Full text of DOI:10.1080/10473289.2000.10464178

This article was downloaded by: [Adams State University]
On: 24 October 2014, At: 08:08
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of the Air & Waste Management Association
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uawm20
Comparison of the Particle Size Distribution of
Heavy-Duty Diesel Exhaust Using a Dilution Tailpipe
Sampler and an In-Plume Sampler during On-Road
Operation
J. Edward Brown a , Matthew J. Clayton a , D. Bruce Harris b & Foy G. King Jr. b
a ARCADIS Geraghty & Miller, Inc. , Research Triangle Park , North Carolina , USA
b National Risk Management Research Laboratory , Office of Research and
Devleopment, U.S. Environmental Protection Agency , Research Triangle Park , North
Carolina , USA
Published online: 27 Dec 2011.
To cite this article: J. Edward Brown , Matthew J. Clayton , D. Bruce Harris & Foy G. King Jr. (2000) Comparison
of the Particle Size Distribution of Heavy-Duty Diesel Exhaust Using a Dilution Tailpipe Sampler and an In-Plume
Sampler during On-Road Operation, Journal of the Air & Waste Management Association, 50:8, 1407-1416, DOI:
10.1080/10473289.2000.10464178
To link to this article: http://dx.doi.org/10.1080/10473289.2000.10464178
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose
of the Content. Any opinions and views expressed in this publication are the opinions and views of the
authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources of information. Taylor
and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses,
damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

TECHNICAL PAPER
Brown, Clayton, Harris, and King
ISSN 1047-3289 J. Air & Waste Manage. Assoc. 50:1407-1416
Copyright 2000 Air & Waste Management Association
Comparison of the Particle Size Distribution of Heavy-Duty
Diesel Exhaust Using a Dilution Tailpipe Sampler and an
In-Plume Sampler during On-Road Operation
J. Edward Brown and Matthew J. Clayton
ARCADIS Geraghty & Miller, Inc., Research Triangle Park, North Carolina
D. Bruce Harris and Foy G. King, Jr.
National Risk Management Research Laboratory, Office of Research and Devleopment, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina
ABSTRACT
INTRODUCTION
Originally constructed to develop gaseous emission factors
for heavy-duty diesel trucks, the U.S. Environmental Pro-
tection Agency’s (EPA) On-Road Diesel Emissions Charac-
terization Facility has been modified to incorporate particle
measurement instrumentation. An electrical low-pressure
impactor designed to continuously measure and record size
distribution data was used to monitor the particle size dis-
tribution of heavy-duty diesel truck exhaust. For this study,
which involved a high-mileage (900,000 mi) truck running
at full load, samples were collected by two different meth-
ods. One sample was obtained directly from the exhaust
stack using an adaptation of the University of Minnesota’s
air-ejector-based mini-dilution sampler. The second sample
was pulled from the plume just above the enclosed trailer,
at a point ~11 m from the exhaust discharge. Typical dilu-
tion ratios of about 300:1 were obtained for both the dilu-
tion and plume sampling systems. Hundreds of particle
size distributions were obtained at each sampling location.
These were compared both selectively and cumulatively to
evaluate the performance of the dilution system in simu-
lating real-world exhaust plumes. The data show that, in
its current residence-time configuration, the dilution sys-
tem imposes a statistically significant bias toward smaller
particles, with substantially more nanoparticles being col-
lected than from the plume sample.
After decades of focusing its attention on visible emis-
sions, mass emissions, and chemical composition, the U.S.
Environmental Protection Agency (EPA) has recently fo-
cused on another important characteristic of PM emis-
sions: particle size. It has been known for some time that
respiratory systems function much like multistage impac-
tors, collecting the largest particles near the inlet (the
nose), and leaving the smallest particles to deposit in the
deepest parts (the air sacs of the lungs).1 It has also been
known for some time that diesel exhaust contains large
quantities of small particles.2 Most often, these observa-
tions have been primarily qualitative, and the impacts
evaluated in terms of how the environment is affected by
the emissions.
Several advanced analytical tools have become avail-
able to quantify and characterize large numbers of very
small nanoparticles (typically defined as less than 50 nm),
which can number in the millions of particles/cm3 of
ambient air.3 These small particles make a negligible con-
tribution to the traditional mass-based PM measurement
techniques that dominate most current regulations. How-
ever, a number of recent health effects studies indicate
that these particles may be considerably more hazardous
than those making up most of the mass. Therefore, fine
PM (less than 2.5 µm in aerodynamic diameter) is a con-
cern because of its effects not only on the environment
in general, but on people in particular.4,5
IMPLICATIONS
Data gathered during this study suggest that the accu-
mulation mode of diesel exhaust is sensitive to the sample
collection procedure. The transfer of laboratory particle
size data to ambient models may need to be conducted
carefully. Further work is needed to establish the com-
plete particle time history in a truck plume.
A number of laboratories across the United States are
studying fine PM emissions from a variety of sources, in-
cluding diesel engines. Many of these laboratories were
originally created for the study of gaseous pollutants and
other emissions where mass is the quantity of interest.
Their sampling equipment and techniques reflect that
Volume 50 August 2000
Journal of the Air & Waste Management Association 1407

Brown, Clayton, Harris, and King
“collect-and-quantify” approach to emissions measure-
ment. Some laboratories collect samples directly from the
source, condition the samples to a constant and defined
state, and measure the components of interest. Others col-
lect and dilute samples, simulating what happens to point
source emissions after they are released to the atmosphere.
It has been repeatedly demonstrated, during annual Rela-
tive Accuracy Test Audits of regulated emissions monitor-
ing systems all over the United States, that both raw gas
and dilution sampling systems can produce accurate and
comparable data when measuring gaseous pollutants.
Many studies have shown, however, that diesel PM
remains in a state of flux for some time after it is emitted
to the atmosphere, in part due to the continuation of in-
stack coagulation and adsorption, but also due to the sig-
nificant quantities of condensable organics and inorganics
usually present in diesel exhaust.6,7 The fate of these
condensables is significantly affected by atmospheric ag-
ing and dilution of the exhaust stream. As such, the
method of exhaust dilution has fallen under a level of
scrutiny that was not necessary when dilution systems
were first deployed to collect samples for gaseous emis-
sions analysis. The same source sampled by two different
dilution systems could yield fundamentally different fine
PM measurements.
This paper describes a field study in which fine PM
samples were collected and analyzed from a heavy-duty
diesel vehicle in operation. The purpose of this study was
to characterize, under as realistic a condition as possible,
the emissions from a tractor trailer operating on an inter-
state highway in moderately hilly terrain. A secondary
purpose was to evaluate a prototype dilution system for its
ability to effectively simulate real-world dilution of
tractor-trailer exhaust. All of the data for this report were
collected using the On-Road Diesel Emissions Character-
ization (ODEC) facility, constructed and operated by the
National Risk Management Research Laboratory’s Air Pol-
lution Prevention and Control Division.
A number of processes occur during atmospheric ag-
ing that can alter the size distribution of an aerosol, in-
cluding nucleation, agglomeration, and adsorption.
Homogeneous nucleation is the spontaneous formation of
a nanoparticle of volatile material within a locally super-
saturated zone. Heterogeneous nucleation involves the
same volatile material and driving force (i.e., saturation),
but also involves “seed” nuclei that considerably lower the
required degree of saturation. Saturation is the condition
at which the partial pressure of a volatile material equals
its vapor pressure; degree of saturation is usually expressed
as a ratio of those pressures. Once a hot exhaust is released
into the air, the partial pressure of its volatile components
decreases with dilution. The vapor pressure of those com-
ponents is a function of temperature, which also decreases
with dilution. Since the vapor pressure-versus-temperature
relationship is nonlinear, it is normal for the saturation
ratio to reach a maximum at some dilution level. The or-
ganic fraction of diesel exhaust reaches its maximum de-
gree of saturation somewhere between a 5:1 and 50:1
dilution, but does not usually achieve the necessary super-
saturation required for homogeneous nucleation.9 H2SO4,
on the other hand, may reach the supersaturation levels
necessary for homogeneous nucleation at ratios between
10:1 and 50:1, possibly serving as “seed” nuclei for hetero-
geneous nucleation.10 Therefore, residence times at vari-
ous dilution levels can be of critical importance to the
formation of nuclei mode nanoparticles.
The agglomeration of ash particles and carbonaceous
products of incomplete combustion forms a majority of
the accumulation mode particles. This process begins oc-
curring immediately upon detonation within the engine;
those solid particles that were not completely burned be-
gin adsorbing hydrocarbons and sulfates, creating an outer
layer that causes any particles that come into contact to
stick together. Agglomeration rates are likely character-
ized by second-order kinetics, meaning that a 1000:1 di-
lution would decrease the rate by a factor of 1,000,000.
Therefore, a majority of the accumulation mode particles
that are going to form are expected to do so in the ex-
haust piping before any dilution takes place.
BACKGROUND
While a detailed description of diesel PM chemistry and
morphology is beyond the scope of this paper, a general
description is provided. Diesel exhaust PM size distribu-
tions are typically trimodal in shape, meaning that the
frequency distribution will exhibit three peaks that may
or may not overlap one another. Figure 1 shows an exag-
gerated view of such a distribution. In reality, for the typi-
cal exhaust from a modern diesel engine, the coarse mode
would be negligible. The accumulation mode would ac-
count for most of the mass, as shown here, but would
often be dwarfed by the nuclei mode on a distribution
graph where the frequency is expressed as a particle num-
ber concentration.8
The accumulation mode particles play an important
role in determining the characteristics of an aerosol’s size
distribution. Adsorption is an alternative path for the
volatile material that would otherwise be subject to nucle-
ation. Since the driving force for adsorption is much stron-
ger than for nucleation, the presence of a significant
accumulation mode can substantially suppress the for-
mation of nanoparticles, at the expense of having larger
particles in the accumulation mode. Ironically, given the
current mass-based regulatory climate, most of the
industry’s reductions in PM emissions have been taken
out of the accumulation mode.11
1408 Journal of the Air & Waste Management Association
Volume 50 August 2000

Brown, Clayton, Harris, and King
Figure 1. Typical trimodal distribution.
STUDY DESIGN
plume “dilution schedule” data that may serve as a guide-
line in future dilution system designs.
The goal of the heavy-duty diesel on-road testing program
is to characterize emissions under the most realistic con-
ditions attainable. For fine PM emissions, those condi-
tions include the natural processes of ambient dilution
and aging as they affect the exhaust of a tractor trailer.
This study followed two approaches to including the di-
lution and aging process in the fine PM sampling. The
most realistic approach, of course, is to sample the ex-
haust plume after it leaves the stack and has already be-
gun to disperse. However, because this process is
impossible to predict or control, the resulting data would
likely include random variations (as well as gaps, where
the collected sample shows no measurable exhaust im-
pact) that must be accounted for during data processing
and interpretation. A more reliable approach would be
to use a dilution system to simulate what happens to
tractor-trailer exhaust as it traverses the wake of the truck
and disperses into the air (the “direct-dilution” approach).
Following this approach involves having some knowledge
of the real-world process.
EXPERIMENTAL SETUP
Figure 2 shows the core capabilities of the ODEC facility,
which is described in detail elsewhere.12 Additional in-
strumentation deployed for this study included an Elec-
trical Low Pressure Impactor (ELPI) and a dual-channel
0–1000 ppm nondispersive infrared CO2 analyzer. The ELPI
is a semi-continuous analyzer (i.e., one reading/sec) that
quantifies 12 bins of fine PM data in the 32-nm to 10-µm
size range. It provides data comparable to a Scanning
Mobility Particle Sizer (SMPS),13 but for a size range more
suited to accumulation mode particles. The purpose of
the CO2 analyzer is to monitor the dilution ratio of the
sample that flows to the ELPI, from either the ambient
probe or the raw stack dilution system. In either case, it is
necessary to monitor the concentrations of both the
diluent and the sample (as well as the raw exhaust CO2)
in order to quantify the dilution ratio.
The ELPI generated all of the particle size data pre-
sented in this report. This instrument uses a cascade im-
pactor to provide real-time size classification and
quantification of particulate with aerodynamic diameters
from 0.03 to 10 µm. As shown in Figure 3, particles enter-
ing the ELPI are first bombarded with ions created by a
corona discharge. These ions are subsequently removed by
the ion trap, which also removes all particles smaller than
20 nm. The impactor itself removes particles beginning with
The ultimate goal of combining the two sampling
approaches is to have the direct-dilution approach pro-
duce results that agree with plume data. The intermedi-
ate goals reflected in this study are to (1) collect ambient
plume data under controlled conditions that will estab-
lish a “reference” measurement to which direct-dilution
measurement can be compared; (2) assess the adequacy
of a “first-guess” prototype dilution system; and (3) collect
Volume 50 August 2000
Journal of the Air & Waste Management Association 1409

Brown, Clayton, Harris, and King
Figure 2. Schematic of ODEC facility capabilities.
the largest aerodynamic cut point of 10,400 nm. As each
particle is collected, its charge is absorbed by the collection
plate, which is electrically isolated and connected to a
highly sensitive electrometer, which, in turn, is connected
to a computerized data acquisition system (DAS) built into
the ELPI. The impactor is operated at low pressure in order
to increase the Cunningham slip correction factor and,
hence, allow particles as small as 30 nm to attain Stoke’s
numbers high enough to impact.14
The impactor and all other internal components of
the ELPI that contact the sample, including the substrates,
were clean on the first day of testing. The substrates were
Figure 3. ELPI schematic.
1410 Journal of the Air & Waste Management Association
Volume 50 August 2000

Brown, Clayton, Harris, and King
not replaced for the second day of testing but were still
not beyond capacity at the end of testing. The ELPI has a
built-in utility to feed filtered air to the impactor and set
each electrometer to zero offset. This utility was run at
the beginning of each evening’s testing.
The ELPI and dual-channel CO2 analyzer were oper-
ated in two locations during this study. As shown in Fig-
ure 4, the cab location used the dilution system with a
probe inserted directly into the stack. No attempt was
made to sample the stack isokinetically because (1) the
stack flow changes too rapidly to make isokinetic sam-
pling possible, and (2) the particulate of interest is less
than 0.5 µm in diameter and therefore not expected to
experience inertial sampling biases. In the cab location,
the “Diluent” CO2 channel was used to monitor the con-
centration of the dilution air.
The trailer location used a probe of 0.625-in. (1.6 cm)
outside diameter (OD) stainless steel tubing placed 10 cm
above the top of the trailer and 3.8 m from the back of
the trailer (~11 m from the stack when the truck and trailer
are in line). This sample was fed directly to the instru-
ments with no additional dilution (beyond what occurred
in the air between the exhaust pipe and the probe). For
this location, the “Diluent” CO2 channel was connected
via a long run of 0.25-in. (0.6 cm) OD tubing to a probe
near the front of the tractor cab, thus providing a sample
of the dilution gas (i.e., ambient air) that was reasonably
unaffected by the truck’s exhaust.
will contribute to that goal. Lacking experimental data,
we chose to extrapolate a theoretical dilution schedule
using the Kittelson study as a guide.15 The cargo van trailer
should create three distinct stages of dilution: a rapid mix-
ing stage (caused by the turbulent region between the trac-
tor and trailer), followed by a steady dilution stage (similar
to the flatbed’s first stage), and by a second rapid mixing
stage at the rear of the trailer. The selection of the dilution
ratios used in the dilution system is based upon equipment
availability rather than detailed flow modeling.
The dilution system is shown in Figure 6. Two close-
coupled stainless steel vacuum transducer pumps (E1 and
E2) provide the primary dilution. Each of these pumps
draws sample at 65 L/min and mixes it with 115 L/min
of dry, filtered air, providing an approximately 7:1 over-
all dilution ratio and a residence time on the order of 3
msec.16 The second stage of dilution occurs as 1.6 L/min
of the primary diluted sample is introduced through a
0.25-in. tube (D) into a 0.5-in. (1.3-cm) inside diameter
(ID), 1 m long vinyl tube (V), where it mixes with 14
L/min of clean air provided through a pressure regulator
(P) and a rotometer (R). This second stage of the dilu-
tion is designed to provide a 10:1 dilution with a 500
msec residence time. Two more stainless steel vacuum
transducer pumps (E3 and E4) keep the second stage
under vacuum. One of these (E3) pulls 5.4 L/min of the
dilute sample through a 0.060-in. (0.2 cm) orifice (O),
providing another 7:1 dilution with a residence time of
less than 2 msec. The diluted sample is sent through
0.5-in. ID vinyl tubing to one channel of the CO2 ana-
lyzer and to the ELPI. Except for this tube and the vinyl
tube (V) used in the second stage, all components of the
dilution system that contact the exhaust are stainless
steel, to minimize electrostatic losses of the nanoparticles.
The vacuum transducer pumps are powered by com-
pressed dilution air (50 psi or 3.5 kg/cm2) that passes
through a dryer (which also removes most of the CO2)
and through a high efficiency particulate air filter.
PROTOTYPE DILUTION SYSTEM
A literature search on tractor-trailer plume dilution re-
vealed that Kittelson et al. reported the only similar study
in 1988,15 involving a flatbed trailer and a mobile labora-
tory; some of the data were collected with the laboratory
driven behind the truck, and some were collected with
the laboratory riding on the flatbed. The data (Figure 5)
for 50–55 mph roadway conditions suggest two distinct
stages of dilution. From the time it is released, the raw
exhaust (dilution ratio = 1) steadily
dilutes over the course of the next 10
m (~420 msec) to a ratio of 100:1. At
some point between there and 23 m
(~1000 msec), the exhaust plume is dis-
turbed by the vehicle wake, bringing
about a rapid 3:1 dilution beyond what
was already taking place. After that, the
plume continues to dilute at a rate only
slightly slower than what was occur-
ring above the flatbed.
Eventually, we intend to collect
a similar data set and derive a similar
dilution schedule for a cargo van
Figure 4. Sampling locations.
trailer, and the data from these tests
Volume 50 August 2000
Journal of the Air & Waste Management Association 1411

Brown, Clayton, Harris, and King
Turnpike between Pittsburgh and Harrisburg. This section,
which passes through the Tuscarora Mountain tunnel, was
traversed 3 times in both directions during each of four
1-hr test runs. Both tests were conducted with the truck
loaded to its maximum weight of 80,000 lb (36,000 kg)
GCW (gross combined weight). The ODEC facility was
operated in accordance with its facility manual with two
exceptions: gas analyzers were challenged with only two
upscale gas standards, and total hydrocarbon measure-
ments were not included. Both of these procedural modi-
fications were necessary to meet the requirements for
passing through the tunnel (limited to 1500 lb [680 kg]
of compressed gas cylinders, none of them flammable).
The May 18 test was conducted with the ELPI con-
nected to the dilution system. That evening’s testing was
affected by several malfunctions of the DAS computer,
which left several gaps in the gaseous emissions and op-
erating data but did not affect the ELPI data, which came
from a different computer. It was determined, however,
that the dilution system was not functioning properly
during the first hour of testing. As such, that portion of
the ELPI and CO2 data was removed from the processed
dataset. The daylight hours of May 19 were spent relocat-
ing the ELPI/CO2 sampling location to the trailer. The ELPI
functioned normally throughout both evenings’ testing.
SAMPLING DISTANCE, m
Figure 5. Relationship between dilution ratio and sampling distance.
ON-ROAD TESTING SUMMARY
Testing was conducted on two consecutive nights, May
18 and 19, 1999, with both tests extending past midnight.
The weather was fair both nights. The “test loop” con-
sisted of a 9.5-mi (15.3-km) section of the Pennsylvania
RESULTS
Table 1 summarizes the test results. Each ELPI record nomi-
nally represents 1 sec of data. The “Total ELPI Records”
include all of the time that
the ELPI computer collected
data, including “off course”
time spent paying tolls and
turning around. The
“Records w/ Location” are
those records where the
truck is actually “on course”
traveling between toll sta-
tions (the toll stations serve
as markers in the data file
because the velocity drops to
zero). All further analysis is
limited to those records. The
“location” data consist of a
one-dimensional displace-
ment along the test loop;
this displacement was calcu-
lated during data processing
from the ODEC facility’s ve-
locity signal. For the May 18
tests, DAS computer mal-
functions caused gaps in the
location data, which further
Figure 6. Schematic of dilution system.
1412 Journal of the Air & Waste Management Association
Volume 50 August 2000

Brown, Clayton, Harris, and King
reduced the number of “Records w/ Location” available
on that day.
The emissions measured from the tractor plume show
quite a bit of variation from run to run. Even the dilution
system data appear scattered in some areas. Nonetheless,
at least four location intervals exhibited enough repeat-
ability to be considered “representative” of the truck op-
erating at those locations. The intervals are shown on the
top plot, while the bottom plot provides a visual indica-
tion of what the truck was doing (e.g., “uphill” climbing
versus “downhill” motoring) at that location.
The goal was to have the test conditions as similar as
possible between the two test days. That way, the only
variable would be the location of the dilute sampling in-
struments. By driving the truck under the same load at
the same time of day on the same highway, the test
achieved operating conditions that were quite acceptable,
as illustrated by the average speeds. The total miles dif-
fered, of course, because the May 19 test produced twice
as many valid records.
The remaining summary statistics provide the first
glimpse of how comparable the prototype dilution sys-
tem was to ambient dilution. The median dilution ratio
is shown because it better represents the “typical” ratios
that were measured; the mean dilution ratio was too
strongly influenced by outliers, especially during the
ambient plume sampling of May 19. The “typical” PM
measurements are based on measurements near the me-
dian dilution ratios, and are back-adjusted for that dilu-
tion (i.e., the quantities shown indicate pre-dilution
concentrations 381 or 253 times what were actually mea-
sured by the ELPI). This adjustment is performed for com-
parison purposes only, as there is no reason to believe
that either data set actually represents the PM in raw ex-
haust that came out of the stack.
The number concentrations showed up much higher
in the dilution system samples, indicating that the sys-
tem artificially promotes nucleation. As expected, the
mass loading numbers are more similar, given that there
is a similar amount of mass there to begin with (i.e., be-
fore any dilution takes place). Nonetheless, there is a
limit to the amount of information that can be gathered
from summary statistics. So the next step is to inspect
the data more closely.
Figure 7 shows the data organized by location. In the
top two plots, each dot represents one ELPI record. The
purpose of these plots is to assess the repeatability of the
ELPI measurements at various locations along the test loop.
The bottom plot shows the topography of the test loop as
measured with the ODEC’s on-board in-
Figure 8 shows the size distributions for each of the se-
lected intervals. Distributions of this type are typically log-
normal or bimodal/lognormal, hence the logarithmic x-axis.
The y-axis unit, dN/dLogDP, essentially allows bar chart data
to more meaningfully illustrate the true shape of the distri-
bution (by forcing the bar heights to be proportional to their
areas). In these plots, the bars represent the bin data as re-
corded by the ELPI (with no adjustment for dilution). The
smooth curves were derived from the bin data using a
best-fit algorithm. The data were fit to both the bimodal/
lognormal shape shown here, and to a simple lognormal
that would not include the “tail” to the left. In all cases,
the bimodal/lognormal distribution model provided a
better fit, thus illustrating the comparability between these
test data and the distribution shown in Figure 1.
The size distribution does not appear to be substan-
tially affected by the vehicle operating mode. All eight of
the distributions appear similar. It is only when the top
and bottom plots for each mode are compared that a pat-
tern emerges. The more obvious difference is that the 32–
67 nm stage shows more counts (relative to the other bins)
in the direct-dilution data than in the ambient probe data.
This suggests that there may be a more substantial nuclei
mode peak for those data.
A careful inspection of the best-fit curves shows that
the ambient probe accumulation peaks are consistently
farther to the right than the dilution system peaks, by as
much as 20 or 30 nm. This result would be consistent
with the theory that accumulation mode particles tend
to adsorb the volatile material that otherwise might form
nanoparticles by nucleation. If the nuclei mode peak were,
struments. The numerical location iden- Table 1. Testing summary.
tifiers were arbitrarily assigned based on
Sampling Location:
Tractor Stack
5-18-99 21:52:52
5-19-99 02:57:08
11,851
Sampling Location:
Test Start:
Ambient Probe
5-19-99 18:51:00
5-20-99 01:55:45
17,193
the “typical” round-trip distance from the
western toll plaza eastward and back to
Test Start:
Test End:
Test End:
that same toll plaza. The gap between 9.4
and 9.75 is the distance traveled while
Total ELPI Records:
Total ELPI Records:
Records w/ Location:
Average Speed, MPH:
Total Miles:
Records w/ Location:
turning around at the eastern end of the
Average Speed, MPH:
6247
12,888
50.3
51.1
loop (i.e., beyond the toll plaza). A close
inspection of the bottom plot will reveal
that the topography is not perfectly sym-
metrical; this is not unusual for divided
highways in mountainous areas.
Total Miles:
89.0
179.9
Median Dilution Ratio:
381
Median Dilution Ratio:
Typical Particles/cm3:
Typical Loading, g/m3:
253
Typical Particles/cm3:
Typical Loading, g/m3:
2.69E+07
0.62
2.73E+04
0.51
Volume 50 August 2000
Journal of the Air & Waste Management Association 1413

Brown, Clayton, Harris, and King
Figure 7. Emissions and elevation along route.
in fact, smaller for the ambient probe samples, then the
accompanying effect would be slightly larger particles in
the accumulation mode.
showed that the data sets’ 99.9999% confidence inter-
vals do not overlap, meaning that there is virtually no
likelihood that the direct dilution data and the ambient
probe data come from the same population.
Examination of the subtleties of these selected data
points has strongly indicated that a bias exists and is
consistent with current theory on the formation and
transformation of fine PM. Nonetheless, to draw a more
defensible conclusion, the analysis should be expanded
to include the entire data set. Since the most direct indi-
cation of nuclei mode presence is the 32–67 nm bin,
those values are a logical choice for a comparison vari-
able between the ambient plume samples and the dilu-
tion system samples. For each data set, these last stage
particle counts were calculated as a proportion of total
counts. The resulting data were used to generate the his-
tograms shown in Figure 9. Obviously, the histograms
do not match up very well. In fact, a statistical analysis
CONCLUSIONS AND RECOMMENDATIONS
This study successfully captured and characterized the
plume of a heavy-duty tractor trailer while operating at
highway speeds. The characterization included dilution
ratio, particle size distribution, and overall variability.
Data collected from the prototype dilution system,
however, show that it has not adequately simulated the
ambient dilution that occurs in tractor-trailer exhaust. It
is likely that allowing 500 msec for the sample to dilute
from 7:1 to 70:1 leaves too much time for nucleation to
occur. This study’s plume data indicate that, for a cargo
van configuration, the ratio has already reached ~250:1
1414 Journal of the Air & Waste Management Association
Volume 50 August 2000

Brown, Clayton, Harris, and King
Figure 8. Average particle distributions.
by 500 msec. Extrapolating from Kittelson’s flatbed study15
shows that the dilution ratio surpasses 50:1 (the top end
of the “critical range” for nucleation) at about 85 msec.
Further characterization of the particles and dilution ra-
tio in the plume, as it develops along the trailer and dis-
perses into the wake, is needed to define a performance
envelope for any dilution sampling system.
larger complement of equipment can be deployed in any
combination of five sampling locations: the tractor stack
(with dilution system), the 11-m trailer location (used
in this study), and three intermediate trailer locations at
1.5, 5.5, and 7.5 m. The accessory hardware, dilution
measurement instruments, and available computers can
support up to three ELPI sampling locations (one in the
cab, two in the trailer). This additional hardware will
provide both concurrent data and enhanced plume par-
ticle characterization.
One of the difficulties involved in this study was
drawing meaningful comparisons between nonconcur-
rent data sets. Since this test series was conducted, the
project team has acquired an additional ELPI and addi-
tional instrumentation for measuring dilution. This
One of the biggest shortcomings of the ELPI is its lower
size limit of 32 nm, which does not allow measurement of
Volume 50 August 2000
Journal of the Air & Waste Management Association 1415

Brown, Clayton, Harris, and King
Figure 9. Plume vs direct dilution distributions.
11. Abdul-Khalek, I.S.; Kittelson, D.B.; Graskow, B.R.; Wei, Q. Diesel Ex-
haust Particle Size: Measurement Issues and Trends; SAE Paper 980525;
Society of Automotive Engineers: Warrendale, PA, 1998.
12. Harris, D.B.; King, F.G., Jr.; Brown, J.E. Development of On-Road Emis-
sion Factors for Heavy-Duty Diesel Vehicles Using a Continuous Sam-
pling System. In The Emission Inventory: Programs and Progress;
A&WMA: Pittsburgh, PA, 1996; VIP-56, pp 290-308.
13. Maricq, M. Vehicle Particulate Emissions: A Comparison of ELPI, SMPS
and Mass Measurement. Presented at International Seminar on Par-
ticle Size Distribution Measurement From Combustion Engines,
Espoo, Finland, May 1998.
the nuclei mode. Other researchers have used an SMPS
system to explore the nuclei mode on dynamometer test
stands. We have not been able to successfully implement
those test procedures during on-highway operations. Sev-
eral investigators and instrument manufacturers are cur-
rently addressing this problem. Successful results of their
efforts will be incorporated into our project.
14. Electrical Low Pressure Impactor product literature; Dekati Ltd.,
Tampere, Finland, 1999.
15. Kittelson, D.B.; Kadue, P.A.; Scherrer, H.C.; Lourein, R.E. Character-
ization of Diesel Particles in the Atmosphere; Coordinating Research
Council AP-2 Project Group Final Report; Coordinating Research
Council: Atlanta, GA, 1988.
16. Air-Vac product literature; Air-Vac Engineering Co. Inc., Milford, CT,
1999.
REFERENCES
1. Lippmann, M. Size-Selective Sampling for Inhalation Hazard Evalua-
tion. In Fine Particles: Aerosol Generation, Measurement, Sampling, and
Analysis; Liu, B.Y.H., Ed.; Academic Press: Orlando, FL, 1976; pp 287-
310.
2. Kittelson, D.B.; Dolan, D.F.; Verrant, J.A. Investigation of a Diesel Ex-
haust Aerosol; SAE Paper 780109; Society of Automotive Engineers:
Warrendale, PA, 1978.
3. Graskow, B.R.; Kittelson, D.B.; Abdul-Khalek, I.S. Characterization of
Exhaust Particulate Emissions from a Spark Ignition Engine; SAE Paper
980528; Society of Automotive Engineers: Warrendale, PA, 1998.
4. Bagley, S.T.; Baumgard, K.J.; Gratz, L.D.; Johnson, J.J.; Leddy, D.G.
Characterization of Fuel and Aftertreatment Device Effects on Diesel Emis-
sions; Research Report Number 76; Health Effects Institute: Topsfield,
MA, 1996.
5. Samut, J.M. The National Morbidity Mortality and Air Pollution Study;
Health Effects Institute: Topsfield, MA, 1996.
About the Authors
6. Kittelson, D.B.; Abdul-Khalek, I.S. Formation of Nanoparticles dur-
ing Exhaust Dilution. In Proceedings of 2nd ETH Workshop on
Nanoparticle Measurement; Zurich, Switzerland, August 1998.
7. Ahlvik, P.; Ntziachristos, L.; Keskinen, J.; Virtanen, A. Real Time Mea-
surements of Diesel Particle Size Distribution with an Electrical Low Pres-
sure Impactor; SAE Paper 980410; Society of Automotive Engineers:
Warrendale, PA, 1998.
8. Kittelson, D.B. Engines and Nanoparticles: A Review; J. Aerosol Sci.
1998, 29(5/6), 575-588.
9. Kittelson, D.B.; Watts, W.F., Jr.; Arnold, M. Aerosol Dynamics, Labo-
ratory and On-Road Studies. EPA grant report; National Vehicle and
Fuels Emission Laboratory: Ann Arbor, MI, July 1998 (accessed at http:
//www.me.umn.edu/centers/cdr/Proj_EPA.html).
J. Edward Brown and Matthew J. Clayton are senior engi-
neer and engineer, respectively, at ARCADIS Geraghty &
Miller, P.O. Box 14109, Research Triangle Park, NC 27709.
D. Bruce Harris (corresponding author, phone: 919-541-
7807; e-mail: harris.bruce@epa.gov) and Foy G. King, Jr.,
are senior environmental engineer and senior mechanical
engineering technician, respectively, for the Air Pollution
Prevention and Control Division of the National Risk Man-
agement Research Laboratory, Mail Drop 61, Research Tri-
angle Park, NC 27711.
10. Kittelson, D.B.; Arnold, M.; Watts, W.F., Jr. Review of Diesel Particulate
Matter Sampling Methods: Final Report. EPA grant report; (accessed at
http://www.me.umn.edu/centers/cdr/Proj_EPA.html).
1416 Journal of the Air & Waste Management Association
Volume 50 August 2000