Group Title: 7th International Conference on Multiphase Flow - ICMF 2010 Proceedings
Title: P3.59 - Image Based Measurement of Particle Phase Reynolds Stresses in a Laboratory Scale Circulating Fluidized Bed
ALL VOLUMES CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00102023/00543
 Material Information
Title: P3.59 - Image Based Measurement of Particle Phase Reynolds Stresses in a Laboratory Scale Circulating Fluidized Bed Droplet Flows
Series Title: 7th International Conference on Multiphase Flow - ICMF 2010 Proceedings
Physical Description: Conference Papers
Creator: Peltola, J.
Kallio, S.
Honkanen, M.
Saarenrinne, P.
Publisher: International Conference on Multiphase Flow (ICMF)
Publication Date: June 4, 2010
 Subjects
Subject: circulating fluidized bed
time-averaged modeling
Reynolds stresses
shadowgraphy
particle image velocimetry
 Notes
Abstract: To fully account for the complicated flow patterns of the dense gas-solid suspension in a circulating fluidized bed (CFB) fluid dynamic simulations of CFBs are typically conducted in the transient mode. The computational mesh in these simulations must be reasonably fine, which in the case of large industrial processes is not feasible. A better approach for large processes seems to be time-averaged modeling facilitating steady-state simulation of fluidization. Time-averaging of the transport equations creates fluctuation terms that require modeling. Of these terms the Reynolds stress terms usually are among the largest terms to be modeled. In the present paper, experiments were carried out at a laboratory scale pseudo-2D CFB at Åbo Akademi University, Finland. Image based measurement methods were used to simultaneously determine particle velocities and volume fraction. Backlight illuminated shadowgraphy was the imaging method of choice. From the images, particle velocities were measured with Particle Image Velocimetry (PIV). The sub-pixel accuracy of the PIV-algorithm with the large particle images was verified with synthetic images. The local instantaneous particle volume fraction was determined from the gray-scale value of the shadowgraphy images with a correlation suggested by Grasa and Abanades (2001). The simultaneous measurement of particle velocities and volume fraction allows calculation of particle phase Reynolds stresses and volume fraction weighted average velocities. The results are presented at different particle sizes and fluidization velocities in the present paper.
General Note: The International Conference on Multiphase Flow (ICMF) first was held in Tsukuba, Japan in 1991 and the second ICMF took place in Kyoto, Japan in 1995. During this conference, it was decided to establish an International Governing Board which oversees the major aspects of the conference and makes decisions about future conference locations. Due to the great importance of the field, it was furthermore decided to hold the conference every three years successively in Asia including Australia, Europe including Africa, Russia and the Near East and America. Hence, ICMF 1998 was held in Lyon, France, ICMF 2001 in New Orleans, USA, ICMF 2004 in Yokohama, Japan, and ICMF 2007 in Leipzig, Germany. ICMF-2010 is devoted to all aspects of Multiphase Flow. Researchers from all over the world gathered in order to introduce their recent advances in the field and thereby promote the exchange of new ideas, results and techniques. The conference is a key event in Multiphase Flow and supports the advancement of science in this very important field. The major research topics relevant for the conference are as follows: Bio-Fluid Dynamics; Boiling; Bubbly Flows; Cavitation; Colloidal and Suspension Dynamics; Collision, Agglomeration and Breakup; Computational Techniques for Multiphase Flows; Droplet Flows; Environmental and Geophysical Flows; Experimental Methods for Multiphase Flows; Fluidized and Circulating Fluidized Beds; Fluid Structure Interactions; Granular Media; Industrial Applications; Instabilities; Interfacial Flows; Micro and Nano-Scale Multiphase Flows; Microgravity in Two-Phase Flow; Multiphase Flows with Heat and Mass Transfer; Non-Newtonian Multiphase Flows; Particle-Laden Flows; Particle, Bubble and Drop Dynamics; Reactive Multiphase Flows
 Record Information
Bibliographic ID: UF00102023
Volume ID: VID00543
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: P359-Peltola-ICMF2010.pdf

Full Text

7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


Image Based Measurement of Particle Phase Reynolds Stresses in a Laboratory Scale
Circulating Fluidized Bed


J. Peltola**, S. Kalliot, M. Honkanen* and P. Saarenrinne*

Department of Energy and Process Engineering, Tampere University of Technology, Tampere, Finland
t VTT Technical Research Centre of Finland, Espoo, Finland

Current affiliation
juho.peltola vtt.fi and sirpa.kallio@vtt.fi



Keywords: circulating fluidized bed, time-averaged modeling, Reynolds stresses, shadowgraphy, particle image velocimetry





Abstract

To fully account for the complicated flow patterns of the dense gas-solid suspension in a circulating fluidized bed
(CFB) fluid dynamic simulations of CFBs are typically conducted in the transient mode. The computational mesh in these
simulations must be reasonably fine, which in the case of large industrial processes is not feasible. A better approach for large
processes seems to be time-averaged modeling facilitating steady-state simulation of fluidization. Time-averaging of the
transport equations creates fluctuation terms that require modeling. Of these terms the Reynolds stress terms usually are among
the largest terms to be modeled.
In the present paper, experiments were carried out at a laboratory scale pseudo-2D CFB at Abo Akademi University,
Finland. Image based measurement methods were used to simultaneously determine particle velocities and volume fraction.
Backlight illuminated shadowgraphy was the imaging method of choice. From the images, particle velocities were measured
with Particle Image Velocimetry (PIV). The sub-pixel accuracy of the PIV-algorithm with the large particle images was verified
with synthetic images. The local instantaneous particle volume fraction was determined from the gray-scale value of the
shadowgraphy images with a correlation suggested by Grasa and Abanades (2001). The simultaneous measurement of particle
velocities and volume fraction allows calculation of particle phase Reynolds stresses and volume fraction weighted average
velocities. The results are presented at different particle sizes and fluidization velocities in the present paper.


Introduction

Even if chemical reactions and heat transfer are ignored, the
physics of the multiphase flow in fluidized beds is very
complex. In current practice, fluidized beds are always
simulated with time dependent, transient simulations with
reasonably fine computational meshes to capture the
complicated physical phenomena. In industrial applications
the fluidized beds are large and their size, together with long
averaging times necessitated by the unsteady nature of the
flow leads to unfeasibly large computational costs. These
computational costs have created a demand for an efficient
simulation method that could be used as a design tool in
industry. The most attractive approach seems to be
time-averaged modeling facilitating steady-state simulation of
fluidization.
Time-averaging of the transport equations creates fluctuation
terms that require modeling. In the averaging approach, the
averaged velocities are typically weighted with volume
fraction. By using these Favre-averages of velocities it is
possible to avoid fluctuation terms in the continuity equation.
In the time-averaged solid phase momentum equation the
largest terms to be modeled are the drag term, the Reynolds


stresses and the term resulting from the correlation between
pressure and voidage fluctuations. Reynolds stresses
dominate momentum transfer in the lateral direction and
even in the vertical direction they are significant. (Kallio et
al., 2008.) The development of the time-averaged closure
models requires extensive transient simulations and these
have to be validated with experimental results. For this
purpose a laboratory scale CFB was constructed at Abo
Akademi University in Turku, Finland and an extensive
measurement campaign was carried out.
In the present paper results of image-based particle velocity
and volume fraction measurements are presented with two
different particles sizes and bed masses. The bed masses
and particle sizes for measurement sets Ml and M2 are
listed in Table 1. For the larger particles the results are
presented at two different fluidization velocities.


Table 1. Bed masses used in the measurements.
Particle Sauter mean Bed mass Fluidization
diameter, dSMD [pm] [kgl velocity [m/s]
Ml 442 3.8 3.25 and 3.75
M2 255 1.9 2.75









Nomenclature


dsMD
I
R
u
Uft


Sauter mean diameter
image grey-scale value
Reynolds stress (m2s-2)
velocity (ms-1)
fluidization velocity (ms-1)


Greek letters
a volume fraction
p density (kgm-3)

Subsripts
max maximum
min minimum
ref reference
s solids


Experimental Device


The design and construction of the laboratory scale CFB is
presented in detail in Matias Gulden's Master of Science
thesis (2" '") and the device has also been described by Kallio
et al. (2'" "'. The objective was to create a reasonably large
but relatively two-dimensional fluidized bed: reasonably large
to facilitate the scaling of the results to industrial scale and
two-dimensional to allow quick simulation and image based
velocity and volume fraction measurements.
The riser section of the CFB is 3.0 m tall and 0.4 m wide. The
distance between the front and back walls of the riser is 0.015
m. The wall material is clear, 10 mm thick, hardened
polycarbonate. The fluidization air is injected from eight
equally spaced 0.013 x 0.013 m injectors at the bottom of the
bed. For adequate pressure loss in the injectors, 4 mm
diameter restrictors are placed below the injector nozzles. The
device has been designed for fluidization velocities of up to
4.0 m/s. Instead of a cyclone the CFB has a simple separator
to separate the particles from the gas outflow. A schematic
and pictures of the CFB are shown in Figure 1.
To eliminate static electricity, water was injected into the
fluidization air so that a relative humidity of 40-50% was
maintained at the gas outlet. To further reduce the static
electricity a dose of 0.08%mass of Larostat 519 (BASF SE)
antistatic powder was mixed with the particles.
The bed material is approximately spherical glass beads.
Particles were sieved to obtain desired particle size and the
diameter distribution was measured from an external sample
using shadowgraphy (Peltola, 2009). Density of the material
is 2480 kgm- and the maximum solids volume fraction was
measured to be 0.625 for the larger particles (Ml) and 0.627
for the smaller ones (M2).


Measurement Methods and Setup

The easy optical access provided by the laboratory scale CFB
enables the use of image based measurement methods.
High-speed imaging gives an excellent visualization of the
flow structures of the particle phase and their interaction.
Furthermore, if the recorded image frames have a short
enough time delay between them, it is possible to calculate a
two dimensional displacement field for the solid particles


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

using Particle Image Velocimetry, PIV As the time delay
between the frames is known, the velocity is easily
calculated from the displacement. PIV is a commonly used
method in flow measurements. Usually in PIV a sideways
laser light sheet is used to illuminate the particles in the
flow. In case of the CFB this is impossible because the
dense falling particle layers on the side walls would prevent
optical access. In the pseudo-2D geometry of the
experimental CFB backlight illumination provides a short
enough penetration depth for the light to allow velocity
measurement even at high solids volume fractions.
With backlight illuminated shadowgraphy imaging the local
solids volume fraction can also be estimated by correlating
the recorded light intensity with the volume fraction. This
allows simultaneous measurement of particle velocities and
volume fraction, which is a requirement to measure volume
fraction weighted i.e. Favre-averaged mean velocities and
Reynolds stresses for the particle phase.


Figure 1. A schematic and pictures of the experimental
device. The solids return and loop seal can be seen in the
rightmost picture. (Picture: Matias Gulden, 2008)


Imaging setup

The imaging was carried out with an ImperX Lynx 2M30
CCD-camera with 1600x1200 pixel resolution. A schematic
of the measurement setup is shown in Figure 2. It proved
vital for the measurements to have a Depth-Of-Field (DOF,
focal depth) thick enough to cover the whole depth of the
riser to prevent the measurement signal from being blurred
by unfocused particles (Peltola, 2009). With the desired
measurement window a small aperture (f/16) has to be used
to achieve an adequate DOF. A Sigma 105mm f/2.8 lens
was used on the camera. At this focal length the projection
error across the riser depth is 3% (Peltola, 2009).
Illumination was provided by a pulsed CAVILUX Smart
690 nm diode-laser (Cavitar Ltd., Finland). Other visible
light wavelengths were filtered out with an optical filter on
the camera lens.
A sample of a measurement image frame recorded with this
setup is shown in Figure 3. In the measurements
double-frame images with a time delay of 300 us were
recorded at 5 Hz. In the Ml measurement set the image
scale in the middle plane of the riser was 0.025821
mm/pixel. In the M2 measurement set a smaller






7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


measurement window was used because of the smaller
particle size and the image scale was 0.020855 mm/pixel.


I s,ref
log---
a, = a,Tref t-
logI
s,ref


Figure 2. Measurement setup.


Figure 3. A sample measurement image frame from the CFB
riser.


Particle Volume Fraction Estimate


When the particles in the riser are illuminated from behind
and the light penetrating the suspension is recorded with a
digital camera, it is possible to estimate the volume fraction
of the particles from the recorded light intensity. Grasa and
Abanades (2001) compared several correlation functions and
found out that a logarithmic function as presented in Equation
1 gives a good quality estimate, without the need for any
fitting constants. In the Equation 1 I is the local mean
recorded light intensity, Imax and Imin are the light
intensities obtained from calibration reference images and C
is the concentration of the particles.


I
log -
C=
log m
,mm


When this equation is applied to the backlight illuminated
CFB images to determine the volume fraction of the particles
it is written as Equation 2, where as,ref is the volume
fraction of the particles at the minimum intensity reference,
as,ref and las,o is the light intensity corresponding with the
zero particle volume fraction.


Figure 4 shows the particle volume fraction field calculated
from the particle image in Figure 3 by applying the
Equation 3.2. The local average intensity was calculated
from a 128 x 128 pixel sample.


02
018
016
10
014
15 012

20 01
008
25
006
30 004
002
35
5 10 15 20 25 30 35 40 45 50
Figure 4. Particle volume fraction calculated from the
sample image shown in Figure 3 with Equation 2.

A validation and calibration study was conducted by
recording the light penetration through packed layers of the
glass beads with varying thickness. It was found out than
the Equation 2 overestimates the solids volume fraction if
images of the empty riser are used as the low volume
fraction reference value, as can be seen from Figure 5. The
error was mostly eliminated by defining separate reference
values for the low volume fractions and for the high volume
fractions.








04-- I - -
3 AA
S




135704
03 -

02 j Particle layer thickness
SGrey-scale, single correlation, center
0 1 a a A Grey-scale, dilute and dense correlations center
SGrey-scale, single correlation, off-center
o Grey-scale, dilute and dense correlations, off-center
0 01 02 03 04 05 06 07 08 09
Volume fraction based on particle layer thickness
Figure 5. Calibration and validation results of the image
grey-scale value and particle volume fraction correlations.

For the low solids volume fraction correlation the reference
values were set based on an empty riser and a 1 mm thick
packed layer of particles, which correspond to a solids
volume fraction of 0.042 in the CFB riser. For the high
solids volume fraction correlation the low volume fraction
reference intensity was chosen based on the calibration
results with 1- 15 mm layer thicknesses. The differences
between the center and off-center values in Figure 5 are
caused by uneven particle layers in the calibration








experiment, spotlight nature of the illumination and lens
vignetting.
The problems with this method are: the generation of an
adequately powerful, evenly distributed backlight, the
determination of the correct reference intensities with
corresponding volume fractions and the inability to resolve
volume fractions above the lowest void fraction where the
light penetrating the suspension can be reliably detected by
the camera. With the setup used in the measurements, the
highest measurable volume fraction is very close to the
packing limit and therefore the results are quite accurate.
However, these limitations should still be kept in mind when
interpreting the results.
The major benefit of this method is that the volume fraction
fields are recorded simultaneously with the velocity fields,
which is vital for the calculation of volume fraction weighted
averages, i.e. Favre-averages, and Reynolds stresses in a
multiphase flow. It would be beneficial to combine the
grey-scale VOF estimate with another more accurate -
method to measure the mean volume fraction. The non-time
resolved mean VOF values could then be used to calibrate
and validate the grey-scale estimate in the actual experimental
device, and the VOF values given by the grey-scale estimate
could be used as a weighting factor in the calculation of the
derived quantities.
The volume fractions presented in the results were calculated
from the mean grey-scale value of a 128x128 pixel
interrogation area with Equation 2, using separate correlations
for dilute and dense regions. Constant values were used as
references. With the constant reference values the effect of the
spotlight nature the illumination and lens vignetting can be
seen in the measured volume fraction profiles (Figures 8, 9
and 16) of the results section. The procedure gives higher
volume fraction values on the edges of the image frame, but
the error is reasonably small.


Particle Velocity Measurement

When PIV is applied to the measurement of fluid flow,
images of a flow seeded with small particles are recorded
with a short time delay between individual frames. The
images are then divided into smaller interrogation areas, and
the interrogation area intensity fields are cross-correlated
between the consequent frames. The displacement of the
particles can be calculated from the displacement of the
correlation peak. In fluid velocity measurements small
seeding particles, with a small Stokes number, that
consistently follow the fluid flow are used.
In gas-solid flow where the solid particles are visible to the
camera, their velocity can be measured similarly to the
seeding particles used in traditional PIV measurements. In
this case the measured velocities don't represent the fluid
velocity, but the local expected, most probable velocity of the
particle phase in the measurement volume.
The particle velocities were determined with commercial PIV
software (LaVision DaVis 7.2, LaVision Gmbh, Gottingen,
Germany) and a multi pass correlation was used because of
the unpredictable nature of the flow. At first, a single pass
with a 256x256 pixel interrogation area was carried out and
then two passes with a 64x64 pixel interrogation area. 50%
overlap between the interrogation areas was used in all three
passes. Before the correlation the images were locally


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

inverted (scale length 30 pixels) and normalized (scale
length 40 pixels) using the DaVis built-in image
pre-processor. Because of the low noise level of the camera
the correlation produced only very few spurious vectors and,
as such, further postprocessing of the measured velocity
vectors was deemed unnecessary.
The accuracy of the DaVis peak detection algorithm with
the large particle images (10-30 pixels) was verified with
synthetic images. Single particle intensity profiles were
isolated from the actual measurement images and used to
create synthetic image pairs with random particle position,
displacement and desired particle count, mean displacement
and displacement standard deviation. The error in measured
mean values was less than 0.1 pixels when the standard
displacement of the random component of the particle
motion was less than 4 pixels. The accuracy was mostly
unaffected affected by large solids volume fractions.
(Peltola, 2009.)


Results and Discussion

All the measurements were carried out at three different
heights: 0.40, 0.80 and 1.20 m. At each height ten separate
image sets spread over the whole width of the riser were
recorded. The measurement points are visualized in Figure
7. The Figure 7 also includes instantaneous images of the
particle flow in the measurement regions with both bed
masses. They show a difference in the scale of the flow
structures between the two particle sizes: smaller particles
produce smaller structures. With the larger bed mass and
particles there are large, dense particle cluster on both side
walls up to 0.7-1.0 m height, but with the smaller bed mass
and particles these cluster are almost completely absent.
There is only a relatively steady, dense and narrow
downwards particle stream on the walls.

Ml M2






-Ul








Figure 7. Measurement points and visualization images of
the those regions from measurement sets Ml and M2. The
same color codes are used in all the results Figures for each
measurement height: red for 0.40 m, green for 0.80 m and
blue for 1.20 m.

When the recorded images are analyzed with the methods
described in the previous sections, a large database of
instantaneous particle volume fraction and velocity fields
are obtained. From these it is possible to calculate









volume-fraction-weighted average i.e. Favre-average
velocities, Equation 3, and Reynolds stresses, Equation 4. In
these equations s,a,,li is the mean particle volume fraction
including only those volume fraction values that correspond
with a valid velocity measurement points. The Reynolds
stresses are normalized with the measured mean volume
fraction to minimize possible effects of the inaccuracies in the
volume fraction measurement method.


Nvalid
S1 a s()uivaial) xy
u = N- S a = x,y
1valid 1 s,vatiLd





R j s,va idvatlid )U )
I=o
Mds,vaidati(l),tj ()], ij = xy



M1: 442 pm particles, bed mass 3.8 kg


In the first measurement set, Ml, a large 3.8 kg bed mass and
particles with a Sauter mean diameter, dsm, of 442 pm were
used. The number and mass distribution of the particles is
shown in Figure 6. The particle diameter is defined as the
diameter of a circle with surface area equal to the projected
area of the particle.
Measurements were carried out on three different riser heights
and all the measurements were carried out with two different
fluidization velocities: 3.25 m/s and 3.75 m/s.
In all measurements of the Ml set, a sampling period of 200 s
was used and for each measurement set at least 1000
double-frame images were recorded at imaging frequency of
5 Hz. At this frequency the measurements have to be
considered as discrete samples. The parameters of the
measurement are summarized in Table 2.

15 -
Number distribution i
Mass distribution


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

Table 2. Parameters of the measurement set Ml
Operational parameters:
Fluidization
Fluidization 3.25 m/s and 3.75 m/s
Velocity
Bed mass 3.8 kg
Bed mass in the
riser section 3.2 kg (3.25 m/s) and 3.1 kg (3.75 m/s)
riser section
dsMD 442 Im

Measurement parameters:
Image resolution: 1600x1200 pixels
Image scale: 0.025821 mm/pixel
Measurement
Measurement 41.3x31.0 mm
window:
Measurement Vertical: 0.40, 0.80 andl.20 m
locations Horizontal: 0.02, 0.06, 0.10, 0.14, 0.18,
0.22, 0.26, 0.30, 0.34 and 0.38 m
Time delay: 300 ps
Imaging frequency: 5 Hz
Sampling period 200 s
Images: 60 x 1000 = 60000 double-frame
images



In the CFB the densely packed wall regions at the bottom
funnel the fluidization air up the middle of the riser as a
high speed jet that rapidly accelerates the particles caught in
the flow. This is reflected in the mean particle velocity and
volume fraction profiles in Figures 8 and 9. Highest particle
vertical mean velocities are found in the middle of the riser
at heights of 0.40 (Uf = 3.25 m/s: 2.00 m/s) and 0.80 m (Ufl
= 3.75 m/s: 2.49 m/s). With the higher fluidization velocity
the acceleration zone continues further up and particle
mean volume fractions are lower. The expected dense wall
layers and dilute middle portion are clearly visible in the
mean particle volume fraction profiles A more surprising
result is that the peak volume fractions at the upper
measurement heights are not right at the wall but slightly
inwards. This is a valid result and it can also be seen in
visualization videos. A possible cause for this behaviour is
minute imperfections in the walls of the experimental
device.


0 01 02 03 04


Diameter [mm]
Figure 6. Particle number and mass distribution by diameter
of measurement set Ml was measured from an external
sample with shadowgraphy.


0 01 02 03 04
Horizontal position [m]


0 0 01 2 03 04




H 01 02 03 04
Horizontal position [m]


Figure 8. Measured particle vertical velocity and volume
fraction profiles. Fluidization velocity is 3.25 m/s.


2







7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


2

04



06
D 2

> C-
0 01 02 03 04 0 01 02 03 04





-2
0 01 02 03 04 0 1 02 03 04 4
Horizontal position [m] Horizonta position [m]

Figure 9. Measured particle vertical velocity and volume
fraction profiles. Fluidization velocity is 3.75 m/s.


There is a consistent difference between the Reynolds
averaged velocities in Figures 8 (Ufz = 3.25 m/s) and 9

(Ufz = 3.75 m/s) and the volume-fraction-weighted Favre
averaged velocities in Figure 10. The Favre-averaging results
in lower particle velocities, especially in the dilute middle
portion of the riser. In the dense regions on the walls the
differences are smaller. This is expected behaviour as the
dense particle clusters are weighted more in the averaging
procedure. The effects are very similar at both fluidization
velocities.


U, = 3.25 m/s
2












w 0 01 02 03 04



2


0 01 02 03 04
Honzontal position [m]


Ul = 325 m/s


U, = 3.75 m


S2/ 1 02 03 04


0 01 02 03 04
Horizontal position [m]


0 01 02 03 04
Horizontal position [m]


Figure 11. Reynolds stress vertical component calculated
with Equation 4. Fluidization velocities are 3.25 m/s and
3.75 m/s.


, = 3.25 m/s


E 0 01 02 03 04


Ul = 3.75 m/s
2 02 03 04




0 1 02 o3 04


H2
0 01 02 03 04
Horizontal position [m]


0 01 02 03 04






0 01 02 03 04
Horizontal position [m]


U = 3.75 m/s






0 01 02 03 04






0 01 02 03 04






0 01 02 03 04
Horizontal position [m]


Figure 12. Reynolds stress horizontal component
calculated with Equation 4. Fluidization velocities are 3.25
m/s and 3.75 m/s.


U = 3.25 m/s


U = 3.75 m/


Figure 10. Favre-averaged vertical particle velocities,
fluidization velocities 3.25 m/s and 3.75 m/s


The Reynolds stresses, as defined by Equation 4, are another
term that is important for the time-averaged modelling and
requires simultaneous velocity and volume fraction
measurements. The results for these in Figures 11 (Ryy), 12
(R,) and 13 (Rxy) demonstrate the strongly anisotropic nature
of the Reynolds stresses. The vertical component is larger
than the horizontal and cross components by an order of
magnitude. The vertical and cross components are
comparable in magnitude. The general behaviour is similar at
both fluidization velocities, but the stresses are larger at the
higher fluidization velocity, as expected. The highest values
for the vertical component are recorded at heights 0.40 m (Uf1
= 3.25 m/s) and 0.80 m (Up = 3.75 m/s), while the horizontal
and cross components consistently weaken while moving
upwards.


0 01 02 03 04






0 01 02 03 04
Horizontal position [m]


0 01 02 03 04





2
0 01 o 02 03 04
Horizontal position [m]


Figure 13. Reynolds stress cross component calculated
with Equation 4. Fluidization velocities are 3.25 m/s and
3 75 m/s


)4

2 ^_________}


S 01 02 03 04 E 0 01 02 03 04
15 1 5

SE0




00 01 02 03 04 00 01 02 03 04
1 1

5



tr 15------- Ir


01 02 03 04


)1 02 03 04


02


S-02









M2: 255 pm particles, bed mass 1.9 kg

For the second measurement set smaller particles closer in
size to those in industrial CFBs were used with a lower bed
mass. The size distribution of the particles is shown in Figure
14 as it was measured from an external sample. The Sauter
mean diameter of the particles is 255 tm. A summary of the
parameters of the second measurement set, M2, are listed in
Table 3. Correspondingly with the lower terminal velocity of
the smaller particles the fluidization velocity was reduced to
2.75 m/s. It is important to note that, while compared to the
measurement Ml the bed mass was halved to 1.9 kg, the
portion of the bed mass in the riser section of the CFB is only
approximately one third of the mass in Ml.
In the measurements a smaller, 33.0x25.4 mm, measurement
window was used because of the smaller particle size. Riser
cross-sections were covered with ten measurement locations
and measurements were carried out at the same heights as in
the previous measurement set. Compared to the Ml the
sampling period was lengthened to 300s to achieve more
stable Reynolds stress values and Favre averages.


Table 3. Parameters of the measurement set M2
Operational parameters:
Fluidization 2
2.75 m/s
Velocity
Bed mass 1.9 kg
Bed mass in the
riser section 1.0 kg
dsm _255 Im

Measurement parameters:
Image resolution: 1600x1200 pixels
Image scale: 0.020855 mm/pixel
Measurement 33. 0
33.4x25.0 mm
window:
Measurement Vertical: 0.40, 0.80, andl.20 m
locations Horizontal: 0.016, 0.06, 0.10, 0.14,
0.18, 0.22, 0.26, 0.30, 0.34 and
0.384 m
Time delay: 300 gs
Imaging frequency: 5 Hz
Sampling period 300 s
Images: 30 x 1500 = 45000 double-frame
Images


Number distnbution
1 Mass distnbution


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

With the lower bed mass and smaller particles the measured
particle mean velocity profiles, Figure 15, show clear,
qualitative differences to those in the previous measurement
set, Ml. In the lower portion of the riser the peak particle
velocities are no longer found in the middle of the riser, but
closer to the walls. The less dense particle suspension at the
bottom also means the solids return affects the
time-averaged flow and the velocity profiles are no longer
symmetric like in the measurement set Ml. The particle
velocities are generally higher on the left i.e. opposite to the
wall where solids return is located. At the lowest two
heights 0.40 and 0.80 m the highest particle rise
velocities are found approx. 70 mm from the left wall.
In the measurement set Ml the difference between the
Reynolds- and Favre-averaged velocities was larger in the
dilute regions of the flow and the same can be said for the
current measurement set, M2. However, because of the
lower bed mass, larger portions of the riser are dilute and,
as a result, the difference between the Reynolds- and
Favre-averaged velocities is more obvious. In the middle
portion of the riser the Favre-averaged rise velocities are
generally 0.3 0.5 m/s lower than the Reynolds-averaged.
In the falling and dense wall-layer the differences are
smaller.


N.----
/


/J- -I.


03 04 W O
2
S 0

03 04 0 o
1S


0 01 02 03 04
Horizontal position [m]


0 01 02 03 o0
Horizontal position [m]


Figure 15. Comparison of Reynolds- and Favre-averaged
vertical particle velocities.

The mean particle volume fraction profiles, Figure 16, of
measurement set M2 are generally very flat, with only
narrow denser wall layers. Because of the lower bed mass
the particle volume fractions obviously are lower than in
the measurement set Ml across the board. The asymmetry
caused by the solids return in the mean velocity profiles can
also be seen in the volume fraction profiles. The mean
particle volume fractions exceed 0.25 only right at the wall
below the solids return and close to the opposite wall is a
region of lower particle volume fraction that corresponds
with the higher rise velocities in Figure 15.
The measured Reynolds stresses, Figure 17, are still
strongly anisotropic with a dominant vertical component.
The magnitude of the vertical component is generally lower
(0.3 1.4 : i'- -) than in the previous measurement set, Ml.
At the lowest measurement height the vertical component
has peak values (0.8 1.4 :ri- _-) approx. 20 mm from the
walls. These peaks level off at higher measurements heights.
Close to the left wall the region of higher mean particle rise
velocity (Figure 15) coincides with increased vertical


Diameter rmml
Figure 14. Particle number and mass distribution by diameter
of measurement set M2 was measured from an external
sample with shadowgraphy.










Reynolds stress. In the Ml measurement set the horizontal
and cross components of the Reynolds stresses were similar
in magnitude, but here the cross component is much weaker.
Only exception is the right hand side of the riser at height of
0.40 m, where the effect of the solids return is clearly visible
as a strong Reynolds stress cross component.


S02 03 04


0 01 02 03 04





0 01 02 03 04
Horizontal position [m]


Figure 16. Solids
logarithmic scales.


Horizontal position [m]


mean volume fraction with linear and


1.5


E 0 0.2
I

15
1 1


o
- 0 02
0) 15
c:
oJrx-'


02
0.4 E 0



o
-
02
0.1


04 t 0

j ~ 02


0 0.2 0.4
Horizontal position [m]


S -0 2
02 04 0
E


02

-A 02
02 04 0
A I_ 0 2


0 02 04
Horizontal position [ml


).2 0.4


02 04


-0 2
0 0.2 0.4
Horizontal position [m]


Figure 17. Reynolds stresses calculated with Equation 4. The
colors for each measurement height are the same as in the
previous figures: red for 0.40 m, green for 0.80 m and blue
for 1.20 m.



Conclusions

In the measurement set Ml the particles form thick dense
regions on the walls at the bottom of the riser. These clusters
direct the fluidization air up through the middle of the riser as
a high speed jet, quickly accelerating the particles. In all
measurement in set Ml the peak particle velocities are
recorded in the middle of the riser. The Reynolds stresses are
strongly anisotropic: the vertical component is larger than the
horizontal and cross component by a order of magnitude. All
the measured profiles are very symmetric, which implies that
the solids return has little effect on the time-averaged flow


D-- 02 0


7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

because of the dense wall regions at the bottom.
The nature of the flow is quite different in measurement set
M2. The dense wall region no longer exist at the bottom
and in the lower portion of the riser the peak particle rise
velocities are recorded 50-80 mm from the walls. The
Reynolds stresses behave similarly to the measurement set
M2 with the exception of generally weaker cross
component. With the more dilute suspension the solids
return significantly affects the time-averaged flow in the
riser section creating asymmetry. This asymmetry is visible
in all the measured profiles, but most apparent it is in the
particle vertical velocity at the bottom and the cross
component of the Reynolds stresses just below the solids
return.
In both cases there is a significant difference between
Reynolds and Favre averaged velocities. In dilute regions
the Favre averaging gives lower particle rise velocities,
while in the dense wall layers the difference is small. There
are small regions where the sign of the particle vertical
velocity changes between the Reynolds and Favre averages.
All in all, the image-based measurement methods proved
highly capable in producing detailed numerical data from
the laboratory scale CFB. Most of the value of the
measurement results is in comparison with the
time-averaged and transient simulations, and for this
purpose they provide an important benchmark. The limiting
factor for the scope this kind of experimental studies is the
time consumption and computational costs of recording,
transferring and analyzing the image data.



Acknowledgements

This study was financially supported by Tekes (the Finnish
Funding Agency for Technology and Innovation), Fortum
Power and Heat Oy, Foster Wheeler Energia Oy, Metso
Power Oy, Neste Jacobs Oy, Neste Oil Oyj, and VTT
Technical Research Centre of Finland. The authors would
like to thank Mr. Alf Hermanson and Mr. Jonatan Hassel for
their assistance and for running the CFB unit at Abo
Akademi University during the experiments.



References

Grasa G. and Abanades J.C., A calibration procedure to
obtain solid concentrations from digital images of bulk
powders, Powder Technology, Vol. 114, pp. 125-128 (2001)

Gulden, M., Pilotmodell av en cirkulerande fluidiserad bad.
Master of science thesis, Abo Akademi. 65 p. (2" I")

Kallio S., Gulden M., Hermanson A. Experimental study
and CFD simulation of a 2D circulating fluidized bed,
FBC20 Proceedings of the 20th Int. Conf. on Fluidized Bed
Combustion, Xi'an, China, May 18-21, 2009

Kallio, S., Taivassalo, V, Hyppinen, T., Towards
time-averaged CFD modelling of circulating fluidized beds,
9th Int Conf. on Circulating Fluidized Beds, Hamburg, May
12-16, 2008






7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

Peltola, J., 2009. Dynamics in a circulating fluidized bed:
experimental and numerical study, M.Sc. Thesis, Tampere
University of Technology, 95 p.
http://ur.fi/URN:NBN:fi:tty-200910136897




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs