You are on page 1of 6

Aerosol Science and Technology

ISSN: 0278-6826 (Print) 1521-7388 (Online) Journal homepage: http://www.tandfonline.com/loi/uast20

Venturi Scrubbers: Pressure Loss and Regain


David Leith , Douglas W. Cooper & Stephen N. Rudnick
To cite this article: David Leith , Douglas W. Cooper & Stephen N. Rudnick (1985) Venturi
Scrubbers: Pressure Loss and Regain, Aerosol Science and Technology, 4:2, 239-243, DOI:
10.1080/02786828508959052
To link to this article: http://dx.doi.org/10.1080/02786828508959052

Published online: 10 Mar 2011.

Submit your article to this journal

Article views: 1159

View related articles

Citing articles: 9 View citing articles

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


http://www.tandfonline.com/action/journalInformation?journalCode=uast20
Download by: [191.113.127.124]

Date: 13 July 2016, At: 13:47

Venturi Scrubbers: Pressure Loss and Regain


David Leith, Douglas W. Cooper, and Stephen N. R~idnick

Downloaded by [191.113.127.124] at 13:47 13 July 2016

Physical Sciences and Engineering, Harvard School of Public Health, Boston, M A 02115

Pressure loss in the throat of a venturi scrubber is offset


by regain in the diverging section. Available models for
venturi scrubber pressure drop that consider regain are
difficult to use because they require numerical integration
or reference to published figures. This note presents an
expression for pressure regain in the diverging section
that, with an expression for pressure loss in the throat,
gives a single, simple equation for overall pressure drop

in a venturi xmbber. Predictions made with this equation, and with models from the literature, are compared
with data for three venturi scrubbers operated at throat
velocitier from 20 to 160 m/s at liquid-to-gas ratios
Predictions from the new equafrom 0.2 to 2.3 x
tion are at least as good as predictions of the complex
models that also consider regain, and about as easy to
obtain as analytical models that ignore regain.

INTRODUCTION

velocity between gas and droplet causes a


frictional drag force that accelerates the
slower droplet. Both gas and droplet experience drag that is equal and opposite in direction. The resultant work on the gas equals
the work done to accelerate the droplet. The
work on the droplet increases its kinetic energy, whereas the work on the gas is lost as
heat.
Thus, the total work required to accelerate
a droplet in a scrubber is the sum of the
work done on the gas plus that done on the
droplet. Because the work on the gas equals
that on the droplet, total work is m(u;, uj,), twice the droplet's increase in kinetic
energy. The effect of this work is to decrease
static pressure through the portion of the
scrubber where droplets accelerate.
In the diverging section of the venturi
scrubber, gas slows. If droplet velocity becomes higher than gas velocity, droplets decelerate. Some of the kinetic energy a droplet
loses as it decelerates will increase gas potential energy, causing static pressure regain.
The remaining kinetic energy will be lost as
heat. If the gas becomes stationary, as would
occur on expansion to the atmosphere, all
the droplet kinetic energy dissipates as heat.

A venturi scrubber consists of a converging


duct, a straight duct called the throat, and a
diverging duct. Liquid enters in or upstream
of the throat and atomizes into droplets that
capture particles by impaction. Venturi
scrubbers are compact and comparatively inexpensive to construct, but can be expensive
to operate because they require hlgh gasphase pressure drop to achieve high collection efficiency for small particles. Pressure
drop is due to energy consumed as friction
from gas flow, a loss that would occur even if
no water were injected, and as energy to
accelerate the droplets.
Although correct expressions for pressure
drop can be derived from conservation of
momentum equations, an analysis of work
done on the droplets and on the gas as the
droplets accelerate is clearer. If a droplet of
mass rn accelerates from velocity u,, to v,,,
using a frictionless connection between the
power source and the droplet, the necessary
energy is the droplet's gain in kinetic energy,
rn(u;, - uj,)/2. A scrubber, however, does
not provide a frictionless connection between
power source and droplets. The relative

Acrosol Scicncc and

Technology 4:239-243 (1985)

240

THEORY

Virtually all models for pressure drop in a


venturi scrubber assume that:

Downloaded by [191.113.127.124] at 13:47 13 July 2016

1. Gas is incompressible and isothermal;


2. Flow of gas and liquid is one dimensional;
3. No liquid fed to the scrubber evaporates
or flows along the scrubber walls;
4. All droplets are the same size.
Boll (1973) considered the effects of
changing momentum of gas and liquid and
the effect of energy loss due to gas flow on
pressure drop in a venturi scrubber. T h s
analysis led to

Here, p, and p, are the densities of the gas


and liquid, u, and u, are the velocities of the
gas and droplets, Q, and Q, are the volumetric flow rates of gas and liquid, f is a
friction factor, D is the hydraulic diameter
of the duct, and z is axial position.
With the further assumptions that:
5. Pressure loss due to gas flow is negligible;
and
6. Liquid is injected at the throat entrance
with no velocity in the axial direction,
Eq. (1) reduces to

If the further assumption is made that:


7. Gas velocity is constant,
and if puG,is the maximum velocity attained
by droplets in the throat where u,, is gas
velocity in the throat, integration of Eq. (2)
yields

Equation (3) is the starting point for several


theories for pressure drop in a venturi

Leith, Cooper, and Rudnick

scrubber. If the droplets accelerate to the gas


throat velocity, then /3= 1 as assumed by
Calvert (1968, 1970). Calvert (1977) later
stated that P = 0.85 except at low liquid
rates where pressure losses due to gas flow
are significant.
Hollands and Goel (1975) developed an
analytical expression for P, also given later
by Yung et al. (1977):

Hollands and Goel (1975) gave an equation


for x that simplifies to Eq. (5) if liquid inlet
velocity in the axial direction is zero:

where I is length of the scrubber throat, CD


is the droplet drag coefficient at the point of
liquid injection, and d, is droplet diameter.
CD can be evaluated from the droplet
Reynolds number at the axial position where
liquid is injected using the standard curve as
described, for example, by the equation of
Dickenson and Marshall (1968). Droplet diameter, d,, can be determined using the
equation from Nukiyama and Tanasawa
(1940).
Boll's model for pressure drop, Eq. (I),
must be integrated numerically for each
scrubber geometry and operating condition
because no analytical solution exists. This
integration accounts for static pressure loss
caused by droplet acceleration as well as
static pressure regain caused by droplet deceleration. Although t h s is probably the most
complete model to describe venturi pressure
drop, it is difficult to use because the equation must be solved numerically. Hollands
and Goel (1975) present generalized solutions to Eq. (1) through six plots, three for
rectangular and three for circular venturi
scrubbers. These plots make the numerical
integration unnecessary, but their use is still
cumbersome.
The model by Calvert (1968, 1970) and
the simplified model by Hollands and Goel
(1975), later described by Yung et al. (1977),

24 1

Downloaded by [191.113.127.124] at 13:47 13 July 2016

Venturi Scrubbers: Pressure Loss and Regain

have analytical solutions and are easy to use;


however, these models consider the venturi
scrubber to be a straight duct of constant
diameter. They do not consider static pressure regain caused by droplet deceleration
downstream of the throat. When regain is
important, these models will overestimate
pressure drop if pressure loss due to gas flow
is negligible.
An analytical model can be developed for
static pressure regain that can be used with
Eqs. (3), (4), and (5) for static pressure loss,
to give a more accurate estimate of pressure
drop in a venturi scrubber. This model, also
based on Eq. (2), requires the additional
assumptions :

gas velocity is half that in the throat (uGf/uGt


= 0.5). In this case, regain in the diverging
section "saves" 25% of the pressure loss
across the throat. On the other hand, if
droplets do not reach the gas velocity in the
throat ( p -,0 ) and if the scrubber has little
or no diverging section (u,,/uG, + I ) , droplets will continue to accelerate after leaving
the throat, and pressure drop across the
throat and diverging section together will be
higher than the pressure loss across the throat
alone. In general, Eq. (6) will have a negative
value, and regain in the diverging section will
"save" some of the pressure loss across the
throat only if P > (uGf/uGt).

8. Gas expands abruptly in the venturi diverging section; that is, uG= u, where
uGf is the gas velocity in the venturi outlet

EXPERIMENTS

duct.
9. Droplets decelerate and ultimately reach
the gas velocity in the outlet duct, ud = u,.

Assumption (8) is approximately true if the


diverging section is short. Assumption (9) is
approximately true if the venturi outlet duct
is long.
With these assumptions, Eq. (2) yields

Total pressure drop is the sum of loss and


regain from Eqs. ( 3 ) and (6):

where j? from Eq. (4) can be used.


The ratio of pressure loss with regain predicted from Eqs. (4) and (7) to pressure loss
without regain predicted from Eqs. (3) and
(4) has a minimum of 0.75 if droplets accelerate fully in the throat ( P = 1 ) and the final

Three venturi scrubbers were used; each had


a circular cross section and was machined
from transparent acrylic plastic. All had inlet
and outlet diameters of 127 mm, a converging section with half-angle of 10SO,and a
diverging section with half-angle of 4.8". The
diameters of the small, medium, and large
throats were 32, 54, and 76 mm; their respective throat lengths were 32, 51, and 76 mm.
Water entered the venturi scrubber in three
ways. In the "inward" method, water entered
through an annular slot in the duct wall; this
slot was 0.6 mm wide and 20.9 mm above the
entrance of the converging section. In the
"downward" method, water was introduced
through a downward-facing slot that protruded 1.0 mm into the duct 15 mm above
the entrance of the converging section. In the
"center7' method, the water was introduced
downward through a 14.2-mm pipe, located
along the axis of the duct with its outlet at
the entrance of the converging section.
Pressure drop across the venturi was measured with a manometer. The upstream pressure tap was 264 mm before the entrance to
the venturi converging section; the downstream tap was 387 mm after the elbow that
connected the vertical venturi to the horizontal inlet to the 'cyclone demister. The downstream pressure tap was on the top of the

242

Leith, Cooper, a n d Rudnick

duct and designed to prevent the manometer


line from filling with water from the walls of
the ductwork.
Pressure drops were measured over a range
of operating conditions. Gas velocity in the
throat varied from 21 to 180 m/s; QJQ,
varied from 0.19 x lop3 to 4.25 x lop3. The
three scrubbers and water injection methods
described above were used. In all, 117 measurements of pressure drop were made.

TABLE 1. Results of Tests on Logarithm


of Observed a n d Predicted Pressure Drops
d,

Model
Calvert
Simplified Hollands
and Goel
Boll
Equation (7)

Geometric std. dev. for d,

0.45

1.47

0.83
0.89
0.91

1.41
1.40
1.40

termined for the 117 measurements from

Downloaded by [191.113.127.124] at 13:47 13 July 2016

RESULTS AND DISCUSSION

For the same conditions examined experimentally, pressure drop was calculated: (a)
using the Calvert model, Eq. (3) with P =
0.85; (b) using the simplified Hollands and
Goel model (identical to the model published
by Yung et al.), Eq. (3) with P given by Eq.
(4); (c) the Boll model, Eq. (1); and (d) the
model developed here for pressure loss and
regain, Eq. (7), with P given by Eq. (4).
Figure 1 is a plot of the data against the
pressure drop predicted using Eq. (7).
For each pressure drop theory discussed
here, the geometric mean difference between
observations and predictions, d,, was de-

FIGURE 1. Measured pressure d r o p versus predicted pressure d r o p with regain included, using

Eq. (7).

PREDICTED

PRESSURE

DROP,

kPa

log d,

(log 'observed

- log A ~ ~ r e d i c t e d )
117

The geometric mean difference was used to


give equal weight to the relative errors found
over the range of pressure drops investigated.
Values of d, for each model are given in
Table 1. Also given are the geometric standard deviations for the d, values.
The geometric mean differences between
observed and predicted pressure drops, d,,
are less than one for all models tested; that
is, all models tended to predict pressure drops
higher than those observed. The Calvert
model with P = 0.85 overestimated pressure
drop most; Eq. (7), the least. The geometric
standard deviations for the d, values were a
little higher for the Calvert model than for
the other models.
As a practical matter, predictions using
the Boll model and Eq. (7) gave comparable
results; however, the Boll model, Eq. (I), is
cumbersome to use whereas Eq. (7) is much
simpler. Both the Boll model and Eq. (7)
gave pressure drop predictions somewhat
better than those from the simplified Hollands and Goel model, and considerably better than those from the Calvert model.
Although the research described in this note has been
funded in part by the United States Environmental
Protection Agency under grant R809003010 to Harvard
University, it has not been subjected to the Agency's
required peer and administrative review. This note does

Venturi Scrubbers: Pressure Loss and Regain

not necessarily reflect the views of the Agency, and no


official endorsement should be inferred. We acknowledge gratefully the assistance of Kenneth P. Martin and
Stephen Ferguson, who carried out much of the experimental work reported here, and of John Koehler of
Harvard University and Thomas J. Overcamp of
Clcmson University, who provided useful comments.

REFERENCES

Downloaded by [191.113.127.124] at 13:47 13 July 2016

Roll, R. H. (1973). Znd. Eng. Chem. Fundum. 12:40-50.


Calvert, S. (1968). In Azr Pollutlon. (A. C. Stem, ed.),
2nd cd., Volumc 111, Academic Press, New York:
474-475.
Calvert, S. (1970). AIChE J . 16:392-396.

Calvcrt, S. (1977). In Alr Pollutlon. (A. C . Stern, ed.),


3rd ed., Volume IV, Academic Press, New York:
271
Dickenson, D. R., and Marshall, W. R. (1968). AIChE
J. 14:541-552.
Hollands, K. G. T., and Goel, K. C. (1975). Ind. Eng.
Chem. Fundum. 14:16-22.
Nukiyama, S., and Tanasawa, Y. (1938-1940). Experiments on the atomization of liquids in an air stream.
Trans. Soc. Mech . Engrs. (Jupun ): 4-6. Translated
by E. Hope, Defense Research Board, Dept. of
National Defense (Canada), Ottawa, Mar. 18, 1950.
Yung, S.-C., Barbarika, H. F., and Calvert, S. (1977).
J. Azr Pollut. Control Assoc. 27:348-351.
Received 12 January 1984; accepted 27 August 1984

You might also like