vvEPA
United States      Industrial Environmental Research  EPA-600/7-79-186
Environmental Protection  Laboratory    \      August 1979
Agency        Research Triangle Park NC 27711      L > '

Setting Priorities
for Control of
Fugitive Particulate
Emissions from
Open Sources

Interagency
Energy/Environment
R&D  Program Report
                                        * "

-------
                  RESEARCH REPORTING SERIES


Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

    1.  Environmental Health Effects Research

    2.  Environmental Protection Technology

    3.  Ecological Research

    4.  Environmental Monitoring

    5.  Socioeconomic Environmental Studies

    6.  Scientific and Technical Assessment Reports (STAR)

    7.  Interagency  Energy-Environment Research and Development

    8.  "Special" Reports

    9.  Miscellaneous Reports

This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort  funded under the 17-agency  Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems.  The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments  of, and development of, control technologies for  energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
                       EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for  publication. Approval does not signify that the contents necessarily reflect
the  views and policies of the Government, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.

This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                     EPA-600/7-79-186

                                            August 1979
 Setting  Priorities for  Control
      of  Fugitive  Particulate
Emissions from  Open  Sources
                       by
          D.W. Cooper, J.S. Sullivan, Margaret Quinn,
             R.C. Antonelli, and Maria Schneider

                  Harvard University
                School of Public Health
              Boston, Massachusetts 02115
                   Grant R805294
              Program Element No. EHE624A
            EPA Project Officer: Dennis C. Drehmel

          Industrial Environmental Research Laboratory
            Office of Energy, Minerals, and Industry
             Research Triangle Park, NC 27711
                    Prepared for

          U.S. ENVIRONMENTAL PROTECTION AGENCY        	   .
                                     1T^v_.*'._.W,r..«^i.-j-vs /--~>-ir fs > i-.'1"""1:'1";:"•.' =-! '' • f 1 f  >•? If JJV
             Office of Research and Development linviio.m^..^-^-—-  _> o
                 Washington, DC 20460     teton. v, :,r--';j-y

-------
                              ABSTRACT
     Emission rate estimates of suspended particulates from open sources
in the United States were obtained from emission factors and source ex-
tents in the literature.  The major open sources, with their estimated
total emission rates (in millions of tons per year), are:  unpaved roads,
3 x 10^; construction activities, 3 x 10^; wind erosion of cropland,
4 x 10*; paved roads, 8; wild fires, agricultural tilling, and mineral
extraction, each 3.  (For comparison, point sources of particulates in the
U.S. are estimated to emit about 20 million tons per year.)  Open
source emission rates are estimated for each state.  Correlations among-
these rates (and with state area and population) show that most open source
rates are correlated with each other and that state population is strongly
correlated with the total rate and with most of the source types.  The
use of cost-effectiveness is defended.  It is shown that the paving of un-
paved roads should reduce emissions at an average of less than $0.01 per
pound for such states as RI and DE  (for rural roads) and AK, AZ, CA, DE,
MI, NV, PA, CO, FL, IL, IN, KY,MD, MA, NJ, NM, OH, RI, TN, TX, UT, VA, WA,
WV  (for remaining unpaved municipal roads).  These cost figures are esti-
mates, and particular situations may differ greatly from the conditions
assumed.  Linear regression of state-by-state annual geometric mean TSP
values versus open and point source emissions rates showed that variations
in open sources emissions contributed less to variations in TSP readings
(per ton of material emitted) than did variations in point source emissions.
Particle toxicity and transport characteristics deserve consideration, too,
but the literature indicates that the predominant contribution to total sus-
pended particulates  (TSP) measurements is soil-like material.  The control
of unpaved road emissions (generally by paving), especially in cities, and
the control of emissions from construction activities are concluded to de-
serve high priority  in the effort to reduce TSP levels.
                                  ii

-------
                               CONTENTS
Abstract .	ii
Figures	^v
Tables 	   v

    1.  Introduction and Summary 	   1
    2.  Defining Goals	15
    3.  Problem Identification 	  19
    4.  Figures of Merit	39
    5.  Evaluation of Control Strategies: Cost-Effectiveness of the
        Control of Unpaved Roads 	  45
    6.  Conclusions	61

Appendix A:  Setting Priorities for the Control of Particulate
             Emissions from Open Sources	63
Appendix B:  Bibliography  	  82
Appendix C:  Open Sources of Particulates in the Mineral Industries:
             Extraction and Milling	102
Appendix D:  Fugitive Emissions of Toxic Particles 	 110
Appendix E:  Formulas for Estimating Open Source Emission Rates •,• 118
Appendix F:  Estimating the Impact  of Irrigation on
             Agricultural Emissions 	  	 121
Appendix G:  Linear Regression Analysis 	 123
Appendix H:  Metric Conversion  Factors 	 127
                                  iii

-------
                                    FIGURES

Number

 1.1    Trends of peak daily total suspended particulate concentrations
          from 1970 to 1976 at 2,350 sampling sites	2

 1.2    AQCR status of compliance with ambient air quality standards for
          suspended particulates                                          3

 1.3    Flow chart: estimation of source impact 	 7

 1.4    Nontraditional source increments at different site types	11

 2.1    General decision framework for air pollution control. .	17

 3.1    Histogram: total emissions. . ... . . . . . ... .	27

 3.2    Histogram of state total emissions per area .	28

 3.3    Histogram of state total emissions per population	29

 3.4    Total emissions by state (c!2) versus unpaved road emissions
        (clO)	30

 3.5    Histogram of state non-road emissions rates 	32

 5.1    Unpaved road dust emission reduction versus control cost	58

 A-l    Unpaved road dust emission reduction versus control cost.........77
                                      iv

-------
                                    TABLES

Number                                                                 Pages

 1.1    Estimated totals of anthropogenic open source particulate emis-
          'sion rates for the United States	5

 1.2    Comparison of fugitive dust emissions from unpaved roads with tra-
          ditional  source emissions in major counties of 14 AQCR's.  .  .  9

 1.3    Estimates of average filter loadings by site classification.  .  .12

 1.4    Composite summary of particle size by components	13

 3.1    Terms used in tables 3.2 to 3.6	21

 3.2    State emission rates:  tilling,  wind erosion, construction,
          wild fires and prescribed fires	22

 3.3    State emission rates:  mineral extraction:  coal and
          other,  tailings, paved and unpaved roads	23

 3.4    State emission rates:  total open source emissions and total
          non-road open source emissions	24

 3.5    State populations, areas,  emission rates:  total per area, non-
          road per area,  total per-population,  non-road per population  .25

 3.6    Matrix of correlation coefficients	35-37

 5.1    Cost estimates for paving  unpaved roads	47

 5.2    Cost estimates for oiling unpaved roads	50

 5.3    Estimated costs and dust emission reductions of various
          treatments for unpaved roads	54

 5.4    Cost-effectiveness ratios   for  unpaved  road treatments	55

 5.5    Estimated costs of control for  unpaved  road dust for urban
          and rural roads,  by  states	57

 A.I    Open source emission  factors by state,...,.,..,..,.,,,. <„,,,,,.. ,65

 A. 2    Open source extents by state	71

 A. 3    Open source emission rates by state	73
                                     v

-------
A.4  Cost efficiency of alternate methods of dust control: unpaved
     roads	75

A.5  Estimated costs of control for unpaved road dust for urban and
     rural roads, by states	76

C.I  Particulate emission rates in the mining industry	104-105

C.2  Approximate areas enclosed by the iso-exposure rate lines at
     inactive mill site.	107

D.I  Particulate material having TLV values (ACGIH ) lower than "nuisance
     dust"	.	110-111

D.2  Comparison of lead results"	.112

D.3  The United States asbestos mines 	. ....		113

D.4  Typical waste dump ambient air asbestos levels measured by electron
     microscope, optical microscope, and atomic absorption	115

D.5  Fugitive asbestos waste emission estimates based on field
     measurements	115

E.I  Formulas for estimating open source emission rates	119

F. 1  Corrected wind erosion emission rate estimates	122

G. 1  Simple regression results and summary of inputs	125

G. 2  Examination of residuals	126
                                      vi

-------
                               SECTION 1

                        INTRODUCTION AND SUMMARY
     Of the 3215 counties in the U.S., 424 are not meeting the current
EPA standards for particulates, and this is true of the majority of the
country's urbanized area% with populations greater than 200,000.*  Our
research goal  was  to investigate setting priorities for the reduction
of particulate emissions from open sources, following a systems analysis
approach:  goal definition,  problem identification, formulation of assess-
ment criteria, comparison of alternatives.  As the work progressed, it
became evident to us that the first three steps of such an analysis were
not sufficiently advanced that we would be able to compare all the major
alternatives for control of open sources.  Instead, having settled on
cost-effectiveness as our criterion for comparison of alternatives, we
performed a cost-effectiveness analysis on the source we estimated to have
the greatest emission rate of suspended particulates: emissions from unpaved
roads.

BACKGROUND

     One reason open sources have received increased attention is that the
steady improvement in total suspended particulate (TSP) concentrations in
the early 1970's was at least temporarily reversed in 1976.  Figure 1.1
(from reference 2) shows a box plot of the national geometric mean daily
TSP readings from 1970 to 1976.  (The upper and lower edges of the boxes
are the 75th and 25th percentiles; the triangle is the average; the point in
the box is the median; the dark squares connected by lines to the box are the
90th and 10 percentiles.)  In 1976, TSP values increased rather than decreased.
Analysis of the events associated with 1976 indicated:2  "Large areas of
the country experienced drought during 1976, and these extremely dry soil
conditions increased the likelihood of wind-blown dust contributing to am-
bient TSP levels....dry soil conditions existed in those general areas that
had TSP increases.   Further, the results of an assessment of the particulate
material captured on filters in 14 major cities in the U.S. indicated that
mineral matter (mostly soil-like material) predominated over other types of
particulate material, including combustion products.-'

     Figure 1.2 (from reference 6) shows the Air Quality Control Regions
(AQCR's) having violations in the first half of 1976 of the  EPA  TSP
primary standards.   (In some instances, only a few of the sampling locations,
such as in a major city, within an AQCR may exceed the standard yet the en-
tire AQCR has been shaded.)  The violations in the West are surprising if
only point sources are considered.  The pattern is more easily understood
when the effects of  emissions from open sources - roads, agricultural areas -
are considered.

-------
     400
     350
     300
   3 200
   £
   a
    3150
     100
     SO
          I
I     T
I
T
I
          1970     1971      1972
                                  1973


                                  YEAR
                                          1974      197S     197S
Figure  1.1.  Trends  of peak daily total  suspended particulate

              concentrations from 1970 to 1976 at 2,350 sampling  sites.

-------
                                                                      Areas in Compliance
                                                                      Areas with Violations
Figure 1.2.  AQCR status of compliance with ambient  air quality standards  for  suspended  particulates.6
                                                     .Jtt

-------
GOAL DEFINITION

     A proper goal would be the equitable minimization of the costs of pol-
lution control and the costs of pollution damage.  The difficulties in find-
ing such an equitable minimization are discussed in Chapter 2.  Essentially,
one requires some agreed-upon measure of equity and accurate information on
the emissions, emissions control, and     the connection between the reduc-
tion of emissions and the reduction of pollution damage in terms of costs.
Such an equitable minimization is probably impossible to determine.

PROBLEM IDENTIFICATION

     One way of stating the problem is that emissions from open sources
contribute to failures to meet the E.P.A. standards for particulate con-
centrations in many localities nationwide.  Determining which sources con-
tribute most would require a knowledge of the connection between emission
rates and concentration measurements.  At least an initial understanding
of the problem could be obtained by determining which open sources contri-
buted most ( as  mass  rate  emitted).    The results of our using emission
factors and source extents from the literature to calculate emission rates
are given in Table 1.1.  More details are presented in Chapter  3 and in
Appendices A, C, D> E, and G.

     Unpaved roads are estimated to have the largest emission rates of the
major open sources.  Total emissions of particulates from point sources
have been less than 20 million tons per year for several years.  In con-
trast, our unpaved road emission estimate is fifteen times this, and con-
struction emissions and wind erosion of cropland are also estimated by us
to exceed the point source emissions.  Admittedly, the estimates of open
source emissions are crude, but it is unlikely they are wrong by an order
of magnitude; such sources are certainly greater in magnitude than point
sources.  Whether or not their impact on particulate concentrations in
populated areas is as important as those from point sources depends upon
particle transport and relative toxicities, matters that deserve further
study.  Results of air sampling in cities suggests that soil-like particles
are the major contributors to suspended particulate concentrations, as dis-
cussed below.

     Our emission estimates were aggregated to the state-wide level, as it
is generally on this level that governmental control decisions will be
made.  Chapter 3 presents frequencies and correlations by state for the
various types of open sources of particulates and also identifies those
states with the greatest and least mass rates of emission in  each category.

ASSESSMENT CRITERIA

     In Chapter 4, it is argued that cost-effectiveness—achieving the
maximum reduction in emissions  (or in concentrations or in doses) for a
given expenditure-^is the appropriate criterion  to apply to evaluate con-

-------
TABLE  1.1.  ESTIMATED TOTALS OF ANTHROPOGENIC OPEN  SOURCE PAR-
            TICULATE EMISSION RATES FOR THE UNITED  STATES

                       ESTIMATED EMISSION RATE
EMISSIONS TYPE             (10^ tons per year)            RANK

agricultural tilling           3.2                        6

wind erosion of crop-
  land                        44-                         2

construction                  27.                         3

wild fires                     3.4                        5

prescribed fires               0.43                      10

minerals: extraction
    -coal                      0.4                        9
    -other                     2,8                        7

minerals: tailings             0.8                        8

paved roads                    7.9                        4

unpaved roads                320.                         1

TOTAL                        409.

TOTAL,  NON-ROAD               81.

-------
trol alternatives, in conjunction with equity considerations.

     Initially, we intended to use a cost-benefit approach to optimize
control of open sources.  The costs of control were to be determined as
well as the benefits.   The benefits would be estimated by the procedure
outlined in Figure 1.3.  Having obtained emission rates (from emission
factors and source extents), we would use a (simple) dispersion model to
predict mean concentrations, then factor in population distributions and
dose-response information to predict health effects.  Health costs and
values assigned to person-years of healthy or sick life would then be
used to estimate the benefits to be obtained from changes in emissions.
Recently, the relationships between total suspended particulate concentra-
tions and health effects have been summarized by Lave and Seskin.^  Medi-
cal costs and lost earnings are not adequate for estimating the value of -
lost years of life.f as we discussed in a progress report for this project, and
as  Howard" has argued in a recent paper on the proper methodology for
estimating the value an individual places on a small incremental probabil-
ity of premature death.  Almost as good a measure would be the cost of
alternative ways of saving lives.  Lacking proper information for evaluat-
ing the benefits of pollution control, we have chosen to investigate the
least-cost reduction of emissions.

COMPARISON OF ALTERNATIVES: UNPAVED ROADS

     In Chapter 5, cost-effectiveness is applied to the control of unpaved
road emissions.  The least-cost technique is judged to be paving of roads,
and states are ranked from those for which paving    would produce the
greatest reduction in emissions per dollar to those for which the reductions
would be least.  (Rural and municipal roads are distinguished.)  The state-
wide average costs per pound of particulates prevented from being emitted
are estimated to range from <_ $0.005/lb. (DE, CA, MD, NJ) to about $0.06/lb.
(SC).

CONCLUSIONS

     The conclusions in Chapter 6 relate  primarily to the areas in which
more investigation is warranted.  The conclusions others will draw from
the results in Chapters 3 and 5 and Appendices C}  p, and G will depend on
their perspectives and responsibilities.  It seems incontestible, however,
that open sources are major contributors of particulate material which is
measured as total suspended particulates, both in urban and in rural areas,
and that control of such particulates in many areas can be much less costly
than control of.some industrial emissions in those areas.

DISCUSSION

     This analysis indicates that open sources may well contribute more
primary particulate material than do point sources and may be less expensive
to control, at least in one important case, unpaved roads.  The emission in-
ventory approach used in this report estimates mass rates of emissions as

-------
EMISSION FACTOR
SOURCE EXTENT
                   EMISSION RATE
DISPERSION MODEL
DOSE/RESPONSE
COST MODEL
                        V
                   CONCENTRATIONS
POPULATION MODEL
                   HEALTH EFFECTS
                         V
                   HEALTH,COSTS
Figure 1.3.  Flow chart:  estimation of source impact.

-------
a means of assessing the relative impact of different source types on air
quality.  This necessarily neglects transport.  A complementary approach
is that of using air sampling and analysis to determine the sources of
particulate materials that are airborne in populated areas.  The results
of several such sampling and analysis studies corroborate our conclusion
that sources of soil-like particles, such as unpaved roads, construction,
wind erosion, and paved roads are primary   contributors to measured TSP
levels.
                3
     Lynn et al.  selected fourteen cities in the contiguous United States
and analyzed 300 filters from high-volume samplers in these cities.  They
used these analyses to determine the major contributors to measured TSP
concentrations.  They concluded that the traditionally-studied point
sources are sometimes dominant, especially in heavily industrialized areas,
and they estimated such sources contributed to about 15 yg/m3 in residential
areas and over 60 yg/m3 in heavily industrialized areas.  Non-traditional
sources, primarily those open sources which are the subjects of this report,
were estimated to contribute about 25 to 35 yg/m3 to citywide TSP levels.
Lynn et al.3 found that the particulates, by mass, had an average of
about 65 percent mineral matter versus about 25 percent combustion products.
Denver and Oklahoma City, "both areas of dry climate with acknowledged fu^-
gitive dust problems," had averages of over 80 percent minerals.  They con-
cluded, "Even the cleanest cities (with respect to traditional sources)
will be 30 to 40 yg/m3 above the local nonurban levels due to nontraditional
sources."
                                                                   3
     With respect to reentrained dust from paved roads, Lynn et al.  con-
cluded, "The average impact on monitors reviewed in this study was about
10 to 15 yg/m3 in residential areas and 15 to 20 yg/m3 in commercial and
industrial areas...."
                3
     Lynn et al.  found it difficult to specify the contribution of dust
from unpaved roads: "...the current emission factors given in AP-42 pro-
vide emission estimates that can be up to 30 times the emissions from tra-
ditional sources."  Table 1.2  shows emission inventory (not air sample
analysis) estimates of emissions from unpaved roads and from traditional
sources for fourteen major counties in the United States.  Unpaved road
emission estimates ranged from much larger than to much smaller than those
of traditional sources, from city to city.  It was noted that unpaved
road emissions and traditional emissions have different particle sizes
and release heights and therefore may not make the same contribution to
airborne concentrations per mass emitted.

     Lynn et al. also concluded that construction activities can be im-
portant:^ "Monitors within half a mile of construction may have annual
geometric means 10 to 15 yg/m3 higher than normal.  On an annual basis,
the effect on the citywide TSP level is expected to be only 1 to 3 yg/m3
.... As with reentrained dust from unpaved roads, the currently available
emission factors for construction activity provide estimates of fugitive
dust emissions well above what would appear logical in terms of the TSP

-------
      SABLE 1.2.  COMPARISON OF FUGITIVE DUST EMISSIONS FROM UNPAVED ROADS
                 WITH TRADITIONAL SOURCE EMISSIONS IN MAJOR COUNTIES OF
                 14 AQCR'S3
AQCR
Baltimore
Birmingham
Chattanooga
Cincinnati
Cleveland
Denver
Miami
Oklahoma City
Philadelphia
Providence
San Francisco
Seattle
St. 'Louis
Washington, D.C.
Major county
o
Baltimore City
Jefferson
Hamilton
Hamilton
a
Cuyahoga
Denver
Dade
Oklahoma
Philadelphia3
Providence
•a
San Francisco
King
St. Louis Citya
ft
District of Columbia
Emissions from
unpaved roads ,
tons per year
510
193,000
N.A.
76,990
18,800
1,270
70,940
85,920
2,660
139,830
660
197,770
l,220b
100
Emissions from
traditional
sources ,
tons per year
7,000
110,000
10,300
56,100
210,000
9,700
8,000
2,600
31,600
7,800
4,800
7,300
15,600
5,600
Ratio
0.07
1.75
—
1.37
0.09
0.13
8.88
33.55
0.08
17.88
0.14
27.17
0.08
0.02
 These counties are almost totally urbanized so the travel on unpaved roads is  ex-
pected to be minimal.

 This represents the emissions from St. Louis County,  a much larger,  less urbanized
area; actual reported emissions for St. Louis City were 159,900 tons/year suggesting
a coding error confusion with the county.

-------
levels actually observed."
                                                                    o
     Figure 1.4 and Table 1.3 summarize the estimates by Lynn et al.  of
the impact of non-traditional sources on air quality. (They cited a figure
of 0.66 million tons of vehicle tire tread wear per year, which is a much
smaller mass rate than those of the major open sources we have identified;
these particles were generally the largest found on the filters and there
is some question as to their true contribution to the mass concentration.)
Table 1.3  gives particle size information on minerals, combusion products,
biological material, and rubber from the 300-filter analysis.

     Confirmatory evidence that soil-like particulates are major contribu-
tors to the TSP levels in cities as well as rural areas came from work re-
cently published by Klappenbach and Goranson .         Noting that 15 Octo-
ber 1976 had many reported TSP values (24-hour samples) exceeding secondary
(150 yg/m ) standards, they analyzed by microscope material caught that day
on the filters of hi-vol samplers for six sites: Fish Creek, WI; Madison,
WI; Milwaukee, WI; Chicago, IL; Moline, IL; Peoria, IL.  The particles were
examined and identified as to chemical type (such as quartz, fly ash,
spores, salt) and the percentages (by count) were listed by the detailed
descriptions under four rather general categories: minerals, combustion
products, biological materials, miscellaneous.  The minimum detectable par-
ticle size was estimated to be 1-2 ym.  The percentages attributed to min-
erals (the kind of material emitted by open sources) ranged from 72 percent
(Fish Creek, WI) to 91 percent (Moline, IL; Milwaukee, WI); the Chicago
value was 84 percent.  The authors conjectured that agricultural activity
was responsible, but their evidence was inconclusive.  It was, however, a
period ±n the area which was more dry and more windy than average, which
would increase emissions from-agriculture, roads, and construction.

     Almost all open sources are at ground level.  If, as we believe, they
contribute substantially to the ambient concentrations of particulates in
the cities as well as in rural areas, then one would expect that urban con-
centrations, would decrease from the ground level up.  This is exactly what
Pace et al.   found in a study of the concentrations measured by 31 high-
volume air samplers in 7 urban commercial areas.  Subtracting estimated
nonurban and secondary particulate contributions, the residual concentrations
were found to decrease with high height, following a relationship well-fitted
(r2=0j47) by the equation:

                          c(z) = b + a/z        (10 ft <_ z <_ 100 ft)
where

c(z)  = adjusted concentration at z,  yg/m3

z = height of monitor above ground level, ft

b - 23.5 yg/m3

a = 381.6 ft-yg/m3.
                                    10

-------
    40 r
    30 -
10
 ,E

  4.
 LU
 CC
 O
 z
 Q.
 CO
    20 -
     10 -
-

REENTRAINED
OTHER
TIRE WEAR
TAILPIPE
CONSTRUCTION

REENTRAINED
OTHER
TIRE WEAR
TAILPIPE
CONSTRUCTION
                                                       REENTRAINED
                                                            +
                                                         OTHER
                                                        TIRE WEAR
                                                         TAILPIPE
                                                       CONSTRUCTION
               RESIDENTIAL
COMMERCIAL
                                                       INDUSTRIAL
       Figure 1.4.  Nontraditional source increments at different
                    site types.3
                                       11

-------
 TABLE 1,3. ESTIMATES OF AVERAGE FILTER LOADINGS BY SITE CLASSIFICATION3
                                                                      _3 -

Components
Mineral
Combustion products
Biological material
Misc. (mostly rubber)
Assumed < 1 \m
Total
3
Average loading, |jg/m
Commercial
64
27
2
9
19
120
Residential
51
19
3
5
14
92
Industrial
87
42
3
9
25
166
Undeveloped
66
6
<1
<1
13
86
aBased on a total of 300 filters analyzed.
                                    12

-------
TABLE 1.4. COMPOSITE SUMMARY OF PARTICLE SIZE BY COMPONENTS3
Component
Minerals
Quartz
Calcite
Hematite
Combustion Products
Oil soot
Coal soot
Glassy fly ash
Biological Material
Pollen
Rubber
Average
size, urn
( 8)
11
9
3
(5)
13
30
12
(24)
35
(43)
Average size
range , urn

-------
Pace et al. also analyzed data taken in other cities at different heights
at the same sites.  They concluded,4 "roughly 12 to 20 yg/m3 may be added
to the concentrations at typical hi-volume sites due to the influence of
nearby ground-level sources...."

     Major open sources within the cities would include paved roads, un-
paved roads, and construction.  As Table 1.2 demonstrates, there are urban
areas where the expected emissions from unpaved roads exceed those estimated
to be emitted from "traditional" sources.

CONCLUSIONS

     Emissions from unpaved roads greatly exceed those from point sources.
In relatively dry regions of relatively high traffic density, they can be
controlled for as little as $0.01/lb.  Other open sources which rival point
sources in magnitude are: construction emissions, wind erosion.of  crop-
land, and emissions from paved roads.  Controlling such sources may be more
cost-effective in reducing TSP concentrations than controlling point sources
further.  Whether such a strategy would also be optimal in a cost-benefit
sense depends upon factors yet to be determined: transport of particulates
and their relative toxicities, the value of increased life expectancy, the
costs of alternative methods of saving lives.
                                     14

-------
REFERENCES


1.  EPA-Released State Air Quality Data Show Areas Not Meeting Federal
    Standards.   J.  Air Pollution Control Assoc.   Vol.  28,  pp.  378-380,
    April 1978.

2.  U.S. Environmental Protection Agency.  National Air Quality and Emis-
    sions Trends Report, 1976.   Report #EPA-450/l-77-002.   U.S. Environ-
    mental Protection Agency,  Research Triangle Park,  NC.   December 1977.

3.  Lynn, D.A.,  G.L. Deane, R.C. Galkiewicz, and R.M.  Bradway.  National
    Assessment of the Urban Particulate Problem.  Volume 1: Summary of  Na-
    tional Assessment.  Report #EPA-450/3-76-024.  U.S. Environmental Pro-
    tection Agency, Research Triangle Park, NC.   June 1976.

4.  Pace, T.G.,  W.P. Freas, and E.M. Afify.  Quantification of Relationship
    Between Monitor Height and Measured Particulate Levels in 7 U.S. Urban
    Areas.  Paper 77-13.4 presented at the 70th Annual Meeting of the Air
    Pollution Control Association.  Toronto, Ontario,  Canada.   June 1977.

5.  Klappenbach, E.W. and S.K. Goranson.  Evaluation of Midwest Sources of
    High Particulate Concentrations on October 15, 1976: A Case Study.
    J. Air Pollution Control Assn. Vol. 29, pp.  47-50, 1979.

6.  Kalinowski,  T., R. Myers,  M. Ellenbecker, and J. Spengler.  The Impact
    of EPA's Interpretive Ruling on Industrial Growth and the Siting of New
    Plants.  Harvard School of Public Health, Boston,  MA.  March 1977.

7.  Lave, L.B. and E.P. Seskin.  Air Pollution and Human Health.  Johns
    Hopkins University Press.   Baltimore, MD.  1977.  368 pp.

8.  Howard, R.A.  Life and Death Decision Analyses.  Department of Engineering,
    Stanford University, Stanford, CA.  October 1978.
                                    15

-------
                                SECTION 2

                             DEFINING GOALS
INTRODUCTION

     To set priorities, one first needs to identify one's goals.  In
air pollution control these are the reduction of air pollution damages
in an economical and equitable manner.  Because of the complexity of the
source-transport-receptor-effects-valuation chain in air pollution control,
often the criteria for choosing one approach over another represent
rather imperfect surrogates for the optimally economic and equitable
reduction of damage.  This is discussed at greater length next.
DECISION FRAMEWORK

     Figure 2.1 (from Deininger ) presents a general decision frame-
work for decisions about control technology and regulations.  This in-
formation would be required for determining whether the added cost of
the controls/regulations was justified by the added benefits.  Con-
siderations of equity would also enter into such a decision.  The major
aspects of the decision framework are:

1.  sources - emissions types, sizes, locations, future growth;

2.  transport - concentrations created in time and space;

3.  receptors - population magnitude and distribution;

4.  effects - dose-response relationships;

5.  impact - value and implications of effects and controls.

     An approach to optimal control would be to use cost and benefit
estimates for each alternative.  Benefit estimation is quite difficult,
however, and should involve equity considerations, which are unfortun-
ately still subjective.  A cost-effectiveness approach is less complete,
but more tractable:  this indicates the minimum cost to reach any particular
set of conditions.  For example, it could be determined what the minimum
necessary cost was for achieving the National Ambient Air Quality Standards
for a city like New York through a linear programming analysis.1
Another cost-effectiveness example would be the determination of the maximum
reduction of unpaved road emissions which could be obtained for one (or
several) level(s) of expenditure; this we have done in Section 5.
                                     16

-------
                               REGIONAL INPUTS
                                                4  Meteoro-
                                                  logical
                                                  Data
             8  Control
             Technologies
              and Costs
                I
             9 Alternative
              Controls
              for each
              Source
5 Demo-
  graphic
  Data
                             EVALUATED FOR EACH COMBINATION OF CONTROLS

                                                     I
      I	
Figure  2.1.  General decision framework  for  air  pollution  control.
                                          17

-------
     Ideally, then, we would like information about the reduction of
damages from various policies.  Short of that we would prefer, in
descending order:

-reduction of doses

-reduction of concentrations

-reduction of emission rates.

The cost-effectiveness approach, which we use here, would then ask for
the maximum reduction achievable at various expenditures.

     The sources and the costs of controlling them vary greatly geographic-
ally.  It may be much more cost-effective to control unpaved road emissions
in one region than another, for example.  In some sense, the more de-
tailed information we could provide about open sources, the better for
those who have to make control decisions, but this is offset by information
limitations as well as by the financial and manpower resources available
for this project.  We have chosen to aggregate to the state level;
the states have primary responsibility in pollution control, so that
those making the decisions based partly upon information in this document
may well have responsibility at the state level, and the state-wide
figures may be of most use.


CONCLUDING REMARKS

     This investigation of the setting of priorities in open source
control has led to the identification, state by state, of open sources
contributing most to the total emission emission rate of the state.
Further analysis is required to approximate the costs and benefits of con-
trol to make an optimal decision. For the overwhelmingly largest of these
emitters, unpaved roads, we have performed a cost-effectiveness analysis
(Section 5)  to set priorities among the states for control and to serve as
an illustration of the cost-effectiveness approach we recommend.


REFERENCES

1.  Deininger,  R.A.,  ed.  Models for Environmental Pollution Control.
    Ann Arbor Science Publication, Ann Arbor, Michigan 1974.  448 pp.
                                     18

-------
                                SECTION 3

                         PROBLEM IDENTIFICATION
INTRODUCTION

     One approach to systematic problem solving involves determination of
goals, identification of problems, specification of criteria, evaluation of
alternatives, implementation of the alternative selected, and monitoring
of the performance of that alternative.  This section is directed to
characterizing some aspects of the problem of open sources of particulate
emissions.

     Three elements of air pollution are the sources, the transport of
the material, and the effects on receptors.  An optimizing approach to
air pollution control would involve minimizing the effects (weighted for
value) at a given budgetary level, or 	 better yet 	 continuing to
add to control until the marginal cost of additional control just matched
the marginal benefits of that control.  Because of the complexity of
evaluation of dose-response relationships and even of predicting dose
given the sources, the approach taken in this study has been to focus on
reduction of the emissions rather than reduction of the concentrations, the
effects, or the cost of the effects.

     We do not think that priorities for pollution control should be set
solely on the basis of emission rates.  Information on control costs and
benefits and other impacts is clearly needed.  However, to decide on
which of the open sources we should first gather such additional information,
we were strongly influenced by the dominant emission rate from unpaved
roads, so that our cost-effectiveness investigation focused on this source.

     We have divided emissions into those which are particularly harmful
("toxic") and those which are not particularly harmful, though not
necessarily harmless.  Industrial hygienists, for example, distinguish
between "nuisance dusts" and those that are more toxic.  We have adhered
to the same distinction.  Toxic substances are discussed in Appendix D.
In this section, we deal with emissions which are not especially toxic,
akin to "nuisance dusts" though not necessarily synonymous with them.

     Emission rates are the masses of material emitted per period of time.
Emission factors give the emission rates per unit of source extent.  The
determination of emission factors and source extents done for this study
is described in  Appendices A and  E9 and more information will be made
available on request.  This section presents emission rates by state for
the major anthropogenic open sources of particulate emissions.
                                     19

-------
 EMISSION FACTORS

      The derivation of these is outlined in Appendix A, and they are
 presented by state in Table 1 of that appendix.


 SOURCE EXTENTS

      These,  too, are explained in Appendix A;  they are listed in Table
 2 of that appendix.


 EMISSION RATES

      In Table 1.1 are presented the emission rate totals for the major
 open sources we identified.   A partial listing of the state-by state
 emission rates appears in Table 3 of Appendix  A.   The general categories
 are:   agriculture,  construction,  fires,  mineral  industries,  and roads.
 As indicated in Table 1.1,  unpaved road  emissions were an order of  magnitude
 larger than  those from any other  source,  but the estimation  of their
 magnitude is quite difficult (see Appendix A), thus suspect.   Next  largest
 were the rates from agriculture and from construction,  followed by
 emissions from paved roads,  wild  fires,  and the  extraction of minerals.
 Relatively small contributions came from mineral tailings piles and from
 prescribed burning.

      The national totals  are less useful  for control implementation than
 are the state totals,  on  which we concentrate  in this section.   The Clean
 Air Act of 1970 requires  the states to develop State Implementation Plans
 to meet the  National Ambient Air  Quality  Standards  for  criteria pollu-
 tants,  including "total suspended particulates"  (TSP).   Therefore,  much
 of the control decision-making takes  place at  the state level.   Further-
 more,  as will be seen, the  states differ  greatly  one from another,  so that
 a  major source in one state  may well  be minor  in  another.

      To facilitate the presentation of the emission rate estimates  by
 computer,  we  present  in Table  3.1  an  explanation  of the lables  used in
 the four succeeding  tables.  The emission  estimates  are  for the  masses of
 particulates  smaller  than about 30  ym in  diameter and of  specific gravity
 2.5, which is  equivalent  to  particles  smaller than  about  48 ym  in
 aerodynamic diameter.  Although these  are  thought to be the correct cri-
 teria  for  describing particles  captured in the standard Hi-Vol  used for
measuring  total  suspended particulates, only particles  smaller  than about
 10 ym  aerodynamic diameter are respirable  and capable of being  transported
miles  or more under usual atmospheric  conditions.   Thus, a substantial
portion  of the mass of particulates emitted by open  sources as  indicated
 in these  tables may not make an important  contribution  to the hazards
of air pollution or even to  the measures of total suspended particulates
miles away from where they are generated.

     Table 3.2 gives the emission rates for tilling, wind erosion,  con-
struction, wild fires, and prescribed fires.  Table 3.3 continues these
                                     20

-------
               TABLE 3.1. TERMS USED IN TABLES 3.2 to 3.6


AREA:  State area, land and water, in thousands of square miles.

CONSTRUCTION:  Emissions due to construction activities.3

EXTRACTION:  Emissions from overburden removed to facilitate surface
             mining.

NON-ROAD:  TOTAL minus the sum of PAVED and UNPAVED ROADS; in other
           words, the non-transportation contribution.

PAVED ROADS:  Emissions     from transportation over paved roads, due
              to re-entrainment of surface dirt.

POPULATION:  State population, in millions.

PRESCRIBED FIRES:  Emissions due to fires set for agricultural or silva-
                   cultural reasons.

TAILINGS:  Emissions due to the re-entrainment of dust from tailings
           piles.

TILLING:  Emissions due to tilling cropland.

TOTAL:  Sum of the emission rate estimates from all major anthropogenic
        open sources of particulates.

UNPAVED ROADS:  Emissions due to transportation over, and wind erosion of,
                unpaved roads due to re-entrainment of surface material.

WIND EROSION:  Emissions due to wind erosion of cropland.
a.  Note:  all emission rates are in thousands of tons per year (909
    x 103 kg per year).
                                     21

-------
          TABLE 3.2.  STATE EMISSION RATES:  TILLING, WIND EROSION,
                      CONSTRUCTION,  WILD FIRES AND PRESCRIBED FIRES

State   Tilling   Wind Erosion  Construction  Wild Fires   Prescribed Fires
AL ;
AK '
AZ
AR !
CA '•
CO !
CT
DE
FL
GA
HI
ID
IL
IN
IA
KS
KY
LA
ME
MO
HA
MI
MN
MS
MO
MT
ME
NV
NH
NJ
NM
NY
NC
ND
OH
OK
OR
PA
RI
SC
SB '
TN
TX
UT
VT
MA
WA
WU
WI
WY
5,30
0,10
117.00
14,60
126,00
158,00
0,50
0,60
5,00
8,45
2,30
88.80
151,00
52,40
142,00
239.00
15.10
14.40
1,20
2.10
0.50
21.20
99.30
12.90
61.70
200.00
306,00
68.00
0.40
1.60
50.30'
20,10
6,30
292,00
44.20
89.00
39.40
21.60
0.10
4.30
210.00
10.90
386,00
78,40
1,90
5,30
14,60
2,84
49.00
56.30
47.5
0.0
713,1
• 159.3
3945,8
881.6
4.1
8.6
92.2
91.5
0.0
987.8
1721.4
842 . 6
2077.7
4769.0
. 122,1
90,7
10.4
44,6
6,4
410,7
1610,6
71,9
858.8
1264,1
5218,9
1157,1
0.5
15.7-
842,3-'
149,6
118,2'
3553,3
822,8
1177,9!
48,2
81,6
0*9
54.0
3119,6
41,2
5000,7
653,4
8,1
, 71,0
64,3
5.6i
663,1
638,8
273,
59.
139,
141.
4854.
117.
333.
10.
766.
455.
62.
24.
1415,
773.
253.
245.
270.
870.
87.
128.
1260T
743.
452.
138.
346.
31.
197.
22.
69.
844,
61.
1462.
640.
80.
1365.
332,
212,
2090,
82.
177.
39.
333.
3760,
203,
17,
212,
447,
215,
218,
59,
102.00
646.80
28.50
94.50
218.80
25.00
1.00
0.05
316.60
41.30
0,00
544.00
11.40
8.90
1.90
65.80
62.50
73 . 00
2.50
1.40
8.40
7.70
30.80
78.60
119.80
99.60
20.00
14.80
0.46
23.50
19.90
6.00
75,10
0 47
5.20
115.80
196,50
12,90
0,70
43.60
4,80
27,70
19,10
10,00
0,20
5 , 60
164,30
74,20
6,90
8 , 30
15.60
0.00
5,20
4,10
21,20
0,50
0,00
0.04
71.90
54.30
0 , 00
45,40
0,00
0,00
0,00
0,00
0,00
16.60
0.00
0.00
0,00
0.30
0.00
12,70
0.00
52,90
0 , 00
0,00
0,00
1.50
1,40
0,00
8,80
0.10
0.00
0.00
21.60
0,00
0,00
29.10
0,00
0,00
6,20
0.00
0,00
3,90
56,80'
0»00<
0 , 00'
0,00'
                                      22

-------
TABLE 3.3.  STATE EMISSION RATES:  MINERAL EXTRACTION:
            COAL AND OTHER, TAILINGS, PAVED AND UNPAVED ROADS
ate

AL
AK
AZ
AR
CA
CO
CT
DE
PL
GA
HI
ID
II.
IN
I A
KS
KY
LA
ME:
MD
MA
MI
MN
MS
MO
MI-
NE
NV
NH
NJ
.NM
NY
NC
ND
OH
OK
OR
PA
RI
BC
SD
IN
TX
HI
"..' t
UT
V 1
UA
LIA
w r"
UIM
WI
WY
Extraction:
Coal
16*9
0*6
0*2
0*4
0*0
0 * 2
0*0
0 * 0
0 * 0*
0*0'
0*0'
0 * 0
49*5'
20*2
0*5
0*6
116*6'
0 * 0
0*0
2*0
0*0
0*0
0*0
0*0
3*9
0*4
0*0
0*0
0*0
0*0
0.3
0*0
0*0
0*0
38*6
2*0
0*0
68*4
0*0
0*0
0*0
6*4
0*0
0*2
0*0
29*2
0*1
87*1
0*0
0*6
Extraction:
Other
39*3
127*0
187*0
38*0
171*0
33*8
15*1
2*4
235*0
54*6
9*0
17.8'
104*0'
58*1'
49*4
28*5
36*6'
24*5
5*4
30*5 ,
25*7
139*0
216*0
18*2
56*1
33*9
16*9
36*8
6*8
44*9
42*5
75*6
57*1
5*1
95*9
32*0
45 * 0
90*7
3*2
24*3
12*4
56 * 0
116*0
55*8
5*3
58*7
38*1
14*6
53*9
17*9
Tailings

1*1
0*0
156*7
1 * 0
141*5
17*0
0*4
0*0
2 * 6
1*3
0*0
30*9
5*5
2*6'
4 * 8'
22*5
2*1'
0*4-
0*2
0*8
0.5-
6*7'
20*4'
0*2'
19*3'
11*6
2*8
142*2-
0*0
1*8'
44*0
10*0
2*0
1*5
6*3
6*5
1*4
4*0
0 * 1
0*6
5*3
1 * 5
25*9
118*0
0*7
3*9
0*7
1*2
2*9
5*9
Paved Roads

149*
11*
92*
73*
803 *
89*
117*
23*
386*
212*
26*
30*
370*
229 *
104*
78*
146*
121*
43*
154*
182*
345*
.1.39*
81*
175*
28*
56*
21*
32*
304*
48*
414*
220*
17*
403*
123*
83*
424*
36 *
126*
24*
191*
476*
' 42*
1 8 *
214*
137*
60*
176*
19*
Unpaved Road

3760*
1990*
9740*
9340*
27530*
.1.0080*
320 *
~yy\
/ O * ..
70.1.0*
7010*
/"\ f\
80*
4120*
7960*
6450 »;
11310*
14330*
4380*
3790*
7 1 0 *
710*
680 *
8830*
11560*
4260*''
10330>
773*'
10610*
4890 *
1250*
.1.950*
10640*
5160*
3860*
8800*
3820*
.1.1860*
12110*
10100*'
390*'
1350*'
6370*'
4380*
28640*
5410*
1180*
2340*
6710*
3920*
3310*
2440*
                               23

-------
       TABLE 3.4.  STATE EMISSION RATES:  TOTAL OPEN SOURCE EMISSIONS
                 AND TOTAL NON-ROAD OPEN SOURCE EMISSIONS
States

 AL
 AK
 AZ
 AR
 CA
 CO
. CT
 BE
 PI-
 CA
 HI
 ID
 IL
 IN
 IA
 KS
 KY
 LA
 ME
 HD
 MA
 MI
 MN
 MS
 MO
 MT
 ME
 MV
 NH
 NJ
 NM
 NY
 NC
 NO
 OH
 OK
• OR
 PA
 RI
 SC
 SD
 TN
 TX
 UT
 YT
 MA
 WA
 WV
 MX
 WY
  Total

  4409,7
  2834,5
 11178,7
  9865,9
 37811,3
 11402,1
   791,1
   114,7
  8885.3
  7928,4
   179,3
  5888,7
 11787,8
  8436,8
 13943,3
 19778,4
  5151,0
  5000,6
   859*7
  1073,4
  2163,5
 10503.6
 14128,1
  4773.5
 11970,6
  2494,5
 16427.6
  6351,9
  1359,2
  3187.0
 11749,7
  7297,3
  4987,5
 12749,5
  6601,0
 13738,2
 12757,1
 12893.2
   513,0
  1808,9
  9785,1
  5047.7
 38429.9
,  6570.8
  1231,2
  2943,6
  7632,9
  4380,5
  4479.8
  3245,8
 Non-Road

  500,70
  833,50
 1346,70
  452.90
 9478,30
 1233,10
  354.10
   21,69
 1489,30
  706,45
   73,30
 1738,70
 3457,80
 1757.80
 2529.30
 5370,40
  625.00
 1089.60
  106,70
  209,40
 1301,50
 1328,60
 2429,10
  332.50
 1465,60
 1693,50
 5761,60
 1440.90
   77.16
  933,00
 1O61.7O
 1723,30
  907,50
 3932,47
 2378.00
.1755,20
  564,10
 2369.20
   87,00
  332,90
 3391,10
  476,70
 9313,90
 1118,80
   33,20
  389,60
  785,90
  400,54
 993,80
  786,80
                               24

-------
TABLE 3.5.   STATE POPULATIONS,  AREAS,  EMISSION RATES:
            TOTAL PER AREA, NON-ROAD PER AREA, TOTAL
            PER POPULATION, NON-ROAD PER POPULATION
STATES

Ai
r 1 1.«
AK
n 1 \
A7
II *»
AR
1 1 1 \
CA
CO
w \»t
CT
DE
Fl...
GA
HI
1 11
I !...
I N
:I:A
KS
KY
1 A
I.H. n
ME
MD
MA
M:C
MN
MS
MO
MT
NE
NV
NH
NJ
MM
NY
NC
ND
OH
OK
OR
P A
R I
3C
SO
IN
TX
UT
VT
VA
WA
WV
WI
WY
POPULATION

3*44-
0*30
1 * 77
1*92
19*95
2*21
3*03
0*55
\f 9 W W i
6 * 79
4*59
0*77
0*71
11*11
5*19.
2 » 83'
2*25
3 « 22
-* , 44,
\.J V W I
0*99'
3*92
5*69
8*88
3«81<
2.22'.
4*68
0*69
1 * 48
0 * 49
0*74
7*17
1*02
18*24
5 * 08
0 * 62
10*65'
2*56
2*09
11*79
0*95
2 .51' ,
0*67' '
3*92
1 1 * 20*
1*06
0*44i
4*65
3*41
1*74'
4*42<
0*33.:
AREA TOTAL PER NON-ROAD PER
AREA
51*600 85*459
586*000 4*837
114*000 98*059
53*100 185*799
159*000 237*807
104*000 109*636
5*010 157.904
2*060 55*675
58*600 151*626
58*900 134*609
6*450 27*798
83*600 70*439
56*400 209*004
36*300 232.419
56*300 247*661
82*200 240*613
40*400 127.500
48.500 103*105
33*200 25*895
10*600 101*264
8*260 261.925
58.200 180.474
84.100 167.992
47.700 100.073
69.700 171.745
147*000 16*969
77*200 212*793
111.000 57*224
9*300 146*146
7*840 406.505
122*000 96*309
49*600 147*123
52*600 94*819
70*700 180*332
41*200 160*218
69*900 196*541
97*000 131*516
45*300 284*618
1*210 423.967
31.100 58*164
77.000 127*079
42.200 119*614
267*000 143*932
84*900 77*395
9,610 128.117
40*800 72*147
68*200 111*919
24*200 181.014
56*200 79*712
97*900 33.154
AREA
9*703
1*422
11*813
8*529
59*612
11*857
70*679
10*529
25*415
11*994
11*364
20*798
61*309
48*424
44*925
65*333
15*470
22*466
3,214
19,755
157*567
22*828
28*883
6*971
21*027
11,520
74*632
12*981
8*297
119*005
8*702
34,744
17*253
55*622
57*718
25*110
5*815
52 * 300
71*901
10*704
44,040
1 1 * 296
34*884
13,178
3 , 455
9*549
.1.1*523
16*551
17,683
8,037
TOTAL PER
POPULATION
1281.9
9448,3
6315,6
5138,5
1895.3
5159.3
261.1
208,5
1308,6
1727,3
232.9
8293.9
1061,0
1625,6
4927,0
8790,4
1599,7
1373*8
868*4'
273*8
380,2
1182*8
3708,2
2150.2
2557.8
3615.2
11099*7
12963*1
1836*7
444*5
,11519*3
400*1
981*8
20563*7
619,8
5366.5
6103.9
1093*6
540*0
720*7
14604*6
1287.7
3431,2
6198,9
2798*2
633*0
2238*4
2517.6
1013.5
9835.8
NON-ROAD PER
POPULATION
145.55
2778.33
760.85
235.89
475*10
557,96
116,86
39*44
• 219*34
153*91
95*19
2448*87
311*23 .
338*69
893*75
2386*84'
194,10
299 , 34
107,78
53,42
228 . 73
149,62
637*56
149*77
313.16
2454.35
3892.97
2940.61
104*27
130,13
1040*88
94,48
178*64
6342*69
223*29
685*63
269.90
200,95
91*58
132*63
5061*34
121*61
831*60
1055,47
75.45
83.78
230,47
230,20
224,84
2384,24
                           25

-------
emission rate estimates for mineral extraction and beneficiation (mill
tailings) and for roads (paved and unpaved, including wind erosion of un-
paved roads).

     Table 3.4 presents the total anthropogenic open source emission rates
of particulates and the non-road contribution to that total.  Table 3.5 pre-
sents the 1970 populations of the states, their total areas, and emissions
per population and per area.  The material in Tables 3.2-3.5 is discussed
next.

Total Emissions

     Total emissions by state ranged from 115 thousand tons per year (Dela-
ware) to 38,400 thousand tons per year (Texas), with the median value
being 6,460.  Figure 3.1 shows the histogram derived from these data.  The
intervals are 1000 thousand tons per year.  Note that Texas and California
together contribute 19 percent of the total emissions.

     Emissions per area (tons per square mile per year) is another useful
measure.  If one were to formulate a simple box model for the air above
the state, the concentrations would be proportional to the rate of emissions
per area (of material which stayed suspended).  These values ranged from
4.8 to 424 tons/sq. mi./yr.  A measure of the relative dispersion of the
values is the coefficient of variation (COV), the standard deviation divided
by the mean.  For the state total emissions, the COV was 0.95.  If a major
factor in the emissions were the area of the state, we would expect that to-
tal emissions per area would have a smaller COV, which it does, 0.60.  The
comparative coefficients of variation show that the states are more alike
in their emissions per area than they are in their total emissions, not un-
surprisingly.  Figure 3.2 is a histogram of total emissions per area.  The
two largest values are for Rhode Island and New Jersey.


     States differ greatly in population.  Emissions per person indicate not
only something about the relative contribution of individuals between states
to the air pollution problem, but also the relative doses, to the degree to
which concentrations correlate with emission rates.  Figure 3.3 shows
a histogram of emissions per population, which range from 208 to 20,500
tons/year/thousand, with a mean   value of  3,884.  The COV is 1.14.  The
three highest states are North Dakota, South Dakota, and Nevada, principally
due to their unpaved road emissions and relatively small populations.

     On a. state-by-state basis, as well as on an aggregated national basis,
the dominant contributor to our emission rate estimates is unpaved road emis-
sions.  Figure 3.4 shows total emissions versus unpaved road emissions, for
the states.  The correlation is evident; in fact, the correlation coefficient
is r=0.99 and the coefficient of determination is r2=0.98, indicating that
98 percent of the state-to-state variation can be "explained" by the varia-
tion in unpaved road emissions.  The regression equation is

     c!2 = -107 +1.30 clO

where
                                           •j
c!2 = total state open source emissions, 10  tons/yr

                                    26

-------
middle of
interval
1000*
3000*
5000*
7000*
9000 *
.11000,
13000,
15000,
17000,
19000,
21000,
23000 ,
25000,
27000 ,
29000 ,
31000,
33000,
35000 ,
37000 ,
39000,
number
of
observations
9
6
9
6
4
6
5
1
1
i
0
0
0
0
0
0
0
0
1
1
If* /|% /p /|t /p Jfi Jp /P ^%
?p ^p Jp * JJ\ ?p
*********
******
****
rf|\ /P /p J|\ ?|% Jj%
*****
*
*
*








*
*
Figure 3.1.  Histogram: total emissions.
                  27

-------
middle of
interval
10*
30,
50,
70,
90,
110,
130,
150.
170,
190,
210,
230,
250,
270,
290,
310,
330.
350,
370.
390,
410.
430,
number
of
observations
o
AV,
3
3
4
4
6
5
5
3
5
2
2
2
1
1
0
0
0
0
0
1
1
**
**#
*#*
****
* * ^ ^
^ ^ ^ *n ^ ^
#*#))(*
^ ^ ^ ^ ^
***
*****
**
**
**
*
*





*
*
Figure 3.2.  Histogram of state total emissions per area,
                            28

-------
middle of
interval
500 ,
1500*
2500.
3500*
4500,
5500,
6500,
7500,
8500,
9500 ,
10500,
11500,
12500,
13500,
14500,
15500,
16500,
17500,
18500,
19500,
20500,
number
of
observations
13
13
5
3
1
3
3
0
2
2
0
2
1
0
1
0
0
0
0
0
1
*************
*************
*****
***
*
***
*##

**
**

*#
*

*





*
Figure 3.3.   Histogram of state total emissions per population.
                               29

-------
     c!2
40000» +
                                             * *
30000*+
20000*+
10000,+
                       *
                        2*
                    ***4 *
                 *   **
              5**
           ***4
    0,+  4*
         +~"
       0»
                         16000.
               8000*
           32000*
24000*                40000*
            Figure  3.4.  Total emissions by state (c!2) versus
                        unpaved road emissions (clO).
                                30

-------
                                      3
clO = state unpaved road emissions, 10  tons/yr.

The intercept is not statistically significantly different from 0.0.  Not
only are unpaved roads the dominant source, but they tend to increase in
rate when agricultural tilling and wind erosion increase, due to the cli-
matic determinants the emission factors have in common.(See Appendix E.)

     Paved roads also contribute substantially.  The correlation coefficient
for total emissions versus paved road emissions is r = 0.58.  In turn, paved
road emissions are strongly correlated with population (r = 0.95).  Note that
population and the emission rates for construction (r = 0.86) and unpaved
roads (r = 0.48) have high correlation coefficient values.  For example,
r^ >. 0.25 means that the influence of population "explains" at least 25 per-
cent of the variation among the states.  A predictive linear equation for
total emissions based just upon population and area accounted for 45 percent
of the variation among the states (i™ = 0.37).

     Thus, unpaved roads are the dominant contribution to state emission
rates, as estimated here.  Along with paved roads, they contribute 84 percent
of the estimated total emissions.

Non-Road Total Emissions

     There is value in separating the total into transportation and non-
transportation contributions.  For one thing, if the emission rate estimates
for unpaved roads are later shown to be greatly in error, the non-road es-
timates here would not be affected.  Agricultural emissions, construction
emissions, fires, and emissions from the mineral industries all are related
to climatic conditions, although the relations differ.  Where measures to
reduce transportation emissions are infeasible, proposals for action might
center on this broad group.

     Figure 3.5 is a histogram of non-road emissions.  They range from 21 to
9478 thousand tons per year, with a median of 1075 and a coefficient of varia-
tion of 1.23.  These values and the histogram show that the emissions rates
have great variation.  In decreasing order, the states with the largest non-
road rates are:  California, Texas, Illinois, and Nebraska.  In these states,
the contributions are primarily from wind erosion of cropland and from con-
struction.  Accounting for irrigation would substantially lessen the Cali-
fornia rate  (Appendix F).

     Wind erosion, the second largest source, is highly correlated with non-
road emissions.  We regressed the non-road emission rates (c!5) versus the
wind erosion emission rates  (c4).  The slope and the intercept of the follow-
ing regression equation were both statistically significant  (P < 0.05):

     c!5 =488+1.32 c3

where
                                  3
c!5 = non-road emissions rates,  10  tons/yr
c3  = wind erosion emissions rates, 10-^ tons/yr
                                   31

-------
       middle  of    number of
       interval     observations
          250,       14    **************
          750,       10    **********
         1250.       10    **********
         1750,        5    *****
         2250,        3    ***
         2750.        1    *
         3250.        2    **
         3750,        l    *
         4250,        0
         4750,        0
         5250,        1    *
         5750,        1    *
         6250,        0
         6750.        0
         7250.        0
         7750.        0
         8250.        0
         8750,        0
         9250,        2    **
Figure 3.5.  Histogram of state non-road emissions rates.
                        32

-------
The correlation coefficient was r = 0.81.   It is this high partly be-
cause wind erosion emissions contribute heavily to the non-road emissions,
partly because they covary with other important emissions.

     The non-road emission rates are better correlated with the simple
measures of extent, state area and state population, than were the total
emission rates.  For the non-road sources the regression equation for to-
tal emissions:

     c!5 = 154 + 241 c!3 + 7.33 c!4

where

c!3 = state population, millions
c!4 = state area, millions of square miles,
                                    f\
had a coefficient of determination r  =0.37 (meaning it accounted for 76
percent of the variation).  The intercept and the slopes were all statis-
tically significant with the population being the most significant.

     Having looked at the totals and the non-transportation totals, we next
examine briefly the individual emission rates.

Agricultural Tilling

     These rates depend primarily on soil type, climatic conditions, and
acres of cropland harvested yearly.  The five states with the largest emis-
sion rates (in decreasing order) are:  Texas, Nebraska, North Dakota, Kan-
sas, and South Dakota.  The two states with the smallest are Alaska and Rhode
Island.

Wind Erosion of Cropland

     Wind erosion depends upon soil type, climatic  factors, crop type and
crop acreage, among other factors.  The states with the five largest emission
rates are (decreasing order):  Nebraska, Texas, Kansas, California, and North
Dakota.  Accounting for irrigation would substantially reduce the rates for
California (Appendix F).

Construction

     These emissions were estimated from very approximate emission factors
for the states, adjusted for type  of construction, with source extents taken
from construction activity types and expenditures.  The top five state rates
were (decreasing order):  California, Texas, Pennsylvania, New York, and Illi-
nois.  The two smallest:  Vermont and Delaware.

Paved Roads

     As noted above, paved road emissions were very highly correlated with
population.  Thus  the  top five states are about as  expected:  California,
                                    33

-------
Texas, Pennsylvania, New York,  and Ohio.   The rate for New York,  the second
most populous state, would be expected to be higher;  Figure 3.5 shows paved
road emission rates by state versus population,  and it can be seen that New
York's value lies well below the trend.  This suggests there is relatively
less driving per capita there than in most other states,  due to demographic
and socio-economic factors.  The states with the least paved road emissions
were Alaska and North Dakota.

Unpaved Roads

     Texas produced only slightly more unpaved road emissions by our esti-
mates than did California, followed by Kansas, Oregon, and Oklahoma.  The
least were produced by Delaware and Hawaii.

Wild Fires

     The top five states are: Alaska, Indiana, Florida, California, and
Oregon.  The smallest two are Vermont and Delaware.

Prescribed Fires

     The biggest five emission rate estimates were for: Florida, Washington,
Georgia, Missouri, and Indiana; many states had negligible rates.

Minerals:  Coal

     Kentucky is estimated to have the greatest emissions rate due to coal
mining, followed by West Virginia, Pennsylvania, Illinois, and Ohio.  Many
states had negligible coal mining emissions.

Minerals:  Other Than Coal

     Phosphate rock extraction helped make Florida the leader  in this
emission rate category, with the next highest being Minnesota  (iron ore),
then Arizona, California, and .Michigan.   The smallest are Rhode Island and
Delaware.

Minerals;  Tailings

      The biggest five are:   Arizona, Nevada,  California, Utah, and New Mexico,
all in  the Southwest.  The smallest two are Rhode Island and Maine.

Correlation  Matrix

      Table 3.6 contains all  the correlation coefficients  for the various  pos-
sible combinations  of emission rates.  The first column is  a list  of  the
variables.   The  first row contains a similar  list.  The rest of the  table gives
the correlation  coefficients.  The first column of numbers  has correlationu
coefficient  values  for  the emission  rates, etc., versus the "state numbers,"
the alphabetical  rank of  the state  (Alabama=l,  Wyoming=50).  These state
numbers should be essentially  randomly associated with any  of  the  other
variables, but   one  of  the  numbers  in the first  column of  coefficients would
be significant at  P<0.05  (r>0.28), which shows  that if you  test enough random
                                    34

-------
                                     TABLE 3.6,  MATRIX OF CORRELATION COEFFICIENTS
                           State No.
Tilling      Wind Erosion      Construction
                                                                                     Wild Fires
Prescribed Fires
T:i. 1.1. ins.«
Wind  Erosion
Construct,.! on
W :i. 1 d  i::' i r (•:? s
Prescribed Fires
Extraction S Cosi
Extraction t Other
Paved Roads
U n P s v 6? d R o 3 d s
T a i .1. :i. r i ;3 •:•;
Tot-, ail.
Population
Area
Nojv-RoBd
To t a1 per Are s
M o n •- R o 3 d per Are 3
Total per Population
Non-Road per Populstion
0,
0,
-0,
-0,
-0 ,
0,
•-0,
-0 ,
-0,
-0,
-0,
-0,
•-0,
-0,
. 0,
-0,
0,
0,
053
012
039
31.9
130
096
257
067
067
1.39
061.
040
1.69
036
037
044
093
086
0,
0,
-0,
•-0,
-0,
0,
0,
o±
0,
0,
0,
0,
922
267
085
059
1.37
083
056
630
203
686
052
286
0., 792
0,
0,
0,
0,
098
2O5
578
654
0,
™'0 *
""0 *
""0 *
0,
0,
p_i
0,
0,
0,
0,
394
055
1.06
1.40
123
1.92
707
246
775
1.73
254
0,90:1.
0,
0,
0,
0,
208
301
531
609

0,
0,
0,
0,

0
0
....<>
-0
0
0


,458
, 080
,271
,001
,072
,035
,066
,051.
,643
,042
,220
,191
,1.58
,161



-0,
0,
0 ,
0,
"" \) *
0 ,
0,
0,
-0,
-0,
-0,



1.68
217
1.31
029
023
022
029
049
01.2
200
1.99
-0,109
-0,043

-------
                        Extraction:  Coal
TABLE 3,6,   (cont'd)

Extraction:  Other    Paved Roads
            Unpaved Roads   Tailings   Total
Extraction? Other            0*008
Paved Roads                  0*167
Unpsved Roads               -0.053
Tailings                    -0*135
Total                       -0,051 ,
P o P i..i 1 a t i o n                   0*170
Area                        -0*134
Non-Road                    -0*055
Total per Ares               0*133
Non-Rosd per Ares            0,012
Total per Population       -0,202
Non-Road per Population    -0*179
     0 * 468
     0.367
     0.458
     0*517
     0,360
     0 * 349
     0*077
     0*004
     -0*131
     -0.157
 0.542
 0,164
 0*575
 0*953
 0,028
 0,551
 0*354
 0,309
-0,423
-0,349
0,386
0.989
0*478
0,322
0,841
0.259
0,089
0 , 229
0,121

0 * 382
0*137
0 , 207
0,328
-0,095
-0,066
0,26'9
0,135


0,517
0 , 329
0,910
0.276
0*167
0*239
0*181

-------
                                              TABLE 3.6.   (cont'd)

                         Population   Area    Non-Road   Total Per Area   Non-Road Per Area    Total Per Population
Ares                        0.010
Non-Road                    0*5:16    0,322
Total per  Area             0,354  -0,291    0,277
Norr-Rosd per  Ares          0,340  -0,197    0,356       0,748
Total Per  Population     -0,401    0,366    0,285      -0,133           -0,044
Non-Road Per  Population  -0,318    0,347    0,369      -0,109            0,105               0,909
   GO

-------
 correlations,  chance will produce some "significant"  ones.   In the remainder
 of the table,  correlation coefficients greater than r =  0.5  have been under-
 lined, not because they are statistically significant (though they are,  at
 P < 0.01), but because the relationship is strong enough that it "explains"
 25 percent or  more of the variation in the two variates  compared.

      Inspecting the matrix, we find the following to  have high correlations
 (r > 0.5):
 1.  Agricultural tilling emissions and: emissions from wind  erosion,  unpaved
     roads, total emissions, total per population, non-road and non-road  per
     population.
 2.  Wind erosion emissions and:  emissions from tilling,  unpaved roads,'total
     emissions, non-road emissions.
 3.  Construction emissions and:  emissions from paved  roads,  from unpaved roads,
     total emissions, total per population, non-road,  and non-road per popula-
     tion.
 4.  Wild fires and area.
 5.  Prescribed fires and none.
 6.  Coal extraction emissions and none.
 7.  Emissions  from extraction of minerals other than  coal and: emissions from
     paved roads, total emissions, and population.
 8.  Paved road emissions and: unpaved road emissions, total  and non-road emis-
     sions, and population (as well, of course, as those  named above:   wind
     erosion,  construction, and mineral extraction other  than coal.)
 9.  Unpaved road emissions and:  emissions from tilling,  wind erosion, con-
     struction, paved roads, non-road emissions, and total emissions.
10.  Mineral tailings and none.
11.  Total emissions and: agricultural tilling, wind erosion  of cropland,
     construction, mineral extraction other than coal,     both paved and un-
     paved roads, population,and non-road emissions.
12.  Non-road emissions and: all the same factors as for  total emissions, ex-
     cept for wind erosion of cropland.
13.  Population and: wind erosion of cropland, construction,  paved and unpaved
     road emissions, and non-road and total emisssions.
14.  Area and:  wild fires.
15.  Non-road total and: wind erosion of cropland, construction, paved and
     unpaved road emissions,tailings, and total emissions.

      The large number of correlations and the fact that  all  those having
 r = 0.50 or larger are positive indicate that states  with one open source emis-
 sion problem have others as well.

 CONCLUDING REMARKS

      The state-by-state emission rate estimates presented here can be useful
 in setting priorities for control.  Many of the emissions are correlated with
 other emissions as well, suggesting that those states with any open sources
 problem probably have several.  The major emission sources were generally un-
 paved roads, wind erosion of cropland, construction,  and paved roads.
                                    38

-------
                                 SECTION 4

                             FIGURES OF MERIT
INTRODUCTION

     Particles impair health, increase rates of weathering and corrosion
of materials, contribute to the damage of vegetation, and decrease visi-
bility.  The decision to control open sources of emissions, and the selec-
tion of a strategy of control, affect society in two ways: the amount,
type, and location of damage done by particles is altered, and resources
are diverted from other uses to control open sources of emissions.  In an
ideal world, the optimal level of control would be easily determined.   Con-
trol expenditures would be increased to the point at which the marginal so-
cial  benefit (associated with emission reductions) was exactly equal to
the marginal social cost (of reducing emissions).  In this ideal world
costs and benefits would be measured in units of social welfare, i.e., the
system of measurement would incorporate simultaneous consideration of equi-
ty and economic efficiency.  In the real world imperfect information pre-
vents us from knowing how close our actual decisions are to these optimal
decisions, and even with such information it might be impossible to achieve
an ideal solution.  Nonetheless, consideration of these abstract models may
help us avoid grossly inefficient or inquitable social policies.
DOSE-RESPONSE RELATIONSHIPS

     In order to develop an open source model  useful in the selection of
an optimal control policy, we would first consider the physical system's be-
havior, and later translate measures of physical performance into measures
of social desirability.  The elements of the physical system are: source,
 control  system, transport system, and receptor.  A control policy consists
simply of the specification of which sources to control, the selection of
a control method (and possibly, degree of control for each source requiring
control)?
*In a systems model we hope to relate measures of output (here, air pollu-
tion damages) to parameters of the system which we can vary (control sys-
tem expenditures).  To complete a systems model, it must be possible to
mathematically describe the behavior of each component of the system.
                                    39

-------
     As the value of any control policy is directly contingent upon reduc-
tion in physical damages to receptors, we feel that a brief discussion of
the response of various receptors to particles is in order.t

     As air is such a prevasive medium, it would be difficult to make a
comprehensive listing of receptors.  It is conceivable that certain or-
ganisms and/or materials     benefit   by the changes in air quality in-
duced by man's activities.  However, traditionally discussions of the "re-
ceptors" of "air pollution" have been limited to those materials and/or
organisms which are thought to be adversely affected by air pollution, e.g.:

          i. humans
          ii. vegetation
          iii. materials

Although a large number of crops (including potatoes, corn, tomatoes, green
beans, lima beans, soy beans, grapes, oranges, tobacco, spinach, peanuts,
and alfalfa) and many other plants are known to be injured by air pollution,
oxidants (e.g.,ozone,nitrogen dioxide, and peroxyacetyl nitrate)  and sulfur di-
oxide are most frequently implicated.  Similarly, although Yocum and McCal-
din  have identified damage to metals, building materials, paint, leather,
paper, textiles, dyes, natural and synthetic rubber, glass, ceramics, and
electrical equipment as a portion of the social costs of air pollution,
sulfur dioxide, acid gases, and oxidants are the most frequently cited
causes of damage.

     In contrast, physiologists have identified three mechanisms by which
exposure to particles could affect human health:

     i. direct chemical or toxic action of the particle, per se: e.g.j,
        arsenic, lead, beryllium;

     ii. indirect effects which are mediated by the physical properties
         of the particle; e.g./mucous flow, slowed ciliary beat, over-
         loading macrophages, or penetration of the  alveolar membrane^

     iii. adsorption or absorption of gases or solutes, thereby increasing
          their effect by holding them more focally in the deeper portions
          of  the lungs for longer periods of time.

     Exposure to air pollution is thought to influence the rates of inci-
dence and prevalence of disease, and  to affect its severity.  Asthma, em-
physema, chronic bronchitis, respiratory infections, lung  cancer, angina
pectoris, arteriosclerosis, myocardial infarction, and conjunctivitis  are
diseases which are currently suspected of being induced (either directly,
or indirectly by increasing susceptibility to a causal factor) or aggra-
 tEven if we were able to complete the physical systems model, a full social
 model       would include evaluation of the social desirability of alterna-
 tive policies.  Currently there are both theoretical and practical limita-
 tions to attempts to place a value on reductions of mortality and morbidity,
 effects of great significance in any air pollution model.


                                    40

-------
vated by community air pollutants.  This increased incidence, prevalence,
and severity of disease is thought to indirectly contribute to reductions
in life expectancy.

     Although much effort of epidemiologists, toxicologists, physicians,
biochemists, and physiologists has been devoted to the analysis of these
relationships, many questions remain largely unanswered.  These would in-
clude:

       i. At the levels of air pollution which currently prevail in
          the U.S., how much (if any) additional life shortening is
          produced by each incremental unit of air pollution exposure?

       ii. Which pollutants are responsible for any reductions in life
           expectancy or increments in morbidity?

       iii. Are the observed effects due to chronic exposure to annual
            mean levels of pollution, or are all effects due to the
            occasional periods of peak exposure?

Until these issues are clarified, we are left simply speculating about the
appropriate form of weighted human dose  function.   A few hypothetical weighted
human dose          functions are outlined here simply to illustrate the
wide range of intuitively attractive alternatives.

     FORM 1:  DT = Q/T CTSpdt

                     T
     FORM 2:  D,,, =  /  C    dt
               T   oj   mrp

     FORM 3:  D  =  /T [C    - K   ]dt
               T   cr    mrp    mrp

     FORM 4:  D  =  JT [C    - K   ]Pdt
               T   cr    TSr    mrp
                         m                       rp
     FORM 5:  D_ = a  ,  J [C  , - K Jdt  + aD _  / [CL  „ - KD  ']dt + ...
               T    cd 
-------
     K = a threshold for given pollutant

Clearly, more complex measures of dose (e.g.tincluding synergisms and an-
tagonisms, non-linear relationships with concentration)  may be required to
adequately reflect human physiological response to particles in the at-
mosphere.


COST-EFFECTIVENESS CRITERION

     This uncertainty as to the response of the system to increased ambient
concentrations of particles limits our ability to carry out a full cost-
benefit analysis.  We must be content with an analysis of the cost-effective-
ness of alternative control strategies.  In a cost-effectiveness analysis
we seek to minimize the costs associated with achievement of a specified
level of effectiveness.  By combining several cost-effectiveness analyses,
we can determine the minimum cost associated with any specified level of
effectiveness.  Alternatively stated, for any level of control expenditure
we will know how to achieve the maximum level of effectiveness:
           LEVEL
             OF
       EFFECTIVENESS
                                       COSTS
      Many measures of effectiveness might be suggested: reduction of
 mean annual TSP concentrations, reduction of peak TSP concentrations,
 reduction of population-weighted TSP levels, reduction of inhalable
 particulate levels, reduction of respirable lead levels, reduction
 of asbestos emissions, reduction of frequency of exceeding a threshold
 TSP concentration, reduction of mass emissions of soil-like particles,
 reduction of emissions of particles with aerodynamic diameters less
 than 10 ym, or a weighted sum of these.  In the selection of a criterion
 we balance two factors: the degree to which we believe the criterion is
 correlated with human response, and the amount of effort required to com-
                                    42

-------
plete an analysis involving the criterion.*  Relatively complex criteria
are likely to be highly correlated with human response, but may require
considerable computational efforts.


CONCLUSIONS

     We recommend a sequential analysis of the open source problem.   At-
tractive alternatives could be identified in a preliminary cost-effective-
ness analysis using reduction of mass emissions of particles not signifi-
cantly more toxic than soil as the criterion of efficiency.t  In the second
stage of analysis, promising alternatives could be more thoroughly inves-
tigated using a more sophisticated criterion, such as a weighted sum of
reductions in population-weighted annual mean TSP exposures and peak TSP
exposures.  By allowing the weights to vary, a "production possibility
frontier" could be sketched for each budgetary level (B,.,) :
       reduction in
       peak exposure
                              reduction in mean exposure


     In the  following section, we apply the cost-effectiveness approach to
 the  largest  source of soil-like particles, reentrainment of road dust from
 unpaved roads.
 *In this balancing we must bear in  mind that  policies  based  on distant  sur-
 rogates  for human response may involve socially  inefficient  expenditures  of
 control  resources.

 tThis could be accompanied by a qualitative analysis of the  feasibility of
 control   and estimation of control  costs for  sources emitting potentially
 toxic particles,  e.g., asbestos, lead,  cadmium....
                                    43

-------
REFERENCES
.1. Yocom,  J.  E.  and  R.  0. McCaldin.   Effects  of  Air  Pollution on Materials
   and  the Economy.   In: Air Pollution,  Stern, A.  C.  (ed.),  Academic Press,
   Vol.  1  (2nd Ed.),  1968.
                                     44

-------
                                SECTION 5

                    EVALUATION OF CONTROL STRATEGIES:
           COST-EFFECTIVENESS OF THE CONTROL OF UNPAVED ROADS
INTRODUCTION

     Dust arises from unpaved roads through three mechanisms:   wind erosion
of the unpaved surface, action of the tires,  and surface disturbance by the
aerodynamic wake behind the vehicle.  Heinsohn, Birnie,  and Cuscino20 note
that estimates of crude emission factors range from 0.04 to 55.9 pounds per
vehicle mile  and postulate that the extreme variability in these estimates
stems from underlying variability in:

     1.  speed and shape of the vehicle,
     2.  physical characteristics of the road surface,
     3.  meteorological conditions,
     4.  size distribution of the aerosol, and
     5.  uncertainty in sampling and estimation methods.

     By parameterizing the emission factor, uncertainty can be separated from
true variation.  The emission factor currently sanctioned by the EPA includes
such parameterization:

     EQ - (0.60) (0.81) (Sr)(^

where
     E  = emission factor (Ibs/veh-mi)
     S° = road surface silt content (percentage < 75 ym diameter)
     S  = average vehicle speed (mph)
     w  = number of days per year with > .01 inch of rain, or snow cover.

Considering the wide fluctuations in climate and soil types across the nation,
it should not be surprising that the average emission factors for states vary
from 3.9 pounds per vehicle mile in South Carolina to 13.2 pounds per vehicle
mile in California.

     Generic methods of pollution control include alteration of the source,
the transport mechanism or the receptor.  In air pollution control, efforts
have been concentrated on source control.  Open source control is no excep-
tion.  Widely-used methods of road dust control include paving, oiling,
watering, and the application of calcium chloride.  Speed reduction is often
suggested as an emission control method, since emission factors increase
at least proportionally with vehicle speed.  Although many methods of chemi-
cal stabilization have been tested (and appear to be more effective than
oiling), not enough data is available to permit evaluation of their cost ef-
ficiency.  Briefly, we summarize the estimates of control cost and effi-
ciency for each of the feasible control methods below.

                                   45

-------
CONTROL COSTS AND EFFICIENCIES

Paving

     The term "paving" is very ambiguous.  Paved surfaces vary from what
is referred to by the industry as a "single chip seal" to portland cement
surfaces over several inches of bituminous material.   The costs of paving
and the life of the paved surface will clearly depend not only upon the
surface type, but also on the condition of the existing surface, traffic
density, severity of climate, and regional costs of aggregate and labor.
The estimates of initial cost and expected surface life which we have com-
piled are presented in Table  5.1.  In order to facilitate comparison of
these cost estimates we have inflated capital costs to equivalent mid-1977
levels using the "bituminous surface cost index," and have made similar ad-
justments in the annual operations and maintenance figures on the basis of
the "highway operating and maintenance price index."8*9  These estimated
1977 equivalent costs are shown in brackets.

     The efficiency of paving as a control method depends very directly
upon the characteristics of the unpaved surface being considered for treat-
ment.  By simply comparing the range of emission factors for paved and un-
paved roads developed by Chatten Cowherd et al. at Midwest Research In-
stitute, we can begin to estimate the range of control efficiency:  »

                                            paved emissions

                              industrial      residential       commercial

                                 .0243             -0108            .0026
unpaved emissions              Ib/veh-mi        Ib/veh-mi       Ib/veh-mi

 South Carolina:3.9 Ib/veh-mi  99.38%           99.72%           99.93%

California:13.2 Ib/veh-mi     99.82%           99.92%           99.98%
                                                                  o
In his M.S.E.  thesis, J.W.Roberts presented emission factors  for:J

      -  a  gravel road                       6.0-8.1    Ibs/veh-mi
      -  a  paved road without curbs             0.83     Ibs/veh-mi
      -  a  paved road with curbs-               0.14     Ibs/veh-mi
        flushed weekly and swept  bi-weekly

 These data indicate that simple  paving of the main road  surface might  yield
 86 percent or more reduction of  emissions,  and  that  the  addition of curbs
 in combination with periodic cleaning could bring the efficiency above 98%.

 Watering

      High rates of infiltration and evaporation combine  to limit the general
 effectiveness of watering as a long-term means of  dust supression.  Nonetheless,
 it may be an appropriate control method for short-term supression of nuisance
 dust, such as during highway construction.   Support  for  these beliefs is


                                    46

-------
               TABLE 5.1.   COST ESTIMATES FOR  PAVING  UNPAVED ROADS
Surface Type


1. 3" asphalt
                     Change in Operating
Initial Cost  Life  and Maintenance Costs  Year  Source
  ($/mile)(yrs)        ($/mile/yr)
   26,400
  [40,890]*
20   -2665[-3627Jgravel     1973  Roberts"
2.  low-type bituminous
              5-12  -124[-267]gravel
                    +219[+471]macadam
                                                                    1965  Winfrey
3. high-type bituminous
             12-20  +195[+419]gravel
                    +538[+1158]macadam
                                                                    1965  Winfrey
4. double bituminous

5. asphalt bituminous

6. 2" hot bituminous
   19,000
                         -1365 gravel

                         -1020 gravel
7. "hot mix"  15,134[15,470]+6278[6417]  -
                     per mi paved
                            1977  Mississippi

                            1977  Mississippi

                            1978  North Dakota

                            1975  Vermont
 Figures in brackets are the corresponding dollar value in mid-1977 dollars.
                                         47

-------
found in both the open literature and    the correspondence we received
from the state highway departments:

     "...watering is not a feasible method of effective dust con-
      trol on public roads because of the high frequency of treat-
      ment required.  However it may be used advantageously on un-
      paved roads under special circumstances...." - Jutze and Axetell

     "Water, as a means of controlling dust is too costly, time con-
      suming, and ineffective to use when dealing with the many miles
      of non-hard surfaced roads we have. ' This is due to the quick
      evaporation of water, therefore necessitating many applica-
      tions...." - State of Virginia12

To provide some quantitative basis for comparison of watering with other
control methods, we cite the 1978 estimate of the State of North Dakota,
"The average cost of water is $200/mile/day and, rating effectiveness,
on a scale of 1-10 rates about 4."6  Using this daily cost, and assuming
that is used as a permanent control watering would be required on each
dry day of the year, we find annual costs ranging from $41,400/mile in
West Virginia to $62,800/mile in Arizona.  Regional variations in water
prices, which are not considered here, would be expected to increase the
geographic variation in these annual costs.

Oiling

     Oil may either be sprayed onto the road surface, or it can be mixed in-
to the top few inches of surface material.  Recycled oils, standard commer-
cial oils, or specially formulated dust control oils, which typically are
petroleum resins cut back with light hydrocarbon solvents, may all be used
for dust control.  Although oiling may intuitively seem an attractive inter-
mediate between watering and paving, practice does not seem to support
intuition:11

     "Application of a surface chemical treatment for dust suppression
      is a relatively inexpensive control method.  However, in tests
      on public roads conducted by several different highway depart-
      ments, no commercial material has been found which retains its
      effectiveness over a reasonable period of time (e.g., two months)
      under traffic conditions....  An alternative intermediate in cost
      and effectiveness between paving and surface treatment is working
      the stabilization chemicals into the roadbed to a depth of two
      to six inches.  This construction technique has been used extensive-
      ly in the San Joaquin Valley, where locally available petroleum
      by-products provide a cheap material for oiled earth roads... the
      results so far are not encouraging.  The construction cost approaches
      that of a single bituminous chip seal surface, and the resulting
      road has a shorter lifespan with comparable maintenance..."

From our correspondence with state highway departments, it appears that
oiling is not even particularly attractive for short-term control of dust:
                                   48

-------
     "On construction projects  where temporary  gravel  roads
      are being maintained,  watering seems  to be the most
      cost-effective.  Normally a contractor is given  the
      option to either water or oil the road and he usually
      elects the watering option." - State  of Michigan

     "Very little dust oil has  been used in our State...we            ^
      require watering during construction..."  - State of  South Dakota

Estimates of the cost and effectiveness of  various methods of  oiling  for
dust control are presented in Table 5.2.

     The costs in brackets in Table 5.2 were adjusted  to mid-1977  price
levels using the "refined petroleum products index" to adjust  materials
costs, and the "general index of hourly and weekly earnings" to adjust
applications costs.17  Estimates of the required frequency of  application
necessary to maintain desirable surface properties and continuous  control
of dust are sparse, and highly varied.   Whereas Roberts's  estimate of the
average life of an oiled surface is four to six years, Sultan states  that
oil controls dust effectively for at most 25 days.  An estimate of the dif-
ference in annual maintenance costs for gravel  and oiled road surfaces is
provided by Roberts^:  gravel,  $2910/mile;  oiled, $1106/mile.

     The control efficiency of oiling has been  reported to be between 50
and 90 percent, as indicated in  Table  5.2  presented below.  Although  we
have not seen data reported in this form, it would seem clear that for any
material and method of application the average  efficiency of control  is a
function of the frequency of application.

Calcium Chloride

     Calcium chloride has a deliquescent effect.  At any relative humidity
above 29 percent it will absorb moisture into  the road surface from the
atmosphere.  In addition to the direct reduction of emissions due to  this
increased soil moisture, soil  stability is increased as a result of com-
paction of  clays.  In arid and semi-arid regions it has found widespread
application,as  it  is relatively  effective and inexpensive.  In regions which
undergo frequent wetting and drying, it is impractical since calcium chloride
is  readily  leached from  the treated road surface.

      The information which we  obtained  in  correspondence from the state high-
way departments permits  estimation of  the  cost  and effectiveness of this
control method:

      "...Calcium chloride is used  in special cases and works best
      when  the humidity  is high.   It costs $400/mile/week [1978]
      and rates 6  [on  a  scale  of 10].   Calcium  chloride is  con-
      sidered  one  of the best  and  most  economical methods of dust
      control	" - State of  North Dakota6

      "Presently the  bulk of  our  dust control is done with calcium
      chloride...  A  calcium  chloride treatment  of  one pound per
      square  yard  costs  approximately  5c...."  - State of Virginia


                                    49

-------
                                 TABLE 5.2  COST ESTIMATES FOR OILING UNPAVED ROADS
                  Material
       Method of
Control
Cost
Year
Source
Ui
o

1.
2.

3.
4.
5,

PS 300 oil
StandardHDust
Control Oil

oil
emulsified asphalt
S t and ar cruus t
Control Roil
application efficiency
not specified(n/s) n/s
0.5 gal./sq. yd. - (98%)
sprayed

not specified n/s
not specified (80%)
0.5 gal./sq. yd. -
mixed 3 inches fol- (88-91%)
lowed by 0.1 gal./
sq. yd. spray
($ /mile/application)
2000 [3202]
1100-3153 [2650-7597]
+I492[l930]cost of
application
5950
2000
1320-3784[3181-9117]
SF +U72[>1930]
application

1973 Roberts3
1974 Sultan15
1973 Roberts3
1978
1978
1974 Sultan15
1973 Roberts3
        6. not specified
a. applied to prepared    (50%)
   surface

b. applied to unprepared  (50%)
   surface
                               c. worked into the
                                  roadbed
                          (50%)
             2000-3000[3322-4164]    1974
                                                                     1000-2000[1661-3322]    1974
            5000-12000[8305-19932]   1974
                            Jutze and
                                                                                                        Axetell
                                                                                                               11

-------
If these treatments were repeated weekly, the annual costs of application
per mile treated would be $20,800 in North Dakota and $38,116 in Virginia.

     A commercially available material, known as Soil-Lok, combines the
properties of calcium chloride with those of sodium silicate.  Experiments
in arid regions demonstrated that sodium silicate is a very good dust-
proofing agent.  Sodium silicate is a good inorganic cementing material
capable of bonding or hardening the soil.  In combination, the two chemicals
react within the sand pores to form an impervious gel which hardens, binding
the sand particles into a solid mass.  Although Sultan reported that Soil-
Lok showed no damage after six months and that it affords good protection
against wind erosion, he did not provide any information on costs or re-
sistance to traffic and rain. ^

Speed Control

     As emission factors varyat least proportionally with vehicle speed,
speed control is often suggested as a means of reducing dust emissions from
unpaved roads.  Although some have naively suggested that the only costs
associated with speed control are the costs of administration and enforce-
ment, we believe that social costs are imposed in the form of increased travel
t imes.

     The increased travel times are easily calculated.  Using 40 mph as a
reference, we can see that 0.5 minutes per mile will be added to travel times
by a reduction to 30 mph, and 3.0 minutes per mile by a reduction to 20 mph.
In order to estimate the social costs associated with these increases, we
must consider traffic densities and the economic value of this "lost time."
            4
     Winfrey   in 1969  recommended valuing generalized passenger car travel
between $1.00  [$1.68] and $4.00 [$6.67] per car hour.   The Interstate Commerce
Commission (1965) estimated that the value of time losses for commerical ve-
hicles varied from $3.65 [7.66] to $8.29 [17.39] per hour.4  These values
have been adjusted to mid-1977 price levels using the "index of general hourly
and weekly earnings. "*-'

     The control efficiency which we attribute to various degrees of speed
reduction follows directly from the functional form of the relationship be-
tween average vehicle speed and emissions.  On the basis of proportional rela-
tionships developed by Cowherd et al. we would expect a 25 percent reduction
in emissions with a decrease from 40 to 30 mph, and a 50 percent reduction
if average speeds were reduced from 40 to 20 mph.   Other researchers have
suggested non-linear relationships:

     1.  Jutze, Axetell, and Parker of PEDCo Environmental Specialists
     developed emission factors for six AQCR's in New Mexico, Nevada,
     Arizona, California, and Texas.    Their equation of best fit for
     vehicle-generated dust was:

                         E = 0.27 x (1.068)S

     where:  E = emission factor (Ib/veh-mi)
             S = average vehicle speed (mph)

                                    51

-------
    To this must be added a term accounting for wind erosion of
    the unpaved surface.  This second term was derived from the
    empirical wind erosion equation of Woodruff and Siddoway   .
    Using a wind erosion rate of 3 tons/acre/year, or 9 tons/mile/
    year, we have combined wind and vehicle generated emissions of

              E = [0.27 x (1.068)8] +
10
20
30
3.5
7.0
22.0
    where:  ADT = average daily traffic (veh/day)
            and all other terms are defined above.

    2.  Roberts, Watters, Mangold, and Russano in Seattle, Wash-
    ington developed the following emission factors for gravel
    roads: ^

        speed   observed emission  factor     £stimated_emission_factor
        (mph)        (Ib/veh.-mi)              "   ~   (Ib/veh.-mi)

                                                       3.3
                                                       8.2
                                                      20.7

          40             -                             52'3

       We have fit  these data to the  equation:

               E =  1.29 x  (1.097)8

Exploring these relationships, we see that Robert's  data would suggest  a 60
percent reduction in emissions accompanying a speed  reduction from 40 to
30 mph, and a 84% emissions reduction with a reduction to 20 mph.   The
PEDCo data demonstrate the dependency of emissions reduction efficiency on
traffic density:11

              Speed Reduction    Average Daily Traffic
                                  50      500
                   40-»30 mph      38%      47%     48%
                   40+20 mph      58%      71%     73%

      Intuitively it might seem  that speed reductions  should be credited
with  reduction of the  frequency and severity of  accidents.  However, we have
not been  able to find  reliable  data on accident  frequency or  severity as a
function  of average vehicle  speed, applicable to unpaved roads.

COST-EFFECTIVENESS RATIO

      From this basic  information  on  the components  of cost  and anticipated
 efficiency of control  we can compute  a  single number  which  characterizes  the
 relative cost-efficiencies of the various control methods.  Let us  arbitrarily
 select a mile of unpaved road as  the  unit of  interest.


                                    52

-------
     First,  we must adjust all costs to the same price levels.   We have
chosen to adjust to the levels which prevailed in mid-1977.   Then we must
combine the costs of capital investments with the recurring  costs associ-
ated with maintenance.  To combine costs which occur at different points
in time, one can compute annualized cost.  The total annualized cost is
the average yearly cost of items which involve repeated yearly  expenditures,
such as maintenance, plus a yearly pro-rating of costs which do not,  in
fact, occur yearly, such as initial capital investment.  To  calculate the
total annualized cost we add the average annual cost of operation and main-
tenance, AC, to the product of the capital recovery factor, CRF, and the
initial capital investment, K.  (The capital recovery factor is the per-
centage of the initial investment which one would pay yearly on a loan or
mortgage, at an interest rate, i, for the life of the loan,  n years.)  The
relevant equations are:

                         TAG = AC 4- (CRF)K
                         CRF -
                               (1 +
In Table 5.3 we summarize the range of estimates of the total annual costs
of application of each of the feasible control methods to one mile of un-
paved road.  The figures in the table are based upon a discount rate, i,  of
10 percent per year.  Without further computation, it is clear that any
method which has both higher annualized costs and lower control efficiency
than some other method, can be culled.  Similarly if one method has both
lower annualized costs and higher control efficiencies than the others, it
clearly dominates.  Generally, however, it will be helpful to calculate a
single index of relative cost-efficiency.  In Table 5.4 we have divided the
total annualized costs by the control efficiency of each method to derive
such a measure.  The wide ranges in these cost estimates reflect two in-
fluences:  regional variation in the costs, lifetimes and efficiencies of
control; and imprecision and uncertainty in the estimates of these param-
eters. In the face of uncertainty, an initial selection of the dominant con-
trol technology may be made by examining the mid-ranges of the cost estimates.

     Although speed control from 40 mph to 30 mph appears to be economical
for roads with low traffic density (- 80 vehicles/day), for permanent or
continuous reduction of emissions from roads with more typical traffic den-
sities*   paving appears to dominate, with a standardized total annual unit
cost of $88. 98/yr-mile-percent control, or $7653/yr-mile at 86 percent con-
trol efficiency.  This analysis suggests that regardless of location or
existing road surface type, by paving we can reduce mass emissions more
for each dollar spent on control than through the application of any other
control technique.

     Now,  if we are interested in maximizing the nation-wide reduction of
dust emissions for various levels of national control expenditure, we must
account for regional variations in emissions factors and traffic densities.
For example, it is clear that paving a road in California (average statewide
emission factor = 13.2 Ibs/veh-mi) will reduce national dust emissions more
than paving a road with similar traffic density in South Carolina (average

                                   53

-------
          TABLE 5.3.  ESTIMATED COSTS AND DUST EMISSION REDUCTIONS
                 OF VARIOUS TREATMENTS FOR UNPAVED ROADS.
Control Method
paving

oiling

watering

calcium chloride

speed control:
   40 to 30 mph

speed control:
   40 to 20 mph
Total Annualized Cost Estimates

     ($/yr /mile @ i=10%)

    low               high
  -1,810

  -1,423

  41,400

  20,800


  5.08 x ADT


  30.60 x ADT
 +16,861

+291,413

  62,800

  38,116


52.68 x ADT


317.35 x ADT
                Range of Efficiencies

                    (% reduction)
86-99.98

50-98

40

60


25-60


50-84
                                     54

-------