Sunteți pe pagina 1din 5

Removal of Si02 Particles with an Ejector Venturi Scrubber

D. A. Marshall
Falconbridge Ltd., Falconbridge, Ontario, Canada and

R. J. Sumner Northern Tel com , Saskatoon , Saskatchewan, Canada


and

C. A. Shook
Chemical Engineering Depart ment, University of Saskatchewan, Sas katoon, Saskatchewan, Canada
INTRODUCTION
Over the last thirty years, public awareness and concern over air pollution has escalated dramatically. To ensure the protection of public health and welfare, legislated emission standards now control mass emissions of particulates. Particulate matter has been defined as any finely divided solid or liquid material emitted into the air Fine particles, less than 5 pm in diameter, have been identified as a significant source of air pollution. Not only do they pose a major health hazard to respiratory systems in themselves, but because of their high surface area, they can act as transport vehicles for other gaseous pollutants. They can remain airborne for extended periods of time and their ability to obstruct light can intensify haze or smog conditions. Venturi scrubbers have been used for the removal of submicron particulate matter found in dusts, fogs, fumes, odours, or smoke from gas streams. Cunic [4] investigated the performance of commercial Venturi scrubbers used to control submicron particulate emissions from a Fludized Catalytic Cracking Unit in a modern refining complex (Exxon Research and Engineering Company). He reported efficiencies in the range of 93-97%. Stadnick and Drehmel (1975) reported on the performance of Venturi scrubber systems on the Shawnee Wet Limestone Scrubbing Test Facility and on the Mystic Power Generating Station (Everett, Mass.). They found efficiencies decreased from 94% for particles of diameter 1.73 pm to 81% at 0.65 pm and 29% at 0.29 pm. In the production of fiber optic cable, very fine Si02particles are produced and these must be removed from the emission gases. At the Northern Telecom Fibre Optics Plant in Saskatoon, Saskatchewan, an ejector Venturi scrubber has been used to control these particulate emissions. The size distribution of particles encountered in the process range from agglomerates larger than 2 pm to very small particles less than 0.2 pm [II]. Although a 95% removal efficiency by the Duo-Flo HydroKinetico Venturi Scrubber was anticipated in the design of the plant, much lower efficiencies have been observed. The purpose of this investigation was to examine the operation of the scrubber in an attempt to identify procedures to improve its performance.

[a.

Principles of Operation
The Venturi scrubber studied in this investigation is of the ejector type with water recirculation. It is shown schematically in Figure 1. This type of scrubber consists of a converging section, a throat section, and a diverging section. The operation of this type of scrubber has been described by Harris [q, but there appear to have been few subsequent reports of performance. The scrubbing liquid is introduced at high velocity in the converging section by a pneumatic spiral spray nozzle. Particulate collection occurs when the high velocity droplets intersect the slower moving particles in the gas stream. The purpose of the spiral nozzle is to provide a combination of axial and tangential velocities to the scrubbing liquid during

28

February, 1995

Environmental Progress (Vol. 14, No. 1)

Air to Pneumatic Nozzle

Water to

100,

O A

7 Pneumatic 7 Nozzle

Recirculating Water

Separator

'"i
0

, , , , ,,,

, ,A A , b,A,A, ,Inlet , , 0 0 0 0 0 Outlet

, , , , , , ,,
100

1 10. Particle Diameter (microns)

FIGURE 1. Duo-Flo HydroKineticQVenturi Scrubber.


atomization. Adjusting the angle of the spiral, produces a conical spray which fills the Venturi throat. The size and velocity of the liquid droplets directly affect overall scrubbing efficiency [8]. The liquid flow rate ca.n be controlled by the size of the orifice in the nozzle. The droplet size can be varied by changing the atomizing air pressure. The size and velocity of the droplets should be optimized to maximize collection efficiencies and separation properties 181. This optimum droplet size and velocity is influenced by the density, shape, and velocity of the particles. Particle collection mechanisms by drops were studied by Johnstone et al. [ 9 ] . They determined that the main collection mechanism for particles greater than about 0.3 pm was inertial impaction with negligible electrostatic, induction, and diffusion effects. This conclusion is in agreement with the subsequent literature [ I , 3, 5 , 8, Z5]. For particles less than 0.1 pm, diffusion becomes the predominant collection mechanism. The overall contribution of any one mechanism depends on the particle and droplet sizes and their relative velocity.

FIGURE 2. Particle size distributions at the scrubber inlet and outlet.

Experimental Procedures
Scrubber performance was evaluated by measuring the inlet and outlet particulate concentrations. Since elbows or bends upstream of the sampling site can cause swirling flow or irregular flow profiles, both the inlet and the outlet sampling ports were preceded by approximately 10 metres (30 pipe diameters) of straight pipe. Gas velocity distributions were measured at both locations using a pitot-static tube (United Sensor, Type PAE-T, 6.35 mm 0.d.) and a micromanometer (Flow Corporation, Model MM-3). These distributions were smooth and of the shape expected for turbulent flow without irregularities or secondary flow [IZ]. Thus the concentration distributions could be assumed to be uniform over the pipe crosssection at both locations. The particle density was 2265 kg/m3. A microscopic examination showed the particles to be spheres with a few aggregates of relatively large diameter. Particle size distributions had been measured previously [I41 at the scrubber inlet and outlet when the scrubber was operating at an overall efficiency of 88%. These results are shown in Figure 2. The uncertainty in the measurements is actually somewhat greater than the difference between the inlet and outlet results, showing that there was little tendency to remove the larger particles preferentially. Preferential withdrawal of particles with respect to size occurs if there is a velocity difference between the gas stream and the gas entering the sample probe. To ensure the collection

of unbiased samples, an isokinetic sampling procedure based on the EPA Method 5 Standard for compliance testing was developed [II]. A schematic drawing of the sampling train is shown in Figure 3. The gas velocities for isokinetic sampling were established by the aforementioned pitot tube traverses. The manufacturing process was adjusted to produce a fixed solids loading of soot particles at a concentration reflecting typical operating conditions. A DuoSeal vacuum pump (Model 1405, The Welch Scientific Co.) withdrew the sample gas through an 8.0 rnm (i.d.) stainless steel nozzle and 0.66 m probe assembly (Apex Instruments). The probe assembly consisted of a stainless steel sheath with a quartz liner. The particulate matter was collected with a glass fibre filter (Whatcom) in a 10 cm glass filter assembly (Apex Instruments). The cleaned gas was then passed through an impinger (Apex Instruments) to remove any particulate passing through the filter. A drying tube filled with silica gel was finally used to remove any moisture from the gas stream. The ball valve was used as an on/off valve in sampling. The sampling velocity was controlled by a metering valve with a Vernier handle (Nupro "L" Series). The vacuum gauge was used to monitor the pressure in the vacuum line and to check for any air leaks in the system. The orifice downstream of the vacuum pump and the micromanometer was used to fix the sampling velocity. The volume of gas sampled was calculated from this velocity, the cross-

li
Stack Wall

sampling Probe

Finer
Air Dying Tube

Ball Valve

Micromanometer Metering

Vacuum Gauge Vacuum Pump

orifice

FIGURE 3. Particulate Sampling Train. February, 1995


29

Environmental Progress (Vol. 14, No. 1)

Table 1 Results from the 24-' ExDerimental Desian Air Water Recirc. Orifice Collection NTU Flowrate Flowrate Water Diameter Efficiency (Yo) (m3/min)* (L/min) (L/min) (mm) 2.475 2.475 1.484 1.484 I .294 1.294 1.286 1.286 2.475 2.475 1.295 1.295 1.286 1.286 1.274 1.274 0.665 0.665 5.7 5.7 5.7 5.7 5.7 5.7 5.7 5.7 5.7 7.6 7.6 7.6 7.6 7.6 7.6 7.6 1.9 I .9 18.9 18.9 37.9 37.9 37.9 37.9 18.9 18.9 37.9 37.9 18.9 18.9 37.9 37.9 18.9 18.9 18.9 18.9 2.45 2.45 1.3 1.3 2.45 2.45 1.3 1.3 2.45 2.45 1.3 1.3 1.3 1.3 2.45 2.45 0.8 0.8 85.5 85.8 68.8 69.1 66.7 64.6 62.5 62.9 86.6 87.0 72.3 71.4 71.0 66.4 67.2 68.0 47.9 51.2 1.93 1.95 1.16 1.17 1.10 1.04 .98 .99 2.01 2.04 1.29 1.25 1.24 1.09 1.11 1.14 0.65 0.72
i
r
v)

2'2 2.0

1
o
0 0

1.6

1.4

k
0

1.2 1.0

#A

$0.8

0.6

4 4

ooooo Waterflow lo Nozzle = A A A A A Waterflow to Nozzle =


I I I I

7.6 L/rnin 5.7 L/min

444P4 Waterflow to Nozzle = 1.9 Lfrnin


0.4
O ?
I I I I I I

a I

-81

I I I

0.5

1.0

1.5

' 2.0

I I

'

I 1 I 1

2.5

3.0

Airflow to Nozzle (m'/rnin)

FIGURE 4. Effect of air flowrate to the pneumatic nozzle on the number of transfer units at various nozzle water flowrates.

'Standard cubic meters

NTU= -In

outlet concentration inlet concentration

sectional area of the probe and the sampling time. The ratio of the sampling velocity to the stack velocity provided a measure of isokineticity.

Operating Parameters
The four Venturi scrubber operating variables considered were the air and water flowrates to the pneumatic nozzles, the recirculating water rate, and the orifice diameter on the air side of the pneumatic nozzle. The operating conditions were: X,-Air flowrate to pneumatic nozzle i) 1.274 m3/min (45 SCFM) ii) 2.475 m3/min (87 SCFM) X2-Water flowrate to pneumatic nozzle i) 5.7 L/min (1.5 GPM) ii) 7.6 L/min (2.0 GPM) X,-Recirculating water flowrate i) 18.9 L/min (5 GPM) ii) 37.9 L/min (10 GPM) &-Orifice size in air side of pneumatic nozzle i) 2.45 mm ii) 1.3 mm The pneumatic nozzle exit diameter was 8.7 mm. The Venturi scrubber dimensions were: upstream diameter 300 mm, throat diameter 140 mm, throat length 330 mm.

The isokineticity of these measurements ranged between 96% and 102%. With the exception of one set of data the reproducibility was within 5% (some were better than 2%). The uncertainty in the number of transfer units is k0.06. The two-level factorial design was used to examine the individual effects of each operating variable and possible interactions between the variables. Because each variable was tested at two values, only a first-order term can be obtained for each variable. Although this type of design could not provide a definitive model of process behavior, it was a very useful screening tool to detect the important operating variables. A linear empirical relationship correlating the effects of the variables on the Venturi scrubber efficiency was obtained from these results using a least squares estimation. From the regression analysis, it was determined that the interaction between operating variables was insignificant. The effects of the recirculating water flowrate and the orifice diameter were also negligible. Omitting these variables yielded the following empirical model:

NTU= 1.525 + 0.457X; + 0.059X;


where
X,' = (X,- 1.88)/0.60; Airflow to Nozzle, m3/min and Xi = (X, - 6.65)/0.95; Waterflow to Nozzle, L/min.

(2)

The analysis shows no significant lack of fit to this model (F-test; R2=0.9906). Confidence intervals (95%) for the parameter estimates are: 1.525 k0.028

Experimental Results
In the first series of tests, a 24-' fractional factorial experimental design was used to determine the effect of the operating variables. Eight experiments were performed at the upper and lower levels of the operating conditions. Each experiment was replicated once, for a total of 16 experiments. Two additional experiments were performed using an airside nozzle insert with 0.8 mm orifice openings. A summary of the results is shown in Table 1. The scrubber performance is described in terms of percent efficiency as well as by the number of transfer units (NTU), defined as:
30

0.457&0.031 0.059&0.025

+ 1 and - 1 is so that the magnitudes of the parameter estimates


reflect their importance in the final equation. The sign of the parameter estimate indicates whether the corresponding operating variable should be increased or decreased to improve performance. From the magnitude of the parameter estimates for the air and water flowrates, it can be concluded that the air flowrate Environmental Progress (Vol. 14, No. 1)

These operating variables have been scaled to range between

February, 1995

2.2
2.0

3
2.0 2'2
+ . ._
(0

'c 1.8
+ .

ffl

1.8

1
Waterflow (previous experimenlal l o nozzle CON/ data)

3
L L

ar
ffl

1.6
1.4

Best Fit: Y = 0.7887 X - 0.0002

"

t
~

F
0

1.2

i
b

; 1.0

8 P

Z
0.6
AAAAA

(new ewerimental dala) Best F d : Y = 0.7559 X - 0.019

0.8 Air Flowrote to Nozzle = 2.48 rn'/min

O b b V O Air Flowrale to Nozzle = 1.48 rn'/rnin


0.0
1 .o

2.0

5.0

4.0

5.0

0 . 0

0.5

1.0

1.5

2.0

2.5

3.0

Orifice Diameter (mrn)

Airflow to Nozzle (rns/min)

FIGURE 5. Effect of the orifice diameters in the air side of the pneumatic nozzle on the number of transfer units at various pneumatic nozzle air flowrates.

FIGURE 6. Linear relationship between the air flowrate to the pneumatic nozzle and the number of transfer units.

n,, is the number of particles per unit volume,


is the most important operating variable; the effect of the water flowrate is small, but significant. Because both parameter estimates are positive, the output (NTU) is highest when both of these variables are maximized within the constraints of plant operation. The number of transfer units is plotted against the air flowrate in Figure 4.This graph shows how the number of transfer units increases with increasing air flowrate. The relationship appears to be linear, although more data would be required to verify this. Figure 5 shows that there was no effect of orifice diameter on the number of transfer units. An additional series of experiments was performed to examine further the effect of the most important operating variable, the air flowrate to the pneumatic nozzle, on the efficiency of the Venturi scrubber. In these experiments, the other operating variables were held constant: water flowrate at 5.7 L/ min, recirculating water flowrate at 18.9 L/min, and orifice plate diameters at 2.45 mm. The air flow rate was tested at 2.48, 1.84, 1.42, and 1.27 m3/min with replicates at the first three conditions. A summary of these experimental results is given by the solid line in Figure 6. Again, there is a linear relationship between the air flowrate and the number of transfer units. The dashed line shows the locus of the previous results for the same water flowrate. Although the two lines are very similar, there is a slight variation. This could be explained by process noise in experiments which were performed several weeks apart. It would have been preferable to include higher air flowrates in these experiments, since according to Licht (1980), at some point an increase in air flowrate will result in a decrease in scrubber efficiency when the water droplet size falls below its optimum value. However, the air flow was limited by the capacity of the compressor. Because it was not possible to increase the air flow rate further, this optimum could not be determined. The minimal effect of the water flowrate indicates that the throat coverage by the jet was satisfactory. The nearly linear relationship between NTU and airflow to the nozzle is consistent with Harris' [2]correlation. The material balance equation governing particle capture by 1 0 1 is: droplets I

n, is the number of droplets per unit volume, v, and u, are the gas and droplet velocities, D is the droplet diameter, q T is the capture efficiency, and x is the distance downstream.
Equation (3) assumes the particle concentration in the gas phase is uniform in the radial direction. Because of entrainment and expansion of the jet, u, will decrease as x increases and this wit1 cause the droplets to decelerate. The quantities in the brackets in equation (3) are each likely to vary with position. The volume fraction of liquid droplets in the expanding jet, $ , is related to the droplet diameter by:

(4)
so that equation (3) can be written:

(3)

Since the droplets and the air will leave the nozzle with nearly identical velocities, the relative velocity in the expanding, entraining jet will be determined primarily by the nozzle exit velocity, u,,". This velocity can be calculated from the liquid and air flowrates assuming adiabatic and reversible flow for the air and neglecting slip between the air and the droplets. The calculations show that although 4, at the nozzle varied threefold, between 0.013 and 0.04 as the liquid flowrate varied, the product varied much less, between 1.59 and 2.14 m/s, during the experiments. This calculation provides a partial explanation for the fact that the liquid flowrate had little effect on scrubber performance: the term 4,(v,- v,) is relatively insensitive to liquid flowrate where particle capture takes place. The effect of the air flowrate to the nozzle on scrubber performance is consistent with the presence of the droplet diameter in equation (S), since it is known that an inverse relationship exists [12]. Since the entrained gas flowrate was constant in the experiments, these experiments suggest that an approximate performance equation for the scrubber is:
,, & J

where

Environmental Progress (Vol. 14, No. 1)

February, 1995

31

where G is the gas flowrate, van is the nozzle exit velocity and $ is a dimensional factor, proportional to the mean capture efficiency, for a given scrubber. Changes in particle size distribution between the inlet and the outlet of the scrubber can be computed because v,,, G and the other factors in $ remain constant. For particle species i, the penetration is:

us = gas velocity, m/s uI = droplet velocity, m/s

x = distance measured in the downstream direction X i = Venturi scrubber operating variables, i = 1 - 4 X, = Scaled Venturi scrubber operating variables, i=1-4 pp = gas viscosity, poise qr = individual droplet capture efficiency 4, = volume fraction liquid pp = particle density, g/cm3 $ = dimensional factor for given scrubber

According to Calvert 121, the capture efficiency is related to the inertial impaction parameter K by:

LITERATURE CITED
?IT=

[A]:
(9)

where

where pp and d, are the particle density and diameter, C is the Cunningham correction for small particles and pg is the gas velocity. Over the course of the experiments, D varied as u,, varied, but this change in droplet size would lead to changes in the relative velocity so that ( u I- v E ) / D would vary less than D . It seems likely that K was always substantially greater than 0.7, so that qr and yr, were always high. This would account for the minor difference between the inlet and outlet size distributions shown in Figure 2.

CONCLUSIONS
The most important operating parameter in the performance of this scrubber is the air flowrate to the pneumatic nozzle. The relationship between the air flowrate to the pneumatic nozzle and the number of transfer units appears to be linear. The best performance, approximately 85-87% removal, was observed at the maximum attainable air flowrate. It seems probable that the efficiency of the scrubber is controlled by the size of the droplets and by the relative velocity between the droplets and the particles.

NOTATION
C = Cunningham correction factor D = droplet diameter, m G = volumetric flowrate of gas, m3/min K = inertial impact parameter, dimensionless np = number of particles per unit volume n , = number of droplets per unit volume NTU = number of transfer units v,, = nozzle exit velocity, m/s

1. Behie, S. W. and J. M. Beeckmans, On the Efficiency of a Venturi Scrubber, Can. J. Chem. Eng., 51, pp. 430433 (Aug. 1973). 2. Calvert, S., J. Goldschmid, D. Leith, and D. Mehta, Scrubber Handbook, NTIS, PB 213-016 (1972). 3. Calvert, S., Venturi and Other Atomizing Scrubbers Efficiency and Pressure Drop, AZChE J., 16(3), pp. 392396 (May, 1970). 4. Cunic, J . D., M. G . Bienstock, and A.M. Edelman, Wet Gas Scrubbing: State of the Art in FCCU Emission Control, Env. Prog., 6(4), pp. 267-273 (1987). 5. Goel, K. C. and K. G. T. Hollands, Optimum Design of Venturi Scrubbers,Atm. Env., 11,pp. 837-845 (1977). 6. Harrington, R. E., Fine Particulates-The Misunderstood Air Pollutant, J. Air Pol. C. A . , 24(10), pp. 927929 (Oct., 1974). 7. Harris, L. S. and G. R. Haun, The Ejector Venturi Scrubber, Chem. Eng. Prog., 60(5), pp. 100-103 (May, 1964). 8. Harris, L. S., Fume Scrubbing with the Ejector Venturi System, Chem. Eng. Prog., 62(4), pp. 55-59 (April, 1966). 9. Johnstone, H. F., R. B. Field, and M. C. Tassler, Gas Absorption and Aerosol Collection in a Venturi Atomizer, Ind. and Chem. Eng., 46(8), pp. 1601-1608 (Aug., 1954). 10. Licht, W., Air Pollution Control Engineering, Basic Calculations for Particulate Collection, Dekkar, pp. 343368 (1980). 11. Marshall, D. A., The Operation of a Venturi Scrubber, Masters Thesis, Department of Chemical Engineering, University of Saskatchewan (1993). 12. Nukiyama, S., and Y. Tanasawa, Trans. SOC. Mech. Engrs. (Japan),4, p. 86 (1938). 13. Statnick, R. M. and D. C. Drehmel, Fine ParticleControl Using Sulfur Oxide Scrubbers, J . Air Pol. C. A . , 25(6), pp. 605-609 (June, 1975). 14. Wallace, K., R. Bergrand, R. Jahren, and R. Bouvier, Source Testing of Northern Telecoms CVD and 110 Scrubber Systems, Saskatchewan Research Council, E-2800-7-(2-91 (July, 1991). 15. Yung, S. C., S. Calvert, and H. F. Barbarika, Venturi Scrubber Performance Model, Env. Sci. Tech., 12(4), pp. 456-459 (April, 1978).

32

February, 1995

Environmental Progress (Vol. 14, No. 1)

S-ar putea să vă placă și