- Email: [email protected]
Evaluation of the TSI small-scale powder disperser
J.AerosolSci..
Pergamon
EVALUATION
OF THE TSI SMALL-SCALE
Vol. 26. No. 8. pp. 1303-1313. 1995 Elsevier Science Ltd Printed in Great Entain
POWDER
DISPERSER
B. T. Chen,* H. C. Yeht and B. J. Fan Inhalation
Toxicology
Research Institute, Lovelace Biomedical and Environmental P.O. Box 5890, Albuquerque, NM 87185, U.S.A.
(First
received
14 October
1994;
and
in finalform
3 July
Research Institute,
1995)
Abstract-The TSI small-scale powder disperser was evaluated to study the effects of total flow rate, capillary flow rate, and particle shape, size, and density on dispersion efficiency and internal losses. Dry powders, including carbon fibers, talc, alumina, and polystyrene microspheres, were used to represent materials with different physical properties. Polystyrene microspheres in the size range of 3-97 pm were used to study the effects of particle size on performance. Results indicated that, for the same operating conditions, dispersion efficiencies for carbon fibers and polystyrene spheres were higher than talc and alumina. Dispersion efficiency increased as the total flow rate increased and reached a constant value when the low rate was >8emin-‘. Also the dispersion efficiency decreased as particle sizes increased at a given operating conditions. As for the internal losses in the disperser, particles were deposited primarily at the expansion zone of the venturi tube. Adjustment of the capillary tube location to increase the capillary flow rate did not result in any noticeable change in the efficiency, while poor alignment of the tube resulted in a significant loss as compared with a proper alignment. Along with these data, computer simulations of the flow field and theoretical predictions of the lift force as a result of high flow shear at the venturi throat were successfully used to interpret particle deposition in the TSl small-scale powder disperser.
INTRODUCTION
Various techniques involving air streams, venturi tubes, fluidized beds, and fluid energy jet mills have been used to disperse dry powders (Hinds, 1982). The ideal powder disperser is one that delivers discrete (deagglomerated) particles to the air with a generation efficiency of 100%. A successful dispersion process requires constant feeding, complete deagglomeration and aerosolization, and a minimal loss of powders in the generators. The most common method of dispersion is to feed the powders into a high-velocity air stream. The shear forces in the turbulent air stream disperse the powders and break up the agglomerates. In the case of venturi tubes, a high-velocity air jet blows across a constriction in a pipe to create a suction that draws clumps of powder into the shear flow of air. Shear forces are generated between the high-velocity air jet that passes through the venturi and the low-velocity particle gas stream that leaves the suction pipe. These forces are generally strong enough to break up the agglomerates. For jet mills, the fluid energy is delivered in high-velocity streams, which circulate around a grinding and classifying chamber where turbulence and centrifugal forces deagglomerate particles. Fine particles follow the gas streamlines and exit in the middle of the chamber, while coarse particles are recirculated for further size reduction. For a given dispersion technique, the dispersion characteristics of a powder depend on the powder material, particle size and shape, moisture content, and buildup of charge on the particles during dispersion. It has been found that dry, hydrophobic materials are easier to disperse than humidified, hydrophilic materials (Hinds, 1982). In addition, dispersion efficiency increases with particle size. Several dry powder generators, including the dust-feed disperser, the fluidized bed, the venturi tube, and the jet mills, have been used to generate aerosols with mass concentrations as high as 1 g me3 for inhalation toxicity studies (Carpenter and Yerkes, 1980; Cheng et al., 1985; Koch et al., 1986; Cheng et al., 1989; Nikula et al., 1993; Timbre11 et al., 1968). In
*Current address: GRAM Inc., 8500 Menaul Blvd., NE, Albuquerque, ‘To whom correspondence should be addressed.
NM
87112,
U.S.A.
B. T. Chen et al.
To Filter (7)
Exhaust Tube
DepositionSufiaces Being Analyzed: (1) Turntable (2) CapillaryTube (3) First Expansion Cone (4) Second Expansion Cone
Fig.
1. Schematic
of the TSI small-scale
powder
disperser.
laboratories, however, a powder generator that can produce a limited quantity of test aerosols is more useful for fundamental studies such as calibration of instruments, development of air samplers, and testing of air-cleaning equipment. In this study, we evaluated the TSI small-scale powder disperser. The objectives were to investigate the effects of flow rate, particle size, and particle shape on the powder generation efficiency and internal losses, and to optimize generation efficiency by adjusting operating parameters.
SMALL-SCALE
POWDER
DISPERSER
Figure 1 shows a schematic diagram of the TSI small-scale powder disperser (TSI Inc., Model 3433, St Paul, MN). The powders to be dispersed were brushed over the surface of an abrasive paper on top of a turntable and were removed by venturi aspiration via a capillary delivery tube. This design arrangement enables this device to produce a limited quantity of test aerosols; thus, only a small amount of test material is required. The lower end of the capillary tube was positioned just above the abrasive paper, while the upper end was at the throat of the venturi. A region of low pressure was created by increasing air velocity through the venturi throat which drew particles up through the capillary tube; thus, the tube operated like a vacuum cleaner as the turntable rotated beneath it. The aerosol particles were deagglomerated in the venturi throat. Shear forces were generated as a result of the difference in velocities between the high-velocity air passing through the venturi throat and the low-velocity particle air stream leaving the capillary tube. The air stream passing through the venturi throat tended to break up the agglomerates and also acted as a sheath around the aerosol stream in the throat section, reducing particle losses to the walls of the expansion cone and the sampling probe. Although a total flow rate of 18.5 4 min- ’ was recommended by the manufacturer, different output flow rates were achieved at the
Evaluation of the TSI small-scale powder disperser
1305
sampling probe exit by adjusting the gauge pressure and bypassing the excess air flow through an exhaust tube. In the present study, a flow plug was used to prevent any air flow passing through the exhaust tube of the small-scale powder disperser (Fig. 1). The aerosol passing through the expansion cone was drawn through the sampling probe, and the particles were collected on a filter. The pressure regulator in the small-scale powder disperser was adjusted to achieve different total flow and capillary flow rates. Total flow rates were varied between 5 and 25 ernin-l to study the effects of flow rate on the aerosol generation efficiency. In addition, the location of the capillary tube opening in the venturi nozzle was varied to study its effect on the generation efficiency and internal losses. At a total flow rate of 18.5 emin- l, the capillary tube was situated either at a normal location to achieve a capillary flow rate of 1.4 emin-’ or was elevated to result in a flow rate of 2 emin- ‘, as recommended by the manufacturer. The effects of these two flow rates on dispersion efficiency and internal losses were investigated using powders of various particle sizes. Also, the effects of tube alignment in the venturi throat on the dispersion efficiency and internal losses were examined. POWDER
GENERATION
AND
CHARACTERIZATION
Four types of dry powders, including carbon fibers (Hercules, Inc., Wilmington, DE), talc (commercial grade), alumina (Al,O,) (Duke Scientific, #347, Palo Alto, CA), and polystyrene latex (PSL) microspheres (Duke Scientific), were used to study the effects of particle shape (cylindrical, disc, irregular, and spherical shapes, respectively) on the dispersion efficiency of the small-scale powder disperser. Monodisperse PSL or polystyrene divinylbenzene (PSDB) spheres with fluorescent green dye (excitation maximum at 459 nm, emission maximum at 512 nm) and nominal particle diameters of 3, 6, 10, 22, and 97 ,cm were used to study the effects of particle size on powder production. The densities of carbon fiber, talc, and alumina powders were measured using a Beckman air pycnometer (Beckman Industries, Fullerton, CA). The densities of PSL and PSDB powders are both 1.05 gcmm3, as provided by the manufacturer. Prior to dispersion, lo-15 mg of each powder were weighed in a microbalance (Model C-26, Cahn Instruments, Cerritos, CA), then put on the abrasive paper of the turntable. A polonium-210 radioactive source was passed over the powder on the turntable to minimize any buildup of electrostatic charge in the particles due to brushing powders onto the abrasive paper of the turntable. This will help subsequent suction of powders through the capillary tube. During dispersion, the powder was sucked through the capillary tube, deagglomerated at the venturi throat and expansion cone, then was drawn into the sampling probe, and collected on a fluoropore filter (Millipore Corp., Bedford, MA) placed at the top of the small-scale powder disperser. After each run, the small-scale powder disperser was disassembled into six parts (Fig. 1): the turntable (part l), the capillary tube (part 2), two expansion cones (parts 3 and 4), the exhaust tube (part 5) and the sampling probe (part 6). The internal surface of each disassembled part was rinsed with acetone into a pre-weighed aluminum container. The containers were left in a ventilated hood to complete the evaporation ofthe acetone. Once dried, the material in each container was weighed to determine the mass of particles deposited on the corresponding part of the small-scale powder disperser. The mass on the fluoropore filter (part 7) was also determined gravimetrically to assess the mass balance of the powders before and after generation and the fractional internal losses at different parts of the small-scale powder disperser. The advantages of using acetone in the gravimetric analysis were two-fold: first, no reaction between acetone and the powders could be detected; and, secondly, the acetone evaporated quickly. For the fluorescence-tagged PSL and PSDB powders, both the gravimetric and fluorometric methods were used for mass analysis. The fluorometric method was processed by extracting the samples with ethyl acetate. The samples were obtained by directly rinsing the internal surfaces of each part or rinsing the samples from the gravimetric analysis. Three ml of ethyl acetate wereadded to each sample to extract the fluorescent pigment from
B. T. Chen et al.
1306
the particles; the sample was placed into a vial, and the preparation was sonicated in a water bath for 3 min (Chen et al., 1991). The solution was then poured into a cuvette, and the fluorescence intensity (proportional to the PSL or PSDB mass) was determined using a fluorescence spectrophotometer (Hitachi, Model F-1200, Danbury, CT). The excitation and emission wavelengths were set at 459 and 512 nm, respectively, for maximum fluorescence detection. DATA
ANALYSIS
For each experiment, the masses (or fluorescence intensities) of the samples collected from the small-scale powder disperser and the filter were used to determine the generation efficiency of the small-scale powder disperser and the percentage of internal losses at each disassembled part. The efficiency (q) was computed from:
M,
’= [x7=2 Mi]
’
(1)
where Mi represents the mass (or fluorescence intensity) of the ith sample indicated by subscripts 226 (denoting various disassembled parts) and 7 (the filter) (Fig. 1). Internal loss at the ith part of the small-scale powder disperser was computed from: WL,
z
Mi
[Cl!=,Mil ’
where WLi (i = 2-6) represents the fraction of internal loss at ith part of the small-scale powder disperser. The mass (or fluorescence intensity) on the turntable (i = 1) was not used in the calculation of the generation efficiency and internal loss because the residual amounts of powders on the turntable was arbitrary after each run. The mass could be biased by the location and placement of the powder on the turntable prior to the run, and by the alignment and adjusted distance between the capillary tube and the turntable. However, the mass of powder on the turntable was included to estimate the mass balance in each run. Results indicated that, depending on the powders, 95-100% of the mass balance was achieved in both gravimetric and fluorometric analyses. THEORETICAL
CONSIDERATIONS
At the venturi throat, aerosol particles encountered drag, gravity, and flow shear-induced lift force. The lift force was generated as a result of the difference in velocities between high-velocity air passing through the venturi throat and the low-velocity particle stream exiting the capillary tube. Particle behavior in the small-scale powder disperser was governed by the equation of motion:
where m,, is particle mass; up is particle velocity; and Fdrag, F,iftr and Fg are the drag, lift, and gravitational force, respectively. Because of the vertical orientation of the internal surfaces, the influence of gravitational force on particle deposition was ignored. The lift or Saffman force (Saffman, 1965) was perpendicular to the flow direction. It appeared in shear flow and was expressed by the following equation: F,ift = 1.165 ~d~((KI/V)1’2W,
(4)
where p and v are the air dynamical and kinematical viscosities, respectively; d, is the particle diameter, K the flow shear, and W the relative velocity between particles and flow. The lift force acted on particles to drive them laterally and to deposit on the wall; therefore, this force could be an important deposition mechanism for particles moving through the small-scale powder disperser. A simple way to understand the significance of the lift force in
Evaluation of the TSI small-scale powder disperser
1307
Table 1. The ratios of lift force to drag force for particles of different sizes Particle size (pm)
Lift/drag ratto, R,, (%)a
3 6 10 22 97
8.4 16.7 27.8 61.2 270
’Equation (5) was used to calculate the ratio. In the calculations, the diameters of the venturi throat and capillary tube in the small-scale powder disperser were 2.06 and 1.65 mm, respectively (provided by the manufacturer), and the total and capillary flow rates were 18.5 and 1.4 ! min- ‘, respectively.
influencing particle behavior was to compare its magnitude with the drag force. A ratio of lift force to drag force, Rid, expressed as F,vt R ,d=-= F drag
1.615 ,~d;(l~l/v)i’~ W 3zpd, W
(5)
could be used to estimate the magnitude of the lift force. Table 1 shows the calculated ratios using different particle sizes. In the calculations, a venturi throat diameter of 2.06 mm and a capillary tube diameter of 1.65 mm were used with a total flow rate of 18.5 emin-’ and a capillary flow rate of 1.4 emin -I. The value of K was calculated to be 4.9 x lo5 s- ‘. The data in this table clearly indicate that the effect of the lift force on the particle deposition could be significant and should be taken into account in estimating internal losses in the small-scale powder disperser. RESULTS
AND
DISCUSSION
Air pressure and flow rates
Figure 2 shows the total and capillary flow rates as functions of the air pressure. Both the flow rates and the difference between them increased with air pressure. At a gauge pressure of 140 in of Hz0 (356 cm of H,O), the total flow rate and capillary flow rate were 18.5 and 1.4 emin- l, respectively. The corresponding air velocities through the venturi throat and capillary tube were calculated to be 240 and 15 ms- ‘, respectively. This difference in velocities resulted in a high shear force that broke up the agglomerates as the powders were dispersed. To achieve a capillary flow rate of 2 emin-’ at a total flow rate of 18.5 emin- ‘, the capillary tube was elevated by 1.5 mm, and the pressure gauge was adjusted to 115 in of Hz0 (292 cm of H20). With the tube elevated, both flow rates were higher at a given pressure compared to those under the normal tube location. E$ect of powder types on generation efJiciency and internal losses
Table 2A shows aerosol production efficiencies and fractional internal losses at different parts of the small-scale powder disperser using the four powders: carbon fibers, talc, alumina particles, and PSL/PSDB microspheres. The results indicated that the efficiency increased when the total flow rate was varied from 5 to 8 emin- ’ and remained relatively constant as long as the flow rate was greater than 8 8 min- ‘. The majority of internal losses occurred at the flow expansion region of the venturi (parts 3 and 4), and the losses decreased when the flow rate was increased from 5 to 8 emin- ‘. At a flow rate of 5.5 / min- ‘, the internal losses of talc and alumina powders in the expansion cones amounted to 5060% of the total materials dispersed. Depending on the powders and flow rates used, the total losses
B. T. Chen et al.
1308
0
I
I
I
I
I
50
100
150
200
250
6
300
Air Pressure 4iiches of 401 Fig. 2. Relationship between air pressure and flow rates in the small-scale powder (0) and (A) represent the total and capillary flow rates, respectively.
disperser.
in other regions contained only 3-S% of the powders dispersed. The data in Table 2A were used to produce Fig. 3 that shows the relationship between the generation efficiency and the flow rate. Among the four powders, the carbon fibers and polystyrene microspheres exhibited higher dispersion efficiencies (60-80%) than the talc and alumina particles (40-60%). Although it was likely that the spherical polystyrene particles would have higher production efficiencies than the irregularly shaped particles, it was interesting to note that the cylinder-like fibers were also dispersed as efficiently. The results indicated that the major axis of the fibers might have been oriented parallel to the airflow through the capillary tube, the venturi throat, and the expansion cone; thus, the fiber length did not affect the dispersion and transport characteristics of fibers in the small-scale powder disperser. The lower efficiency of the talc and alumina particles could be attributed to the differences in particle shape, electrical property, cohesive and adhesive features, as well as the particle size of the powders. It is beyond the scope of the present study to identify the contribution of each parameter; however, the effect of particle size and density on deposition losses as a result of lift force will be discussed later.
Efect of particle size on generation ejjiciency and internal losses
Table 2B shows aerosol generation efficiency and fractional internal losses in the small-scale powder disperser using monodisperse PSL or PSDB microspheres in five different particle sizes. The results indicated that the generation efficiency decreased as the particle size increased for a given flow rate (Fig. 4). Similar to the trend for the irregularly shaped powders as shown in Table 2A and Fig. 3, the efficiency curves for polystyrene particles of 3,6, and 10 pm increased from 5 to 8 c!’ min - ‘. For the PSDB particles of 22 and 97 pm, however, the efficiency remained relatively low and consistent at the flow rates tested. At the total flow rate of 18.5 emin-‘, the generation efficiency of the polystyrene aerosol decreased from 74 to 24% by increasing the particle size from 3 to 97 pm. Internal losses occurred primarily at the flow expansion region (parts 3 and 4) of the venturi tube except for the 97 pm polystyrene microspheres, in which the particles had high inertia and
Evaluation
Table
2. Aerosol
production
efficiency
of the TSI small-scale
powder
among
Powder aerosol
Particle density (g cm _ 3, and size CMD(pm)/GSD
SSPD flow rate ((min-‘)
Aerosol production efficiency q(%)
at i=2
Density: I .83 Fiber width: 3.7411.06 Fiber length: 35.8/2.08
5.46 6.46 8.06 10.5 18.5 22.1
60.0 63.4 73.5 13.4 76.6 73.3
1.9 1.4 0.2 0.1 0.0 0.0
Talc particles
Density: 3.04 Projected area diameter: 2.0511.63
5.49 6.52 8.11 13.1 18.6
41.7 54.5 56.1 56.0 53.7
Alumina (AlsDs) particles
Density: 3.9 Nominal diameter range: 1.2-4.6
5.45 12.1 18.6
Green polystyrene latex particles
Density: Nominal 3.0/1.05
6.12 12.2 18.5
(B) Comparison
between
fibers
Green polystyrene latex spheres, (Density = 1.05 gcm3) Green fluorescent polystyrene divinylbenzene spheres, (Density = 1.05 gcm3)
1.05 diameter:
monodisperse
disperser
Fractional
deposition
at i=4
ati=
at i=6
6.0 7.1 9.2 3.4 1.3 3.4
30.6 26.3 14.8 20.0 18.5 18.7
0.9 1.0 I.7 2.3 2.0 2.5
0.7 0.9 0.7 0.9 I.8 2.2
0.6 0.4 1.0 0.7 1.9
38.2 21.5 24.8 21.1 14.3
13.2 13.8 14.2 18.8 24.1
5.5 2.5 3.1 2.2 4.2
0.6 1.4 0.9 I.2 1.9
33.2 56.6 50.7
0.8 2.1 0.9
37.7 7.5 1.5
20.3 27.9 39.9
5.9 3.8 3.9
2.2 2.1 3.1
64.1 73.9 13.5
1.4 I.3 1.1
14.3 6.4 6.8
14.9 12.6 13.9
4.3 4.9 3.5
1.0 0.9 1.2
polystyrene
spheres
at i=3
losses, WLi (i = 2-6), (%)
with different
particle
Fractional
deposition
sizes
SSPD flow rate (Cmin-‘)
Aerosol production efficiency q(%)
at i=2
ati=
at i=4
3.Ol1.05
6.12 12.2 18.5
64.1 73.9 73.5
1.4 1.3 1.1
14.3 6.4 6.8
14.9 12.6 13.9
4.3 4.9 3.5
1.0 0.9 1.2
6.0/l .09
5.5 12.2 18.6 5.96 12.0 18.5 5.55 12.2 18.6 5.57 12.0 18.5
59.0 55.5 66.4 47.4 54.4 52.0 27.9 21.7 29.6 18.8 25.5 23.8
1.2 0.1 0.0 3.8 1.2 0.6 0.7 0.0 0.0 0.0 0.0 0.0
8.4 4.9 0.4 15.0 2.5 5.3 0.6 0.3 0.2 1.1 0.4 0.7
30.5 36.6 31.1 22.3 32.6 32.8 63.1 62.1 52.4 45.8 18.9 12.4
0.8 1.8 I.5 6.9 6.9 6.3 6.4 5.1 9.1 27.0 46.0 57.0
0.2 1.0 0.7 4.6 2.6 3.0 1.3 4.8 8.8 7.3 9.2 6.2
Fractional
deposition
10.0/1.14
22.011.14
97.OJ1.07
(C) Etfects of capillary
tube location
losses, WLi (i = 2-6) (%) ati=
at i=6
and alignment Aerosol production efficiency n(%)
at i=2
Powder aerosol
Total flow rate (tmin-‘)
Capillary flow rate (Lmin-‘)
PSL powders CMD=3pm
18.5 18.5e
1.4 2.0
13.5 60.8
PSDB powders CMD=6bm
18.5 18.5b 18.5’
1.4 1.4 2.0
PSDB powders CMD = 1Opm
18.5 18.5”
Carbon
18.5 18.5”
fibers
powder
four powders
Particle size CMD(pm)/GSD
Powder aerosol
1309
and fractional deposition losses in the small-scale (SSPD) as a function of flow rate
(A) Comparison
Carbon
disperser
losses, WLi (i = 2-6) (%)
at i=3
at i=4
at i=5
at i=6
1.1 0.9
6.8 9.6
13.9 19.1
3.5 6.7
1.2 3.0
66.4 19.3 46.8
0.0 0.1 0.0
0.4 I.7 2.2
31.1 15.6 44.5
I.5 1.4 3.6
0.7 2.0 2.9
1.4 2.0
52.0 46.6
0.6 0.5
5.3 5.2
32.8 38.4
6.3 5.0
3.0 3.9
1.4 2.0
16.6 17.9
0.0 0.1
I.3 2.7
18.5 16.6
2.0 1.5
I.8 1.2
‘The location of capillary tube in the venturi throat was adjusted to achieve a capillary flow rate of 2.0 / minunder a total flow rate of 18.5 emin-‘, recommended by the manufacturer. “The capillary tube was aligned so that the tube opening was not at the center of the venturi throat.
’
1310
B. T. Chen et al.
00
0
5
10
15
20
25
Flow Rate (L/min) Fig. 3. Dispersion efficiency of four different types of powders in the TSI small-scale powder disperser: (O), (V), (m) and (A) represent carbon fiber, polystyrene spheres, talc, and alumina, respectively.
t?
5 ._
._ i
50
A __-_---
1 0
6
_____----
&I
____-a
5 40
‘Z
s
E g
p.....................y...................67
20
0
00 0
5
_._._._.-.-.
10
_._.“Q
15
20
25
Flow Rate (L/min) Fig. 4. Dispersion efficiency of monodisperse polystyrene microspheres in the TSI small-scale powder disperser. (0) (O), (A), (V) and (0) represent the particles with 3, 6, 10, 22 and 97 Mm, respectively.
deposited significantly inside the exhaust tube (art 5) due to impaction. The losses in the exhaust tube ranged between 27 and 57% of the total dispersed powders for 97 pm microspheres, compared to < 10% for smaller spheres. Along with these experiments, the flow field in the powder disperser were simulated by computer using FIDAP@ software (Fluid Dynamics International, Version 6, Evanston, IL); the result for the total flow rate of 18.5 emin - ’ is shown in Fig. 5. This flow field illustrates the formation of eddies in the small-scale powder disperser that are initiated from the inner lower portion of the expansion cone with up-and-outward circulation to the region of the sampling probe and the surfaces of the expansion cones. Because smaller particles (3 and 6 pm) tended to entrain with the air stream, they had higher upward velocity compared to the larger particles (22 and 97 pm), which as expected lagged behind because of higher inertia. In addition, the smaller particles tended to have smaller lateral migration due to
Evaluation
of the TSI small-scale
powder disperser
1311
Air Out
Air In Fig. 5. Flow streamlines
in the small-scale powder disperser (total and capillary flow rates are 18.5 and 1.4 e min- ‘, respectively).
a smaller lift force (described below). As a result, the small particles followed the upward streamlines were drawn into the sampling probe and became dispersed, whereas large particles were trapped in the eddies and deposited on the wall. EfSects of capillary tube location and alignment on generation ejiciency and internal losses
The location of the capillary tube opening in the venturi nozzle was varied to study the effects of tube location and capillary flow rate on the generation efficiency and internal losses. At the normal tube location, the capillary flow rate was 1.4 emin-‘; at the elevated location, the flow was 2.0 e min - r, as recommended by the manufacturer. Table 2C shows the aerosol generation efficiency and the internal losses at the two different locations and alignments. For polystyrene powders, the results indicated that the normal tube- location achieved a somewhat higher efficiency than the elevated location. In the case of carbon fibers, there was no noticeable difference between the two tube locations. Although a2emin-’ capillary flow rate produced a higher aerosol output than the 1.4 emin-r rate, the generation efficiencies of both rates were very similar. Unless a high powder output was required, there was no reason to elevate the tube for a higher capillary flow. The effects of tube alignment in the venturi throat on the dispersion efficiency and internal losses were examined by comparing the straight position, when the tube was at the center of the venturi throat, with the skewed position, when the tube was off-center. The results from the 6 pm PSDB powders (Table 2C) showed that the generation efficiency was 66% for the straight position and only 19% for the skewed position with an enhanced deposition loss occurring in part 4. This loss indicated that proper alignment of
1312
B. T. Chen et a/
the capillary tube in the venturi throat was critical, and poor alignment could result in a significant loss. EfSect of lift force on generation eficiency and internal losses
As described earlier, the difference between flow velocities.from the venturi throat and capillary tube produced a flow shear to deagglomerate the aerosol particles. However, this shear produced a particle lift (equation (4)) that could affect the particle behavior (movement), thus affecting internal losses of particles and the performance of the small-scale powder disperser. This fact explains the failure of sheath flow from the venturi throat in preventing particle deposition. The aerosol particles aspirated into the expansion cone experienced the lift force and acquired momentum from the force to migrate laterally. The velocity (V,ift) was determined by F,,,B (Hinds, 1982), where B is the particle mobility, (3rcpd,)- ‘, and the lateral migration distance (Slirt) was obtained by
(6) where z is the particle relaxation time and p the air density. From equation (6) it is clear that the lateral migration of the particle or its capability to deviate from its original position toward the wall was proportional to the particle density and particle size. Consequently, internal losses should increase and generation efficiency should decrease with the particle density and size. These predictions were verified by our experimental data. In the present study, the lower density materials, including carbon fibers and polystyrene spheres, had densities of 1.83 and 1.05 g cm- 3, and the higher density materials, including talc and alumina particles, 3.04 and 3.9 g cm- 3. Figure 3 clearly shows that the generation efficiency was higher for the lower-density materials and lower for the higher-density materials; this fact agrees with our analyses. Furthermore, the decrease in generation efficiency with particle size, which is shown in Fig. 4, also agreed with our analysis. In summary, the particle density and size were two parameters in performance characterization. Although a high flow shear could enhance powder deagglomeration, the particle lift force resulting from the shear also increased particle deposition, which played a role in powder dispersion. The influences of flow pattern and lift force on the particle trajectory and the disperser’s performance are complex and need further study. For a given particle material with known size, the effect of operating conditions, such as capillary flow and total flow rates, was negligible when the total flow rate is larger than 8 Lmin- ‘. For the current TSI design, the operating conditions suggested by the manufactory are adequate. However, the results obtained in this study, which shows the major locations of particle losses (zones 3 and 4, Fig. l), can be based to design an improved powder disperse that has minimal particle losses. Acknowledgements-The authors are indebted to Mr Maynard Havlicek of TSI, Inc. for useful discussions. We thank Mr J. A. Stephens for conducting the experiments, and MS P. L. Bradley for editorial assistance. This research was supported by the Office of Health and Environmental Research of the U.S. Department of Energy under Contract Number DE-AC04-76EV01013.
REFERENCES Carpenter, R. L. and Yerkes, K. (1980). Am. Ind. Hyg. Assoc. J. 41, 888-894. Chen, B. T., Cheng, Y. S., Yeh, H. C., Bechtold, W. E. and Finch, G. L. (1991) Am. Ind. Hyg. Assoc. J. 52, 75-80. Cheng, Y. S., Marshall, T. C., Henderson, R. F. and Newton, G. J. (1985) Am. Ind. Hyg. Assoc. J. 46, 449-454.
Evaluation of the TSI small-scale powder disperser
1313
Cheng, Y. S., Barr, E. B. and Yeh, H. C. (1989) Inhal. Toxicol. 1, 365-371. Hinds, W. (1982) Aerosol Technology. Wiley, New York. Koch, W., Lodding, H., Oenning, G. and Muhle, H. (1986) .I. Aerosol Sci. 17, 499-504. Nikula, K. J., Sun, J. D., Barr, E. B., Bechtold, W. E., Haley, P. J., Benson, J. M., Eidson, A. F., Burt, D. G., Dahl, A. R., Henderson, R. F., Chang, I. Y., Mauderly, J. L., Dieter, M. P. and Hobbs, C. H. (1993) Fundam. appl. Toxicof. 21, 127-139. Saffman, P. G. (1965) J. Fluid Mech. 22, 385400.
Timbrell, V., Hyett, A. W. and Skidmore, J. W. (1968) Ann. occup. Hyg. 11, 273-281.
Pergamon
EVALUATION
OF THE TSI SMALL-SCALE
Vol. 26. No. 8. pp. 1303-1313. 1995 Elsevier Science Ltd Printed in Great Entain
POWDER
DISPERSER
B. T. Chen,* H. C. Yeht and B. J. Fan Inhalation
Toxicology
Research Institute, Lovelace Biomedical and Environmental P.O. Box 5890, Albuquerque, NM 87185, U.S.A.
(First
received
14 October
1994;
and
in finalform
3 July
Research Institute,
1995)
Abstract-The TSI small-scale powder disperser was evaluated to study the effects of total flow rate, capillary flow rate, and particle shape, size, and density on dispersion efficiency and internal losses. Dry powders, including carbon fibers, talc, alumina, and polystyrene microspheres, were used to represent materials with different physical properties. Polystyrene microspheres in the size range of 3-97 pm were used to study the effects of particle size on performance. Results indicated that, for the same operating conditions, dispersion efficiencies for carbon fibers and polystyrene spheres were higher than talc and alumina. Dispersion efficiency increased as the total flow rate increased and reached a constant value when the low rate was >8emin-‘. Also the dispersion efficiency decreased as particle sizes increased at a given operating conditions. As for the internal losses in the disperser, particles were deposited primarily at the expansion zone of the venturi tube. Adjustment of the capillary tube location to increase the capillary flow rate did not result in any noticeable change in the efficiency, while poor alignment of the tube resulted in a significant loss as compared with a proper alignment. Along with these data, computer simulations of the flow field and theoretical predictions of the lift force as a result of high flow shear at the venturi throat were successfully used to interpret particle deposition in the TSl small-scale powder disperser.
INTRODUCTION
Various techniques involving air streams, venturi tubes, fluidized beds, and fluid energy jet mills have been used to disperse dry powders (Hinds, 1982). The ideal powder disperser is one that delivers discrete (deagglomerated) particles to the air with a generation efficiency of 100%. A successful dispersion process requires constant feeding, complete deagglomeration and aerosolization, and a minimal loss of powders in the generators. The most common method of dispersion is to feed the powders into a high-velocity air stream. The shear forces in the turbulent air stream disperse the powders and break up the agglomerates. In the case of venturi tubes, a high-velocity air jet blows across a constriction in a pipe to create a suction that draws clumps of powder into the shear flow of air. Shear forces are generated between the high-velocity air jet that passes through the venturi and the low-velocity particle gas stream that leaves the suction pipe. These forces are generally strong enough to break up the agglomerates. For jet mills, the fluid energy is delivered in high-velocity streams, which circulate around a grinding and classifying chamber where turbulence and centrifugal forces deagglomerate particles. Fine particles follow the gas streamlines and exit in the middle of the chamber, while coarse particles are recirculated for further size reduction. For a given dispersion technique, the dispersion characteristics of a powder depend on the powder material, particle size and shape, moisture content, and buildup of charge on the particles during dispersion. It has been found that dry, hydrophobic materials are easier to disperse than humidified, hydrophilic materials (Hinds, 1982). In addition, dispersion efficiency increases with particle size. Several dry powder generators, including the dust-feed disperser, the fluidized bed, the venturi tube, and the jet mills, have been used to generate aerosols with mass concentrations as high as 1 g me3 for inhalation toxicity studies (Carpenter and Yerkes, 1980; Cheng et al., 1985; Koch et al., 1986; Cheng et al., 1989; Nikula et al., 1993; Timbre11 et al., 1968). In
*Current address: GRAM Inc., 8500 Menaul Blvd., NE, Albuquerque, ‘To whom correspondence should be addressed.
NM
87112,
U.S.A.
B. T. Chen et al.
To Filter (7)
Exhaust Tube
DepositionSufiaces Being Analyzed: (1) Turntable (2) CapillaryTube (3) First Expansion Cone (4) Second Expansion Cone
Fig.
1. Schematic
of the TSI small-scale
powder
disperser.
laboratories, however, a powder generator that can produce a limited quantity of test aerosols is more useful for fundamental studies such as calibration of instruments, development of air samplers, and testing of air-cleaning equipment. In this study, we evaluated the TSI small-scale powder disperser. The objectives were to investigate the effects of flow rate, particle size, and particle shape on the powder generation efficiency and internal losses, and to optimize generation efficiency by adjusting operating parameters.
SMALL-SCALE
POWDER
DISPERSER
Figure 1 shows a schematic diagram of the TSI small-scale powder disperser (TSI Inc., Model 3433, St Paul, MN). The powders to be dispersed were brushed over the surface of an abrasive paper on top of a turntable and were removed by venturi aspiration via a capillary delivery tube. This design arrangement enables this device to produce a limited quantity of test aerosols; thus, only a small amount of test material is required. The lower end of the capillary tube was positioned just above the abrasive paper, while the upper end was at the throat of the venturi. A region of low pressure was created by increasing air velocity through the venturi throat which drew particles up through the capillary tube; thus, the tube operated like a vacuum cleaner as the turntable rotated beneath it. The aerosol particles were deagglomerated in the venturi throat. Shear forces were generated as a result of the difference in velocities between the high-velocity air passing through the venturi throat and the low-velocity particle air stream leaving the capillary tube. The air stream passing through the venturi throat tended to break up the agglomerates and also acted as a sheath around the aerosol stream in the throat section, reducing particle losses to the walls of the expansion cone and the sampling probe. Although a total flow rate of 18.5 4 min- ’ was recommended by the manufacturer, different output flow rates were achieved at the
Evaluation of the TSI small-scale powder disperser
1305
sampling probe exit by adjusting the gauge pressure and bypassing the excess air flow through an exhaust tube. In the present study, a flow plug was used to prevent any air flow passing through the exhaust tube of the small-scale powder disperser (Fig. 1). The aerosol passing through the expansion cone was drawn through the sampling probe, and the particles were collected on a filter. The pressure regulator in the small-scale powder disperser was adjusted to achieve different total flow and capillary flow rates. Total flow rates were varied between 5 and 25 ernin-l to study the effects of flow rate on the aerosol generation efficiency. In addition, the location of the capillary tube opening in the venturi nozzle was varied to study its effect on the generation efficiency and internal losses. At a total flow rate of 18.5 emin- l, the capillary tube was situated either at a normal location to achieve a capillary flow rate of 1.4 emin-’ or was elevated to result in a flow rate of 2 emin- ‘, as recommended by the manufacturer. The effects of these two flow rates on dispersion efficiency and internal losses were investigated using powders of various particle sizes. Also, the effects of tube alignment in the venturi throat on the dispersion efficiency and internal losses were examined. POWDER
GENERATION
AND
CHARACTERIZATION
Four types of dry powders, including carbon fibers (Hercules, Inc., Wilmington, DE), talc (commercial grade), alumina (Al,O,) (Duke Scientific, #347, Palo Alto, CA), and polystyrene latex (PSL) microspheres (Duke Scientific), were used to study the effects of particle shape (cylindrical, disc, irregular, and spherical shapes, respectively) on the dispersion efficiency of the small-scale powder disperser. Monodisperse PSL or polystyrene divinylbenzene (PSDB) spheres with fluorescent green dye (excitation maximum at 459 nm, emission maximum at 512 nm) and nominal particle diameters of 3, 6, 10, 22, and 97 ,cm were used to study the effects of particle size on powder production. The densities of carbon fiber, talc, and alumina powders were measured using a Beckman air pycnometer (Beckman Industries, Fullerton, CA). The densities of PSL and PSDB powders are both 1.05 gcmm3, as provided by the manufacturer. Prior to dispersion, lo-15 mg of each powder were weighed in a microbalance (Model C-26, Cahn Instruments, Cerritos, CA), then put on the abrasive paper of the turntable. A polonium-210 radioactive source was passed over the powder on the turntable to minimize any buildup of electrostatic charge in the particles due to brushing powders onto the abrasive paper of the turntable. This will help subsequent suction of powders through the capillary tube. During dispersion, the powder was sucked through the capillary tube, deagglomerated at the venturi throat and expansion cone, then was drawn into the sampling probe, and collected on a fluoropore filter (Millipore Corp., Bedford, MA) placed at the top of the small-scale powder disperser. After each run, the small-scale powder disperser was disassembled into six parts (Fig. 1): the turntable (part l), the capillary tube (part 2), two expansion cones (parts 3 and 4), the exhaust tube (part 5) and the sampling probe (part 6). The internal surface of each disassembled part was rinsed with acetone into a pre-weighed aluminum container. The containers were left in a ventilated hood to complete the evaporation ofthe acetone. Once dried, the material in each container was weighed to determine the mass of particles deposited on the corresponding part of the small-scale powder disperser. The mass on the fluoropore filter (part 7) was also determined gravimetrically to assess the mass balance of the powders before and after generation and the fractional internal losses at different parts of the small-scale powder disperser. The advantages of using acetone in the gravimetric analysis were two-fold: first, no reaction between acetone and the powders could be detected; and, secondly, the acetone evaporated quickly. For the fluorescence-tagged PSL and PSDB powders, both the gravimetric and fluorometric methods were used for mass analysis. The fluorometric method was processed by extracting the samples with ethyl acetate. The samples were obtained by directly rinsing the internal surfaces of each part or rinsing the samples from the gravimetric analysis. Three ml of ethyl acetate wereadded to each sample to extract the fluorescent pigment from
B. T. Chen et al.
1306
the particles; the sample was placed into a vial, and the preparation was sonicated in a water bath for 3 min (Chen et al., 1991). The solution was then poured into a cuvette, and the fluorescence intensity (proportional to the PSL or PSDB mass) was determined using a fluorescence spectrophotometer (Hitachi, Model F-1200, Danbury, CT). The excitation and emission wavelengths were set at 459 and 512 nm, respectively, for maximum fluorescence detection. DATA
ANALYSIS
For each experiment, the masses (or fluorescence intensities) of the samples collected from the small-scale powder disperser and the filter were used to determine the generation efficiency of the small-scale powder disperser and the percentage of internal losses at each disassembled part. The efficiency (q) was computed from:
M,
’= [x7=2 Mi]
’
(1)
where Mi represents the mass (or fluorescence intensity) of the ith sample indicated by subscripts 226 (denoting various disassembled parts) and 7 (the filter) (Fig. 1). Internal loss at the ith part of the small-scale powder disperser was computed from: WL,
z
Mi
[Cl!=,Mil ’
where WLi (i = 2-6) represents the fraction of internal loss at ith part of the small-scale powder disperser. The mass (or fluorescence intensity) on the turntable (i = 1) was not used in the calculation of the generation efficiency and internal loss because the residual amounts of powders on the turntable was arbitrary after each run. The mass could be biased by the location and placement of the powder on the turntable prior to the run, and by the alignment and adjusted distance between the capillary tube and the turntable. However, the mass of powder on the turntable was included to estimate the mass balance in each run. Results indicated that, depending on the powders, 95-100% of the mass balance was achieved in both gravimetric and fluorometric analyses. THEORETICAL
CONSIDERATIONS
At the venturi throat, aerosol particles encountered drag, gravity, and flow shear-induced lift force. The lift force was generated as a result of the difference in velocities between high-velocity air passing through the venturi throat and the low-velocity particle stream exiting the capillary tube. Particle behavior in the small-scale powder disperser was governed by the equation of motion:
where m,, is particle mass; up is particle velocity; and Fdrag, F,iftr and Fg are the drag, lift, and gravitational force, respectively. Because of the vertical orientation of the internal surfaces, the influence of gravitational force on particle deposition was ignored. The lift or Saffman force (Saffman, 1965) was perpendicular to the flow direction. It appeared in shear flow and was expressed by the following equation: F,ift = 1.165 ~d~((KI/V)1’2W,
(4)
where p and v are the air dynamical and kinematical viscosities, respectively; d, is the particle diameter, K the flow shear, and W the relative velocity between particles and flow. The lift force acted on particles to drive them laterally and to deposit on the wall; therefore, this force could be an important deposition mechanism for particles moving through the small-scale powder disperser. A simple way to understand the significance of the lift force in
Evaluation of the TSI small-scale powder disperser
1307
Table 1. The ratios of lift force to drag force for particles of different sizes Particle size (pm)
Lift/drag ratto, R,, (%)a
3 6 10 22 97
8.4 16.7 27.8 61.2 270
’Equation (5) was used to calculate the ratio. In the calculations, the diameters of the venturi throat and capillary tube in the small-scale powder disperser were 2.06 and 1.65 mm, respectively (provided by the manufacturer), and the total and capillary flow rates were 18.5 and 1.4 ! min- ‘, respectively.
influencing particle behavior was to compare its magnitude with the drag force. A ratio of lift force to drag force, Rid, expressed as F,vt R ,d=-= F drag
1.615 ,~d;(l~l/v)i’~ W 3zpd, W
(5)
could be used to estimate the magnitude of the lift force. Table 1 shows the calculated ratios using different particle sizes. In the calculations, a venturi throat diameter of 2.06 mm and a capillary tube diameter of 1.65 mm were used with a total flow rate of 18.5 emin-’ and a capillary flow rate of 1.4 emin -I. The value of K was calculated to be 4.9 x lo5 s- ‘. The data in this table clearly indicate that the effect of the lift force on the particle deposition could be significant and should be taken into account in estimating internal losses in the small-scale powder disperser. RESULTS
AND
DISCUSSION
Air pressure and flow rates
Figure 2 shows the total and capillary flow rates as functions of the air pressure. Both the flow rates and the difference between them increased with air pressure. At a gauge pressure of 140 in of Hz0 (356 cm of H,O), the total flow rate and capillary flow rate were 18.5 and 1.4 emin- l, respectively. The corresponding air velocities through the venturi throat and capillary tube were calculated to be 240 and 15 ms- ‘, respectively. This difference in velocities resulted in a high shear force that broke up the agglomerates as the powders were dispersed. To achieve a capillary flow rate of 2 emin-’ at a total flow rate of 18.5 emin- ‘, the capillary tube was elevated by 1.5 mm, and the pressure gauge was adjusted to 115 in of Hz0 (292 cm of H20). With the tube elevated, both flow rates were higher at a given pressure compared to those under the normal tube location. E$ect of powder types on generation efJiciency and internal losses
Table 2A shows aerosol production efficiencies and fractional internal losses at different parts of the small-scale powder disperser using the four powders: carbon fibers, talc, alumina particles, and PSL/PSDB microspheres. The results indicated that the efficiency increased when the total flow rate was varied from 5 to 8 emin- ’ and remained relatively constant as long as the flow rate was greater than 8 8 min- ‘. The majority of internal losses occurred at the flow expansion region of the venturi (parts 3 and 4), and the losses decreased when the flow rate was increased from 5 to 8 emin- ‘. At a flow rate of 5.5 / min- ‘, the internal losses of talc and alumina powders in the expansion cones amounted to 5060% of the total materials dispersed. Depending on the powders and flow rates used, the total losses
B. T. Chen et al.
1308
0
I
I
I
I
I
50
100
150
200
250
6
300
Air Pressure 4iiches of 401 Fig. 2. Relationship between air pressure and flow rates in the small-scale powder (0) and (A) represent the total and capillary flow rates, respectively.
disperser.
in other regions contained only 3-S% of the powders dispersed. The data in Table 2A were used to produce Fig. 3 that shows the relationship between the generation efficiency and the flow rate. Among the four powders, the carbon fibers and polystyrene microspheres exhibited higher dispersion efficiencies (60-80%) than the talc and alumina particles (40-60%). Although it was likely that the spherical polystyrene particles would have higher production efficiencies than the irregularly shaped particles, it was interesting to note that the cylinder-like fibers were also dispersed as efficiently. The results indicated that the major axis of the fibers might have been oriented parallel to the airflow through the capillary tube, the venturi throat, and the expansion cone; thus, the fiber length did not affect the dispersion and transport characteristics of fibers in the small-scale powder disperser. The lower efficiency of the talc and alumina particles could be attributed to the differences in particle shape, electrical property, cohesive and adhesive features, as well as the particle size of the powders. It is beyond the scope of the present study to identify the contribution of each parameter; however, the effect of particle size and density on deposition losses as a result of lift force will be discussed later.
Efect of particle size on generation ejjiciency and internal losses
Table 2B shows aerosol generation efficiency and fractional internal losses in the small-scale powder disperser using monodisperse PSL or PSDB microspheres in five different particle sizes. The results indicated that the generation efficiency decreased as the particle size increased for a given flow rate (Fig. 4). Similar to the trend for the irregularly shaped powders as shown in Table 2A and Fig. 3, the efficiency curves for polystyrene particles of 3,6, and 10 pm increased from 5 to 8 c!’ min - ‘. For the PSDB particles of 22 and 97 pm, however, the efficiency remained relatively low and consistent at the flow rates tested. At the total flow rate of 18.5 emin-‘, the generation efficiency of the polystyrene aerosol decreased from 74 to 24% by increasing the particle size from 3 to 97 pm. Internal losses occurred primarily at the flow expansion region (parts 3 and 4) of the venturi tube except for the 97 pm polystyrene microspheres, in which the particles had high inertia and
Evaluation
Table
2. Aerosol
production
efficiency
of the TSI small-scale
powder
among
Powder aerosol
Particle density (g cm _ 3, and size CMD(pm)/GSD
SSPD flow rate ((min-‘)
Aerosol production efficiency q(%)
at i=2
Density: I .83 Fiber width: 3.7411.06 Fiber length: 35.8/2.08
5.46 6.46 8.06 10.5 18.5 22.1
60.0 63.4 73.5 13.4 76.6 73.3
1.9 1.4 0.2 0.1 0.0 0.0
Talc particles
Density: 3.04 Projected area diameter: 2.0511.63
5.49 6.52 8.11 13.1 18.6
41.7 54.5 56.1 56.0 53.7
Alumina (AlsDs) particles
Density: 3.9 Nominal diameter range: 1.2-4.6
5.45 12.1 18.6
Green polystyrene latex particles
Density: Nominal 3.0/1.05
6.12 12.2 18.5
(B) Comparison
between
fibers
Green polystyrene latex spheres, (Density = 1.05 gcm3) Green fluorescent polystyrene divinylbenzene spheres, (Density = 1.05 gcm3)
1.05 diameter:
monodisperse
disperser
Fractional
deposition
at i=4
ati=
at i=6
6.0 7.1 9.2 3.4 1.3 3.4
30.6 26.3 14.8 20.0 18.5 18.7
0.9 1.0 I.7 2.3 2.0 2.5
0.7 0.9 0.7 0.9 I.8 2.2
0.6 0.4 1.0 0.7 1.9
38.2 21.5 24.8 21.1 14.3
13.2 13.8 14.2 18.8 24.1
5.5 2.5 3.1 2.2 4.2
0.6 1.4 0.9 I.2 1.9
33.2 56.6 50.7
0.8 2.1 0.9
37.7 7.5 1.5
20.3 27.9 39.9
5.9 3.8 3.9
2.2 2.1 3.1
64.1 73.9 13.5
1.4 I.3 1.1
14.3 6.4 6.8
14.9 12.6 13.9
4.3 4.9 3.5
1.0 0.9 1.2
polystyrene
spheres
at i=3
losses, WLi (i = 2-6), (%)
with different
particle
Fractional
deposition
sizes
SSPD flow rate (Cmin-‘)
Aerosol production efficiency q(%)
at i=2
ati=
at i=4
3.Ol1.05
6.12 12.2 18.5
64.1 73.9 73.5
1.4 1.3 1.1
14.3 6.4 6.8
14.9 12.6 13.9
4.3 4.9 3.5
1.0 0.9 1.2
6.0/l .09
5.5 12.2 18.6 5.96 12.0 18.5 5.55 12.2 18.6 5.57 12.0 18.5
59.0 55.5 66.4 47.4 54.4 52.0 27.9 21.7 29.6 18.8 25.5 23.8
1.2 0.1 0.0 3.8 1.2 0.6 0.7 0.0 0.0 0.0 0.0 0.0
8.4 4.9 0.4 15.0 2.5 5.3 0.6 0.3 0.2 1.1 0.4 0.7
30.5 36.6 31.1 22.3 32.6 32.8 63.1 62.1 52.4 45.8 18.9 12.4
0.8 1.8 I.5 6.9 6.9 6.3 6.4 5.1 9.1 27.0 46.0 57.0
0.2 1.0 0.7 4.6 2.6 3.0 1.3 4.8 8.8 7.3 9.2 6.2
Fractional
deposition
10.0/1.14
22.011.14
97.OJ1.07
(C) Etfects of capillary
tube location
losses, WLi (i = 2-6) (%) ati=
at i=6
and alignment Aerosol production efficiency n(%)
at i=2
Powder aerosol
Total flow rate (tmin-‘)
Capillary flow rate (Lmin-‘)
PSL powders CMD=3pm
18.5 18.5e
1.4 2.0
13.5 60.8
PSDB powders CMD=6bm
18.5 18.5b 18.5’
1.4 1.4 2.0
PSDB powders CMD = 1Opm
18.5 18.5”
Carbon
18.5 18.5”
fibers
powder
four powders
Particle size CMD(pm)/GSD
Powder aerosol
1309
and fractional deposition losses in the small-scale (SSPD) as a function of flow rate
(A) Comparison
Carbon
disperser
losses, WLi (i = 2-6) (%)
at i=3
at i=4
at i=5
at i=6
1.1 0.9
6.8 9.6
13.9 19.1
3.5 6.7
1.2 3.0
66.4 19.3 46.8
0.0 0.1 0.0
0.4 I.7 2.2
31.1 15.6 44.5
I.5 1.4 3.6
0.7 2.0 2.9
1.4 2.0
52.0 46.6
0.6 0.5
5.3 5.2
32.8 38.4
6.3 5.0
3.0 3.9
1.4 2.0
16.6 17.9
0.0 0.1
I.3 2.7
18.5 16.6
2.0 1.5
I.8 1.2
‘The location of capillary tube in the venturi throat was adjusted to achieve a capillary flow rate of 2.0 / minunder a total flow rate of 18.5 emin-‘, recommended by the manufacturer. “The capillary tube was aligned so that the tube opening was not at the center of the venturi throat.
’
1310
B. T. Chen et al.
00
0
5
10
15
20
25
Flow Rate (L/min) Fig. 3. Dispersion efficiency of four different types of powders in the TSI small-scale powder disperser: (O), (V), (m) and (A) represent carbon fiber, polystyrene spheres, talc, and alumina, respectively.
t?
5 ._
._ i
50
A __-_---
1 0
6
_____----
&I
____-a
5 40
‘Z
s
E g
p.....................y...................67
20
0
00 0
5
_._._._.-.-.
10
_._.“Q
15
20
25
Flow Rate (L/min) Fig. 4. Dispersion efficiency of monodisperse polystyrene microspheres in the TSI small-scale powder disperser. (0) (O), (A), (V) and (0) represent the particles with 3, 6, 10, 22 and 97 Mm, respectively.
deposited significantly inside the exhaust tube (art 5) due to impaction. The losses in the exhaust tube ranged between 27 and 57% of the total dispersed powders for 97 pm microspheres, compared to < 10% for smaller spheres. Along with these experiments, the flow field in the powder disperser were simulated by computer using FIDAP@ software (Fluid Dynamics International, Version 6, Evanston, IL); the result for the total flow rate of 18.5 emin - ’ is shown in Fig. 5. This flow field illustrates the formation of eddies in the small-scale powder disperser that are initiated from the inner lower portion of the expansion cone with up-and-outward circulation to the region of the sampling probe and the surfaces of the expansion cones. Because smaller particles (3 and 6 pm) tended to entrain with the air stream, they had higher upward velocity compared to the larger particles (22 and 97 pm), which as expected lagged behind because of higher inertia. In addition, the smaller particles tended to have smaller lateral migration due to
Evaluation
of the TSI small-scale
powder disperser
1311
Air Out
Air In Fig. 5. Flow streamlines
in the small-scale powder disperser (total and capillary flow rates are 18.5 and 1.4 e min- ‘, respectively).
a smaller lift force (described below). As a result, the small particles followed the upward streamlines were drawn into the sampling probe and became dispersed, whereas large particles were trapped in the eddies and deposited on the wall. EfSects of capillary tube location and alignment on generation ejiciency and internal losses
The location of the capillary tube opening in the venturi nozzle was varied to study the effects of tube location and capillary flow rate on the generation efficiency and internal losses. At the normal tube location, the capillary flow rate was 1.4 emin-‘; at the elevated location, the flow was 2.0 e min - r, as recommended by the manufacturer. Table 2C shows the aerosol generation efficiency and the internal losses at the two different locations and alignments. For polystyrene powders, the results indicated that the normal tube- location achieved a somewhat higher efficiency than the elevated location. In the case of carbon fibers, there was no noticeable difference between the two tube locations. Although a2emin-’ capillary flow rate produced a higher aerosol output than the 1.4 emin-r rate, the generation efficiencies of both rates were very similar. Unless a high powder output was required, there was no reason to elevate the tube for a higher capillary flow. The effects of tube alignment in the venturi throat on the dispersion efficiency and internal losses were examined by comparing the straight position, when the tube was at the center of the venturi throat, with the skewed position, when the tube was off-center. The results from the 6 pm PSDB powders (Table 2C) showed that the generation efficiency was 66% for the straight position and only 19% for the skewed position with an enhanced deposition loss occurring in part 4. This loss indicated that proper alignment of
1312
B. T. Chen et a/
the capillary tube in the venturi throat was critical, and poor alignment could result in a significant loss. EfSect of lift force on generation eficiency and internal losses
As described earlier, the difference between flow velocities.from the venturi throat and capillary tube produced a flow shear to deagglomerate the aerosol particles. However, this shear produced a particle lift (equation (4)) that could affect the particle behavior (movement), thus affecting internal losses of particles and the performance of the small-scale powder disperser. This fact explains the failure of sheath flow from the venturi throat in preventing particle deposition. The aerosol particles aspirated into the expansion cone experienced the lift force and acquired momentum from the force to migrate laterally. The velocity (V,ift) was determined by F,,,B (Hinds, 1982), where B is the particle mobility, (3rcpd,)- ‘, and the lateral migration distance (Slirt) was obtained by
(6) where z is the particle relaxation time and p the air density. From equation (6) it is clear that the lateral migration of the particle or its capability to deviate from its original position toward the wall was proportional to the particle density and particle size. Consequently, internal losses should increase and generation efficiency should decrease with the particle density and size. These predictions were verified by our experimental data. In the present study, the lower density materials, including carbon fibers and polystyrene spheres, had densities of 1.83 and 1.05 g cm- 3, and the higher density materials, including talc and alumina particles, 3.04 and 3.9 g cm- 3. Figure 3 clearly shows that the generation efficiency was higher for the lower-density materials and lower for the higher-density materials; this fact agrees with our analyses. Furthermore, the decrease in generation efficiency with particle size, which is shown in Fig. 4, also agreed with our analysis. In summary, the particle density and size were two parameters in performance characterization. Although a high flow shear could enhance powder deagglomeration, the particle lift force resulting from the shear also increased particle deposition, which played a role in powder dispersion. The influences of flow pattern and lift force on the particle trajectory and the disperser’s performance are complex and need further study. For a given particle material with known size, the effect of operating conditions, such as capillary flow and total flow rates, was negligible when the total flow rate is larger than 8 Lmin- ‘. For the current TSI design, the operating conditions suggested by the manufactory are adequate. However, the results obtained in this study, which shows the major locations of particle losses (zones 3 and 4, Fig. l), can be based to design an improved powder disperse that has minimal particle losses. Acknowledgements-The authors are indebted to Mr Maynard Havlicek of TSI, Inc. for useful discussions. We thank Mr J. A. Stephens for conducting the experiments, and MS P. L. Bradley for editorial assistance. This research was supported by the Office of Health and Environmental Research of the U.S. Department of Energy under Contract Number DE-AC04-76EV01013.
REFERENCES Carpenter, R. L. and Yerkes, K. (1980). Am. Ind. Hyg. Assoc. J. 41, 888-894. Chen, B. T., Cheng, Y. S., Yeh, H. C., Bechtold, W. E. and Finch, G. L. (1991) Am. Ind. Hyg. Assoc. J. 52, 75-80. Cheng, Y. S., Marshall, T. C., Henderson, R. F. and Newton, G. J. (1985) Am. Ind. Hyg. Assoc. J. 46, 449-454.
Evaluation of the TSI small-scale powder disperser
1313
Cheng, Y. S., Barr, E. B. and Yeh, H. C. (1989) Inhal. Toxicol. 1, 365-371. Hinds, W. (1982) Aerosol Technology. Wiley, New York. Koch, W., Lodding, H., Oenning, G. and Muhle, H. (1986) .I. Aerosol Sci. 17, 499-504. Nikula, K. J., Sun, J. D., Barr, E. B., Bechtold, W. E., Haley, P. J., Benson, J. M., Eidson, A. F., Burt, D. G., Dahl, A. R., Henderson, R. F., Chang, I. Y., Mauderly, J. L., Dieter, M. P. and Hobbs, C. H. (1993) Fundam. appl. Toxicof. 21, 127-139. Saffman, P. G. (1965) J. Fluid Mech. 22, 385400.
Timbrell, V., Hyett, A. W. and Skidmore, J. W. (1968) Ann. occup. Hyg. 11, 273-281.