EPA/600/R-12/646 | October 2012 | www.epa.gov/ord
United States
Environmental Protection
Agency
Determination of the
Sampling Efficiency of
Biosamplers to Collect
Inhalable Particles
Cooling
Hofitycomb
Sampler
Office of Research and Development
National Homeland Security Research Center
-------
EPA/600/R-12/646
September 2012
Determination of the
Sampling Efficiency of
Biosamplers to Collect
Inhalable Particles
U.S. Environmental Protection Agency
Office of Research and Development
National Homeland Security Research Center
Research Triangle Park, NC 27711
Office of Research and Development
National Homeland Security Research Center
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development's National Homeland Security Research Center, funded and directed the research
described herein under Contracts EP-D-05-065 and EP-D-10-070 with Alion Science and
Technology. It has been reviewed by the Agency but does not necessarily reflect the Agency's views.
No official endorsement should be inferred. EPA does not endorse the purchase or sale of any
commercial products or services.
For questions about this report, please contact Dr. Russell W. Wiener of the U.S. Environmental
Protection Agency, National Homeland Security Research Center, 109 T.W. Alexander Drive, Mail
Code: D205-03, Research Triangle Park, NC 27709, 919-541-1910, wiener.russell@epa.gov.
-------
Foreword
Following the events of September 11, 2001, EPA's mission was expanded to address critical needs
related to homeland security. Presidential Directives identify EPA as the primary federal agency
responsible for the country's water supplies and for decontamination following a chemical,
biological, and/or radiological attack.
As part of this expanded mission, the National Homeland Security Research Center (NHSRC) was
established to conduct research and deliver products that improve the capability of EPA to carry out
its homeland security responsibilities. One focus area of this research is the compilation,
development, and evaluation of information on the ability to measure potential exposure to pathogens
that might be used by terrorists. Such information is critical to understanding the risks associated with
biological contamination and supporting the development of site-specific cleanup goals, treatment
technologies, and detection limits.
NHSRC has made this publication available to assist the response community in preparing for and
recovering from disasters involving microbial contamination. This information is intended to move
EPA one step closer to achieving its homeland security goals and its overall mission of protecting
human health and the environment while providing sustainable solutions to our environmental
problems.
Jonathan Herrmann, Director
National Homeland Security Research Center
-------
-------
Contents
Notice ii
Foreword iii
List of Figures vi
List of Tables vii
Acronyms and Abbreviations viii
Executive Summary ix
1.0 Introduction 1
2.0 Experimental Methods 3
2.1 Aerosol Wind Tunnel 3
2.2 Selected Bioaerosol Samplers 4
2.2.1 XMX Bioaerosol Sampler 5
2.2.2 Portable Sampling Unit 5
2.2.3 DryClone™ Sampler 5
2.3 Liquid Aerosol Generation 5
2.4 Aerosol Uniformity 7
2.5 Aerosol Quality 8
2.6 Fluorometric Analysis 8
2.7 Test Protocol 9
3.0 Results and Discussion 11
3.1 Aerosol Uniformity Results 11
3.2 XMX Sampler Results 13
3.3 PSU Sampler Results 13
3.4 DryClone™ Sampler Results 13
3.5 Sampling Efficiency 13
4.0 References 23
-------
List of Figures
Figure 1 Plan view of the EPANHSRC aerosol wind tunnel 3
Figure 2 Aerosol generation system 6
Figure 3 Critical dimensions of the rake samplers showing locations of aerosol
sampling points for two configurations 7
Figure 4 Isokinetic sampler filter showing collected uranine-tagged particles 10
Figure 5 PSU sampler efficiency data plotted against particle size for all three test
wind speeds 14
Figure 6 DryClone™ sampler efficiency data plotted against particle size for all three test
wind speeds 17
Figure 7 PSU sampling efficiency plotted as a function of the square root of Stk 20
Figure 8 DryClone™ sampling efficiency plotted as a function of the square root of Stk 20
-------
List of Tables
Table 1 Bioaerosol Samplers and Operating Specifications 4
Table 2 Aerosol Uniformity Results 12
Table 3 Sampling Efficiencies of the PSU Sampler 15
Table 4 Sampling Efficiencies of the DryClone™ Sampler 18
Table 5 Gompertz Parameter Values from Regression Fitting for the PSU Sampler 19
Table 6 Gompertz Parameter Values from Regression Fitting for the DryClone™ Sampler 19
-------
Acronyms and Abbreviations
°C degrees Celsius lb
og geometric standard deviation m
um micrometer MFC
AE aspiration efficiency min
APS Aerodynamic Particle Sizer mL
ATF Aerosol Test Facility mm
AWT aerosol wind tunnel mph
CE collection efficiency NaOH
CFR Code of Federal Regulations ng
cm centimeter NHSRC
CV coefficient of variation PSL
dpae particle aerodynamic diameter PSU
EPA U.S. Environmental Protection Agency R2
h hour RH
hp horsepower rpm
HPLC high-performance liquid chromatography SD
Hz hertz SE
in. inch SOP
iso isokinetic Stk
K Kelvin STS
km kilometer TE
kPa kilopascal TI
kW kilowatt VOAG
L liter
pound
meter
mass flow controller
minute
milliliter
millimeter
miles per hour
sodium hydroxide
nanogram
National Homeland Security Research Center
polystyrene latex
portable sampling unit
coefficient of determination
relative humidity
revolutions per minute
standard deviation
sampling efficiency
standard operating procedure
Stokes number
sampler test section
transmission efficiency
turbulence intensity
vibrating orifice aerosol generator
-------
Executive Summary
Three bioaerosol samplers were chosen for evaluation of sampling efficiency (SE) in the aerosol wind
tunnel according to 40 CFR Part 53 Subpart D (U.S. EPA, 1998): the XMX/2L-MIL, manufactured
by Dycor Technologies, Ltd.; the portable sampling unit (PSU) sampler, manufactured by Hi-Q
Environmental Products; and the DryClone™ sampler, manufactured by Evogen, Inc. Wind tunnel
testing of the samplers was carried out with monodisperse liquid aerosols (oleic acid and uranine)
with particle diameters of 5, 10, 15, and 20 um. The DryClone™ and PSU samplers were tested in the
wind tunnel in triplicate with each of the four particle sizes at wind speeds of 2, 8, and 24
kilometers/hour (km/h). The sampling efficiency for inhalable particulate (5 to 20 (im aerodynamic
diameter, dpae) is reported versus dpae and the square root of the Stokes number (a nondimensional
number relating velocity and aerodynamic particle size to relaxation time). Neither of the samplers
showed wind speed insensitivity, and both had a rapid decline in efficiency with rising Stokes
number.
The XMX sampler was tested with 5-um aerosol at 2 and 8 km/h, but the SE did not compare well
with the manufacturer's results for solid monodisperse aerosols under the same experimental
conditions. The manufacturer indicated that testing with liquid aerosol was not appropriate. For this
reason, we suggest retesting the XMX sampler using solid particles prepared from fluoroscein and
ammonium hydroxide.
-------
-------
1.0
Introduction
This research was conducted as part of the
efforts of the U.S. Environmental Protection
Agency (EPA) National Homeland Security
Research Center (NHSRC) to determine the
sampling capability of biological aerosol
samplers expected to be used for homeland
security field support to collect respirable and
inhalable particles resuspended after an
accidental or intentional release of bioagents.
The specific objective of this research was to
determine the sampling efficiency (SE) of
several biological samplers to characterize their
performance for a number of particle
aerodynamic diameters and wind speeds. For
this work, the PMi0 test criteria published in
Title 40 Part 53 Subpart D of the Code of
Federal Regulations (40 CFR Part 53 Subpart D)
formed the basis of the wind tunnel evaluation.
Direct assessment of particle concentration in
the atmosphere is based on aerosol sampling by
different measurement devices. Small airborne
particles are well entrained in the airflow and are
collected by the sampling device with high
efficiency. In contrast, larger particles can
deviate from fluid stream lines (aspiration loss)
and can deposit inside sampler inlets and not
reach the filter or measurement zone
(transmission losses). The presence of the
sampler and the action of sampling can cause
disturbances in the surrounding air and affect
particle movement, leading to a significant
difference between the particle concentration
measured by the sampler and the particle
concentration in the free-stream air or ambient
atmosphere. The SE, defined as the ratio of the
measured mean particle concentration to that in
the free-stream air, characterizes these
discrepancies.
The SE of a sampler can be further broken down
into three factors: aspiration efficiency (AE),
transmission efficiency (TE), and collection
efficiency (CE), as SE = AE x TE x CE
(Kesavan et al., 2003; Brockmann, 2011). AE is
the efficiency of transport of particles from the
free-stream air into the inlet of the sampler (ratio
of concentration that enters the sampler inlet to
free-stream concentration) and is influenced by
the wind speed and configuration of the sampler
inlet. TE is the efficiency of transport of
particles from the sampler inlet to the filter or
measurement zone. CE is the efficiency with
which the sampler collection medium (filter,
impactor, impinger, etc.) collects and retains
particles. All three efficiencies are particle size
dependent. In the wind tunnel testing portion of
this study, the SE of each sampler was
determined by comparing the test samplers to
reference measurements.
-------
-------
2.0
Experimental Methods
2.1 Aerosol Wind Tunnel
To evaluate the samplers according to the 40
CFR Part 53 Subpart D acceptance criteria, a
wind tunnel is required to provide a controlled
environment with well-defined velocity profiles
and monodisperse particle concentrations. An
overview of the Aerosol Test Facility (ATF) wind
tunnel, part of NHSRC, is shown in Figure 1. In
plan view, the aerosol wind tunnel (AWT) is
rectangular with outside dimensions of
approximately 20 meters (m) by 14 m. Flow
through the recirculating wind tunnel during all
operations is counterclockwise with few flow
obstructions. A number of doors with magnetic
locks allow access to all sections of the wind
tunnel for cleaning. At the sampler test section
(STS), the wind tunnel cross-section is 1.75 m
wide by 1.45 m high by 6.1 m long.
"\
ff
Turning
fr
\-
Aerosol
injection
Co<
c<
=^
li
il
-
|
ig f-i
Fil
ta
1
C
4
I
Honeycomb
^
Airflow
Honeycomb
Sampler
test section
\
•*/
Figure 1. Plan view of the EPA NHSRC aerosol wind tunnel.
Since the wind tunnel is of fixed geometry,
varying wind speeds are achieved by controlling
the volumetric flow rate. Major flow through the
wind tunnel is provided by a direct-drive,
adjustable-blade, vane-axial fan (Twin City Fan
and Blower, Minneapolis, MN) capable of
providing approximately 2002 m /min (71,500
ft3/min) against 0.97 kPa (3.89 inches of water)
pressure drop at 1133 rpm at a power
requirement of 56 kW (75 hp). This blower is
capable of producing wind speeds up to 48 km/h
(30 mph) in the STS. Wind speed is controlled
through a variable-speed drive combined with a
fan pitch system that regulates the rotational rate
-------
of the fan. High-speed operation of the wind
tunnel adds a significant amount of heat to the
recirculating air stream, so a cooling coil/chilled
water system is used to control the tunnel
temperature. Controlled recirculation of chilled
water through the cooling coil counteracts the
continued heat input and allows the wind tunnel
to be operated at specified temperatures.
Temperature was maintained at 20 °C for the
experiments presented here.
Humidity in the wind tunnel can also be
controlled and is achieved by the combination of
a desiccant dehumidifier and a deionized water
steam humidifier. After the humidity reaches the
target condition, the dehumidifier operates at a
constant (low) setting and the humidifier output
is automatically controlled to maintain the target
set point. Relative humidity (RH) was
maintained at 50% for the experiments described
here.
The wind tunnel includes a bank of high-
capacity, mini-pleated filters downstream of the
test section to remove aerosols not collected by
the samplers. This primary filter bank effectively
prevents the continuous accumulation of
material in the tunnel interior, dramatically
reducing the background level of particles in the
air stream.
Performance specifications outlined in 40 CFR
Part 53 require that the velocity profile in the
test section be well characterized at each of three
wind speeds—2, 8, and 24 km/h—and meet
specific acceptance criteria. Wind speed and
turbulence intensity (TI) must be measured by
hot-wire anemometry at a minimum of 12 points
in a cross-sectional area of the test section. The
wind tunnel meets velocity acceptance criteria if
the mean wind speed in the test section is within
10% of the desired mean and the variation at any
test point in the test section does not exceed 10%
of the measured mean. TI must be measured and
recorded, but there are no specifications for
either the magnitude or the scale of the
turbulence. Wind velocity measurements were
made in the ATF wind tunnel during the sampler
evaluation using a sonic anemometer and were
found to meet the specified acceptance criteria.
TI was calculated during the evaluation and was
found to be 2.8% at an average wind speed of 2
km/h, 1.0% at 8 km/h, and 0.80% at 24 km/h.
2.2 Selected Bioaerosol Samplers
Three bioaerosol samplers were selected and
evaluated during the project. These samplers are
listed in Table 1 along with their respective
operating specifications.
Table 1. Bioaerosol Samplers and Operating Specifications
Vendor
Dycor Technologies
Ltd. (2006)
HI-Q Environmental
Products
(2012)
Evogen, Inc.
(2012)
•Reported to be 800
Sampler
ID
XMX
PSU
DryClone™
Collection
Media
47-mm filter,
5-mL liquid
extraction
47-mm filter
13-mL liquid
extraction
Flow
(L/min)
650 primary*
12
secondary
100
400
L/min in another evaluation (Black, 2011) and
Particle
Size
Range
(Mm)
1-10
Not
available
<100
measured to
Cabinet
Dimensions
(in.)
11 x 17.5x13.5
18x18x12
8x8.5x16.5
Inlet
Dimensions
(in.)
Inlet height
above cabinet,
9.125
Inlet height
above cabinet,
44
Inlet height
above cabinet,
33.25
Total
Weight
(Ib)
37.5
55
17
be 720 L/min for this study.
-------
2.2.1 XMX Bioaerosol Sampler
The XMX/2L-MIL is manufactured by Dycor
Technologies, Ltd. (Edmonton, Alberta, Canada;
www.dycor.com) and was designed as a
weatherproof, outdoor aerosol collector to
sample and concentrate aerosols in the respirable
particle size range. The XMX has a high
sampling flow rate (measured to be 720 L/min
volumetric for this study) and concentrates
aerosols between 1 and 10 um using a two-stage
virtual impactor. The particles are impinged into
a centrifuge tube containing a liquid (e.g.,
phosphate-buffered saline) that the user removes
for subsequent analysis. For this study, we used
the XMX dry filter unit option to replace the
liquid vial. This unit collects the particles onto a
filter (37-mm mixed cellulose ester), which is
then removed and extracted in liquid for further
analysis.
2.2.2 Portable Sampling Unit
The portable sampling unit (PSU) (U.S. DHS,
2009) is manufactured by Hi-Q Environmental
Products (San Diego, CA; http://www.hi-q.net)
and is designed to pull air samples through a
filter for the collection of biological aerosols.
The filters used for the PSU were 47-mm-
diameter glass-fiber filters produced by
Millipore Corporation (Billerica, MA). The PSU
encloses a pump, flow controller, flowmeter,
filter holder, and all electronics inside two
individual lockable, weather-resistant boxes.
The PSU sample inlet is adjustable between 48
and 72 inches above the ground and samples air
at a flow rate of approximately 100 L/min.
2.2.3 DryClone™ Sampler
The DryClone™, developed by Evogen, Inc.
(Kansas City, MO; www.evogen.com), is a
prototype instrument designed to collect and
concentrate airborne aerosol in a liquid medium.
The specially designed system is particularly
important for collecting viable samples of
biologies and other low vapor pressure aerosols.
The aerosol is separated by inertial impaction
onto the inner (dry) surfaces of the cyclone. The
aerosol is collected into the liquid medium by
means of an automated rinse phase. The air
velocity decreases and liquid is injected into the
air stream to wash the walls of the cyclone. The
particle-laden rinse solution is then removed for
further analysis.
2.3 Liquid Aerosol Generation
Adhering to the guidance in 40 CFR Part 53.42,
liquid aerosols were generated using a vibrating
orifice aerosol generator (VOAG). The volume
of the ATF wind tunnel required several VOAGs
to produce sufficient quantities of monodisperse
aerosols to challenge each candidate sampler
adequately. The VOAGs used to generate the
aerosols in the wind tunnel were equipped with a
20-um nominal diameter orifice (see Figure 2).
Liquid flow through the orifice (0.160 cm3/min)
was provided by a Scientific Systems (State
College, PA) high-performance liquid
chromatography (HPLC) pump that was
calibrated independently at the design flow rate.
Their applied AC signal was produced by a B-K
Precision (Yorba Linda, CA) Model 4040A
frequency generator. The VOAGs were
routinely operated at a vibrational frequency of
60,000-200,000 Hz. Monodisperse aerosols of
the required diameter were produced through
proper selection of liquid feed rate, vibrational
frequency, and solution composition. Flow rate
for the VOAGs was determined by fluorometric
analysis.
The liquid solution for production of liquid
aerosols consisted of a mixture of oleic acid and
uranine dissolved in pure ethanol. Liquid flow
rate, vibrational frequency, and solution
-------
concentration were varied to produce particles of
the desired aerodynamic diameter. The aerosol
solution was prepared by measuring quantities
of oleic acid and uranine and adding enough
200-proof ethanol to reach a total volume of
2000 mL. The mixture was homogenized for use
with the VOAG. The product of each vibration
was an oleic acid droplet "tagged" with uranine
in a 10:1 ratio. This fluorescent tracer provided a
metric for quantitative analysis.
The liquid droplets that were produced were
then dispersed using a dispersion air flow rate of
1700 cmVmin and diluted at a flow rate of
3 m3/h. Because the liquid nebulization process
produced electrically charged droplets, a Kr-85-
based neutralizer (Model 3054, TSI, Inc.,
Shoreview, MN) was used to produce an
electrically neutral aerosol. The neutralized
aerosols were then introduced into the wind
tunnel through heated copper pipes.
Pump
Solution reservoir
Function generator
Dilution and dispersion
air control panel
—
»
i
v>
1
5
o
'5
c
I
i
VOAG
•=
i
I
0
•>
'5
|
2
~
~
i
I
Q
"^
'5
1
0
"•
~
~
i
I
D
•^
r!5
1
2
3
1 VOAG 2 VOAG 3 VOAG 4
@
Set at 15PSI
Hand
valve
©
Pressure
5
c
1
~
1>
a
(A
5
Jl
i
g
1
'"-
regulator VOAe g
••
;,,''r4*
*..«.*. '•.•l>T*
—
VOAG jX^1
head
Drying^
column
Aerosol^
neutralizer
VOAG
control panel
n
Power
-------
2.4 Aerosol Uniformity
The 40 CFR Part 53 Subpart D requires that the
test aerosol within the test section be spatially
uniform. Aerosol concentrations must be
measured at a minimum of five test points, and
an acceptable uniformity is defined as a
measured coefficient of variation (CV) not to
exceed 10% at each wind speed. Title 40 CFR
Part 53.42d sets the criteria for proper sample
collection to confirm aerosol uniformity within a
wind tunnel as follows:
(d) The concentration of particles in the wind
tunnel is not critical. However, the cross-
sectional uniformity of the particle concentration
in the sampling zone of the test section shall be
established during the tests using isokinetic
samplers. An array of not less than five evenly
spaced isokinetic samplers shall be used to
determine the particle concentration uniformity
in the sampling zone. If the particle concentra-
tion measured by any single isokinetic sampler
in the sampling zone differs by more than
10 percent from the mean concentration, the
particle delivery system is unacceptable in terms
of uniformity of particle concentration. The
sampling zone shall be a rectangular area having
a horizontal dimension not less than 1.2 times
the width of the test sampler at its inlet opening
and a vertical dimension not less than 25
centimeters. The sampling zone is an area in the
test section of the wind tunnel that is
horizontally and vertically symmetrical with
respect to the test sampler inlet opening.
In the ATF wind tunnel, an array of isokinetic
nozzles, called a rake, samples in a symmetrical
pattern around the sampler test section to assess
particle uniformity. The diagrams in Figure 3
illustrate two rake configurations used during
this work. The first was designed for a previous
study where similar wind speed and particle size
combinations were attempted. Then 40 CFR Part
53.42d was revisited and a second configuration
was designed. Each sampler in the second
configuration consists of a commercial 47-mm-
diameter in-line filter holder with a conical
nozzle replacing the usual inlet of the filter
holder. Each nozzle projects approximately
10 cm upstream of the filter holder and is made
with a sharp-edged entry leading to a gradually
expanding conical section. Different nozzles
were used for conducting uniformity tests at 2,
8, and 24 km/h. The flow rate through each
nozzle was controlled using mass flow
controllers (MFCs) (Model 585IE, Brooks
Instrument, Hatfield, PA). Each MFC was
adjusted to provide isokinetic sampling at each
wind speed. The filters used in the rake samplers
were 47-mm glass-fiber filters (Type A/E, Pall
Life Sciences, Ann Arbor, MI), which were
chosen because of their collection characteristics
and low background fluorometric content.
T Sin
=f
18 a,
14 Mn
22 i in
<8in
I 1.
1 SinT
* ° 1
— J
•
— 7
,1
1
29 in
24 in.
19m
I 22 in In in
Figure 3. Critical dimensions of rake samplers
showing locations of aerosol sampling points
for two configuration.
-------
2.5 Aerosol Quality
To verify the quality and mean size of the
generated aerosols, samples were collected in
the test section using a specially designed slide
impactor. This impactor withdrew a
representative sample of the aerosol from the
test section and then impacted the sampled
particles on a microscope slide. The slides were
pretreated using a solution of 2% NyeBar®
(Type CT or Type K) oliophobic surfactant,
which resisted the spread of the oleic acid
droplets. The slides were then baked for 1 h at
100 °C.
Following the aerosol collection, a polarizing
optical microscope (Labophot™ Pol, Nikon,
Melville, NY) was used to verify that the mean
particle size and aerosol uniformity met
acceptance criteria outlined in the CFR. The
eyepiece micrometer of the microscope was
calibrated periodically using a certified stage
micrometer (Model 1400, American Optics,
Burlington, ON, Canada).
Two Aerodynamic Particle Sizer (APS™)
spectrometers (TSI, Inc., Shoreview, MN) were
installed in the wind tunnel to allow continuous
observation of the aerosol quality and to check
for the presence or absence of satellites. If large
secondary particles (larger than one-fifth of the
desired aerosol) were detected, first an attempt
was made to locate the malfunctioning VOAG
and then, if the problem could not be resolved
expeditiously, the experiment was aborted.
A spread factor established by Olan-Figueroa et
al. (1982) was used to calculate the spherical
droplet size from the deposited drop and was
checked against the particle diameter calculated
by knowledge of the feed rate, the frequency,
and the oleic acid and uranine concentration in
the feed solution. The calculated value was used
in plotting experimental sampling effectiveness
results. The target agreement between the
microscopy check and the calculated size was
generally within 0.5 (im or 5% and well within
the EPA specification of 0.5 um or 10%. The
geometric standard deviation (og) of the
generated particle size distribution was typically
less than 1.05 and also within the EPA
specification of 1.10.
2.6 Fluorometric Analysis
Per the CFR and the Tolocka et al. (200 la)
method, the aerosol particle mass was measured
by fluorometry. Following aerosol collection,
the 47-mm-diameter glass-fiber filters from the
isokinetic samplers were each placed in clean,
labeled 50-mL centrifuge tubes equipped with
resealable lids, and 30 mL of 0.0IN sodium
hydroxide (NaOH) was added to each container.
The containers were then sealed and placed in an
ultrasonic bath for 10 min. To ensure that the
extract was homogeneous, the contents of each
container were gently swirled with a spiraling
motion. Disposable pipettes were used to
transfer an aliquot of each container into
separate cuvettes for fluorometric analysis.
Comparison tests showed that centrifugation of
the solutions was not required to obtain reliable
test results.
The fluorometric content of the solutions was
determined using a calibrated Quantech™
fluorometer (Barnstead-Turner, Dubuque, IA)
containing an excitation filter (NB360) and an
emission filter (NB460). Different calibration
curves were used depending on the
concentration and particle size being analyzed.
Each curve consisted of at least five analytical
standards. A coefficient of determination (R2) of
1.00 was used as the indicator that the curves
were linear. After the curve was set, a sample
was read in the fluorometer to determine if the
samples were over-range or if the working
standards were sufficient. If the sample was
over-range, a new curve was set or the sample
was diluted. Three fluorometric measurements
were recorded for each sample, the curve was
checked intermediately, and blank filters (at
least one per run) and spikes were analyzed for
quality assurance.
-------
The aerosol concentration in the wind tunnel
was calculated from fluorometric measurements
of the uranine concentrations. If the glass-fiber
filter fluorescence was significant compared to
the collected aerosol mass, then the calculated
concentration would significantly overpredict
the actual concentration. To quantify the effect
of the filter's natural fluorometric content,
unexposed filters were analyzed as laboratory
blanks. Since the values measured from the
blank filters were significantly lower than from
the sample filters, blank corrections were not
made to the measured filter concentrations
during the course of the project.
To determine percent recovery from the filter, a
solution of known uranine concentration was
prepared and spiked onto a blank 47-mm glass-
fiber filter (simulating the rake and sampler
filters). The solution also was added to an empty
centrifuge tube to serve as a blank. The filters
and the blank were washed with 30 mL of 0.0IN
NaOH. The average percent recovery on the
spikes was 91%.
2.7 Test Protocol
To evaluate the aerosol sampling performance of
the biosampler monitors appropriately, most of the
reference and equivalent method instrument test
protocols for PMi0 were used, as described in 40
CFR Part 53 Subpart D. The CFRprovides strict
test protocols and procedures for the
determination of reference sampler or equivalent
sampler status for PMi0. The test protocol required
a comparison between particle concentration
measured by isokinetic nozzles and the test
sampler performed at fixed wind velocities and
with monodisperse particles (40 CFR Part 53;
Tolocka et al., 2001a, 2001b; Ranade et al., 1990).
The experiments were conducted according to the
CFR except that only four particle sizes were
tested: 5, 10, 15, and 20 urn.
The isoknetic sampling rake was installed in the
wind tunnel. Prior to a sampling run, new
47-mm filters were installed in each isoknetic
(iso) sampler and clean isokinetic sampling
nozzles were installed. The flow on the iso
samplers was set at the isokinetic flow rate using
certified MFCs.
The biosampler was installed in the wind tunnel.
A clean filter was loaded in the filter assembly
located within the instrument housing. For the
experiments, each sampler was oriented so that
its inlet was on the upwind side of the sampler.
The samplers were placed in the wind tunnel so
that the inlet was centered both horizontally and
vertically in the test section, approximately 2 m
downwind of the iso sampler rake. Because of
the designs of the samplers and locations of the
motors, only a portion of the PSU and
DryClone™ samplers could be placed in the
wind tunnel. Plywood plates were built to
surround the sampler housings and fit into the
floor of the tunnel. Caulking and tape were used
to fill the gaps in the tunnel floor around the
samplers. The XMX sampler was placed
completely inside the wind tunnel. Using this
configuration the blockage of the cross-sectional
area of the STS was 4.8% for the XMX, 1.3%
for the PSU, and 6.0% for the DryClone™. Each
met the CFR requirement of no more than 15%
blockage.
The typical test procedure for the biosampler
evaluation was as follows. At the start of the
day, ambient temperature and pressure
conditions were checked. Checks were then
made to verify that the wind tunnel was ready to
operate and that all mixing fans were operating
and in the correct orientation. The wind tunnel
temperature and humidity controls were turned
on. The tunnel's blower was turned on and its
rotational speed adjusted to attain the desired
wind speed. The wind tunnel air temperature and
relative humidity were monitored throughout the
tests. The monodisperse aerosol generators were
brought online and allowed to stabilize for at
least 30 min. Aerosol was then introduced into
the tunnel and a representative sample was
collected on the slide impactor during a
sampling time of at least 30 min. The aerosol
-------
quality (particle size and geometric standard
deviation) was then verified using the optical
microscope, and any necessary adjustments were
made to the aerosol generation and dispersion
system. Aerosol samples were monitored
continuously using an APS spectrometer (TSI,
Inc.), and the aerosol quality was verified based
on the geometric standard deviation. All routine
wind tunnel operations were carried out
following established ATF procedures. The
aerosol concentration in the wind tunnel was
then allowed to stabilize for at least 15 min
before sampling began.
Once the system was verified as ready for
sampling, the test biosampler and iso rake
samplers were turned on and operated for 1 h. At
the end of the sampling period, the iso samplers
and the test sampler were turned off and the stop
time and the final volume sampled were
recorded for each. Once the experimental runs
were completed, the iso filters, iso nozzles,
biosampler filter, and biosampler inlet unit were
collected from the wind tunnel and prepared for
analysis. Samples were collected from the iso
rake from right to left as standard practice to
avoid mislabeling of samples. Figure 4 shows an
example of the uranine-tagged particles
collected on the iso sampler filter.
Filters were placed in prelabeled centrifuge
tubes with 30 mL of 0.0IN NaOH and sonicated
for 10 min. The internal walls of the iso nozzles
were washed into prelabeled centrifuge tubes,
each with 30 mL of 0.0IN NaOH. The collected
aerosol deposits were extracted and quantified
fluorometrically using procedures described in
Section 2.6.
Figure 4. Isokinetic sampler filter showing collected
uranine-tagged particles.
-------
3.0
Results and Discussion
Sampling efficiency (SE) was calculated for the
biosamplers from the wind tunnel experiments.
SE is defined as the ratio of the measured mean
particle concentration (Cm) to that in the free-
stream air (C0):
SE=-
C0
All concentrations were calculated as mass
concentration of uranine in units of nanograms
per liter (ng/L). C0 was calculated from the iso
sampler results as
where Mfiso was the mass collected on the iso filter
(ng), MWiiso was the mass collected from the iso
nozzle wash (ng), Qiso was the iso calibrated flow
rate (L/min), and t was the sampling time (min).
The measured concentration for the biosampler
was calculated as
Mf
C=- f
Q-t
where A^-was the mass collected on the
biosampler filter (ng), Q was the biosampler
calibrated flow rate (L/min), and t was the
sampling time (min). The inlet concentration for
the biosampler was calculated as
Mf+Mw
c f
Q-t
where Mw was the mass collected from the
biosampler inlet wash (ng).
The SE data were plotted as a function of the
aerodynamic particle size. To determine the
significance of the data, data were plotted as a
function of the square root of Stokes number
(Stk). The Stokes number is a dimensionless
number that describes the ability of particles to
follow the flow of the carrier fluid. Stk is defined
as the ratio of the stopping distance of a particle
to a characteristic dimension of the obstacle, or
Stk = •
where dp is the calculated particle diameter
(|im), pp is the density of the particle (kg/m3), U
is the free-stream velocity (m/s), (i is the
viscosity of air (Pa-s), and D is the inlet diameter
of the sampler (m). The larger the Stk, the less
the particles will follow the air stream lines,
making them less likely to be captured by the
sampler. Since the sampler inlets are not simple
circular openings, D is defined for each sampler
as the diameter of a circle with the same area as
the opening in the sampler inlet. For this
purpose, the dp was calculated using the VOAG
operating conditions, as it is considered more
accurate than measuring the particles
microscopically (Liu et al., 1982). The Stokes
numbers were calculated using AEROCALC, an
Excel spreadsheet calculator for aerosols written
by Dr. Paul Baron (Baron, 2001).
3.1 Aerosol Uniformity Results
Acceptable uniformity was achieved for wind
speeds of 2, 8 and 24 km/h for particle sizes of
5, 10, 15, and 20 um with all samplers tested.
Stairmand disks and axial fans were employed to
homogenize the aerosol within the air stream to
properly challenge the candidate samplers. The
rake, which provides a measure of the aerosol
homogeneity, was reconfigured after the
addition of the mixing instruments. The aerosol
uniformity data are presented in Table 2.
-------
Table 2. Aerosol Uniformity Results
Mass Concentration (ng/mL)
Date
2/10/2011
2/10/2011
2/10/2011
2/11/2011
2/14/2011
3/8/2011
8/24/2011
8/24/2011
8/24/2011
6/7/2011
6/9/2011
6/10/2011
6/8/2011
6/9/2011
6/13/2011
8/9/2011
8/9/2011
8/10/2011
8/19/2011
8/22/2011
8/22/2011
8/18/2011
8/19/2011
8/19/2011
8/18/2011
8/19/2011
7/27/2011
8/5/2011
8/29/2011
8/15/2011
8/15/2011
8/17/2011
8/15/2011
8/16/2011
8/16/2011
dp
(Mm)
5
5
5
5
5
5
5
5
5
10
10
10
10
10
10
10
10
10
15
15
15
15
15
15
15
15
15
20
20
20
20
20
20
20
20
U
(km/h)
2
2
2
8
8
8
24
24
24
2
2
2
8
8
8
24
24
24
2
2
2
8
8
8
24
24
24
2
2
8
8
8
24
24
24
Nozzle
1
11.79
14.32
12.87
8.99
9.27
14.89
5.78
6.19
3.70
6.17
4.85
5.14
11.15
8.13
26.31
25.18
24.22
14.54
14.94
16.62
14.16
56.67
56.14
63.47
21.94
21.81
33.79
5.00
1.50
27.04
22.76
26.01
12.29
14.36
14.27
Nozzle
2
11.79
14.54
12.65
9.19
8.84
13.60
5.33
6.05
3.73
5.66
4.56
4.86
10.09
7.81
26.57
26.98
24.47
14.99
16.46
19.49
16.69
56.23
56.37
61.68
23.94
24.05
37.98
5.43
1.39
26.86
24.39
28.89
15.09
17.45
17.54
Nozzle
3
11.88
14.03
12.49
8.90
8.57
14.80
5.81
6.43
3.96
5.30
4.75
4.81
10.91
8.83
26.22
27.27
25.85
15.64
16.53
18.98
15.22
57.22
56.88
63.36
22.89
24.19
38.22
5.24
1.51
27.31
24.94
30.17
13.50
16.01
15.67
Nozzle
4
12.35
15.13
13.34
8.78
8.38
13.09
6.13
6.82
3.89
6.03
5.03
5.33
12.28
9.52
27.36
30.15
27.51
17.43
14.67
17.57
15.25
57.88
56.72
63.26
25.10
25.20
39.90
5.37
1.56
26.92
25.28
29.95
14.48
17.91
17.97
Nozzle
5
12.16
15.04
13.50
8.71
8.08
13.25
6.09
6.43
3.75
5.72
4.56
4.61
10.44
8.69
28.19
27.15
25.46
15.75
17.56
21.25
18.25
58.75
58.20
65.52
23.71
23.11
37.33
5.96
1.66
29.51
26.82
31.27
12.98
16.01
15.16
Average
11.99
14.61
12.97
8.91
8.63
13.92
5.83
6.38
3.81
5.78
4.75
4.95
10.97
8.60
26.93
27.35
25.50
15.67
16.03
18.78
15.91
57.35
56.86
63.46
23.51
23.67
37.44
5.40
1.52
27.53
24.84
29.26
13.67
16.35
16.12
SD
0.25
0.47
0.44
0.19
0.45
0.86
0.32
0.29
0.11
0.34
0.20
0.28
0.84
0.66
0.84
1.78
1.31
1.10
1.21
1.79
1.59
1.00
0.80
1.37
1.18
1.28
2.25
0.35
0.10
1.12
1.47
2.00
1.13
1.40
1.58
cv
2.1%
3.2%
3.4%
2.1%
5.3%
6.2%
5.5%
4.6%
2.9%
5.9%
4.2%
5.7%
7.6%
7.7%
3.1%
6.5%
5.1%
7.0%
7.5%
9.5%
10.0%
1.7%
1 .4%
2.2%
5.0%
5.4%
6.0%
6.5%
6.5%
4.1%
5.9%
6.8%
8.2%
8.6%
9.8%
-------
3.2 XMX Sampler Results
The total flow rate for the XMX sampler during
the tests was 720 L/min. Five test runs were
performed with 5-(im liquid particles (oleic acid)
at wind speeds of 2, 8, and 24 km/h using the
filter collection option on the sampler. Data
collected for two of the tests with this sampler
were not acceptable because a uniform
distribution of the particles across the STS could
not be achieved. During the runs with acceptable
aerosol distribution, the SE was less than 30%.
The manufacturer, Dycor, stated that the sampler
was not designed for liquid particles and that the
standard EPA tests as defined in this report
would likely not yield representative SEs. In
tests performed under the direction of Dycor, the
SE for 1.9-(im polystyrene latex (PSL) spheres
was reported to be approximately 54% at both 2
and 8 km/h. For 5-(im PSL spheres at 2 and
8 km/h, the results were reported to be 88% and
83.6%, respectively. We suggest retesting the
sampler using solid monodisperse particles
prepared from fluoroscein and ammonium
hydroxide to determine SE.
3.3 PSU Sampler Results
The Hi-Q PSU bioaerosol sampler was
configured to collect samples on 47-mm glass-
fiber filters. The flow rate was set to 100 L/min.
Test runs using 5-, 10-, 15-, and 20-(im particles
at wind speeds of 2, 8, and 24 km/h were
completed in triplicate. Some runs were repeated
when the aerosol uniformity or particle size did
not meet the method criteria. Data for the failed
runs are not included here. SE was calculated
from the analytical data, and the averages are
plotted in Figure 5 and presented in Table 3
along with the square root of Stk.
3.4 DryClone™ Sampler Results
Test runs with the DryClone™ sampler using 5-,
10-, 15-, and 20-(im particles at wind speeds of
2, 8, and 24 km/h were completed in triplicate.
Additional test runs were performed if the data
did not meet QA requirements. These runs were
marked as outliers and the data were not
included in the averages. SE was calculated
from the analytical data, and the averages are
plotted in Figure 6 and presented in Table 4
along with the square root of Stk. In tests
performed under the direction of Evogen, the SE
of the DryClone™ for 5-(im solid particles was
reported to be 93-103%. The SE from our tests
at 2 km/h compares well with the reported
Evogen data.
3.5 Sampling Efficiency
The SE data for each sampler were fitted to a
Gompertz function for regression analysis
(Ratkowsky, 1983). The Gompertz curves are
given by
SE = aQb^5,
where Stk is based on the VOAG particle size
estimate, e is the base of the natural logarithm,
and a, b, and c were fit to the data by regression
with the limits 0 < a < 1, -1< b < 0, and c> 0.
The values for a, b, and c from the regressions
are given in Table 5 for the PSU sampler and
Table 6 for the DryClone™1 sampler.
In Figures 7 and 8, the Gompertz regressions are
plotted against the square root of the calculated
Stk along with the SE data for each particle size
and wind speed. These plots demonstrate the
quality of the overall data by providing a
nondimensional estimate of the performance of
the sampler. The square root of Stk provides a
nondimensional relationship between the
particle relaxation time and the SE. The particle
relaxation time is important because it indicates
how a particle will change its velocity in a
changing flow field. In the case of these
samplers, the particle is accelerating from its
initial velocity (the wind speed) to its collection
velocity in the sampler.
-------
PSU Sampler Efficiency, 2 km/h
100
80
60
en
oo
f 40
E
re
in
•M
c
-------
Table 3. Sampling Efficiencies of the PSU Sampler
Replicate No.
5 um_2 km/h
1
2
3
Average
SD
CV
5 um_8 km/h
2
3
4
Average
SD
CV
5 |jm_24 km/h
1
2
3
Average
SD
CV
10 \im_2 km/h
1
3
4
Average
SD
CV
10 |jm_8 km/h
1
2
4
Average
SD
CV
10 |jm_24 km/h
1
2
3
Average
SD
CV
15 |jm_2 km/h
1
2
3
Average
SD
CV
Date
2/10/2011
2/10/2011
2/10/2011
2/11/2011
2/14/2011
3/8/201 1
8/24/201 1
8/24/201 1
8/24/201 1
2/15/2011
2/17/2011
2/22/201 1
2/17/2011
2/21/2011
2/23/201 1
8/23/201 1
8/23/201 1
8/23/201 1
8/19/2011
8/22/201 1
8/22/201 1
dp
(Mm)
5.42
5.42
5.42
5.42
5.42
5.18
5.01
5.01
5.01
10.36
10.83
10.83
10.83
10.83
10.83
11.64
11.49
11.49
14.45
14.45
14.45
Rake
(ng/L)
0.2654
0.3233
0.2869
0.1230
0.1191
0.1922
0.0624
0.0684
0.0407
0.3721
0.5938
0.4168
0.6794
0.4839
0.6158
0.2193
0.1961
0.1924
0.3547
0.4155
0.3521
Spatial
Uniformity
CV (%)
2.1%
3.2%
3.4%
2.1%
5.3%
6.2%
5.5%
4.6%
2.9%
10.8%
7.5%
7.1%
2.5%
2.3%
4.4%
5.9%
5.0%
5.9%
7.5%
9.5%
10.0%
Sampler
(ng/L)
0.2428
0.3020
0.2548
0.0663
0.1118
0.1988
0.0587
0.0630
0.0339
0.3359
0.4988
0.3502
0.3319
0.3012
0.3460
0.0773
0.0566
0.0662
0.1953
0.2891
0.2243
Sampler
Efficiency
(%)
91.5
93.4
88.8
91.2
2.3
0.03
53.9
93.9
103.5
83.7
26.3
0.31
94.1
92.2
83.1
89.8
5.9
0.07
90.3
84.0
84.0
86.1
3.6
0.04
48.8
62.2
56.2
55.8
6.7
0.12
35.2
28.8
34.4
32.8
3.5
0.11
55.1
69.6
63.7
62.8
7.3
0.12
Square Root
of Stk*
0.0257
0.0257
0.0257
0.0516
0.0516
0.0493
0.0827
0.0827
0.0827
0.0487
0.0487
0.0487
0.1023
0.1023
0.1023
0.1904
0.1880
0.1880
0.0678
0.0678
0.0678
Continued on next page
-------
Replicate No.
15 |jm_8 km/h
1
2
3
Average
SD
CV
15 |jm_24 km/h
1
2
3
Average
SD
CV
20 \im_2 km/h
1
20 |jm_8 km/h
1
2
3
Average
SD
CV
20 |jm_24 km/h
1
2
3
Average
SD
CV
*Stk was calculated
and pp = 933 kg/m3.
Date
8/18/2011
8/19/2011
8/19/2011
8/18/2011
8/18/2011
8/19/2011
8/29/201 1
8/15/2011
8/15/2011
8/17/2011
8/15/2011
8/16/2011
8/16/2011
dp
(Mm)
14.45
14.45
14.45
14.45
14.45
14.45
19.78
19.78
19.78
19.78
19.78
19.78
19.78
using the following values:
Rake Spatial
.„„,,. Uniformity
(ng/L) cv(%)y
0.7915
0.7848
0.8758
0.3173
0.2518
0.2534
0.0169
0.3799
0.3428
0.4038
0.1463
0.1750
0.1726
viscosity of air (/u)
1.7%
1.4%
2.2%
15.0%
5.0%
5.4%
6.5%
4.1%
5.9%
6.8%
8.2%
8.6%
9.8%
at temperature
Sampler
(ng/L)
0.4326
0.4266
0.4262
0.0464
0.0427
0.0290
0.0105
0.1223
0.1100
0.0902
0.0041
0.0043
0.0021
= 293.15 K and
FSamP'er Square Root
Efficiency ^^
54.7
54.4
48.7
52.6
3.4
0.06
14.6
16.9
11.4
14.3
2.8
0.19
62.3
32.2
32.1
22.3
28.9
5.7
0.20
2.8
2.5
1.2
2.1
0.8
0.39
pressure = 101.3kPa,
0.1362
0.1362
0.1362
0.2361
0.2361
0.2361
0.0926
0.1861
0.1861
0.1861
0.3227
0.3227
0.3227
D= 0.145m,
-------
1 in
>•
y
c 1 nn
n
4-f
s u
£ C
Q.
DryClone Sampler Efficiency, 2 km/h
4<^^
^^^ t
^^^
^
) 5 10 15 20 25
Particle Size (urn)
1 in
>»
u
c: i nn
QJ 1UU
'G
^~ on
LJJ
GO fin
°- zin
n
l/l in
^ ZU
01 n
w
(U r
Q. I
DryClone Sampler Efficiency, 8 km/h
4i^^
^^^^
^^^
^^^
^>
1 1 1 1 t
) 5 10 15 20 25
Particle Size {(am)
1 in
>»
u
c i nn
5 luu
*o
— on
LJJ
Q. /in
E 40
CD
rrt Tfl
M
-------
Table 4. Sampling Efficiencies of the DryClone Sampler
Replicate No.
5 um_2 km/h
1
3
4
Average
SD
CV
5 um_8 km/h
1
2
3
Average
SD
CV
5 |jm_24 km/h
1
2
3
Average
SD
CV
10 \\m_2 km/h
1
2
4
Average
SD
CV
10 um_8 km/h
2
3
5
Average
SD
CV
10 |jm_24 km/h
1
2
3
Average
SD
CV
15 |jm_2 km/h
4
5
6
Average
SD
CV
Date
6/1/2011
6/3/201 1
6/3/201 1
6/1/2011
6/1/2011
6/2/201 1
8/8/201 1
8/8/201 1
8/9/201 1
6/7/201 1
6/9/201 1
6/10/2011
6/8/201 1
6/9/201 1
6/13/2011
8/9/201 1
8/9/201 1
8/10/2011
19-Jul
21-Jul
25-Jul
dP
(Mm)
5.42
5.42
5.42
5.42
5.42
5.42
5.02
5.01
5.01
10.36
10.36
10.36
10.36
10.36
10.36
9.84
9.84
11.60
14.67
14.67
14.37
Rake
(ng/L)
0.1544
0.1524
0.1494
0.1315
0.1384
0.1572
0.0601
0.0594
0.0627
0.2556
0.2102
0.2190
0.3029
0.2373
0.7433
0.2928
0.2731
0.1678
0.123
0.200
0.226
Spatial
Uniformity CV
(%)
5.9%
7.9%
5.9%
7.7%
4.7%
6.5%
1 1 .9%
2.0%
3.4%
5.9%
4.2%
5.7%
7.6%
7.7%
3.1%
6.5%
5.1%
7.0%
7.2
12.9
13.7
Sampler
(ng/L)
0.1664
0.1569
0.1497
0.1412
0.1507
0.1542
0.0465
0.0477
0.0562
0.2165
0.1697
0.1782
0.2506
0.1723
0.6540
0.1817
0.1802
0.0773
0.100
0.169
0.185
Sampler
Efficiency (%)
107.8
102.9
100.2
103.6
3.8
0.04
107.3
108.9
98.1
104.8
5.8
0.06
77.3
80.3
89.6
82.4
6.4
0.08
84.7
80.8
81.4
82.3
2.1
0.03
82.7
72.6
88.0
81.1
7.8
0.05
62.0
66.0
46.1
58.0
10.6
0.18
81.8
84.3
81.8
82.1
1.5
0.02
Square Root
of Stk*
0.0219
0.0219
0.0219
0.0439
0.0439
0.0439
0.0706
0.0704
0.0704
0.0415
0.0415
0.0415
0.0833
0.0833
0.0833
0.1372
0.1372
0.1616
0.0575
0.0568
0.0568
Continued on next page
-------
Replicate No.
15 um_8 km/h
2
3
4
Average
SD
CV
15 |jm_24 km/h
1
2
3
Average
SD
CV
20 |jm_2 km/h
1
2
4
Average
SD
CV
20 um_8 km/h
1
2
3
Average
SD
CV
20 |jm_24 km/h
1
2
3
Average
SD
CV
*Stk was calculated
and pp = 933 kg/m3.
Date
7/18/2011
7/19/2011
7/21/2011
7/26/201 1
7/27/201 1
7/27/201 1
8/3/201 1
8/4/201 1
8/11/2011
8/2/201 1
8/3/201 1
8/4/201 1
8/2/201 1
8/2/201 1
8/4/201 1
dp
(Mm)
14.23
14.23
14.23
14.23
14.23
14.23
19.78
19.97
19.78
19.78
19.97
19.97
19.78
19.78
19.97
using the following values:
Rake
(ng/L)
0.6124
0.4462
1.7550
0.1816
0.2673
0.2549
0.0354
0.0183
0.0565
0.3689
0.4102
0.2860
0.0570
0.1007
0.0945
viscosity of air
Spatial
Uniformity CV
(%)
8.2%
5.9%
9.5%
1 1 .2%
6.0%
9.2%
10.4%
18.6%
15.3%
3.7%
8.8%
6.2%
10.1%
12.1%
9.2%
(ju) at temperature
Sampler
(ng/L)
0.4239
0.3102
1.2411
0.0532
0.0757
0.0736
0.0165
0.0095
0.0355
0.1413
0.1610
0.0992
0.0045
0.0086
0.0100
= 293.1 5 K
Sampler
Efficiency (%)
69.2
69.5
70.7
69.8
0.8
0.01
29.3
28.3
28.9
28.8
0.5
0.02
46.6
51.7
62.9
53.7
8.3
0.15
38.3
39.2
34.7
37.4
2.4
0.06
7.9
8.5
10.6
9.0
1.4
0.16
and pressure = 101
Square Root
of Stk*
0.1142
0.1142
0.1142
0.1980
0.1980
0.1980
0.0789
0.0796
0.0789
0.1583
0.1600
0.1600
0.2747
0.2747
0.2774
3 kPa, D = 0.2m,
TableS. Gompertz Parameter Values from Regression Fitting for the PSU Sampler
Parameter
a
b
c
Estimate
1.0
-0.1010
13.1459
Lower 95tn Bound
1.0
-0.1405
10.5746
Upper 95tn Bound
1.0
-0.0615
15.7172
TM
Table 6. Gompertz Parameter Values from Regression Fitting for the DryClone Sampler
Parameter
a
b
c
Estimate
1.0
-0.0500
17.0515
Lower 95tn Bound
1.0
-0.0741
13.7347
Upper 95tn Bound
1.0
-0.0258
20.3682
-------
PSU Sampler
• Sampler Efficiency
» Gompertz Function
Regression
— •* — Lower 95th bound
on a new prediction
— •* — Upper95th bound
on a new prediction
0.25
0.3
0.35
Figure 7. PSU sampling efficiency plotted as a function of the square root of Stk. Regression lines are the
Gompertz functions with values of a, b, and c from Table 5.
1.2 -
1 -
& 0.8
-------
The data quality is indicated by the smoothness
of the curves. The noise indicated by the
distance of the particles off the curve is due to
methodological error. For comparison, the
isokinetic precision tests in Table 2 show CVs
ranging from 1.4% to 9.8%, which represents
the methodological error.
A few comments are in order regarding the
regression fits. First, parameter a is estimated for
both samplers to be 1.0 with a 95% confidence
interval of (1,1), which reflects that this
parameter provides the upper bound on the
efficiency. Note that the efficiency at an Stk of 0
(i.e., the vertical axis intercept) is given by
j, b
a*e .
Although many of the sampling efficiencies at
low Stk05 values were quite close to 1, with
some exceeding 100%, parameter a was
constrained to be less than or equal to 1 (and
greater than 0) in estimating the curves. The
reason is the interplay seen between parameters
a and b in this expression for the intercept at Stk
= 0. (See also the description of the further role
of b below.) Without the restriction of a being
no greater than 1, the interplay between a and b
led to unreliable estimates of both the
parameters themselves and their confidence
intervals for both sampler types. The estimates
obtained for a and b yielded intercepts of 95%
and 90% for the DryClone™ and PSU samplers,
respectively.
Another point worth noting is that residual
analyses indicated that some of the original data
points were suspiciously aberrant. Upon review,
these outlying data points had been collected
during the very earliest phase of sampling and
were more likely to be subject to error than data
collected later after more experience had been
gained with the sampling. Because of the
aberrant data, the Gompertz curves were
estimated after eliminating three data points
from the DryClone™ sampler and one data point
from the PSU sampler from the analyses. The
following data points were deleted:
DryClone™: 20 urn at 2 km/h, replicates 1, 2,
and 4
PSU: 5 um at 8 km/h, replicate 2
The accompanying figures show that the
Gompertz curves fit the data reasonably well.
The estimated curves reflect the sigmoidal
behavior of the data, and the R2 values were
quite high at 90% and 92% for the PSU and
DryClone™1 samplers, respectively.
The Gompertz curve has an inflection point at
The interpretation of the coefficients in this case
is as follows. The b parameter reflects an
exponential decay portion of the behavior, and
the c parameter mitigates the exponential decay
at lower Stk. At lower Stk (i.e., to the left of the
inflection point) the sampler efficiency might be
dominated by properties of the sampler itself,
whereas at higher Stk (i.e., to the right of the
inflection point) the exponential decay,
representing particle impaction, takes over.
-------
-------
4.0
References
Baron, P. 2001. National Institute for Occupational
Safety and Health, Center for Disease Control
and Prevention, Cincinnati, OH 45226, USA,
http://aerosols.wustl.edu/AAARworkshop08/
software/AEROCALC-1 l-3-03.xls, last accessed
March 19, 2012.
Black, J.E. 2011. Evaluation of XMX/2L-MIL
Virtual Impactor Performance and Capture and
Retention of Aerosol Particles in Two Different
Collection Media. Thesis. Department of
Systems and Engineering Management, Graduate
School of Engineering and Management, Air
Force Institute of Technology, Air University,
Air Education and Training Command.
Brockmann, J.E. 2011. Aerosol transport in sampling
lines and inlets (Chapter 6 pp 69-103). In
Aerosol Measurement: Principles, Techniques,
and Applications, Third Edition, edited by
Kulkarni, P.; Baron, P.A.; Willeke, K. John
Wiley & Sons, Inc. DOI: 10.10027
9781118001684.ch6.
Dycor Technologies, Ltd. 2005. XMX/2L-MIL
Operator's Manual, V. 1.5, Rev. August 4, 2006.
Edmonton, Canada. http://www. dycor.com, last
accessed May 9, 2012.
Evogen, Inc., 8301 State Line Road, Suite 101,
Kansas City, MO, http://www.evogen.com/
products/dryclone.html, last accessed
March 20, 2012.
Hi-Q Environmental Products, 7386 Trade St. San
Diego, CA 92121, http://www.hi-q.net/
products/continuous-duty-air-samplers/psu-
series-outdoor-mobile-continuous-duty-
samplers/default.html, last accessed May 23,
2012.
Kesavan, J.S.; Doherty, R.W.; Bottiger, J.R. 2003.
Performance Characterization Methods of
Aerosol Samplers. Aberdeen Proving Ground,
MD: Edgewood Chemical Biological Center.
Liu, B.Y.H.; Pui, D.Y.H.; Wang, X.Q. 1982. Drop
size measurement of liquid aerosols. Atmos.
Environ. 16(l):563-567.
Olan-Figueroa, E.; McFarland, A.R.; Ortiz, C.A.
1982. Flattening coefficients for DOP and oleic
acid droplets deposited on treated glass slides.
Am. Ind. Hyg. Assoc. J. 43:395-399.
Ranade, M.B.; Woods, M.C.; Chen, F-L; Purdue, L.J.;
Rehme, K.A. 1990. Wind tunnel evaluation of
PMio samplers. Aerosol Sci. Technol. 13(1):54-71.
Ratkowsky, D.A. 1983. Nonlinear Regression
Modeling: A Unified Practical Approach. Marcel
Dekker, New York, 276 pp.
Tolocka, M.P.; Tseng, P.T.; Wiener, R.W. (200la).
Optimization of the wash-off method for
measuring aerosol concentrations. Aerosol Sci.
Technol. 34:416-421.
Tolocka, M.P.; Peters, T.M.; Vanderpool, R.W.;
Chen, F.L.; Wiener, R.W. (200Ib). On the
modification of the low flow-rate PM10
dichotomous sampler inlet. Aerosol Sci. Technol.
34:407-415.
U.S. Department of Homeland Security (U.S. DHS).
2009. Bio Watch Field Operations Standard
Operating Procedure. Version 1.2 (FOUO).
U.S. Environmental Protection Agency (U.S. EPA).
1998. Code of Federal Regulations, Title 40, Part
53, Subpart D. National Archives and Records
Administration.
-------
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
------- |