You are on page 1of 10

Aerosol Science & Technology

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

Design and Calibration of the EPA PM2.5 Well


Impactor Ninety-Six (WINS)

Thomas M. Peters , Robert W. Vanderpool & Russell W. Wiener

To cite this article: Thomas M. Peters , Robert W. Vanderpool & Russell W. Wiener (2001) Design
and Calibration of the EPA PM2.5 Well Impactor Ninety-Six (WINS), Aerosol Science & Technology,
34:5, 389-397, DOI: 10.1080/02786820120352

To link to this article: https://doi.org/10.1080/02786820120352

Published online: 30 Nov 2010.

Submit your article to this journal

Article views: 2700

View related articles

Citing articles: 36 View citing articles

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


https://www.tandfonline.com/action/journalInformation?journalCode=uast20
Aerosol Science and Technology 34: 389 – 397 (2001)
c 2001 American Association for Aerosol Research
°
Published by Taylor and Francis
0278-6826 =01=$12.00 C .00

Design and Calibration of the EPA PM2:5 Well Impactor


Ninety-Six (WINS)

Thomas M. Peters,1 Robert W. Vanderpool, and Russell W. Wiener2


1 Research Triangle Institute, Research Triangle Park, North Carolina
2 National Exposure Research Laboratory, US EPA, Research Triangle Park, North Carolina

2.5 ¹m (PM2:5 ) (EPA 1997). This method speciŽ es a reference


The EPA well impactor ninety-six (WINS) was designed and method sampler that is deŽ ned by design and performance re-
calibrated to serve as a particle size separation device for the EPA quirements, set forth in the Code of Federal Regulations (CFR).
reference method sampler for particulate matter under 2.5 ¹m Detailed design and fabrication speciŽ cations, freely available to
aerodynamic diameter. The WINS was designed to operate down-
stream of a PM 10 inlet at a volumetric  ow rate of 16.7 Lpm. any manufacturer, are presented in 40 CFR Part 50, Appendix L
For design simplicity and ease of construction, fractionation of describing all components through which the aerosol passes: in-
the aerosol in the WINS is provided by a single-stage, single-jet, let, downtube, PM2:5 separator, upper portion of the Ž lter holder,
round-hole impactor. Particles >2.5 ¹m (aerodynamic diameter) and Ž lter cassette.
have sufŽ cient inertia to be impacted upon a circular 37 mm di- The design goals for the internal separation of PM2:5 in the
ameter glass Ž ber Ž lter immersed in 1 mL of a low volatility oil.
The relatively large amount of oil is intended to minimize sub- FRM sampler were as follows:
strate overloading and subsequent particle bounce experienced by
some conventional impactors and to represent an easily Ž eld dis- ² operate at 16.7 actual liters per minute (aLpm) down-
pensable quantity of deŽ ned tolerance. The novel geometry of the stream of the Graseby-Andersen 246B PM10 inlet,
impaction reservoir (or well) is designed to capture any reentrained ² require a single air ow control and measurement
material from the impaction surface and to prevent loss of oil should
the unit be inadvertently turned over or onto its side. The penetra- system,
tion curve of the Ž nal WINS design has a 50% cutpoint diameter ² possess a design which is easy and inexpensive to man-
equal to 2.48 ¹m and a geometric standard deviation of 1.18. ufacture and simple to operate,
During development, several nozzle designs and well geometries ² require no moving parts during sampling,
were evaluated to optimize the performance of the WINS. Addition- ² provide a unit that is easy to maintain,
ally, two candidate oils (Neovac and Dow Corning 704 diffusion
pump oils) and three types of Ž lters (glass Ž ber Ž lters, drain discs,
² require no maintenance over multiple days,
and polycarbonate membrane Ž lters) were evaluated for use as im- ² provide sustained operation at high loadings, and
paction substrates in the WINS. The performance of the WINS was ² provide separation of the aerosol by inertia with sep-
similar for the two oils in combination with a glass Ž ber Ž lter and a aration characteristics such that 50% of the particles
drain disc; however, a polycarbonate Ž lter demonstrated elevated below 2.5 ¹m in aerodynamic diameter penetrate to
penetration values. Based on these tests, a Gelman Type A/E glass
Ž ber Ž lter immersed in Dow Corning 704 diffusion pump oil was the sample collection Ž lter (D50 ) and have a shape de-
selected as the best impaction substrate. Further testing showed scribed by a geometric standard deviation (GSD) of
that the penetration curve was essentially the same when operated <1.2 (Price 1996).
with quantities of oil ranging from 0.75 to 3 mL.
The GSD is calculated as the square root of the particle diameter
associated with 16% penetration (D16 ) divided by the particle
diameter associated with 84% penetration (D84 ).
INTRODUCTION Devices that provide inertial separation of aerosols are well
On July 18, 1997, the U.S. Environmental Protection Agency documented to include virtual impactors, cyclones, and conven-
promulgated a new federal reference method (FRM) for tional impactors. The dichotomous virtual impactor (Loo and
measuring ambient concentrations of particulate matter under Cork 1988) has been the benchmark method for collecting Ž ne
and coarse mode aerosol in the atmosphere for many years. The
Received 6 May 1999; accepted 21 September 1999.
virtual impactor is dependent upon two air ows and was there-
Address correspondence to Robert W. Vanderpool, Research Trian- fore eliminated from consideration due to design complexity
gle Institute, Research Triangle Park, NC, 27709. issues.

389
390 T. M. PETERS ET AL.

The loading characteristics of a cyclone, where relatively substrate, thereby resulting in reduced efŽ ciency (or increased
large quantities of coarse aerosol may be deposited in a col- penetration) after a period of operation.
lection cup without reentrainment, makes the cyclone a favor- The well impactor ninety-six (WINS) was developed to pro-
able method of aerosol separation. John and Reischl (1980) pre- vide an impactor that met the FRM design goals and to provide
sented a cyclone now used for collecting Ž ne particulate mass an aerosol separator that could be used in the National Air Mon-
in the National Park’s IMPROVE sampling network that op- itoring Network for PM2:5 . The WINS addresses the issues of
erates at a design  ow rate of 23.1 Lpm. Cyclone design has bounce and reentrainment during extended periods of operation
been primarily empirical in nature until recently. Moore and common to conventional impactors. The primary aim of this pa-
McFarland (1990, 1993) present a model based on extensive per is to present the design, development, and evaluation of the
testing of the Stairmand family of cyclones. A second effort to WINS under clean conditions. Vanderpool et al. (2001) presents
model a family of cyclones is presented by Kenny and Gussman the behavior of this separator after sampling laboratory-created,
(1997). These modeling efforts have resulted in two cyclones coarse mode aerosol at elevated concentrations and after ex-
offering a D50 size equal to 2.5 ¹m at the desired  ow rate of tended operation in several US cities.
16.7 Lpm.
Traditional impactor designs were rejected due to their prob-
lems with particle bounce (Pak et al. 1992) and reentrainment EXPERIMENTAL METHODS
of previously deposited particles (Markowski 1984). One ap- WINS Impactor Design
proach to solving these problems has been to use an oil-soaked
The Ž nal design of the WINS, shown schematically in
sintered metal disk as the impaction substrate (Reischl and John
Figure 1, was designed to operate downstream of the Graseby-
1978). This type of surface was utilized by Marple et al. (1987)
Andersen Model 246B PM10 inlet at a volumetric  ow rate of
to reduce or eliminate particle bounce in a micro-environmental
16.7 actual Lpm. This separator consists of an upper housing,
impactor designed for indoor air studies and further utilized
an impaction reservoir (or well), and a lower housing. Separa-
in personal samplers (Marple et al. 1989). Markowski (1984)
tion of the aerosol is provided by a single-jet, round hole that is
similarly utilized two greased stages in series in a cascade im-
integral to the upper housing. A conical entrance was adopted to
pactor to reduce or eliminate artifacts caused by particle bounce
and blow-off. Turner and Herring (1987) observed that Te on
membrane Ž lters with 2 ¹L of oil could be used to reduce or
eliminate bounce. They further identiŽ ed that the viscosity of
the oil was a determining factor in whether an oil would wick up
through a pile of deposited particles (viscosity of <300 cSt) or
behave more like grease (viscosity of approximately 1000 cSt or
greater). Randtke et al. (1990) used a single-stage, single-hole
impactor with a polycarbonate Ž lter coated with 5 ¹L of sili-
cone oil as the impaction substrate to separate Ž ne particulate
for subsequent chemical analysis for species of interest in dry
deposition and acid aerosol sampling.
Further investigations into reduction of particle bounce have
focused on the modiŽ cation of the impaction surface. Biswas
and Flagan (1988) describe a particle trap impactor in which
separation of the aerosol is achieved as particles impinge upon
a stagnant cavity, thereby eliminating the need for a greased im-
paction substrate. This is a particularly desirable feature when
the particles collected above the cutpoint of the impactor are to
be analyzed for their chemical makeup. Unfortunately, the sepa-
ration characteristics of this type of impactor were observed to be
not sharp and resulted in maximum collection efŽ ciency below
100% for particles with large Stokes numbers. Tsai and Cheng
(1995) extended this concept to increase the lifetime of the im-
paction substrate for high particulate loadings. These researchers
studied a variety of impaction surface geometries coated and un-
coated with silicone grease. They identiŽ ed an inverted conical
cavity in the collection plate that was observed to alleviate parti-
cle bounce and reentrainment problems. Again, problems were
identiŽ ed after a monolayer of particles built-up on the greased Figure 1. Schematic view of the Ž nal design of the WINS.
DESIGN AND CALIBRATION OF WINS IMPACTOR 391

provide a smooth transition from the inner nozzle entrance bore throat length on separation characteristics. The T = W ratio was
to the Ž nal nozzle diameter to minimize disruption of the air reduced from 5.5 to 1.2 in this design. A third nozzle design,
streamlines moving through the nozzle and encourage a more the three-hole WINS, was constructed to elucidate the effect
uniform velocity proŽ le at the jet exit. Particles with sufŽ cient of reduced Reynolds number operation on the nozzle design.
inertia are impacted into the well upon a circular 37 mm diame- In this design, the central bore culminated in three holes of
ter glass Ž ber Ž lter immersed in 1 mL of low volatility diffusion W D 0:300 cm to yield a Reynolds number of 2600. Addition-
pump oil. ally, reduced jet velocity of the three-hole design was believed
The novel geometry of the impaction reservoir (or well) to reduce particle bounce. The Ž nal design of the WINS has a
and substrate was designed to minimize overloading and sub- nozzle width equal to 0.391 cm to yield a Reynolds number of
sequent particle bounce-off experienced by some conventional 6000 at a design  ow rate of 16.7 Lpm. The jet-to-plate dis-
impactors. An oil-immersed porous surface was selected for tance was selected to yield an S = W ratio of 3.1 to minimize
an impaction substrate because of the ability of an oil to wick interaction between the impactor air jet and the potential con-
through previously deposited particulate and present a contin- ical build-up of particles on the impaction surface. The throat
uously wetted surface to the incoming aerosol (Marple et al. length, T D 0:61 cm, was kept short in relation to the jet width
1987; Turner and Herring 1987). The nontraditional well-type (T = W D 1:6).
impaction plate geometry was designed to hold a relatively large The nonstandard shape of the well presented additional op-
quantity of oil (up to 3 mL) in the impaction area even if the de- portunity for particle collection on the walls of the well and the
vice is inadvertently turned on its side or even upside down. antispill ring. Three conŽ gurations were used in the develop-
Additionally, the well is intended to retain any blow-off from ment of the WINS: a relatively small well with a diameter of
the tip of the conical build-up that is typical after the collection 3.0 cm, a larger well with a diameter of 3.7 cm, and a  at plate.
of large quantities of particulate matter with diameters above the During the initial phases of development, an oil (Dow Corning
cutpoint of the impactor. 704 diffusion pump oil) saturated glass Ž ber Ž lter was used as
Conventional impactor design guidelines presented in Marple the impaction substrate for most of the tests. Alternatively, alu-
and Rubow (1986) were closely considered during the design minum surfaces were coated with Apiezon grease to elucidate
of the WINS nozzle section. This reference recommends that the effects of surface roughness of the glass Ž ber Ž lter on sepa-
the ratio of the jet-to-plate distance to the nozzle width (S = W ) rator characteristics.
should be >1.0 and the jet Reynolds number should be in the
range of 500 – 3000 to achieve optimal impactor performance. Evaluation of the WINS Impaction Substrate
Further, altering the number of nozzles and nozzle width can Two oils and three Ž lters were selected following a limited
control the jet Reynolds number for a given  ow rate. Marple survey of available oil and Ž lter candidates. The criteria for se-
and Willeke (1976) suggest that the throat length to nozzle width lection of the oil were reasonable cost, low vapor pressure to
ratio (T = W ) should be >1.0 to provide sufŽ cient time for the minimize volatilization, and a viscosity great enough to prevent
particles to accelerate to the  uid velocity and, if possible, the migration from the well while maintaining the ability to wick
entrance of the nozzle should be tapered or conical. through the particulate deposit. Selection of the oil was lim-
The well geometry of the impaction surface that was adopted ited to diffusion pump oils because of their low vapor pressures
for use in this design led to the use of a nonstandard nozzle (typically <10¡7 torr at 20± C).
design. A single-hole nozzle was viewed as favorable due to The two diffusion pump oils selected as the candidate oils
the relative ease of construction, adaptation to the well geome- were Neovac (P/N 4120-K6948-301 , Varian Vacuum Products,
try, and lesser propensity to clog. Unfortunately, it is necessary Woburn, MA) and Dow Corning 704 (P/N 704 diffusion pump
to operate a single nozzle outside the recommended Reynolds oil, Dow Corning Corporation, Midland, MI). Table 2 shows
number upper limit of 3000 to achieve a D50 size equal to 2.5 ¹m that these two single-component oils have similar characteris-
(a Reynolds number of approximately 5500 – 6000 is required to tics with the exception of composition. The Neovac oil is hydro-
achieve this separation). carbon based, while the Dow Corning 704 oil is silicon based.
Figure 2 displays several prototype upper housing designs The estimated cost per mL (the oil required in the well of the
that were evaluated to investigate the in uence of nozzle de- WINS) is 8 cents and 17 cents for the Neovac and Dow Corning
sign on separation characteristics. The critical parameters of oils, respectively. The viscosity of the oils under consideration
each of these prototype designs and the Ž nal design of the is 25 cSt at 40± C and 39 cSt at 25± C for the Neovac and Dow
WINS are presented in Table 1. Figure 2a presents the long- Corning 704 oils, respectively. These values fall into the lower
throat prototype nozzle. The well impaction surface required end of the oil viscosity range (30– 300 cSt) tested by Turner and
use of a nozzle with an unusually long throat length, T D 2:5 Hering (1987); therefore they are expected to behave in a desir-
cm. This design had a nozzle width of 0.452 cm, a T = W ratio able manner in that they can be wicked up into the particulate
equal to 5.5, a S = W ratio of approximately 1.8, and a Reynolds deposits.
number of 5200. A second nozzle, the bored WINS (shown in The selection criteria for the Ž lter were modest cost, the abil-
Figure 2b) was constructed to evaluate the effect of a reduced ity to wick the selected oil into the central region of the Ž lter,
392 T. M. PETERS ET AL.

Figure 2. Prototype nozzles tested: (a) the long-throat nozzle, (b) the bored nozzle, and (c) the three-hole nozzle.

and the ability to maintain structural integrity when immersed in cause they offer an extremely  at surface, but at a signiŽ cantly
oil. Table 3 displays the vendor and other pertinent information higher cost (66 cents each as compared to 14 cents each for
for the candidate Ž lters selected for this series of experiments. glass Ž ber Ž lters). Drain discs, a medium typically used under
Glass Ž ber Ž lters were selected because of the relatively low membrane Ž lters to provide mechanical support and to improve
cost and dense Ž ber weave that can easily retain oil. Two  atter the spatial uniformity of particulate matter deposits, were se-
Ž lter media were selected as candidates because of the possibil- lected as the third candidate due to their  at surface (more like
ity that the surface features of glass Ž ber Ž lters could affect the the Nuclepore Ž lter) and more porous polyester weave. Porous
penetration for particle sizes below the cutpoint of the impactor. sintered metal was considered as an alternative for the Ž lter
Nuclepore polycarbonate membrane Ž lters were selected be- media; however, it was eliminated from consideration because
DESIGN AND CALIBRATION OF WINS IMPACTOR 393

Table 1 Table 3
Critical dimensions of impactors tested Critical parameters of Ž lters tested

Throat Jet-to-plate Nozzle Glass Ž ber Nuclepore Drain disc


length; distance; width; Jet;
ConŽ guration T , cm S, cm W , cm S = W T = W Re Vendor Gelmana Costar Corp. Costar Corp.
Catalog or Type A/E cat # 1108110 cat # 231100
Long-throat 2.5 0.80 0.452 1.8 5.5 5200 model no.
WINS Composition Glass Ž ber
Polycarbonate Spun-bonde d
Bored WINS 0.56 0.80 0.452 1.8 1.2 5200 membrane polyester
3-Hole WINS 0.38 0.44 0.300 1.5 1.3 2600 Pore size, ¹m 1.0 1.0 Not available
Final WINS 0.61 1.22 0.391 3.1 1.6 6000 Thickness, ¹m 457 10 Not available
Qualities Dense weave Flat, featureless Flat, thin,
of the difŽ culties associated with periodic cleaning of such a of Ž bers surface porous
surface. Approx. price $0.14 $0.66 $0.36
A qualitative evaluation of each candidate oil and Ž lter com- each
bination was performed by visually inspecting the movement of a
Whatman type 934AH is an equivalent media.
the oils in the well as a jet of air impinged on the surface of
the impaction substrate. The jet diameter was roughly equiva- 1.35 g/cm3 . The number concentration of the primary particles
lent to the WINS impaction nozzle, 1 mL of oil was used in all downstream and upstream of the separation device was mea-
cases, and the air ow was metered at 16.7 aLpm. Measurement sured with an Aerosizer (MN Aerosizer LD, API, Amherst, MA).
of the penetration curve for each oil/Ž lter combination in the The Aerosizer was used to verify the aerodynamic particle size
Ž nal design WINS provided a more quantitative assessment of distribution of the challenge aerosol as well. Penetration for a
each candidate. given size was calculated by dividing the number concentration
of the primary particles penetrating through the separator by that
Evaluation of the WINS with Various Oil Quantities bypassing the separator. An entire penetration curve was then
developed by stepping through discrete particle sizes until an
The WINS is speciŽ ed to operate with a glass Ž ber Ž lter and
adequate number of sizes were evaluated to describe the D50
1 mL of oil. The performance of the Ž nal design WINS was
and GSD of the separator.
evaluated at 0.5, 0.75, 1.0, and 3.0 mL oil quantities to deter-
It should be noted that this method correctly accounts for par-
mine a range over which similar performance can be expected.
tial impactor jet and interstage losses and thus characterizes the
Additionally, the performance of the WINS was evaluated with
performance of the entire separator. Validation of this technique
no oil (0.0 mL) and with no Ž lter present.
was achieved by comparing penetration measurements made
with the Aerosizer to those made by collecting ammonium  u-
Measurement of Separator Characteristics orescein aerosol onto glass Ž ber Ž lters followed by subsequent
Penetration measurements were performed with the static  uorometric analysis (Peters et al. 2001). Development and eval-
testing apparatus and procedures presented in Peters et al. (2001). uation of the WINS was performed using ammonium  uores-
Brie y, this protocol involved challenging the separator with cein. During this development process, a more rapid manner of
either monodispersed aerosol created with a vibrating oriŽ ce evaluating separator performance was adopted by using suspen-
aerosol generator (VOAG, TSI, Inc., St. Paul, MN) or by nebuliz- sions containing multiple sizes of PSL spheres. The time for
ing solutions containing polystyrene latex (PSL) microspheres measuring a single penetration curve was reduced from a week,
with a density of 1.05 g/cm3 . The VOAG aerosols were com- using VOAG aerosols, to fewer than 2 h by using PSL aerosols,
posed of solid ammonium  uorescein with a known density of thereby increasing the number of substrate conŽ gurations that

Table 2
Critical parameters of oils tested

Neovac Dow Corning 704


Vendor Matsumura Oil Research Corp. Dow Corning
Viscosity, cSt 25 @ 40± C 39 @ 25± C
Vapor pressure, torr 1 £ 10¡8 @ 25± C 2 £ 10¡8 @ 25± C
Composition Hydrocarbon (mono-N-alkyldiphenylether ) Single-component silicon
(tetramethyltetraphenyltrisiloxane )
Cost per impactor oil change 8 cents 17 cents
394 T. M. PETERS ET AL.

could be tested. For this reason, the evaluation of Ž lter and oil
substrate candidates and the operation at various oil levels was
performed using PSL aerosol.

RESULTS AND DISCUSSION


Development of the WINS
Figure 3 displays the percent penetration versus aerodynamic
diameter for selected conŽ gurations. Figure 4 displays the same
selected conŽ gurations expressed as a penetration in percent ver-
sus the aerodynamic diameter divided by the cutpoint diameter.
This Ž gure is useful for penetration slope comparison. Table 4
presents the D50 , D84 , D16 , and GSD for each conŽ guration
tested.
The long-throat WINS design, with a 3.0 cm well, possessed
a cutpoint diameter of 2.55 ¹m and a GSD of 1.24. Operating
this nozzle design with a  at greased plate as the impaction
surface identiŽ ed that losses occurred in the small 3.0 cm well
because the cutpoint shifted upward from 2.55 ¹m to 3.05 ¹m.
The penetration curve became somewhat sharper, resulting in a
reduction in the GSD from 1.24 to 1.22. Figure 4. Penetration versus aerodynamic diameter divided
The bored nozzle, possessing a T = W of 1.2 as opposed to 5.5 by the cutpoint size for selected tests conducted during WINS
in the long-throat design, had a GSD equal to 1.14 when operated development.
with a  at greased plate. This demonstrates that an upper limit
on T = W should be established in the design criteria established
With greased wells, the measured GSD for the bored WINS
by Marple and Rubow (1986). It was conjectured that the long
nozzle was 1.21 and 1.18 for the 3.0 cm well and the 3.7 cm
throat length encourages the development of a large boundary
well, respectively. This indicates that the presence of the well
layer at the nozzle walls where the air and particles would have
does have an impact on the shape of the penetration curve; how-
lower velocities than in the center of the jet. This being the
ever, enlarging the diameter of the well reduced the effect. The
case, the differing velocities would give the particles different
well modiŽ es the air ow patterns as compared to an ideal,  at-
momentum and result in a nonsharp cut.
plate impactor and thereby presents additional opportunities for
particle collection. Figure 4 shows that the greased  at plate
demonstrated similar collection characteristics, above the cut-
point, to the large diameter well. Therefore it is conjectured that
the wall effect is no longer a factor for the larger well and that
deviation from ideal performance, for particles above the cut-
point, is likely due solely to the Reynolds number being greater
than the ideal 500 – 3000 range.
The GSD with the oil and a glass Ž ber Ž lter impaction sub-
strate was greater than with a greased surface in all cases exclud-
ing the small diameter well. Close examination of Figure 4 shows
that the presence of the Ž lter slightly lowers the penetration curve
for particle sizes smaller than the D50 of the impactor and hence
the GSD is increased. It is believed that the surface roughness
of the glass Ž ber Ž lter media may be the cause of this effect.
The three hole WINS nozzle showed extremely high GSD
values for both a  at greased plate (GSD D 1.30) and a small
well with an oil/Ž lter surface (GSD D 1.32). The deposits of the
particles directly under the impaction nozzle were observed to
be irregular in shape, indicating that the nozzles were likely too
closely spaced and thus resulting in jet interaction. The multiple
hole design was abandoned because of the increased complexity
Figure 3. Penetration versus aerodynamic diameter for se- of multiple-hole construction and the incompatibility with the
lected tests conducted during WINS development. well design.
DESIGN AND CALIBRATION OF WINS IMPACTOR 395

Table 4 Table 5
Summary of results of testing various WINS designs Equation (1) parameters of the best-Ž t curve
and conŽ gurations for the Ž nal design of the WINS

ConŽ guration D50 , ¹m D84 , ¹m D16 , ¹m GSD Curve parameter WINS


Long-throat WINS a 1.001
(3.0 cm well) b ¡1.001
Oil/Ž lter in well 2.55 2.05 3.15 1.24 c 2.48
Flat greased plate 3.05 2.55 3.8 1.22 d 0.3093
Bored WINS e 3.3683
greased surface:
Flat plate 3.2 2.75 3.55 1.14
A reverse asymmetric sigmoid equation was selected from nu-
3.0 cm well 2.8 2.4 3.5 1.21
merous equations based on the sum of the squares, R 2 , as our Ž t
3.7 cm well 3.0 2.6 3.6 1.18
analysis criteria. This equation is expressed as
Oil with glass
Ž ber Ž lter: " " Á¡ ¡ 1 ¢ ¢ !#¡e #
x C d ln 2 e ¡ 1 ¡ c
Flat plate 2.8 2.45 3.5 1.20 y D a C b 1 ¡ 1 C exp ;
3.0 cm well 2.7 2.2 3.2 1.21 d
3.7 cm well 2.9 2.35 3.35 1.19 [1]
Three-hole WINS
Flat greased plate 2.4 1.8 3.05 1.30 where y is the penetration represented as a fraction of one, x is
Oil/Ž lter in well 2.3 1.65 2.9 1.32 particle aerodynamic diameter, and a – e are curve parameters.
Final WINS design 2.48 2.05 2.85 1.18 Although simpler algorithms have been found to Ž t symmetrical
penetration curves adequately, they do not Ž t nonsymmetrical
performance curves as well as the function provided by Equation
Evaluation of the Final Version of the WINS (1). Table 5 presents the equation parameters for the best-Ž t
The Ž nal design of the WINS was designed using the results curve for the WINS.
from the prototype nozzle tests. Figure 5 displays the penetra- Based on this best-Ž t curve through the experimental data,
tion curve for the Ž nal WINS design. TableCurve 2D (a computer the Ž nal design of the WINS has a D50 cut-size of 2.48 ¹m and
software program for Ž tting curves to experimental data, Jandel a GSD of 1.18. This is in good agreement with the theoretical
ScientiŽ c Software, San Rafael, CA) was used to Ž t the exper- cutpoint of this design, D50 D 2:40 ¹m, which was calculated
imental data with a best-Ž t curve (Maynard and Kenny 1995). using the impaction theory presented by Marple and Willeke
(1976) (a 50% square root of a Stokes number of 0.475 was
used in this calculation). Further, the fact that the experimentally
determined cut-size is somewhat larger than that calculated from
theory supports the supposition that the larger diameter well does
not interfere with the separation characteristics of this WINS
Ž nal design.

Evaluation of WINS Impaction Substrate


There was no visible difference in the behavior of the two oils;
however, the Ž lter media had an effect on the retention of the oil.
The dense weave of the glass Ž ber Ž lter retained the bulk of the
oil in the center of the well, with a slight ring of excess oil pushed
to the edge of the Ž lter by the force of the impinging air jet. The
drain disc demonstrated similar dynamic qualities, but with a
larger quantity of excess oil due to the lower retention qualities
associated with this media. The Nuclepore Ž lter appeared to have
an even lower holding capacity with a more pronounce d ring of
oil with the visible oil cleared from the surface of the Ž lter.
Figure 6 displays the effect of the various impaction sub-
strates on the shape of the penetration curve of the WINS. There
was no consistent difference observed between the two oils.
Figure 5. Evaluation of the Ž nal version of the WINS. The drain disc and the glass Ž ber Ž lter also behaved with no
396 T. M. PETERS ET AL.

Operation of the WINS at Various Oil Quantities


Figure 7 displays the performance of the WINS at various
quantities of oil in the well. The penetration of 3.4 ¹m PSL
particles is increased dramatically from approximately 4% to
28% for 0.75 mL oil and 0.0 mL with no Ž lter, respectively.
This is likely due to the PSL particles bouncing off the alu-
minum surface, being reentrained into the airstream, and sub-
sequently carried to the detector. WINS performance was less
sharp when operated with 0.5 mL oil, possibly due to an in-
creased effect of surface roughness at this lower oil quantity.
The performance of the WINS was essentially identical for 0.75,
1.0, and 3.0 mL.

SUMMARY AND CONCLUSIONS


The WINS impactor was designed and calibrated to serve as a
particle size separation device for the FRM for PM2:5 . For design
simplicity and ease of construction, fractionation of the aerosol
is provided by a single-stage, single-jet, round-hole impactor.
Particles >2.5 ¹m have sufŽ cient inertia to be impacted upon a
Figure 6. Performance evaluation of candidate oils and Ž lters circular 3.7 cm diameter glass Ž ber Ž lter which is immersed in
for use as an impaction surface in the Ž nal WINS. 1 mL of a low volatility oil. The relatively large amount of oil
is intended to minimize substrate overloading and subsequent
particle bounce experienced by some conventional impactors
noticeable differences. The only observable difference in sub- (refer to Vanderpool et al. (2001) for more information). The
strate performance was detected for the Nucleopore Ž lter. The novel geometry of the impaction reservoir (or well) is designed
penetration measured with this media was consistently greater to capture any reentrained material from the impaction surface
than that measured with the other Ž lters. From this experi- and to prevent loss of oil should the unit be inadvertently turned
ment, it was decided that a glass Ž ber Ž lter immersed in Dow over or onto its side.
Corning 704 diffusion pump oil was an acceptable choice of The nonstandard geometry of the long-throat prototype
substrates. WINS led to a rather shallow separation curve. Several addi-
tional prototype nozzles and wells of varying sizes were used
to demonstrate that a combination of nozzle throat length, well
size, and surface roughness of the glass Ž ber Ž lter caused the
deviation from the ideal case. As a result, the Ž nal WINS used
a nozzle possessing a throat-to-nozzle width of 1.2 and the well
size was increased from 3.0 cm to 3.7 cm. Additional testing of
 atter impaction substrates revealed that the glass Ž ber mat pos-
sesses desirable oil retention characteristics and was therefore
maintained as the impaction substrate. The Ž nal WINS design
was shown to have a cutpoint of 2.48 ¹m with a GSD of 1.18.
This design meets all speciŽ ed design objectives and should thus
serve as an effective particle size separator during conditions of
its intended Ž eld use.

ACKNOWLEDGMENT
The authors gratefully acknowledge the valuable comments
provided by Mr. Robert Gussman of BGI, Inc. and Dr. Dale
Lundgren of the University of Florida during the development
of the WINS.
This work was conducted by Research Triangle Institute with
support provided by the U.S. Environmental Protection Agency
through contract no. 68-D5-0040 . It has been reviewed in accor-
Figure 7. Performance of the WINS for various oil quantities. dance with the Agency’s peer and administrative review policies
DESIGN AND CALIBRATION OF WINS IMPACTOR 397

and approved for presentation and publication. Mention of trade Maynard, A. D., and Kenny, L. C. (1995). Performance Assessment of Three
names or commercial products does not constitute endorsement Personal Cyclone Models, Using an Aerodynamic Particle Sizer, J. Aerosol
or recommendation by RTI or the Agency. Sci. 26:671 – 684.
Moore, M. E., and McFarland, A. R. (1990). Design of Stairmand-Type Sampling
Cyclones, Am. Ind. Hyg. Assoc. J. 51:151 – 159.
REFERENCES Moore, M. E., and McFarland, A. R. (1993). Performance Modeling of
Biswas, P., and Flagan, R. C. (1988). The Particle Trap Impactor, J. Aerosol Sci. Single-Inlet Aerosol Sampling Cyclones, Environ. Sci. Technol. 27:1842 –
19:113 – 121. 1848.
EPA. (1997). National Ambient Air Quality Standards for Particulate Matter; Pak, S. S., Liu, B. Y. H., and Rubow, K. L. (1992). Effect of Coating Thick-
Final Rule, Federal Register, Vol. 62, No. 138, July 18, 1997, pp. 38651 – ness on Particle Bounce in Inertial Impactors, Aerosol Sci. Technol. 16:141 –
38752. 150.
John, W., and Reischl, G. (1980). A Cyclone for Size-Selective Sampling of Peters, T. M., Vanderpool, R. W., and Wiener, R. W. (2001). Methodology for
Ambient Air, JAPCA 30:872 – 876. Measuring PM2:5 Separator Characteristics Using an Aerosizer, Aerosol Sci.
Kenny, L. C., and Gussman, R. A. (1997). Characterization and Modelling of a Technol. 34(5):398 – 406.
Family of Cyclone Aerosol Preseparators, J. Aerosol Sci. 28:677 – 688. Price, J. H. (1996). Letter to Dr. George T. Wolff, Chair, Clean Air ScientiŽ c
Loo, B. W., and Cork, C. P. (1988). Development of High EfŽ ciency Virtual Advisory Committee Regarding Reopt of the Clean Air ScientiŽ c Advisory
Impactors, Aerosol Sci. Technol. 9:167 – 176. Committee (CASAC) Technical Subcommittee for Fine Particulate Monitor-
Markowski, G. R. (1984). Reducing Blowoff in Cascade Impactor Measure- ing, August 1, 1996.
ments, Aerosol Sci. Technol. 3:431 – 439. Randtke, S. J., Lane, D. D., and Baxter, T. E. (1990). Developmen t of a Sampling
Marple, V. A., Liu, B., Behm, S., Olson, B., and Wiener, R. (1989). A Personal Procedure for Large Nitrogen- and Sulfur-Bearing Aerosols, EPA/600/S3-
Environmental Monitor (PEM). Presented at The American Association for 90/048.
Aerosol Research Annual Meeting, October 10 – 13, 1989, Reno, NV. Reischl, G. P., and John, W. (1978). The Collection EfŽ ciency of Impaction
Marple, V. A., and Rubow, K. L. (1986). Theory and Design Guidelines. In Surfaces: A New Impaction Surface. Staub-Reinhalt, Luft 38 – 55.
Cascade Impactor, edited by J. P. Lodge, Jr. and T. L. Chan. American Indus- Tsai, C. J., and Cheng, Y. H. (1995). Solid Particle Collection Characteristics on
trial Hygiene Association, Akron, OH, pp. 79 – 101. Impaction Surfaces of Different Designs, Aerosol Sci. Technol. 23:96 – 106.
Marple, V. A., Rubow, K. L., Turner, W., and Spengler, J. D. (1987). Low Flow Turner, J. R., and Hering, S. V. (1987). Greased and Oiled Substrates as Bounce-
Rate Sharp Cut Impactors for Indoor Air Sampling: Design and Calibration, Free Impaction Surfaces, J. Aerosol Sci. 18:215 – 224.
JAPCA 37:1303 – 1307. Vanderpool, R. W., Peters, T. M., Natarajan, S., Gemmill, D. B., and Wiener,
Marple, V. A., and Willeke, K. (1976). Impactor Design, Atmos. Environ. R. W. (2001). Evaluation of the Loading Characteristics of the EPA WINS
10:891 – 896. PM2:5 Separator, Aerosol Sci. Technol. 34(5):444 – 456.

You might also like