Paper No 7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010


LDV techniques to simultaneously measure gas and particulate phase velocities in a jet
plume in a 2D bubbling fluidized bed


Alexander Mychkovsky, Deepak Rangarajan and Steven Ceccio

University of Michigan, Department of Mechanical Engineering
1231 Beal Ave, Ann Arbor, MI, 48109, USA
almych@umich.edu


Keywords: Fluidized Bed, LDV, Jets



Abstract

A laser Doppler velocimetry (LDV) technique has been developed to simultaneously measure the gas phase and particulate
phases of a jet in a 2D bubbling fluidized bed. The LDV system was configured to mitigate spurious optical intensity
fluctuations, which can contaminate velocity measurements in optically dense flows. Also, the jet gas was seeded with ice
crystals, which were formed by rapidly condensing and freezing the moisture in the jet air just prior to injection in order to
produce small seed particles which track the gas phase. LDV bursts from the bed particles and gas tracers were
simultaneously acquired to yield the particulate and gas phase velocities at a given location within the jet plume. The bursts
from the tracer crystals and bed particles were differentiated based on their intensity and coincidence. Example gas and
particulate phase velocity profiles are presented.


Introduction


Nomenclature


Fluidized beds are commonly used as chemical reactors and
solid fuel combustors due to their rapid transport. In many
configurations, gas jets are injected into a bubbling
particulate emulsion. The mixing of the jet is often key to
the efficiency of the process since the jet plume is a region
of rapid mass, momentum, and energy transfer.

In order to understand the mixing and transport, the velocity
profiles of both phases at various axial locations must be
known. An overview of transverse and axial profile
measurements is given by Massimilla (1985). Typically,
gas velocities are obtained with pitot tubes, and high-speed
video is used to determine the particle velocities. Pitot
tubes are intrusive and their pressure measurements can
only be correlated to gas velocities if the gas and solids
momentum contributions can be distinguished from one
another. And, in order to measure the relatively high
particle velocities in a jet plume (-10 m/s), significant
illumination is needed for video techniques. This is often
difficult in bubbling beds due to their limited optical access;
therefore, these measurements tend to be limited to particles
near the wall. Other optical techniques, such as optical
fiber probes (Zhu et al., 2001) and Laser Doppler
velocimetry (LDV) (Werther et al., 1996; Breault et al.,
2008) have been successfully applied to dilute, circulating
fluidized beds. The current work describes the
development and implementation of an LDV technique to
obtain simultaneous gas and particulate phase velocity
measurements in a jet plume in two-dimensional (2D)
bubbling fluidized bed.


C speed of light (m/s)
f frequency (Hz)
I intensity (mV)
L laser cavity length (m)
t time (s)
v velocity (m/s)
x transverse coordinate (mm)
y axial coordinate (mm)
Greek letters
8 LDV fringe spacing (t)m)
Subsripts
B Bragg
fl fluidization
g gas
j jet
p particulate


Experimental Facility

Experiments were conducted in a two-dimensional bubbling
fluidized bed, which is shown in Figure 1. The bed
dimensions are 457 mm wide by Im tall with a 12.7 mm
gap. The walls are transparent acrylic with 102 mm by
153 mm by 5 mm thick quartz viewing windows inserted 50
mm above the vertical jet inlet orifice, which is 9.2 mm in
diameter. The vertical jet is located midway across the
porous polyethylene fluidization distributor.

The particles used in the emulsion are 838 jtm Sauter mean
diameter high-density polyethylene (HDPE) micropellets,
which have a density of 900 kg/m3. The minimum





Paper No


fluidization velocity for these particles was experimentally
determined to be 29 cm/s. The velocity profiles presented
in this paper are obtained with a jet inlet velocity of 92 m/s
and a distributor fluidization velocity of 34 cm/s.


figure 1: vertical jet (v, = vY m/s) in a numalzea emulsion
of 838ytm HDPE particles (vf = 34 cm/s).

LDV Configuration

Bed particle and gas phase velocities were obtained with a
two-component LDV system employing an argon-ion laser
for illumination of the seed particles. The axial velocities
were recorded on channel 1 using the green beam (514.5
nm) and the transverse velocities were recorded on channel


2 using the blue (488 nm) beam.
settings are listed in Table 1.


The LDV parameters and


Ch 1 Ch2
Laser Power per Beam (mW) 90 55
Beam Diameter (microns) 90 85
PMT Gain (mV) 450 450
Burst Threshold (mV) 250 150
Frequency Downmixing (MHz) 0 0
Band Pass Filter (MHz) 5-50 20-65
Bragg Shift Frequency (MHz) 40 40
Fringe Spacing (microns) 3.74 3.55
Velocity Range (m/s) 131 to -37 71 to -89
Coincidence Interval ([s) 10 10
Table 1: LDV parameters and settings.

LDV is an established measurement technique and
overviews of the fundamentals are provided in Durst et al.
(1976) and Stevenson (1982). The basic operating
principle is that an LDV signal is recorded when a particle
scatters light as it traverses the interference fringe pattern
established by intersecting monochromatic laser beams.
Therefore particle velocity is measured as the product of the
frequency of the scattered light and the fringe spacing.

In order to remove the directional ambiguity of the velocity
associated with a specific Doppler frequency, one of the
intersecting laser beams is frequency shifted by an
acousto-optic element, usually a Bragg Cell operating at
40MHz. This slight optical frequency shift causes the
fringe pattern to propagate in space at a velocity of 3f/
within the measurement volume. Therefore, a stationary
particle would produce a Doppler signal at fB, a particle


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

moving in the direction of the fringe motion would produce
a Doppler signal less than fB, and a particle moving in the
opposite direction of the fringe motion would produce a
Doppler signal greater than fs. Typically, the beams are
oriented so that the fringe pattern moves counter to the bulk
particle motion.

Another feature of the Bragg Cell is that it can also
modulate the intensity of the frequency shifted beam,
creating an intensity fluctuation that occurs at twice the
Bragg frequency (Mychkovsky et al., 2009). This
intensity fluctuation will mix with the Doppler burst signal
according to trigonometric relationship expressed in
equation 1.

cos(2;r2fB t)cos(2r ft)=

cos [2ff(2fB +f)t]+ cos[2(2fB -f)t] (1)
2 2


As illustrated in the FFT shown Figure 2, the Bragg shifted
beam intensity fluctuation at 80 MHz mixes with a Doppler
burst signal at 47 MHz to create spurious peaks at 33 MHz
and 127 MHz. The false peak at 2fB-f can be problematic
due to its proximity to the Doppler burst frequency.

In addition to the Bragg Cell intensity modulation, all beams
emerging from ionized gas lasers will experience intensity
fluctuations due to mode hopping, which occurs at a
frequency of C/2L, where C is the speed of light and L is the
cavity length. For a 1.2m laser, the mode hopping
frequency will occur at 125 MHz. When mixed with the
Bragg cell intensity fluctuation, this produces a peak at
45 MHz, as shown in Figure 3. Note that for this figure
the Doppler burst frequency occurs at 36 MHz and thus a
Bragg-Doppler mixed signal occurs at 44 MHz, which is
very near the C/2L-2fB mixed peak. Note that the laser
used to obtain the data in Figure 2 was smaller with a cavity
length of approximately 300 cm, and therefore had a mode
hopping frequency around 500 MHz.


33 47


80
FFT of PMT Signal


127 MHz


Figure 2: Frequency mixing of the Doppler burst signal
with the Bragg shifted beam intensity fluctuation.






Paper No


2fs C/2L







S C/2L -f,


36 45 80 125 lvHz
FFT of PMT Signal
Figure 3: Frequency mixing of the laser mode hopping
intensity fluctuation with the Bragg shifted beam intensity
fluctuation.

These intensity fluctuation frequency mixings are an optical
phenomena and are only noticeable when a significant
portion of the Bragg shifted beam is reflected and detected
by the LDV photomultiplier, which is the case with
measurements of large particles, measurements in optically
dense regions, or measurements near reflecting boundaries.

In order to eliminate these spurious optical signals, the
intensity fluctuation due to the Bragg cell was minimized by
optimizing the laser beam diameter entering the Bragg cell
as well as the Bragg angle as described by Mychkovsky et
al. (2009). The smaller 300 cm argon-ion laser was not
powerful enough to detect the Doppler bursts of the small
gas tracer ice crystals, therefore the larger 1.2 m laser was
necessary. The residual peak at 45 MHz was avoided by
orienting the axial velocity measurement beams (channel 1)
so that the fringes move in the direction of the jet flow and
the band pass filter was set to 5-50 MHz. Any false bursts
detected at 45 MHz where subsequently omitted when
determining velocity values. No frequency down-mixing
was used in the post-processing of either channel to avoid
further complications.

Jet Gas Seeding

The jet gas was seeded with ice crystals, which were formed
by rapidly condensing and freezing (T, = -50C) the moisture
in the jet air just prior to injection via a dry ice heat
exchanger. These ice crystals are on the order of a few
microns in size (Sasaki et al., 1980) and therefore reflect
much less laser light than the larger bed particles, which are
several hundred microns in diameter. Furthermore, due to
the small size and low density of the ice crystals, they track
the jet gas velocity well, which in the axial direction is
significantly higher than the bed particle velocities.
Therefore, the gas tracer ice crystals will have low intensity,
high axial velocity bursts whereas the larger bed particles
will have high intensity, low axial velocity bursts as shown
in Figure 4.

As a safety precaution, it should be noted that rapid
condensation of moisture in air is accompanied by
separation of electric charges which can produce large
voltage discharges, essentially forming lightening, if the
apparatus is not properly grounded (Grosu et al., 2007)


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


Bed Particles and Ice Crystals (Ops Coincidence)
1000
900
SIntensity \ Bed Particles
700 Fluctuation -
600 at 45 -Hz-
g 500 Ice Crystals
400
300
200

0
-40 -20 0 20 40 60 80 100
v (m/s)

Figure 4: Doppler burst intensities and velocities for bed
particles and gas tracer ice crystals measured in a jet plume.

LDV bursts

Intensity subranging has been done by Lee and Durst (1982)
in ducts Barlow and Morrison (1990) in open air,
particle-laden jets. Large particles span several fringes and
reflect a larger fraction of the laser light when passing
through the LDV measurement volume and therefore have
larger burst pedestals than tracer particles whose size is on
the order of the fringe spacing (Figure 5).

When the burst pedestal is removed with a high pass filter, it
is evident that the smaller tracer particles have much cleaner
Doppler signals (Figure 6). This is because particles that
are much larger than the fringe spacing require surface
inclusions, which are often erratic, to scatter light. If the
entire surface continuously produced a Doppler signal then
the burst gate time would be on the order of the particle
residence time in the measurement volume, which is
approximately Dplv since Dp>>D. However, Figure 7
shows that this is not the case and that particle burst gate
time tends to be limited by the size of the LDV
measurement volume, which is indicative of Doppler bursts
caused by small surface inclusions.

Furthermore, since burst gate time does not necessarily
correspond to the residence time, gate time weighting is not
an appropriate way to correct for any LDV velocity bias
effects that may occur. Therefore, the velocity values
reported for both phases are simply based on the arithmetic
average of the bursts recorded at a given location. This
simple data analysis method has been found to have an
insignificant bias effect on LDV measurements (Ahmed et
al., 1996).



large particle burst

S small particle f I
burst


Figure 5: LDV burst signals from small and large particles.
Modified from Lee and Durst (1982)






Paper No


High Pass Filtered Doppler Bursts



S i -- Gas Tracer Ice
SCrystal Burst
-- Bed Particle Burst
,L., .i u.. Al1Burst Threshold



-5 -4 -3 -2 -1 0 1 2 3 4 5
time (Ips)

Figure 6: Doppler bursts from ice crystals and bed
particles with the pedestal removed.


Bed Particles Only (10is Coincidence)

50 -
40

S30 LDV Bursts
S- ----Dm/v
20 Dp/v
."5:: d ':. .

10

0 10 20 30 40 50
v (m/s)

Figure 7: Doppler burst gate times for bed particles
measured in a jet plume.

Because large particle bursts are not clean as small tracer
bursts, temporal coincidence between the two directional
component channels is more difficult to achieve. For
example, and elongated fissure may only cause a Doppler
burst with one set of fringes depending on its orientation.
In this particular experimental setup, bed particles can move
with velocities up to about 10 m/s. The measurement
volume diameter is approximately 100 pm, therefore a
coincidence interval of 10 us is used for the subranged
particle bursts. Strict coincidence criteria (0 ps), which
requires that both channels detect a valid Doppler burst
simultaneously, is maintained for the gas tracer ice crystals
bursts. Note that the 'Burst Threshold' settings listed in
Table 1 are applied to the high pass filtered Doppler bursts
and that the 'Burst Intensity' values (referenced in the
graphs) are recorded prior to the pedestal removal.

In order to determine the intensity threshold values for the
bed particles and gas tracer ice crystals, the bubbling bed
was: 1) run with a room temperature jet so that no ice
crystals were present, and then 2) run with a cold, seeded jet
in an empty bed so that no bed particles were present. An
intensity histogram of the bed particle only run is shown in
Figure 8. The corresponding probability distribution
(Figure 9) indicates that over 99% of the bed particle bursts
have intensities greater than 200mV.


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



An intensity histogram of the gas tracer ice crystal only run
is shown in Figure 10. The corresponding probability
distribution (Figure 11) indicates that over 99% of the ice
crystal bursts have intensities less than 500mV. Therefore,
in order to eliminate cross-contamination of the data, the
LDV bursts are subranged according to the following
criteria:

1. Bed particles: I>500mV, 10 ps coincidence
2. Gas tracer ice crystals: I<200mV, 0 ps coincidence


Bed Particles Only (10ps coincidence)

200

150

S 100
U
50
-.. W iW 1111 .J ,

0 100 200 300 400 501 600 700 800 900
I (mV)

Figure 8: Histogram of bed article DoDDler bi


urst


intensity.


Bed Particles Only (10ps coincidence)

100%

80% -

60% -

40% -

20% -

0%
0 100 200 300 400 500 600 700 800 900 1000
I (mV)

Figure 9: Probability distribution of bed particle Doppler
burst intensity.


Ice Crystals Only (Ops Coincidence)

200

150 -

e loo
100 -

50 -


0 100 200 300 400 500 600 700 800 900
I(mV)

Figure 10: Histogram of ice crystal Doppler burst intensity.


irst






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


Ice Crystals Only (Ops Coincidence)

100%

80%

60%

40%

20%

0% -
0 100 200 300 400 500 600 700 800 900 1000
I (mV)

Figure 11: Probability distribution of ice crystal Doppler
burst intensity.

The velocity histogram of the two-phase gas-solid flow is
shown in Figure 12. (This the same LDV data set that is
plotted in Figure 4.) Using the aforementioned subranging
criteria, the particulate and gas phase velocity histograms
are presented in Figures 13 and 14, respectively. The
shapes of the clearly distinguishable velocity profiles in
Figure 12 are maintained by this subranging process.
Furthermore, very few ice crystal bursts are discarded with
the intensity thresholding. However, Figure 9 indicates
that roughly 40% of the bed particle bursts are below 500
mV This reduction in particle burst count is offset by the
weaker coincidence criteria. The effect on measured
particle velocity values due to this change in burst
processing criteria was determined to be negligible by
comparing the bed particle only measurements. This can
be physically explained as follows. Since the bed particles
are much larger than the LDV measurement volume, low
intensity bed particle bursts are caused by particles that
graze the measurement volume rather than by smaller, faster
moving bed particles, which would be the case if the bed
particles were on the order of or smaller than the
measurement volume. Figure 15 supports the claim that
there is no relationship between burst intensity and velocity
for the bed particle bursts.


Bed Particles and Ice Crystals (Ops Coincidence)

2000
Bed Particles
1500
1 | Ice Crystals
2 1000
Intensity
500 Fluctuation
at 45 ...Hz
01
-40 -20 0 20 40 60 80 100
v (m/s)

Figure 12: Velocity histogram for bed particles and gas
tracer ice crystals measured in a jet plume.


Bed Particles Subranged (I>500mV, 10ps coincidence)


)0


30

)0
)o i --------------------



OE _111I.,


-40 -20 0


20 40
v (m/s)


60 80 100


Figure 13: Subranged velocity histogram for bed particles
measured in a jet plume.


Ice Crystals Subranged (I<200mV, Ops Coincidence)

2000

1500 -

| 1000 -

500 -

0 . . .
-40 -20 0 20 40 60 80 100
v (m/s)


Figure 14: Subranged velocity histogram
crystals measured in a jet plume.


giF ure 15: Doppler burst intense
bed particles measured in a jet plume.


gas tracer i


velocities f


ce



















or


Results and Discussion

Gas and particulate phase velocity profiles were
simultaneously recorded in a jet plume in a bubbling bed of
838jim HDPE particles. The inlet jet velocity was 92 m/s
and the fluidization velocity was 34 cm/s, which is
approximately 15% greater than then minimum fluidization
velocity for the emulsion. The velocity profiles for the two
phases are juxtaposed at increasing axial distances in
Figures 16-19. The scales in these figures are kept
constant for the purpose of visual comparison.

The axial development for the velocity profiles of the gas
and particulate phases is shown in Figures 20 and 21. As
would be expected, the gas phase centreline velocity decays
and the profile expands with axial distance. The


Paper No


Bed Particles Only (10 is Coincidence)

1000


Z 600 -


0
400

200 "-"

0 ----i
0 5 10 15 20 25 30
v (m/s)


3






Paper No


particulate phase velocity profile also broadens downstream.
One would expect the particulate centreline velocity to
increase as momentum is transferred from the jet gas to the
bed particles via drag. However, it is important to realize
that the LDV method records the Eularian particulate phase
velocity as a function of space and time rather than track the
Lagrangian velocity of a single particle as it accelerates in
the jet plume. Therefore, the average particulate phase
velocity decreases downstream as particles in the emulsion
are entrained from rest into the jet plume.




y = 60mm, V = 92 m/s, Vfl = 34cm/s, 838pm HDPE

70
60 -
50
40- gas phase
S30 -- particdate phase
20
10 --
0
-20 -15 -10 -5 0 5 10 15 20
x (mm)

Figure 16: Gas and particulate phase velocity profiles at
y = 60mm in the vertical jet.


y = 70mm, Vj = 92 m/s, Vfl = 34cm/s, 838pm HDPE

70 -
60 -
50 -
S40- -- gas phase
30 -- particulate phase
20
10 a
0
-20 -15 -10 -5 0 5 10 15 20
x (mm)

Figure 17: Gas and particulate phase velocity profiles at
y = 70mm in the vertical jet.


y = 100mm, Vj = 92 m/s, Vfl = 34cm/s, 838pm HDPE

70 -
60 -
50 -
40 gas phase
304
30 -- particulate phase
20
10
0
-20 -15 -10 -5 0 5 10 15 20
x (mm)

Figure 18: Gas and particulate phase velocity profiles at
y = 100mm in the vertical jet.


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


y = 130mm, Vj = 92 m/s, Vfl = 34cm/s, 838pm HDPE


- gas phase
-a- particulate phase


-20 -15 -10 -5 0 5 10 15 20
S(mmn)

Figure 19: Gas and particulate phase velocity profiles at
y = 130mm in the vertical jet.


Gas Phase, Vj = 92 m/s, Vf = 34cm/s, 838pm HDPE

70
70 -----------____------------____________
60
50 y- = 60mm
S40 ---y= 70mm
30 -y= 100mm
20- --y= 130mm
10-4 -
0
-20 -15 -10 -5 0 5 10 15 20
x (mm)

Figure 20: Gas phase velocity profiles the vertical jet.


Particulate Phase, Vj = 92 m/s, Vfl = 34cm/s, 838pm HDPE

14
12
10
y 10-60mm
8 -e- y=70mm
6 ^-- -y= 100mm
--y= 130mm

2
0
-20 -15 -10 -5 0 5 10 15 20
x (mm)

Figure 21: Particulate phase velocity profiles the vertical
jet.


Conclusions


A procedure has been developed to simultaneously measure
gas and particulate phase velocities based on LDV intensity
and coincidence subranging. This technique has been
implemented to obtain velocity profiles in a gas jet in a
bubbling fluidized bed. Work in the near future involves
calculating mass and momentum fluxes of the gas and
particulate phases. This can then be used to indirectly
determine the solids fraction and coefficient of drag of the
particles in the jet plume.


Acknowledgements


This project is sponsored by the Department of Energy's
Office of Fossil Energy's University Research Program






Paper No 7th International Conference on Multiphase Flow
ICMF 2010, Tampa, FL USA, May 30-June 4, 2010

under project number DE-NT0007649. The research is
also performed under appointment to the Rickover Graduate
Fellowship Program sponsored by Naval Reactors Division
of the U.S. Department of Energy (A. Mychkovsky).

References

Ahmed, S.A.; Nejad, A.S.; & Al-Gami, A. Near-field
Study of a Turbulent Free Jet and Velocity Bias Effects.
J. Propul. Power, Vol. 12, 155-158 (1996)

Barlow, R. S. & Morrison, C. Q. Two-phase velocity
measurements in dense particle-laden jets. Exp. Fluids.,
Vol. 9, 93-104 (1990)

Breault, R.W.; Guenther, C.P & Shadle, L.J. Velocity
fluctuation interpretation in the near wall region of a dense
riser. Powder Tech., Vol. 182, 137-145 (2008)

Durst, F.; Melling, A. & Whitelaw, J.H. Principles and
Practice of Laser-Doppler Anemometry. Academic Press
(1976)

Grosu, F.P; Bologa, M.K.; Polikarpo, A.A. & Motorin, O.V
On modeling of processes of moisture circulation and
electric charge separation in the atmosphere. Surf. Eng.
Appl. Electrochem., Vol. 43, 176-181 (2007)

Lee, S.L. & Durst, F. On the motion of particles in
turbulent duct flows. Int. J. Multiphase Flow, Vol. 8,
125-146 (1982)

Massimilla, L. Gas Jets in Fluidized Beds. in Davison, J.F;
Clift, R. & Harrison, D. Fluidization, Second Ed. pp.
133-171, Academic Press (1985)

Mychkovsky, A.G; Chang, N.A. & Ceccio, S.L. Bragg cell
laser intensity modulation: effect on laser Doppler
velocimetry measurements. Appl. Opt., Vol. 48,
3468-3474 (2009)

Sasaki, O.; Oda, T. & Sato, T. Simultaneous determination
of particle size and density by LDV Appl. Opt., Vol. 19,
2565-2568 (1980)

Stevenson, W.H. Laser Doppler Velocimetry: A Status
Report. IEEE, Vol. 70, 652-658 (1982)

Werther, J.; Hage, B. & Rudnick, C. A comparison of laser
Doppler and single-fiber reflection probes for the
measurement of the velocity of solids in a gas-solid
circulating fluidized bed. Chem. Eng. Process., Vol. 35.
381-391 (1996)

Zhu, J.-X; Li, G.-Z.; Qin, S.-Z.; Li, F.-Y; Zhang, H & Yang,
Y-L. Direct measurements of particle velocities in
gas-solids suspension flow using a novel five-fiber optical
probe. Powder Tech., Vol. 115, 184-192 (2001).