United States
                    Environmental Protection
Air and Energy Engineering
Research Laboratory
Research Triangle Park, NC 27711
                    Research and Development
EPA/600/SR-93/019  March 1993
&EPA       Project  Summary

                    Characterization of PM-10
                    Emissions from Antiskid
                    Materials Applied To  Ice  and
                    Snow-Covered  Roadways
                    John S. Kinsey
                     Several areas of the U.S. in violation
                    of the National Ambient  Air Quality
                    Standard for PM-10 (participate matter
                    with diameters < 10 IJJTI) have conducted
                    studies that have Identified the resus-
                    pension of antiskid material applied to
                    paved roadways as an important source
                    of PM-10. The application of antiskid
                    materials creates a temporary but sub-
                    stantial increase in the amount of fine
                    part leu late on the  road surface over
                    and above that which  Is  normally
                    present. Measured  emission  data are
                    lacking for all types of antiskid materi-
                    als; therefore, an appropriate  field pro-
                    gram whose objective was to  establish
                    a predictive model for PM-10 emissions
                    was undertaken. A source-oriented
                    emissions sampling procedure  was
                    conducted on a section of US 53 just
                    west of Duluth, MN, during March/April
                    of 1992. The measured emission fac-
                    tors varied from 1 to 11 g/VKT (vehicle
                    kilometer traveled)  for the three tests
                    conducted. The data were not sufficient,
                    however, to develop any specific cor-
                    relation  between the measured emis-
                    sion factors and source  parameters.
                    The only general observation made was
                    PM-10 emissions appear  to  increase
                    with the  amount of antiskid material
                    applied. A comparison of measured
                    emission factors with those  predicted
                    by an EPA compilation of air pollutant
                    emission factors (AP-42) indicated that
                    most of the measured factors are higher
                    than those that would be predicted from
                    silt-loading  values  alone.  Due  to the
                    marginal test conditions during storms,
                    no  definitive assessment of this in-
crease can  be  made until  additional
data are obtained.
   This Project Summary was devel-
oped by EPA's Air and Energy Engi-
neering Research Laboratory, Research
Triangle Park, NC, to announce key find-
Ings of the research project that Is fully
documented In a separate report of the
same title (see Project Report ordering
Information at back).

  Several areas of the U.S. in  violation of
the National Ambient Air Quality Standard
(NAAQS) for PM-10 (airborne particles <10
u,m in  diameter)  have conducted studies
to determine the sources of these emis-
sions.  One source of PM-10 emissions
identified in these studies  is  the
resuspension of antiskid  material applied
to paved roadways. Antiskid materials may
consist of abrasives (e. g., sand, stone, or
cinders applied to the road surface to im-
prove traction) or deicers, which  restore
pavement traction by preventing the for-
mation of  ice films, weakening  the  ice/
pavement bond, and/or by melting ice and
  The application of antiskid materials cre-
ates a temporary but substantial increase
in the amount of fine particles on the road
surface over and above that which is  nor-
mally present. Prior research  has estab-
lished  a direct  relationship between the
loading of silt-size fines (particles < 75 \im
in physical diameter) and the PM-10 emis-
sion generated  by vehicular traffic.  "The
empirical relationship between silt loading
and PM-10 emissions is reflected in EPA-
                                                                   Printed on Recycled Paper

 recommended PM-10 emission factors for
 paved urban roads.
   In  a recent EPA  study,  a literature
 search, an  engineering  analysis,  and a
 laboratory testing program were performed
 to provide air pollution control agencies
 with information on how to identify  an ap-
 propriate antiskid  material that is both du-
 rable and effective and produces  lower
 PM-10 emissions.  The primary objectives
 of this study were  to provide guidance on
 methods to  determine (a)  the  physical
 properties and durability of antiskid mate-
 rials selected for  use on ice- and  snow-
 covered  roadways,  and (b)  criteria for
 defining the elements of an effective PM-
 10 emission control strategy associated
 with use of antiskid materials.
   Although the above study provided guid-
 ance for the selection  of antiskid materi-
 als,  no direct information was developed
 regarding the actual PM-10 emissions re-
 lated to their use, the changes in surface
 silt loading  resulting from such applica-
 tion,  or the degree of  control actually
 achieved by compliance with  the material
 selection criteria developed in the  study.
 Measured emissions data are lacking for
 all types  of  antiskid material, but deicing
 chemicals have received the least amount
 of attention. Therefore, an appropriate field
 program was needed to establish a  suit-
 able method for predicting PM-10 emis-
 sions from the use of  antiskid materials.
 This is the primary objective  of the work
 reported here.
  Source-oriented  emissions sampling
 was employed in the program using Mid-
 west  Research Institute's (MRI's) "expo-
 sure profiling" approach. The data obtained
 by this technique were coupled  with the
 results of road surface  sampling and ma-
terials application  data in an  attempt to
develop a method for predicting PM-10
 Site Selection

   The original test site  was located on
 northbound US 53 at Mile Post (MP) 13.35
 (or approximately Station  No. 452+50).
 This particular location has good orienta-
 tion with respect to the wind direction an-
 ticipated for  periods  after  storms, good
 exposure to  ambient  winds (i.e.,  lack  of
 trees in the upwind direction), and  a rela-
 tively flat median for installation of the air-
 sampling equipment. However, due to the
 mild winter in Duluth, the original test site
 could not be  used. The site was selected
 on the assumption that the ground would
 remain  frozen throughout  the sampling
 period,  allowing equipment to be safely
 placed on the median. When the ground
 remained thawed, the sampling site  was
 moved  approximately 1/4 mi  (0.4  km)
 northbound (i.e., to the west) to an area
 near  the  intersection of US 53  with  a
 county  road.   This area  provided  a  firm
 surface for the sampling equipment but
 had a substantial stand of trees bordering
 the northern   portion of  the right-of-way
 which did  not conform to criteria  estab-
 lished by  MRI for exposure profiling. For
 this and other related reasons, only one
 of the three  test series  was  conducted
 under suitable conditions with respect to
 wind  speed and direction. This noncom-
 pliance with established criteria adversely
 affected data  quality as discussed below.
 Overall Study Design
  The source-directed field sampling con-
 ducted in this study employed the expo-
 sure profiling  approach to quantify source
 emission contributions. The exposure pro-
filing technique for particulate source test-
 ing is based  on  the  isokinetic profiling
concept used in conventional (stack) test-
 ing. The passage of airborne pollutant im-
 mediately  downwind  of the  source is
measured directly  by  means of simulta-
 neous multipoint sampling over the effec-
 tive cross section of the dust source plume.
 This  technique uses a  mass flux mea-
 surement scheme similar to EPA Method
 5 stack testing rather than requiring  indi-
 rect emission rate calculation through the
 application of a generalized atmospheric
 dispersion model.

 Air Sampling Equipment
  For measurement of particulate emis-
 sions from the test  road, a vertical  net-
 work of samplers (Table 1) was positioned
 5 m downwind and 10 m upwind from the
 edge of the pavement. Four downwind
 vertical sampling  arrays  (D1 through D4)
 and two vertical upwind arrays were used.
 Three of the downwind  arrays (D1, D3,
 and D4) and  one upwind array (U2) made
 use of high-volume  (hi-vol)  air samplers
 equipped with constant-flow electronic con-
 trollers and cyclone  pre-separators.  The
 remaining two arrays (D2 and U1) used
 hi-vols equipped with Wedding PM-10 in-
 lets and critical orifice flow controllers.

 Testing Procedures
  Prior to actual sample collection, a num-
 ber  of decisions  were  made as  to the
 potential for acceptable source-testing con-
 ditions. These decisions were based on
 forecast information  obtained either from
 the local U.S. Weather Service office and/
 or from the Roadway Weather Information
 System (RWIS) located at the site. If con-
 ditions were  considered acceptable, the
 sampling equipment was prepared for test-
 ing. Pretest preparations included calibra-
tion of the various air sampling instruments
 and insertion  of filters. All sampling activi-
ties followed  quality  control  (QC) proce-
dures approved by EPA prior to testing.
  The particulate samples were collected
on Type AH grade glass fiber filters. The
filters were analyzed gravimetrically at a
constant  temperature and humidity in  a
Table 1. Sampler Deployment
Sampler array ID

D1, D3, D4


' Certain downwind sat
No. of



mnlinn Ofininmont muM nnt K
height(s) (m)

1, 3, 5, 7



Type of sampler
or instrument
Hi-vol +
Wedding inlet
Hi-vol + Cyclone
Hi-vol + Cyclone

Warm wire
Wind vane
Hi-vol +
Wedding inlet
PM-10, Pb

PM-10, Na+,CI
(Selected arrays
Wind velocity

Wind direction
PM-10, Pb

special laboratory. Selected filters were
also extracted and  chemically analyzed
by  an  independent  laboratory to deter-
mine the concentration of chloride (Cr),
sodium (Na*), or lead (Pb) in the particu-
late sample collected. Appropriate quality
assurance/quality control (QA/QC) proce-
dures were  used throughout all sample
Ancillary Sampling and
  The types of ancillary samples and in-
formation collected in the program fall into
two broad categories:  antiskid materials
and roadway surface samples, and source
activity levels.  Each  is described below.
  In conjunction with the emissions tests,
samples were  taken of  the antiskid mate-
rial (abrasive/rock salt mix  and straight
rock salt) applied to the road and the dust
remaining on  the surface after it dried.
These  samples were needed not  only to
evaluate the performance of existing emis-
sion models but also to develop improved
models for  antiskid materials. Antiskid
samples were  analyzed for silt and mois-
ture content as well as a  number of key
physical and chemical  properties thought
to be important in the generation of silt
loading (e. g., mass of material < 75 u,m in
physical diameter) on  the  road  surface.
Road surface  samples  were analyzed for
silt content using established MRI proce-
  Source  extent and activity data were
collected with  a variety of tools. For ex-
ample, in addition to visual observation
and note taking, pneumatic traffic counters
were used to determine source activity on
US 53. A radar gun was  used to deter-
mine the average speed of vehicles pass-
ing the sampler array.
   In conjunction with  the program,  de-
tailed information was  collected indepen-
dently  by the Minnesota  Department of
Transportation (MNDOT) on the condition
of the  weather and  pavement during the
course of the storm, the types and amounts
of  antiskid materials applied to the test
road, and a general indication of the re-
sidual  deicing chemical on the road sur-
face at various  times. These  data were
used to augment the information obtained
on source activity mentioned above. Addi-
tional  surface sampling  was  conducted
between storms to  develop a silt-loading
"history" during data analysis.

Calculation Procedure
   To calculate  emission  rates from  the
measured  PM-10 concentrations (mass/
volume), a conservation of mass approach
was used. The  passage of airborne par-
ticulate (i.e., the quantity of emissions per
unit of source activity) was  obtained by
spatial integration of distributed measure-
ments of exposure (mass/area) over the
effective cross section of the plume. Ex-
posure is the point value of the flux (mass/
area-time) of airborne paniculate integrated
over the time of measurement, or equiva-
lent ly, the  net paniculate mass  passing
through a unit area normal to the mean
wind direction during the test. To be con-
sistent with the NAAQS, all concentra-
tions  and flow  rates were expressed  in
standard conditions (25C and 101 kPa or
77F  and 29.92 in. Hg).

Field Sampling Program

Source Description and Activity
  The test site used in the experimental
program  was located  on northbound US
53 on the outskirts of metropolitan Duluth,
MN. US  53 is an unlimited-access,  four-
lane, high-speed roadway carrying  com-
muter traffic to  and from Duluth at an
approximate volume of 5,000 vehicles/day.
Data collected  during  field sampling
showed  that most  traffic was  two-axle,
light-duty  vehicles traveling  between 88
and 97 km/h (55 and 60 mph).  Normal
surface loadings, determined both visually
and by sampling, were generally very low,
with silt loadings in the range of 0.2 g/m2.
  Exposure profiling was performed after
two  minor  storms that occurred  on  April
10, and  on April 21 and 22, 1992. One
test series  was conducted  on April 11,
with additional testing on April 23 and 26.
  During the April 10 storm, approximately
4 to 6 in. (10 to 15 cm) of wet snow fell on
US 53 in 24 h.  Four applications of  an
abrasive/salt mixture  were made to the
northbound lanes over a period of ap-
proximately 10 h totaling 395 kg/lane-km
(1,400  Ib/lane-mi).  During  the  second
storm, a combination  of wet snow and
freezing  rain fell during  approximately a
36-h  period. In this case, only one appli-
cation of 197 kg/lane-km (700 Ib/lane-mi)
was made to each lane especially for test
  As determined in previous  research, the
application rates used by MNDOT are far
below those of most other transportation
agencies. Also,  much of the material ap-
plied was either cast off by the snow plow
during clearing  operations or eliminated
by  melted  precipitation. The material that
remained on the surface was quickly lost
to the atmosphere by the action of pass-
ing vehicles. Thus, surface loadings were
generally low,  with  higher  loadings ob-
served in the passing (left) lane, com-
pared to the driving (right) lane.

Exposure Profiling Results
  The results of the three  PM-10 expo-
sure-profiling tests are shown in Table 2.
These results were calculated  using  the
outcome  of the  gravimetric analyses per-
formed to obtain the measured (i.e., blank-
corrected) PM-10 concentration at each
sampling location. Net (i.e., upwind-cor-
rected) PM-10  concentrations were then
determined  at  each vertical  sampling
height by subtracting estimates from a
straight line fit  of the measured  upwind
concentrations.  (Note that net concentra-
tions were calculated only for arrays with
adequate information for integration pur-
poses.) Using these net concentrations,
the PM-10 exposure was calculated for
each measurement location  and exposure
integration performed using a two-step pro-
cess.  Finally, PM-10 emission factors were
calculated from  the data.

Results of Chemical Analyses
  As  mentioned above, selected filters (in-
cluding blanks)  were submitted for chemi-
cal analysis for either Pb or Na*  and Cr
content.  The concentration  of  paniculate
Pb was determined both upwind and down-
wind  of the road during Tests AY-4 and
AY-5, using samples collected by the two
Wedding PM-10  instruments. (Similar
analyses could  not be performed for Test
AY-3  due to a downwind sample invali-
dated by generator problems.) In the case
of Na* and Cr,  filter  sets from one down-
wind  profiler array for each of the three
tests  were submitted for chemical analy-
sis. The purpose of these analyses was to
determine the relative contribution  of rock
salt to the total  PM-10 emissions from the
  As  shown by the experimental data, the
analyte mass found on most of the filters
does  not exceed the blank values  for any
of the three species. Of the  few filters
which show a net increase in analyte mass
over the blank value, the quantity deter-
mined is generally so slight as to be negli-
gible.  Therefore, for the purpose of this
study, it was assumed that the contribu-
tion of both Pb and NaCI to the total PM-
10 emissions  from the road was  not
significant and  thus could be ignored. For
this reason,  no specific emission  factors
were  developed for  either Pb or  NaCI in
the current program.
Results of Ancillary Sampling
and Analysis
   Samples of both the abrasive/rock salt
mixture  and straight rock salt were  col-

  Table 2.  Results of Emission Factor Calculations'

AY -4


""" f*> J""" 	




(\ig/std m3)"
1 .11
Integrated No. of
exposure vehicle
(m-\ig/cm*) c passes
459 1,175

1,038 983

31.8 220

149 650





1 Parentheses denote inter/extrapolated values. ~~~ 	    	
   , w^, ~, u OUWM m,u,ai ow ** uw trm uo mssn measured oownwina ana upwina concentrations. Upvrind data inter/extrapol
                    ""                '^^^ fmm0to1m ^9*. with Simpson's rule used for integration be^een
 lected from the MNDOT storage piles and/
 or spreader trucks. These samples were
 then analyzed for moisture and silt con-
 tent as well as various key material prop-
   Using the  data obtained  for the abra-
 sive/salt mixture, a comparison was made
 of properties with EPA  material selection
 criteria. This comparison showed that the
 abrasive/salt mixture used by MNDOT did
 not meet three out of the four key param-
 eters thought to be important for high ma-
 terial  durability.  For  this  reason, the
 abrasive/salt mixture analyzed in the cur-
 rent program  was classified as being  of
 questionable quality, based on the  results
 of previous testing.
   Road  surface sampling was also per-
 formed throughout the  period that field
 testing was  attempted  in  Duluth, MN.
 Samples were collected and analyzed  to
 determine silt loading using accepted pro-
 cedures. Surface samples were collected
 from both the  driving (right) and passing
 (left)  lane on  the  northbound and
 southbound sides of the median near the
 air sampling site with the majority  of the
 sampling performed  on  the northbound
   Using the silt loading data. PM-10 emis-
 sion factor  estimates were calculated for
the various samples using the  AP-42 pre-
dictive model.  Emission factors were pre-
dicted  for both northbound (abrasive/salt
mixture) and southbound (rock salt only)
 lanes in terms of grams per vehicle kilo-
 meter traveled (g/VKT).
   As shown in Table 3, predicted emis-
 sion  factors, based on silt-loading data,
 were generally low, ranging from less than
 1 to a maximum of 4 g/VKT. The highest
 emission factors (resulting  from the high-
 est silt loadings) are for samples collected
 from  the northbound  lanes (abrasive/salt
 mixture) after a fairly major storm on Feb-
 ruary 24, 1992. (Note that this storm oc-
 curred priono field testing. After that time,
 the predicted emissions generally drop to
 < 1  g/VKT until  the  next storm in early
 April. This decrease in loading would be
 expected due to the effects of traffic, which
 tends to reestablish a silt-loading "equilib-
 rium" on  the road surface after external
 influences (e.g., the application of antiskid
 material)  have  been  eliminated. As also
 found, the estimated emission factors are
 of the same approximate  magnitude as
 the test results provided in Table 2  This
 lends some  additional credibility to the
 data obtained by  exposure profiling in the
 current study.

 Discussion of Results
  As shown by the above test results, the
 measured emission factors varied from 1
to 11 g/VKT for the three tests conducted.
The data were  not sufficient, however, to
develop any specific correlation  between
the measured emission factors and source
parameters such  as quantity of antiskid
 material applied,  time  since  application,
 and silt loading. The only general obser-
 vation that can  be made from the data is
 that the PM-10 emissions appear to  in-
 crease with the amount of antiskid mate-
 rial applied during the two storms tested
 (i.e.,  the emission factors for Test AY-3
 are higher than for Tests AY-4 and AY-5).
 (Also note that  Test AY-5 had  the most
 acceptable  wind conditions of the three
 tests conducted and thus  represents the
 most reliable  emission  factor  determined
 in the study.) Otherwise,  the results do
 not seem to follow a consistent relation-
 ship.  A major cause for the lack of asso-
 ciation  was the marginal test conditions
 discussed above.
   A comparison of  the measured  emis-
 sion factors (Table 2) with those predicted
 by the current AP-42 equation (Table  3)
 also yields some interesting results. With
 only a relatively few exceptions, most  of
 the measured factors are higher than those
 which would be predicted from silt-loading
 values alone In  the case of the test (i.e.,
 northbound) lanes themselves, the mea-
 sured factors  either  equal  or exceed the
 lane average  emissions predicted by the
 AP-42 equation, except for the February
 26 samples that were  collected after a
 fairly major storm.  This suggests that the
 application of antiskid material results in a
 short-term increase  in PM-10 emissions
from the roadway  in  an amount greater
than that which would be predicted by silt

Table 3. Predicted PM- 10 Emission Factors for Measured Surface Silt Loadings 









Lane sampled
NB Driving
NB Passing

SB  Driving
SB Passing

NB Driving
NB Passing

SB Driving
SB Passing

NB Driving
NB Passing

SB Driving
SB  Passing

NB Driving

NB Passing

SB Driving
SB Passing

NB Driving

NB Driving *
NB Passing"

NB Driving "
NB Passing *

NB Driving
NB  Passing

SB-Driving and Passing

Road surface silt
loading (g/nf)













PM-10 emission
factor (g/VKT)













 PM- 10 emission factors predicted using the AP-42 predictive equations and silt-loadings obtained in
the program.

" Loadings measured after application of antiskid material.

loading alone. Although no definitive as-
sessment of this increase can be made
based on available data, the PM-10 emis-
sions could be almost a factor of 3 higher
(i.e., 11 vs. 4 g/VKT) than would be pre-
dicted from the silt loading. This propor-
tional increase in emissions is considerably
higher than the precision factor of ap-
proximately 2 for the current AP-42 pre-
dictive equation and thus worthy of further
study. Therefore, a more explicit evalua-
tion would help determine the extent of
the emissions increase due to the appli-
cation of antiskid material to paved road-
The following conclusions were reached
as a result of the current study:
1. Because the measurements con-
ducted were performed under diffi-
cult environmental conditions that
did not meet all of the applicable
QC criteria for exposure profiling,
data quality was adversely affected.
2. Although the emission factors de-
termined in the program ranged
from 1 to 1 1 g/VKT for the three
tests conducted, these values
should be used with extreme cau-
tion due to the lack of suitable test
conditions during sample collection
and resulting poor data quality.
3. The contribution of both particulate
Pb and NaCI to the total PM-10
emissions from the road did not
appear to be significant based on
the limited chemical analyses per-
formed during the program.
4. The dry silt loadings determined on
US 53 tended to (a) be relatively
low and (b) drop off rapidly after a
storm, which is typical of high-speed
roadways. The amount of silt avail-
able for resuspension was also
found to be very low due to the
minimal application of antiskid ma-
terial during the experimental pro-
5. Emission factors determined in the
program were generally higher than
those predicted by the current AP-
42 equation, based on road sur-
face silt loading. The magnitude of
this increase could not be defini-
tively determined from available
data, but it may be almost a factor
of 3.
U.S. Government Printing Office:  1993  750-071/60218



  J. S. Kinsey is with Midwest Research Institute, Kansas City, MO 64110-2299.
  Charles C. Masseris the EPA Project Officer (see below).
  The complete report, entitled "Characterization ofPM-10 Emissions from Antiskid
    Materials  Applied To Ice and Snow-Covered Roadways," (Order No. PB93-
    150209; Cost: $19.50; subject to change) will be available only from
          National Technical Information Service
          5285 Port Royal Road
          Springfield, VA 22161
          Telephone: 703-487-4650
  The EPA Project Officer can be contacted at
          Air and Energy Engineering Research Laboratory
          U.S. Environmental Protection Agency
          Research Triangle Park, NC 27711
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268

Official Business
Penalty for Private Use
   PERMIT No. G-35