BackscatterImagingLDVsystem FluidizedBedPandey06

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Particulate Science and Technology

ISSN: 0272-6351 (Print) 1548-0046 (Online) Journal homepage: https://www.tandfonline.com/loi/upst20

Evaluation of a Backscatter Imaging LDV System


and its Application to a Pilot-Scale Circulating
Fluidized Bed

Preetanshu Pandey , Richard Turton , Paul Yue & Lawrence Shadle

To cite this article: Preetanshu Pandey , Richard Turton , Paul Yue & Lawrence Shadle
(2006) Evaluation of a Backscatter Imaging LDV System and its Application to a Pilot-
Scale Circulating Fluidized Bed, Particulate Science and Technology, 24:1, 1-22, DOI:
10.1080/02726350500212947

To link to this article: https://doi.org/10.1080/02726350500212947

Published online: 24 Feb 2007.

Submit your article to this journal

Article views: 70

View related articles

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=upst20
Particulate Science and Technology, 24: 1–22, 2006
Copyright # Taylor & Francis LLC
ISSN: 0272-6351 print/1548-0046 online
DOI: 10.1080/02726350500212947

Evaluation of a Backscatter Imaging LDV System


and Its Application to a Pilot-Scale Circulating
Fluidized Bed

PREETANSHU PANDEY
RICHARD TURTON
Department of Chemical Engineering, West Virginia University,
Morgantown, West Virginia, USA

PAUL YUE
LAWRENCE SHADLE
U.S. Department of Energy, National Energy Technology Laboratory,
Morgantown, West Virginia, USA

This study focuses on the evaluation and application of a backscatter imaging LDV
(laser Doppler velocimeter) system (BILS) built by TSI Inc. (Shoreview, Minn.).
This nonintrusive instrument has the ability to record particle images and velocity
data simultaneously. A series of validation tests was conducted prior to measuring
particle velocity in a turbulent system. A set of experiments was conducted to mea-
sure terminal velocity and particle size for five different-size cuts of near-spherical,
Nu-Pareil particles, and the results were compared with the standard drag curve,
which showed good agreement. The imaging data were used to estimate the spher-
icity (U) of the particles. The data confirmed that U was close to 1
(0.98 < U < 0.99). The system was then used to measure particle velocity in a cold
flow circulating fluidized bed at various operating conditions. The effect of operating
conditions, such as superficial gas velocity and solids circulation rate, on the solids
velocity was studied. The mean solids velocity near the wall was found to increase
with increasing superficial gas velocity and decreasing solids circulation rate. The
granular temperature was also estimated and was found to increase with increasing
distance from the wall. The granular temperature was found to lie in the range of
0.1–1.0 m2=s2 for the operating conditions studied. The granular temperature calcu-
lations also showed that the flow of particles moving downwards was more ordered
than the flow moving up.

Keywords LDV, imaging, particle velocity, particle size, sphericity, circulating


fluidized bed, granular temperature

The support for this study by the University=NETL student partnership program,
through the National Energy Technology Laboratory (NETL), is gratefully acknowledged.
The authors are thankful to Dr. Joseph Shakal from TSI Inc., Minn., for his invaluable help
with the instrument operation, and Dr. James Spenik from NETL for the high-speed pressure
drop data and turbulence intensity estimates.
Address correspondence to Richard Turton, 453 Engineering Science Building, Chemical
Engineering Department, P.O. 6102, West Virginia University, Morgantown, WV 26506,
USA. E-mail: richard.turton@mail.wvu.edu

1
2 P. Pandey et al.

Introduction
Gas-solid circulating fluidized beds (CFBs) have widespread application in a variety
of industrial processes such as coal combustion and fluidized catalytic cracking
(FCC). The study of such complex two-phase flows requires precise analysis of the
local behavior of both the gas and solid phases (Van den Moortel et al., 1998).
The characteristics of the solid phase, such as particle size and velocity, directly
affect the performance of such systems. Particle velocities are very important and
affect the heat and mass transfer, as well as the erosion, at the walls of a CFB (Zhou
et al., 1995).
Computational fluid dynamic models of gas-solids flow systems have been inac-
curate in describing flow dynamics in CFBs (Knowlton & Karri, 2001; Gidaspow,
1994). As the circulation rate is increased and the riser bed density exceeds 5% vol-
ume of solids, the solids concentration increases near the wall and the riser develops
a core-annular radial flow profile. The large concentration of solids near the wall
results in negative solids flux conditions, which can dominate the flow distribution.
Simulations must accurately capture the particle interactions in this annular region
to successfully capture the overall flow behavior. One common approach is to apply
the continuum theory (Goodman & Cowin, 1972) to flowing gas solids systems and
solve for each phase separately. The properties for the solids phase can be estimated
by analogy to the kinetic theory of gases in which the particles behave like gas mole-
cules. The constitutive law required to simulate core annular behavior in a riser is the
solids viscosity, which can be considered a function of the granular temperature
(Gidaspow, 1994). The granular temperature is a measure of the relative motion
of individual particles in the mixture. Thus, measurements of the radial dependence
of particle velocities, especially near the wall, are important to establish the shear
rate or viscous dependence of the gas-solids mixture on this granular temperature.
The techniques used in the past to measure solids velocity can be broadly classi-
fied into intrusive and nonintrusive. Some of the intrusive techniques include fiber
optic probes (Herbert et al., 1994; Horio et al., 1988; Liu et al., 2002; Zhou et al.,
1995; Zhu et al., 2001), extraction probes (Miller & Gidaspow, 1992), and capaci-
tance probes (Louge et al., 1990). Nonintrusive techniques include the use of LDV
(Wang et al., 1998; Wei et al., 1998; Yang et al., 1993), particle imaging velocimetry
(PIV) (Shi et al., 2002), and various optical techniques (Viitanen, 1993; Zheng et al.,
1992).
Zhou et al. (1995) developed a five-fiber optical probe to measure vertical velo-
cities and the fraction of rising and falling particles in a CFB riser of cross section
146  146 mm. They found that the ascending particle velocity increased in the core
region and decreased near the wall region with increasing solid circulation rate. The
magnitudes of descending-particle velocities increased with solid circulation rate, but
was almost unaffected by an increase in superficial gas velocity. Monceaux et al.
(1985) showed that a characteristic flow regime in CFB was a core-annulus configur-
ation in the dilute zone of the riser. Arastoopour and Yang (1991) used a LDV (laser
Doppler velocimeter) to obtain velocity profiles in a CFB under relatively dilute con-
ditions (solid volume fraction <3%). At the wall boundary, an instantaneous rever-
sal of flow was observed. Van den Moortel et al. (1998) used a 1-D PDPA (phase
Doppler particle analyzer) to measure velocities in the dilute zone of the riser (solid
volume fraction <1.5%) and confirmed a core-annulus flow structure in the riser.
Wei et al. (1998) used a modified LDV system to develop velocity profiles in a riser.
Evaluation of a Backscatter Imaging LDV System 3

Although the modified LDV lost its ‘‘nonintrusive’’ nature, it was able to measure
flows with higher solids fractions. Thus the flow in a CFB riser, in general, can be
characterized by a rapidly rising, relatively dilute suspension in the core that is sur-
rounded by a slow-falling, denser suspension towards the wall (Rhodes et al., 1988).
The instrument used in this study is a TSI-built backscatter imaging LDV system
(BILS). BILS incorporates and extends the particle velocity measurement ability of
the LDV with the ability to measure particle size. The combined measurement of
particle size and velocity allows the construction of the size-velocity correlation.
BILS differs from a conventional PDPA in that it requires the particle to have rough
surfaces in order to obtain non-specular scattering. Besides particle size and velocity,
BILS provides particle shape information, such as aspect ratio and circularity, and
requires only one optical window to access the flow (TSI Inc., 2000). The LDV unit
of BILS has both probes (transmitter and receiver) combined into a transceiver unit
so that optical alignment between them is automatically maintained.
In order to obtain accurate particle velocities, it is important that the flow is
unperturbed while making measurements. This can be attained easily with a LDV
system, which is a nonintrusive optical technique and measures in-situ local instan-
taneous particle velocities. It can be used to provide accurate, real-time velocity mea-
surements with high resolution and has the ability to detect flow reversals. Even
though the LDV cannot be used to obtain velocity profiles across the whole riser
for very dense flows, it can still be used to obtain velocities in the near-wall region
in a nonintrusive way. This can provide experimental verification for the boundary
condition used by various CFB modelers. In addition, most of the previous studies
were conducted on small-scale risers, which may not be an accurate representation of
an industrial CFB riser due to scale-up effects. Thus, there is still interest in obtain-
ing experimental velocity data measured nonintrusively, in a relatively large-scale
CFB riser.
The aim of this study is to evaluate the performance of this BILS instrument and
to assess its potential for measuring in-situ particle velocities and sizes in the riser
section of a pilot-scale, cold-flow, circulating fluidized bed.

Material and Methods

Measuring Technique
The LDV component uses an air-cooled Ar ion laser with 750 mW peak output
power as the light source. The laser beam passes through a fiber drive where a Bragg
cell is used to split it into two equal intensity beams. The green (k ¼ 514 nm) and the
blue (k ¼ 488 nm) portions of each beam are then separated, producing two pairs of
color beams. These beams are steered via separated optical fibers to the transceiver.
The Bragg cell also shifts the frequency of one of the beams, resulting in a moving
pattern of fringes. Thus, the system is able to distinguish between particles moving
in the up and down directions. This distinction plays a critical role in turbulent flow
measurements. Each pair of color beams emerges from the transceiver and intersects
to form interference fringes in the measurement volume. Doppler light pulses are
generated when a particle crosses these fringes. These are captured by an optical fiber
residing inside the transceiver and transmitted to real-time fast Fourier transform
(FFT)-based signal analyzers (RSAs) where they are amplified by photomultipliers
and processed (Pandey et al., 2004). The velocities of the particles derived from
4 P. Pandey et al.

Figure 1. Components of BILS.

the Doppler frequencies are displayed on the computer by DataView NT software.


The components of the system are shown in Figure 1.
The imaging system consists of three main components: a backscatter imaging
receiver (containing CCD camera and focusing lens system), a backscatter imaging
transmitter (containing diode laser and focusing lens system), and a backscatter ima-
ging controller. The backscatter probe volume, a region well under 1 mm2 in size, is
aligned with the LDV probe volume. The pulsed diode laser has a wavelength of
905 nm and a pulse width of 15 ns. Its maximum repetition rate is 5 kHz. The focal
length of the front lens is 750 mm. The area scan CCD camera has a speed of 1000
frames per second and a resolution of 256  256 pixels. The pixel size is 10  10 mm.
The lenses provide approximately 1:1 imaging.

Operating Principles
LDV Component
When a particle crosses the two orthogonal sets of interference fringes set up by the
two pairs of color beams, Doppler bursts are generated. The frequencies of these
bursts are proportional to the particle velocity components perpendicular to the
plane of the fringes. Each component of velocity is calculated using the following
equation:
V ¼ dFD ð1Þ
where FD is the Doppler frequency, V is the particle velocity, and d is the fringe
spacing given by:
k
d¼ ð2Þ
2 sinða=2Þ
where k is the wavelength of the laser beam and a is the beam-crossing angle, which
for this instrument is approximately 6.
Evaluation of a Backscatter Imaging LDV System 5

Imaging Component
The pulsed diode laser (Power Technology Inc., Model ML-100H15) from the trans-
mitting probe illuminates the particle. The CCD camera (Dalsa Inc., Model CA-D6-
0256) that resides in the receiver probe captures images of the particle through a
750 mm focal length front lens. A long pass filter is used to block any reflections
from the Ar-ion beam and other light sources. Particle imaging is initiated by the
RSA processor, which sends a trigger to the backscatter imaging controller. Gener-
ally, several hundreds of images are grabbed in a single run, and both the LDV and
imaging data are saved on the hard drive. Statistical analysis of the images by post-
processing is accomplished using the BackscaPP software, supplied by TSI Inc. The
particle images aid in the calculation of particle circularity, aspect ratio, and other
shape features. Since the information is based on 2-D images, this works well for par-
ticles with high sphericity but may not be accurate for highly nonspherical particles.

Velocity Validation
The equipment used to validate the velocity measurements consisted of a stiff wire
attached to the rim of a rotating disc, perpendicular to the plane of rotation, as
shown in Figure 2. The rate of rotation of the disc was measured using a digistrobe
(Power Instruments Inc., Skokie, Ill.). The tangential velocity of the wire, v, was cal-
culated by knowing the radial distance of the wire, r, from the center and the
rotation rate, x, (v ¼ r x). Each time the wire passed through the measurement vol-
ume, BILS measured and recorded the wire’s tangential velocity. Two wires with
diameter 0.5 mm and 1 mm were used. The experimental and the calculated values
were compared and a good match was observed, as shown in Figure 3. It was also
observed that wire diameter did not have any significant influence on the readings.
This is also illustrated in Figure 3.

Imaging Calibration and BILS Alignment


For imaging calibration, a calibration scale (1 division ¼ 0.1 mm) was used to pro-
vide the conversion factor from pixels to length in the BackscaPP imaging software.
The alignment of the LDV measurement volume and the imaging volume requires

Figure 2. Rotating disc with a wire attached for instrument velocity data validation.
6 P. Pandey et al.

Figure 3. Random runs to check repeatability of data and effect of wire diameter (particle
size) on velocity readings.

great accuracy to ensure that the camera obtains simultaneous images of the particle
whose velocity the LDV records. This was achieved by observing the velocity and
image of a thin stiff wire passing through the measurement volume. Simultaneous
data on both the LDV and imaging channels (sharp images) ensured that the system
was aligned.

Particle Terminal Velocity and Sizing Tests


After calibration, the instrument’s performance was evaluated by recording terminal
velocity and particle size of Nu-PareilTM sucrose particles (qs ¼ 1.2 g=cc). The dia-
meter of particles used in this study ranged from 500 to 1200 mm. These particles
were chosen because they were close to spherical and had rough surfaces, making
them suitable for imaging. The near-spherical shape of these particles allowed com-
parison of the experimental results with predictions from the standard drag curve.
The particles were placed on a metal screen to which a pneumatic vibrator was
attached. The frequency of vibration was controlled by adjusting the airflow from
a 20–30 psi air supply. Minimal particle-to-particle interaction was ensured by exam-
ining the images of the particles and observing that only single, isolated particles
passed through the measurement volume. A narrow size cut of particles was
obtained by sieve analysis.
Correlations between terminal velocities of particles and particle diameters,
developed by Haider and Levenspiel (1989), were used to compare the results
obtained from BILS. They used two dimensionless numbers, u*, the dimensionless
terminal velocity, and d*, the dimensionless particle diameter. These parameters
are defined by Equations (3) and (4).
" #1=3
q2f
u ¼ ut ð3Þ
glðqs  qf Þ
 1=3
gqf ðqs  qf Þ
d ¼ dsph ð4Þ
l2
Evaluation of a Backscatter Imaging LDV System 7

where ut is the terminal velocity of the particle, dsph is the equivalent spherical par-
ticle diameter, qf ( ¼ 1.17 kg=m3) is the density of fluid (air), qs ( ¼ 1200 kg=m3) is the
particle density, and lf ( ¼ 1.85  105 kg=m-s) is the air viscosity. The terminal velo-
city and diameter measured by BILS were converted to corresponding u* and d*
values using Equations (3) and (4). The experimental values of u* plotted against
experimental values of d* are shown in Figure 4. The error bars shown correspond
to two standard deviations about the mean value, which gives a confidence interval
of approximately 95%. The correlation of Haider and Levenspiel (1989) between the
two parameters, u* and d*, is given by Equation (5):
 
18 ð2:3348  1:7439 UÞ 1
u ¼ 2 þ ð5Þ
d d0:5

where U is the sphericity of particle. This relationship is shown in Figure 4 for a


sphericity of 1. The results obtained were in good agreement with the drag curve
from Equation (5). The imaging data were then used to estimate the sphericity of
the Nu-Pareil particles.
A typical example of an image for Nu-PareilTM particle, recorded by BILS, is
shown in Figure 5. The sphericity of a particle (U) is given by Equation (6), where
s is the surface area of a sphere having the same volume as the particle and S is
the actual surface area of the particle.
s
U¼ ð6Þ
S
Assuming the particle to be an oblate ellipsoid, the actual surface area (S) of the
particle can be estimated from Equation (7).
S ¼ 2pa2 ½1 þ ð1  e2 ÞatanhðeÞ=e ð7Þ
where a is the major axis of the ellipse, b is the minor axis, and the eccentricity, e, is
given by Equation (8).
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e ¼ ð1  b2 =a2 Þ ð8Þ

Figure 4. Experimental mean dimensionless particle velocity versus mean dimensionless


particle diameter with 95% confidence interval (log-log scale).
8 P. Pandey et al.

Figure 5. Typical image of a near-spherical Nu-Pareil particle obtained from BILS.

The major and minor axes were calculated from the values of the projected area
(Aproj) and the ellipsivity (e) of the particle, which were obtained by the analysis of
the particle images from the BackscaPP software and are given by:
b
Aproj ¼ pab and e¼  100 ð9Þ
a
The surface area of a sphere with the same volume as that of the particle (s) was
obtained by equating the volume of the ellipsoid to that of a sphere. An equivalent
2
diameter (dequv) was found from Equation (10) to evaluate s ð¼ pdequv Þ.

dequv ¼ 2ða2 bÞ1=3 ð10Þ

Using these correlations, the sphericities of the particles were obtained. The results
confirm that the sphericities of the particles used were close to 1 (0.98 < U <
0.99). Thus, BILS could also be successfully used to obtain information on the
particle’s shape.

Application to Circulating Fluidized Bed (CFB)


After successful validation, BILS was tested for its ability to measure particle velo-
cities and size near the wall region of the riser section of a cold flow circulating flui-
dized bed. This experimental test facility is situated at the National Energy
Technology Laboratory (NETL), U.S. Department of Energy, Morgantown, W.V.
The system is described in detail by Monazam et al. (2001). The unit is an atmos-
pheric, cold flow model of a CFB combustor and is shown in Figure 6. The riser
has an internal diameter of 30.5 cm and a height of 15.2 m, with a standpipe of
Evaluation of a Backscatter Imaging LDV System 9

Figure 6. Overview of the cold flow circulating fluidized bed setup at NETL, Morgan-
town, W.V.

internal diameter 25.4 cm. The solid material used was cork with a Sauter mean
diameter of 812 mm and a particle density of 189 kg=m3 (Shadle et al., 2002). The size
distribution of cork is shown in Figure 7. The CFB walls are made of metal with
some sections made of PlexiglasTM to allow for visual observation and provide clear
optical access to instruments such as BILS. A test was conducted to study the effect

Figure 7. Size distribution of the cork material used in the CFB, Sauter mean diameter ¼
812 mm.
10 P. Pandey et al.

Figure 8. Comparison of velocity readings recorded by BILS with and without Plexiglas
section to study the curvature effect of the walls and its comparison with the y ¼ x line.

of PlexiglasTM wall curvature on the velocity measurements in the axial direction.


The velocity values recorded by BILS with and without the PlexiglasTM section
under similar conditions are shown in Figure 8. It can be seen that the curvature
and the quality of the PlexiglasTM does not significantly affect the velocity measure-
ments. However, due to the high attenuation and roughness of the PlexiglasTM
surface in the fluid bed unit, sharp images of the particles could not be obtained.
The LDV transceiver was fixed on a translation stage that allowed precise posi-
tioning (þ=10 microns) of the measurement volume radially inside the bed (10
microns). Measurements deep inside the bed were not possible due to solid particles
obstructing the path of the laser beam. The penetration of the laser beam into the
bed decreased with increasing solids loading.

Results and Discussion


The CFB was operated under a wide range of operating conditions, including the
dilute, core-annular, and fast fluidization regimes (Bi & Grace, 1995). These tests
were conducted at gas velocities above which an S-shaped axial pressure profile
could be found. The tests were conducted using a central composite design with
two independent variables, namely riser superficial gas velocity, Ug, and solids circu-
lation rate, Ms. A dimensionless load ratio (m) was defined to characterize the oper-
ating conditions. This is given by Equation (11), where A is the cross-sectional area
of the riser and qf is the density of fluid (air).

Ms
m¼ ð11Þ
qf AUg

Figure 9 is a typical example of the velocity data obtained 1 mm from the wall of
the riser section at an axial location of z ¼ 9 m from the bottom. A negative velocity
Evaluation of a Backscatter Imaging LDV System 11

Figure 9. Typical temporal evolution of particle axial velocity obtained 1 mm from the wall of
the riser section of the CFB at a location of z ¼ 9 m for Ug ¼ 5.4 m=s and Ms ¼ 2722 kg=h
(negative velocity here indicates that the particle is moving downwards).

on the graph indicated that the particle was moving downwards. Instantaneous velo-
city data for each solid particle passing through the measurement volume was
recorded with time. The scattered velocity values demonstrate the turbulent flow
conditions in the riser. It is clear from the graph that the majority of the particles
near the wall were moving downwards. Most of the downward-moving particles
had velocities in the range of 1–2 m=s. The typical duration for which data was taken
for measurements near the wall generally varied from 30 to 45 s. There is always a
substantial overestimation of the number of particles passing through the probe vol-
ume when investigating two-phase flows due to multiple recording of the same par-
ticle as it passes through the interference pattern (Van den Moortel, 1998; Lazaro,
1991; Gould & Loseke, 1993). This causes a ‘‘velocity bias’’ in the measurements
when taking a number average. This ‘‘bias’’ was eliminated from the measurements
by a post-processing step, proposed by Pandey et al. (2004), where multiple record-
ings by the same particle were discarded. The second kind of ‘‘velocity bias’’ associa-
ted with LDV measurements is associated with long measurement times (Johnson
et al., 1983). To minimize this bias, the velocity measurements were done for a short
span. The typical time for which data was taken for measurements near the wall was
in the range from 30 to 45 s. The acquisition time was chosen individually for all
operating conditions in order to acquire a sufficient number of measurements to
undertake a statistical analysis. Such a data set provides a little over 1000 data
points. This is typical of LDV measurements made on a CFB by various other
researchers (Van den Moortel, 1998; Liu et al., 2002). Zhu et al. (2001) report to have
collected data for 1200 points to get an estimate of the mean velocity from their fiber
optic probe, which would encounter the same velocity bias problem.
The data reproducibility of the instrument for a turbulent system such as CFB is
of significance and was tested under steady-state conditions. The velocities of
ascending and descending particles were averaged separately and the reproducibility
of each one, along with the mean velocity, was tested. This is shown in Figure 10 for
12 P. Pandey et al.

Figure 10. Data reproducibility of the instrument for Ug ¼ 4.58 m=s and Ms ¼ 2722 kg=h,
z ¼ 9 m. The upward- and the downward-flowing particles were studied separately, along with
the mean velocity.

the condition Ug ¼ 4.58 m=s and Ms ¼ 2722 kg=h. The system is seen to reproduce
data remarkably well even under turbulent flow conditions.
The turbulence intensity in the CFB was determined from high-speed pressure
differentials measured over a 61 cm interval for 180 s at 25 Hz for various operating
conditions. It was expressed as a dimensionless number FP, the normalized ampli-
tude of the pressure drop fluctuations, given by Equation (12), where DPi is the
instantaneous pressure drop (Sun & Grace, 1992).
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ffi
Pn 
i¼1 DP i  DP =n
FP ¼ ð12Þ
DP
The variation of the turbulence intensity with load ratio is shown in Figure 11. It can
be seen that FP from pressure drop fluctuations reveals no difference for a relatively
wide range of load ratios. However, the difference in flow pattern with load ratio was
significant and was readily obtained from the particle velocity data collected by the
LDV system. Data sets displaying particle velocities, measured over the sampling
time, were obtained at a location 1 mm from the wall for a series of conditions. These
are shown in Figure 12 in order of increasing load ratio, where N is the number of
particle hits obtained in the sampling time of 30 s. As expected, N is found to increase
with increasing load ratio. Thus, more particles pass through the measurement vol-
ume at a higher load ratio. Also, the distribution of velocities across the sampling
period varied as the load ratio increased. For very dilute flows, the individual
particle velocities varied more or less randomly with time; however, as the solids load
ratio increased, there appeared to be more time-correlated particles traveling at
or near the same velocity. In other words, the denser flow conditions apparently
gave rise to more streamers or groups of particles that traveled together in groups
or clusters.
The instantaneous velocity data, as a function of the distance from the wall, is
shown in Figure 13 for the condition Ms ¼ 907 kg=h and Ug ¼ 4.58 m=s. For this
Evaluation of a Backscatter Imaging LDV System 13

Figure 11. Pressure drop variation with the increase in load ratio at z ¼ 9 m.

dilute condition, the majority of particles were found to be moving upwards, even
very close to the wall. The number of particles moving downwards decreased with
increasing penetration distance. Also, N was found to decrease with increasing dis-
tance from the wall, indicating higher solids concentration closer to the wall. Here
it must be pointed out that the reduction in the number of data points will be a com-
bined effect of the reduction in concentration as well as the obscuring of the laser
beam caused by the particles as the measurement volume is moved deeper inside
the bed.
The mean velocity of solids was found to increase with increasing Ug, as shown
in Figure 14. This is consistent with the core-annular flow pattern anticipated in the
riser section (Van den Moortel et al., 1998; Nieuwland et al., 1996). The particles
moving upwards and downwards were averaged separately, and the effect of Ug is
shown in Figure 15. An increase in Ug is seen to increase the mean velocity of the
particles moving upwards at any location, but the mean velocity of particles moving
downwards remained relatively unaffected. This is consistent with the work of Zhou
et al. (1995).
The mean velocity of solids was found to increase with decreasing Ms, as shown
in Figure 16. This observation agrees with the results of Tadrist and Cattiew (1993),
Yang et al. (1993), and Zhou et al. (1995) and was attributed to the fact that an
increase in the solids circulation rate caused the gas velocity near the walls to
decrease.

Granular Temperature
The granular solids temperature (T ) uses the analogy between particles in a fluidized
bed and thermal motion of particles to describe two-phase systems (Cody et al.,
1996). It has been described as the ‘‘single most important key to understanding
the behavior of rapid granular flows’’ (Campbell, 1990). It can be defined as the
ensemble average of the squared fluctuation velocity. If the instantaneous z-direction
particle velocity is denoted by vz,i and the average velocity by Vz, then the fluctuation
14 P. Pandey et al.

Figure 12. Variation in measured particle velocities for different load ratios at a distance 1 mm
from the wall.

velocity in one dimension, wz,i, is given by wz,i ¼ vz,i  Vz. Here, it was assumed that
the velocity in the z-direction was the predominant velocity in comparison to the
Evaluation of a Backscatter Imaging LDV System 15

Figure 13. Variation in particle velocity distribution with the distance from the wall for the
operating condition: Ms ¼ 907 kg=h and Ug ¼ 4.58 m=s.

velocity in the other directions. Thus the net average velocity can be approxi-
mated by the average of the velocity of particles in the z-direction. The granular
16 P. Pandey et al.

Figure 14. Variation in the mean solids velocity with superficial gas velocity at Ms ¼
4005 kg=h at z ¼ 9 m.

temperature can be defined by Equation (13) (Gidaspow, 1994; Cody et al. 1996).
1
T  ¼ hw2z;i i ð13Þ
3
Using this definition, the granular temperature was evaluated for various oper-
ating conditions at 1 mm from the wall from the LDV velocity data and is shown in
Figure 17. For most cases, T decreases with an increase in load ratio thereby indi-
cating a more ordered flow near the wall region for higher loadings. In general, for
modeling purposes, a single value of the granular temperature is used at any given

Figure 15. Variation of the mean velocity of up-flowing and down-flowing particles with
superficial gas velocity at Ms ¼ 2722 kg=h at z ¼ 9 m.
Evaluation of a Backscatter Imaging LDV System 17

Figure 16. Variation in the mean solids velocity with solids circulation rate at Ug ¼ 5.16 m=s
at z ¼ 9 m.

location in the bed. The values of T determined here, 0.2 to 1.0 m2=s2, compare
favorably with theoretical estimates and measurements of FCC materials (Gidaspow
& Huilin, 1996). Granular temperatures theoretically represent the fluctuations in
particle velocities on the time scale required for particle collisions or length scales
on the order of the mean free path between collisions. Ideally, measurements would
be grouped together based only upon these criteria, but additional information
regarding solids concentration is necessary to better qualify the data set.
In lieu of such additional data, a lumping of the same directional flow fields was
adopted to obtain better insight into the flow structures. The particles flowing

Figure 17. Variation of granular temperature with load ratio for the mean flow of particles
along with upflow and downflow separately.
18 P. Pandey et al.

upwards and downwards were analyzed separately, and a T value corresponding to


each of them was also evaluated and shown in Figure 17. The low values of the T
for up and down flow, in comparison to the T for the net flow, were not unexpected,
indicating that both flows were more ordered in comparison to the net flow. The T
from these subsets were smaller by a factor of two or three, and roughly within an
order of magnitude of the theoretical values. In general, the T value for particles
moving upwards was found to decrease with increasing load ratio whereas for the
particles moving downwards T seemed to be unaffected by the load ratio. Also,
for all cases, T for downflow was found to be lower than that of the upflow, indicat-
ing that downflow of particles is more ordered. This is consistent with the idea of the
formation of ‘‘clusters’’ near the wall region whose movement is unaffected by the
operating conditions (Rhodes et al., 1992; Noymer & Glicksman, 2000; Pandey
et al., 2004).
The variation of T with the distance from the wall was also studied and is
shown in Figures 18 and 19 for two sets of operating conditions. For the near-wall
region, T was found to increase with the increase in distance from the wall. This has
been observed in the past by Neri and Gidaspow (2000), using CFD (computational
fluid dynamics) code. These findings must be qualified by the extent of obscuration
since this interference will be greater as the load ratio increases.

Conclusions
Particle velocity and image measurements recorded by BILS were successfully vali-
dated. BILS demonstrates the ability to record simultaneous particle velocity and
size data that agree well with theoretical calculations. It was shown that the imaging
component of BILS provides valuable information on the particle’s shape (spher-
icity) and size (limitation: only two-dimensional images are obtained). BILS was suc-
cessfully used to obtain particle velocities in the near-wall region of a pilot-scale
CFB. The velocity distribution obtained from LDV measurements was seen to

Figure 18. Variation of granular temperature with the distance from the wall along with
upflow and downflow separately for Ms ¼ 2722 kg=h and Ug ¼ 3.75 m=s.
Evaluation of a Backscatter Imaging LDV System 19

Figure 19. Variation of granular temperature with the distance from the wall along with
upflow and downflow separately for Ms ¼ 2722 kg=h and Ug ¼ 5.40 m=s.

change with a change in load ratio, while the turbulence intensities, as extracted from
variations in the pressure drop, were invariant.
The LDV velocity data results show that the mean velocity of solids increased
with increasing Ug and decreasing Ms. The mean velocity was found to be affected
more by the change in Ug than in Ms. Also, the mean velocity of particles moving
downwards was relatively unaffected by the operating conditions and was found
to be in the range of 1–2 m=s. The granular temperature was best determined by
separately analyzing the upward and downward flow particles. The T values for
the upward flowing particles in the near-wall region were found to decrease with
increasing load ratio (Figure 17). Like the mean velocities for the downward-flowing
particles, the granular temperature for particles moving downwards remained
relatively constant ( 0.15 m2=s2) even with varying load ratios from dilute upflow
to very densely loaded core-annular flow regimes.

Nomenclature
A cross-sectional area of the riser, m2
d fringe spacing, mm
d* dimensionless particle diameter
dp particle diameter, m
dsph equivalent spherical particle diameter, m
D riser diameter, m
FD Doppler frequency, Hz
FP dimensionless normalized amplitude of the pressure drop fluctuations
H height of riser section, m
L length of riser, m
m dimensionless load ratio
Ms solids circulation rate, kg=h
N total number of particles that pass through the measurement volume
20 P. Pandey et al.

T granular temperature, m2=s2


u* dimensionless terminal velocity
ut terminal velocity, m=s
Ug superficial gas velocity, m=s
vi,z instantaneous z-direction velocity, m=s
V particle velocity, m=s
Vz average z-direction velocity, m=s
wi fluctuation velocity, m=s
z distance from the bottom of the riser, m
a two times the beam crossing angle, degree
DP pressure drop across riser, kg=ms2
k wavelength of the laser beam, nm
qf density of fluid, kg=m3
qs particle density, kg=m3
U sphericity of particle
Abbreviations
BILS backscatter imaging LDV system
CCD charged coupled devices
CFB circulating fluidized bed
CFD computational fluid dynamics
FCC fluidized catalytic cracking
FFT fast Fourier transform
LDV laser Doppler velocimeter
NETL National Energy Technology Laboratories
PDPA phase Doppler particle analyzer
RSA real-time system analyzers

References
Arastoopour, H. & Y. Yang. 1991. Experimental studies on dilute gas and cohesive particles
flow behavior using Laser doppler anemometer. In Fluidization VII, ed. by O. Potter & D.
Nicklin. New York: Engineering Foundation. pp. 723–730.
Bi, H. & J. Grace. 1995. Flow regime diagrams for gas-solid fluidization and upward trans-
port. Int. J. Multiphase Flow 21 (6): 1229–1236.
Campbell, C. 1990. Rapid granular flows. Annu. Rev. Fluid Mech. 22: 57–92.
Cody, G., D. Goldfarb, G. Storch, Jr., & A. Nooris. 1996. Particle granular temperature in
gas-fluidized beds. Powder Technol. 87: 211–232.
Gidaspow, D. 1994. Multiphase Flow and Fluidization: Continuum and Kinetic Theory Descrip-
tions. New York: Academic Press.
Gidaspow, D. & L. Huilin. 1996. Collisional viscosity of FCC particles in a CFB. AIChE J.
42 (9): 2503–2510.
Goodman, M. & S. Cowin. 1972. A continuum theory for granular materials. Arch. Rational
Mech. Anal. 44: 249–266.
Gould, R. & K. Loseke. 1993. A comparison of four velocity bias correction techniques in
laser Doppler velocimetry. Trans. ASME 115: 508–514.
Haider, A. & O. Levenspiel. 1989. Drag coefficient and terminal velocity of spherical and non-
spherical particles. Powder Technol. 58: 63–70.
Herbert, P., T. Gauthier, C. Briens, & M. Bergougnou. 1994. Application of fiber optic reflec-
tion probes to the measurement of local particle velocity and concentration in gas-solid
flow. Powder Technol. 80: 243–252.
Evaluation of a Backscatter Imaging LDV System 21

Horio, M., K. Morishita, O. Tachibana, & N. Murata. 1988. Solid distribution and movement
in circulating fluidized beds. In Circulating Fluidized Bed Technology II, ed. by P. Basu &
J. Large. Oxford: Pergamon. pp. 147–154.
Johnson, D., D. Modarress, & F. K. Owen. 1983. An experimental verification of laser velo-
cimeter sampling bias and its correction. Paper presented at the Annual Meeting of
ASME, Nov. 14–19. pp 153–162.
Knowlton, T. & S. Karri. 2001. Results of CFB riser modeling Challenge Problem 2: Indus-
trial applications of circulating fluidized beds. Paper presented at AIChE meeting,
November, Reno, Nevada.
Lazaro, B. 1991. Evaluation of phase Doppler particle sizing in the measurement of optically
thick, high number sprays, UTRC91-11. East Hartford, Conn.: United Technologies.
Liu, J., J. Grace, & H. Bi. 2002. Radial distribution of local particle velocity in a high-density
circulating fluidized bed riser. In Circulating Fluidized Bed Technology VII, ed. by J.
Grace, J. Zhu, & H. Lasa. pp. 341–348.
Louge, M., D. Lischer, & H. Chang. 1990. Measurements of voidage near wall of a circulating
fluidized bed riser. Powder Technol. 62: 269–276.
Miller, A. & D. Gidaspow. 1992. Dense, vertical gas-solid flow in a pipe. AIChE J. 38:
1801–1815.
Monazam, E., L. Shadle, & L. Lawson. 2001. A transient method for determination of satu-
ration carrying capacity. Powder Technol. 121: 205–212.
Monceaux, L., M. Azzi, Y. Molodtsof, & J. Large. 1985. Overall and local characteristics of
flow in a circulating fluidized bed. In Circulating Fluidized Bed Technology, ed. by P.
Basu. New York: Pergamon. pp. 185–191.
Neri, A. & D. Gidaspow. 2000. Riser hydrodynamics: Simulation using kinetic theory. AIChE
J. 46 (1): 52–67.
Nieuwland, J., R. Meijer, J. Kuipers, & V. Swaaij. 1996. Measurements of solids concentration
and axial solids velocity in a gas-solid two-phase flows. Powder Technol. 87: 127–139.
Noymer, P. & L. Glicksman. 2000. Descent velocities of particle clusters at the wall of a cir-
culating fluidized bed. Chem. Eng. Sci. 55: 5283–5289.
Pandey, P. 2002. Application evaluation of a prototype backscatter imaging LDV system.
Masters thesis, West Virginia University.
Pandey, P., R. Turton, P. Yue, & L. Shadle. 2004. Non-intrusive particle motion studies in the
near-wall region of a pilot-scale circulating fluidized bed. Ind. Eng. Chem. Res. 48 (18):
5582–5592.
Rhodes, M., P. Laussmann, F. Villain, & D. Geldart. 1988. Measurement of radial and axial
solids flux variations in the riser of a circulating fluidized bed. In Circulating Fluidized Bed
Technology II, ed. by P. Basu & J. Large. Oxford: Pergamon. pp. 155–164.
Rhodes, M., H. Mineo, & T. Hirama. 1992. Particle motion at the wall of a circulating flui-
dized bed. Powder Technol. 70: 207–214.
Shadle, L., E. Monazum, & J. Mei. 2002. Circulating fluid bed operating regimes. In Circulat-
ing Fluidized Bed Technology VII, ed. by J. Grace, J. Zhu, & H. Lasa. pp. 255–262.
Shi, H., Q. Wang, C. Wang, Z. Luo, M. Ni, & K. Cen. 2002. PIV measurement of the gas-solid
flow pattern in a CFB riser. In Circulating Fluidized Bed Technology VII, ed. by J. Grace,
J. Zhu, & H. Lasa. pp. 161–168.
Sun, G., & J. R. Grace. 1992. Effect of particle size distribution in different fluidization
regimes. AIChE J. 38: 716– 722.
Tadrist, L. & P. Cattiew. 1993. Analysis of two-phase flow in a circulating fluidized bed. In
Circulating Fluidized Bed Technology IV. pp. 702–707.
TSI Inc. 2000. Backscatter Imaging LDV System, Hardware, Software and Analysis, Users
Manual. Shoreview, Minn.: TSI inc.
Van den Moortel, T., E. Azario, R. Santini, & L. Tadrist. 1998. Experimental analysis of the
gas-particle flow in a circulating fluidized bed using a phase Doppler particle analyzer.
Chem. Eng. Sci. 53 (10): 1883–1899.
22 P. Pandey et al.

Viitanen, P. 1993. Tracer studies on a riser reactor of a fluidized catalyst cracking plant. Ind.
Eng. Chem. Res. 32: 577–583.
Wang, Y., F. Wei, Z. Wang, Y. Jin, & Z. Yu. 1998. Radial profiles of solids concentration and
velocity in a very fine particle (36 mm) riser. Powder Technol. 96: 262–266.
Wei, F., H. Lin, Y. Cheng, Z. Wang, & Y. Jin. 1998. Profiles of particle velocity and solids
fraction in a high-density riser. Powder Technol. 100: 183–189.
Yang, Y., Y. Jin, Z. Yu, J. Zhu, & H. Bi. 1993. Local slip behavior in the circulating fluidized
bed. AIChE Symp. Ser. 89 (296): 81–90.
Zheng, Z., J. Zhu, J. Grace, C. Lim, & C. Brereton. 1992. Particle motion in circulating flui-
dized beds via microcomputer-controlled color-stroboscopic photography. In Fluidization
VII, ed. by O. Potter & D. Nicklin. New York: Engineering Foundation. pp. 781–789.
Zhou, J., J. Grace, C. Lim, & C. Brereton. 1995. Particle velocity profiles in a circulating flui-
dized bed riser of square cross-section. Chem. Eng. Sci. 50 (2): 237–244.
Zhu, J., G. Li, A. Qin, F. Li, H. Zhang, & Y. Yang. 2001. Direct measurements of particle
velocities in gas-solids suspension flow using a novel five-fiber optical probe. Powder
Technol. 115: 184–192.

You might also like