PB-220 433
DEVELOPMENT OF METHODOLOGY TO PERMIT
PROJECTION OF AIR POLLUTION EMISSIONS FOR
GEOGRAPHIC AREAS
RESEARCH TRIANGLE INSTITUTE
PREPARED FOR
ENVIRONMENTAL PROTECTION AGENCY
FEBRUARY 1973
Distributed By:
National Technical Information Service
U. S. DEPARTMENT OF COMMERCE
-------
600R73004
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
APTD-1464
3. Recipient's Accession No.
PB-22;0 433
4. Title and Subtitle
The Development of Methodology to Permit Projection of Air
Pollution Emissions for Geographic Areas
5. Report Date
February 1973
6.
7. Author(s)
8. Performing Organization Rept.
N°- 41U-723
9. Performing Organization Name and Address
Research Triangle Institute
Research Triangle Park, North Carolina 27709
10. Project/Task/Work Unit No.
11. Contract/Grant No.
68-02-0253
1 2. Sponsoring Organization Name and Address
ENVIRONMENTAL PROTECTION AGENCY
Applied Technology Division
Research Triangle Park, North Carolina 27711
13. Type of Report & Period
Covered
Final
14.
15. Supplementary Notes
16. Abstracts The purpose of this study was to provide a conceptual design of a model to
project regional air pollution emissions. Existing national economic forecasting models
were examined to determine the extent to which such a model could be disaggregated to
provide regional forecasts. Only the OBERS model used by the U.S. Department of Commerce
was found to be appropriate, and its use was primarily to provide control totals by
State. It was determined that new model components would be required to project region-
al values for area and mobile sources and for each industrial source to be forecast.
Mobile Source emissions can be estimated on the basis of population, the stock of motor
vehicles and its age composition, and vehicle use patterns. Area source emissions can
be estimated on the basis of population, fuel use patterns, and projected employment.
Industrial point source emissions will require" the application of industrial growth and
location theories to each industry individually;- The required data is available or can '
be developed for each model component. The analysis sets forth the functional relation-
ships of the proposed model, its structural and data requirements, and evaluates its
limitations.
17. Key Words and Document Analysis. 17a. Descriptors
Air pollution
Emission
Models
Computer programs
Economic forecasting
Projection
Vehicles
Electric power generation
Regions
Geography
17b. Identifiers/Open-Ended Terms
Regional projection model
Design
Fuel consumption
Population (statistics)
Industrial plants
NATIONAL TECHNICAL
INFORMATION SERVICE
US O.p»rtm.nl of Comm.it*
. VA. 22151
17e. COSATI Field/Group 133
18. Availability Statement
Unlimited
19.. Security Class (This
Report)
UNCLASSIFIEt
££-
Thi:
20. Security Class (This
Page
UNCLASSIFIED
21. No.
L22—R.:-
FORM NTIS-33 IREV. 3-72)
USCOMM-qff M952-P72
-------
ACKNOWLEDGEMENT
RTI gratefully acknowledges the Information and assistance provided
by many individuals and agencies in the conduct of this study. Among those
whose contributions were especially significant in evaluating existing
economic models were Robert Graham, Bureau of Economic Analysis, U. S.
Department of Commerce; Curtis Harris, University of Maryland; Wesly Long,
Economic Development Administration, U. S. Department of Commerce; Manny
Helsner, National Planning Association; Niel FitzSimons, Office of Civil
Defense; and Hugh Pitcher, Institute for Defense Analysis. James Cavender,
who served as Project Monitor for the Environmental Protection Agency, was
consistently understanding and helpful. Other personnel of EPA who
assisted with parts of the analysis included David Kircher, Wayne Ott, and
Alan Basala.
The project was under the direct supervision of Alvin Cruze and
.David LeSourd. Responsibility for analysis and writing of the sections
of this report was as follows: Alvin Cruze, for the evaluation of existing
models; Tayler Bingham, for model evaluation and analysis of mobile sources;
Paul Mulligan, for the treatment of industrial point sources; David LeSourd,
for the treatment of area sources; and Robert Thornton and Philip Cooley,
for defining the model structure and computer requirements.
ii
-------
TABLE OF CONTENTS
LIST OF TABLES .................... ..... vll
LIST OF FIGURES . . ........ .............. vlii
SECTION 1 - INTRODUCTION AND SUMMARY .............. 1-1
1.1 Scope of the Research ............... 1-1
1.2 Summary ............ , ......... 1-2
1.3 Model Structure ............... ... 1-4
1.3.1 Mobile Sources ... ............ . . 1-4
1.3.2 Area Sources ............. ...... 1-4
1.3.3 Point Sources .................. 1-5
SECTION 2 - EVALUATION OF MODELS TO PROJECT REGIONAL ECONOMIC
ACTIVITY ................ ...... 2-1
2.1 Introduction ........... ........ 2-1
2.2 Evaluation .................... 2-2
, SECTION 3 - PROJECTING EMISSIONS FROM MOBILE SOURCES ...... 3-1
3.1 Introduction ......... . ......... 3-1
3.2 Gasoline Powered Motor Vehicles .......... 3-1
3.2.1 General .................... . 3-1
3.2.2 Passenger Cars ................. 3-2
3.2.3 Trucks .................. ... 3-14
3.3 Diesel Powered Motor Vehicles ........... 3-19
3.3.1 General ..................... 3-19
3.4 Aircraft ............... ...... 3-20
3.4.1 General ... .................. 3-20
3.4.2 Aircraft Activity Projection Model ..... . . 3-22
SECTION 4 - PROJECTING EMISSIONS FROM AREA SOURCES AND
FOSSIL-FUELED ELECTRIC GENERATING PLANTS ...... 4-1
4.1 Introduction ................... 4-1
4.2 Estimation of Fuel Consumption .......... 4-2
4.3 Residential Heating ..... .... ...... . 4-2
4.4 Commercial, Institutional, and Industrial Heating . 4-3
4.5 Fossil Fueled Electric Generating Plants ..... 4~4
iv
Preceding page blank
-------
TABLE OF CONTENTS (Continued)
Page
SECTION 5 - PROJECTING POINT SOURCES OF AIR EMISSIONS
ON A REGIONAL BASIS 5-1
5.1 Introduction . . 5-1
5.2 Significance of Point Sources of Emissions .... 5-1
5.3 Projection Requirements . 5-7
5.3.1 Expected Changes in Point Sources of
Air Emissions 5-7
5.3.2 Additional Data Needs 5-11
5.3.3 Relative Efforts in Making Projections
of Point Sources 5-11
5.3.4 Decision Criteria 5-15
5.4 Review of Point Sources of Emissions 5-17
5.5 Computer Model . 5-21
SECTION 6 - CONSTRUCTION OF THE COMPUTER MODEL 6-1
6.1 Scope of Work . , 6-1
6.2 Projection Strategy . 6-1
6.2.1 General 6-1
6.2.2 Operational Aspects: Model Methodology .... 6-2
6.3 Computer Model Characteristics 6-3
6.3.1 Introduction 6-3
6.3.2 Model Input 6-4
6.3.3 Model Outputs 6-5
6.3.4 Model Documentation 6-6
6.3.5 Other Considerations 6-6
6.4 Computer Model Implementation 6-7
6.5 Validation of the Model 6-8
SECTION 7 - FEASIBILITY, CAPABILITY, AND LIMITATIONS 7-1
APPENDIX A - EVALUATION OF NATIONAL MODELS
APPENDIX B - IRON AND STEEL INDUSTRY
APPENDIX C - PRIMARY COPPER
APPENDIX D - PRIMARY LEAD
APPENDIX E - PRIMARY ZINC
-------
TABLE OF CONTENTS (Continued)
APPENDIX F - PRIMARY ALUMINUM
APPENDIX G - PETROLEUM REFINING
APPENDIX H - SECONDARY NONFERROUS METALS
APPENDIX I - THE PULP AND PAPER INDUSTRY
APPENDIX J - PHOSPHATE FERTILIZER INDUSTRY
APPENDIX K - SULFURIC ACID
APPENDIX L - RUBBER TIRES
APPENDIX M - COAL CLEANING
APPENDIX N - FEED AND GRAIN INDUSTRY
APPENDIX 0 - ASPHALT BATCHING
APPENDIX P - CEMENT
APPENDIX Q - BRICK MANUFACTURING
APPENDIX R - LIME INDUSTRY
APPENDIX S - GASOLINE MARKETING AND BULK STORAGE
vl
-------
LIST OF TABLES
Table Page
1 Annual Mileage as a Function of Vehicle Age 3-9
2 Percentage Distribution of Annual Vehicle-Miles
of Truck Travel 3-16
3 Annual Truck Miles and Size Class and Year
Model of Truck: 1963 • • • » 3-19
4 Emissions from all Sources for 298 Metropolitan
Areas and for the Average Metropolitan Area . . 5-4
5 Industrial Process Sources - Estimates of Emission
Levels, 1967 C298 Metropolitan Areas) 5-5
6 Average Emissions from each Point Source, 1967 . . . 5-6
7 Emissions from Average Point Source as a Percentage
of Background Emissions for the Average
Metropolitan Areas .... 5-8
8 Expected Significant Changes in the Location.,
Production, and Process of Each Point
Source, 1967-1980 5-9
9 Additional Data Needs for Project Point Sources
of Air Emissions 5-12
10 Relative Effort in Making Projections of Selected
Aspects of Point Sources 5-13
11 Decision Table . 5-16
vii
-------
LIST OF FIGURES
Figure Page
1 Tasks for Projection Regional Motor Vehicle
Activity 3-3
2 New Automobile Sales 3-4
3 Stock of Automobiles 3-4
4 Stock of Automobiles Per Capita 3-5
5 Family Automobile Ownership 3-6
6 Automobile Survival Probabilities 3-8
7 Stock of Trucks 3-15
8 Stock of Trucks Per Capita 3-15
9 Truck Survival Probabilities 3-18
10 Air Traffic Hubs 3-21
viii
-------
SECTION 1
INTRODUCTION AND SUMMARY
The purpose of this project, as stated in the contract, has been
to "perform a study of the feasibility of using modelling techniques in
order to project air pollution emissions for a given geographic area."
To a great extent, feasibility has been determined by defining the
elements required by such a model and the availability of the
requisite data and projection methodologies. The research has produced
a conceptual design of a feasible regional projection model. This
report describes the evaluations that have been made and the model
design that has been evolved.
1.1 Scope of the Research
At the outset of this study it was apparent that what would be
most immediately useful to the Environmental Protection Agency (EPA)
was a computerized model that would project probable air pollution
emissions by source for areas no larger than air quality control
regions (AQCR's) and for dates well into the future. This would
require the design of a means of relating forecasts of aggregate
economic activity to population changes, the growth of output in
specific industries, changing technology, and other variables, in
specific geographic locations. Projections of these variables
could then be used to estimate the pollutant emissions from identified
mobile, area, and point sources. These emissions estimates would
permit EPA to identify those AQCR's in which emissions might be
expected to rise above acceptable levels and, therefore, to initiate
policy changes in cooperation with State authorities before critical
levels of pollution occur.
The scope of this project was limited to consideration of the
feasibility of constructing a model that would (a) provide projections
for two points in time, i.e., 10 years and 20 years from date;
(b) project emissions of particulates, sulfur oxides, nitrogen oxides,
carbon monoxide, and hydrocarbons; (c) be interconnected to projections
1-1
-------
of the aggregate of the national economy; and (d) provide estimates
disaggregated to the county or air quality control region level.
The analysis examined the availability of national econometric
models with regional components that could be disaggregated to the
required level of geographic and Industrial detail. Studies were
made of the required methodology and data availability for projections
of mobile, area, and point sources of significant emissions of the
candidate pollutants. The structure of the computer model that would
be required was also examined as well as the capabilities and limita-
tions of the total model.
1.2 Summary
A number of existing national econometric models were evaluated
to determine their usefulness as a basis for constructing the desired
model. Of these, only that employed by the Bureau of Economic Analysis
(BEA), U. S. Department of Commerce appeared to offer substantial
support for this effort. That model provides estimates of population,
personal income, employment and earnings by Industrial sector for 173
areas of the U.S. These projections can provide essential control totals
against which to check AQCR projections. Some of the BEA data series
can also be disaggregated to provide AQCR data.
In dealing with the problems associated with projecting mobile
source emissions most of the effort in this study has been focused on
passenger cars, gasoline powered trucks, and aircraft, although consi-
deration is also given to other trucks and buses. A method is proposed
for providing projections of automobile emissions based on automobile
registrations and estimated annual vehicle miles. It appears that the
essential data for this method are available and that it can provide
more accurate estimates and be more useful for sensitivity analysis
and policy evaluation than any alternative system. The model design
discussed, however, will not provide full compensation for the inter-
region variations caused by commuting patterns.
A similar projection model is proposed for trucks and its speci-
fications and limitations discussed. Aircraft emissions in the vicinity
of most airports can be projected directly from presently published
Government data.
1-2
-------
Projections of area source emissions from fuel combustion for heat
and power require a large volume of detailed calculations but are not
conceptually difficult. They can be based on fuel consumption patterns
available by State in regularly published data. These can be allocated
to residential, commercial, institutional, and industrial heating use
and projected on the basis of population and employment patterns.
Fossil fueled electric generating plants comprise a special case
of fuel consumption sources. For modelling purposes they have some of
the characteristics of both the area sources and the point sources.
However, a very large body of detailed data is available on electric
utilities and, except for the location problem, projections of these
emissions are relatively easy to construct.
Emissions arising from a number of point sources that are Industrial
plants are considered in this analysis. Projection of emissions from
industrial production processes must be based largely on projections
of each industry's production and capacity and regional allocations
provided by the application of economic location theory. No general
method of disaggregation from national or area forecasts seems likely
to provide the desired level of accuracy at the specified 10 and 20
year time horizons. Projections of these emissions will almost certainly
prove to be the most difficult part of the overall model construction
effort. This analysis provides a detailed evaluation of the significance
of various point sources and the relative complexity of the problems
related to projecting their emissions. It is clearly established that
construction of this segment of the model is feasible.
This report also provided a detailed discussion of the structure
and capability of the computer programs that may be used in the actual
construction and implementation of the desired projections model.
1-3
-------
1.3 Model Structure
1.3.1 Mobile Sources
1.3.1.1 Passenger Cars. The model segment for estimating emissions
from passenger cars will require the following inputs:
1. Equilibrium stock of automobiles per capita, as a function
of income, population density, and other.
2. Retirement rate, as a function of vehicle age.
3. Average annual mileage per vehicle by model year, divided
i
into two average speed classes (rural and urban).
4. Emissions factors by model year and average speed class.
5. Regional distribution of the base year national stock of
automobiles.
The equilibrium stock of automobiles for any projected year includes
new purchases and the carry over of stock from the previous year. Knowing
the retirement rate by model year, it is possible to calculate the age
distribution of the projected stock. This, multiplied by the average
annual mileage per vehicle by model year, gives total mileage, which is
then divided into speed categories, and multiplied by emission factors, to
provide the national emission estimate. When the equilibrium stock is
disaggregated to regions, the same procedure will yield projected regional
emissions.
1.3.1.2 Trucks. Emissions generated by the operation of trucks may
be calculated in the same way as outlined for passenger cars. It is
necessary, however, to classify trucks as light and medium duty gasoline
powered, heavy duty gasoline powered, and diesel powered, and project the
stocks, vehicle miles, and emissions for each category.
1.3.1.3 Aircraft. National and regional ten year projections of
aircraft activity at airports are provided by FAA. Data Is also available
on aircraft types relative to types of use and airport size and future
aircraft use is also forecast by FAA. These projections may be extended to
the years required by this model and the appropriate emission factors
applied to derive emission projections by region.
1.3.2 Area Sources
Required inputs for the calculation of emissions from area
sources are:
1-4
-------
1. Base year consumption of fuels for heating, by state.
2. Population, by AQCR.
3. Average family size.
4. Average number of persons per dwelling.
5. Commercial-institutional and industrial employment.
6. Emission rates by fuel and heating class.
1.3.2.1 Residential Heating. Data are available on consumption of
fuels for heating by state. Heating demand in 1970 for the average residence
in each state may be calculated by dividing total fuel consumption for
residential heating, in Btu's, by the number of dwelling units. Assuming
that this relationship remains constant, future demand can be projected
on the basis of population, family size, and dwelling unit size. Total
residential heating demand can be apportioned to fuel types on the basis of
the market share trends by state, and may be disaggregated to AQCR by county
population. The product of this analysis will then be fuel consumption by
AQCR, from which projected emissions may be estimated.
1.3.2.2 Commercial-institutional and Industrial Heating. Commercial-
institutional and industrial consumption of fuels for heating may be
calculated in similar manner. Baseline data-on fuel consumption for these
categories can be developed from regular Government reports. The ratio
of commercial-institutional heating demand in Btu's to commercial employment
by state, can be used to estimate future consumption as commercial employment
changes. In the same way, industrial heating demand may be projected on the
basis of Industrial employment. The two employment series may be used to
disaggregate state fuel consumption to AQCR's. Emission estimates may be
made by applying appropriate emission factors to the fuel consumption patterns.
1.3.3 Point Sources
Inputs required for the analysis of emissions from Industrial
point sources are:
1. Present location of plants, by industry.
2. Production rates, by plant and process.
3. Process use, by plants in each Industry.
4. Level of controls in place, by plant.
5. Emission rates, by production process.
1-5
-------
Each industrial classification will have to be projected separately.
National output, by 2-digit SIC, can be projected based on available
Government and private forecasts. Data are available from other research
to permit disaggregation of current industry output figures to the
required detail of product type and production process for most industry
categories. Estimates of future production may be based on trend analysis
modified exogenously to allow for expected changes in technology. Geographic
disaggregation will require, for some industries, careful application of
industry location theory to estimate growth of those industries in new
locations. Where limited growth is predicted or where location is determined
by resource availability, and therefore unlikely to change, the model will
assume that any growth occurs where plants are already in existence.
Emission factors related to production rates may then be applied to production
estimates to provide estimated future emissions by AQCR.
1-6
-------
SECTION 2
EVALUATION OF MODELS TO PROJECT REGIONAL ECONOMIC ACTIVITY
2.1 Introduction
In its efforts to improve the quality of the envirpnment, the Environ-
mental Protection Agency (EPA) faces a continually changing situation as
economic and social forces create growth, stagnation or possibly decline
in various areas of the Nation. These changes may cause changes in
industrial production, housing density, motor vehicle operation, and other
activities giving rise to air pollution emissions. Through economic growth
and uncontrolled emissions of pollutants, some areas of the Nation already
have experienced problems associated with critical levels of air pollution.
These areas for the most part are known and for them the primary task is
to reduce or prevent further increases in pollution. Other areas, that
are marginal in terms of pollution problems today, may reach critical levels
of air pollution in the future through economic growth and changes in
production techniques.
Early identification of areas with potential for pollution is required
so that steps can be taken to forestall major pollution problems. This
identification requires a model of regional growth, expressing the inter-
relationships among economic, demographic and air emission factors, within
a framework of projected national growth. This study is the first phase
of a multi-phase undertaking and is designed to conceptualize the overall
model and to specify the work required in later stages.
In order to project air emissions from the variety of sources, a
procedure to forecast the distribution of population and economic activity
among the various regions of the Nation is required.— Several models and
techniques for projecting population and economic activity for isolated
regions—river basins, states, groups of counties—have been developed and
are currently operational. However, since EPA is interested in projections
for regions distributed over the entire Nation, these models are not
appropriate for this project.
— The terms, project and forecast, are used interchangeably in this report.
Each has the same meaning—indicating what is likely to happen in the future
under stated assumptions and the continuation of recent trends in the
Nation's social and economic system. It should be noted that the results
of this process do not indicate what will happen in the future, only what
is likely to happen.
2-1
-------
Several models developed on a nationwide basis, which include pro-
jections of selected variables on a sub-national or regional basis, are
appropriate for this project, «nd have b7een reviewed and evaluated. One
model is recommended for later stages of the overall effort to develop
procedures to project air emissions.
It should be noted that none of the currently existing models is
completely acceptable for EPA purposes. In particular, none provides
projections at the appropriate level of detail, either at the industrial
level or, more Importantly for projecting air emissions, at the industrial
process level of detail. This was anticipated and the survey was planned
to:
1. Ascertain which of the existing models was most appropriate
to provide a) the required regional projections developed within
the framework of national projections and b) the necessary control
totals for more detailed projections, and
2. Recommend additional model development efforts required to
implement the entire system for projecting regional air emissions.
2.2 Evaluation
In order to establish a rational basis for reviewing, comparing,
and evaluating the various existing models for projecting population and
economic activity, a set of evaluation criteria was established. These
include: time frame for which projections are currently available;
geographic and Industrial detail of the model's projections; basic assump-
tions underlying the projections; units in which economic activity is pro-
jected, including consistency among the various measures projected; data
requirements; ease of updating the model and revising the associated pro-
jections; and acceptability for sensitivity analyses in which the changes
in projections of air emissions resulting from changes in model inputs
are evaluated.
Five models were reviewed during this effort:
1. The projections developed for the United States Water Resources
Council by the Office of Business Economics of the U. S. Department
2-2
-------
of Commerce and the Economic Research Service of the U. S. Depart-
21
ment of Agriculture (the OBERS projections),—
2. A multi-regional, multi-industry forecasting model developed
by Curtis Harris at the University of Maryland (the Harris model),
3. The multi-regional input-output model sponsored by the Economic
Development Administration of the U. S. Department of Commerce and
developed at Harvard University (the MRIO model),
4. The economic and demographic projections developed by the
National Planning Association (the NPA projections), and
5. Projections developed at the Institute for Defense Analyses
and sponsored by the Office of Civil Defense of the Defense Depart-
ment (the OCD projections).
In addition, 1980 projections of labor force, aggregate and industry
demand, output, and the industrial and occupational structure of employment,
developed by the Bureau of Labor Statistics (BLS) of the U. S. Department of
Labor are briefly discussed. These projections result from the cooperative
efforts of several government agencies working under the auspices of the
Interagency Economic Growth Project and are termed to BLS projections in
this memorandum. Although the BLS projections do not contain any regional
detail, they do provide certain types of industrial detail at the national
level which might prove useful in the EPA overall model development effort.
Based on the above criteria, the OBERS projections and projection
methodology are recommended for use by EPA to provide the overall frame-
work for projecting air emissions. The OBERS projections provide an
excellent set of consistent regional projections of population, personal
income, employment and earnings at an adequate degree of industrial detail.
The projections are provided for a time frame that falls within the EPA
period of interest (10-20 years into the future). Furthermore, the data
base associated with the OBERS methodology is structured in such a manner
that the projections can be provided for a variety of geographical areas.
21
— The identifiers in parentheses indicate the means by which the various
models are referenced throughout this report. Complete references to
published documentation are provided with discussions of the models in
Appendix A.
2-3
-------
The OBERS projections have been developed for the United States
Water Resources Council and will serve as the official projections of
population and economic activity which must be followed by all agencies
planning water resource projects. Thus, the use of these projections as
the overall framework for air emission projections will insure consist-
ency in the longe-range planning efforts of both air and water programs.
In addition, the Bureau of Economic Analysis (BEA) of the U. S.
Department of Commerce (formerly the Office of Business Economics) is
committed to maintaining the required data base, improving the projection
techniques, and updating the projections on a regular basis. The fact
that these tasks will be accomplished by another organization will permit
EPA to devote its resources to developing and refining the portions of
the overall methodology for projection air emissions.
Review of the OBERS methodology, indicates that the associated
projections will be most useful for projecting air emissions from mobile
and area sources. Emissions from these sources are functions of a
number of variables, two of the most important of which are population
and income. As noted above, projections of these key variables are
contained within the OBERS framework. With respect to projecting emissions
from point sources, the OBERS projections are less complete. The industry
sector and industrial process detail necessary for projecting industry
are not available within the OBERS framework. Therefore, although the
OBERS projections can serve as overall control totals for the necessary
projections of detailed industrial activity, efforts will be required to
develop procedures by which these detailed projections- can be provided
within the OBERS framework.
Details of the review and evaluation of each of the six models are
presented in Appendix A. For each model the discussion is organized
around major headings of general characteristics, detailed characteristics,
operating procedures and overall evaluation for EPA use. To provide
appropriate information for the recommended methodology, the OBERS model
is discussed in greater detail than the other models.
2-4
-------
SECTION 3
PROJECTING EMISSIONS FRO^ MOBILE SOURCES
3.1 Introduction
Mobile sources, especially gasoline powered motor vehicles, are one
of the major sources of air pollution. Projection of emissions from
mobile sources on a regional level is an especially challenging task, not
only due to their mobile character, but a^so because emissions vary with
vehicle utilization, vehicle speed, and the age distribution of the
vehicle population, all of which are difficult to forecast. Methodologies
for projecting gasoline and diesel powered motor vehicle and aircraft
activity and their associated emissions are discussed below- General
equations are presented and variables identified. During the actual
estimation of the relationships the form of the equations and the
variables included may change somewhat. Most of the emphasis is on
passenger cars because of their dominant role in mobile source emissions.
3.2 Gasoline Powered Motor Vehicles
3.2.1 General
Gasoline powered motor vehicles consist of passenger cars, trucks,
buses and various other vehicles. Emissions for this category of mobile
combustion sources vary with vehicle miles, model year and speed of the
vehicle.
Because passenger cars and trucks account for almost all mptor
vehicle emissions it appears most cost-effective to project emissions
only from these sources. Buses, however, can prpbably be fairly easily
and reasonably included with trucks if desired.
2
EPA has a computer program (PAVE-I) for calculating motor vehicle
emissions, given certain inputs. The program essentially solves the
following equation:
n
i=l 1 X 1
4> = total annual emissions,
N. = number of motor vehicles by model year,
q. - emissions factor for each model year,
v. = average annual mileage per vehicle by model year,
n = number of model years.
3-1
-------
3.2.2 Passenger Cars
In 1969 of 105 million registered motor vehicles, 83 percent (87
million) were passenger cars. Because they are the dominant source of
emissions from mobile combustion sources, they have received the most
attention in this study.
The objective of an emissions projection methodology for passenger
cars is to provide a procedure for projecting, first on a national basis
and subsequently on a regional basis, the number of passenger cars by
model year, the vehicle miles traveled by each model year, and the
distribution of those vehicles miles between urban and rural travel.
Figure 1 illustrates the major elements in the suggested model and their
relationships.
3.2.2.1 Passenger Car Projection Model. Most economists working
on the problem of automobile demand have been concerned with projecting
the short run demand for automobiles. They have examined such variables
as: consumer income, automobile prices, credit conditions, used car
allowances, operating costs and dealers' used car stocks in an effort to
explain the shifts in demand for new automobiles. Some insight into the
difficulty of such projections of new automobiles can be gained by
observing the large changes in the sales* of new automobiles £rom year
to year, shown in Figure 2.
Since we are interested in long :term trends in automobile ownership
rather than year-to-year changes in the demand for new automobiles.and
because it is quite difficult to project many of the variables identified
above, a more straight forward approach should be sufficient. The
approach set forth below would use the data provided in the BEA projections
to make the automobile emissions projections.
Since automobiles yield a transportation service over a period of
several years, and since we are interested in emissions projections for
the entire stock of automobiles, the most useful approach to developing
trend projections is to begin by projecting the demand for a stock of
automobiles and then to project the age distribution of that stock. As
shown in Figures 3 and 4 on both an absolute and per capita basis the
stock of automobiles has shown a fairly regular annual increase since 1930.
3*2
-------
u>
NATIONAL AUTOS
PER CAPITA PRO-
JECTION EQUATION
REGIONAL AUTOS
PER CAPITA PRO-
JECTION EQUATION
*
FUTURE REGIONAL
VALUES FOR
INDEPENDENT
VARIABLES
-
FUTURE NATIONAL
VALUES FOR
INDEPENDENT
VARIABLES
\
REGION/
JECTIOT
CAPITA
r
U. PRO-
IS OF
>ER
NATIONAL PRO-
JECTION OF
AUTOS PER
CAPITA
FUTURE
NATIONAL
VALUES FOR
POPULATION
1
NATIONAL PRO-
JECTIONS OF
THE STOCK OF
AUTOMOBILES
1
COMPARE M-
PREVIOUS PERIOD
NATIONAL STOCK
OF AUTOMOBILES
DISPLACE STOCK
ONE YEAR
SURVIVAL
PROBABILITIES
i.
SURVIVED NATIONAL
STOCK OF AUTO-
MOBILES
L
FUTURE REGIONAL
VALUES FOR
POPULATION
REGIONAL PRO-
JECTIONS OF THE
STOCK OF AUTO-
MOBILES
NEW NATIONAL
PURCHASES OF
AUTOMOBILES
TOTAL NATIONAL
STOCK OF AUTO-
MOBILES
PREVIOUS PERIOD
REGIONAL STOCK
OF AUTOMOBILES
I COMPARE I-
~^
DISPLACE STOCK
ONE YEAR
NEW REGIONAL
PURCHASES OF
AUTOMOBILES
SURVIVED
REGIONAL STOCK
OF AUTOMOBILES
TOTAL REGIONAL
STOCK OF
AUTOMOBILES
COUNTY POPU-
.LATION SHARE
COUNTY STOCK
OF AUTOMOBILES
AVERAGE
VEHICLE
MILES BY
AC! 0?
AUTOMOBILE
VEHICLE
MILES PRO-
JECTIONS
BY AGE OF
AUTOMOBILE
NATIONAL
URBAN/RURAL
TRAVEL
DISTRIBUTION
REGIONAL
URBAN/RURAL
TRAVEL
DISTRIBUTION
URBAN AND RURAL
VEHICLE MILES
PROJECTIONS BY
AGE OF AUTOMOBILE
COUNTY
URBAN/RURAL
TRAVEL
DISTRIBUTION
AQCR
SUMMATION
Figure 1. Tasks for Projection Regional Motor Vehicle Activity.
-------
I I I I I I I 1 1 I J I
SO 91 82 S3 S4 39 96 37 96 99 60 61 62 63 64 69 66 67 66 69 70
Figure 2. New Automobile Sales (Source: Automobile
Manufacturers Association, 1970-Automobile
Facts and Figures).
100
90
§ •
3.o
O 70
8 60
90
40
30
TOTAL STOCK OP
AUTOMOBILES '
90 92 94 96 96 60 62 64 66 66
rc»»
70
Figure 3. Stock of Automobiles (Source: Automobile Manufacturers
Association, 1970 Automobile Facts and Figures).
3-4
-------
.4000
.3*00
£ J»oo
3 .3100
S .3600
J .3500
9 J<<)0
S .5300
3
£ .5200
jj .3100
S JOOO
| JtOO
.MOO
.2700
.MOO
s
/
no C»HT» S
f
/
/
/
/
•_^S
S
^ — '
/r
/
/
/
/
f
1 i i i i i i i i i i i i i i i i i i i '
M SI U M S4 »5 M »T M W U
YtAR
U O (4 «9 *• «r U » TO
Figure 4. Stock of Automobiles Per Capita (Source:
Automobile Manufacturers Association,
1970 Automobile Facts and Figures).
Stock adjustment models have been applied to demand analysis for
consumer durables by several researchers, the most complete being a set
of 82 consumer demand models by Houthakker and Taylor. The behavioral
hypothesis underlying these models of consumer demand for durable
expenditures is that current purchases depend in part on the pre-existing
stock of the item in question. Current purchases are treated as an
attempt by consumers to adjust this stock toward some equilibrium level.
The rate of adjustment is usually a function of anticipated economic
conditions.
Consumers can be viewed as expected to possess an equilibrium
automobile stock based on their income, location, and various other
factors as represented by the following model:
SA - f1(Y,L,U) (2)
where:
SA = the expected equilibrium automobile stock
Y = the expected Income
L = the expected population density
U = the error term which is the combined effect of the omitted forces.
3-5
-------
Expressing equation (2) entirely on a per capita basis and assuming
linear relationships, with the variables expressed in first differences
to reduce the problems caused by serial correlation, we have the
following model specifications
CA
A ~
A - +
A L + U-
(3)
It is especially useful to present the stock of automobiles on a per
capita basis because such a formulation allows for the possibility of a
saturation in the demand for automobiles. For example, as shown in
Figure 5, on a family basis the percent of all families owning
automobiles has been stable at 80 percent over the last seven years.
The source of the increase in the per capita stock of automobiles has
apparently been the increase in family multi-car ownership.
100
90
80
70
60
»-
£ so
u
25 40
OL
30
20
10
-FAMILIES OWNING
AUTOMOBILES
FAMILIES OWNING
ONLY ONE AUTOMOBILE-
FAMILIES OWNING
TWO OR MORE
I
50
52
54
58 60
YEAR
62
64
66
66
70
Figure 5. Family Automobile Ownership. (Source: The
University of Michigan, Survey Research
Center, Survey of Consumer Finances)
3-6
-------
Because of differences in automobile emissions from model year to
model year, the age distribution of the stock of automobiles is
particularly important. This places special emphasis on (RA), the
number of automobiles retired. It is, therefore, necessary to project
(RA) for each model year.
A review of the literature did not yield any studies of automobile
retirement. One of the major reasons for retirement is probably the
cost of repair. As the ratio of the cost of replacement to the cost of
repair falls the rate of retirement would be expected to increase. For
newer cars, severe accidents are the primary reason for their retirement.
For older models, less severe accidents or repair costs due to mechanical
failures are important factors.
While a thorough study of the trends in accidents, their severity,
cost of repair, and replacement costs would be possible, it would appear
that a more cost-effective approach would be to assume (RA) to be a
stable function of vehicle age. Therefore:
DA = If CA
**lt i i(t-l) (i = 1,2,3...n model years) (4)
(t = time by year)
Figure 6 shows survival probability functions on both a cumulative
and an annual incremental retirement basis. While the function is usually
expressed on a cumulative basis, the annual incremental basis is more
useful for sensitivity analysis. For example, as shown in Figure 6, about
84 percent of the seven year old vehicles are expected to be registered in
the eighth year. However, it may be useful to examine the effect on
emissions if the percent rose to 95 percent due to the higher new car
prices expected with emissions control systems. The annual change
function provides the best method for allowing for the incorporation of
such possibilities.
Using the survival probability function it is possible to compute the
total number of cars retired in each year:
n
RA = I RA.. (5)
C 1=1 "
3-7
-------
The number of new automobiles (XA) purchased at time (t) can be obtained
by subtraction:
XA - SA - SA_ + RA (6)
E aeo
V
\
\
\
\
CUUULATNt '
MTflUMNT
\
I I I I
10 II II II 14
VtWCLt A«C -71*11$
Figure 6. Automobile Survival Probabilities.
(Source: Research Triangle institute)
3.2.2.2 Vehicle Miles Projections. A passenger car projection model
of the type described above provides projections of the stock of automobiles
by vehicle age. The next step is to convert these stock projections to
vehicle miles. The PAVE-I model has a procedure for calculating vehicle
miles. It uses a function which relates the average annual mileage (V,)
per vehicle by model year to vehicle age. Table 1 shows this, relationship
on A national average basis.
There is probably a significant variation in this functional
relationship from region to region. For example, motor vehicles operated
in the Great Plains States where there are.straight roads, with high speed
limits and low population densities probably have a higher average annual
mileage than do motor.••vehicles operated-in urban areas.
3r-8
-------
Table 1. ANNUAL MILEAGE AS A FUNCTION OF VEHICLE AGE
Car Age
(years)
1
2
3
4
5
6
7
8
9
10
11 or more
Average Annual Milage
Per Car
17,500
16,100
13,200
11,400
11,700
10,000
10,300
8,600
10,900
8,000
6,500
Source: U.S. Department of Transportation, Nationwide Personal
Transportation Study, Annual Miles of Automobile Travel,
Report No. 2, April 1972, p 9.
The regional differences may vary inversely with population
densities and if so, functional relationships may be able to be developed
to adjust the national relationship for each region based on its
projected population density.
3.2.2.3 Vehicle Speed Distributions. Motor vehicle emissions of
carbon monoxide and hydrocarbons vary inversely with vehicle speed. It
is desirable, therefore, to distribute the projected vehicle miles by
speed. Two speed regimes are typically used, urban and rural. Urban
travel is assumed to average 25 miles per hour, rural travel 45 miles
per hour.
Historical distributions of travel are available on a state basis.
From these distributions it may be possible to determine functional
relationships which could be applied on a regional basis. Population
density may be a good explanatory variable'.
Another possibility is to use speed distributions presented in the
state traffic forecasts each state must provide the Department of
Transportation.
3-9
-------
3.2.2.4 Commuting Patterns. The most difficult adjustment to the
vehicle miles projections will be to allow for commuting across regional
boundaries. Since the BEA regions have been drawn so as to minimize
intercounty commuting, the vehicle miles projections for these regions
probably do not require any adjustment for commuting; however, ideally
commuting should be taken into account when disaggregating the BEA
regional projections to counties and their reaggregation to AQCR's.
There appears to be no simple way to adjust the vehicle miles
projections for intercounty commuting. For some metropolitan areas,
empirical data on commuting patterns may exist which could be directly
imputed into the emissions projection model. For most areas of the
nation, however, such data will not exist in a useable form. It may be
possible to use the information on trip lengths provided in the Nationwide
Personal Transportation Survey, however, this would be a fairly tedious •
effort.
Because of the lack of a computationally simple method.for projecting
commuting patterns, it does not appear cost-effective at this time to
estimate ot project commuting patterns. Provision should be made,
however, for directly inputing any known commuting patterns for a county
into the projection model.
3.2.2.5 National-Regional Integration. The per capita stock of
automobiles, the total stock and the age distribution should first be
projected on a national basis. The next step would be to determine the
distribution of the total national stock among the regions.
Annual data on the distribution of the automobile stock are
available on a county basis. In order to account for the various factors
which would influence this distribution at a given point in time and changes
in these factors with time, it will be necessary to employ appropriate
statistical procedures for combining cross section and time series data.
The use of analysis of covariance techniques in the problem of
pooling cross section and time series data has no* become a common
practice in econometric work. Suppose we have data on N counties over
3-10
-------
I periods of time. The model usually followed in pooling procedures is:
yij = ai + °J \l^ *rij + Uij
(i = i,2 N;
j = 1,2,...,T),
where the a are the county "dummies", o. are the time "dummies", and x
the "covariates". An argument frequently made against the use of the
dummy variable technique is that it eliminates a major portion of the
variation among both the explained and explanatory variables if-the
between county and between time-period variation is large. In some
cases there may also be a loss in the number of degrees of freedom
available for estimating the sampling variance of the parameter estimates.
In addition, it is seldom possible to give a meaningful interpretation to
the dummy variables.
A second approach is to recognize possible correlations among the
error terms of the model and, in conjunction with restrictions, account
for these in order to increase the asymptotic efficiency of the estimates
of the causal parameters. A third approach to combining cross section
with time series data has been termed the components of error or variance
components model. With this approach, the regression error is assumed to
be composed of three independent components - one associated with time,
the second with the cross section units and the third being an overall
component variable both in time and cross sectional dimensions.
Wallace and Hussain present an analysis of the error components
model as an approach to combining cross section and time series data.
Comparisons of generalized least squares and iterative estimates with
those produced by analysis of covariance techniques are provided.
Maddala extends this approach by investigating some aspects of the
analysis of variance components models that arise from the use of
likelihood methods and the presence of lagged dependent variables as
covariates. However, for all their efforts., Wallace and Hussain
concluded that covariance estimates compare fairly well with the other
approaches that they investigated. Given that covariance estimates
3-11
-------
are easily obtained without the time-consuming iterative estimation
procedures necessary for some of the alternative approaches, it appears
that the covariance technique would be the most appropriate approach to
combining cross section and time series data in estimating the regional
air emissions model.
Using the covariance approach the regional per capita stock of
automobiles equations can be related to national conditions in order to
minimize any differences between the sum of the regional projections
and the national projection. A possible form of the functional
relationship is:
A £=
SA"
p
T".
ii _ « 4. g
81 + S2 A
^
V
p
Y
.P.
y
1
+ 6
AL
a
(7)
(i = 1,2,...,N regions;
j = t - 1,2,...,T time periods).
In this regression the relationship of a region's change in the
stock of automobiles to the change in the national average depends not
only on the covariates of change in the relationship of regional income
per capita and population density to the change in the national averages
but also on a variable a. which is peculiar to the ± region and on a
th
variable o. which is specific to the t year.
3.2.2.6 Ease of Model Revision. The model would be easy to revise.
The inputs would consist of exogenously estimated: automobiles per capita
functions, survival probability function, vehicle miles by vehicle age
function, and an urban/rural travel function. In addition, beginning
period estimates of the stock of automobiles by age would be a primary
Input as would the BEA income and population projections and county land
area data. Periodically, the.equations should be reformulated using the
latest data available.
3-12
-------
3.2.2.7 Use for Sensitivity Analysis. The model should have the
capability of shifting the values projected for any of the parameters on
a percentage basis in order to determine the sensitivity of the emissions
projections to the values of the model parameters. For example, it would
be desirable to be able to shift the projected values of the per capita
stock of automobiles a given percentage in a simple manner so that the
sensitivity of the emissions projections to the projection of the stock
of automobiles could be observed. Likewise, it would be desirable to be
able to arbitrarily shift the retirement, average vehicle miles, and
urban/rural travel functions in order to determine the sensitivity of the
emissions projections to these variables.
3.2.2.8 Data Sources for Predictive Variables. Stock of Auto-
mobiles - The R. L. Polk Co. of Detroit, Michigan maintains automobile
registration data on computer tape by county of registration. The data
are for July 1 but can be adjusted to January 1 by subtracting new
registrations.
Personal Income - The Regional Economics Division, Bureau of Economic
Analysis, U.S. Department of Commerce maintains an annual personal income
series on a county basis. The values are in current dollars and should be
deflated using the GNP Implicit Price Deflator for Personal Consumption
Expenditures. Future values for personal income are provided in the BEA
projections for 10 year intervals. Values for the intermediate years
can be obtained through interpolation.
Population - The Regional Economics Division, Bureau of Economic
Analysis, U.S. Department of Commerce has historical county population
estimates. Future values for population in the BEA regions are provided
in the BEA projections which are for 10 year intervals. Values for the
intermediate years can be obtained through interpolation.
Land Area - The City and County Data Book, U.S. Department of
Commerce contains a list of the land area of all counties.
Automobile Retirement Function - The retirement function can be
calculated from R. L. Polk data of automobile registrations by year model.
It would be desirable to examine the function for several time periods in
order to analyze its stability.
3-13
-------
Average Annual Vehicle Miles Per Automobile - The U.S. Department of
Transportation's Nationwide Personal Transportation Study has average
miles per vehicle by year-model of automobile. The national averages are
published in Report No. 2, April 1972, Annual Miles of Automobile Travel.
Primary data from the survey are on computer tape and could be used to
adjust the national average for regional variations.
3.2.3 Trucks
Trucks account for the greatest portion, after automobiles, of
registered motor vehicles, 17 percent. In 1969 there were 17.9 million
trucks registered of which 97 percent were gasoline-powered. The
remainder were diesel-powered or used other special fuels. Buses are
less than one percent of all motor vehicles.° About 18 percent of the
buses are diesel-powered. Because they are a small source of emissions
the most cost-effective approach may be to not include them in the
projection model. If desired, however, they could be combined with trucks
or treated in an analogous manner to trucks.
The objective of an emissions projection methodology for trucks is
to provide a procedure for projecting vehicle miles by gasoline-powered
trucks and fuel consumption by diesel-powered trucks since the emissions
factors for gasoline-powered trucks are related to vehicle miles while
emissions factors for diesel-powered trucks are related to fuel
consumption.
3.2.3.1 Truck Projection Model. The stock of trucks on both a
total and per capita basis has increased quite regularly over the last
20 years as shown in Figures 7 and 8. The stock of all trucks can be
projected in a manner similar to that used for automobiles. However,
since there is a clear differentiation in the area of operation of light
and heavy trucks it is necessary to project the truck size distribution.
As shown in Table 2, light trucks are primarily operated locally whereas
heavy trucks are most typically used for intercity transport.
3-14
-------
18
-» 17
I-
1 15
1 l4
K 13
S l2
-------
Table 2. PERCENTAGE DISTRIBUTION OF ANNUAL VEHICLE-MILES
OF TRUCK TRAVEL
United States - 1963
AREA OF OPERATION
Total
Local
2
Intermediate
3
Long distance
Not reported
LIGHT
54.5
66.8
27.8
5.9
71.3
PER CENT OF
MEDIUM
8.0
9.6
8.3
0.8
5.4
TRAVEL BY
HEAVY
31.3
18,6
55.2
82.5
18.4
TRUCK SIZE
MISCELLANEOUS
6.2
5.0
8.7
10.8
4.9
TOTAL
100.0
100.0
100.0
100.0
100.0
Urban, immediate environment; farm use.
2
Distances up to 200 miles.
3
Distances over 200 miles.
Source: Truck Inventory and Use Survey, Vol. II, Table 27. 1963
Census of Transportation, U.S. Bureau of the Census, 1965.
The stock of all trucks (ST) is expected to be related to the area
income, population density, employment (E) and an error term for the
omitted factors (V) as represented by the following model:
ST
- f2 (Y, L, E, V) (8)
The BEA projections include projections of earnings for trucking and
warehousing. This industry is primarily involved in furnishing local or
long distance trucking. The stock of registered heavy trucks (SHT) can
probably be related to earnings in trucking and warehousing (ETS) and an
error term for the omitted factors (W).
SHT - f3 GETS ,10 (9)
3-16
-------
Expressing equations (8) and (9) on a per capita basis and assuming
linear relationships, with the variables expressed in first differences
to reduce the problems caused by serial correlation, we have the
following model specification:
A^=b + bA+bAL + V (10)
A S|T = GI + c2 A |1 + w
The light and medium trucks can be obtained by subtraction:
SLT = ST - SHT (12)
The diesel share (SHTD) of all heavy trucks can be projected by
extrapolating time trends:
ir • di + d2 T * z
Where T = time in years
Z = the error term which is the combined effect of the omitted
factors
The age distribution for the stock of trucks can be projected in
the same manner as for automobiles. Figure 9 shows the annual
incremental and the cumulative retirement functions for all trucks. The
number of trucks retired each year by model year is :
•pni = i_ CT
it i i(t-l) (i = 1,2,3 ... n model years) (14)
The total number of trucks retired is :
n
RT = I RT (15)
C i=l 1C
The number of new trucks purchased at time (t) can be obtained by
subtraction.
XTfc = STt - ST x + RTfc ' (16)
3-17
-------
YtM-TO-YIUt
MTIKHCNT
cuMuumvt
ftCTHUMtNT
I I I I
Figure 9. Truck Survival Probabilities.
(Source: Research Triangle Institute)
3.2.3.2 Vehicle Miles Projections. The average annual vehicle
miles for trucks tend to decrease with increases in vehicle age and, for
a given age, to increase as the size of the truck increases. Table 3
shows this relationship on a national basis. State data are available
and can be used to adjust for regional variations.
3.2.3,3 Area of Operation. All trucks, other than heavy trucks,
can be reasonably assumed to operate locally. Heavy truck mileage can be
distributed on the basis of primary and urban road mileage in each region.
3.2.3.4 Vehicle Speed Distribution. Since there does not appear to
be a significant difference in automobile and truck speeds, the same
urban/rural speed distributions as determined for automobiles can be used
for distributing the truck miles.
3.2.3.5 Data Sources for Predictive Variables. Most of the data
sources applicable to the projections of automobile emissions also apply
to the truck projections. The annual vehicle miles per truck, however,
are available in the Census of Transportation.
3-18
-------
Table 3. ANNUAL TRUCK MILES AND SIZE CLASS AND
YEAR MODEL OF TRUCK: 1963
(Percent distribution of motor trucks)
Size class and year model
1960-61 models
1955-59 models
1950- 54 models
Pre-1950 models
1962-63 models
1960-61 models
1950-54 models
Pre-1950 models
1960-61 models
1955-59 models
Pre-1950 models
1962-63 models
1960-61 models
1955-59 models
1950-54 models
Total
Less
than
5.000
miles
5.000
to
9,999
miles
10.000
to
19,999
miles
20.000
to
29.999
miles
30,000
miles or
more
J"°L
reported
Distribution by truck miles
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
22.7
22.8
6.0
9.2
18.6
33.8
39.2
27.7
6.8
7.1
17.6
39.2
50.4
27.3
8.9
11.6
20.5
39.8
49.1
7.2
2.4
2.2
6.1
14.8
25.0
23.4
24.6
26.4
20.5
23.2
30.3
29.2
21.7
22.2
16.4
17.6
27.4
26.8
15.6
23.0
16.5
19.8
27.3
26.5
16.8
11.9
5.9
6.9
12.5
19.4
20.5
22.4
23.9
25.0
39.2
40.5
29.6
14.1
7.9
23.3
35.8
42.3
31.3
14.5
5.2
21.3
30.7
32.3
26.2
13.4
6.9
18.1
15.3
16.4
19.6
20.5
14.9
21.5
6.4
5.6
14.1
11.5
4:9
1.7
0.9
7.4
15.7
15.4
9.3
3.2
1.0
8.3
17.0
15.6
9.8
2.9
0.9
10.6
9.0
10.8
12.2
9.2
7.2
6.4
6.4
2.7
6.1
4.4
2.5
1.3
0.9
5.6
16.0
11.3
6.4
1.5
0.9
7.4
18.5
14.9
7.2
2.2
0.8
43.8
59.3
.58.6
43.0
23.0
11.0
10.3
16.0
17.5
14.1
11.2
14.1
19.9
29.4
13.8
9.3
6.3
8.0
14.8
26.9
12.7
8.4
' 5.3
9.0
15.2
25.5
8.4
8.1
5.1
6.6
13.1
21.4
15.5
Source: RTI
3.3 Diesel Powered Motor Vehicles
3.3.1 General
Most diesel powered motor vehicles are trucks. To a lesser extent,
however, buses are also diesel powered. It appears most cost-effective
to project emissions from diesel trucks since their number is about eight
times greater than that for buses. The methodology for projecting the
stock of diesel trucks was set forth above. Since almost all diesel
trucks are in the heavy size category the national projections of
emissions should be projected in the same manner as for heavy gasoline
trucks.
Relationships between the number of diesel trucks and the quantity
of diesel fuel consumed can be established from the data presented in
the American Petroleum Institute publication, Petroleum Facts and Figures.
3-19
-------
3.4 Aircraft
3.4.1 General
Aircraft are propelled by emissions emitting; gasoline-powered
reciprocating engines and Jet fuel-powered turbine engines. Aircraft
emissions vary by the type of aircraft and number of engines.
Aircraft emissions factors are based on a landing-takeoff cycle (LTO).
The LTO cycle used to develop emissions factors includes all the typical
pre-takeoff ground operations, takeoff, climb to 3,500 feet, approach from
3,500 feet, and touchdown. The objective of an emissions projection
methodology for aircraft is to provide a procedure for projecting, first
on a national basis and subsequently on a regional basis, the number of
LTO cycles by type of aircraft.
In 1969 there were 11,050 airports on record with the Federal
Aviation Administration (FAA), of which 38 percent (4,155) were publicly
9
owned. The remainder (6,895) were privately owned.. However, only a small
percentage of the public airports, those with FAA air traffic control
facilities, account for virtually all of, the ai# ca^rietr traffic and the
greatest majority of the general aviation traffic. In 1969 there were
328 such airports. By 1971 the number had risen to 346.
The trend in operations at the airports with FAA controlled towers
is upward, growing about 7.4 percent annually. An "operation" is a
landing, takeoff or missed approach and, therefore, must be divided by
two (2) to get the number of LTO cycles. All of the air carrier traffic
is itinerant, while about half the general aviation and military traffic
is itinerant, the remainder being local.
Airport activity is greatest in the nation's population centers.
The areas served by the air carriers have been divided Into air traffic
hubs (large, medium, small, and nonhub). Figure 10 shows the location
of the air traffic hubs. About 38 percent of all! itinerant_^eratlon8_jre __:
accounted for by the 22 large hubs, 20 percent by the 38 medium hubs,
22 percent by the 86 small hubs and 16 percent by the approximately 200
hon hubs with FAA traffic control towers. .In selecting an aircraft
emissions projection methodology it will be most cost-effective to
3-20
-------
V
ISJ
\ ___ •-. r-
—y j
Uf mtr^f
X / I
v L._r~-;'rr'i~: J •
'v ' i '—-••
....w. v ; i
^:-:^.rr- --" v/
1 „., y c.-»-V^
LEGEND
• LMCC HUK a
^
•- r—-~\.«. ^-^ — r
\ —/ i j
/ i -.-= \— m="if .->—*
• SUAU.KUIS «
Figure 10. Air Traffic Hiobs. (Source: Department of Transportation,
Federal Aviation Administration, Airport Activity Statistics)
-------
confine the projections to the 346 moat active airports. Within this
total it may well be desirable to have a less sophisticated procedure
for the smaller airports than for the larger airports.
3.4.2 Aircraft Activity Projection Model
• *»
There are three general types of aircraft activity: air carrier,
general aviation and military. The air carrier activity is for-hire air
transportation by trunk, regional or commuter airlines. General
aviation consists of business and pleasure aircraft.
The air carrier fleet consists of about 2,700 aircraft (1970) of
which 92 percent are turbine powered. Most of these are 2, 3 or 4
engine.turbojet aircraft.
The general aviation fleet consisted of about 131,000 aircraft in
1970, about 95 percent of which are reciprocating engine powered. Most
of these are single-engine (83 percent); the remainder are multi-engine,.
primarily twins. About 2 percent are turbine powered, almost all of
which have two engines. The remaining aircraft are helicopters or other
types.
Ten year national forecasts of itinerant and local aircraft
operations at airports with FAA Traffic Control Service are developed
annually by the Office of Aviation Economics, FAA. These forecasts will
provide good national control totals for any regional projections.
National projections beyond 1980 can be developed either by simply
extending the historical trends and ten year FAA projections, by using the
more sophisticated technique of relating the aircraft activity to
expected population and income growth or by simply using the forecasts
to be released by the Aviation Advisory Commission at the end of 1972.
Regional projections of aircraft activity are provided by the FAA
for approximately 1000 airports which meet at least one of the following
criteria:
- existing tower airport
- candidate for a tower
- 50 or more based aircraft
- receives certificated route air carrier service
- 20,000 or more general aviation itinerant operations.
3-22
-------
The distribution of operations by type of aircraft can be projected
for large airports by analyzing current air carrier service in terms of
the type of aircraft used and the routes flown. Future aircraft types
are forecasted by the FAA and can be allocated to the airport. An
example of such an approach is shown in the FAA publication Aviation
Demand and Airport Facility Requirement Forecasts for Large Air Trans-
portation Hubs Through 1980. For smaller airports served by air carriers
it can probably be reasonably assumed that all the aircraft are twin-
engined. The distribution can be based on FAA's projection of fleet
composition.
3-23
-------
REFERENCES
1. U. S. Environmental Protection Agency, Compilation of Air Pollutant
Emission Factors, Research Triangle Park, North Carolina, 1972,
p. 3-1.
2. Sauter, G. D. and W. R. Ott, A Program for Computer Calculation of
Current and Projected Vehicular Air Pollutant Emissions in Urban
Areas and Regional Air Basins. Unpublished paper
3. Automobile Manufacturers Association, 1970 Automobile Facts and
Figures^ Detroit, Michigan, p. 19.
4. Houthakker, H. S. and L. D. Taylor, Consumer Demand in the United
States; Analyses and Projections. Harvard University Press,
Cambridge, Massachusetts, 1970.
5. U. S. Environmental Protection Agency, Compilation of Air Pollutant
Emission Factors, Research Triangle Park, North Carolina, 1972,
p. 3-7.
6. Maddala, 6. S., 1971, "The Use of Variance Components Models in
Pooling Cross Section and Time Series Data." Econometrica 39 (2):
pp. 341-358.
7. Wallace, T. D. and Ashiq Hussain, 1969, "The Use of Error Components
Models in Combining Cross Sections with Time Series Data."
Econometrica 37 (1): pp. 55-72.
8. Automobile Manufacturers Association, 1970 Motor Truck Facts.
Detroit, Michigan, p. 13.
9. Department of Transportation, Federal Aviation Administration,
FAA Statistical Handbook of Aviation. 1970 Edition, Washington,
D.C., pp.46, 49.
10. Department of Transportation, Federal Aviation Administration,
FAA Air Traffic Activity. 197I.Washington, D. C. p. 24.
11. Department of Transportation, Federal Aviation Administration,
Terminal Area Forecast 1973-1983. December 1971, Washington, D. C.
3-24
-------
BIBLIOGRAPHY
Automobile Manufacturers Association, 1970 Automobile Facts and
Figures.
Automobile Manufacturers Association, 1970 Motor Truck Facts.
Automobile Manufacturers Association, Motor Trucks in the Metropolis,
by Wilbur Smith and Associates, New Haven, Connecticut, August 1969.
Aviation Week & Space Technology, Vol. 92, No. 25, (June 22, 1970),
pp. 1-262.
Aviation Week & Space Technology, Vol. 94, No. 10 (March 8, 1971),
pp. 1-190.
Civil Aeronautics Board. Measuring The Elasticities of Air Travel,
by Samuel L. Brown, pp. 278-285.
Environmental Protection Agency. The Potential Impact of Aircraft
Emissions, by M. Platt, R. C. Baker, and R. D. Siegel. (Report No.
1167-1) Cambridge, Mass.: Northern Research and Engineering
Corporation, Dec. 29, 1971, 318 pp.
Garrison, W. L., et al. A Prolegomenon To The Forecasting of
Transportation Development - Final Report, Report No. AD 621 514,
Northwestern University, Evanston, Illinois, August 1965, 123 pp.
Moore, James G. "Long Range Forecasting of Commercial Airline
Passengers," Business Economics. (September 1969), pp. 66-70.
National Air Pollution Control Administration. Nature and Control
of Aircraft Engine Exhaust Emissions, (Report No. 1134-1), Cambridge,
Massachusetts: Northern Research and Engineering Corporation, Nov.
1968, 388 pp.
Platt, Melvin and E. Karl Bastress. "The Impact of Air Craft
Emissions Upon Air Quality," International Conference on .
Transportation and the Environment.SAE, EPA, DOT Conference
Proceedings. Washington, D.C.: U.S. Government Printing Office,
May 31 - June 2, 1972, pp. 42-55.
U.S. Department of Commerce, Aeronautics Commission and Arthur D.
Little, Inc., Consultant, "Commercial and General Aviation,"
Transportation Predictive Procedures. Lansing: The State of
Michigan, Technical Report No. 9A, December 1966, pp. 1-30.
3-25
-------
U.S. Department of Commerce Office of The Under Secretary of
Commerce For Transportation. Demand For Inter-City Passenger
Travel In The Washington-Boston Corridor, by Systems Analysis and
Research Corporation, (Report No. PB 166 884), Boston, Mass.:
U.S. Government Printing Office, 1968. .
U.S. Department of Transportation, Federal Aviation Administration.
Airport Activity Statistics of Certificated Route Air Carriers.
Washington, D.C.: U.S. Government Printing Office, June 30, 1970,
304 pp.
U.S. Department of Transportation, Federal Aviation Administration.
FAA Air Traffic Activity. Calendar Year 1971. Washington, D.C.:
U.S. Government Printing Office, February 1972, 274 pp.
U.S. Department of Transportation, Federal Aviation Administration.
FAA Statistical Handbook of Aviation. Washington, D.C.: U.S.
Government Printing Office, 1967 Edition, 249 pp.
U.S. Department of Transportation, Federal Aviation Administration.
FAA Statistical Handbook of Aviation. Washington, D.C.: U.S.
Government Printing Office, 1970 Edition, 278 pp.
U.S. Department of Transportation, Federal Aviation Administration.
The National Aviation System Plan. Ten Year Plan 1971-1980.
Washington, D.C.: U.S. Government Printing Office, March 1970, 147 pp.
U.S. Department of Transportation, Federal Aviation Administration.
The National Aviation System Plan, Ten Year Plan 1972-1981,
Washington, D.C.: U.S. Government Printing Office, March 1971, 217 pp.
U.S. Department of Transportation, Federal Aviation Administration.
Bureau of Public Roads, Airports Service. Aviation Demand And
Airport Facility Requirement Forecasts For Large Air Transportation
Hubs Through 1980, Washington, D.C.: U.S. Government Printing Office,
August 1967, 130 pp.
U.S. Department of Transportation, Federal Aviation Administration,
Office of Aviation Economics, Aviation Forecast Division. Aviation
Forecasts Fiscal Years 1971-1982. Washington, D.C.: Reproduced by
National Technical Information Service, Springfield, Va., January 1971,
49 pp.
3-26
-------
SECTION 4
PROJECTING EMISSIONS FROM AREA SOURCES
AND FOSSIL-FUELED ELECTRIC GENERATING PLANTS
4.1 Introduction
The term "area sources" refers to those sources of emissions that
are widespread and relatively evenly distributed throughout a region in
such a way that the pollution they produce forms a ubiquitous pattern.
This pollution results from fuel combustion for residential, commercial,
institutional, and industrial heating. The pollutants involved are
primarily hydrocarbons resulting from less than perfect combustion of
the fuel, sulfur oxides resulting from the oxidation of the sulfur in
the fuel, oxides of carbon and nitrogen formed in combustion, and
particulates released with the exhaust gases. The quantities of these
pollutants formed depend on the quantity of fuel burned, its ash and
sulfur content, and the type of burner employed.
The problems associated with projecting emissions from steam-electric
plants is discussed in this section because, although they are point
sources, they have many of the characteristics of other stationary fuel
combustion sources. Discussion of the appropriate methodology follows
the analysis of area sources.
There are two general approaches that may be taken to projecting
area source emissions. One is to estimate fuel consumption for each
type of source and the other is to identify the number and type of burners
in use and the fuel consumed in them. The second approach provides the
most accurate estimates of emissions, but involves more complex and
3/
detailed calculations. A study— performed for EPA by the Walden Research
Corporation in 1971 followed this method and produced results apparently
superior to previous studies. However, continuous use of this method
requires use of data that are not readily available. The fuel consumption
approach, on the other hand, uses data that are quite easily obtained
from regularly published sources. Estimates based on projected fuel
consumption may also be adjusted to reflect the influence of fuels
availability and prices, and may be improved by incorporating much of
the data on burner use patterns developed in the Walden study.
3/ John R. Ehrenfeld, et al., Systematic Study of Air Pollution From
Intermediate-Size Fossil-Fuel Combustion Equipment. Contract No.
CPA 22-69-85. Cambridge, Mass., Walden Research Corporation, July 1971.
4-1
-------
It is the recommendation of this report, therefore, that the projection
of area source emissions he based on fuel consumption, patterns. The
remainder of this section is devoted to a discussion of the model to be
used for this purpose. The problems involved in construction of this
portion of the overall model are less complex than those related to
mobile or point sources and this discussion is, therefore, less extensive
than is provided in Sections III and V.
4.2 Estimation of Fuel Consumption
Estimates will be needed of the consumption, by county, of coal,
distillate oil, residual oil, and natural gas. Consumption of these
fuels by State is reported annually by the Bureau of Mines in its
Minerals Yearbook. The figures for fuel oil are also published by the
Journal Fuel Oil and Oil Heat each year, along with analyses of new
and replacement burner installations and other information indicating
heating trends. The Edison Electric Institute provides annual data by
State on electric heating and several trade organizations in the gas
field also have annual data series that are applicable. Although the
classifications under which these sources report their data are not
fully compatible, it is possible to construct quite accurate estimates
of actual consumption of each fuel by residential, commercial (including
institutional), and industrial users for recent years. These series
permit analysis of the time series trends of fuel substitutions. RTI
has used these data over the past four years as the basis for the
estimates furnished to EPA for inclusion in the Economics of Clean Air
reports.
Future demand for these fuels for heating purposes ten and twenty
years from now will be determined by the growth of population and commercial
and industrial activity and by the general trend of energy use for heating.
Construction of these equations should present no conceptual difficulties.
4.3 Residential Heating
Residential heating plants may be divided into two general categories:
those in buildings housing one to four families, which use a relatively
small furnace or space heaters, and larger structures with large central
heating plants that are similar in burner characteristics and emissions
to heating units for commercial, institutional, and small-to-medium
sized industrial heat and power boilers. For purposes of this analysis,
4-2
-------
the discussion of residential heating will be limited to structures
housing four or less dwelling units. Larger residential units are
combined with commercial and institutional units.
The fuels used for residential heating today are coal, distillate
oil, and natural gas. The use of coal for this purpose has declined
very sharply in recent years and is negligible in most areas. Within
10 years no significant quantity of coal will be used for residential
heating except in those few locations where it is exceptionally
economical, such as parts of Pennsylvania, West Virginia, Kentucky,
Ohio, Illinois, and perhaps Minnesota and Montana. The specific
counties in which coal may continue to be used will have to be identified
for regional analysis. For the general model, however, coal can be
considered negligible.
Number 1 and 2 distillate oil are primarily home heating fuels and
are consumed in most counties. Natural gas is also important, but its
use is restricted in some areas where pipeline distribution is not
available. A substantial and growing percentage of home heating is
also done with electricity and allowance must be made for this in
estimating residential fuel use on the basis of population growth.
Residential heating requirements can be estimated on the basis
of population and the average number of persons per family, which
indicates occupied dwelling units. The average structure size (number
of rooms per dwelling unit and units per structure) combined with the
annual heating degree days for each region can be used to estimate
heating requirements in Btu's. This in turn can be divided into shares
for each fuel on the basis of market share trends adjusted for fuel
availability, and expressed in units of each fuel consumed. When
multiplied by appropriate emission factors a reasonable estimate of
annual emissions from residential heating may be obtained. Among the
variables involved, only population will be likely to change substantially
over the projection period.
4.4 Commercial, Institutional, and Industrial Heating
An approach similar to that used for residential sources can be
used in projecting emissions from the other heating sources, except
4-3
-------
that the basis would be employment rather than population. It may
reasonably be assumed that the required heat input per commercial
structure is proportional to the number of persons employed therein,
and that the same is true for institutional establishments. These ratios
can be calculated by region and projections of heating requirements in
Btu's based on projected commercial and institutional employment. The
fuel demand thus projected can be apportioned to fuel types in the same
manner as residential fuel requirements.
In a similar manner the ratio of required heat per year per employee
can be calculated for industrial plants by two digit SIC classifications
and projected fuel consumption projected by industrial employment. In
this instance, however, adjustments will have to be made by classifying
industrial users into broad size ranges to reflect the emission
characteristics of boilers of different sizes and further adjustments
may have to be made to reflect trends in use of boilers of various types.
Although these adjustments require considerable computations, they do
not appear to interfere with the estimation methodology.
It should be noted that the method described above assumes that
variations in building construction characteristics, heat loss through
loading doors and other areas, the use of process heat for space
heating, and other similar variations from plant to plant are success-
fully averaged out so that the resultant emissions estimates have
acceptable error limits. This appears to be an acceptable assumption
at this time.
4.5 Fossil Fueled Electric Generating Plants
The problems of projecting emissions from fossil fueled electric
plants is given relatively little space in this report not because they
are unimportant, but because, in comparison with the other sources
covered, they present few difficulties. These sources are major
sources of particulates and SO emissions especially, and of other
A
pollutants as well. This has been so well documented elsewhere as
to need no elaboration here.
A very large and detailed body of data is available from the reports
of the Bureau of Mines and the Federal Power Commission providing
4-4
-------
information, by plant, on operation characteristics, structure, fuel
consumed, and control of emissions. Similarly, detailed .data on planned
construction are available for 5 to 10 years into the future and some
information on potential plant locations is available for more distant
future dates.
Detailed analyses have also been made of the future demand for
electrical energy and the probable patterns of production by many
public and private authorities. Much attention is being given also ,
to the problems of fuels availability and the effects this may have
on future patterns of electricity production.
These data, when incorporated with the projections of fuel use
for heating, will provide a firm basis for projections of emissions
from electric generating plants.
4-5
-------
SECTION,5
PROJECTING POINT SOURCES OF AIR EMISSIONS
ON A REGIONAL BASIS
5.1 Introduction
The objective of this section is to define techniques and models
for projecting point sources of air emissions on a regional basis.
More specifically, the feasibility of this task is examined in terms
of the location of emissions, total output and process type for each
industry, and the relative effort required to make the projections .
The significance of each point source relative to the background level
of emissions for the average region is also examined. Specification
of the data requirements and the major variables determining the regional
projections of point sources of emissions are also provided.
5.2 Significance of Point Sources of Emissions
The point sources of emissions examined herein generate substantial
quantities of pollutants as a result of the industrial processes involved.
Some of these sources are substantially larger than others. The various
pollutants from each source have been classified according to their
relative significance, in order to provide the Environmental Protection
Agency with one criterion for establishing priorities for projection
modeling. Data availability, the difficulty of projecting output,
location, and process type are other criteria that will be treated in
following sections.
The point sources of air pollution emissions considered in this
study are those that have been identified as being important contribu-
tors of one or more pollutants. Because they vary considerably in
terms of their contribution to aggregate pollution, however, it is
desirable to provide a measure of significance that can be used in
determining priorities for phased development of projection models.
Assignment of priority to some sources to be included in the early
phases of the model may be based also on the availability of data and the
relative effort required to project output, location, or process type
for each industry.
5-1
-------
The measure of significance employed in this study is the ratio
of emissions from an industrial point source within an average metropolitan
AQCR to the total emissions from non-industrial sources, expressed as a
percentage. The calculations of these percentages, shown in Tables 4
i\ I -
through 7, are based on the analyses provided in 1970 by RTI— of
emissions by source for 298 metropolitan regions. These data were
used because they provide the only available analyses of a large number
of regions. The emission estimates reflect emission factors that are
different from those now considered most accurate, but are nevertheless
sufficiently accurate to reflect the relative significance of the
sources measured.
The calculations of significance for each industrial source were
made by finding the average emissions by pollutant for the plants covered
in the 1970 data and dividing those figures by the average background
(non-industrial) emissions for the regions involved.
There will also be non-metropolitan regions in the total regional
projection model so that every county in the United States will be
included in an air quality control'region. Sufficient information is
not available at the present time to determine emissions for the average
non-metropolitan area. The background emissions for such an area would
have the effect of making any point source in these regions more signi-
ficant than it would be in a metropolitan region. The ambient air quality
would probably still be superior in the non-metropolitan region, however.
The significance measure is at best a guide for setting priorities.
There are weaknesses in the technique that limit its application to more
complex tasks. For instance, there may be more than one establishment of a
particular source type in an area and the emissions are additive.
Serious polluters such as iron and steel plants and refineries tend to
be clustered in particular regions. An establishment may be substantially
larger than the average for that industry or it may be less controlled
than most establishments in the industry. On the other hand, the region
may be smaller than the average metropolitan area and this factor would
make the point source more significant.
4_/ For a list of the 298 metropolitan areas and the emission estimates,
see D. LeSourd,e£ al., "Comprehensive Study of Specified Pollution Sources
to Assess the Economic Effects of Air Quality Standards, FR-OU-534, Vol. I,
Research Triangle Institute, Research Triangle Park, December 1970.
5-2
-------
Table 4 contains the emissions by type from major sources for 298
metropolitan areas in 1967. Stationary fuel combustion includes steam
electric power plants as well as heating plants and boilers. Most
particulates originate from stationary fuel combustion and industrial
process sources. Stationary fuel combustion also accounts for approxi-
mately two-thirds of sulfur oxide emissions and industrial process sources
for about one-third. Habile sources generate about 90 percent of hydro-
carbon and carbon monoxide emissions. Petroleum refining and storage
produces most of the hydrocarbons and carbon monoxide emissions from
industrial point sources.
The second half of Table 4 consists of emissions for the average
metropolitan area by type and source. These were calculated by dividing
total emissions from all metropolitan areas by 298. The critical row in
this table is total emissions from all sources excluding industrial
processes. Emissions from the various point sources were divided by
the entries in this row in order to calculate their significance.
Table 5 is a list of total emissions from each of the industrial
process sources for 1967 in the 298 metropolitan areas. Particulates
arise in many different industries, but the largest quantities come from
the grain and feed industry, from the iron and steel industry, from
the kraft pulp industry, and from asphalt batching plants. Each of
the other types of emissions are found in a relatively small number
of industries.
Table 6 lists the emissions from an average establishment for each
point source. These were calculated by dividing the total emissions from
the point sources in the metropolitan areas (Table 5) by the number of
establishments in the metropolitan areas. Some establishments are much
i
larger than the average and others are much smaller, so the averages
shown may be distorted. The largest point sources of particulates
on an average establishment basis are the iron and steel industry and
the pulp and paper industry followed by cement plants and lime plants.
The primary metals industries, petroleum refineries, and sulfuric acid
plants are the major source of sulfur oxide emissions. The only point
source of carbon monoxide is petroleum refining, which is also the
major point source of hydrocarbons.
5-3
-------
Table 4
EMISSIONS FROM ALL SOURCES FOR 298 METROPOLITAN AREAS
AND FOR THE AVERAGE METROPOLITAN AREA
(Thousands of Tons Per Year)
Total for 298
Metropolitan Areas
Solid Waste Disposal
Stationary Fuel —
Combustion
TOTAL
Industrial Process
TOTAL
21
Mobile Sources —
TOTAL
TOTAL EXCLUDING
INDUSTRIAL PROCESS
Average Metropolitan
Area
Solid Waste Disposal
Stationary Fuel
Combustion
TOTAL
Industrial Process
TOTAL
Mobile Sources
TOTAL
TOTAL EXCLUDING
INDUSTRIAL PROCESS
Particulates
1,110
3,247
4,357
4,601
8,958
330
9,288
4,687
3.7
10.9
14.6
15.4
30.1
1.1
31.2
15.7
SO
X
11,416
11,416
5,156
16,572
'
16,572
11,416
_
38.3
38.3
17.3
55.6
- ' •
55.6
38.3
CO
3,770
-
3,770
7,520
11,290
126,000 .
137,290
129,770
12.7.
-.
12.7
25.2
37.9
422.8
460.7
435.5
HC
1,400
-
1,400
1,412
2,812
21,100
23,912
22,500
4.7
-
4.7
4.7
9.4
70.8
80.2
75.5
— Includes commercial-institutional heating plants, industrial boilers, residential
heating plants, and steam-electric power plants.
2/
— Nitrogen oxides are also emitted from mobile sources.
Source: D. LeSourd et al., Comprehensive Study of Specified Pollution Sources.
5-4
-------
Table 5
INDUSTRIAL PROCESS SOURCES - ESTIMATES OF EMISSION LEVELS, 1967
(298 Metropolitan Areas)
Primary Metals
Iron and Steel
Copper
Lead
Zinc
Aluminum
Petroleum Refining
Secondary Nonferrous
'Metals
Pulp and Paper
Chemicals
Phosphate Fertilizer
Sulfuric Acid
Rubber Tires
Coal Cleaning
Feed and Grain
Mineral Products
Asphalt Batching
Cement
Brick Making
Lime
Gasoline Marketing
and Bulk Storage
Quantity of Emissions (Thousands of Tons per year —
Particulates
1,100.0
6.0
80.0
9.8
561.0
63.6
1.2
64.7
1,674.0
452.0
239.0
181.0
-
SO
X
2,140
200
416
1,750
-
650
-
-
-
CO
5,300
-
-
-
-
-
HC
810
-
2/
n.a. —
-
-
600
Source: D. LeSourd, et al., Comprehensive Study of Specified Pollution Sources.
I/
Emissions abbreviated are: particulates (Part.), sulfur oxides (SOX), carbon
monoxide (CO)^ hydrocarbons (HC). Blanks in the table indicate the emission levels
meet the applicable regulation or that emissions are negligible or do not exist.
2/
— Not available.
5-5
-------
Table 6
AVERAGE EMISSIONS FROM EACH POINT SOURCE, 1967
(Thousands of Tons Per Year)
I/
Primary Metals
Iron and Steel
Copper
Lead
Zinc
Aluminum
Petroleum Refining
Secondary Nonferrous
Metals
Pulp and Paper
Chemicals
Phosphate Fertilizer —
Sulfuric Acid
Rubber Tires
Coal Cleaning
Feed and Grain —
Mineral Products
Asphalt Batching
Cement
Brick Making
Lime
Gasoline Marketing
and Bulk Storage
Number of Es-
tablishments
134
10
4
9
14
199
583
81
155
180
54
256
6,253
1,064
138
301
113
14,998
Particu-
lates
8.20
-
-
-
0.40
0.40
0.02
6.90
0.02
0.40
0.02
0.30
0.30
0.40
1.70
"
1.60
-
SO
X
214.0
50.0
46.2
-
8.8
-
-
-
3.6
. -
-
-
-
-
-.
-
-
CO
-
-
-
-
26.6
-
-
—
-
-
-
-
-
-
. '-
-
-
HC
.
-
-
-
-
4.1
- .
-
-
. -
-
-
-
-
<-
-
-
— .
— Elemental phosphorus and phosphate fertilizer emissions have been combined.
21
— Handling and milling.
Source: RTI
5-6
-------
Table 7 contains the percentages that emissions from average point
sources are of background emissions for the average metropolitan area.
Table 7 indicates the importance of a particular point source if there
is one and only one average size establishment present in the average
size metropolitan area. If any of these conditions do not hold, the
significance will be larger or smaller. Given these limitations, the
analysis shows that there are a relatively small number of point sources
that are highly significant as a percentage of background emissions
for the average metropolitan area. Particulates from the typical iron
and steel establishment and from the typical pulp and paper plant are
both more than 40 percent of the average metropolitan area's background
emissions of particulates. The average cement plant and the average
lime plant each produce about ten percent of total emissions for the
average metropolitan area. None of the other point sources of emissions
are more than 2.5 percent of the average metropolitan area. Sulfur oxide
emissions are very significant for copper, lead and zinc and relatively
significant for petroleum refineries and sulfuric acid plants. The
average petroleum refining establishment also generates about five
percent of the total emissions of carbon monoxide and hydrocarbons
for an average metropolitan area.
5.3 Projection Requirements
5.3.1 Expected Changes in Point Sources of Air Emissions
Some critical factors in projecting regional air emissions
from point sources are the location of the point sources, total production
for each source, and the production process used. Table 7 is a list of
the point sources together with expected changes in the location, production
and process types in each of these industries. The "no" column is checked
when virtually no change is expected in any of the three aspects. The
substantial column under "yes" is checked when substantial changes are
expected or when the production process is likely to be a critical factor
in projecting regional air emissions.
The first important factor in projecting regional air emissions is
location. Many of the industries are resource based; that is, their
location is dependent upon a source of raw materials. The location of
some of the other industries are proportional to the distribution of the
5-7 *
-------
Table 7
EMISSIONS FROM AVERAGE POINT SOURCE AS A PERCENTAGE
OF BACKGROUND EMISSIONS FOR THE AVERAGE METROPOLITAN AREASl/
(Percent)
Partic.
52.2
-
• -
-
2.5
2.5
0.1
43.9
0.1
2.5
0.1
1.9
1.9
2.5
10.8
10.2
-
SO
X
558.7
130.5
120.6
•
23.0
-
-
-
9.4
-
-
-
—
_
-
-
CO
.
-
-
-
-
6.1
-
-
-
-
-
-
—
- . •
-
-
-
I1C
..
.
-
-
-
5.4
-
-
-
-
-
-
-
-
-
-
0.05
Primary Metals
Iron and Steel
Copper
Lead
Zinc
Aluminum
Petroleum Refining
Secondary Nonferrous
Metals
Pulp and Paper
Chemicals
Phosphate Fertilizer
Sulfuric Acid
Rubber Tires
Coal Cleaning
Feed and Grain
Mineral Products
Asphalt Batching
Cement
Brick Making
Lime
Gasoline Marketing
and Bulk Storage
JL/ Background emissions include stationary fuel combustion and mobile
sources.
Source: Tables 4 and 6.
5-8 '
-------
Table 8
EXPECTED SIGNIFICANT CHANGKS IN THE LOCATION, PRODUCTION', AIID
PROCESS OF EACH POINT SOURCE, 1967-1980
Primary Metals
Iron and Steel
Copper
Lend
Zinc
Aluminum
Petroleum Refining
Secondary Konferrous
Metals
Pulp and Paper
Chemicals
Phosphate Fertilizer
. Sulfuric Acid
Rubber Tires
Coal Cleaning
Feed and Grain
Mineral Products
Asphalt Batching
Cement
Brick Making
Lime
Gasoline Marketing
and Bulk Storage
Location .
No
X
X
X
Yes
Minor
X
X
X
X
X
X
X
X .
. X
X
X
X
X
Subst.
X
X
Production
No
X
Yes
Minor
X
X
X
X
X
X
X
X
X
X
X
X
Subst.
X
X
X
•
X
X
Process
HO
X
X
X
X
X
X
*T
i c:e>
Minor
X
X
X
X
X
X
Subst.
X
X
X
X
X
X
5-9
-------
population or of general economic activity. The only industry among the
point sources that is not either resource based or determine by the
distribution of population is rubber tires. The substantial column is •
also checked for phosphate fertilizers, sulfuric acid and gasoline
marketing and bulk storage. Phosphate fertilizer is checked because
of the difficulty in distinguishing between plants that produce elemental
phosphorus and those that manufacture only phosphate fertilizers.
Sulfuric acid is checked because the location is likely to.change in
the near future due to new sources of sulfur arising from the control
of air emissions. Gasoline marketing and bulk storage is checked because
very little is known about the location and level of production of these
establishments. Copper, lead and zinc are unlikely to change their
location. All the other industries are subject to small changes in
the location of establishments or in the relative shares of national
output.
The second factor is the level of total production in the nation.
National production may change substantially in the aluminum, iron and
steel, secondary non-ferrous metal, and the lime industries. The pro*
duction and consumption of aluminum is dependent upon a number of factors,
especially its competitive position relative to steel and copper in many .
uses. Domestic production of iron and steel will be greatly affected by
the size and composition of imports. The secondary non-ferrous metal
industry is intricately dependent upon production of primary metals and
upon the amount of scrap that is generated and collected. The lime . •
industry has been growing rapidly due to changes in the type of furnace
used in the steel industry, because of activities in the paper industry,
and because of new uses for lime. Very little is known about the level
. of'production of gasoline marketing by specific location.
The third factor in projecting point sources is the production
process used. If more than one process is used, then the composition
of output from these processes must be known, since emissions can vary
significantly from one process to another. Production process consi-
derations are very critical in iron and steel, aluminum, phosphate
fertilizer, coal cleaning, brick making and lime. Other required process
5-10
-------
information is the presence or absence of a catalytic cracker in petro-
leum refining, the presence or absence of a lime plant or generating
equipment in the pulp and paper industry, the percentage of P?0 i-n
the finished product for each phosphate fertilizer establishment, the
type of operation in the feed and grain establishments, and the age
of kilns in the cement industry. An important element in projecting
emissions from gasoline storage is whether or not the tanks have floating
roofs, though this factor can also be considered a control measure.
5.3.2 Additional Data Needs
Data are required on the location, the production level and
the process used in each establishment. The type and efficiency of
controls applied at each establishment are also necessary. If the
Environmental Protection Agency does not provide information on
controls, then these data will need to be collected for almost all
industries (see Table 9).
Additional data from that already available will be needed on the
location and production levels of phosphate fertilizer plants, feed
and grain elevators and mills, coal cleaning plants, asphalt batching
plants, lime plants, and brick making establishments. A large amount
of detailed information will also have to be collected on gasoline
marketing and bulk storage. This information may have to be collected
directly from the companies involved or it may be available in publica-
tions that have not come to the attention of RTI.
Additional information on process types is required for the aluminum,
petroleum refining, pulp and paper, phosphate fertilizer, coal cleaning,
brick making, and lime industries. General industry averages will not
be useful in making regional projections of air emissions because the
establishments in a region may use one process or another but not the
industry average.
5.3.3 Relative Efforts in Making Projections of Point Sources
The relative effort required in projecting four different
aspects of point sources have been listed in Table 10, together with a
column that combines all four of them into an average. This last column
is not a simple average, but a judgmental weighting of the four aspects
that comprise it. The four aspects are the number of establishments,
their location, total production, and the production process.
5-11
-------
Table 9
ADDITIONAL DATA NEEDS FOR PROJECT POINT SOURCES OF AIR
p.. _•-,...,... ><_«-„i „
1 1 -t.*»i_* A. j . .t~ i. k~ J.w
Iron and 'Steel
Copper
Lead
Zinc
Aluminum
Petroleum Refining
Secondary Nonferrous
Metals
Pulp and Paper
Chemicals
Phosphate Fertilizer
Sulfuric Acid
Rubber Tires
Coal Cleaning
Feed and Grain
Mineral Products
Asphalt Batching
Cement
Brick Making
Lime
Gasoline Marketing
and Bulk Storage
I.oc;:L ion —
X
X
X
X
X
X
X
X
Proc-'G.s
X
X
X
X
X
X
X
Coni.ro!
-------
Table 10
RELATIVE EFFORT IN MAKING PROJECTIONS OF
SELECTED ASPECTS OF POINT SOURCES
Industry Effort-
Primary Metals
Iron and Steel
Copper
Lead
Zinc
Aluminum
Petroleum Refining
Secondary Nonferrous
Metals
Pulp and Paper
Chemicals
Phosphate Fertilizer
Sulfuric Acid
Rubber Tires
Coal Cleaning
Feed and Grain
Mineral Products
Asphalt Batching
Cement
Brick Making
Lime
Gasoline Marketing
and Bulk Storage
rt
Number—'
1
X
X
X
X
X
X
2
X
X
X
X
X
X
n
X
X
X
X
X
X
Locatior
1
X
X
X
X
o
z.
X
X
X
X
X
X
X
3
X
X
X
X
X
X
X
Production
1
X
X
X
X
X
X
2
X
X
X
X
X
X
X
X
X
3
X
X
X
X
j
Process 1 Combined
1
X
X
X
X
X
X
2
X
X
X
X
X
X
3
X
X
X
X
X
1
X
X
X
X
2
X
X
X
3
X
X
X
X
X
X
X
X
X
X
X
21
Relative effort is defined as: 1 = easy, 2 = moderate, and 3 = difficult.
Defined as: 1 = less than 100, 2 = 100-200, and 3 = more than 200.
5-13
-------
The effort required in making projections of point sources of air
emissions is probably related directly to the number of establishments.
Those industries in which the number of establishments is less than 100
have been rated at the number one level of effort, those industries with
100 or more but less than 200 have been rated as number two, and those
industries with more than 200 establishments have been rated as number
three. Only six industries-have more than 200 establishments.
The effort required in projecting the location of the establishments
is a function of expected changes in location as well as the number of
establishments involved. The location of point sources of emission will
require a three level qf effort for seven industries, a two level of
effort for seven industries, and a one level of effort for four industries.
The one level of effort indicates that no new establishments are expected
or that a complete list of establishments is available and is not likely
to change very much. A three level indicates that quite a detailed
analysis will be necessary in order to come up with a reliable projection
of the location of establishments and production at each one. The two
level requires an effort somewhere between one and two.
The effort of projecting the level of production also varies among
industries. Three level of effort will be required in the iron and
steel, aluminum, secondary nonferrous metal, and the lime industries.
The remaining industries are about evenly divided between .levels of
effort one and two.
The fourth aspect in making projections of point sources is the
production process used. There are only five industries in which a three
level of effort will be required. The production process is not important
in the copper, lead, zinc, secondary nonferrous metal, rubber tire, feed
and grain, and asphalt batching industries. All the other industries
fall into the number two level of effort.
The four aspects involved in making projections of point sources
have been combined into a single figure based on judgment. The reason
that a simple weighted average cannot be used is that a level two effort
in one aspect may not be equivalent in terms of man-months to level two
efforts in another aspect. Therefore, this final number is based upon
RTl's best judgment as to the degree of effort required. Projections
5-14
-------
can be easily made for the copper, lead, zinc and rubber tire industries,
rated at the number one level of effort. Number two level of effort
will be required for the phosphate fertilizers, cement, and brick making
industries. The number three level of effort will be required for the
remaining industries. These ratings do not imply the relative amount
of time that should be spent on making projections for each industry,
but are a first approach to the relative amount of time required to
make equally accurate and valid projections for each of the point sources.
The next section will discuss all the factors that should be considered
in deciding the amount of effort to be allocated to each point source.
5.3.4 Decision Criteria
Table 11 consolidates the information contained in the preceding
tables. Relative significance appears in the first section. A check mark in
a column indicates that the emission of a particular pollutant from the
average establishment for that industry is greater than a certain percent-
age of the background emissions of that type pollutant for the average size
metropolitan area. The levels of significance are one percent, five per-
cent, ten percent, and 20 percent. All the figures in the one percent
column are clustered between 1.9 and 2.5 percent. There are only 9
point sources in which the level of significance is greater than five per-
cent for a particular pollutant.
The second section is data needs, divided into a low degree and a
high degree of new data. There are only four industries in the low category,
but three of them are at the five precent level of significance (copper,
lead and zinc). High data needs can be interpreted as those not readily
available from published sources (see Table 9 for details). The greatest
need for information appears to be in the phosphate fertilizer, feed
and grain and the sulfuric acid industries. The six industries of iron
and steel, aluminum, petroleum, pulp and paper, coal cleaning and lime
industries will require an intermediate level of additional data. Signi-
ficant data requirements also exist for asphalt batching and gasoline
marketing and bulk storage.
The third part of Table 11 is the section on relative effort. In
general, the combined figure corresponds very closely to the average
of the four aspects of the relative effort.
5-15
-------
Table 11
Decision Table
Ul
i
Significance
>1%
X
X
X
X
X
X
>5%
X
•
X
>10%
X
X
>20%
X
X
X
X
X
X
Data Needs
Low
X
X
X
X
X
*
High
X
X
X
X
X
X
X
X
X
X
X
Relative Effort
No.
2
1
1
1
1
2
3
1
2
2
1
3
3
3
2
3
2
3
Loc.
.3
1
1
1
1
2
2
2
2
3
2
3
3
3
2
2
3
1
Prod.
3
1
1
1
3
2
3
2
2
2
1
. 2
2
2
O
1
3
3
Proc.
3
1
1
1
3
2
1
2
3
2
1
3
1
1
2
2
3
2
Co TC .
3
1
1
1
3
3
3
3
2
3
1
3
3
3
2
2
3
3
Reliability of
Emission Factors
Yes
X
X
X
X
X
No
X
X
X
X
X
Primary Metals
Iron and Steel
Copper
Lead
Zinc
Aluminum
Petroleum Refining
Secondary Nonferrous Metals
Pulp and Paper
Chemicals
Phosphate Fertilizer
Sulfuric Acid
Rubber Tires
Coal Cleaning
Feed and Grain
Mineral Products
Asphalt Batching
Cement
Brick Making
Lime
Gasoline Marketing and Bulk
Storage
— Defined as emissions from the average point source as a percentage of total emissions for the average
metropolitan area for 1967 emissions. There is one check for each type of pollutant emitted.
-------
The fourth part of the table is the reliability of the emission
factor. Because of differences in production levels, process types and
controls, the emission factors for some industries contain a wide
margin of error. It is generally not reasonable to expend a large
degree of effort in projecting an industry whose emission factors
may be off by 50 or 100 percent.
The decision on the amount of time and money to be allocated to
each industry is a choice for EPA to decide. The four categories
of significance, data needs, relative effort, and emission factor
reliability should be an aid in this decision. However, EPA personnel
may have other insights into the importance of each as a source of
emissions and also of the degree of difficulty in making regional
projections for each of these sources. A particular point source
may be relatively insignificant in terms of the total region but very
significant in the area immediately surrounding it.
5.4 Review of Point Sources of Emissions
In this section each of the point sources of emission will be
briefly analyzed. There is an appendix for each industry in which the
industry is examined in greater detail.
Iron and Steel—The iron and steel industry is a very substantial
emitter of particulates and fluorides. There were 134 establish-
ments in 1967, but 35 of them account for more than half of the
value of shipment in the industry. The critical elements in
projecting this industry will be to project total domestic output
by product and to allocate the share of output to each establishment.
The production process used in each establishment also has to be
projected as does the coking operation. The combined relative
effort is level three.
Copper—Copper smelters are very substantial emitters of sulfur
oxides in the areas in which they are located. There are only 10
copper smelters in metropolitan areas though there are some in
non-metropolitan areas as well. The projection of this industry's
output will require an examination of imports and recycling.
Allocation of output by region will call for a look at possible
new locations. The combined level of effort is one.
5-17
-------
Lead—The lead industry is a substantial emitter of sulfur oxides.
Demand is not growing and is likely to shrink in the future. Regional
allocation of output will be affected by possible closures. The
data needs and relative effort required to project the industry
are both relatively low.
Zinc—The zinc industry is also a substantial emitter of sulfur
oxides and like lead and copper, consists of a small number of
establishments. Little additional data will be required nor is
information on process type required. Projection of demand is
the primary task. Distribution of domestic production will have
to take closures into consideration.
Aluminum—Aluminum smelters are minor emitters of particulates
and very substantial emitters of fluorides. There were 14
establishments in metropolitan areas in 1967. The primary task
will be to project total demand for and domestic supply of the
product. Attention will have to be given to be given to possible
new locations when allocating total output on a regional basis.
The combined level of effort is three.
Petroleum Refining—Petroleum refineries emit significant amounts
of particulates, carbon monoxide and hydrocarbons and very
substantial amounts of sulfur oxides. Projections of demand
can probably be taken from the section dealing with mobile sources.
The role of imports will affect domestic production, and greatly
influence the location of refineries. Data are necessary on the
presence of catalytic crackers as this greatly influences the
amount of emissions. The level of effort required for this industry
is three,
Secondary Nonferrous Metals—The secondary nonferrous metals industry
is a small emitter of particulates on an establishment basis. There
are, however, nearly 600 establishments and thus, if there are a
number of them in one area, the emissions from them may be significant.
Additional data will be required on the location and production level
of each establishment in order to make accurate and reliable projections.
The amount of scrap to be recylced is a major determinant of total
output. The relative level of effort for making projections in
this industry is three.
5-18
-------
Kraft (Sulfate) Pulp Industry—The pulp and paper industry is a
very large emitter of particulates on an establishment basis.
There are 81 sulfate mills Ln metropolitan areas as well as
sulfite and mechanical process mills. Data are necessary on the.
production level of each establishment and on the process used.
Recycling will influence total output and new mills will help
determine the regional allocation among existing plants. The
combined level of effort in this industry is three.
Phosphate Fertilizer—The phosphate fertilizer industry consists
of establishments that manufacture phosphoric acid and also those
that manufacture phosphate fertilizer. This industry is a
substantial emitter of fluorides and a very small emitter of
particulates. There were 155 establishments in this industry
in metropolitan regions in 1967 and so the absolute amount of
fluoride emissions from each establishment is relatively low.
Projections of total and regional production can be made readily
but substantial amounts of data would be required in order to
project emissions from each establishment. The combined level
of relative effort is three.
SuIfuric Acid—The sulfuric acid industry is a significant emitter
of sulfur oxides and a relatively minor emitter of particulates.
There are about 213 sulfuric acid plants scattered around the
country. The industry is changing rapidly due to the growing
production of sulfuric acid as a by-product and regional projections
of emissions will have to take this factor into account. The
relative level of effort in this industry is three.
Rubber Tires—The rubber tire industry is a minor emitter of
particulates. There are 54 rubber tire plants and only the
older ones emit pollutants. Information is required on the
emission controls on old plants in order to project emissions
from this industry. The critical item in making projections
is probably the projection of output from old establishments.
The relative level of effort is one.
5-19
-------
Coal Cleaning-—The coal cleaning industry emits substantial
quantities of particulates during the drying operation; although
the amount varies with the process. Data need to be collected
on the location, production level and processes used at existing
coal cleaning plants. Total output can be readily projected and
allocated regionally based on trends. Possible new locations
will have to be examined for impact on emission levels in these
areas.
Feed and Grain Industry—The feed and grain industry is a very
large emitter of particulates. There are more than 6,000
establishments, but only a relatively small number are big
emitters of particulates. Data are required on the production
levels and number of establishments at each location in order to
make projections. Total output can be readily projected as can
the regional distribution once the necessary data are collected.
The relative level of effort is three.
Asphalt Batching—Asphalt batching plants are significant emitters
of particulates. There are more than 1,000 asphalt batching plants
in metropolitan areas, distributed roughly in proportion to the
population and level of construction activities. Additional data
required are the location,production level, and controls at each
establishment. Some plants are mobile, which further complicates
the task. The combined level of relative effort is three.
Cement—Cement manufacturing plants are substantial emitters of
particulates. There are 138 cement plants located in metropolitan
areas. Good data are available on the location and production
level of each of these plants, as well as the production process
and age of the plant. The primary task will be to predict the
location of the new, larger plants that are replacing existing
plants.
Brick Making—Fluorides are emitted only from brick manufacturing
establishments that use clay containing fluorides. It will be
necessary therefore to survey all brick making establishments
to determine the type of clay that each uses. Projection of total
output and the regional distribution should be readily accomplished.
The relative level of effort is two for this industry.
5-20
-------
Lime—The lime industry is a substantial emitter of particulates.
The amount of particulates emitted from lime plants is a function
of the type of kiln used and the type of fuel. Data needs to be
collected on an establishment basis for production levels, type
of kilns, and the fuel used. Total output and the regional
distribution of production are both substantial tasks.
Gasoline Marketing and Bulk Storage—The gasoline marketing and
bulk storage industry is an emitter of hydrocarbons. There are
about 30,000 establishments in 1967 and very little is known
about their location, throughput, and tank types of controls.
Because emissions from the average establishment are relatively
small and the rate of emissions is declining, a series of
assumptions may be the best way to allocate total throughput.
Otherwise, an extensive data collection effort will be required.
5.5 Computer Model
The approach envisioned in making projections of air emissions from
point sources is to make each of the point sources a submodel. The
process type and production levels would be determined outside the
model. The level of controls could be specified within or without
the model. Proportional changes in the level of production could
be made if this were desirable. Each submodel would have built into
it the output in physical terms, the emissions associated with this
output, and best estimates of the level of controls in the industry.
There is little reason for the regional distribution of output among
point sources to be determined within the large projection model.
The projection equations should probably be expressed in per capita
terms so that total output, and regional output in some cases, will
be a function of population. It will also be desirable, if possible to
make total output a function of GNP (as well as the variable) so
that this variable may be manipulated for the entire model. The model
should be flexible enough to handle different assumptions about imports
and import policy as this will be an important factor influencing total
output in some of the industries. Finally, the model user should be
able to assume a new plant for one or more industries in a region
in order to see the effect on emissions.
5-21
-------
SECTION 6
CONSTRUCTION OF THE COMPUTER MODEL
6.1 Scope of Work
The model for projecting air emissions must provide emission estimates
at a national level. However, it must be possible to disaggregate national
emission estimates by geographical region and polluter source.
There are three general sources of pollution: mobile sources, point
sources, and area sources. Mobile sources are primarily motor vehicles
and aircraft. Point sources include such things as solid waste disposal
facilities, power generation facilities, petroleum refineries, pulp and
paper mills, chemical, gasoline marketing and bulk storage, and food,
agriculture and mineral products industries. Area sources include such
things as residential and commercial heating plants.
The submodels used to compute emission estimates from these sources
must reflect variations for a given type of emission due to such things
as different industrial processes and different types of raw materials.
Thus, in many uses it would not be possible to develop a single submodel
for a given pollution source which would be applicable, to all geographical
regions.
6.2 Projection Strategy
6.2.1 General
There are two basic approaches to projecting air emissions:
extrinsic methods and intrinsic methods. The extrinsic methods project
the variable of interest. In other words, they extrapolate the variable
based on historical trends. The intrinsic methods project related
variables that predict the variable of interest. Often considerable
effort will have been spent in projecting the related variables, i.e.,
the Bureau of the Census statistics on population. The intrinsic
methods are normally preferred when adequate data are available for
the related variables and when the variable of interest can be written
as a function of several related variables. Both extrinsic and intrinsic
methods would probably be used in developing a total national projection
strategy.
6-1
-------
A variation of an intrinsic method is the growth model. Using
growth models to project the increase in air pollution emissions, an
equation can be developed that expresses emission variables as a
function of various demographic and socioeconomic factors whose values
are currently known. Using this approach, estimates of the changes
over time in the value of a given parameter as a function of a
set of independent variables can be made. Alternatively, new values
of the parameter at a given time can be calculated if values of the
parameter at an earlier time are available. These alternatives are.
expressed mathematically as follows:
P -P
A?= ~At~= f(Xl' X2' •••' V
where:
— = change in parameter P over time,
P? = value of emission parameter at time t?,
P, = value of emission parameter at time t1,
At = t2 - t1,
X1 , X , . ..., X = independent variables,
or the alternative expression
P2 = ?i + At(f(Xr X2, ..., Xn)) (2)
6.2.2 Operational Aspects: Model Methodology
The basic steps that would be involved in projecting national
air emissions are the following.
First, national air emissions would be disaggregated into some
logical set of components such as by region, by type of polluter and
by type of pollution.
Second, an equation for each emission category would be developed
using one of the methods identified above in Section 6.2.1, or a vector
of data points would be defined and schemes for interpolation or extrapo-
lation of these data would be determined. In general, the equations
6-2
-------
would consist of a set of coefficients (technological coefficients)
identified by regression techniques and an associated set of independent
variables. Both coefficients and variables would be supplied to the
model by the user or would be defined internally by the projection
model. The technological coefficients would be required for a given
polluter source to reflect different geographical areas.
Third, the equations derived in step two would be used to calculate
air emission for each logical component.
Finally, the emissions would be summed by different subcategories
and/or totaled for national air emission estimates and the results
reported. The equations for the different types of sources (point,
area, and mobile) would be independent; thus, these emission estimates
would be additive.
In addition to using the model to estimate total national emissions,
subcategory emissions, or emissions from individual components, it
should be possible to use the model to perform sensitivity analyses.
With sensitivity analyses, variables which could contribute to a signi-
ficant portion of the total emission would be identified. A sensitivity
analysis should also'be performed on the technological coefficients
discussed earlier and the coefficients that require accurate estimation
would be identified. The model could then be rerun and the effects of
modifications to critical variables or coefficients determined.
6.3 Computer Model Characteristics
6.3.1 Introduction
In the previous section a basic mathematical approach to
calculating air emissions is described. However, the mathematics must
be incorporated into a comptuer model that itself must have a number of
special features in order to provide users with a flexible working tool.
For example, the model should be easy to use, provide a number of output
options, allow for easy modifications of the data, and give clear error
diagnostics. Recommended features of the computer model are discussed
in the remainder of this section.
6-3
-------
6.3.2 Model Input
Input data would include the type of run (such as a projection
run or a catalog update run), information relating to various output
and sensitivity analysis options, and model variables such as the year
that the projection will be terminated, the population, or the gross
national product.
The model should be structured so as to allow a user to spend a
minimum amount of time learning input rules. In most existing models,
data are entered with specified fields, for every variable, an input
approach which is easy to program but which also results in many rules
for input. As an alternative, each piece of data should be introduced
by a descriptive attribute. In other words, each item of data would
be named, and for input the value associated with the name would
immediately follow it. This allows the data to be introduced in name/
value pairs in any order and allows the user to omit input variables
that are not needed for a given run. Variables that are not input can
either be assigned default values by the model, based on national data
estimates or ignored completely if the particular item is not critical
to a given run. The name/value data entry technique provides an
extremely flexible data entry procedure from the standpoint of the
user, and yet input conversion is still straightforward from the stand-
point of the programmer. The existence of the value for a model variable
in the input should override catalog values or values that might normally
be estimated by the model during a run.
A brief example of possible model input is as follows (each line
represents one data card):
TYPE OF RUN: PROJECTION
MODEL VARIABLES: YEAR 1975 POP 2.33E8
GNP 1.09E12 DWELLINGS 6.11E7
EMISSIONS: A24 2.16E-1
SENSITIVITY: GNP .10
6-4
-------
The technological coefficients discussed earlier are integral
parts of the model equations and would not normally be considered a
model input. These coefficients would be derived externally to the
model and cataloged for latter use. During a run, the coefficients
would be accessed by the descriptive attributes associated with each
coefficient set (such as geographical region, etc.) and used in the
computation procedures. However, catalog updates should be allowed
either as part of a model run or on special update passes. The user
should be allowed to replace complete sets of technological coefficients
or to modify a set by introducing a percentage change in the entire set
or possibly to a portion of the set.
6.3.3 Model Outputs
The key model output would be reports describing the emission
estimates produced during the model run. These reports would include
multiway tables (two-to three-way, at least), sorted lists, and key
model results by geographical regions. It is felt that it is more
important to provide the user with the ability to stratify the file
and produce reports according to these strata than it is to provide
the user with a large number of report options or report formats.
In addition to producing reports, the model should automatically
produce a data set of all emission data computed in a given run. This
data set would be maintained for n runs and then rewritten on the n+1
run. That is, a series of n data sets of results from the latest n
runs should be maintained. The existence of these backup files immediately
implies the need for output options that allow reports to be produced
directly from a specified backup data set without recalculation of
emission data. Data from the backup data sets could also be used as
input to statistical analysis routines.
Another important model output is error diagnostics. In addition
to the diagnostics produced by the operating system, the model should
produce diagnostic messages related to the computer model and the data
it processes. Whenever possible, the user should be able to read the
error message from his printout rather than being forced to refer to
an error code book. The model should look for and report invalid
6-5
-------
attributes, invalid ranges on certain variables, infeasible answers,
and, whenever possible, logical inconsistencies between sets: of answers
or sets of input data. Insufficient error diagnostics could make the
model difficult to use and seriously limit its effectiveness.
6.3.4 Model Documentation
The model documentation should consist of a technical report,
a user's manual, and annotation of the computer programs.
The technical report should contain the mathematics used to calculate
emissions and derivations and explanations of the mathematics and
computation procedures whenever necessary.
The user's manual would contain procedures for making computer
runs, descriptions of the various model options, input data formats,
descriptions and examples of reports that could be produced by the
model, system job control language used in making runs, data file
formats, and any other information that is necessary for a user to
effectively operate the computer model. To as great an extent as
possible, the user's manual should be a brief "cookbook" type of report
that is carefully indexed to allow easy access to the various sections.
The computer program annotation should contain enough information
for a programmer to learn how the model operates and to make modifi-
cations to the computer code if needed. At a minimum this documentation
should contain an overall English language description of how the
computer program operates, at least one level of flowcharts, and a
parameter list that defines and describes parameters which are internal
to the computer code as opposed to the named input data variables.
The computer program annotation should not be developed in its final
form until the model has been thoroughly tested and is considered
operational.
6.3.5 Other Considerations
6.3.5.1 Accuracy of Answers. The answers produced by
the model should of course be as accurate as possible. However,
when an answer is questionable, error bounds or estimates of a probable
error should be provided. Numerical stability should be considered;
however, it is not likely to become a critical consideration since the
6-6
-------
proposed methodology does not call for solving large sets of potentially
ill-conditioned equations. Once the model is operational, the computa-
tion scheme for variables to which the model is found to be particularly
sensitive should be reviewed for possible modification of the computa-
tional procedure in order to provide additional accuracy.
6.3.5.2 Hardware and Software. There do not appear to be
any unusual hardware features that must be available in order to operate
the proposed model. However, the model should be compatible with EPA's
computer facility. This compatibility can be promoted by developing
the model for an IBM 360 using secondary tape storage or 2314 disk file
storage. Software compatibility can be promoted by developing the model
in a standard version of a general purpose programming language such
as FORTRAN or PL/1.
6.3.5.3 Flexibility in Altering the Computer Model Code. To
as great an extent as possible, the model designers should develop the
model in a modular fashion to facilitate modifications and extensions
of the model. Persons who are not familiar with the overall model logic
should be able to modify specific model routines (modules). Input
required and output produced by the routine should be described as well
as the function that the routine performs. Whenever possible, discussion
of what is required to extend the routine or modify the routine should
be included.
6.3.5.4 Model Test Procedures. Once initial runs have
been made with the model and it is apparently debugged, the procedures
that will be used to thoroughly test the model should be formally
documented and the results of the acceptance test should be reported.
Whenever possible these tests should include projections of earlier
data to currently known data. If this cannot be done, model projections
should be compared with projections obtained from other sources.
6.4 Computer Model Implementation
The project is divided into three phases. The first phase, the
feasibility study, identifies the methodology that is required for
projecting various emission categories. In the second phase the
computer program will be written and a few industries selected for
6-7
-------
implementation. These industries will then be disaggregated by geography,
and the technological coefficients will be obtained for each selected
emission category. Finally, these selected industries will be incorporated
in the computer program.
The third phase will incorporate the remaining emissions categories
not included in Phase 2. At the completion of Phase 3 a total national
emission estimate for all industries should be provided. It should be
noted that most of the computer program will be developed in Phase 2
except possibly for a few minor details. In Phase 3 the variables
and equation lists will be expanded to include the new emissions
categories but no additional computer code will be required by the
inclusion of these categories.
6.5 Validation of the Model
Any new model designed to estimate the future values of dependent
variables can truly be validated only by testing it against actual empirical
data for time periods for which the model was designed to provide projections.
In other words, true validation can be provided only after sufficient time
has elapsed for the operation of the model to be compared with actual data
and its success in tracking the variables has been demonstrated enough
times to remove most chance elements. Even then, the fact that a model
tracks well does not prove that its logic is correct; only that it
shows a high correlation with actual data.
A kind of preliminary validation can be accomplished, if the base
year of the model is not the current year, by running the model for a
year in which real data is available. In this instance, if the model
is constructed on a base of 1970 and earlier data, the model could be
run for 1972 and the results tested against available emission data
for 1972, in selected AQCR's. Several dangers are inherent in this
approach, however. One is that the model will be designed to provide
10 and 20 year projections and its ability to project with reasonable
accuracy for 2 years would not ensure its validity in projecting for
the longer period. Also, to serve as a test of validity it would be
necessary that the emissions data be derived independently and not
6-8
-------
based on the same types of data and assumptions used in constructing
the model.
The model could also be run backward to some earlier year and its
estimates compared with available empirical data. However, since the
model will be based on known trend data, a good fit with the past
tends to show only that it expresses the past relationships with reasonable
accuracy.
One other effort can be made to provide as valid a model as possible.
That is, each of the data series used as inputs may be examined with great
care for accuracy and for the assumptions used. Any model is only as
good as its inputs.
6-9
-------
SECTION 7
FEASIBILITY, CAPABILITY, AND LIMITATIONS
The discussion in previous chapters of this report have made it
clear that an emissions projection model of the type desired by EPA
is feasible. It has been determined that there is enough known about
the present status of mobile, area, and point sources of air pollution
emissions and their future growth patterns, as well as the theories
of the economic determinants of regional growth to provide a sound
basis for projections. Most of the essential data are available
from Government and private sources, or can be interpolated or syn-
thesized from available sources, thus minimizing the collection of
new data series and the use of assumed values. The nature and
structure of the required computer model has been examined and found
to be conceptually feasible.
The model conceptualized in this study will project, for the tenth
and the twentieth year from date, by county or AQCR, the number and
operating size of each source category, and the resultant emissions at
a prespecified level of control. This would represent the initial
specifications of the total model. The sources included would be
those described in Sections III, IV, and V of this report. Emissions
would be estimated for the five pollutants discussed, i.e., particulates,
SO , NO , hydrocarbons, and CO. The model would be so structured as to
X X
permit the addition of subroutines for additional industrial sources
and additional pollutants. The model would also be capable of aggregating
the emissions data to provide state and national totals by pollutant
and by source category.
The accuracy of such projections at the county or AQCR level will
depend primarily on the accuracy of forecasts of the location of new
plants, of the rate of growth of production in each industry category,
of the patterns of population growth and migration, and of the patterns
of vehicle movements. Projections of any of these factors are necessarily
subject to considerable error.
7-1
-------
The rate of growth of production is now being estimated by many
agencies, both within the Federal Government and by private university,
consulting, and trade organizations and these provide a reasonable basis
for estimates of the total output of each industry within this model.
However, estimation of the future growth of production from the individual
plants in an AQCR has much less basis. Firms may prosper and grow or
fail to do so depending on the effectiveness of management decisions,
degree of competition, and many other factors that cannot be included as
variables in the proposed model. It will be necessary, in making emission
estimates for AQCR's, to assume that each plant in an industry behaves
as does the industry average, unless special information about geographical
trends is available.
Plant closings and expansions can be predicted only very roughly. If
closings are anticipated, it may be assumed that older plants will be the
ones closed, but such an assumption is subject to exception. The expansion
of existing plants and the location of new ones is even less determinable
10 and 20 years in advance.
The proposed model cannot, therefore, be expected to provide more
than a general indication of possible patterns of the growth of emissions.
Operation of the model to answer questions of the "What if..." form, with
alternative values for the significant parameters can be useful to EPA,
however. It will provide a means of quantifying the potential impact of
hypothesized changes in industry emissions sources and of projecting a
large scale picture of impacts of change that could occur in many regions
s imultaneously.
Emissions from areas will depend largely on population movements and
to a lesser degree on industry location, although the two are obviously
interrelated. Population movements are carefully studied by a number of
agencies and have been projected with a limited degree of accuracy in the
past. The 10 and 20 year estimates required for the proposed model will,
however, be subject to substantial possible error. This will be especially
true for small regions (AQCR's and even states) due to potential changes
in the rate of population growth and, more importantly, to shifting
patterns of urbanization and migration.
7-2
-------
Mobile source emission estimates will be subject to the same
limitations as area source estimates, since they are closely linked to
population. In addition, it will be particularly difficult to project
future commuting patterns and changes in transportation technology. Both
of these factors may change quite rapidly and fundamentally, especially
over a 20 year time span.
The model should be tested to determine the sensitivity of various
parameters. It is probable that many of the variables discussed above
will prove to be those which are substantially sensitive, so that small
variations will significantly change the estimates of emissions for in-
dividual AQCR's. These should be subjected to independent analysis to
insure maximum accuracy of data inputs.
Any predicative model should be regarded as providing conditional
estimates of probable values of key variables, not absolute statements
of what will happen. Models, therefore, should be used as aids to
judgment and analysis. The proposed model can be run annually and
continuously revised to incorporate improved estimates of any of its
variables. Used in this way it should provide an excellent means for
evaluating complex interrelationships influencing emission sources and
prove useful for identifying potential future problem areas.
7-3
-------
APPENDIX A
EVALUATION OF NATIONAL MODELS
1.0 THE OBERS PROJECTIONS
1.1 General Characteristics
The objectives of the program under which these projections were pro-
duced were to develop: (1) a regional economic information system covering
both the past and the future, with provision for rapid data retrieval; (2)
short, intermediate and long-term projections of population, economic activity
and land use for the Nation and its subareas; and (3) special analytical
capabilities. The information system and projections form an economic frame-
work within which future economic needs for the development of the Nation's
resources can be estimated. The analytical system provides a systematic
procedure for quantifying the nature and magnitude of economic benefits and
impacts resulting from specified types and scales of resource development.
The projections were developed by the Office of Business Economics
(OBE), presently the Bureau of Economic Analysis (BEA), of the U. S.
Department of Commerce, and the Economic Research Service and the Forest
Service of the U. S. Department of Agriculture. The effort was sponsored
by the United States Water Resources Council. The program was initiated
in 1964. Projections of population, employment and earnings by State,
water resources areas, 173 OBE economic areas, and the SMSA and non-SMSA
portions of the QBE areas are scheduled for publication in the summer of
1972. Documentation of the projection methodology and preliminary pro-
jections of economic activity have been previously published by the United
States Water Resources Council (U. S. Department of Commerce and U.S.
Department of Agriculture, 1971).
The projections are basically developed by a computer model which
projects the share of employment and earnings by industry sector in each
of the 173 OBE economic areas. However, the projected trends of these
shares are examined for consistency and reasonability and adjusted if
necessary. In addition, a preliminary set of projections have been re-
viewed by designated officials in each of the 173 OBE regions. The final
series of projections are being developed and will incorporate the results
of these reviews. Thus, the OBERS projections have been developed by a
combination of man and machine techniques.
A-l
-------
1.2 Detailed Characteristics
The OBERS projections, like all projections, are conditional views
of the future, conditional on the realization of the assumptions employed
in their development and an extension of past relationships believed to
have future relevance for the measure being projected. The projections
are shaped by long-run or secular trends in the economy rather than by the
cyclical fluctuations which characterize the short-run path of development.
Assumptions, either explicit or implicit, which reflect this principle
and which are inherent in the projections, are as follows (U. S. Department
of Commerce and U. S. Department of Agriculture, 1971, pp. 1-4, 1-5):
1. Population growth will be conditioned by a substantial decline
from the fertility rates of the 1962-1965 period;
2. Reasonably full employment will prevail at each of the projection
points;
3. At projected dates, the economy is considered free of the
disruptive effects of foreign conflict;
4. Stability will be maintained in the conduct of international
trade;
5. Continued technological progress and capital accumulation will
support a growth in output per man hour of 3 percent annually;
6. Development of new products will be accommodated within the
existing industrial classification system;
7. Growth in output within the context of the existing industrial
structure can be achieved with environmental balance although this may
require control of energy resources, restriction of the use of
pesticides and other chemical products, and encouragement of population
dispersion; and
8. The historical trends in import/export activity are extended into
the future except for agricultural exports which, though continuing to
increase, will constitute a smaller share of U. S. production.
The OBERS projections deal exclusively with the supply side of
economy. Implicit in the projections, however, is the assumption that
A-2
-------
sufficient demand will be generated by the private and public sectors to
maintain a full employment level of economic activity.
The model is in no way an attempt to explain the process of economic
growth and development. Rather, it begins with projections of national
population and successively refines this total to the working age
population, total labor force, civilian labor force, civilian employment,
private civilian employment, hours worked per year per man, and gross
product per man per hour, , Projected Gross National Product results from
the mathematical combination of these variables.
The foundation for all projections relating to the agriculture sector
is projected national demands for food and fiber. These national estimates
provide both the conceptual and quantitative control for other elements of
the national agricultural structure and all regional distributions of pro-
duction, value, employment, earnings, and land use. Of particular note is
the fact that projections of agricultural product demands for domestic food
use are based on the projected rates of population growth and per capita
consumption with the latter related to projected levels of per capita
personal income. i
The overall system fbr preparing the economic area projections uses
four separate models. Basic or export industries except agriculture are
projected by a variation of the "shift-share" technique for regional
industrial analysis. This technique distinguishes a proportional growth
element and a differential growth element between a region and the Nation
in each industry or income component from historical data and projects these
elements into the future. Thus, employment (or earnings) in industry will
grow faster (slower) in a particular region than in the Nation, due to that
region's comparative advantage (disadvantage) in the production of that
industry's output.
Mathematically the shift-share approach to projecting regional economic
^
activity may be expressed as follows:
' Ei = (Eio/Eio> Eij + Cij
A-3
-------
The subscripts i and j refer to the i industry and th3 j region,
respectively; the subscript o refers to a summation: when in the right hand
position, it is the summation of the regions (= the Nation); when in the
left hand position, it is the summation of industries (= total employment or
earnings). The superscripts t and x refer to the projected and base periods,
respectively. The term C. equals the difference between the level of em-
ployment or earnings attributable to the national growth rate of the industry
and the regional level actually attained in the industry. The first term on
the right hand side of the above equation is the proportional growth element;
the C,. then is called the share or regional share effect.
x
The causal factors associated with the C.. terms are the essence of
industrial location theory. One approach to projecting regional economic
activity is to develop an econometric model which uses multiple regression
techniques to "explain" the variations in the C. effect across regions for
each industry in question. Such a model is currently under development at
the BEA (U. S. Department of Commerce and U. S. Department of Agriculture,
1971, p. 111-13), but was not used for the OBERS projections.—
For the OBERS projections, a more simple method for projecting the
C.. term, one which is less demanding of the data but which makes maximum
use of all presently available information, was employed. For each of the
basic industries, a curve was fitted to each region's percent of the national
total of earnings and employment (separately) for the selected years for
which data are available. These curves were then extended into the future
and values of the region's projected percentages in the target years were
determined. Thus, the approach employed is actually a variation of "shift-
share" analysis with the regional share effects calculated implicitly rather
than explicitly.
I/The BEA model projects the share effect for each 50 industries. Presumably
one of the reasons for the lack of acceptable results from this approach is
the fact that substantial variations in locational factors may exist within
these 50 categories. Accordingly, this approach may prove more satisfactory
if the industry sectors are defined in more detail.
In order to project air emissions, it will be necessary to project
industrial activity at the four-digit SIC level of detail for approximately
15 industries. Since there is a relatively small number of these industries
and since their characteristics are relatively uniform within a four digit
SIC category the feasibility of using a multiple regression approach to
projecting the share component and thus the regional distribution of economic
activity will be explored in the remaining efforts of this project.
A-4
-------
Projected levels of agricultural production for each region are
consistent with the availability of natural resources for agricultural
use. They are based in part on the use of an interregional linear program-
ming system which identifies a pattern of land use and geographic patterns
of output with a minimum outlay of capital, labor and other economic resources.
Employment and earnings in the local-service or residentiary industries in
each area are projected on the basis of historically-developed relationships
with the area's basic industry activity. Finally a consistent projection
of regional population is developed by assuming this total to be a function
of regional employment and income.
Preliminary projections are currently available from the OBERS system
for selected time periods extending almost fifty years into the future:
the years 1980, 1990, 2000, 2010, 2020. A final series of regional pro-
jections for each of these years is scheduled for completion in the summer
of 1972.
The basic geographic areas for which the projections were developed
are the 173 QBE economic areas which completely span the Nation. These
areas are composed of groups of counties and were determined in such a manner
that inter-area commuting from place of residence to place of work was
minimized. In other words, the QBE areas are relatively self-contained.
regions of economic activity, measured in terms of residence/work locations.
One disadvantage of this regional framework is that the QBE regional
boundaries do not coincide with certain political and statistically defined
areas, e.g. states and SMSA's. However, projections for the 173 areas have
been subdivided into approximately 900 sub-sectors in order to provide pro-
jections for various geographic areas. The scheduled summer 1972 publication
will contain projections for states, 20 water resources regions, SMSA's and
the non-SMSA protions of each state and QBE region. Of particular note
is the fact that population, personal income and earnings projections for
55 Air Quality Control Regions, established as of October 1969, have been
developed within this regional projection system (U. S. Department of Commerce,
1970).
A-5
-------
The OBERS system provides internally consistent projections of
population, employment, personal income and earnings. Per capita
personal income and earnings per worker are also provided, both in
absolute terms and in U. S. relative indices. Both earnings and in-
come are projected in constant 1967 dollars.
Employment projections are developed from historical data collected in
the decennial census. The "persons engaged in production" employment series
used in projections of the national industrial structure is conceptually
consistent with other data series used in the overall projection effort and
is therefore the proper series to use in projecting industry relationships.
However, since this series is not available in the required geographical
detail the projected national industry employment was converted to the
Census employment concept.
Employment and earnings projections are available for each of the 173
QBE regions for 37 industrial groups, mostly consisting of two digit level
SIC detail. However, projections for other geographic areas (e.g.
water resources areas, air quality control regions) contain only
employment totals. For all types of geographic area aggregations
earnings for each of the sectors are provided.
The OBERS projections, as with any regional projection effort, require
large amounts of somewhat detailed data. In .the necessary compromise be-
tween theoretical adequacy and data availability in regional projection efforts,
the BEA feels that the approach chosen is usually dictated by the type.s of data
that are either available or can be developed within a reasonable expenditure
of resources. As mentioned previously, their projection methodology confirms
this belief.
The regional population and employment data series for the OBERS pro-
jections are obtained from the results of the decennial population censuses.
The area estimates of personal income and earnings by industry sector are
constructed from a wide variety of statistical information, consisting mainly
of compilations by government agencies, although data are also drawn from
numerous private sources. The income and earnings estimates were developed
A-6
-------
Population
Employment
within the framework of the Commerce Department's official state estimates
of personal income and are available annually on a county basis.
The current BEA data base within Regional Economics Division's Regional
Information System consists of the following items:
Decennial census totals by county for 1930, 1940, 1950,
1960.
Decennial census data for 1940, 1950, and 1960 for each
of the 37 sectors for each of the 173 QBE economic areas.
For the 173 areas, employment totals for the 37 sectors
were estimated for 1966. County employment estimates
for agriculture for 1962, 1965 and 1966. For 1967, 1968,
1969, and 1970 wage and salary employment by 2-digit
SIC classification by county and by place of work are
now available, although these data were not used for the
current projection effort. In addition, estimates of
self-employed persons in agriculture and non-agriculture
industries are available by county for these years.
County data through 1969 available classified into a
variety of sources.
County data through 1969 are available by 2-digit SIC
category. From 1967, annual estimates of wage and salary
income are available in 60-industry detail at the county
level. However, for each of these series, much of the
industry detail cannot be shown at the county level in
order to avoid disclosure of confidential information.
1.3 Operating Procedures
The Regional Economics Division of the BEA plans to update these
projections on a regular basis, hopefully every two years. Future pro-
jection efforts will of course benefit from an improved data base and
projections procedures. In particular, the Division is planning to provide
projections of employment in each of the 37 sectors for sub-QBE regions
during the next projection cycle. They are also planning to improve the
shift-share methodology for projecting regional economic activity.
Personal income
Earnings
A-7
-------
These updating efforts will, of course, require a substantial amount
of resources, most of which are related to updating the appropriate data
bases. However, the Regional Economics Division seems to be committed to
maintaining the data base, as a large portion of its staff of approximately
60 is continuously involved in this effort.
It is doubtful that the OBERS methodology as it now exists can be
efficiently used in a sensitivity analysis model. In order to examine the
effects of alternative assumptions, it would be necessary to generate a
large amount of additional data and to perform numerous additional computer
runs. For sensitivity analyses, it may be more appropriate to take the
relatively aggregated output of the OBERS projections as fixed and perform
the industrial and geographic sensitivity calculations within this frame-
work .
1.4 Evaluation for EPA's Use
The OBERS projections provide an excellent set of consistent regional
projections of population, personal income, and employment and earnings at
an adequate degree of industrial detail. The projections are provided for
a time-frame that falls within EPA's period of interest. Furthermore,
despite the large amount of resources required, the Regional Economics
Division of the BEA appears committed to updating the projections on a
regular basis.
The primary disadvantage of the OBERS projections for the EPA's use is
that the appropriate measures are not provided in the necessary detail to
project air emissions. In order to project these emissions, some measure
of output, measured in physical terms, by industrial sector is required,
whereas the OBERS projections furnish only employment and earnings projections.
However, it is felt that a proxy for physical output can be constructed by
projecting the relationship of some output measure, such as gross product
originating, to earnings for each industry sector, with both variables
expressed in constant dollars.
As mentioned previously, the OBERS projections are based on supply factors
only and do not explicitly include the effects of changes in demand patterns
on the industrial composition of output and employment. However, in order
A-8
-------
to include demand factors in the projection methodology, various components
of demand would have to be projected for each of the 173 QBE regions. In
addition projections of trade flows among the regions would be required as
all regions produce goods for export and import goods and materials from
other regions. Projections of these measures would require a significant
expenditure of resources, which would perhaps be more effectively employed
in developing an improved basis for other components of the overall air
emissions projection framework.
Finally, the OBERS methodology may be criticized in that it is essentially
a trend projection technique. This is true to a certain extent, at least
with respect to projections for the basic industries. However, the trend
projections for these industries are modified by the judgment of both BEA
staff and, if the current practice is continued, by knowledgable persons in
the regions themselves. Again it appears that the necessary compromises with
the available data have resulted in a projection methodology that does not
contain all the features desired but one which, nonetheless, is acceptable
for OAP's purposes.
Although it is difficult to evaluate the accuracy of the regional pro-
jections, the preliminary national projections of economic activity as
developed within the OBERS framework have been compared with the BLS national
projections for 1980 (see Section 6.0 of this appendix for a brief dis-
cussion of the BLS approach to developing projections). Although the two
sets of projections were developed by different approaches, the national
totals for 1930 were reasonably consistent, a finding which helps tc justify
the recommendation of the OBERS projections for the OAP's use.
Disaggregation of the OBERS projections to both the appropriate degree
of industrial and geographic detail may not be so easily accomplished.
However, a more promising approach may be to develop specific procedures for
developing the detailed outputs required to project air emissions and to use
the resulting measures within the framework of the overall OBERS projections.
In summary, the OBERS projections seem most appropriate of those
evaluated for use by the OAP to project air emissions. They form a consistent
set of projections of several key economic and demographic variables that
A-9
-------
can be input to the development of a more detailed procedure for projecting
air emissions. Furthermore, the OBERS earnings projections, which can be
obtained for a variety of county groupings, will be most useful in developing
an index by which projections of physical measures of output can be obtained.
Finally, since the OBERS projections constitute the basis for a long-range
water resource planning effort it seems appropriate to provide the samp, basis
for long-range projections and planning in the area of air emissions.
A-10
-------
2.0 THE HARRIS MODEL
2.1 General Characteristics
The objective of this model is to forecast industry activity on a
regional level along with other regional variables, including population,
income, labor force and unemployment. It is assumed that there is a
tendency for labor and capital to migrate in order to improve their
returns. The migration of capital is explained with a set of industry
location equations that predict the change in output by region. The
migration of labor is explained by a set of population migration equations,
The forecasting model is being developed under the leadership of
21
Dr. Curtis C. Harris, Jr., at the University of Maryland.— The develop-
ment was supported in part by a grant to the University of Maryland from
the Economic Development Administration, United States Department of
Commerce. At the present time, work is continuing under funding from
the National Science Foundation.
The projections are developed by a computer model, which projects
values for each of the regions and insures that they are consistent with
separately derived national projections. Once the equations in the model
have been developed and selected, the projections are generated by the
computer model without' any manual adjustment.
2.2 Detailed Characteristics
At the present time, the Harris model has not been used to make
projections for a future year. Therefore, it is not possible to report
on the assumptions that are used to develop the future projections.
2/Complete documentation of the Harris model is not available at the
present time. Dr. Harris supplied draft chapters from two books, A
Regional Forecasting Model and An Industry Location Model, which are
based on the model building effort. The structure of the model is out-
lined in Harris (1970). A different type of model for projecting county
economic activity, whose development was not completed, is documented in
Harris and McGuire (1969). This latter model is based on the shift-share
technique for projection regional economic activity.
A-ll
-------
The forecasting model starts by projecting the location of firms
in each industry. The change in output of each industry sector is ex-
plained by two sets of variables: (1) input prices that firms face in each
location, and (2) agglomeration variables that help explain location
behavior that is not accounted for by prices. After output has been
determined, then employment, population, earnings, and personal income
are derived. Also, final demand sectors are projected—consumption,
government expenditures, investment and foreign exports.
The model is recursive. The supply and demand data in year t_
are used to forecast variables for the year t^ + 1, then the forecasts
are used as data to make forecasts for the year t^ + 2. Successive
forecasts are developed in this manner until those for the target
year are provided. In any given year, predetermined changes may
also be made in the input data, such as changes in governmental ex-
penditures.
An important set of variables used to determine the location of
output is the transportation variable. These variables are the cost
of transporting a marginal unit of a commodity either into or out of
a region. For example, in explaining the location of the steel industry,
explanatory variables include the marginal cost of shipping a unit of
steel out of each region and the marginal cost of shipping a unit of
iron ore and other inputs into each region. These are derived by deter-
mining the cost of shipping a unit of goods between each pair of regions
by rail and truck for each of several weight classes. The least-cost
method of shipping goods in each weight class for each commodity is
determined and these costs are used in a linear programming transportation
algorithm in order to produce the transportation costs.
The model starts with a set of national input-output forecasts which
serve as control totals for the regional forecasts. The model requires
at least two years of data, although data for additional years would be
useful. The general steps of model operation are as follows:
A-12
-------
1. Estimate demand and supply at each location for the base
year (most recent year for which data are available) and the
year prior to the base year. The demand for intermediate
goods is estimated by applying input-output coefficients to
the industry output of each region and the final demand cate-
gories are estimated directly. Final demand is estimated for
twelve categories of personal consumption expenditures, gross
private equipment purchases by 69 purchasing sectors, private
and public construction of 28 types, federal government purchases
by 7 functional groupings excluding construction, state and local
government purchases excluding construction, gross exports and
Federal defense expenditures. Total supply by industry is the
sum of domestic output and imports, with both measured in dollars.
2. Estimate transportation costs of shipping a unit bundle of
the output between each region. Transportation costs are esti-
mated for both rail and truck shipments and are composed of both
terminal and line-haul costs. These costs are estimated for
different weight classes for each mode of shipment.
3i Apply a linear programming transportation model to obtain
the cost of getting an additional unit of an industry's output to
each region and the marginal cost of shipping an additional unit
of output from each region. The linear programming model is solved
for both the base year period and the year prior to the base year.
4. Using the cost for the year prior to the base period produced
in step 3 and other relevant marginal costs, estimate parameters in
the industry location equations. The other relevant marginal costs
include annual wage rates, value of land, output in the previous
period, equipment investment which serves as a proxy for capital
stock, representations of major buying sectors in the regions that
bought goods from the industry in question, and output of major
supplying sectors in the region that sold goods to the industry in
question. For the non-commodity industries (Services and Construction),
A-13
-------
the change in output is explained by the changes in the sizes of
the markets served.
5. Enter the base year's marginal costs and values" of other
explanatory variables into the industry location equations and
make regional forecasts of changes in industry output.
6. Forecast regional employment and payrolls using the output
forecasts.
7. Adjust each region's total employment and payrolls for com-
muting in order to convert the figures from an establishment to
a residence basis.
8. Forecast regional population using the population migration
equations and birth and death rates for each region.
9. Forecast regional labor force by applying labor force parti-
cipation rates to the population forecasts.
10. Add to payrolls estimates of other components of personal income
using the population and unemployment forecasts.
11. Forecast consumption expenditures using consumption functions
that relate type of consumption with income.
12. Forecast capital expenditures using the output forecasts.
13. Update exports, imports, government expenditures by assuming
that the regional distributions are exogenous.
14. Forecast intermediate demand to complete the forecast of demand
or each region, recompute the marginal costs from the linear pro-
gramming problem, and go back to step five using the forecasts as
data and repeat the steps for next year's forecasts.
The model is based on the assumption that business firms are moti-
vated by the desire to increase profits. They seek locations which will
improve profits, for example, areas with low costs and/or high demands.
A firm located in an area near its major buyers and also near its major
suppliers would have an advantage over firms in the same industry located
elsewhere. At any given time, firms are not located optimally, but they
will always be able to move to better locations. This process is slowed
by the fact that once a firm has located, then the plant and equipment
are fixed at the location and the firm would hesitate to relocate.
A-14
-------
Furthermore, the location of industry influences the location of
final demand. A firm makes income payments and workers spend this money
on consumption items. Therefore, if a new industry were to locate in an
area, the consumption expenditures and other components of final demand
would most likely increase in relation to this new industry. Thus,
final components should not be determined exogenously, but endogenously,
as is done in the model.
At the present time, no projections are available from the Harris
model. The model is currently in the process of being verified, using
1970 data, (the original data base was developed for 1965 and 1966). Once
the model is verified using the 1970 data, it is planned to employ it
to make projections for 1985. These projections will be developed
with the framework of national projections using an input-output model.
The regions referred to in the Harris model are the 3,112 county
or county type areas within the United States. These county projections
can, of course, be aggregated into a variety of regions—standard metro-
politan statistical areas, states, water resource regions or air quality
control regions.
The forecasts provided by the model include output, employment and
earnings by industry sector, unemployment, personal income, consumption,
investment, government expenditures, and population by age, race, and sex.
Output is given in dollar terms. Employment projections are based on
establishment data provided in County Business Patterns, and are adjusted
to national totals provided in the Employment and Earnings series published
by the U. S. Department of Labor. Personal income is consistent with the
official estimates provided by the Bureau of Economic Analysis of the
U. S. Department of Commerce. Population projections are provided for four
age groups—0-14, 15-34, 35-64, and 65 and older—two race groups—white
and non-white—and two sex groups—male and female. Adjustments are made
within the model to correct for differences in the bases for data collection,
some of which are collected on the basis of place of residence and some of
which are collected on the basis of place of work. At various stages in
A-15
-------
the projection methodology, matrices are adjusted in two directions so
that county totals conform to separately developed industrial, national,
or state control totals for various measures of economic activity or sizes
of population subgroups.
There are 99 industry sectors contained in the model. The sectors
correspond closely to the Office of Business Economics (QBE) input-output
sectors. In some case, these sectors are combinations of four digit SIC
industries; however, the majority of the sectors are for two and three
digit SIC industries.
The data requirements for this projection model are enormous. The
presently available data base has been developed for 1965 and 1966 from
a variety of sources. County payroll and employment data are based on
reports contained in County Business Patterns. Estimates of output by
county are based on a variety of sources. In particular output/ payroll
ratios for states were obtained from the Annual Survey of Manufactures,
1965 and 1966, and multiplied by county payroll figures from County Business
Patterns to get a first approximation to county output by industry sector.
These figures were then adjusted proportionately to come to the national
totals. Indicative of the resources required to develop the data base, even
for a single year, is the statement by Dr. Harris that efforts were begun
to update the data base for 1967, but were not completed due to lack of
funds.
2.3 Operating Procedures
As stated above, plans currently exist to verify the model using
1970 county employment data. Once this verification has been completed,
the model will be used to provide projections for the year 1985, working
within the framework of a current effort at the University of Maryland
to provide national projections for that year. No estimate was obtained
of the resources required to develop these projections. However, a
thoughtful examination of the magnitude of the relationships involved
and the size of the data base indicate that substantial effort will be
required to develop any series of projections.
A-16
-------
The model can be used in the sensitivity analysis mode by substitu-
ting various national control totals and altering values of selected
exogenous variables. In fact, it is anticipated by the model developers
that it will be used in an impact anlaysis role. An example cited by
them is to use the model to estimate the impact of alternative regional
patterns of defense expenditures. However, because of the magnitude of
the efforts involved in developing a series of projections, it is doubtful
that the Harris model could be effectively and efficiently used by the
EPA for the type of sensitivity analyses which appear ~to be relevant to
their problems.
2.4 Evaluation for EPA's Use
The Harris model has several advantages when considered for EPA's use
to project air emissions. First, it is based on a more satisfying theoreti-
cal foundation than some of the other types of regional projections models,
e.g., those relying on a relatively unsophisticated form of shift-share
analysis. The model attempts to explain the location of industry under the
assumption that over time industries will change their output patterns in a
desire to maximize profits. Thus, the basic model explains industrial location
as an inter-regional, multi-industry disequilibrium adjustment process rather
than by a partial equilibrium otpimization process.
Second, the model uses the county as a geographic data base. Thus,
the output can be aggregated to any combination of counties desired—
states, standard metropolitan statistical areas, or air quality control
regions. Third, the model includes a fair degree of industrial detail.
Although the industries contained in the model are not sufficiently
detailed for air emissions projection purposes, nevertheless there is
some detail at less than the two-digit SIC level, Fourth, the model
does include a consistent set of projections of both population and
economic activity. This consistency is necessary in order to produce a
consistent set of projections of air emissions from both stationary and
mobile sources.
There are also several disadvantages to use of the Harris model for
projecting air emissions. Perhaps the primary disadvantage from a practical
standpoint is the cost that would be required to maintain and update the
A-17
-------
requisite data base for using the model. The data base presently in
existence is based on only two years of data, 1965 and 1966. It would be
most desirable to provide additional observations in time with which more
reliable estimates of the various relationships in the model. However,
the estimation of an additional year's data base is an expensive under-
taking, has not yet been accomplished, and there is no assurance that
adequate financial support would be available for this undertaking in the
future.
Second, although estimates of both supply and demand are available in
the data base on a county basis, many of them are based on the application
of fixed coefficients for larger geographic areas or categories of in-
dustrial classification. Thus, it is appropriate to question the
reliability of these estimates for the smaller geographic areas. Third,
a most important component ,of the entire projection model is the proce-
dure for estimating the inter-county transportation costs for both raw
materials and finished goods. From the documentation of the model, it
appears that these transportation costs are based on an analysis of the
current rates and practices of common carriers. To the extent that
relative transportation costs among the various modes of transportation
will vary in the future, the transportation costs developed by the linear
program will give a misleading projection of the marginal costs of the
expansion of industry output in each of the counties.
Finally, a procedure for comparing and evaluating the results of
the Harris model with projections provided by other models and metho-
dologies reviewed during this project does not exist. When questioned on
this point, Dr. Harris was vague as to how he intended to compare his
model's estimates of employment and output for 1970 with the ex-post
figures for 1970. Thus, no objective criterion exists by which to
ascertain if the additional resources required to provide the data
base and develop projections on a county basis have been efficiently
employed.
A-18
-------
In summary, although the Harris model provides more geographic
and industrial detail than any of the others reviewed, it has not been
developed to a degree adequate to recommend it for EPA's use in projecting
air emissions. It is necessary to examine the results of the model and
desirable to have an expanded data base underlying the projections in order
to make a thorough and objective evaluation of the model. Neither is
available at present and hence the model is not recommended for use by the
GAP.
A-19
-------
3.0 THE MULTIREGIONAL INPUT-OUTPUT MODEL
3.1 General Characteristics
The multiregional input-output model (MRIO) provides a description
of regional (state or multistate units) economic interrelationships and
can be used to forecast regional economic activity based on exogenous
projections of regional final demand.
The model and the projections of regional final demand for 1980
were developed primarily by individuals at Harvard University and Jack
Faucett Associates. The report was sponsored by the Economic Development
Administration of the U. S. Department of Commerce and also by the U. S.
Departments of Transportation, Defense and Interior; and the Office of
Emergency Preparedness of the Executive Office of the President.
The model is on computer tape and is expected to be released to the
3/
public during the summer of 1972.—
3.2 Detailed Characteristics
The MRIO model is similar to other input-output models in its basic
dimensions. It differs, however, in that it consists of interrelated
input-output tables for 44 state or multistate regions.
Input-output is a method for taking into account the interdependence
among the industries or sectors of an economy. The method of presenting
this interdependence is by arraying the industries in an economy in
matrix. When in a row the industry is a producing industry with the
entries in the matrix across the row showing the industry's distribution
of sales. When in a column the industry is a purchasing industry with
the entries in the matrix down the column showing the industry's dis-
tribution of purchases.
In addition to the interindustry sales and purchases the input-output
table also has a set of final demand columns (consumer purchases, business
investment expenditures, government expenditures, and net purchases by
foreigners) and a value added row (employee compensation, profits, depre-
ciation, and indirect business taxes).
_3/The primary documentation of the model is contained in Polenske (1970).
This document contains numerous references to the various data sources and
working documents that were generated as the model was developed.
A-20
-------
Turning to a mathematical representation of an input-o.ut.put model,
the total output of any industry can be represented by the following
equation:
n
E x - c + i + g + t + x (i=l,...,n), (2)
= J
where
x. . = amount of output industry i sells to industry j,
c. = personal consumption expenditures for the output of
industry i,
i. = private investment expenditures, including inventory
changes of industry i,
g. = government purchases of the output of industry i,
t. = net exports of industry i,
x. = total output of industry i.
Although input-output tables are initially developed with transactions
estimates of interindustry sales and final demand purchases, the table's
usefulness is greatly increased when the transactions are converted into
a system of technical coefficients of production. The technical or input
coefficient is the ratio of input to output and can be written as follows:
(3)
x.
J
where
a.. = technical coefficient,
x.. = amount of output of industry i purchased by industry j,
x. = total input of industry j.
The complete set of technical coefficients arranged in matrix form
show the structure of production of the economy.
Substituting the value of x. . from equation (3) into equation (2) yields
A-21
-------
n
Xi = l aiiXi * (°i + ±i + 8i + ti)'
1 j=l 1J J i i i J. (4)
In matrix notation this can be expressed as:
f X = AX + F
where
F = the final demand vector c + i + g + t.
This is equivalent to:
X - AX = F
or (I-A)X = F
where
I = the identity matrix.
Solving for X, total output, yields
X = (I-A)'1? (5)
or, rewriting equation (5):
Xl= rllfl+r!2f2+ •••
X2 = r21fl + r22f2 + ''•
x = r ..f.. + r Of0 + ..,+r f .
n nl 1 n2 2 nn n
The r.. are the total requirements, direct and indirect, of industry
i necessary for industry j to deliver a dollar's worth of output to final
demand. They differ from the a.. in that they include the indirect require-
ments as well as the direct requirements shown in the a... The difference
in perspective can be illustrated by taking an example from the 1963
national input-output table. The technical coefficient (a..) of the motor
vehicle industry for steel is 0.0863. That is, each dollar of output of
motor vehicles requires 8.6 cents of direct steel purchases.
A-22
-------
However, to build a motor vehicle requires other inputs which, in turn,
require steel as an input. The technical coefficient for the rubber
and miscellaneous plast-ics products used by the motor vehicle industry
is, for example, 0.0223, but to produce rubber requires a direct input
of steel of 0.0051. The total requirements of the motor vehicle in-
dustry for steel (r. .) which include all the indirect steel requirements
of the type cited above as well as the direct requirements is 0.2121.
The MRIO expands on the national model by identifying for every
industry in a region the industry by region from which it purchases
its inputs and the industry by region to which it sells its outputs.
In developing the MRIO three basic sets of data were required
interindustry flows, final demands and interregional trade flows.
1. Interindustry flows. Detailed regional input requirements
were assembled for the agricultural, mining and construction
sectors. These estimates of requirements were based on secondary
data sources. For manufacturing" and services a detailed product-
mix approach was used to determine state-by-state input require-
ments since locational factors were judged as less likely to cause
significant state-to-state variations than for agricultural,
mining, and construction.
The interindustry flows are, as discussed above, converted to
a system of technical coefficients. While these coefficients can
be updated and even projected to reflect future price, technology
and industry mix patterns, such a task is quite complex. Therefore,
the most common current practice is to use the fixed coefficients
for the historical period as a reasonable approximation of the
future patterns, i ' '
2. Final demand. Estimates of final demand were developed for
each region for 1947, 1958, and 1963. Projections of final demand
were made for 1970 and 1980. Final demand consists of: personal
consumption expenditures, gross private domestic investment; net
inventory change; federal government purchases of goods and services;
state and local government purchases of goods and services; and net
foreign exports.
'? A-23
-------
Changes in final demand patterns are easily handled in the input-
output model. These changes can be projected for a future time period
and then used to solve for the industry output necessary to meet
the projected demand level and patterns.
3. Interregional trade flows. Flows were estimated from origin-
destination data on commodity movement, values of total output by
state, imports by port of entry, and the demand for goods by state.
The flows were first estimated in physical quantities and subsequently
converted to value terms. The basic geographic area for which the
model and projections have been developed is for the 50 states plus
the District of Columbia. However, since detailed interregional
trade flows could be obtained only for 44 regions the complete model
has 5 multistate regions. The model has 78 industries.
Projecting trade flows is a very complex task since, of the
three areas discussed here, it has received the least analysis by
economists. For that reason we expect that assumptions regarding
fixed inflow coefficients will be retained by most of the early users
of the MRIO model.
3.3 Operating Procedures
The model has been used by the staff of Harvard University to make
industry output projections for the 44 regions for 1980. These 1980 pro-
jections are based on exogenous projections of final demand, and were
developed under the assumptions of fixed technical coefficients and regional
trading patterns.
The model can be used by others to provide projections for other
years. However, as the projection period increases the assumption of the
fixed coefficients and inflows becomes more open to question.
3.4 Evaluation for EPA's Use^
The MRIO model can provide a consistent set of interrelated regional
projections. It can be explicitly adjusted to account for new industries,
technologies or trading relationships. It is, therefore, particulary use-
ful for determining the economic impacts of alternative policies. For
A-24
-------
example, if an .industry increases its output there will be an increase in
the outputs of the industries supplying inputs to it. The model provides
a convenient means for accounting for these interrelationships and for cal-
culating these increases and, given output-emission relationships, could
estimate changes in emissions levels.
There are, however, several disadvantages to the use of the MRIO model
for projecting regional economic activity and the associated air emissions.
First, as presently developed and implemented, the regions in the model
are states, or in some cases multistate units. Any projections for these
regions would have to be further disaggregated to be of use to the GAP,
an effort which would require a significant amount of resources. Also,
it is questionable whether an input-output approach could be effectively
implemented at a sub-state level because of problems of data availability
(see the discussion of the Harris model in Section 2.0 of this appendix).
Therefore, much of the value of the model would be lost.
Perhaps more important, the MRIO model gives projections of only one
economic measure - the output of the various industry groups measured in
dollar terms. The problem of developing a measure of physical output
remains, one which is not unique to the MRIO model, as discussed in other
sections of this memorandum. Although the MRIO model may possibly provide
a useful framework for projecting air emissions from point sources, other
measures (population, income) that are necessary to project air emissions
from mobile and area sources are not available from the model. Thus
additional model development efforts would be necessary to provide con-
sistent employment, output, population and income projections from the
MRIO model.
In additionj at the present time there are no plans to provide pro-
jections for any other time period except that already developed—1980.
State projections using the input-output model for other years would have
to be calculated during further development of the overall air emissions
projection framework, a task that would require a significant amount of
resources. But the projections already developed are based on the 1963
technology matrix. If projections beyond 1980 were made this assumption of
fixed coefficients becomes even more questionable. Finally, there are no
known plans at present for the government to continue this research or to
update the model.
A-25
-------
In summary, although it is felt that input-output analysis does pro-
vide a valuable tool for regional analysis, for the reasons cited above,
it is not recommended that the MRIO model be used to provide an overall
framework for air emission projections. It is possible, however, that
the MRIO model could be useful in other EPA applications and for these
reasons its characteristics should not be completely overlooked.
A-26
-------
4.0 THE NPA PROJECTIONS
4.1 General Characteristics
For more than two decades the National Planning Association (NPA) has
been a pioneer in economic projections. In addition to short-term fore-
casts for the overall economy, NPA makes detailed five and ten year
projections of the national economy and the economies of regions, states
and metropolitan areas.
The National Economic Projections provide five and ten year projections
of gross national product; industrial sales (shipments); output and
employment; investment, capital stock and productivity; consumption;
personal income and income distribution; and government spending and
revenues. Accompanying the projections are historical data for key series.
The National Economic Projections are published annually in a volume
4/
of some 300 pages ot text and statistical tables.— To facilitate its
use, statistical material is similar in definitions, format and sequence
to the national income and product account tables of the U. S. Department
of Commerce.
The Regional Economic Projections forecast population, employment,
personal income, and consumption for five and ten years for eight multi-
state regions, all states, and 224 metropolitan areas.— Considerable
emphasis is placed on estimating fugure patterns of interstate migration
and industrial location. The Regional Economic Projections are published
annually in volumes containing several hundred pages of text and tables.—
The projections are developed by the National Planning Association in
its Center for Economic Projections. The work of the Center is assisted by a
Research Advisory Committee consisting of experts from business, labor and
government.
4_/The latest series of NPA national projections are available in Al-Samarrie
and Scott (1971b).
5/The most recently published and metropolitan area and state projections
are available in National Planning Association, (1970a and 1970b), respectively.
jj/The "metropolitan areas" as used by the NPA are identical with the
"Standard Metropolitan Statistical Areas" as specified by the Office of
Management and Budget.
A-27
-------
Projections of relevant measures of economic activity for the Nation
are provided by a computer-implemented econometric model of the U. S.
economy.— Projections of economic activity in states and metropolitan
areas are initially developed by extrapolating recent trends in the
relative shares of activity in these areas. These trends, however, are
modified by such considerations as relative stages of industrial develop-
ment in each state, extent of urbanization, physical, natural and labor
resources available, and trends in new technological requirements, as
well as consideration of feedbacks on employment on income growth,
population shifts, and labor force changes. These modifications were
introduced partly because the available data permit judgments of a very
loose sort, and partly because the state and the metropolitan areas are
usually not economic entities providing appropriate analytical units.
Thus, the NPA projections have been developed by a combination of man
and machine techniques.
ft,2 Detailed Characteristics
The NPA national model contains functions for generating projections
of 99 endogenous variables, and requires the stipulation of numerical
values for 34 variables that are exogenous to the model. In forming judg-
ments concerning the merits of the current series of NPA Projections, the
reader should review the key assumptions behind the exogenous stipulations.
These assumptions are: the population projections used are those calculated
by the Bureau of the Census; these projections employ an average of the-
so-called "C" and "D" series fertility rate assumptions that, respectively,
women will bear 2.78 and 2.45 children during their lifetimes.
The armed forces are assumed to decline gradually to 2.6 million, remaining
at that level for the rest of the 1970 decade; and state and local government
employment is assumed to increase by approximately 36 percent, and that
federal government employment is expected to increase by approximately
17 percent between 1971 and 1980.
^/Documentation of this model is available in Al-Samarrie and Scott (1971).
A-28
-------
Other assumptions are that productivity is expected to grow at about
three percent per year over the coming decade, a pace not significantly
different from that experience during the 1960's. The real output per
man hour rate for the first half of the 1970's is expected to be 3.0 per-
cent, decreasing slightly to 2.8 percent during the last half of the
decade. Each of these assumptions is reasonably consistent with those used
by BEA and the Interagency Growth Project in developing their respective
1980 projections of national economic activity.
The NPA projections are developed in two stages. First, the national
projections are determined and then these are disaggregated to the regional
projections. The equations used in the national model are divided into
four groups:
1. Those explaining the calculations of gross national product from
the supply side;
2. Those explaining the expenditure components of gross national
product;
3. Price deflators;
4. Income components.
The supply equations for computing GNP arise from the employment of the
projected labor force and capital stocks at the projected rate of unemployment
and capacity utilization. Alternatively, it is the GNP required if the
stipulated unemployment rate is to be achieved. To compute the expenditure
or demand side of GNP, producers' durable equipment and private non-residential
structures investment data, state and local government purchases, personal
consumption expenditures, and imports are estimated through the use of
behavioral equations. Other sectors of final demand (residential construction,
inventory changes, exports, and Federal purchases) are considered as exogenous
variables outside the model.
The overall price deflator is exogenously determined through equations
to estimate most sectoral price deflators. In considering GNP as the sum of
disposable incomes, the income side of GNP is calculated as the sum of
disposable incomes - personal - business and government. The model employs
A-29
-------
behavioral equations, identities and stipulations to project the items
incorporated in these aggregates.
Turning to the state projections, employment is first divided into
two sectors - commodity-producing industries, (agriculture, forestry and
fishery; mining; and manufacturing) and non-commodity industries, (the other
broad industrial 1-digit groups). A modified version of the differential
shift analysis is used to disaggregate the projected national employment
Q /
in each commodity-producing industry among the states.— As expected,
annual variations in the differential effect for a specific industry
within a state are erratic. However, over longer periods of time the
differential effect has displayed a greater consistency, which is used
along with a strong dose of judgment for trend modification, to project
the values of the industry differential effects in each state. The fact
that the sum of the differential effects for all states nets out to zero
prevents the results from exceeding the national totals. Employment in
non-commodity industries is projected as a function of commodity-producing
industry employment in each state. In this procedure, trends and the
ratio of commodity to non-commodity employment in each state are compared
with those in the nation and projected values are developed.
Members of the labor force by age and sex are projected for each state
and these cohorts are combined to yield the total state labor force for the
projection year. The age-sex cohorts are projected by comparing trends in
labor force participation rates for the state with the national rates. Pro-
jected labor force participation rates for the nation are obtained from the
Bureau of Labor Statistics of the U. S. Department of Labor.
State personal income projections are made on the basis of per capita
relatives to the nation. That is, past trends in state per capita personal
income relative to national per capita personal income are extrapolated to
the projected year. The projected state relatives are multiplied by the pro-
jected level of national per capita personal income to yield projected state
per capita personal income figures.
%j Details of this procedure are provided in National Planning Association
(1965. pp. 68-71).
A-30
-------
Projected values of disposable personal income, that is income after
deductions for personal tax and non-tax payments, are developed by extrapolating
past trends in the ratio of implicit state personal tax rates to implicit
national tax rates.
Personal consumption expenditures are projected by first developing
national consumption functions for each consumption item using a least
squares equation which relates the percent share of disposable personal
income spent on a particular consumption item as dependent variable to
per capita disposable personal income as independent variable. Personal
consumption totals for each state are developed from the national totals
by modifying the national figures to reflect regional differentials in
consumption patterns for a particular consumption item.
The metropolitan area projections are developed by the same general
procedures as those for the states. That is, metropolitan area economic
and demographic activity is related to economic activity in the relevant
state and other analytically relevant control areas.
National, state and regional projections are prepared by the NPA on a
regular basis. At the present time national economic projections are
available for 1980. State economic and demographic projections are
available for 1975 and 1980; Metropolitan Area projections are also available
for 1975 and 1980.
The basic geographic areas for which the sub-national projections have
been developed are the 50 states and the District of Columbia. In addition,
the state projections have been aggregated into eight major regions: New
England, Middle Atlantic, Great Lakes, Southeast, Plains, Southwest,
Mountains, and Far West. Projections are also available for 224 metropolitan
areas.
At the national level the NPA economic projections include the following
variables: gross national product by 17 major components in current and
constant prices, personal income by nine income sources, personal consumption
expenditures by 83 product categories, distribution of consumer units by eight
income classes, gross sales by 86 industry groups, gross output per worker
for 64 industries, gross and net capital stocks by industry sector for 20 types
of equipment and 10 types of structures, purchases of non-residential
A-31
-------
structures and producers durable equipment by type, government expenditures
by 45 program categories by level of government, government revenues,
merchandise exports and imports by 86 industry groups, and population and
labor force by age, sex and color.
At the regional level NPA projections include the following variables:
population by region and state, employment in 30 industry sectors by region
and state, total and per capita personal income by region and state, total
and per capita consumption expenditures by region and state for 80 consumer
product items. Metropolitan area projections include population, population
density, employment, and total and per capita personal income.
The employment series used in the projections is conceptually equivalent
to a count of the number of full - part-time jobs filled, including both
wage and salary workers and self-employed. Comparable national aggregates
were derived from the industry employment series reported in the Survey of
Current Business by the Bureau of Economic Analysis. Population data
represent mid-year (July 1) resident population including both civilian
population and Armed Forces stationed in a given state. In the regional
series, personal income, personal consumption expenditures, personal taxes,
and personal saving are expressed in constant 1968 dollars. In the national
projections, GNP and national income figures are projected in both 1958
and current dollars.
Employment projections are available for each state at the 2-digit level
of detail for manufacturing and at the 1-digit level for other industries.
Employment projections for the metropolitan areas are provided only at the
1-digit level of detail. Exceptions are that manufacturing is broken out
into durable and non-durable good industries, trade is divided into wholesale
and retail components and civilian government employment is provided separately
for the Federal and state and local sectors.
The data base for the NPA national projections series is that of the
national income and product accounts. State employment data are based on
statistics from Employment and Earnings reports supplemented by a variety
of sources to develop estimates in various sectors not covered fully in
these reports. The state personal income series is that provided by the
Bureau of Economic Analysis of the U. S. Department of Commerce. Basic
A-32
-------
data for the personal consumption projections are developed from the 1960-
1961. Survey of Consumer Expenditures, prepared by the Bureau of Labor
Statistics and Market Profiles of Consumer Products^ issued by the National
Industrial Conference Board in 1966.
4.3 Operating Procedures
The National Planning Association has been engaged in a regular program
of providing national and regional economic and demographic projections for
a number of years. At the present time, both the national and regional and
state projections are revised on an annual basis.
As it is presently operating, the econometric model used by NPA to
provide national economic projections cannot easily incorporate alternative
assumptions. However, efforts are under way to modify the model so that the
effects of alternative assumptions about the course of the future can be
reflected in different projection series. For sensitivity analyses at the
regional level, as with other models it may be more appropriate to take the
relatively aggregated national output as fixed and perform the industrial
and geographic sensitivity calculations within this framework.
4.4 Evaluation for EPA1 s Use
Hie WPA projections provide a set of consistent national and regional
projections of population, personal income, employment and personal consumption
expenditures. National projections of gross output per worker for 64
industries, a measure which may be useful in developing indices of physical
output, are also provided. The National Planning Association is committed
to continuing the projection program, as national and regional projections
have been updated on an annual basis in the past and are developed as part
of an on-going program at NPA.
The primary disadvantage of the NPA projections for EPA's use is that the
appropriate measures are not provided in the necessary detail to project air
emissions. In order to project these emissions, some measure of output by
industrial sector is required, whereas the NPA projections furnish only em-
ployment projections. In addition, particularly for the metropolitan areas
employment projections are provided for industries that are too highly
A-33
-------
aggregated to be of practical use in projecting air emissions. Finally,
at the present time projections are available only for 1980. This time-frame
may not be adequate for the EPA's use.
In summary, althouth the NPA projections are developed on an adequate
geographic basis for use in projecting air emissions and are part of a continuing
program, they are not provided in sufficient industrial detail at the sub-
state level and are therefore not recommended for further consideration by the
GAP.
A-34
-------
5.0 THE OCD PROJECTIONS
5.1 General Characteristics
The objectives of the program under which the economic projections
have been produced were to develop a series of models for evaluating the
vulnerability of national systems in the event of a nuclear attack. The
entire series includes separate models for projecting surviving population
and industrial capacity from the attack, post-attack demand, and the degree
to which sufficient capacity exists to meet the post-attack demands for
consumption and recovery. The discussion in this memorandum will be limited
to the projections of economic activity.
The entire series of models were developed by a variety of contractors
under the sponsorship of the Ofice of Civil Defense. The projections of
economic activity were developed by the Institute for Defense Analyses (IDA).
The data base on which the projections are based was developed primarily by
Jack Faucette Associates.
The projections provided by this system are developed from a series of
computer models. Once the models are placed in operation no manual inter-
vention is made.
5.2 Detailed Characteristics
A series of assumptions are implicitly contained in these projection
methodologies. Of particular importance for the economic projections is the
fact that national projections of economic activity are those provided by
the U. S. Department of Labor as part of the Interagency Growth Project.
These projections have been published as a part of an analyses and examination
of growth prospects of the U. S. economy to the year 1980. Population pro-
jections have been provided by the U. S. Bureau of the Census.
Population projections for 1975 are provided for each of the approximately
3,000 counties in the United States by the Bureau of the Census. From the
available documentation, it is difficult to determine what assumptions
have been input to those projections. In particular, it is difficult to
determine if they are in any way consistent with any independent projections
of employment growth by county.
A-35
-------
The other projections contained in the data base are those for value
added and number of firms. Each of these series is projected as a function
of the population totals in the counties for both 1975 and 1960 (Institute
for Defense Analyses, 1971, pp. III-4 - III-7). The projected values for
each county are then adjusted to the national control totals developed from
the Department of Labor projections discussed above.
Projections are currently available for the year 1975. The overall
project at IDA contains projections of other economic variables, for example
personal consumption expenditures, for alternative years. However, the value
added and number of establishment projections are currently available only
for 1975.
The basic geographic areas for which the economic projections are available
are the approximately 3,000 counties in the United States. Projections of
final demand, personal consumption expenditures, and other aggregates are only
available at the national level.
County projections include population, value added, and number of firms.
Value added is projected in 1958 dollars for each of the sectors of the national
input-output table. The number of firms is projected for four size classes,
only for manufacturing industries: 1-249 employees, 250-499 employees, 500-999
employees, and 1,000 employees and over.
The basic data on which the projections are based and the projections
9/
themselves are available on tape at the Institute for Defense Analyses.— In
addition, a series of summary indicators of the concentration and dispersion
of industrial activity by county have been prepared for each of the sectors.
These indicators are available both in terms of single numerical values and
in terms of empirically fitted curves to the geographic distribution of
industrial output.
j)/The value added data base from which the projections are derived was
developed by Jack Faucette Associates (1968) and are based on the results
of the 1963 Census of Manufactures.
A-36
-------
5.3 Operating Procedures
At the present time, there is no assurance that the projections for 1975
will be updated. However, the Office of Civil Defense, subject to the
availability of funds, plans to incorporate the result of the 1970 Census
of Population into the population data base. These results would then
presumably be used to update the projections of value added and number of
establishments for some time period beyond 1975.
In their present form the data series discussed above does not appear
to be. very amenable for sensitivity analyses. As with other modes, it seems
to be more appropriate to take the relatively aggregated output of these
projections as fixed and perform the industrial and geographic sensitivity
calculations within this framework.
5.4 Evaluation for EPA's Use
The projections surveyed above are not appropriate for use of EPA in
projecting air emissions. First, at the present time they exist for only
1975. Although they are available at the county level of detail, the
approach to their development is not very rigorous. More acceptable pro-
jections are available from other organizations, although they are not necessarily
available at the county level of detail. It appears that the most appropriate
use of the data developed and used by the Office of Civil Defense and the
Institute for Defense Analyses will be in developing the outline of techniques
for projecting output at the county level. The summary statistics of
industrial concentration and dispersion and the empirically fitted curves of
the geographic distribution of industrial output should provide useful in-
puts to this portion of the development of the overall national-regional
model for projecting economic activity.
A-37
-------
6.0 THE BLS PROJECTIONS
The final series of projections reviewed in this memorandum are those
prepared by the Bureau of Labor Statistics as part of a large study of future
patterns of growth in the U. S. economy.— These projections are provided
at the tiational level only and therefore are not appropriate to serve as
derailed inputs to a regional air emission projection methodology. Accordingly,
the projection methodology is not reviewed in as great a detail as those
of the other sections of this memorandum. However, highlights of the
projection methodology are briefly reviewed, in particular to summarize
the fairly detailed set of national industrial projections that have been
developed for 1980 which may provide useful control totals in develop-
ing more detailed regional projections for EPA's regional air emission fore-
casting efforts. In addition, a variety of supplementary data have been
developed in association with the overall effort. Certain of these data
may also provide useful inputs to the EPA air emission projection model.
The BLS estimates of 1980 demand, output and employment are not fore-
casts but projections of what the economy might be like under a given set
of assumptions. Projections for four alternative 1980 models have been
developed. These four are grouped into two sets: one set termed the
basic model and the other the high durable goods model. The basic models
represent what is believed to be the more likely projections for 1980 and
reflect the long-term shift toward" the service producing and away from the
goods producing sectors of the nation s economy. As the description implies,
the high durable goods set emphasizes expenditures on durable goods. Each
of the sets has two models with identical characteristics throughout except
for the unemployment rate which is varied: one of the models in each set
has a three percent unemployment rate and the other has a four percent rate.
10/These projections and a discussion of the procedures employed in their
development are contained in U. S. Department of Labor (1970).
A-38
-------
The 1980 projections are made in a series of distinct but closely
inter-related steps. First, potential GNP is developed based on a pro-
jection of the labor force, assumptions regarding the rate of unemployment
and the level of the Armed Forces, and projections of trends in average
hours and output per man hour. Given the potential GNP, projections are
developed of the composition of GNP among demand components, government,
consumption, business investment and foreign demand. Once the composition
of GNP is determined, the detailed distribution of each of these final
demand components is projected.
In order to translate projections of industry demand into industry
output requirements, input-output relationships which have been projected
to 1980 are used. After the calculation of industry output growth rates
are completed, the final step is to derive the projected level of employ-
ment by industry, by using projections of changes in output per man hour
by industry. Finally, demand generated employment is reconciled with that
resulting from the population and labor force projections.
In the BLS projections, all productive activities are classified into
87 industries. In particular, it should be noted that industrial
detail is available for some sectors for combinations of three- and
four-digit SIC industries.
Although no firm schedules have been developed, it is assumed that
updated national projections using this approach will be periodically
prepared in the future. Thus, a series of national projections will be
available in the future to provide useful inputs into the EPA air emissions
projections methodology.
A-39
-------
REFERENCES
Al-Samarrie, Ahmad and Graham Scott.
1971a. An Econometric Model for Long-Range Projections of the
United States Economy. Report No. 71-N-l. Washington,
D. C.: National Planning Association.
1971b. Revised National Economic Projections to 1980. Report
No. 71-N-2. Washington, D. C.: National Planning Association.
Executive Office of the President, Bureau of the Budget. Standard
1957. Industrial Classification Manual. Washington, D. C.: U. S.
Government Printing Office.
1967. Standard Industrial Classification Manual. Washington, D. C:
U. S. Government Printing Office.
Harris, Curtis C., Jr.
1970. "A Multiregional, Multi-Industry Forecasting Model". The
Regional Science Association Papers. XXV: 169-180.
Harris, Curtis C., Jr. and Martin C. McGuire.
1969. "Planning Techniques for Regional Development Policy". The
Journal of Human Resources. IV(4): 466-490.
Institute for Defense Analyses.
1971. Methodologies for Evaluating the Vulnerability of National
Systems. Volume I, Part I. Draft Working Paper. Arlington,
Virginia.
Jack Faucette Associates.
1968. 1963 Output Measures for Input-Output Sectors by County.
Silver Spring, Maryland.
National Planning Association.
1965. State Projections to 1975. Report No. 65-11. Washington,
D. C.
1970a. Metropolitan Area Growth Patterns for the Coming Decade.
Report No. 70-R-2. Washington, D. C.
1970b. State Economic and Demographic Projections to 1975 and 1980.
Report No. 70-R-l. Washington, D. C.
A-40
-------
Polenske, Karen R.
1970. A Multiregional Input-Output Model for the United States.
Washington, D. C.: U. S. Department of Commerce.
U. S. Department of Commerce, Office of Business Economics, 'Regional
Economics Division.
1970. Economic Projections for Air Quality Control Regions.
Washington, D. C.
U. S. Department of Commerce and U. S. Department of Agriculture.
1971. Economic Activity in the United States by Water Resources
Regions and Subareas Historical and Projected 1929-2020.
Washington, D. C.: United States Water Resources Council.
U. S. Department of Labor, Bureau of Labor Statistics.
1970. Patterns of U. S. Economic Growth. Bulletin 1672. Washington,
D. C.: U. S. Government Printing Office.
A-41
-------
APPENDIX B
IRON AND STEEL INDUSTRY
1.0 INTRODUCTION
The iron and steel industry (SIC 3312, Blast Furnaces and Steel
Mills) is a substantial emitter of particulates from the sintering
operation, the coking ovens, the steel making furnaces, and the scarfing
operations. There are good data on the capacity of the industry, the
production processes, and the controls. The major tasks in projecting
emissions on a regional basis are predicting the future output of steel
and allocating this on a regional basis.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
Output in the steel industry is dependent upon the demand for steel
and imports. The demand for steel products is probably easier to project
than the supply from domestic sources. Demand is a function of population,
income, and the price and competitive advantages of the product compared
with other materials such as aluminum, glass, wood, and plastics. The
domestic supply is dependent upon the demand for the product, prevailing
prices, the effect of import quotas, and recycling.
2.1.1 Demand
A very detailed study of the demand for steel mill products
has been published by the Bureau of Mines.— This study projects the demand
for the various types of steel mill products. The demand is dependent upon
such factors as population, income, the prices of steel products and
competing materials, the composition of total demand in 1980, and the
growth of industrial uses of steel. Although this study can be used as
the basis for making projections of the demand for steel mill products,
special care must be taken to insure that the projections are consistent
with the projection of demand for aluminum and other materials that will be
made in this study.
_!/ Mo, William Y. and King-Lee Wang. A Quantitative Economic Analysis
and Long Run Projections of the Demand for Steel Mill Products. Bureau of
Mines Information Circular 8451. Washington, D. C.: Government Printing
Office 1970.
B-l
-------
2.1.2 Supply
The most critical factor in projecting supply, once demand
has been determined, is imports. Prices for imported steel will have to
be compared with the expected prices for domestic steel in order to
determine the percentage of steel demand that will be supplied by imports.
Because all steel mills do not produce the same mix of products, the
composition of imports will have an influence on the location of domestic
steel production. The competitive position of each of the major steel mill
products will have to be examined in some detail.
Another important influence on the supply of steel mill products is
recycling. Recycled steel is a good substitute for virgin steel in most
uses. Any increases in the amount of steel recycled will result in
decreases in the amount of primary steel produced, assuming demand is
constant. The current effort to recycle all types of scrap materials
along with legislative changes to make secondary materials more competitive
will undoubtedly have a significant influence on the amount of steel
recycled. Handicaps to the use of recycled steel such as discriminatory
freight rates, depletion allowances for virgin materials, and purchase
specifications will probably be removed.
Current trends in production capacity for various types of steel mill
products will have to be examined as well as industry projections of this
capacity. Growth in particular types of capacity is probably a good
indicator that the domestic production of these steel mill products will
continue to be healthy.
2.2 Location
2.2.1 Data
Data are available on the current location of steel mills, on
the mills that are growing and contracting, on the processes used, and
on the control devices employed. Most of these data have already been
computerized, although data on coking ovens and their emissions still have
to be added.
2.2.2 Regional Shares of Production
Total output of the domestic steel industry needs to be allocated
among existing plants and among possible new plants. The exact location of
new plants cannot be known. In the case of steel mills, however, the
possibility of the new mills within the next five to ten years can be
learned from examination of trade sources. New mills beyond this time
period can only be hypothesized on the basis of locational factors such as
B-2
-------
access to markets and raw materials. The possible sites can be included
in the projection model. That is, the model can be run under the hypo-
thesis of what will happen to emissions if a steel mill is located in this
or that very likely place.
The allocation of production among existing plants is the primary
locational problem in the iron and steel industry. There are data available
on the production levels of the various steel mills that permit projections
of output at each of these mills. Output by major product will also be
necessary. Some factors that influence production of particular products
at a particular steel mill are the demand, the efficiency of the production
equipment, the access to markets, and the competition from imports. These
various factors will have to be examined in projecting output from each
steel mill.
2.3 Process ,
Although there are currently three types of steel making processes in
use, the projection of output by process should not be a major problem. Data
are currently available on production processes by plant. Furthermore,
there is a very distinct trend away from open hearth furnaces.
The trend away from sintering toward pelletizing will also probably
continue. Projection of this trend and its favorable effects on emissions
should not present any problems.
3.0 SUMMARY
The major tasks in projecting emissions from the iron and steel
industry are the projection of the demand for and the supply of steel mill
products and the allocation of the domestic production among regions. Pro-
jections of the demand for steel mill products can be made readily; projec-
tions of the supply of these products will be more difficult because of
imports and recycling. The allocation of domestic supplies of steel mill
products on a regional basis will be a time consuming, though not par-
ticularly difficult task. Projection of trends of various steel mill
products from each of the steel mills will provide a good guide to the
regional allocation.
B-3
-------
APPENDIX C
PRIMARY COPPER
1.0 INTRODUCTION .
The primary copper industry, SIC 3331, consists of the smelting and
refining of copper ore. Copper smelters are the largest point sources of
sulfur oxide emissions. The relatively small number of smelters is well
documented as to location, production, capacity, -processes, and controls.
The major tasks in projecting emissions; from the copper industry will be
to estimate the supply of copper from domestic sources and the location
of these sources. '-
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The total demand for copper, has been.projected by the Bureau
of Mines. More detailed projections may be necessary in order to insure
that the projections are compatible with the projection of competing
products. Copper can substitute in various uses for aluminum, steel,
plastics, and to a degree, zinc. The primary competition, however, is
with aluminum. Other factors that influence the demand for copper are the
price of the product, the population, the income levels and the increased
uses to which copper is put. .
2.1.2 Supply
The supply of domestic copper is due to the demand for copper,
the supply of imported copper ore and.finished copper, and the amount of
copper that is recycled. The total supply, of copper, of course, is deter-
mined in part by the interactions of price with the demand for copper.
Imports of copper and copper ore are dependent upon the current demand
•. ' .. v-
for copper and the capatibility of the domestic copper industry to supply
that demand. The domestic price for, copper.varies from the world price;
it may be higher, though in recent years it has often been substantially
lower. Projections of copper imports will require ,a detailed study of the
C-l
-------
comparative advantages of -domestic and imported copper as well as the
regulations and legal restirctions on copper imports. The comparative
advantage of domestic producers may change due to the need to install
emission control equipment. Preliminary studies indicate that these con-
trols, while requiring large expenditures of funds, are not likely to
affect the price of copper by more than a few percentage points.
Another factor that has a large influence on the supply of domestic
copper is the amount of copper that is recycled. The high value per
pound of copper is responsible for the large percentage of copper that is
recycled. Furthermore, much of the copper is contained in items such as
automobile radiators and electric power equipment that can readily be
collected and recycled. New regulations and laws affecting the secondary
materials market may have an effect on the amount of copper that is
recycled, and therefore, on the amount of primary copper that is produced.
2.2 Location
Data are available on the location of every copper smelter and on
production levels at these smelters. For projections it will be necessary
to determine the allocation of future copper production among existing
smelters and possible new smelters. Any new smelters likely to be con-
structed within the next five years or so can probably be determined from
industry sources. New smelters beyond this point in time will depend upon
current supply and demand relationships. The most likely location of new
copper smelters can probably be determined through industry sources and
Bureau of Mines data.
Current production rates at existing smelters should prove a guide
to future regional shares of production. Trends in the share of production
among smelters exist because certain smelters have competitive advantages
over others. That is, their equipment is newer, more efficient, or they
are located near more desirable inputs or closer to markets. These trends
in output at each smelter can be projected based on past data and expected'
changes in the future.
3.0 SUMMARY
The major task in this project is to project the domestic supply of
copper. The demand also has to be projected but this can be based largely
on projections made by the Bureau of the Mines. The domestic supply is
C-2
-------
dependent on imports and upon recycling activities. The secondary task
is to allocate production among the existing smelters and to hypothesize
upon the existence of new smelters. The most likely locations for new
smelters can be inserted into the projection model in order to make
projections conditional upon the establishment of the new smelter. No
new data are required for primary cppper.
C-3
-------
APPENDIX D
PRIMARY LEAD .
1.0 INTRODUCTION
The primary lead industry, SIC 3312, consists of the smelting and
refining of lead. The six lead smelters are an important source of sulfur
oxides. Detailed information is available on the capacity, output, pro-
duction processes, and controls of the existing lead smelters. The primary
task will be to project the demands for and the supply of lead and to
allocate this supply among the existing smelters.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand ,
The demand for primary lead has been growing slowly due to
the increasing life of automobile batteries and the recycling of almost
all batteries. The other large use of lead, tetraethyl lead in gasoline,
is declining and may soon disappear due to restrictions on the lead content
of gasoline. The projected demand for lead will have to take these two
factors into consideration as well as other uses for lead. Projections
of the demands for lead have been made by the Bureau of Mines, but these
will have to be examined very closely for their assumptions.
2.1.2 Supply
The domestic supply of lead depends upon demand, imports, and
the secondary lead supply. Imports have shifted over time as prices change
and must be examined very closely. It is likely that imports will cease
if the demand for lead declines. The other important supply factor is the
secondary lead market. About 94 percent of all batteries are recycled and
the lead from these batteries is a perfect substitute for new primary lead.
Because such a high percentage of the recycleable lead is already recycled,
lead imports will probably be the major hurdle to calculating the domestic
supply of lead.
2.2 Location
The location and output of all the lead smelters is known. Trends in
production are also known and can be projected using past data as well as
other factors that determine output. Some other factors are access to raw
D-l
-------
materials and to markets, wage rates, and efficiency of production. Using
these factors, output can be allocated to each of the smelters. The
critical point in allocating output is deciding whether some of these
smelters will close in the future. The possibility of closures can be
examined through industry sources as well as by an examination of the
factors that influence the allocation of shares among smelters.
3.0 SUMMARY
The major task in this industry is to project the demand for lead.
The critical assumptions have to do with the use of tetraethyl lead in
gasoline. The supply of domestic primary lead also has to be projected;
it will depend upon the level of imports and the percentage of lead that
is recycled. The critical point in allocating domestic lead production
among the primary producers is whether any of them will close during the
projection period. Current data and trade journals should provide
sufficient information to project emissions in this industry.
D-2
-------
APPENDIX E
PRIMARY ZINC
1.0 INTRODUCTION
The primary zinc industry, SIC 3333, includes the smelting and
refining of zinc from the ore. The fifteen zinc smelters around the
country are important sources of sulfur oxide emissions. The tasks are
to project the demand for zinc, the domestic supply, and the location
of production. The possibility of closures also has to be considered.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for zinc has been projected by the Bureau of Mines.
*
The demand is dependent upon such things as population, income, and the
growth of end-uses of zinc. There is some substitution for zinc by other
materials such as steel, copper, aluminum, and plastics. The primary
consideration in adopting the projections of the Bureau of Mines is to
ensure that they are consistent with the projections for the other primary
and secondary metals.
2.1.2 Supply
The supply of zinc is dependent upon the demand and upon imports.
.'
Zinc imports are not substantial at the present time. Very little zinc is
recycled because it is used primarily as an alloy in other metals and
recovery is difficult. Trends in domestic zinc production and their impact
on imports will have to be examined in order to determine the expected level
of imports in the future.
2.2 Location
The primary problem of location will be to allocate the production of
zinc among the existing zinc smelters. A secondary consideration will be
to consider the possibility of closure of one or more zinc smelters. An
examination of industry sources as well as the usual measures of economic
performance will provide some guides to the possibility of closure.
E-l
-------
2.3 Production Process
Although various processes are used in zinc smelting, sufficient
data are available to permit projection of trends without great difficulty.
There are also data on the various types of controls.
3.0 SUMMARY
The primary task in projecting emissions from the primary zinc
industry is to project the demand for zinc. A secondary task is to
project the domestic supply of zinc and to allocate it among existing
plants. The major uncertainty will be the possible closure of one or
more zinc smelters. There are no special data requirements for projecting
this industry. The level of effort required for projecting this industry
will probably be substantially less than for the other primary metals.
E-2
-------
APPENDIX F
PRIMARY ALUMINUM
1.0 INTRODUCTION
The primary aluminum industry, SIC 3334, is a significant source
of particulate emissions. There are presently 25 establishments located
mainly in the South and West where cheap electricity is available.
Aluminum production is growing rapidly compared with the other primary
metals, in part because the industry is aggressively seeking out new
uses. The major tasks in this industry will be to project the demand
and supply of aluminum and to allocate production among existing sources.
The location of possible new sources will be hypothesized.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The production of aluminum has grown rapidly during the
last twenty years due to increased demand for the material resulting
from its substitution for other metals. For example, aluminum competes
with steel in engine blocks for automobiles and with copper for radiators
in automobiles. There is also severe competition between aluminum and
copper for electrical transmission lines. Some of the competition is
based on the superior technical characteristics of aluminum while the
other is based almost purely on price. Aluminum prices have been quite
stable and this has permitted the metal to capture a larger share of the
market when the prices of other metals were rising.
Although projections of aluminum have been made by the Bureau of
Mines using relatively unsophisticated techniques, a more detailed study
is probably called for. The demand for aluminum depends upon its price
and the prices of competing metals. It also depends on population,
income, and the growth of the various end-products in which aluminum
is used. Some of these end-products are electrical transmission lines,
automobile radiators, packaging products, and various construction
materials such as structural shapes and siding. There are also small
F-l
-------
amounts of aluminum exported and these exports will have to be
explicitly considered.
2.1.2 Supply
The supply of aluminum from domestic sources depends upon
the demand, imports, and recycling. Aluminum capacity has been keeping
pace with the growth of demand. Production costs in comparison with
foreign sources of aluminum could be a factor leading to an increase
in imports. Imports occur when they offer a price advantage over domestic
aluminum or when domestic aluminum is not able to supply total demand.
In general, imports have not been a significant part of the aluminum
market, but this may change due to rapidly increasing electricity
prices in the United States and a concern for environmental quality.
Much of the bauxite used by the aluminum industry is imported and
could be smelted outside the United States.
A second factor influencing the domestic supply of aluminum is
the amount of recycled aluminum available. There has always been an
effort to recycle aluminum because of the high value per ton of the
metal. A recent effort has concentrated on recycling aluminum beverage
containers which are an important part of total output and are available
for recycling as soon as the beverage is consumed. Because aluminum
production has grown rapidly, the amount of aluminum being released
from long-term uses is relatively small.
2.2 Location
The primary determinant of location in the aluminum smelting industry
is the availability of cheap sources of electricity. Secondary considera-
tions are access to raw materials and access to markets. The number of
establishments has been increasing and is likely to do so in the future.
Because raw materials and market location are not very good indicators
of aluminum smelting locations, new establishments spring up in unexpected
places. It may not be possible, therefore, to predict with any degree
of certainty, where new aluminum smelters will occur. The more likely
locations can be hypothesized and tested for their emission implications.
The major part of this task will be allocation of production among
existing sources. This can be done using the knowledge of plant capacity
and patterns of growth among the plants. There is a tendency to build
F-2
-------
plants with the flexibility to increase capacity and this tendency must
be taken into consideration when allocating the production of aluminum
among domestic sources. An examination of trade data sources and of
electricity supply capacity near existing smelters should provide
guidance to future production levels.
2.3 Production Process
The production processes in use are known at all but two plants.
The trends in the production process are also known. It is unlikely
that production processes and controls will present any great
difficulties in making projections.
3.0 SUMMARY
The primary task in projecting emissions from the aluminum industry
will be to project the demand for the product and the percentage of
demand that will be supplied from domestic sources. The secondary
task will be to allocate production among existing sources. Some
attention should be given to the most likely locations for new
aluminum smelters.
F-3
-------
APPENDIX G
PETROLEUM REFINING
1.0 INTRODUCTION
Petroleum refining and the storage of crude and refined products,
SIC 2091, is a source of particles, sulfur oxides, carbon monoxide, and
hydrocarbons. The industry is growing and its activities are located in
a large number of places throughout the nation. Emissions depend upon the
process type used.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for petroleum products has been growing rapidly.
Some of the major uses are as fuel for vehicles, electric power generation
plants, and other types of combustion. The demand for petroleum can be
related to the growth of the various uses of petroleum products. These in
turn are related to the population and income, to the patterns of vehicle
ownership, and to the production of electrical power from various types
of fuel. The prices of competing fuels are also an important determinant
of the demand for petroleum. Good projections of industry output already
exist for the two-digit industry SIC identifier of which petroleum refining
is a part.
2.1.2 Supply
The supply of petroleum products is more difficult to project
than demand. The source of the crude petroleum is important. The sulfur
content and other characteristics of the crude vary between domestic and
foreign sources, and these characteristics affect the emissions. Present
levels of domestic crude production will have to be projected in order to
determine the amount of foreign crude. Recycling will not be a major
problem because most petroleum is burned and thus cannot be recycled. Off-
shore refining of foreign crude is something that has to be considered as
a possibility of the future. Another consideration will be the possible
conversion of coal or other fuels into petroleum. This would have an
obvious effect on the domestic supply.
G-l
-------
2.2 Location
Petroleum refineries are located in areas of crude petroleum pro-
duction, near seaports, or near population centers served by a pipeline.
There will be a shrinkage of the share of refineries near domestic sources
of petroleum because of increased imports. These imports generally go
directly to the points of major usage. About half of all refineries are
located in Texas, Louisiana, and Ohio. California, Pennsylvania, and
Illinois account for another quarter of the production.
The shares of domestic production may be allocated among existing
refineries on the basis of capacity, crude petroleum supplies, access to
markets, and trends in production. New refineries will have to be predicted
because of the growing significance of the imported crude petroleum.
There will probably be new refineries along the East Coast, but the exact
location can not be known with certainty. The most likely locations can
be hypothesized and the model operated under these hypotheses.
2.3 Process
The major distinguishing characteristic among refineries is the
presence or absence of a fluid catalytic cracker. The catalytic cracker
requires a catalyst regenerator which produces carbon monoxide and other
emissions. Because the emissions from refineries do vary with the particular
process used, process information must be collected and projected.
3.0 SUMMARY
The most critical step in projecting emissions from petroleum
refineries is to project the domestic supply of crude petroleum. Once
this is done, imports can be calculated and production allocated among
the various refineries. Imports are significant because they shift produc-
tion away from refineries in areas of crude petroleum sources toward
population centers. Another aspect that requires attention is the
production process used at each refinery.
G-2
-------
APPENDIX H
SECONDARY NONFERROUS PETALS
1.0 INTRODUCTION
The secondary nonferrous metals industry, SIC 3341, is a source of
particulate emissions. Emissions from any one establishment are rela-
tively small by comparison with most other point sources, but the large
number of plants makes the industry as a whole a significant emission
source.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for most secondary nonferrous metals parallels
the demand for the primary metals; however, the demand for nonferrous
metals such as secondary gold, silver, platinum and various alloys is
independent. Only the secondary production of copper, lead, zinc,
and aluminum will be considered because emission factors for these are
available.
2.1.2 Supply
The supply or output of the secondary nonferrous metal
industry depends on a number of different factors. The cost of
obtaining the supply of raw materials (scrap metal) together with
the price that can be obtained for the finished product is an obvious
determinant of output. Another factor is the foreign market for scrap
metals. In many cases it is more profitable to export the scrap than
to melt it and produce secondary metals that can substitute for the
primary metal. Another factor is the possible changes in the regulations
to encourage the use of scrap metals and to make transportation costs
for scrap metals equitable. These factors taken together make projections
of the supply of secondary nonferrous metals a difficult task.
2.2 Location
There are probably between 400 and 600 establishments that produce
secondary nonferrous metals. These are located near population centers
H-l
-------
where scrap metal is generated. Information on the location of the
establishments is available from various sources but is not very
reliable. Almost no information is available on output, process, or
control equipment.
Total production in this industry can be allocated to the various
regions on some sort of shift and share basis. The population of each
region, the potential supply of scrap metal, and the access to markets
for this metal are all important considerations. Because the industry
comprises many small establishments, the new plant problem is a very
severe one. The best approach is probably to project output on a
regional basis rather than on an establishment basis. If an area
does not presently have an establishment and it appears likely that
it might in the future, then the model can be run under this assumption.
3.0 SUMMARY
The major task in projecting emissions from the secondary nonferrous
metals industry is to project the domestic supply of the output. The
location of production can be done using rather crude shift and share
analysis. Little data are available on production by region and
because the emissions from any one source are so small detailed analysis
is difficult and probably not warranted.
H-2
-------
APPENDIX I
THE PULP AND PAPER INDUSTRY
1.0 INTRODUCTION
The pulp and paper industry is an important source of particulate
emissions. Pulp mills, the primary polluter, may be found in one of at
least three different SIC classifications. Detailed information is not
available on the production processes at all pulp mills. Not as much
data are available on sulfite process mills as on sulfate process mills
because they have not been examined in previous studies of air pollution.
Nevertheless, they are substantial sources of emissions. A complicating
factor in projecting emissions from the paper and pulp industry is the
influence of recycled paper. It appears likely that much larger percentages
of paper will be recycled in the future.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for paper products can be forecast readily on the
basis of growth in population and GNP. The demand for pulp, of course, is
dependent upon the demand for paper. The only factor that requires special
treatment are exports, which are fairly substantial in the area of Kraft
paper.
2.1.2 Supply
The supply of pulp and paper, in general, will be equal to the
demand for it, as there are very small imports. There are some imports
across the Canadian border, but these are very small and can be included
in the projection methodology. The primary problem of supply is to
project the amount of paper that will be recycled. The percentage has been
declining over time, but it appears likely that the trend will be reversed
due to various economic and social pressures. Recycled paper production
does not cause the same emissions as virgin paper production.
2.2 Location
The location of pulp and paper mills depends upon the availability of
water, land, and trees. Transportation and access to market are also
important considerations. Existing plants will probably produce much of
1-1
-------
the future supply of paper pulp, but new mills undoubtedly will be
established. It should be possible to select broad regions where paper
mills will be established, but selecting the particular county in the
region may be difficult.
The allocation of production among existing plants can be based on
their capacities and trends in production. The investment cycle in the
pulp and paper industry is such that there are often large amounts of
excess capacity. Therefore, it is likely that the plants with excess
capacity can absorb the demands for increased production in the future.
The various relevant location factors will have to be studied in order
to allocate production among plants and to estimate where new plants will
be located.
2.3 Process
Both the sulfate and sulfite processes create emissions; the emissions
from the sulfate process are more significant. The mechanical process does
not generate much in the way of emissions. Data are not available on all
the details of production process at each mill and will have to be obtained.
Additional problems in projecting emissions arise from the presence
or absence of lime plants at or near paper mills. Although these lime
plants will be considered under the lime industry, their location may be
determined during the study of the paper industry. A further problem is
that some paper mills generate their own power and emissions from this
activity must be considered.
3.0 SUMMARY
There are several substantial tasks to be accomplished in projecting
emissions from the paper and pulp industry. Data will have to be obtained
on production process at each mill, particularly for mills that employ the
sulfate process. In addition, total supply of pulp and paper will require
in-depth study in order to determine the influence of recycling larger
percentages of paper. Production can be allocated among existing plants
with various techniques, but new plants also have to be considered. Their
locations can be predicted down to relatively small regions of perhaps
fifteen counties, but the exact location within the region is impossible to
predict with any degree of confidence. The projection model can be
operated with different combinations of likely locations.
1-2
-------
APPENDIX J
PHOSPHATE FERTILIZER INDUSTRY
1.0 INTRODUCTION
The phosphate fertilizer industry, found in SIC 2871 and SIC 2819,
is a substantial emitter of particulate and fluoride pollution. Most
of the particles are emitted from grinding, crushing, and drying of the
phosphate rock, which can be handled separately. The phosphate
fertilizer industry is a source only of fluorides, which is not a
set one or set two pollutant and thus the industry could be dropped
from consideration as a point source.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for phosphate fertilizer is dependent upon the
output of agricultural products. Trends in the relationship between
phosphate fertilizer and agricultural output can be projected into
the future. The demand should be expressed in terms of P^O equivalent
which is the actual nutrient. Fertilizer composition has been changing
over time towards a more concentrated form in order to reduce the shipping
charges.
2.1.2 Supply
Almost all phosphate fertilizer is produced from domestic
sources. In fact there is some small amount of exports that have to
be included. Supply can be projected once demand is known. The
only consideration might be the particular form that the fertilizer,
i.e. normal superphosphate, triple superphosphate or ammonium
superphosphate.
2.2 Location
Total supply has to be distributed on a regional basis. The
basic phosphate fertilizer production is found near sources of
phosphate rock in Florida and Texas. Production of mixed fertilizers
J-l
-------
is concentrated near agricultural areas in different parts of the
country. Growth and output can be allocated among existing plants on a
shift and share basis. The most likely locations for new establishments
can be predicted.
2.3 Process
The amount of fluorides emitted varies with the type of process
used. If emissions in the phosphate fertilizer industry are to be
projected, additional data will be required on the production process
used at each establishment.
3.0 SUMMARY
The phosphate fertilizer industry is a source only of fluorides,
if phosphate rock dust a source of particulates is omitted from
consideration. Additional data will be necessary on production process
as emissions vary substantially depending on the process. The demand
for and supply of phosphate fertilizer probably can be calculated quite
readily on the basis of projections of agricultural output. The
allocation of output from existing establishments can be done rather
easily once the production process used at each establishment is
known.
J-2
-------
APPENDIX K
SULFURIC ACID
1.0 INTRODUCTION
The sulfuric acid industry, SIC 28193, is an emitter of particles
and sulfur oxides. Control practices within the industry are changing
and these changes will accelerate in the future. Sulfur recovered
through control can be used in the acid making process. There are
several surveys in progress on sulfuric acid markets, though nothing
definitive has been completed yet. The major task will be to clarify
the supply and demand situation, particularly the supply situation.
Once this has been done allocation on a regional basis will follow
almost automatically.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
Sulfuric acid has been an important low priced inorganic
acid used in many industrial processes. The demand derives from the
level of industrial activity. On the basis of past trends and knowledge
of the various uses for sulfuric acid, projections of demand can probably
be made quite easily. It is likely that the price elasticity of demand
for sulfuric acid is quite low.
2.1.2 Supply
The supply situation in the sulfuric acid market is chaotic
at the present time. Emission controls have resulted in large amounts
of sulfur being recovered from emissions from combustion, smelting,
refining, etc. This sulfur can easily be convered to sulfuric acid
and in fact, much of the sulfur is removed in the form of sulfuric acid.
The sulfur potentially available from copper smelters would be more
than sufficient to overwhelm the sulfuric acid market in the West.
Sulfur from refineries could flood the gulf coast market for sulfuric
acid. East coast markets probably could be served with the by-product
sulfur from refineries and electric power generation plants.
K-l
-------
In any case, the supply of sulfuric acid will be more than adequate
to meet the demand for it. But whether it will be produced from present
sulfuric acid plants or as a by-product is not currently known.
Sulfuric acid production as a by-product would not produce substantial
emissions. If by-product sulfuric acid is the primary source of supply,
the industry can probably be eliminated as a point source of pollution.
2.2 Location
The supply problem with respect to sulfuric acid is directly related
to location. If by-product sulfuric acid supplies the market, the
location of sulfuric acid production will be substantially different
from what it is today. If, for some reason, by-product sulfuric acid
is more expensive or in the wrong location to be competitive with
primary sulfuric acid, production can be allocated among existing plants
on a regional basis through shift and share analysis. There is almost
complete plant data available though it can be improved. Production
process information will not be required because variations within
plants are greater than the variation among plants.
3.0 SUMMARY
The major task in projecting emissions from sulfuric acid will be
to determine how it will be supplied in the future. If it will be
supplied through by-products, then a detailed study will be necessary
to determine which sources of sulfuric acid will be used. Given the
uncertainty at the present time, projection of this industry will be
time consuming and not very conclusive. Perhaps the best course of
action may be to simply project current output at a steady rate until
other studies underway are completed.
K-2
-------
APPENDIX L
RUBBER TIRES
1.0 INTRODUCTION
The rubber tire and inner tube industry, SIC 3011, is a minor
emitter of particles. ' The major emission is carbon black, a valuable
material that the manufacturers try to capture. The newer plants are
successful in eliminating virtually all emissions and the older ones are
already controlling emissions at a high level.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for rubber tires is a function of the number of
vehicles in use, the type of tires used and the number of miles traveled
per vehicle. The latter factor has not changed very much over time. There
have been increases in the mileage per tire due to improvements in the
tire body and the tread rubber.
2.1.2 Supply
The supply of tires from domestic sources may be more difficult
to project than the demand because of the influence of imports. Imported
tires have been increasing as a percentage of total supply, although the
percentage is still very small. Furthermore, as imports of a particular
type tire increase, there is a tendency to establish production facilities
in the U.S.
2.2 Location
Tire production is still concentrated in Ohio due to the historical
accident of the first rubber production occurring there. New plants have
been established in other areas of the country, particularly in the South
where wage rates and other economic conditions are favorable to the
efficient production of tires. In general, the new plants that have been
established and that will be established do not produce emissions or
pollutants. Therefore the only task will be to determine what portion of
output will be produced in existing plants. This can probably be done on
the basis of projecting current trends.
L-l
-------
3.0 SUMMARY
There is some question whether rubber tire production should be
considered as a point source of air emissions because of the small
quantity of emissions and declining trends in these emissions. If it is
projected as a point source, there are two tasks that can be accomplished
very easily. One is to project the demand for rubber tires. Consideration
will have to be given to the type of tire in use and to the percentage
of imports. The other task is to allocate output among existing plants,
particularly among the older plants that are more likely to emit pollutants.
L-2
-------
APPENDIX M
COAL CLEANING
1.0 INTRODUCTION
Coal cleaning is an important source of particles that are emitted
during the drying process. About 65 percent of all coal is cleaned
and this percentage has been increasing. It will be necessary to
determine where the coal cleaning plants are located, their production
levels, and the process used.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
Coal cleaning is a process for removing dirt, clay, rock,
shale, sulfur compounds and other materials from raw mine-run coal.
Cleaning can improve the quality of the coal by increasing the BTU
value, by reducing the ash content, and by reducing the shipping costs.
Coal cleaning has been increasing over time due to the increasing costs
of transportation. At present about 65 percent of all coal is cleaned.
This percentage is likely to increase, at a rapid rate, if coal cleaning
proves effective in reducing the sulfur content of the coal. However,
current opinion is that coal cleaning is not very effective in reducing
the sulfur content and so the trend toward coal cleaning will probably
continue at about the same rate.
The demand for coal cleaning is dependent upon the total demand
for coal and the price of transportation relative to the price of
coal. Ash disposal problems also influence this demand. Examination
and extrapolation of present trends probably will be adequate to project
the percentage and quantities of coal that will be cleaned in the future.
2.1.2 Supply
The supply and demand for coal cleaning has to be considered
as a single problem. The demand for cleaning depends upon the cost at
which coal can be cleaned and the cost of transportation. In general,
coal cleaning will be demanded so long as the coal can be cleaned at
M-l
-------
a lower cost than the cost of transporting, and later disposing of,
dirt and other undesirable materials that are mixed with the coal.
The percentage of coal that is exported may also affect the percentage
that is cleaned.
2.2 Location
The primary task in projecting emissions from this industry will
be to determine the location, production level and process used at
present coal cleaning plants. The Bureau of Mines presents aggregated
data, but does not break it down on a regional basis. Once present and
past data have been collected, it should be possible to project the
regional distribution of future output on the basis of shift and share
analysis. Detailed information may be required on the production rates
of coal at the various coal cleaning plants to determine whether the
present trends of regional output will be maintained.
The most likely location of new coal cleaning plants will have to
be identified. The projection model can be run using these most likely
locations in order to see the effects on regional emissions of air
pollution.
2.3 Process
Emissions from the cleaning operation depend upon the type of process
used. The two methods are the wet wash and the dry wash. In the wet
wash, the coal subsequently is dried by either flash dryers or fluidized
bed dryers.. Emissions vary with the type of dryers as well as the type
of wash. The process types in operation at existing facilities are not
known. Substantial effort will have to be devoted to obtaining these
data.
3.0 SUMMARY
The major task in projecting emissions from the coal cleaning industry
will be to obtain data on the present location, production level, and
processes used by existing coal cleaning plants. A secondary task will
be to project the percentage or quantities of coal that will be cleaned
in the future. Once these two tasks are accomplished, the projected
M-2
-------
output can be allocated to existing plants on the basis of capacity
at these plants, trends in production, and other considerations. Likely
locations for new plants can be hypothesized.
M-3
-------
APPENDIX N
FEED AND GRAIN INDUSTRY
1.0 INTRODUCTION
The feed and grain industry is a major source of particulate
emissions. The type and quantity of emissions is a function of the
elevator type and the particular grain being handled. There are few
data available on country elevators and small, part-time feed mills.
The main problem, however, is with the terminal elevators and the
large feed producers for which information is available.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for output from the feed and grain industry is
a function of agricultural output. All grain production moves through
grain elevators regardless of immediate demand for the product. The
grain may be stored for export or future domestic uses. Feed production
on the other hand, is directly related to the immediate demand for animal
products, most of which is consumed within the United States.
The demand for grain storage and processing can probably best be
related to projections of agricultural output made by the Federal
government. Similarly, these projections can also be related to the
projected demand for animal products.
2.1.1 Supply
The supply of output from the grain and feed industry is
essentially equal to the demand for their services. Because of data
requirements, the most reasonable approach may be to handle country
elevators, and terminal elevators and large feed mills separately.
The country elevators and small feed mills are located around the
country, very little infomration is known about them, and they operate
on an intermittent basis. The terminal elevators and large feed mills
operate on a much more continuous basis, data are available on them,
N-l
-------
and they are likely to be more significant sources of emissions because
of their size and location in populated areas.
2.2 Location
Country elevators are spread evenly across grain producing areas.
Terminal elevators on the other hand, are located at major shipping
points and near large markets. This pattern and, in fact, specific
locations should not change. The locational problem therefore, will
be to allocate projected output among existing sources. This can be
done on the basis of present trends and projections of regional output
made by the Department of Agriculture.
3.0 SUMMARY
Projecting emissions from all country elevators and small feed
mills would be a difficult and probably unsatisfactory task. The
country elevators operate on an intermittent basis and only affect
the area directly around them. Because data are not already available,
effort to project this group of sources is probably not worthwhile.
Efforts should be concentrated on projecting output for the terminal
elevators and the large feed mills. Total output can probably be based
on projections made by the Department of Agriculture. This output then
has to be allocated among various existing establishments. The degree
of control at these various establishments will have to be determined
in order to project emissions.
N-2
-------
APPENDIX 0
ASPHALT BATCHING
1.0 INTRODUCTION
The asphalt batching industry consists of a large number of esta-
blishments producing liquid asphalt and tar paving materials. These plants
are an important source of particulate emission. The biggest problem in
projecting emissions is the lack of detailed data about plant locations,
production levels, and emission controls in use.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for asphalt batching is based on the construction of
roads, the paving of parking lots and driveways, and the maintenance of
roads and other paved areas. Although total road construction and mainte^-
nance can probably be projected with a fair degree of confidence, there is
uncertainty about the location and whether the paving will be done with
cement or asphalt. There have been some shortages of asphalt, which is a
residual of oil refining, because of the increased demand for more volatile
forms of petroleum. This shortage may eventually permit cement to capture
a larger share of the paving market.
2.1.2 Supply
The supply of asphalt batching will generally be equal to the
demand except where supplies of the basic material are scarce due to
increased demand for other petroleum refinery products. There are more
than 1,000 batching plants located in all parts of the country. Whether
large or small, the plants are generally located to serve their immediate
area. However, there are in addition, many mobile plants that move around
to wherever large construction jobs are underway.
2.2 Location
The data problems on the location and production levels of asphalt
plants have already been mentioned. When total output has been projected
the output can be allocated on a regional basis through the use of shift and
0-1
-------
share analysis of the data that are available. If the data are not
available, allocation of the output on the basis of population and
economic activity may be preferable as an alternative to collecting
new data.
3.0 SUMMARY
The asphalt batching industry is a substantial emitter of particles
and is located in all parts of the country. There is a lack of data on
the industry and therefore projections on a regional basis are difficult
at best. A practical solution to projecting these emissions may be to
allocate them on the basis of population and economic activity. If
this course of action is chosen, the critical task will be to estimate
total output in the industry on the basis of demand for asphalt paving
and quantity that can be supplied. Asphalt may be replaced by cement
for some uses because of future shortages.
0-2
-------
APPENDIX P
CEMENT
1.0 INTRODUCTION
The cement industry is a very substantial source of particulate
emissions and is found in all parts of the United States. The industry
is localized because of the low value to bulk ratio, but there are changes
underway in distribution patterns. Forecasting the distribution patterns
will be the most difficult part of this project. Present location of
plants, levels of production, trends in output, production process and
controls are all available.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for cement is a function of construction activity.
It is likely that available projections of construction activity on a
national and regional basis can be used as a means of projecting the output
of cement. It may be necessary to modify these projections in order to take
into account changes in the relative positions of asphalt and cement for .
paving purposes. There are some exports of cement and a prediction of these
exports is necessary.
2.1.2 Supply
Most of the demand for cement is met by domestic production.
There is some small amount of cement coming from Canada and this has to be
taken into consideration. In general, however, projecting the supply of
cement is not a problem.
2.2 Location
The allocation of production among regions is the most important part
of projecting emissions from the cement industry. Detailed information is
available on existing plants and can be used to allocate output among
regions. However, there is a tendency to close some plants and shift pro-
duction to a more advantageous location. Bulk shipments are now made to a
larger market than was true in the past. It will be necessary to project
P-l
-------
such trends toward a terminal distribution pattern. Information on
existing plants should provide guidance in predicting the plants that
are likely to close in the future. Locational analysis can then be used
to estimate where new plants will be located in order to minimize trans-
portation costs. In general, the new plants do control emissions to the
level required by the standards, though there are some emissions that
escape.
3.0 SUMMARY
The major task in projecting emissions from the cement industry will
be to allocate output on a regional basis. In particular, it will be
necessary to estimate new locations for large distribution terminals. The
secondary problem will be to determine the demand for and supply of cement.
In general, the demand element will be more difficult to project.
P-2
-------
APPENDIX Q
BRICK MANUFACTURING
1.0 INTRODUCTION .
Brick making is a potential source of fluoride emissions if the raw
material has a high fluoride content. If fluorides are not considered
in this model, then the brick making industry will drop out.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for bricks has been growing slowly because the
cost of installed brick is higher than the installed cost of other
materials. This is due to rapidly rising labor costs. This trend is
likely to continue unless there is some substantial change in technology
or institutional arrangements. For example, brick layers may lay more
bricks per hour if union rules are changed, or if pre-cast brick panels
come into common use. Neither of these events is likely to occur in the
near future, however.
Demand can be projected by making brick manufacturing a function of
construction activity. This function can be projected and then applied to
the absolute amount of construction activity projected.
2.1.2 Supply
The supply of brick is usually equal to demand because there
are no imports or exports.
2.2 Location
The distribution of brick plants in the United States is determined
by the population distribution, the availability of simple clay for brick
making, and local cost for competitive building materials. The location
of output can probably best be projected by the use of shift and share
analysis based on current trends. It will be necessary to determine which
plants use clayf with a high fluoride content.
Q-l
-------
3.0 SUMMARY
Brick making is a source of fluoride emissions and will be dropped
from this model if fluoride emissions are not included. If they are
included, the projection of emissions will require three relatively
minor tasks. The first will be to determine the national output in the
industry based on the percentage of building materials captured by the
brick industry. The second task will be to allocate this output on a
regional basis through the use of shift and share analysis. The third
task will be to determine those establishments that use clay with a high
fluoride content.
Q-2
-------
APPENDIX R
LIME INDUSTRY
1.0 INTRODUCTION
The lime industry is a very substantial emitter of particles. Lime
plants are found not only as independent plants but also as subsidiaries
that may be adjacent to or removed from pulp and paper mills and steel
mills. The location of all lime plants is not known with certainty nor
are details on plant operations. The most difficult task in projecting
emissions from the lime industry will be to obtain data on production
and process types by plant. The secondary task will be to project total
output and to allocate this on a regional basis.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for lime is growing rapidly because of the .
growing use of basic oxygen furnaces in the steel industry. This process
requires more lime than the open hearth furnaces. Rapid growth in the
paper and pulp industry, and new uses for lime such as soil stabiliza-
tion, water treatment, and sewage treatment also create increasing demand.
Projections of lime production will depend upon the growth of the steel
and pulp industry as well as the success of lime in capturing new
markets. The national demand for lime will be projected as a function
of the various uses to which it is put,
2.1.2 Supply
The quantity of lime supplied will be about equal to the
demand because imports are not an important consideration. There may
be some lime recycled from its use in pollution control equipment.
This is a minor consideration, however. The regional distribution of
lime production will be a more difficult task.
2.2 Location
The most difficult aspect of projecting the location of lime plants
is the extreme shortage of available data on plant location, production
R-l
-------
and processes. One possible reason for this lack of data is the large
number of captive lime plants that belong to steel or paper mills. It
is likely that a direct survey of lime plants will be required in order
to acquire the necessary data.
Even with data on the current location of lime production, projections
of location of this industry will require intensive study. The number
of plants has been declining due to economies of scale. While there is
some evidence that the number of plants has stabilized recently, there
may still be additional closures. The most likely possibility for the
next ten years will be an expansion of output at existing plants. The
task, therefore, is to project the share of national market for each
of these establishments. The plants that enjoy a locational advantage
relative to markets will probably be the ones that increase their
share. A location study employing various economic variables should
be done in order to allocate shares to each of the regions. If this
analysis is not successful in explaning the location of present
production, than a simple shift and share analysis may give better
results for projection purposes.
2.3 Process
Data will be necessary on the type of process used at each lime
establishment. Emission factors vary substantially between vertical
kilns and rotary kilns. Although rotary kilns.have been increasing
in popularity during the last ten years or so, there is some evidence
that vertical kilns may stage a comeback because of the lower emission
control cost.
Data will also have to be collected on the type of fuel used at
lime kilns because the emissions vary with the type of fuel.
3.0 SUMMARY
The major task in projecting emissions from the lime industry will
be to collect data on an establishment basis for production levels, type
of kilns and the fuel used. Trends in production over time will be
particularly useful for projecting the share of the national market that
R-2
-------
will be held by each establishment. Although projecting national demand
will require some effort, the much larger effort will have to be devoted
to the regional distribution of production.
R-3
-------
APPENDIX S
GASOLINE MARKETING AND BULK STORAGE
1.0 INTRODUCTION
The gasoline marketing and bulk storage industry, SIC 5092, is a
substantial source of hydrocarbons. The hydrocarbons evaporate from
the storage tanks and also escape during the filling process. These
emissions are effectively controlled through the use of floating roof
tanks and submerged fill lines. Many of the existing tanks have these
devices because they prevent the loss of a valuable product. It is likely
that this industry will decline over time as a source of emissions.
2.0 PROJECTION CHARACTERISTICS
2.1 Demand and Supply
2.1.1 Demand
The demand for the services of the industry, which may be
measured in terms of throughput of gasoline, depends upon gasoline
consumption in the surrounding market area. This gasoline consumption
can probably be taken from the study of mobile sources. It is likely
that the throughput at the larger terminals is proportionally greater
than gasoline consumption because these terminals also distribute
gasoline to smaller storage establishments. In general, the projection
of demand on a national basis and on a regional basis should not require
great effort.
2.1.2 Supply
The quantity of throughput in each area will be equal to the
demand for the gasoline. There is not very much information available
on the total number of storage tanks or on the type of tanks in use.
Even less data are available on the size of tanks and the controls in .
use by location.
2.2 Location
Distributing output in this industry to each region can be done in
two different methods. The first is to gather detailed data on tank size
>>
•.v
S-l
-------
and the controls for each establishment. There are more than 29,000
establishments in the United States in 1967 and the data gathering
required would be extensive. An alternative and a much cheaper method
consists of assuming how many tanks are required in an area to handle
the throughput expected on the basis of population and gasoline
consumption. A number of assumptions can be made on the type of tanks
and the expected emissions from each area. As mentioned above some
areas will have storage facilities much larger than expected on the
basis of their population. These large terminals can be identified
in particular AQCR's and probably can be projected.
3.0 SUMMARY
Lack of data about the gasoline marketing and bulk storage industry
is the greatest obstacle to projecting emissions on a regional basis.
It is possible to project emissions using a number of assumption about
throughput and about the type of tanks and controls devices in use.
If desired, factual data can be supplied for the bulk terminals and, then
?
used as a basis for making projections in these AQCR's. EPA might also
consider dropping this industry because of the relatively low emissions
from each source and the declining rate of emissions. ,
S-2
------- |