Eulerian description of the gas flow field coupled with
a Lagrangian description of the particle flow field. The gas Equation (7) is the result of assuming that the effective
flow field is determined using TEACHT (Gosman and Pun, Prandtl number is unity. The influence of the particles on the
1973), which solves the conservation equations on staggered effective viscosity and conductivity of the air is neglected.
control volumes with the pressure, density, and temperature (4) An ideal gas equation of state is used and Sutherland
evaluated at control volume centers and the velocities law relationships for the dependence of viscosity and thermal
evaluated at the control volume faces. A two equation (/c-e) conductivity on temperature are prescribed (White, 1974).
model of turbulence is included with constants established for Particle Equations. The particle velocity is calculated
a forced flow (Launder and Spalding, 1972). Convection and assuming that the aerodynamic drag and the gravitational
diffusion of a dependent variable are combined into a single body force are responsible for the motion; that is, the flow is
term, which when integrated over the control volume, dilute. This assumption is justified based on the fact that the
represents the flux of that variable across the faces of the con- particle volume fraction is on the order of 10~2 at distances
trol volume. These terms are evaluated using hybrid differenc- greater than 10 cm from the hopper exit slit, and the particle
ing (Spalding, 1972). This differencing scheme is a combina- volume fraction decreases substantially as the particles ac-
tion of central differencing and upwind differencing, celerate and the curtain widens.

176/Vol.110,JUNE1988 Transactions of the ASME

 The particle momentum and energy equations are given by being evaluated. A similar expression holds for they direction
wdp 3
Du . . . . . , -Kdp 3 source term. The energy equation source term for the gas is
given by:
'"' 6 "of ' ' "—6
V=£(e,,AP)^ d5)
P P ' ^ ' ~ ^ = ^dptiK{v-vp) (9)
where Qpi =Nukfivdp(Tp — T) is the convection heat transfer
mI Nu rate from a particle to the air in the /'th trajectory.
Pe%ar, T~'^r = */*4> (T-Tp)~ eardpHTp* - TJ)
(10) Initial Particle Conditions. As the particles flowed from

Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
where \ = CDRe/24, and Re is the particle Reynolds number the hopper in the experiment, there was a component of the
based on the free stream properties and relative velocity. The particle velocity in both the horizontal and vertical directions.
radiative loss term represents the radiative heat transfer from The (small) horizontal component, which arose due to either
an isolated particle and is a conservative estimate of the particle-particle collisions within the hopper or the nature of
radiative heat loss. The particle energy equation is based on a the discharge from the hopper, was not measured. A range of
uniform particle temperature which is justified by the fact that initial horizontal velocities from 0.3 cm/s to 3.0 cm/s for the
the Biot number for particles in the experiment is less than 0.1 particle flow was provided in the calculation. This range in
(Incropera and DeWitt, 1981). particle velocity was sufficient to reproduce the measured
The drag coefficient used in the model corresponds to the spread of the particle curtain (the measured spread was
steady state drag coefficient for an isolated sphere. The em- roughly 3 cm total at a 2 meter fall height). The particle mass
pirical equation used is flow rate was divided into ten parts, with each part repre-
senting the same fraction of the total flow rate. These parts
24 were distinguishable only by different initial conditions on
CD=~ (l+0.15Re/3) (11)
particle velocity. Each part was given an initial temperature
where Rey is the Reynolds number based on the film corresponding to the measured temperature in the hopper just
temperature (which is the average of the particle temperature above the discharge slit. The particles were considered to be
and the bulk air temperature) and the relative velocity between spherical, with a uniform size of 650 jim, density of 3130
the particle and the air. This equation was developed for kg/m 3 , and a specific heat that was a function of temperature
isothermal conditions with Reynolds numbers up to 103 (Clift (1057 J/kgK at 603 K).
et al., 1978). The effects of pressure gradient and particle ac-
celeration on the drag coefficient are negligible because the Calculational Procedure. The calculations were per-
material density ratio between the air and particles is less than formed in a vertical cylinder or channel, depending upon
10~ 3 . Tsuji et al. (1982) found that the drag of a particle is whether axisymmetric or planar geometry was considered.
unaffected by neighboring particles if the particle separation Symmetry was assumed about the cylinder (channel) centerline
distance is ten or more diameters. This condition is met over with the solution obtained over half of the cylinder (channel).
the greater portion of the particle's trajectory in the falling The cylinder (channel) was 1.5 meters in radius (half-width)
curtain. The effect of turbulence on drag coefficient is dif- and 4.0 meters high. The particles were introduced at the top
ficult to quantify because turbulence intensity data are of the cylinder (channel) near the centerline with initial
unavailable. If it is assumed that the root mean square of the downward velocity and temperature as given in the previous
turbulence generated by the particle is the product of the parti- section. Although the experiment did not have confining
cle diameter and vortex shedding rate, a relative turbulence in- walls, the convergence of the calculations and the specification
tensity of 0.1 is calculated (Clift, et al., 1978). At this value of boundary conditions for the elliptic equations is aided by
and at the relative Reynolds number of the falling particles, the addition of a solid boundary. The effect of the outer wall
the effect of turbulence on the particle drag coefficient is on the results for particle velocity and temperature when the
minimal. Thus the expression used for X is equations (8) and radius (half-width) of the cylinder (channel) was varied from
(9) becomes 1.0 to 2.0 meters was less' than 5 percent. Slightly larger
changes of approximately 10 percent occurred in the air
A=(j< / /e)(l+0.15Re / 2/3 ) (12) temperature. The computational mesh consisted of 20 axial
where v is the kinematic viscosity and the nonsubscripted value grid lines (in the direction of fall of the particles) and 16 lateral
corresponds to free stream conditions. grid lines. The grid spacing was non-uniform in both direc-
The Ranz-Marshall correlation is used for the Nusselt tions with the smallest axial grid spacing (4 cm) at the top of
number, namely (Bird et al., 1960) the cylinder (channel) near the particle source and the smallest
lateral grid spacing (2 cm) at the symmetry axis and within the
Nu = 2 + 0.6Re/ / 2 Pr/ / 3 . (13) particle curtain. Since the spread of the particle curtain was
The effect of neighboring particles on the Nusselt number is only a few centimeters, the computations did not yield detailed
not included. There are no data available to assess this effect information on the distribution of the temperature and veloci-
but it is presumed small because the effect of neighboring par- ty of the air within the curtain nor was this information
ticles on the drag coefficient is small. available from the results of the experiment. The sensitivity of
the calculated results to the grid size was determined to be less
Gas Source Terms. The force in the x direction on the gas than 1 percent for particle velocity and temperature and less
in the computational cell due to the particles is given by: than 5 percent for air temperature when the number of grid
lines was doubled in both directions.
The calculations were initiated by assuming a stagnant air
i
x
field. The particle trajectories were calculated and the source
where F pi = 'i-Kdpiik{up — u) is the aerodynamic force on the terms evaluated. The gas flow equations were solved incor-
gas due to a particle, N, is the particle number flow rate, and porating the particle source terms which gave rise to motion in
At, is the particle transit time across the cell, all for the /th tra- the gas flow field (entrainment). The particle trajectories were
jectory. The sum is applied over every trajectory passing recalculated and the source terms re-evaluated. The gas flow
through the computational cell for which the source term is field was calculated again and the cycle was continued until

Journal of Fluids Engineering JUNE.1988, Vol. 110/177

 t
-"

(m/s)
6.0
^ < ^ ^ " - .E>

VELOCITY
l"-^" " " '* ' . •'

5.0
4.0
s^'"--^" •s 9

i. / y
PARTICLE VER1

HIGH MASS FLOW RATE Tp(i) = 773 K

model - low flow
2.0

axisymmetric
•data - low flow
// planar
:/
1.0

A LDV meosuremenis mode) - high flow

Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
isolated particle a data - high flow
0

O -
' 1 I"' '1 1 - . . ——r.. i - , - 1 L-, ,

FALL HEIGHT (m) FALL HEIGHT (m)

Fig. 9 Numerical prediction a n d experimental data of particle velocity

Fig. 7 Numerical prediction and experimental data of particle velocity in elevated particle temperature flow with two particle mass flow rates.
in isothermal particle-air flow. (Uncertainty in velocity. ± 0 . 2 5 m/s, (Uncertainty in velocity: ± 0 . 2 5 m/s, Uncertainty in fall height: ± 0 . 0 1 m at
Uncertainty in fall height: ± 0 . 0 1 m at 20:1 odds.) 20:1 odds)
7
(m/s)
6.0
PARTICLE VER1TCAL VELOCITY
5.0

« * * '*
4.0

8
3.0

a 3 •
a
LOW MASS FLOW RATE
2.0

model - 603 K LOW MASS FLOW RATE

• data - 603 K model - 603 K
1.0

model - 773 K • data - 603 K

» data - 773 K model - 773 K
0.0

1 1 '
0.0 0.8 1.6 2.4 3.2 4.0 * data - 773 K
FALL HEIGHT (m)
FALL HEIGHT (m)
Fig. 8 Numerical prediction a n d experimental data of particle velocity
in elevated particle temperature flow with low particle mass flow rate.
Fig. 10 Numerical prediction a n d experimental data of dimensionless
The dashed and solid curves w h i c h represent the numerical results at
particle temperature for low particle mass flow rate at t w o initial
different temperatures overlap. (Uncertainty in velocity: ± 0 . 2 5 m/s,
elevated particle temperatures. (Uncertainty in temperature: ± 0 . 0 2 5
Uncertainty in fall height: ± 0 . 0 1 m at 20:1 odds.)
m/s, Uncertainty in fall height: ± 0 . 0 1 m at 20:1 odds.)

convergence was achieved. This procedure took one to two

minutes of computer time on the Cray-1. Heated Flow Comparison. Calculations of particle veloci-
ty and temperature, and air velocity and temperature have
Comparison of Experimental and Numerical Results been made for the two mass flow rates and the two initial
elevated particle temperatures. In this section, comparisons
Unheated Flow Comparison. The geometry of the slit between calculations and experimental results are made for
through which the particles exit the hopper was described particle velocity, dimensionless particle temperature, and
earlier as being rectangular with an aspect ratio of 10:1. It was dimensionless air temperature. Although buoyancy effects for
not clear that this aspect ratio was large enough to produce a the air have been taken into account and a radiation heat loss
purely two-dimensional, planar entrained air flow. Since the term has been included in the particle energy equation, both of
axisymmetric geometry allows for uniform air entrainment at these effects are negligible for the conditions considered here.
a given distance from the axis of symmetry, both axisymmetric
and planar geometries were used initially to determine which, Comparison of Particle Velocity. The calculated and
if either, would result in acceptable agreement with particle measured vertical components of particle velocity as a func-
velocity measurements made in the isothermal drop tests. tion of distance from the hopper for the low mass flow rate
and for the two initial particle temperatures (603 K and 773 K)
Figure 7 shows the results of the comparison between are shown in Fig. 8. The effect of initial temperature on the
calculated and experimental vertical particle velocity as a func- predicted values is not discernible, and the effect on the
tion of distance from the hopper exit for the high mass flow measured values is small. The calculations overpredict the
rate of isothermal air-particle flow. Calculational results for measured values of particle velocity.
both planar and axisymmetric geometries are shown. The
results for the planar geometry are in excellent agreement with Figure 9 shows the calculated and measured particle vertical
measurements whereas the axisymmetric results underpredict velocity profiles for the two mass flow rates at the single initial
the data. Consequently, the planar geometry was used for all particle temperature of 773 K. The calculations overpredict
subsequent comparisons. Also shown in Fig. 7 is the result for the velocity data for the high mass flow rate as well. The trend
a single isolated spherical particle falling in quiescent sur- in both calculations and measurements toward higher particle
roundings obtained by integrating equations (8)~(9) with air velocities as the mass flow rate of particles increases is evident.
velocity components, u and v, set to zero. By comparison, par- The higher particle mass flow rate results in more air entrain-
ticles in a curtain fall faster due to the entrainment of air ment, increased downward air velocity in the curtain, and
which results in a downward component of air velocity in the augmented particle fall velocities.
curtain. Comparison of Particle Temperature. For comparison

178/Vol. 110, JUNE 1988 Transactions of the ASME

 LOW MASS a o w RATE
model - 603 K
• dota - 603 K
model - 773 K
i
• " data - 773 K
••
Tp(i) = 773 K
mode! - low flow
• • ^ ^ .
• data - low flow
model - high flow

Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
A data - high flow
i i ' 1 1 , .
0.0 0.8 1.6 2.4 3.2 4.0
FALL HEIGHT (m) FALL HEIGHT (m)

Fig. 11 Numerical prediction and experimental data of dimensionless Fig. 12 Numerical prediction and experimental data of dimensionless
particle temperature for t w o particle mass flow rates at an initial particle air temperature for low particle mass flow rate at two initial elevated par-
temperature of 773 K. (Uncertainty in temperature: ± 0 . 0 2 5 m/s, Uncer- ticle temperatures. (Uncertainty in temperature: ± 0 . 0 7 , Uncertainty in
tainty in fall height: ± 0 . 0 1 m at 20:1 odds.) fall height: ± 0 . 0 1 m at 20:1 odds.)

purposes, the particle temperature has been non-

dimensionalized: T*= (Tp-T„)/(Tpo — T„). Dimensionless Tp(i) = 773 K
particle temperature is shown as a function of the distance model - low flow
from the hopper in Fig. 10 for the low mass flow rate case and • data - low flow
for the two initial particle temperatures. The difference be- •\ model - high flow

tween the results for the two initial temperatures shown is A data - high flow

quite small as a result of the nondimensionalization and the in-

significance of radiative heat transfer. Calculations of particle
temperature overpredict the measured values. This result in-
dicates that the calculations underpredict the total heat
transfer to the air. • ^ £ _ _ _ _

«
Dimensionless particle temperature as a function of distance
from the hopper for the two mass flow rates and for an initial
particle temperature of 773 K is shown in Fig. 11. Both FALL HEIGHT (m)
calculated and experimental results indicate that as the particle
mass flow rate increases, the dimensionless particle Fig. 13 Numerical prediction and experimental data of dimensionless
air temperature for two particle mass flow rates at an initial particle
temperature at a given fall height increases. The increase in temperature of 773 K. (Uncertainty in temperature: ± 0 . 0 7 m/s, Uncertain-
dimensionless particle temperature at high mass flow rates is ty in fall height: ± 0 . 0 1 m at 20:1 odds.)
due to two factors; the increased particle concentration leads
to a higher air temperature within the curtain and the in-
creased relative particle velocity reduces the time available for true that the air is losing energy because a significant mass
heat transfer. flow of air is occurring due to entrainment by the particles.
For example, for the higher particle temperature and higher
Comparison of Air Temperature. The measured and mass flow rate case, the downward mass flow rate of air at
calculated dimensionless air temperature, T* = distances of 14, 89, and 191 cm from the hopper was
(Tg - T„)/(Tpo — T„), is shown on Fig. 12 for the low mass calculated to be 0.054, 0.39, and 0.64 kg/s, respectively.
flow rate case and for the two initial particle temperatures.
The calculated air temperatures are higher than the measured
values. The fact that the predicted air temperature is higher
than measured does not necessarily imply that more energy Drag Coefficient in Heated Flows
was transferred to the air in the calculation; if this were true it Because the calculations overpredict measured particle
would be inconsistent with the higher predicted than measured velocities in heated particle flows, the correlation used for the
particle temperature. If the predicted air velocities are smaller drag coefficient (equation (11)) needs to be examined in more
than the experimental values, which were not measured, then detail. The calculations included the effect of hot particles by
the predicted air temperatures can be higher than those evaluating the Reynolds number of the film temperature in
measured and still result in a smaller predicted than measured lieu of the bulk air temperature. This increased the calculated
heat transfer to the air. As will be discussed in the next section, drag coefficient of a 650 ^m particle at 773 K moving in air at
if the predictions were made with a particle drag coefficient 293 K with a relative velocity of 3 m/s by 70 percent from the
that was too small then it is conceivable that predicted air value obtained assuming uniform air properties at 293 K. Still,
velocities would be smaller than those in the experiment. the data suggest that the drag coefficient in hot particle flows
The variation of dimensionless air temperature as a function is higher than predicted using the film temperature to deter-
of distance from the hopper for the two mass flow rates and at mine the Reynolds number.
the higher initial particle temperature is shown in Fig. 13. Both Very little information is available in the literature on the ef-
calculated and experimental values show an increase with in- fect of surface-gas temperature difference on the particle drag
creasing mass flow rate. This is expected since a higher mass coefficient. An analytic study by Kassoy et al. (1966) predicted
flow rate of particles provides a larger source of heat to the that the drag coefficient of a particle in the Stokes flow regime
surrounding air, and the air mass flow rate does not increase would increase with increasing surface temperature and that
at the same rate as the particle mass flow rate. the Nusselt number would decrease. The same trends result
Note that although both the particle and air temperature are from the use of the film temperature to evaluate the air pro-
decreasing as a function of distance from the hopper, it is not perties. There appears to be no analytic solution available for

Journal of Fluids Engineering JUNE1988, Vol. 110/179

temperature effects in the Reynolds number range of interest cle in free fall. Further, particle temperature measurements
in this study ( 1 0 < R e / < 100). have shown that particle temperature at a given fall height is
Some experimental studies have been reported in the Rus- higher as the particle mass flow rate increases. Calculations
sian literature on the effect of particle surface temperature on using the PSI-Cell computer code, which accounts for the two-
particle drag coefficient. Babiy and Ivanova (1965) propose way momentum and thermal coupling between the particles
that the increased drag coefficient of burning particles is due and the air, are in excellent agreement with the particle veloci-
solely to the increased surface temperature of the particle. ty measurements in isothermal air-particle flow. The predic-
They further propose that the drag coefficient should be cor- tions show the same trends as the measurements for particle
related with a Reynolds number based on the surface velocity, particle temperature, and air temperature in
temperature, not the average film temperature. They cor- nonisothermal, heated particle flow. For the nonisothermal
related a large volume of experimental data and proposed the cases, these quantities are overpredicted. By using a larger

Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
following empirical relationships drag coefficient, the predictions of particle velocity, particle
temperature, and air temperature are in better agreement with
CD=-p-(0<Re,<50) (16) experimental data. The data suggest that an increased drag
coefficient should be used in flows where the particles are hot
compared to the air. The convection coefficient for a single
CB=^r(Res>50) (17) particle seems sufficient to describe the energy exchange in a
dilute particle curtain when the local conditions are known.
where Re s is the Reynolds number evaluated using the
kinematic viscosity of the gas at the particle surface
temperature. This correlation leads to a lower drag coefficient Acknowledgment
than the one used in this study in the range 40 < Re s < 120. The This work was supported by the U.S. Department of Energy
particles falling in the curtain are in this range over the greater under contract DE-AC04-76DP00789.
portion of their trajectory which leads to an even higher
velocity if Babiy and Ivanova's correlation is used. Calcula-
tions were also done using Re s instead of Re^ in the standard References
drag coefficient correlation given by equation (11) but the in- Babiy, V, I., and Ivanova, I. P., 1965, "Aerodynamic Resistance of a Particle
crease in drag was insufficient to account for the lower veloci- in a Nonisothermal Condition (in Russian)," Teploenergetika, Vol. 9, pp.
ty observed in the experimental data. 19-23.
Basina, I. P., and Maksimov, I. A., 1970, "Effect of Nonisothermicity on the
Basina and Maksimov (1970) point out that the coal par- Aerodynamic Drag of a Spherical Particle," Heat Transfer-Soviet Research,
ticles used in Babiy and Ivanova's correlations were not Vol. 2, No. 6, pp. 69-75.
spherical and that the increased drag may partially be due to Bird, R. B., Stewart, W. E., and Lightfoot, E. N., 1960, Transport
the asphericity. They carried out an experiment to measure the Phenomena, Wiley, p. 409.
drag coefficient of a heated, 2.5 mm diameter spherical par- ingBrewster, B. S., and Seader, J. D., 1984, "Measuring Temperature in a Flow-
Gas-Solids Suspension with a Thermocouple," AIChE Journal, Vol. 30,
ticle suspended in a horizontal stream of cooler air over a No. 4, pp. 676-679.
Reynolds number range of 1 to 150. They showed that the Burolla, V. P., Hruby, J. M., and Steele, B. R., 1984, "High Temperature
drag coefficient for heated particles was larger than the stan- Solar Thermal Energy Absorption with Solid Particles," Proc. IBCEC, Vol. 3,
dard drag coefficient. The increase was as much as 45 percent pp. 1663-1668.
Clift, R., Grace, J. R., and Weber, M. E., 1978, Bubbles, Drops and Par-
for a particle at 723 K with a Reynolds number of 15, the ef- ticles, Academic Press, New York.
fect decreasing with increasing Reynolds number to 26 percent Crowe, C. T., and Pratt, D. T., 1974, "Analysis of the Flow Field in Cyclone
at a Reynolds number of 40. This effect is attributed partially Separators," Computers and Fluids, Vol. 2, pp. 249-260.
to increased viscous effects in the boundary layer as noted by Crowe, C. T., Sharma, M. P., and Stock, D. E., 1977, "The Particle-Source-
Babiy and Ivanova, and partially to free convection effects. At in Cell (PSI-CELL) Model for Gas-Droplet Flows," ASME JOURNAL OF FLUIDS
ENGINEERING, pp. 325-332.
low flow velocities they noted a vertical velocity component Crowe, C. T., 1980, "Modeling Spray-Air Contact in Spray-Drying
induced by free convection effects which probably had a Systems," Advances in Spray Drying, Hemisphere Publishing, Chapter 3, pp.
significant effect on the measured drag coefficient. They point 63-99.
out that their data is only valid for a similar flow configura- El-Bainhawy, V., and Whitelaw, J. H., 1980, "The Calculation of the Flow
Properties of a Confined Kerosine-Spray Flame," AIAA Journal, Vol. 18, No.
tion (horizontal flow in a pipe). In the case of freely falling 12, pp. 1503-1510.
particles, the air velocities induced by free convection would Eschbach, E. J., and Stock, D. E., 1979, "Optimization of Collection Effi-
increase the gas-particle relative velocity, and the effect of free ciency by Varying Plate Spacing with an Electrostatic Precipitator," Pro-
convection on the drag coefficient would be greater. ceedings of the 2nd EPA Conference on Transfer and Utilization of Particulate
Control Technology, Denver.
Good agreement between the numerical and the experimen- Evans, G., Houf, W., Greif, R., and Crowe, C , 1986a, "Gas-Particle Flow
tal results for particle velocity in nonisothermal flows was ob- Within a High Temperature Solar Central Receiver including Radiation Heat
Transfer," Journal of Solar Energy Engineering, accepted for publication.
tained when the constant multiplying Re 2/3 in the drag coeffi- Evans, G., Houf, W., Greif, R., and Crowe, C , 1986b, "Numerical Model-
cient of equation (11) was increased from 0.15 to 0.40, an in- ing of a Solid Particle Solar Receiver," Sandia National Laboratories,
crease of 166 percent. Because the calculated results for parti- SAND85-8249.
cle and air temperature also agreed well with the experimental Gosman, A. D., and Pun, W. M., 1973, "Calculation of Recirculating
results for particle and air temperature when the above change Flow," Lecture Notes, Imperial College of Science and Technology, London,
England.
was made to the particle drag coefficient, the discrepancy be- Houf, W. G., and Grief, R., 1985, "Radiation Transfer in a Solar Absorbing
tween measured and predicted particle and air temperatures Particle Laden Flow," Radiation Heal Transfer, ASME HTD-Vol. 43, eds. Ar-
can be primarily associated with uncertainty in the particle maly, B. F., and Emery, A. F., pp. 9-14.
drag coefficient and not the particle heat transfer correlation. Hruby, J. M., and Burolla, V. P . , 1984, "Solid Particle Receiver Ex-
periments: Velocity Measurements," Sandia National Laboratories,
SAND84-8238.
Conclusions Hruby, J. M., and Steele, B. R., 1986a, " A Solid Particle Central Receiver
for Solar Energy," Chemical Engineering Progress, Vol. 82, No. 2, February,
An experiment has been performed to determine the flow 1986.
characteristics and the convective heat transfer in a curtain of Hruby, J. M., 1986b, " A Technical Feasibility Study of a Solid Particle Solar
Central Receiver for High Temperature Applications," Sandia National
freely falling particles. Measurements of particle velocity in Laboratories, SAND86-8211.
heated and unheated flow at two mass flow rates have shown Incropera, F. P., and DeWitt, D. P., 1981, Fundamentals of Heat Transfer,
that the particles in a curtain fall faster than an isolated parti- Wiley, New York, p. 184.

180/Vol.110,JUNE1988 Transactions of the ASME

 Kassoy, D. R., Adamson, T. C , and Messiter, A. F., 1966, "Compressible Evaluation of the 10 MWe Solar Thermal Central Receiver Pilot Plant," Sandia
Low Reynolds Number Flow Around a Sphere," The Physics of Fluids, Vol. 9, National Laboratories, SAND85-8015.
No. 4, pp. 671-681. Spalding, D. B., 1972, "A Novel Finite-Difference Formulation for Differen-
Launder, B. E., and Spalding, D. B., 1972, Mathematical Models of Tur- tial Expressions Involving Both First and Second Derivatives," Int. J. of
bulence, Academic Press. Numerical Methods Eng., Vol. 4, pp. 551-559.
Martin, J., and Vitko, J., 1982, "ASCUAS: A Solar Central Receiver Utiliz- Tsuji, Y., Morikwa, Y., and Terashima, K., 1982, "Fluid Dynamic Interac-
ing a Solid Thermal Carrier," Sandia National Laboratories, SAND82-8203, tion Between Two Spheres," Int. Journal of Multiphase Flow, Vol. 8, No. 1,
Patankar, S. V., 1980, Numerical Heat Tranfer and Fluid Flow, McGraw- pp. 71-82.
Hill. Wallis, G. B., 1969, One-Dimensional Two-Phase Flow, McGraw-Hill.
Radosevich, L. G., 1985, "Final Report on the Experimental Test and White, F. M., 1974, Viscous Fluid Flow, McGraw-Hill.

Downloaded from http://asmedigitalcollection.asme.org/fluidsengineering/article-pdf/110/2/172/5606501/172_1.pdf by Indian Institute of Technology Bombay user on 01 June 2024
Call For Papers
Symposium on Scaling of Fluids Engineering Models
1989 ASME Fluids Engineering Division Spring Conference
University of California at San Diego, La Jolla, California
July 9-12, 1989

The Fluid Mechanics Committee, the Fluid Transients Committee, and the Coordinating Group for Fluid Measurements
of the ASME Fluids Engineering Division are jointly organizing a Symposium on Scaling of Fluids Engineering Models for
the ASME Fluids Engineering Spring Conference in July, 1989.

SCOPE
The objective of this symposium is to provide a technical forum for the presentation and discussion of the recent advances
in the area of fluids engineering modeling. Typical topics include, but are not limited to:
• Scaling of Closed Conduit Flow Models
• Scaling of Free—Surface Flow Models
• Scaling of Fluid Machinery Models
• Scaling of Unsteady Flow and Transient Flow Models
9
Scaling of Atmospheric Dispersion and Wind Engineering Models
• Scaling of Multi-phase Flow Models
• Scaling of Movable Bed Models
• Scale Effects in Model Tests
• Extrapolation of Model Data

SELECTION OF PAPERS
Prospective authors should submit a 500 word abstract by July 1, 1988. Authors will be notified of acceptance by August
1, 1988. Draft copies of the full papers must be submitted by October 1, 1988 for formal review by the organizing committee.
Authors of accepted papers will be notified by January 15, 1989 and receive a set of mats for typing the final paper draft.
Papers are limited to a maximum length of six mat pages, including figures, and must conform to the editorial standards for
FED publications. The final paper draft on mats must be submitted by March 1, 1989. Accepted papers will be published in a
bound volume available at the Symposium.

ORGANIZERS
Abstracts may be submitted to any of the organizing committee members:
Dr. M. Padmanabhan Dr. T. B. Morrow Dr. S. Vigander
Alden Research Laboratory Southwest Research Tennessee Valley Authority
30 Shrewsbury Street Institute Engineering Laboratory
Holden, MA 01520 P.O. Drawer 28510 P.O. Drawer E
(617) 829-4323 San Antonio, TX 78284 Norris, TN 37828
(512) 522-2355 (615) 632-1946

Journal of Fluids Engineering JUNE1988, Vol. 110/181