See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228621688

Efficiency and pressure drop of cyclones across a range of inlet velocities

Article in Applied Engineering in Agriculture · January 2006

DOI: 10.13031/2013.20191

CITATIONS READS
58 8,013

2 authors:

Bryan W. Shaw William B Faulkner

107 PUBLICATIONS 1,761 CITATIONS

Texas A&M University
61 PUBLICATIONS 843 CITATIONS
SEE PROFILE
SEE PROFILE

All content following this page was uploaded by William B Faulkner on 21 May 2014.

The user has requested enhancement of the downloaded file.

 EFFICIENCY AND PRESSURE DROP OF CYCLONES
ACROSS A RANGE OF INLET VELOCITIES
W. B. Faulkner, B. W. Shaw

ABSTRACT. Cyclone separators are widely used in agricultural processing industries as air pollution abatement devices. The
Texas A&M Cyclone Design (TCD) method is a simple method for designing cyclones based on an inlet design velocity. The
TCD method specifies “ideal” inlet velocities of 975 + 120 m/min (3200 + 400 fpm) and 914 + 120 m/min (3000 + 400 fpm)
for 1D3D and 2D2D cyclones, respectively. There is evidence that higher dust collection efficiencies may be obtained from
cyclones using different inlet velocities than those specified as the “ideal” velocity. This article quantifies the inlet velocities
at which maximum collection efficiencies are obtained for 1D3D and 2D2D cyclones and the marginal pressure drop
associated with reaching these collection efficiencies. It is demonstrated that for large particles, the collection efficiency of
six inch diameter 1D3D and 2D2D cyclones is similar for inlet velocities from 10.16 standard m/s (2000 fpm) up to the design
velocity, with significantly lower pressure drop at lower inlet velocities.
Keywords. Cyclone, 1D3D, 2D2D, Abatement, Inlet velocity, Collection efficiency.

C
yclones are the most widely used air pollution The Lapple cyclone design method also predicts the
abatement devices for removing particulate matter efficiency with which particles of a given size will be
from air streams in agricultural processing indus- separated from the air stream. According to the Lapple
tries. Relative to other abatement systems, cy- method, dust collection efficiency increases unconditionally
clones have low initial costs, maintenance requirements, and with increased air stream velocity. However, according to
energy consumption. Cyclones use centrifugal force to sepa- Parnell (1996), above a threshold airspeed, the vortex inside
rate particulates from an air stream. A dust-laden air stream a cyclone becomes disturbed and collection efficiencies
enters a cyclone tangentially near the top of the cyclone and decline. Furthermore, Parnell (1996) indicates that the
flows downward in a spiral. Inertial and centrifugal forces act Lapple method significantly underestimates collection effi-
on the particulates, forcing them to the wall of the cyclone ciency.
where they slide down the wall to the bottom of the cone and The efficiency of cyclone systems is a function of the
are removed. The air stream spirals to the bottom of the de- particle size distribution (PSD) of entrained dust and the
vice, where it reverses direction and flows upward and out the velocity of the air stream entering the abatement device
top of the cyclone (Cooper and Alley, 2002). (Wang et al., 2000). The particle size distribution of most
The Lapple cyclone design method, often referred to as aerosols can be described by a log-normal distribution
the standard cyclone design method, specifies the dimen- (Hinds, 1999). While they have historically been considered
sions of a cyclone based on the barrel diameter. The most low-efficiency collectors, recent studies have shown that
commonly used cyclones are the 2D2D (Shepherd and cyclones can reach efficiencies exceeding 99% for particles
Lapple, 1939) and 1D3D cyclones (Parnell and Davis, 1979). larger than five micrometers (Cooper and Alley, 2002).
The Ds in the 1D3D and 2D2D designations refer to the The Texas A&M Cyclone Design (TCD) method is a
diameter of the cyclone barrel, while the numbers preceding simple method for designing cyclones based on an “optimal”
the Ds refer to the length of the barrel and cone sections, inlet velocity. According to the TCD method, the optimal
respectively. A 1D3D cyclone, for instance, has a barrel inlet air speed for a 1D3D cyclone is 975 ± 120 m/min (3200
length equal to the barrel diameter and a cone length equal to ± 400 fpm) and 914 ± 120 m/min (3000 ± 400 fpm) for 2D2D
three times the barrel diameter. The design parameters for cyclones under standard conditions (Parnell, 1996). The air
1D3D and 2D2D cyclones are shown in figure 1. stream velocities specified by the TCD method seek to
balance the competing desires for high efficiency and low
pressure drop through the abatement device. Further testing
Article was submitted for review in June 2005; approved for has indicated that higher efficiencies than those predicted by
publication by the Structures & Environment Division of ASABE in the TCD method may be achieved by increasing air speed
November 2005. through the cyclone. However, an increase in exit concentra-
The authors are William B. Faulkner, ASABE Member Engineer, tions has been observed for velocities that are significantly
Graduate Research Assistant, and Bryan W. Shaw, ASABE Member
Engineer, Associate Professor, Department of Biological and Agricultural higher or lower than the design velocities. This higher
Engineering, Texas A&M University, College Station, Texas. collection efficiency is accompanied by increased pressure
Corresponding author: William B. Faulkner, Dept. of Biological and drop across the cyclone, thus requiring higher energy inputs.
Agricultural Engineering, Texas A&M University, 2117 TAMU, College There may be times when it is economically beneficial for a
Station, TX 77843-2117; phone: 979-845-3693; fax: 979-845-3932;
processing industry to incur higher energy costs rather than
e-mail: faulkner@tamu.edu.

Applied Engineering in Agriculture

Vol. 22(1): 155-161 E 2006 American Society of Agricultural and Biological Engineers ISSN 0883−8542 155
 Figure 1. 1D3D and 2D2D cyclone layouts (Wang et al., 2003).

convert to a filter system, the cost of which may be five to ten ESD = equivalent spherical diameter
times higher than that of a cyclonic abatement system ρp = particle density (g/cm3)
(Parnell, 1996). χ = shape factor
The objective of the research described in this article is to The AED of an aerosol particle is the diameter of a unit
characterize the collection efficiency and pressure drop density sphere (i.e.: density = 1.00 g/cm3) that would have the
associated with 1D3D and 2D2D cyclones across a range of same settling velocity as the particle or aerosol in question.
inlet velocities. A secondary objective is to determine the The particle size distribution was described as a log-nor-
maximum collection efficiency and associated inlet velocity mal distribution using a MMD and geometric standard
for 1D3D and 2D2D cyclones with varying loading rates of deviation (GSD), where MMD is the particle size for which
two dust samples. Additionally, the “optimal” inlet velocities half of the mass is contributed to by particles smaller than the
specified by the TCD method will be evaluated based on MMD and half by particles larger than the MMD (Hinds,
collection efficiency and operating costs, which are a 1999). The GSD is calculated using equation 2:
function of both inlet velocity and pressure drop.
d84.1% d d84.1%
GSD = σ g = = 50% = (2)
d 50% d15.9% d15.9%
MATERIALS AND METHODS where dn% is the particle size for which n percent of the mass
A three-factorial experiment was conducted for both is contributed by particles less than d.
1D3D and 2D2D cyclones, varying inlet loading concentra- 1D3D and 2D2D metal cyclones with a diameter of
tion, aerosol PSD, and inlet velocity. Commercially available 0.1524 m (6 in.) were used to conduct experiments with the
cornstarch and alumina were used as the test dusts in these pull system shown in figure 2.
trials (table 1). Cornstarch was used because the PSD of A laminar flow element (LFE) (Meriam Instrument Model
cornstarch is similar to that of corn dust from grain elevators. 50MC2-2, Serial No. 773880-NI, Cleveland, Ohio), cali-
Alumina was also used because a second dust with a brated by the manufacturer, was used to determine the flow
relatively low mass median diameter (MMD) was desired. A rate of air through the system, and a voltage controller was
Beckman Coulter Counter MultisizerTM 3 (Beckman Coul- used to obtain the desired flow rate. The actual flow rate was
ter, Miami, Fla.) was used to determine the PSD of each dust. corrected for air temperature using a correction factor
The Coulter Counter is calibrated by the manufacturer furnished by the LFE manufacturer.
annually and by laboratory technicians every 100 runs.
The Coulter Counter measures Equivalent Spherical Table 1. Properties of experimental aerosols.
Diameter (ESD) which is then converted to Aerodynamic Particle MMD
Equivalent Diameter (AED) using equation 1: Density Shape (AED[a])
Aerosol (g/cm3) Factor (µm) GSD
ρp Cornstarch 1.5 1.00 17.95 1.41
AED = ESD (1)
χ Alumina 3.9 1.44[b] 9.96 1.42
[a] Aerodynamic equivalent diameter.
where [b] Source: (Mark et al., 1985).
AED = aerodynamic equivalent diameter

156 APPLIED ENGINEERING IN AGRICULTURE

 Figure 2. Cyclone testing system.

measured and used in conjunction with the volumetric flow

For each test, the temperature, barometric pressure, and rate through the system to determine the loading concentra-
relative humidity were recorded. Wang et al. (2003) demon- tion according to equation 4:
strated that for highest efficiency, cyclones should be min
designed based on inlet velocities of dry-standard air rather C= (4)
than actual conditions. A digital magnehelic (Dwyer Instru- Q×t
ments, Series 475 Mark III, Michigan City, Ind.) was used to where
determine the negative pressure created by the cyclone, filter, C = inlet loading concentration (g/m3)
and piping upstream of the LFE, and the standard flow rate min = mass of dust entrained in the system (g)
through the system was calculated. The standard flow rate of Q = system flow rate (m3/s)
air was calculated based on the actual air density and a t = test duration (s)
standard density of 1.20 kg/m3 (0.075 lb/ft3) using equa- The pressure drop across the cyclone and the laminar flow
tion 3: element were measured using digital magnehelics with a
ρ act 0.01-kPa resolution (Dwyer Instruments, Series 475 Mark
Qstd = Qact (3) III, Michigan City, Ind.). Teflon filters (with 2-µm pore size)
ρ std were used for replications requiring PSD analysis because of
where the lower background particle count as compared to glass
Qstd = flow rate of standard air fiber filters. The background particle count quantifies the
Qact = measured flow rate number of particles within the size range of interest which are
ρact = measured density of air (kg/m3) generated from analysis of a blank (unused) filter and are
ρstd = density of standard air (kg/m3) detected during particle size analysis. Evaluation of blank
For each trial, a filter was placed in the filter holder and glass fiber filters by Buser (2004) indicated a background
the fan was turned on. When equilibrium at the desired flow concentration of 12,388 particles in 500 µL of electrolyte,
rate was reached, a measured mass of the aerosol was fed into while 2-µm Teflon filters yielded only 58 particles in an equal
the system using a vibratory feeder at a specified rate. The amount of electrolyte. Glass fiber filters were used for the
feed rate was controlled by adjusting the frequency of remaining replications requiring only gravimetric analysis as
vibration and verified by visual inspection. Test durations they require less fan power and are significantly less
were 3 min for glass fiber filters (for gravimetric analysis) expensive than Teflon filters.
and 10 min for Teflon filters (for gravimetric and particle size Particle size distributions of the aerosol emitted from the
analysis). Tests were longer for filters used in particle size cyclone were obtained by placing ten 1.6-cm diameter
analysis so that enough dust would accumulate on the filters samples containing particulates cut from random locations
to run three sets of PSDs. Three minute tests were deemed on a given filter into a lithium-chloride methanol electrolyte
sufficient for gravimetric analysis because the amount of dust solution and placing this combination into an ultrasonic bath
that accumulates on a filter during a 3-min test is larger than for 3 min. The solution was then filtered through a 100-mm
the measurement variability of the scale used, and prelimi- screen to remove large particles. The strained electrolyte/
nary tests demonstrated that the dust collection efficiency of particulate solution was injected into a beaker of electrolyte
the cyclone is statistically similar (a < 0.05) for 3- and 10-min until a 6% to 7% concentration of dust in the electrolyte was
tests. obtained (Herber, 1988). For every PSD analysis in this
Between tests the system was run for 3 min to clear any article, three replications of three runs each were conducted
residual dust out of the system. During each trial, the mass of for all filters and the average used as the reported PSD. Each
dust entrained in the system and the test duration were run measured approximately 300,000 particles.

Vol. 22(1): 155-161 157

 Initial experiments were conducted for the 1D3D cyclone fin,j = fraction of dust entering the cyclone in the jth size
at nominal inlet velocities of 10.16, 16.26, and 20.32 m/s range
(2000, 3200, and 4000 fpm), at loading rates of 1 and 2 g/m3, mfilter = mass of dust deposited on the filter
and PSDs analyzed using the Coulter Counter. Loading rates ffilter,j = fraction of dust deposited on the filter in the jth
were selected to approximate the loading expected in a size range
process stream at a cotton gin and/or grain elevator. For each The size distribution of the fractional collection efficiency
combination of inlet velocity and loading rate there were five is the difference between two lognormal distributions (the
replications (one using a Teflon filter and four using glass PSD of dust entrained in the air stream and the PSD of dust
fiber filters). By conducting experiments at inlet velocities collected on the filter) and can also be described using a
lower than, equal to, and higher than the design velocity lognormal distribution. The size range at which 50% of the
specified by Parnell (1996), we established the magnitude of particles are collected is known as the cyclone cut point. The
change expected. This allowed for appropriate inlet veloci- slope of the cyclone collection efficiency curve is described
ties to be selected for testing. Similar initial tests were by equation 7:
conducted for the 2D2D cyclone at inlet velocities of 10.16,
15.24, and 20.32 m/s (2000, 3000, and 4000 fpm) at loading d 84.1% d d84.1%
Slope = = 50% = (7)
rates of 1 and 2 g/m3. d 50% d15.9% d15.9%
For both cyclones, subsequent similar tests were run with
1- and 2-g/m3 loading rates at inlet velocities ranging from where dn% is the particle size for which n percent of the
10.16 m/s (2000 fpm) at standard conditions, corresponding particles are collected by the cyclone.
to the lowest recommended conveying velocity for seed
cotton (Baker et al., 1994), to 20.32 m/s (4000 fpm), 2.04 m/s
(400 fpm) above the maximum design velocity of the 1D3D RESULTS AND DISCUSSION
cyclone according to the TCD method. Using fractional efficiency curves, characteristics were
For gravimetric tests, filters were weighed before and determined for the 0.154-m (6-in.) cyclones used in these
after each trial using a 10-mg resolution scale (AG245, experiments (table 2). As expected, the measured cut points
Mettler Toledo, Greifensee Switzerland). In order to reduce of both cyclones were proportional to the MMD of the
uncertainty associated with filter weights, three pre-weights experimental aerosol.
and three post-weights were taken and the averages were
used to determine the amount of dust collected on the filter.
COLLECTION EFFICIENCY
The efficiency of cyclonic separation was calculated for For both the 1D3D and 2D2D cyclones and for both dusts
each test using the equation 5:
at all tested inlet velocities, no significant difference was
∆m filter detected in cyclone collection efficiency between the inlet
η = 1− (5) loading rates of 1 and 2 g/m3.
min For tests in which cornstarch was used as the test aerosol,
where no significant difference in cyclone efficiency was detected
h = cyclone efficiency for either the 1D3D (fig. 3) or 2D2D (fig. 4) cyclones for inlet
Dmfilter = post-weight of the filter minus the pre-weight velocities ranging from 10.16 to 20.32 m/s (2000 to
of the filter (g) 4000 fpm). However, for tests in which alumina was used as
min = mass of dust entrained in the system (g) the test aerosol, significant differences were detected in the
An analysis of variance test was conducted on the collection efficiencies of both the 1D3D and 2D2D cyclone.
collection efficiencies of all treatments, and the MMDs and Tests of the 1D3D cyclone with alumina (fig. 3) showed
GSDs of all Teflon filters for each treatment. A post hoc that for all inlet velocities between 10.16 and 15.75 m/s (2000
Tukey’s HSD procedure was conducted to analyze the data. and 3100 fpm), collection efficiencies were not statistically
The null hypotheses tested (α = 0.05) were that the collection different. However, for inlet velocities greater than 17.27 m/s
efficiencies of each treatment were equal and the MMD and (3400 fpm), cyclone collection efficiency was significantly
GSD of emitted aerosols were equal. lower than that found at 15.24 m/s (3000 fpm).
A lognormal curve was used to describe the PSD data from Tests of the 2D2D cyclone with alumina (fig. 4) showed
each exposed Teflon filter. These characteristic lognormal that the collection efficiency at 17.78 m/s (3500 fpm) was
distributions were subsequently used to determine the statistically lower than at 15.24 m/s (3000 fpm), at which the
fractional efficiency curve (FEC) of the cyclone for each test. highest collection efficiency was observed. However, collec-
A fractional efficiency curve describes the efficiency with tion efficiencies were not statistically different for all inlet
which a cyclone separates particles of a given size from the velocities between 10.16 and 15.24 m/s (2000 and 3000 fpm).
entering aerosol mixture. The cyclone collection efficiency The standard deviation of measured collection efficiencies
for a given size range was determined using equation 6: increased as inlet velocity increased above 16.26 m/s
(3200 fpm) for the 1D3D cyclone and 15.24 m/s (3000 fpm)
m in × f in , j − m filter × f filter , j for the 2D2D cyclone.
h j = (6)
m in × f in , j Table 2. Cyclone characteristics.
where Cornstarch Alumina
hj = fractional collection efficiency of jth size range Cyclone d50 Slope MMD/d50 d50 Slope MMD/d50
min = mass of dust entrained in the entering air stream 1D3D 4.7 1.10 3.8 2.67 1.11 3.7
2D2D 5.0 1.11 3.6 2.49 1.13 4.0

158 APPLIED ENGINEERING IN AGRICULTURE

Figure 3. Collection efficiencies with 95% confidence intervals of the 1D3D cyclone for all replications at each inlet velocity and inlet loading rate com-
bination.

Figure 4. Collection efficiencies with 95% confidence intervals of the 2D2D cyclone for all replications at each inlet velocity and inlet loading rate com-
bination.

Vol. 22(1): 155-161 159

 The collection efficiency of cyclonic separators is a Based on the TCD method, 1D3D and 2D2D cyclones
function of the aerosol PSD and the inlet velocity of the air should operate at inlet velocities of 975 ± 120 m/min (3200
stream. These results showed that the collection efficiency ± 400 fpm) and 914 ± 120 m/min (3000 ± 400 fpm),
was not impacted by loading rate for the loading rates used respectively. The results of this research indicate that
(less than or equal to 2 g/m3). All treatments demonstrated collection efficiencies similar to those obtained at the TCD
collection efficiencies above 99%, regardless of the inlet design inlet velocity may be obtained at lower inlet velocities
velocity, cyclone, or aerosol being tested. when separating aerosols with MMDs that are much larger
For aerosols with a MMD greater than 3.5 times larger than the cyclone cut point. This reduction in inlet velocity is
than the cut point of the cyclone and GSD less than or equal accompanied by a reduction in the pressure drop across the
to 1.42, collection efficiencies equal to those obtained using cyclone, resulting in lower required energy requirements.
the TCD design inlet velocity may be obtained at lower inlet The potential energy savings that result from operating a
velocities. However, above the TCD design inlet velocity, the 0.457 m (18 in.) 1D3D and 2D2D cyclone below the TCD
variability of collection efficiency increases. This increased design inlet velocity are shown in table 3. These figures
variability is due to increased turbulence and disruption of represent the energy savings only through the cyclone and
the vortex within the cyclone. Further research is needed to disregard the remainder of the system.
determine the lowest MMD to cut point ratio for which this
relationship holds.

PRESSURE DROP
CONCLUSIONS
When separating large aerosols from process air streams,
The pressure drop across the cyclone is directly related to
cyclones may be operated at inlet velocities well below the
the fan power required to operate a cyclonic abatement
TCD design specifications. The results of this research show
device. Therefore, it is important that the pressure drop that agricultural processing industries such as cotton gins and
associated with each inlet velocity be measured so that an
grain elevators can operate cyclones at lower inlet velocities
estimate of the operating cost at each inlet velocity may be
obtained. The pressure drops for each replication as mea- Table 3. Potential energy savings.
sured across the 1D3D and 2D2D cyclones are shown in
% Energy
figures 5 and 6, respectively. Inlet Pressure Use v.
As expected, the pressure drop through both the 1D3D and Velocity Flow Rate Drop Energy TCD[a]
2D2D cyclones increased as inlet velocity increased (figs. 5 Cyclone (m/s) (m3/s) (kPa) (kW) (%)
and 6). However, the experimentally observed pressure drop 1D3D 16.26 0.0425 1.03 0.44 100
was higher than that predicted by the TCD method for both 13.21 0.0345 0.62 0.22 49
cyclones. Given the wide range over which collection 10.16 0.0265 0.32 0.09 20
efficiencies were equal, it is desirable to operate at the lowest 2D2D 15.24 0.0398 1.18 0.47 100
flow rate possible for which the collection efficiency of the 12.70 0.0332 0.73 0.24 52
cyclone is acceptable in order to reduce operating costs of the 10.16 0.0265 0.41 0.10 23
abatement device. [a] TCD = Texas A&M Cyclone Design.

Figure 5. Pressure drop measured across the 1D3D cyclone for all replications at each inlet velocity.

160 APPLIED ENGINEERING IN AGRICULTURE

 Figure 6. Pressure drop measured across the 2D2D cyclone for all replications at each inlet velocity.

and easily obtain collection efficiencies equal to those Mark, D. J., J. H. Vincent, H. Gibson, and W. A. Witherspoon.
predicted by the TCD method. These findings make it much 1985. Applications of closely graded powders of fused alumina
less critical for these industries to maintain the narrow as test dusts for aerosol studies. J. of Aerosol Sci. 16(2):
window of flow rates specified by the TCD method in order 125-131.
Parnell, C. B. 1996. Cyclone design for air pollution abatement
to be in regulatory compliance with federal and state permit
associated with agricultural operations. Proc. of the 1996
guidelines. All treatments demonstrated collection efficien- Beltwide Cotton Production Conferences. Nashville, Tenn.:
cies above 99% for the tested aerosols, regardless of the inlet National Cotton Council.
velocity or cyclone. Further research is needed to determine Parnell, C. B., and D. D. Davis. 1979. Predicted effects of the use of
how the results of these tests may be scaled to larger diameter new cyclone designs on agricultural processing particle
cyclones. emissions. ASAE Paper No SWR79040. St. Joseph, Mich.:
ASAE.
Shepherd, C. B., and C. E. Lapple. 1939. Flow patterns and
pressure drop in cyclone dust collectors. Ind. and Eng. Chem.
REFERENCES 31(8): 972-984.
Baker, R. V., E. P. Columbus, R. C. Eckley, and B. J. Stanley. 1994. Wang, L., C. B. Parnell, and B. W. Shaw. 2000. 1D2D, 1D3D,
Pneumatic and mechanical handling systems. In Cotton Ginners 2D2D Cyclone fractional efficiency curves for fine dust. Proc. of
Handbook, Agriculture Handbook No. 503, 143-171. the 2000 Beltwide Cotton Conferences. San Antonio, Tex.:
Washington, D.C.: Agricultural Research Service, U.S. National Cotton Council.
Department of Agriculture. Wang, L., M. D. Buser, C. B. Parnell, and B. W. Shaw. 2003. Effect
Buser, M. D. 2004. Errors associated with particulate matter of air density on cyclone performance and system design.
measurements on rural sources: Appropriate basis for regulating Transactions of the ASAE 46(4): 1193-1201.
cotton gins. PhD diss. College Station, Tex.: Texas A&M
University, Department of Biological and Agricultural
Engineering.
Cooper, C. D., and F. C. Alley. 2002. Air Pollution Control, A NOMENCLATURE
Design Approach, 3rd ed. Prospect Heights, Ill.: Waveland. FEC = Fractional efficiency curve
Herber, D. J. 1988. Performance evaluation of PM10 and GSD = Geometric standard deviation
high-volume air samplers using a Coulter Counter Particle Size LFE = Laminar flow element
Analyzer. MS thesis. College Station, Tex.: Texas A&M MMD = Mass median diameter
University, Department of Agricultural Engineering.
PSD = Particle size distribution
Hinds, W. C. 1999. Aerosol Technology: Properties, Behavior, and
Measurement of Airborne Particles. New York: John Wiley & TCD = Texas A&M cyclone design.
Sons.

Vol. 22(1): 155-161 161

 162 APPLIED ENGINEERING IN AGRICULTURE

View publication stats