EPA-450/3-74-020-a
AIR QUALITY FOR URBAN
AND INDUSTRIAL PLANNING
by
John C. Goodrich, Scott T. McCandless,
Michael J. Keefe, William P. Walsh, and Alan H. Epstein
Environmental Research & Technology, Inc.
429 Marrett Road
Lexington, Massachusetts
Contract No. 68-02-0567
EPA Project Officer: John Robson
Prepared for
ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Air Quality Planning and Standards
Research Triangle Park, N. C. 27711
March 1974
-------
This report is issued by the Environmental Protection Agency to report technical
data of interest to a limited number of readers. Copies are available
free of charge to Federal employees, current contractors and grantees,
and nonprofit organizations - as supplies permit - from the Air Pollution
Technical Information Center, Environmental Protection Agency, Research
Triangle Park, North Carolina 27711; or, for a fee, from the National
Technical Information Service, 5285 Peart Royal Road, Springfield, Virginia
22161.
This report was furnished to the Environmental Protection Agency by
the Environmental Research & Technology, Inc. , in fulfillment of Contract
No. 68-02-0567. The contents of this report are reproduced herein as
received from the Environmental Research & Technology, Inc. The
opinions, findings, and conclusions expressed are those of the author
and not necessarily those of the Environmental Protection Agency.
Mention of company or product names is not to be considered as an endorsement
by the Environmental Protection Agency.
Publication No. EPA-450/3-74-020-a
-------
PREFACE
This Final Report for EPA Contract No. 68-02-0567, Air Quality for
Urban and Industrial Planning (AQUIP: Extension) is divided into three
parts. Part 1 presents a summary of work undertaken, including sections
describing the proposed scope of work for each of the three major tasks
for the study, together with a summary of the actual work undertaken and
an explanation of deviations, if any, from the intended scope of work.
The detailed findings of Tasks 1 and 2 are found in Parts 2 and 3 of
this Final Report, respectively] the detailed findings for Task 3 are
found in a separately bound report entitled, A Guide for Considering
Air Quality in Urban Planning. The report on Task 3 is available free
of charge to Federal employees, current contractors and grantees, and
nonprofit organizations - as supplies permit - from the Air Pollution
Technical Information Center, Environmental Protection Agency, Research
Triangle Park, North Carolina 27711; or, for a fee, from the National
Technical Information Service, 5285 Port Royal Road, Springfield,
Virginia 22161.
Ill
-------
CONTENTS
Page
LIST OF TABLES v
LIST OF FIGURES vi
PART 1. SUMMARY OF WORK UNDERTAKEN 1
TASK 1. DEVELOPMENT OF IMPROVED EMISSIONS PROJECTION
ACTIVITY INDICES 2
TASK 2. DEVELOPMENT OF A METHODOLOGY FOR INCORPORATING
COST DATA INTO THE EVALUATION OF THE AIR
POLLUTION IMPACT OF LAND USE PLANS 5
TASK 3. PERFORMANCE OF LAND USE - AIR POLLUTION IMPACT
SENSITIVITY STUDIES 6
REFERENCES FOR PART I 12
PART 2. TASK 1 REPORT. DEVELOPMENT OF IMPROVED EMISSIONS
PROJECTION ACTIVITY INDICES 13
TERMINOLOGY 14
INTRODUCTION TO PART 2 16
DATA USED IN THE STUDY 23
ANALYSIS OF EMISSIONS DATA 34
PROJECTING FUTURE EMISSIONS 55
REFERENCES FOR PART 2 69
PART 3. TASK 2 REPORT. COST-EFFECTIVENESS PLANNING FOR
ACCEPTABLE AIR QUALITY 70
SUMMARY 71
INTRODUCTION TO PART 3 72
BACKGROUND AND PROBLEM DEFINITION 72
THE ECONOMICS OF AIR POLLUTION 74
Trade-off: Air Quality versus Economic Viability. . . 76
The Cost of Achieving Economic Viability in Terms
of Air Pollution Impact 78
The Cost of Controlling Pollutant Emissions 80
The Damage Function 82
LITERATURE SURVEY AND EVALUATION 84
A METHODOLOGICAL APPROACH TO COST EFFECTIVENESS AIR QUALITY
IMPACT LAND USE PLANNING 86
SCOPE AND OBJECTIVES 86
CONCEPTUAL DESIGN '. 88
PROCEDURAL DESIGN 90
APPLICATION GUIDELINES 92
TECHNICAL REPORT DATA SHEET 94
IV
-------
LIST OF TABLES
Table Page
2-1 Two-digit Standard Industrial Classifications (SIC) Codes
for Manufacturing 26
2-2 Parameters Examined 27
2-3 Massachusetts 4-Digit SIC Fuel Data 35
2-4 Massachusetts Total Sample for SIC 20 38
2-5 New Jersey SIC 28 Emissions per Hour per Employee 43
2-6 New Jersey SIC 28 Emissions per Hour per Square Foot 44
2-7 Massachusetts 2-Digit SIC Fuel Data 48
2-8 Propensities to Use Different Fuels 50
2-9 Emissions per Hour per Employee 51
2-10 Inventory of Air Quality Planning Data for Implementation
Plan Questionnaires 63
3-1 Reference Material 85
3-2 Scope and Objectives 87
-------
LIST OF FIGURES
Figure Page
2-1 Five-step Procedure for Determining Emissions 17
2-2 Example Input Data 32
2-3 Data Coding Sheet 33
2-4 SIC 20 Space Heating BTU's per Hour versus Employment 39
2-5 SIC 20 Total Heating BTUs per Hour versus Employment 40
2-6 Detail of SIC 20 Total Heating Bit's per Hour versus Employment 41
2-7 SIC 28 Emissions versus Employment 47
2-8 Two-phase Procedure for Projecting Emissions 56
2-9 Permutations of Power Generation - a Sample 59
2-10 Petroleum Refining Variation - a Sample 60
3-1 The Dependence of Air Quality on Land Use 73
3-2 The Economic Implications of Land Use Planning in Terms of Air
Pollution Impact 75
3-3 The General Relationship of Level of Urban Activity to Air Quality. . . 77
3-4 The Total Cost of Air Pollution 79
3-5 The Cost of Controlling Pollutant Emissions 81
3-6 Pro Forma Damage Functions 83
3-7 Conceptual Methodological Design 89
3-8 Procedural Methodological Design 91
VI
-------
PART 1
SUMMARY OF WORK UNDERTAKEN
-------
TASK 1 DEVELOPMENT OF IMPROVED EMISSIONS PROJECTION ACTIVITY INDICES
The general approach to this task was to be based upon the most recent
point and area source inventories compiled for the development of state
implementation plans by the Environmental Protection Agency (EPA) and its
Basic Ordering Agreement (BOA) contractors with emphasis on a large statis-
tical sample drawn from the industrial point source inventories. Such data
were to be divided into specified land use categories and subcategories.
Activity indices were to be derived empirically from the relationships
between activity levels and fuel use or process emissions. Because of the
massive data-handling requirements, computer data processing was to be used.
Determination of activity indices provides factors for estimating current
emissions which can then be used as a basis for projecting activity indices
for future time periods. This task emphasizes determination of current
activity indices, but also establishes the procedures for projecting activity
indices. These projections may involve detailed assumptions concerning
changes in emission control regulations, industrial processes, fuel switching,
productivity, consumer preferences, and many other factors affecting the
relationship between land use and resultant air quality
Subtasks proposed included the following:
1-a Defining principal land use categories and specifications
for statistical validity.
1-b Collecting 1970 implementation plan point and area source
inventories, particularly those available in machine readable
format. Conducting supplementary surveys of particular
industrial, utility and commercial sources, and surveys of
planning and governmental agencies
1-c Compiling and processing the available data using standard
tabulation and statistical computer programs.
1-d Analyzing the data and deriving activity indices for current
emissions estimation.
1-e Developing the methodology for projection of activity indices
for future emissions estimation.
-------
1-f Documenting the data, methodology, and derived activity
indices
1-g Incorporating the resulting activity indices into the AQUIP
System (an acronym standing for Air Quality for Urban and
Industrial Planning).
Early in the study the following were ascertained relative to data for
Task 1:
1) Data from the National Emissions Data System (NEDS) inventories for
the 1970 Implementation Plan (IPP) information was not available in suffi-
cient quantity and appropriate form for our study. The availability of this
data had been stressed as a necessity for successful completion of the
project.
2) The only data available in computer form was the point source
inventories for certain states from BOA contractors; there were significant
problems of confidentiality, making it difficult, if not impossible, for
EPA to release this data to us for our use in the study. Moreover, most of
the states were not in a position to release the data at the time due to
litigation over the question of confidentiality.
Accordingly, it was decided to obtain data for New Jersey and Massa-
chusetts directly from the states since ERT already had worked with agencies
in these states. As documented in the monthly progress reports, this process
resulted in (1) significant delays, (2) unexpected efforts in data gathering,
verifying, and manipulation, and (3) certain redirection of subsequent sub-
tasks relying upon these data.
By reducing the desired sample number and relying upon national indi-
cators of floor space per employee for industrial firms, the initial data-
gathering was concluded with the hope of maintaining sufficient detail of
information to carry out Task 1. Efforts that would have otherwise been
devoted to suppler.eutal surveys and literature searches (Subtask 1-b) were
instead concentrated on coding, keypunching and data preparation. Emphasis
was placed on industrial point source data to the exclusion of area source
data or other types of point source data.
-------
The activity categories to be investigated were broadly defined as
"industrial" (Subtask 1-a); this was ta be narrowed according to the dis-
tribution of Standard Industrial Classification (SIC) contained in the final
sample obtained.
Unfortunately, errors and missing information were found in the data.
Field trips to each of the major data sources in Massachusetts and New Jersey
were then necessary to obtain supplementary information. Time was spent in
the New Jersey Department of Environmental Protection, Bureau of Air Pollution
Control Office, reviewing printouts of point source emission rates, employ-
ment, fuel consumption and process rates. A review of Bureau of Air Pollution
Control questionnaire files was needed to supply information not currently
stored in computer data banks. Likewise, in Massachusetts a review of
Department of Public Health source questionnaires was made to supplement data
in existing printouts.
It became clear as the technical work was concluded that the data
analysis of Task 1 would yield less than the desired statistical sample and
analysis results because of the data-gathering problems of Subtask 1-b.
Therefore, the documentation stresses the approach, problems, opportunities
and recommendations for further work, and de-emphasizes concrete statistical
conclusions. This re-orientation resulted from problems in obtaining requested
data and has been consistently documented in the progress reports.
When all Massachusetts data was identified and received and the status
of all New Jersey data determined, definition of the required tabulations
and statistical computer routines was begun. As a part of the previous AQUIP
study computer algorithms for comparing fuel use, emissions, floor space
and employment by SIC had been partially developed and, since the changes in
Subtask-lb required that data be coded, a form compatible with this software
was used. The software requirements for Subtasks 1-c and 1-d were, thus,
simultaneously formulated.
After initial computer analyses were performed for both the New Jersey
and Massachusetts data, errors corrected, and the final computer runs per-
formed, the interpretation of statistical analyses (Subtask 1-d) was brought
to the disappointing and abbreviated conclusion that the accuracy and com-
pleteness of the data were not sufficient to derive activity indices at the
level of detail anticipated.
-------
The projection methodology (Subtask 1-e) was most affected by data-
gathering delays. Preliminary work began with a literature survey and com-
parison of information with the initial Hackensack study and concluded with
the postulation of the kinds of information and decisions necessary for
emission projection. Given the late start for this subtask, the disappointing
results of Subtask 1-d (necessary as an input >to Subtask 1-e) , and the lack
of other readily usable information, no definitive results were forthcoming
from this subtask.
Work related to documentation (Subtask 1-f) is contained in Part II of
this Final Report. The small degree to which information was shown to be
statistically conclusive and, therefore, documentable, and the inability to
incorporate revised activity indices into AQUIP to any great degree (Sub-
task 1-g) has been a function of the disappointing results of Subtask 1-d
which came ultimately from the data gathering problems of Subtask 1-b.
Moreover, the necessity of concentrating on a specific source category-
industrial point sources due to time and budget constraints precluded the
examination of activity indices which might have been more appropriate for
other source categories.
TASK 2 DEVELOPMENT OF A METHODOLOGY FOR INCORPORATING COST DATA INTO
THE EVALUATION OF THE AIR POLLUTION IMPACT OF LAND USE PLANS
The general approach to this task was to focus on collecting and
assembling the results of EPA-sponsored studies on source control costs and
on evaluation of economic impacts of various control strategies. ERT was to
examine the data available and formulate a framework for presenting such
data relative to land use plans, developments, or facility designs. Param-
eters to be examined were expected to include land use zone cost indices per
unit emission control
Subtasks proposed included the following:
2-a Surveying of literature and federal contract research efforts
to compile source air pollution control cost data and data on
costs and economic impact of implementation plan control
strategies.
2-b Developing a methodology for including cost data in the
evaluation of land use plans.
-------
2-c Compiling representative cost data to test and demonstrate
the techniques for a hypothetical application.
2-d Documenting results of the test and demonstration and making
recommendations for incorporation of the cost data base and
methodologies into the AQUIP system for planning use.
The research showed that existing data and literature were sparse but
that reasonable methodologies could be developed for including cost data in
the evaluation of land use plans. Literature pertinent to the economic
implications of air pollution may generally be categorized into three dis-
tinguishable areas of concern:
1) The tradeoff of economic activity with air quality.
2) The functional relationships; of emission control costs and
damage costs to air quality,,
3) Cost/benefit analyses associated with the control and damage
costs of air pollution for a given level of economic activity
as a function of land use strategy.
The findings of Subtasks 2-a and 2-b determined the depth possible in
demonstrating the methodology. Accordingly, work on the hypothetical appli-
cation of the methodology (Subtask 2-c) concentrated on a review of the
literature on air pollution damage functions. Documentation of the method-
ology (Subtask 2-d) was done in the form of Part III of this Final Report,
entitled "Cost Effective Planning for Acceptable Air Quality."
TASK 3 PERFORMANCE OF LAND USE - AIR POLLUTION IMPACT SENSITIVITY
STUDIES
The proposed approach to this task was to utilize the AQUIP System to
carry out certain sensitivity analyses. ERT was to use the existing land
use data base for the Meadowlands, and, in addition, incorporate the improved
activity indices resulting from Task 1. Several planning agencies were to
be contacted to identify a set of specific and meaningful small area and
facility design alternatives. On the basis of such alternatives, ERT would
specify the number of scenarios or case studies and trade-off parameters to
be modeled. Air quality would be projected for each of these scenarios and
the results correlated with changes in different design parameters.
-------
Subtasks proposed included the following:
3-a Identifying case studies, land use configuration alterna-
tives, and facility design choices to be included in the
sensitivity analysis.
3-b Defining the parameters and scope of specific sensitivity
analyses.
3-c Preparing inputs for the AQUIP System appropriate to each
sensitivity study.
3-d Running the AQUIP System and processing resulting air
quality as a function of parameters described in Subtask 3-b.
3-e Documenting results in the form of guidelines of general
applicability for planners.
Visits to planning agencies in New Hampshire, New York, New Jersey, and
Massachusetts indicated the specific need for a document containing guide-
lines dealing with air pollution for planners. Accordingly, such a guide-
lines document became the main goal of Task 3, with the sensitivity studies
a major contributor to this final product. This guidelines document is a
separately bound report, entitled "A Guide for Considering Air Quality in
Urban Planning."
ERT conducted several interviews of planning agency personnel projected
to be potential users of the findings of any air quality planning guidelines.
ERT initiated a dialogue with these professionals concerning the goals,
methods and data to be used in the study procedure and attempted to determine
which tools and format the intended beneficiaries of the investigations would
prefer to see developed. By acquainting potential users of the guidelines
document with the early stages of data collection, it was hoped that the
guidelines might be more usefully tailored to the needs of such users.
In addition to the air quality agencies in Massachusetts and New Jersey,
as well as EPA, four regional planning agencies w.ere interviewed: Southern
New Hampshire Regional Planning Commission (Manchester, New Hampshire),
Central Massachusetts Regional Planning Commission (Worcester, Massachusetts),
Tri-State Regional Planning Commission (New York City), and the Hackensack
Meadowlands Development Commission (Hackensack, New Jersey).
-------
The information obtained from speaking to these agencies can be sum-
marized as follows. Traditional planning agencies appear to be relatively
ignorant of the relationships between planning practices and air quality.
While each knew that transportation vehicles and industries are the primary
sources of air pollution emissions, the concept of planning for air quality
was a new one. Each agency seemed eager to have some capabilities in the
area of planning for air quality. Motives among the agencies for wanting
these capabilities were diverse, however. Some saw AQUIP a.s a tool for use
in zoning, public hearings, or transportation planning. Each agency
wished to have the study provide a simplistic approach to air quality
planning. The planners seemed much mere interested in receiving a manual
that would provide a cookbook methodology for studying air quality than a
semi-theoretical discussion of air qua.lity parameters. Specifying land uses
by SIC code (Standard Industrial Classification) was universally accepted
as desirable since both land use planners and air quality agencies were
familiar with them. Other parameters received mixed reactions as being
acceptable indices. Some preferred floor area, etc., as a means for
projecting emissions activity. Each agency was anxious to see the guide-
lines document to be produced.
Subtask 3-a required the identification of "case studies, land use
configuration alternatives, and facility design choices to be included in
the sensitivity analyses." As described in greater detail in the guide-
2
lines document, these were chosen to be the following.
Case Study No. 1
Relative effects on annual average air quality resulting from clustered
versus dispersed area sources. Land use configuration alternatives: (1)
clustering, and (2) dispersal. Facility design choices: (1) single area
source, and (2) four dispersed area sources.
Case Study No. 2
Relative effects or Annual average air quality of clustered versus
dispersed point sources. Land use configuration alternatives: (1) clustering,
-------
and (2) dispersal. Facility design choices: (1) single point source, and
(2) four dispersed point sources.
Case Study No. 5
Relative effects on air quality of clustered versus dispersed sources
in a worst-case situation. Land use configuration alternatives: (1)
clustering, and (2} dispersal. Facility design choices: (1) one point
source and one area source, and (2) one point source and four dispersed
area sources.
Highway Sources
Facility design choices: (1) elevated, (2) depressed, and (3) at-
grade.
Subtask 3-b required the definition of "the parameters and scope of
specific sensitivity analyses." As described in greater detail in the
guidelines document, these were chosen to be the following.
Case Study No. 1
Comparison of annual average air quality for a concentrated area
source with annual average air quality for four smaller dispersed area
sources having the same total source strength. Parameters included:
1) Meteoro1ogica1 Cond itions - annual stability wind rose for
Newark, New Jersey.
2) Source Strengths - (a) area source of 4,000 grams/sec, (b)
four dispersed area sources, each emitting 1,000 grams/sec.
Case Study No. 2
Compar-\,.->n of annual average air quality for ,a concentrated point
s^'^ue with annual average air quality for four smaller, dispersed point
sources having the same total source strength. Parameters included:
-------
1) Meteorological Conditions - annual stability wind rose for
Newark, New Jersey.
2) Source Strengths - (a) point source of 4,000 grams/sec, (b)
four dispersed point sources, each emitting 1,000 grams/sec.
Case Study No. 5
Comparison of worst-case air quality for a concentrated point source
and a concentrated area source with worst-case air quality for the same
concentrated point source and four dispersed area sources. Parameters
included:
1) Meteorological Conditions (identical for both source configurations)
(a) high stability, (b) low wind speed, and (c) southwest wind.
2) Source Strengths - (a) point source of 5,000 grams/sec, and area
source of 4,000 grams/sec; (b) point source of 5,000 grams/sec,
and four dispersed area sources, each emitting 1,000 grams/sec.
Highway Sources
Discussion of worst-case air quality for different highway designs
under different meteorological and traffic conditions. Parameters included:
1) Meteorological Conditions - (a) stability, and (b) wind speed.
2) Highway Designs - (a) elevated, (b) depressed, and (c) at-grade.
3) Source Strenth Dependence on Traffic Characteristics: (a) volume,
(b) speed distribution, and (c) vehicle year and make mix.
Subtask 3-c required the preparation of "inputs for the AQUIP System
appropriate to each sensitivity study" while Subtask 3-d required the
running of the AQUIP System and the processing of "resulting air quality
as a function of parameters described in 3-b." The revised scope of the
sensitivity studies led to the decision to use them in illustrating an
important factor in planning for air quality: the influence of source
configuration for both stationary and highway sources. For this purpose,
the most effective use of the AQUIP System (Subtasks 3-c and 3-d) lay in
using the diffusion modeling capability, MARTIK (based upon the EPA Martin-
Tikvart Model), independently of the rest of the system. For describing
10
-------
the influence of source configuration on highway sources, results obtained
from the ERT numerical simulation model, EGAMA (developed ^y Qr- B_ ^m Egan
and Dr. J. R. Mahoney).were used. EGAMA is not part of the AQUIP System,
but its results have been found to be most useful in determining the impact
of alternative highway configurations on air quality.
The improved activity indices resulting from Task 1 were not available
to the sensitivity studies on account of data limitations. This precluded
the incorporation of such improved indices into the existing land use data
base for the Meadowlands. The processed air quality results for the
sensitivity studies are depicted graphically in the guidelines document.
As discussed above, the guidelines document itself represent the results of
Subtask 3-e.
11
-------
REFERENCES FOR PART 1
1. Willis, B. H. et al. The Hackensack Meadowlands Air Pollution
Study, Summary Report and Task Reports, Tasks 1-5, prepared
for the State of New Jersey, Department of Environmental
Protection, Trenton, 1973.
2. Epstein, A. H. et al. A Guide for Considering Air Quality in
Urban Planning, prepared for USEPA, March 1974,
Contract No. 68-02-0567.
12
-------
PART 2
TASK 1 REPORT
DEVELOPMENT OF IMPROVED EMISSIONS PROJECTION
ACTIVITY INDICES
by
John C. Goodrich
Scott T. McCandless
Michael J. Keefe
William P. Walsh
-------
TERMINOLOGY
Because the terminologies of 5>everal different professions are used
in this task report, often in unfamiliar ways, this brief discussion of
terminology is presented to show the context within which different terms
were used.
The basic land use and transportation planning units of intensity of
use - such as square feet of industrial plant space - are called the activities
or the activity level. The parameters which translate the activity levels
into demand for fuel for heating purposes are called activity indices; for
instance, BTUs (British Thermal Units of heat demand) per square foot for
industrial plant space.
A distinction has been made between fuel-related and nonfuel-related
activities or sources of emissions. The fuel-related sources use fuel for:
1) Heating area, for example, heating a building in the winter. The
amount of heat required and the fuel consumed is a function of the
temperature or the number of degree-days (the sum of negative
departures of average daily' temperature from 65°F). This fuel
use is that required for heating, or space heating (or cooling).
2) Raising a product to a certain temperature during an industrial
process. The amount of fu
-------
particular fuel or fuels (the fuel use propensity) determines the actual
fuel used to satisfy the heat requirement.
Different types of activities may have varying activity indices or per-
cent space heat or fuel use propensities; for instance, each industrial
category in the U. S. Census 4-digit Standard Industrial Classification
(SIC) may have a unique value. However, we may know information only by
broad industrial groups comprising aggregates of the 4-digit classification
(e.g., 1-digit or 2-digit SIC codes). Using the value applied to the
larger or broader group for the smaller or more detailed group, when the
unique value is not known, has been termed a default parameter in these
studies.
Emissions from sources that do not result from the burning of fuel;
for example, evaporation from a refinery storage tank, are termed separate
process emissions or process emissions. Note the distinction between
process heating related or combustion emissions and separate process or non-
combustion emissions. Although the combustion of fuel is involved, trans-
portation emissions have been considered as process rather than fuel emissions
in this study for simplicity, since they do not vary with heating degree
days.
15
-------
INTRODUCTION TO PART 2
The Environmental Protection Agency (EPA) and the New Jersey Department
of Environmental Protection (NJDEP) sponsored a study, commencing in 1970,
addressed to their mutual concern for improving future air quality through
the planning of land use and transportation activities. The two fundamental
objectives of the Air Quality for Urban and Industrial Planning (AQUIP) study
were: (1) to develop a broadbased methodology for considering air pollution
in the formulation and evaluation of alternative urban plans; and, (2) to
demonstrate this methodology in detail by applying it directly to the
planning alternatives developed for the New Jersey Hackensack Meadowlands
District.
One of the major goals of the AQUIP study was to develop a methodology
to aid planners in determining air pollutant emissions directly from land
use and transportation activity data. Procedures traditionally used to
estimate emissions from land use and transportation planning data often
emphasize empirical derivation of emissions indices as a one-step function
of "activity categories (e.g., activity times and index yields emissions).
In the AQUIP study, however, a multistep approach was developed so that:
(1) all assumptions and constraints involved in transforming the levels of
activities into emissions could be examined; and (2) procedures for updating
the information which the planner doe:; not directly input could be specified.
In response to the study objectives a five-step procedure was formulated
in the AQUIP study as shown in Figure 1.
Step 1 - Activities. For each land use or transportation planning
category identified for analysis, the "level of activity" is specified,
such as 10 dwelling units per acre for residential density.
Step 2 - Activity Indices. For each category of activity, "default
parameters" for determining fuel requirements are developed, such as
1C tiTUs (British Thermal Units of heat demand) per hour per square
foot of residential floor area.
Step 3 - Fuel Use. For each category of activity (and geographical
subregion of the study area) default parameters for the "propensity"
16
-------
in
to
a
o
^
en
w
rl
60
c
H
c
H
V-i
(U
4-1
0)
Q
ro
CVJ
i!
0)
M
OJ
a
o
14
CM
a.
0)
w
w
I
Q)
I
CN1
bO
17
-------
to use different fuels are applied to the fuel requirements, such
as the degree to which oil is used (65%) rather than natural gas
(25%) for home heating.
Step 4 - Emission Factors. For each category of activity, engineering
estimates of fuel and nonfuel (process) related "emission factors"
are developed and applied directly to fuel use and process rates to
determine emissions, such as 10 Ibs of participates per 1000 gallons
of fuel oil burned.
Step 5 - Emissions. Emissions calculated from fuel and process sources
are adjusted for season of the year, based on temperature variation
(degree-days) and default parameters representing the percent of fuel
used for "space heating" purposes.
Of particular importance in the AQU1P study was the application of these
five-step procedures in two distinct and consecutive phases. In the first
phase current planning data and current fuel use are correlated to produce
projecting indices - the default parameters. In the second phase these
projecting indices are modified to reflect future time periods and are
applied to planning data so as to generate future fuel demand and emission
levels. Current data on fuel use and emission factors are likewise used to
predict future information when better estimates are not known. The first
phase analysis provides the majority of the default parameters to be used
in the second phase in conjunction with the planner's own inputs.
The application of the emissions projection methodology to the Meadow-
lands plans showed that the five-step procedures were workable and, in fact,
quite adaptable to the land use considerations that were encountered. In
particular, the development of a "conversion factors catalog" and sets of
"default parameters" demonstrated that the planner need input only planning-
related data to use a tool such as the AQUIP System.
However, it was found that the planner must specify data he does not
normally deal with, such as the sizes of developments in terms of their
heating requirements, and the types of manufacturing operations anticipated.
Furthermore, the level of detail available for empirically deriving the
default parameters was unsatisfactory for discerning between related activi-
ties; this was particularly true for deciding fuel use and determining
-------
process-related emissions. Consequently, the greatest need for further work
was shown to involve the empirical derivation of activity indices and default
parameters.
Activity indices as shown in Figure 2-1 are the basic factors used to
convert land use activities to either fuel combustion or process emissions.
This is precisely the area, however, for which it was shown in the previous
AQUIP study that there is the greatest lack of data, and for which there is
the greatest need for detailed and accurate data (fuel use data and emissions
factors are generally readily available from EPA).
The objective of Task 1 of the current AQUIP Extension Study was, there-
fore, to define, collect and process activity data so as to derive improved
activity indices for the AQUIP System.
In previous studies (including the AQUIP Study) it has been demonstrated
that one of the principal variables in determining existing and future air
quality levels is land use. This is because land use dictates emission
source characteristics in terms of both the levels of activity and the
activity indices. In order to develop the desired activity indices for
land uses, it is necessary to define the land use categories to be studied
and to arrange these categories in a manner that facilitates statistical
analysis.
Of the major categories of emission sources we determined that indus-
trial point sources most warranted development of activity indices that could
be readily used in the planning process.
The two largest contributors to source emissions have been shown to be
transportation and industry. Most other source categories are either rela-
tively insignificant or have been studied to some extent. Since research
was already being supported by EPA in regards to transportation sources and
emission patterns, it was decided in conjunction with the project officer
to concentrate the efforts of Task 1 of the AQUIP Extension on industrial
sources. For several years research with respect to air quality has focused
on mobile sources. As a result, there are several documents now in publi-
cation that enable one to calculate existing vehicle emissions and to fore-
cast emissions from given numbers and types of transportation vehicles for
2
future time periods . It was also decided that the air quality impacts of
transportation systems are both very specialized and usually localized.
This being the case, major transportation studies should be accompanied by
19
-------
individual air quality assessments based on much more specific and detailed
data than that which could be collected, analyzed, and organized in the
AQUIP Extension Study. In addition, transportation sources cannot reason-
ably be grouped as a land use for air quality considerations. Transportation
systems are products of other land uses and activities, but the variables
which affect pollutant emission fron mobile sources are riot at all similar
to the variables that affect emissions from point sources. Accordingly,
land use and transportation need to be treated as separate, but interdependent
emission sources.
Quasi-public activities, such £.s incinerators and power plants, are
point sources which warrant individual analysis. Such activities do not
lend themselves well to statistical analysis since designs, capacities
and, therefore, emissions, may differ radically. Generally, the relatively
small number of such emission sources makes this individual assessment
feasible.
Other sources may generally be considered insignificant on a regional
scale. Although it is obvious that commercial and residential facilities
have a space heating function, the levels of emissions due to the space
3
heating needs are both small and fairly predictable using current information .
These sources do, however, need to be considered as inducers of transportation
sources.
The previous AQUIP study concluded that industrial sources are of
special significance because of the uncertainty as to the amount of fuel
required for process heating and the incidence of separate process emissions.
Efforts to develop a statistical sample of the propensity to use fuel for
process heating by industrial category, based upon the existing emission
inventories, were not successful in the previous study. It was possible
to divide the industrial SIC codes into only two major categories of
"relatively clean" and "relatively unclean" industries. The clean indus-
tries were assumed to operate fewer hours per year and use a greater
percentage of their fuel for space heating.
A separate study of process emissions corresponding to the industries
proposed for the New Jersey Hackensack Meadowlands was made as a part of
the previous study. With the exception of possible sources in the chemical
and petrochemical and prir.ury rretals area, the SICs proposed for the
Me;u!o\ lanes were not found to be significant separate process emitters.
20
-------
There were some potential emissions of participates and hydrocarbons from
selected industries. These were accounted for by adding an arbitrary best-
estimate percentage of the fuel-burning emissions to the fuel emission
estimates, since no information was available on process rate. The Meadow-
lands planners felt that there would be no petrochemical or primary metals
smelting operatins in the Meadowlands.
As a result of missing data, the concept of "default parameters" was
developed in the previous study. If information is desired according to
a detailed industrial classification for the propensity to use different
fuels and the data is only available in aggregate form for all industries
in the region, a default parameter is used to assign the industry-wide
factor to each individual industry. If, at a later date, specific information
for an industry is available, it can be used in place of the default parameter,
As a result of the previous AQUIP Study, it was clearly evident that
the activity indices required better specification. It was hypothesized
that industrial point source emissions are a category particularly in need
for further study and that such sources would probably lend themselves to
some reasonable statistical analysis. These hypotheses were based on the
fact that industries have several inherent similarities that should make
planning for air quality a reasonable undertaking. The industrial activity
indices that were considered the statistical analysis were:
1) Employment
2) Floor area
3) Process weights (or production rates)
4) Fuel use
5) Hours of operation
6) Classification by product
The fact that industries are conveniently organized by type of activity
is probably the factor that makes this system of emission projection seem
most feasible. The U. S. Census Standard Industrial Classification (SIC)
codes aggregated land uses into numerical groupings by product. It was
hypothesized, therefore, that industries with similar process emission
characteristics are grouped together. These same SIC codes were chosen as
the common denominator since they are fai.iiliar to both planning agencies, as
21
-------
a tool to classify land uses, and to air pollution control agencies as a
classifier of point source activities.
The following chapters present the data used, analysis of emissions
data, and discussion of projecting future emissions.
22
-------
DATA USED IN THE STUDY
In order to derive activity indices for emissions estimations, a
substantial data base is required. The original scope of work specified
that EPA would provide the desired data. It was intended that the EPA
data collected for the National Emissions Data System (NEDS) would provide
sufficient data to perform statistical analyses. Early in the study,
however, it was discovered that the EPA data could not be made available.
Much of the information in the NEDS files is confidential material.
Although its nature is limited to pollutant-related statistics, industries
are very sensitive about releasing any information to their competitors.
Likewise, some industries are reluctant to have figures on emissions and
fuels available to groups and organizations that may lobby against the
interests of .the industries. EPA was not able to provide the general
data files without ensuring that the confidence of the contributing
industries would not be violated. Accordingly, the General Counsel of
EPA decided that the data files would not be released during the course
of the study. Without these data the entire AQUIP Extension Study was
threatened.
After consulting with the project officer, the consultant elected to
apply directly to several state air quality agencies for the data required
for the study. The constraints of budget and time limited the number of
state air quality agencies that could be petitioned for information. It
was decided that two reasonable sources of emissions data would be
Massachusetts and New Jersey, since both were familiar with the contractor.
Massachusetts data was filed in the Boston Office of the Massachusetts
Department of Public Health, only a few miles from the consultant's office
and through previous contracts in the state, a working relationship with
Massachusetts officials had been established. New Jersey had been the
site of the previous AQUIP study, and was therefore considered an appropri-
ate state from which emissions data could be solicited.
The consultant requested a fairly comprehensive set of data from both
states, including point source information for each of the following cate-
gories:
1) SIC code for each pc'r.t source
2) Er.plC;.T..I;\" of the pcint source
23
-------
3) Annual operating hours of the source
4) Fuel use by type and amount of fuel
5) Percent of fuel used for space heat
6) Product process rate
7) Floor area of the point source building.
8) Emmission rates for sulfur dioxide (SO ), total suspended
particulates (TSP), carbon monoxide, (CO),hydrocarbons (HC),
and nithogen oxides (NO,,).
As in previous air quality studies, the air quality agencies from
each of the states were assured that all data used would be confidentially
handled. All point sources -were, accordingly, coded to a source number,
and reference to the sources is by number and SIC codes only. Since all
data was supplied by the two states and since only EPA will review results
there was no loss in confidentiality of the data.
Although each of the states was quite cooperative, assembling the rele-
vant data presented some unavoidable problems. The commltant had hoped to
expedite the data-gathering proces.s by securing the relevant materials in
the form of computer tapes or computer card decks, but this was not possible.
In each case compiling the data inventory involved obtaining clearances to
examine state air quality data and extracting the appropriate information by
hand from existing printouts and questionnaire files. Tn many cases,
accumulating the data required large amounts of time for cross-referencing
and for normalizing the inputs into congruent sets of units. In its final
form the data inventory reflects the problems that were encountered. Major
shortcomings of the data set include problems in the following areas:
Percent space heat or process heat data was not available for New
Jersey. Findings in this area are, therefore, extrapolated from Massachu-
setts data alone. Employment figures for the point sources from both
states were fragmented. (In some cases the consultant does not feel
confident that employment figures supplied are entirely accurate either,
but the figures obtained from the state agencies were used without review.)
Floor areas for the industries were not available. Since floor areas were
not available from the industries, they were derived using employment
figures and a table uhich estinates unit floor areas per employee by SIC
code. This obviously ir
-------
After screening and computer runs to organize the material, the final
form of the data consisted of some 5500 computer cards which outline informa-
tion (although fragmented and incomplete) for some 868 point sources of
industrial emissions. Of these, 555 sources were from Massachusetts and
313 from New Jersey. This included point sources with a wide range of SIC
code numbers, however. The final working file of data was trimmed to
include only manufacturing industries - those whose SIC codes begin with
the digits "2" or "3". Table 2-1 shows the two-digit classification for such
manufacturing industries. The analysis concentrated on the general area of
industrial sources, whle institutional, commercial, and all other sources
which were presumed to be relatively minor were excluded from subsequent
analysis. This decision left at least some statistical input for each of
a total of 644 manufacturing point sources.
Management of the data collected presented the study with the prodigious
tasks of recording, keypunching, checking, sorting, aggregating and analyzing.
The consultant made use of its electronic data processing capabilities to
record, compile, aggregate, and tabulate the various point source-statistics.
The data management capabilities of the algorithms used are as follows:
1) For each point source the statistical analysis program produces
a summary, by source, of identifying information such as source identifica-
tion number and the following parameters of interest.
SIC code
Number of employees
Operating hours/year
Gross plant area (sq.ft.)
Enclosed floor area (sq.ft.)
Percent process heat (e.g., nonspace heating fuel use)
Missing data, if any , is so indicated. Also, supplementary information
is generated as follows: For each type of fuel, the amount of fuel, the
corresponding BTUs supplied and the percentage of the total BTUs supplied
by this fuel, are so indicated. Table 2-2 shows the parameters in terms
of their computer algorithms.
Also, for the sar.o fuel-u^e data, given the total BTUs supplied (by
all fuels), the totrl picccss K-it BTUs and the total space heat BTUs arc
co.vp'..: Led (if the percent proecv.s heat is kuov.n); the corresponding BTU/hcLii
25
-------
TABLF! 2-1
TWO-DIGIT STANDARD INDUSTRIAL CLASSIFICATION (SIC)
CODES FOR MANUFACTURING
Industry Type SIC
Food products 20
Textiles 22
Apparel 23
Lumber and wood 24
Furniture 25
Paper products 26
Printing and publishing. 27
Chemical 28
Petroleum 29
Rubber and plastics 30
Leather products 31
Stone, clay and glass 32
Primary metals 33
Fabricated metals 34
Machinery 35
Electrical machinery 36
Transportation equipment 37
Professional, scientific
precision-made instruments 38
Miscellaneous manufacturing 39
26
-------
TABLE 2-2
PARAMETERS EXAMINED
general parameters
PROC. RATE
PERC. PROC. HEAT
HOURS/YEAR
heat content
parameters
FUEL BTU/HR
pUEL BTU/HR_EMPL
FUEL BTU/HR -ENC.*
(Broken down by Process, Space and Total heating,
where TOTAL = PROCESS + SPACE)
fuel emission
parameters
FUEL AMT/HR
FUEL AMT/HR- ENC*
(Broken down by pollutant: SO , Particulates, HC, CO, NO )
Z A
separate
process emission
parameters
PROC AMT/HR
AMT/HR-EMPL
PROC AMT/HR-ENC*
(Broken down by pollutant; S02, Particulates, HC, CO, NO )
-X
*BTU/hour per unit enclosed floor space
AMT/hour per unit enclosed floor space
27
-------
(BTU divided by operating hours/year), BUT/HR-EMPL (BTU/HR per employee),
BTU/HR-ENC. (BTU/hour per unit enclosed floor area) are also computed for
the process heat and space heat portions of total BTUs. Note that if
such variables as percent process heat, operating hours/year, number of
employees, or enclosed floor area are missing, the related categories will
not be created or shown.
Emissions data are also analyzed; the amounts of fuel ('FUEL...) and/or
separate process ('PR0C...') emissions, by pollutant, are given. Additional
parameters such as AMOUNT/ HR, AMT/HR-EMPL, AMT/HR-ENC may be computed and
printed, for both fuel and process emissions but only if certain variables,
as mentioned in the previous paragraph, are not missing.
For compatibility with the data sets from the previous AQUIP study,
summary data may also be given for FUEL BTU/HR-GR0., FUEL AMT/HR-GR0.,
PROC AMT/HR-GR0, where GR0 is the gross plant area.
2) The statistical program will also aggregate or group, by 4-digit
SIC codes, the total sample size, the number of non-missing data,
the mean and the standard deviation of all of these variables. The mean
and standard deviation are determined as follows, where each of the 42
parameters examined (e.g., PROCESS FUEL BTU/HR) is symbolically represented
as XL, L = 1,...,N:
N
Z x
and
N - 1
To conserve storage of data in the computer files, the sums Zx. and Zx^
are formed, by point source (N is the number of sources per SIC). It is
apparent that Ex. and Zx, may be determined by the following iterative
procedure:
28
-------
For the k source,
I x. = x, ,k=l
L=l L X
k k-1
I x. = x, + Z x. ,22lklN.
The means and standard deviations of all relevant variables (see Table 2-2)
are computed, using these intermediate sums, and a tabulation of all para-
meters, with nonzero means is then given; this includes the total sample
size, the number of non-missing data, the mean and the standard deviation
of all parameters, by activity code (SIC).
3) The statistical program can also truncate the activity codes to a
specified number of characters (LENGTH-X) and reaggregate the total sample
size, the number of valid data, the mean and the standard deviation,
according to these truncated activity codes. In this way, statistics are
determined at the 1, 2, and 3-digit SIC levels. The statistical program
performs this function in the following way:
Given that Mm is the number of sources for the m truncated activity
code, and xm and am are the corresponding mean and standard deviation,
respectively,
J _
I N x
x =
where
J
N = £ N
i
m=l
intermediate sums (by activity code):
29
-------
I N
m
E N x
m m
J
I ' {N - 1) a + N x } NX 2
, m J m mm
a =\ |H£i
N - 1
intermediate sum:
- 1) a 2 + N x 2}
m m mm
where there are J sets of parameters (activity codes) per truncated activity
code.
A tabulation of these variables, similar to that described above for
4-digit SICs, will then be produced.
4) The statistical program also determines the percent of total BTU
satisfied by each of the possible fuels. In calculating the percentage of
total BTU, by fuel, the following relationships are used:
Given that btu. = I btu (for all fuels)
and f.: = btu (by fuel)
for all N sources,
= f
J
F 'BTU
= L-1.J 'L-1,J + J.'JJ > L-2.3.....N
BTUL-1,J + btUJ
PL,J= 10° ' FL,J, L=1,2,...,N
30
-------
where
L = source number
j - activity code number
FL-1 J and BL 1 J are the fraction of tne previous total BTU,
by fuel, (up to and including the L-lst. source) for activity
code number J and that total BTU, respectively; f and btu
*J J
are the fraction of the total BTU, by fuel, for the present
source and that" total BTU, respectively.
BTU for activity code J, for each fuel, is then calculated as:
BTU.. T = btuT
i j «J «J
BTU. T = btuT + BTU. . _ L = 2,3, ... ,N
L f *J J Li 1 j«J 9
5) The statistical program took 320,000 bytes of capacity on an IBM
360/75. In the course of the study the program was used in some seven
individual operational runs. Typical run time was between 10 and 15 minutes.
The following pages give illustrative examples of example input data (Figure 2-
2) and a data coding sheet (Figure 2-3).
31
-------
PARAMETERS HACKENSACK FOLLOW-ON
&INPUT
S C21,1 PARTIC'r' C 0',' HYDRUC',I N OX1. OUNIT=S*'
FNAhsiCOAL't 'Rfc'S. OIL'.'DIS. OIL1, 'NAT, GAS',
&END
RESET
SRCE 1970 MASSACHUSETTS & NEw JERSEY POINT SOURCES
POINT 2001
1 3111
2 1970 10 4460, 6!i>83 50,
3 , 158, , ,
. 4 1 12, , , ,
99999
POINT 2002
1 3111
2 19700000014Q 0062440. 0087600095,
3 t 0000630, , ,
4 10000049. ' , , t ,
99999
POINT 2003
1 3999
2 197000000020 0011220, 0030000100, 2,
4 2 , , ,0000002. ,
99999
POINT 2004
1 3271
* 197000000025 0023850, 0045500100,
3 , 0000164. , ,
4 10000012, » , t
99999
POINT 2005
1 3443
2 197000000053 0044732, 002D000100, 3.
4 £ , . ,0000003, .
99999
STATISTICS 3 & 4 DIGIT SIC'S
PARAMETERS TRUNCATE TO 2 DIGITS
&INPUT LbNGTn=2» SEND
STATISTICS 2 DIG IT SIC'S
PARAMETERS TRUMCATE TO 1 DIGIT
&INPUTLENGTH=1, S.END
STATISTICS 1 DIGIT SIC'S
ENDJOB
Figure 2-2. Example of Input Data.
32
-------
WS/C POINT SOURCL DATA
COLS
50URCL I.D..
ADD.^iio
n-12,
46-70
D
,
0
*
.V.
7"'
fill
LQ .
Mill!
:L 1 LLL
271
A/0 f>, O
1 !
1
FORLIA^.
Acr.v/TY CODE.
nufJ/ciPA i, i TV. AW D couwrv c^D£
U7~f~) K - coort D/ A/ /A r £. - /("« CAST
U T t~~( Y-COORD/"ATl_ . K!~\ HOfirt-l
BASL {.LLVATfOH ffT. /(£>.'£. 5£AL-)
ZOfvL,Ar/D option COOL
5
H~IS
21- 3O
31-36
51-56
6S-- 7f)
n
1 .
i i
i ! _j
i
| ;
i , 1
'
HICW DATA APPj.lE.5
J
7-/0
2
:
!
P£RtLNT
U3L
3/-38
5/-S5
l.a ~ 3 -
£-AT/5570/V CO,Q£ ^CTO^
ri/LL J.
£
3_
^.
j
6
_T
9-/0
il-ld
2'-20
3' -3 8
:r/-_rs
<5/-^3
3
|
FIT
11 DATA
OK z
TV PS,
V
5
fi/f.L
31-38
4
I
//-/$
Figure 2-3. Data Coding Sheet.
. PAOC- £..M.l±L;2
I !
33
-------
ANALYSIS OF THE EMISSIONS DATA
Using data and procedures described in the preceding chapter, the
relationship of industrial categories; (SIC code) to fuel and process emis-
sions for SO-, TSP, CO, HC, and NO,, was to be analyzed, In terms of fuel
emissions the relationship of employment and floor area by SIC code to the
demand for space heating and total Bill demand was to be examined. Likewise,
any propensity to use different fuels; was of interest as this might vary by
SIC code. As previously mentioned, several constraints existed in examining
the data. First, the distinguishing of emissions between fuel and process
was possible only for the New Jersey sample. On the other hand, the variable
"percent process heat" was available for the Massachusetts sample only.
This parameter is necessary to distinguish between space heating and process
heating uses of fuels. Finally, the sample sizes were not very good at the 3-
4-digit SIC level. In fact, the sample sizes were not very good at the 3-
digit or 2-digit levels as well. This made it quite difficult to analyze
very many subsamples or to perform meaningful statistical tests.
Table 2-3 shows the results of the analysis at the 4-digit level for
certain of the Massachusetts data. The first column show:; the SIC code and
the second column the total sample size for each of the SIC codes. The
next two columns show the mean for the parameter "percent process heat"
and the variance for that parameter as defined in the preceding chapter.
The next three columns show the sample size, mean and variance for the parameter
space heating in terms of BTUs per hour per employee. The following three
columns show the same variables for the parameter total heating BTUs per hour
per employee. The next six columns show the same variables, respectively,
for the parameters space heating BTU:; per hour per square foot, and total
heating BTUs per hour per square foot. The final four columns show the
propensity to use different fuels by SIC code. It can be clearly seen that
for the Massachusetts industries contained in the sample, residual oil was
by far the most commonly used fuel.
For industry group SIC 20, the data is shown at the 4-digit SIC level
with the first code being SIC 2010. A sample of only one existed here and,
therefore, no variance or analysis can be shown. For SIC 2013 a sample size
of five existed. The percent process heat is shown to be 88% with a variance
of seven. The space heating BTUs per hour per employee is shown to be 3617
with a variance of 1972. The total 3TUs per hour per employee is 30540
-------
o en o o o o
o o o o 0-.
CM
UJ
CQ
H
f- r- T o
< < 00 01 T M
Z 2 00
-------
with a variance of 9285. For many of the industries shown in this sample,
information on floor area in terms of square feet was obtained from a U. S.
4
Department of Transportation publication which relates employment to square
footage by 4-digit SIC. Accordingly, the means and variances shown in the
columns for BTUs per hour per square foot are directly proportional to the
means and variances shown in the columns for BTUs per hour per employee;
the only variation is that introduced by the varying square feet per
employee by 4-digit SIC as contained in the DOT publication.
In Table 2-3, only three 4-digit SIC categories were shown to have a
sample size greater than 10. These are SIC 2821 with a sample of 10, SIC
3069 with a sample of 13, and SIC 3111 with a sample of 14. In each case
some variance is shown for the parameter percent process heat, varying
between 19% and 24%. For SIC 3069 and SIC 3111 a sample size of eight
exists for the BTU per hour variables. For the space heating BTUs per
hour per employee and per square foot variables, significantly high vari-
ance is seen relative to the mean. In the case of SIC 3111 the variance
is actually higher than the mean. On the other hand, for the variables
total BTUs per hour per employee, and per square foot., the variance is
reasonably good, particularly in the case of SIC 3069. For these SIC
codes and for others which were examined, this type of pattern often
existed, leading to the tentative conclusion that the information on total
fuel use and, therefore, BTU demand, is probably more reliable and
accurate than the parameter percent process heat.
One would expect to find fairly good correlation between employment and
the heat demand for that number of employees, or between square footage of
plant area and the heat demand for that square footage. In other words heat
demand in BTU/hr/employee or BTU/hr/square foot should not vary significantly
from plant to plant within the same industry and climate. However, better
correlations were found with total fuel demand than with the demand for fuel
heating. This may be expla^r-cj by the inaccuracies for the variable percent
process heat: when th<* percent process heat shows very little variance, then
the relations!.-i-p between space heating BTUs per hour and total heat BTUs per
hour is shown to be quite good. This is true, for example, for SIC 2071 with
a sample size of four.
36
-------
On the 3-digit level SIC 201 shows the same information as SIC 2013
because the only valid information at the 3-digit level comes from that
particular 4-digit SIC. On the other hand, SIC group 208 includes the
information from SICs 2084, 2085 and 2086. Here it is found that for a
sample size of four the variance for space heating BTUs per hour is greater
than the mean, whereas for total BTUs per hour it is approximately one-half
the mean. This may be due in large part to the very large variance seen in
the percent process heat variable.
At the two digit level there is a sample size of 30, a fairly large
variance for percent process heat for SIC 20, and, again, variances greater
than the mean for both space heating BTUs and total heat BTUs.
Finally, if one looks at the variable space heating BTUs per hour per
square foot for all the 4-digit SICs shown in Table 2-3, one sees a variation
in the mean from 6 to 102 BTUs per hour per square foot, although this
should be a fairly uniform number if the data were accurate and the reporting
of the percent process heat, in particular, were accurate. In the full
sample there is even greater variation as exhibited by the variance to the
mean. This was to have been the base variable in the sense that if anything
could be thought to be uniform in terms of these parameters, it would be
the amount of heat required to heat "X" number of square feet of floor space.
The fact that one sees so much variation at the 4-digit level makes both
the use of this sample size and the reporting of the parameter "percent
process heat" in particular, suspect. The relatively good results shown
for the variable_total heat BTUs per hour underlines the weakness in this data.
Table 2-4 shows the entire sample for SIC 20 for Massachusetts. It shows
the variables percent process heat, space heating BTUs per hour, total heating
BTUs per hour, and employment. The data shown in Table 2-3 was derived from
this information. Figures 2-4, 2-5 and 2-6 show the plots of the variables space
heating BTUs per hour, and total heating ETUs per hour vs. employment.
This information was plotted to examine graphically the variance so as to
provide better insight into the analysis of the information shown in Table 2-3.
Figure 2-4 shows, on semi log paper, space heating BTUs per hour vs.
employment; whereas Figure 2-5 shows total heating BTUs per hour vs. employ-
ment. Finally, for the center area of the scale, Figure 2-6 shows on normal
graph pciper total BTUs per hour vs. employment. Again, the variable total
BTUs per hour exhibits better correlation than docs the variable space
heating BTUs pc-r hour, although intuitively the reverse v.oulu be expected.
37
-------
TABLE 2-4 MASSACHUSETTS TOTAL SAMPLE FOR SIC 20
SIC Code
2010
2013
2013
2013
2013
2013
2026
2026
2033
2042
2042
2052
2062
2062
2062
2071
2071
2071
2071
2084
2085
2085
2086
2087
2094
2094
2095
2095
2098
2099
Percent Process
Heat
100
97
90
81
80
90
60
88
68
N/A
N/A
N/A
95
N/A
N/A
62
50
55
70
0
50
90
55
N/A
90
98
100
N/A
N/A
N/A
Space Heat
105 BTUs/hr
2.2
5.2
28.0
7.6
9.0
91.0
23.0
24.0
N/A
N/A
N/A
100.0
N/A
N/A
32.0
1150.0
:>7.o
152.0
200.0
46.0
4.3
37.0
N/A
7.3
1.9
N/A
N/A
N/A
N/A
Total Heat
105 BTUs/hr
90.0
75.0
52.0
140.0
38.0
91.0
230. Q
200.0
7S.O
N/A
N/A
N/A
2100.0
N/A
N:/A
84.0
360.0
59.0
174.0
200.0
92.0
43.0
82.0
N/A
73.0
93.0
N/A
N/A
N/A
N/A
Employment
150
250
200
700
125
200
200
500
250
N/A
N/A
N/A
500
N/A
N/A
100
500
200
1109
262
310
52
250
N/A
40
58
50
N/A
N/A
N/A
38
-------
1400
1200
1000
800
CD
Q.
UJ
GOO
400
200
.2013
62071
2O/3
2026
2O85
2OI3 2071
- 2094
20/3
o
0 4-
2062 *207I
\2084
2O26
I ! I I
10'
BTU/!ir.
Figure 2-4. SIC 20 Space Heating BTUs Per Hour vs. Employment
39
-------
1400
c
Q>
O
Q.
e
LLJ
1200
1000
800
600
400
200
0
_ t
2O33 ,
2013*- **
ornf...n .
2071
20/3
2026* *207I
2084
2O85
*086
,, -2026
- o2071
20?C^ 20S1 ~2C:4
! ! 1 1 1 T I! \ \ \ \ \ 1 ! I
2062
\
\ \ 1 1 1 1 M 1
I07
I0b
BTU/Hr.
Figure 2-5. SIC Total Heating BTUs
Per Hour vs. Employment
40
-------
o
§
O
o
rO
O
O
OJ
o
o
o
4-1
C
0)
§
8.
r
00
§
UJ
t/)
1
1
u>
O
X
X
\
Z)
h-
tn
Cu
oo
0>
o
Cxi
u
0)
Q
CVJ
I
tN
0
^
W>
H
P-,
41
-------
Interestingly, in Figure 2-6, based only on a sample size of four, SIC 2013
shows a decent fit between the four data points, giving a nearly one-to-one
correlation between total heat demand and employment. The fit for SIC 2085
and SIC 2071 is not quite as good but is also promising, as is the case for
SIC 2094.
In summary, based on the information in Tables 2-3 and 2-4, as well as
Figures 2-4, 2-5 and 2-6, there seems to be some promise for finding good fit
between the variables under investigation. However, the sample sizes
and the lack of detailed information about how the data were originally
determined makes it impossible to prove or disprove any usefulness in
these correlations.
The next set of information exa-nined was the 4-digit SIC data for
New Jersey for fuel and process emissions per hour per employee and per
square foot. The data for SIC 28 at the 4-digit and 3-digit levels are
shown in Table 2-5, for emissions per hour per employee, and Table 2-6 for
emissions per hour per square foot. In each case, the first column shows
the sample size, mean and variance for total heating BTUs per hour per
employee. The remaining 10 columns show the mean and variance for S0_,
TSP, CO, HC'and NOY, respectively. For each SIC code the first line shows
A
the fuel emissions and the second line shows process emissions. Immedi-
ately preceding the SO,, mean column in parentheses are found the sample
sizes for the fuel and process emissions. At the 4-digit level SIC 2818
has a total sample size of 10 with valid data for nine sources for fuel
emissions and seven for process emissions. The mean for total heating
BTUs per hour per employee is shown to be 222,000 with a variance of 134,000.
Referring to Table 2-3 for SIC 2818 in Massachusetts, there was a sample size
of two and a mean of 74,000 total heating BTUs per hour per employee with
a variance of 53,000. This does not agree favorably with the 222,000
mean shown for the New Jersey sample, but is quite characteristic of the
results that were found for all SIC categories.
Looking at the emissions per hour per employee for fuel emissions,
the variance is greater than the mean for S0_ and'is less than the mean
for the other four pollutants. For TSP and CO in particular, the variance
is one-third to one-quarter of the mean. For the process emissions,
based on a sample size of seven, only NO, shows a variance greater than "the
mean. At the 3-digit level for SIC ,281 for a sample size of 19 the total
-------
UJ
PJ
6
,-j U
w 2
g !
OH *<
OH "^
1 "
PH
UJ
n.
SSIONS 1
PJ
CN
f-H
03
UJ
03
UJ
z
1
TABLE 2
tal BTU/hr/erap
o
i-
vo»-t coo o ooofOf^^ KIOCT^ *o *r
rg-rNr-i/i -.-- -^jrooior-j O1* --*
ooo oo oo «-< oooooo ooo oo
i O d fNto om O o^CTi-for^»o»-i LOi-Of~^ LT, rr
. t.(fN oo^ r-in co o^o*.rch*oQi ,-H\.D-T ^o
O OOO^t OtOO - 4OOOOO O O O O C; C C, O
O^ to O -« CTi tO O O vDOOOO^t OOTTO-rCOO
oOt o»OoO.-H f^OO vOOOOOcO OOCN. -if-r LO-T
,~. I to .i rsi o fNOCN'Or-j'T'.O--^
OQ p1-. CN » -i un » « 14 -- »
o
o ooocn ooo . «oooi^o oooooo oc
^r oo o CN cri LOO ooo Oio o \o i---*....,? mo rj -r r -
t-^ | ^- fO fTl *O O CN f-rt *O O ^T O r^* vO L/) OO \O O C"i i r, i_n n 1
tN rOvJD. < , t t-O oor-o
i fN in tofNtor-, i
^ oo .-i a. VD
t-4 ^-,
OO OO O -T--O O O OCN OO O OfN OO OO O O C~
oo r--o o cio to o rrro oo o oo oo oo ceo oc
oco .*o LO mo o o to ^t oo o o»o too oo OT oo
*<-t --H fN oo * r^ CN o r*- LO \GO
fM f-n (N i < r~j
oo oo o o^r o o oto oo o oo oc oo oo oo
OO OO O rJ-fN O O ONO OO C OfO OO CO OO O O
010 \oo o r-^\o r*- o or*- o\o cc o * oo oo f^v^: co
o'ooo. 4«-H-*t >-4 o o f-^i'Lnor'-iotovDotOr-.oc
O fN \D *-H O CN LO sO tO r^"; u-.
OOOOOx-D O O OOOO OOOOC 00
OOr-OO^^f CO i I OOOCO 1/1OOCCO too
r--fNcoco\DfN -^ i in OOO.-H . « CM c O m r--n
^J^O^O^-D »( r-H r-* o^oo« *f fN rO t-tOO vOU*ld- « ' \£>-T
r*~ t-n
O1 h* vOl/l tO IO i -^ i- 4. -I (OtO i t t~4 i i tOfN fOijQ CTlin LT, tO ('-co'
OOO O O OO CN O OO O O ^£> CD
oooorjcNo *tooooo-H o
^t to « co to CNO L.O o Ovo o i « to oo
totnoo.-o.-Hto in in 01 01 *? .-.
«- » to fM r-t o rN o
OOO O^O O^-T fN O LOO O O " O
ooooo \o a) m o i*-*o o c oo o
r-J «* to .» \.o ,-Hr-- o i- « rjn- a-. a. ^c. r i
ojr-\o GJ CN crj ^ CN r-to ^t rj cc oo
fNtOi «. i I-H fN v.D to L/~,
Ot vo rr to '...j ,- 1 to.-.-.jr.Htotoc'. v. r~~
fN I-H rj
o r- COT un .-i ^- 1 . ( to ,« ra^ ai . t oc L",
"-1 fN n ro
oo CTI _ *T ^^rj to **.( ci ,_....-j a^ . -* ^
oo cc oo to cooo cocooo oo ccco oo ri ri n
r' r- j >** i ri ri rj (j r^j CN CN rj rj CN
-------
u.
01
CO
o:
c.
o:
U4
O.
UJ W
<£ W
r_" tO
u-l
00
(M
u
KM
to
I
(N
OO1
C. 1
a o
H *-)
m x
x
o
_o>
C 1
a o
V i
x
oo
C 1
a o
H --4
> *
y
o
t-H
C t
rt o
""" X
O-H
o * <
c: >
a o
H _~4
f-,
_ ca x
^ Q>
\ rH
X §0
\ O
GrH
C I
cfl O
H i-H
ed X
O, r>
H o
rH
C I
CO 0
JJ rH
X
cu
ucn
C 1
ca o
H rH
Cfl X
rM
S
cn
£ i
o) O
4) rH
X
0
c
CO
H
i1 1
o -^
.c
B ">
* S2
"^ EX
o 0
>- -So
6 -H
r) LO
t/!
^ 0
rt 10
JH
* o
WU
^J
*» ro o
rH tO
rH CM
'"*
tO \O O
rH xf
^.
"* ^
O\ O rH CM
to o r^-
o TT
rH
r"- r^-
f*-) cn \o to
NO CM rM O
t-- rH
°1
o o o
*t fM 00
to to
rH CN
°i
oo ^ o
f^* cn o
*f 01
cn
cn o fM
NO 0 to
LO in
en
cn o o fM
OO \O rH OO
^ rH rH CM
^~
fi
r^ to vo <^j-
fM O CM 00
CM in LO
in
*"* ^f to in CM
NO Tt Lft LO O
r-i rH tO tO
rH ^J-
uo to
rH
o CN cn CN
CN NO tO CM
O *t rH fM
"
-H (N rH O
i-H
O tO LO OO
CO CO 00 CO
rj rj r-j ri
tO-HOCOO CiOOO CO
cn .0^ CNO m ooorHor*- KjrHto r *
rHJ^ rH rH rOCMOOf^-rHCN'^r
r-» rH OO CM «J
rH
NO NO r^. r^- cn CN r-.
tO ')Q ^ ^ rH »
* rH -Hr Trr^NOCM (OfMrH vO ^
oo 'tj-cn mr*x rH fMCNrncntooo otorM too
rH»O CM LOO> rHCM rHrH
LO
%O f*^ ^O CM O O O O C5 O CM
r-^tr-*^- rHOO '310000^ r^-int^-OrH ^oo
O*3- rH, O OOOCNNOOrHfNrH
fM f>- CNCMrH^fsJi <
m
rH
NO cn r^- LO r cn
LOLOtTrHtOCN I-O LO OO 'O°O t^ tO
oo rH r*j r*1- \o to CN in o LO o Is- \o oo o* ^j \o to i* LO TJ-
cn t-*> rH to cn rH ^-t \o cn ^r NO to rH cn rsi ^r
tO tO Ol tO rH rH rH
Cn rH
\O ' CN
h- rj NO cn o CM cn
fMO^rCN O COO O tOOOOOOOOO
OU^ rH rH f--tOO^fOrHrHO
o NOOcnorHTrrsi
CM LO in fN rM rH
t-* r--
rH fM cn CN
(--I/JO O fM O
rHND^-rH LO f-tco o toooo^ror-.r--
CntO rH CNrHOLOr^-OTj-O
oo r-orM^o\Oi-H
I/) 00 LO
rH C-J
CN
NO tO rH
Tj-OO OOrHO L/ICnOOOCN OO
KJrHCf tooOO f)rHOiDOOO^J-OOOCnrHOO
t*-rnrvrH \o o otocnLncMOLooorjto
CMCN CnfMrHOO'3' OO
t CM \O NO CN
NO NO
LO rHtOrHNO OO rH tO I/I (M Cn
. .*.. [S]
("-NOr-rMrHfMrN fJrHO)/^LOO tOOLO
\DNor-^to cn "d- cn
CNCnt*-- tO rH »O rH\O NOLOO'O-CN LO*3-
rH CO CN fM t-t
cn cn CM NO
CN CN O TJ- O O rH
rHOOtOCNO inOONDOrHtOOO
oomrHro in incncn^rH
(O CM rH O CM O
NO ^ CM m ^~
o \o Q} in t*^ oo
tOrH\OrH(^. OrHfMTtCnCnNOfN
NOcncNOfM rrfMr-tOTrCNOooo
tOrHrH rHfNNOtO LO
NO TTtO ^TrH -tO rH i-HIO tO Ch LO h-
fM "~"t CN
t^- COTJ- L/JrH rHrHtO rH CM*3- Cn rH OQ LO
CM CN tO
C, - - ^.-, « - ^ CT - fM 0.
OT OCO" C-'O CnoCOO CO XiOO OO OO CO &
r i r-i CN r i r j fN (N rM fN r-i oj CN CN CN r- 1
-------
BTUs per hour per employee shows a mean of 329,000 with a variance of
291,000. Again, for fuel emissions only S02 shows a variance greater than
the mean. However, for process emissions a larger variability is shown
with SO , HC, and NO all showing variances greater than the mean. One
Z A
would expect that the process emissions would show greater variability
than the fuel emissions.
Table 2-6 shows the comparable data for emissions per hour per square
foot. Again, at the 4-digit level the numbers are directly proportional
to those shown in Table 2-5, since the square footage data is based upon
the U. S. DOT relationship of employment to square feet at the 4-digit
level.
For those SICs that end in 9, such as 2899 or 289, the data would be
expected to show additional variation because the industrial categories
themselves are aggregations of disparate industries. If one examines SIC
289 it is seen that for a sample of 27, the mean for total heating BTUs
per hour per employee is 582 with a variance of 1018. For fuel emissions
both S0_, HC, and NOY show variances greater than the mean, whereas for
£ A
process emissions one also sees that S09, HC and NO show variances greater
-------
Figure 2-7 shows the plotting of S02 fuel and process emissions per hour
vs. employment for SICs 2810, 2813, 2815, 2818, 2819 and 28121. For fuel
emissions all but three of the sampled industries shown used 100% residual
oil. The percentages of residual oil vs. other fuels are shown on the graph
for the three other sources sampled. With the exception of these three
sources, a good correlation between fuel emissions and employment would be
expected, since fuel emissions should correlate well with total BTUs per hour
and vary with space heating Bills per hour only insofar as percent process
heat varies. A fairly good correlation is shov.n for SIC 2818 and SIC 2819,
the two A-digit SICs with a reasonable sample size. Insufficient data
exists for an analysis of the process emissions in Figure 2-7. There are two
data points for SICs 2818 and 2319.
In summary, as in the case of the Massachusetts data, no definitive
conclusions can be reached because of the sample sizes. However, where the
samples are reasonably good the variance relative to the means for the
fuel emissions seems reasonable. This is borne out by the information shown
in Figure 2-7. The expected variability and lack of information for process
emissions relative to fuel emissions was also found,
The variance between the SIC categories seems high enough to make the
analysis by SICs worthwhile. In other words, the variation shown in the
means at the 4-digit SIC level is quite high relative to the variance from
the mean for any particular SIC. As a result further research in this
direction should prove valuable in determining useful indices. In particular,
it is seen that the variances relative to the mean at the 3-digit level for SIC
281 are quite high relative to the variances at the 4-digit level for the SIC
281 series.
Table 2-7 summarizes the BTU per hour parameters analyzed at the 2-digit
and 1-digit SIC levels for the Massachusetts sample. The first column shows
the SIC code; the second, the sum of the New Jersey and Massachusetts samples
for each SIC category; the third column, the Massachusetts sample; the fourth
column, the percent process heat; the next three columns, the sample size,
mean and variance for space heating BTUs per hour per employee; the next
three columns, the sample size, mean and variance for space heating BTUs
per hour per square foot; and, the last three columns, the sample size, mean
and variance for total BTUs per hour per square foot. It shows the sair.e
type of information as Table 2-3 except at the 2-digit rather than the 3-digit
and 4-digit levels.
46
-------
|
<4
Lu
88
00
CD
*S
(O
I
O
o
x
CD
CL
CO
T3
C
3
O
CL
.b u
6
X
O
UJ
G
O
H
I/>
t/)
H
e
U4
ro
01
C
o
"to
to
'E
UJ
OJCQ
C\|Pg
C\J
00
i
r-j
0)
!H
3
bo
H
O
O
<£>
O
O
in
o
o
ro
o
o
CM
o
o
,o
-------
w
3
U-.
u
t t
tn
Q
I
(N
PJ
t/)
co
co
i
(N
as
IS
i-
01
CO
CM
ot
o
r-
^o
o
in
o
tn
* (tOOO
4-> (/I
C f>
-------
No systematic findings can be made; the BTUs per hour and per employee
vary quite a bit with some fairly high variances. In fact, it is not clear
whether one introduces more accuracy by using 2-digit indices than would
be found with a single index for manufacturing as a whole. The variability
between 2-digit SICs is high for both the total heating BTUs per hour and
the space heating BTUs per hour. Thus, it is not just a matter of inaccurate
reporting of the percent process heating variable.
Table 8 shows the propensities to use different fuels: residual oil,
distillate oil, natural gas and coal for both the Massachusetts sample and
the New Jersey sample for 1-digit and 2-digit SIC categories. From an
examination of Table 2-8, it is clear that variation in geography and supply
of fuels rather than SIC code predominate. There is a much higher propensity
across the board for SIC categories to use residual oil in Massachusetts than
in New Jersey. However, it should be noted that the lowest percent residual
oil occurs for SIC 34 in both New Jersey and Massachusetts, and that in all
cases in New Jersey where the percent of residual oil is over 90, the percent
of residual oil in Massachusetts is over 99%. It is difficult to examine the
fuel propensities in much greater detail because of the high dependence upon
residual oil for all industries sampled in both states.
Table 2-9 shows fuel and process emissions per hour per employee on a
2-digit SIC level for both New Jersey and Massachusetts data. The New Jersey
data is an extension of Table 2-5 but at the 2-digit rather than the 3-digit
and 4-digit levels. The Massachusetts data shows aggregated fuel and process
emissions since no breakdown could be determined from the sample. The
numbers in this table were used to derive the industrial emissions in the
Task 3 Report.5 Both the fuel and process emissions from the New Jersey
data were used as a guide as to whether an industry was an A, a B, or a C
classification and, further, whether it were a B-, a B, or B+ subclassifi-
cation. Where insufficient data for a particular pollutant was available
from the New Jersey data, the Massachusetts information was used but less
reliability was attached to it. Finally, for certain SIC categories (parti-
cularly SIC 23 and SIC 25) default information was used, based upon other
parameters in the table, with a general caveat that such industries were of
the cleaner variety.
In general, it was found that at the 2-digit level the variances are
quite high relative to the means although in the case of TSP the variances
49
-------
CO
_3
U4
li.
H
W
OS
w
u,
PH
tI
Q
W
oo
3
B
co
UJ
It
H
ii
co
w
o,
§
tx
OO
I
CN
OJ
J
OQ
<:
C3
tn
rt ra LO
Z in
CM
00
OO (N OO OO
in o
oo oo co
o o ca
<
o z
ooooo
o o
o o
-3 .-
.-i 4-J
c/l ^
50
-------
UJ
UJ
I
a,
OJ
UJ
ex
a:
1
UJ
a.
C.O
O
UJ
.J
CO
o
a.
1
CJ
X C CO
O CO i
'S. i O
CO
> X
OO
c o
ra i
CJ
y. x
o
"c.
rt
o
u
C t-
U rt i
X -H O
rt
> X
00
c/l CO
(H CO rH
_r* Q)
1 ~ X
V
>, <1>
O I
rH O-
o. s
a) to
tn Q>
XI O
«J C oo
O rt i
U -H O
> X
00
1
c o
rt I-H
E x
o
a.
to
X
CO
1
C O
CO rH
E X
o
a-
£j
to
o
0
CN C \O
O rt i
tO "H O
rt
> X
1
c o
-r x
o CN K> oo
T ui
ro
O O O *T O
O 00 O tO CO
O o vO »-H
r** ^-< *t
fNl K)
f-H
r-» irt ^f
rH
O O O
o o o
r^ ^ o
oo rj- -rr
*-< OO rH
0 00
oo to o
O 00 CO
O Tf (NI
r** ift _.
o r>j --H
i/l O CT>
*-* .-t M
O \O vO
o -a- Tt
ro ^
r- *t to ^t r-4
O O O O U")
O \.O rT LO rH
tO (N CS| \O
rH
O O O O fl O
ci o tr> to r^ o^
LO fs) 00 CM tO
00 rH >O Is- f*-4
^ r\i
t^ «-H I/) Psl CN rH
*-* ^1
o r^- ci ^H
*- r^ i>4
O \£> O^ *O -«O
urt »-H r-4 t^- LO
sD C^4 IN
O OO ^TO \o>*O
(M M i-H K)
r-. O
u c
3 r*
a -HO ' y u. a-. ^ o
QO oo o o tiajo
P-"< PrH« 3 *H , 21-
u-a. t/i , rZ.i/1 r_ a, ^ LI, t/i
i I m i i -n i I tr i i i/i
*~i n "~~; r3 *~: ~j *; r3
^ - - -- x » :: -^ s
f~J rj ri t^j
--H \D
-H
000
o m oi
o r-- oo
to
Oi
o o o
o m O)
o o r-
oo
rH,,
*
o o »-*
o o o%
~H VO CTl
o
o
o
Xf
o
o
10
rH
O ^t v-O
rH
O O 00
o o to
Cft 1*^ (N
irt o
rO rH
o o o
O O r-
o r- 'ct
U) t- rH
ro oo
rH
rH * \O
rH
O -cJ- 0
O d ^O
LO tO
i i
o o \o
O 00 -0-
TJ- IO
LO
O O TT OO -O
rH fM
i t CJ
. . <*> O
- U rH J r3 ^ rH J
S 2 . % o - 7 ^ g
u, a. v) u, CL ... a. a*
4 ' « 4 ' S 4 '
'' ^ x =- z.
rj r-t rj
(N
oo
CM
a»
rj
\O
. » (j
C* 3
i T
(/i
--'
51
-------
52
UJ
UJ
o
rJ
OH
UJ
OS
UJ
DH
O
X
U3
DH
CO
h- 1
CO
co
S
UJ
0
3
H
*-^
C
0
(J
V /
en
i
CN
UJ
03
H
H m
vO rf
o o
O 0
Ol O
*3- Ol
r-~ CN
l-H
O O
o o
CN vO
vO to
in CN
rH in
0 0
0 0
o o
00 CN
CN 00
rH CN
0 0
m o
r-H VO
rH in
in
r-H in
vo rr
0 0
r- o
O O
CN vO
to
^
t-~ 0
Tf 0
in in
in
^
rH in
o o
o o
in oo
o O O
o u^ o o m
OO (N O >^
^o rf rr
~-t o r**- rH co co
rH CM \O VO h- CM
r*- o o m o
(N CM tO
1 S ^ j 1 S ^ j 1 S rt j
C_ J ^C-"~*0 L_O. 7j
'. ' 3 2 . '. ' 3 i2 't ' _3 2
U) rl~ « J^ 1/1 ri ^
^ 3 i ;~ >~- -T; ^
O r-H CN
rO to to
O
o
o
o
in
o
o
o
o
to
CN
to
to
vO
^
00
^
00
to
l-H
0
3
LL.
t/>
(/,
^
O
o
o
o
in
0
o
o
o
o
o
in
to
to
U
o
"7
o
o
00
to
to
o
o
to
to
O 00
to OO
in in
CM
o to
vO vO
fO rH
CN
O O
t- O
VO O
t- CN
01
O 0
r^ o
t-~ o
rH f-H
vO
CN
O O
CM rH
CM CN
rH in
o o
o o
in CM
Ol to
in
Oi to
rH l-H
f- rH
^f vo
Ol
00 I-H
CN r-l
in
oo
CM
-H U
O O
3 rn
U. C_
1 1
^
0
O
0
o
i H
ID
O
IO
m
o
o
D
O
O
'31
'*
O
O
o
CN
If-
o
o
o
o
00
o
o
to
to
in
o
o
o
in
in
-
o
o
l-H
Ol
CM
O
O
0
l-H
rH
CN
CN
rH O
d> o
3 rH
1 1
*
tn
'* .
o
to
01
Ol
o
vO
vD
r-- o
in o
in o
Ol
o
CM
rH O
O O
00 O
Ol
rH
in CN
to r--
i-H O
O O
in in
CM l-H
CN T
in to
rr
to
o o
in o
l-H OO
rH OO
r~t
CM O
Ol C~-
vO CN
Ol
o o
r^ o
r- CN
CM OO
to
o o
m o
o oo
I-H r~
CM
O
OO
f-
CN
00
O
I-,
Csl
r^ 01
1-
»
rH
rH
[-^
l-H
CN
-H O
CO O
3 rH
U. O-
1 I
c/1
in
"*-
-------
Pi]
UJ
o
(X
UJ
CN
UJ
ex
a;
O
33
Q£
UJ
d.
to
IZ
o
[TTI^
C/5
w
1 1
s
UJ
nd
tD
3
H
P
c
o
0
N"~-/
O1
1
(N
w
_J
CQ
H
j
O O O fS O O
o or- o\ o
tO in -!
O O O *O O O
o o (Ni *t o r-.
^ Q) -*t K> sO O
o co in o oo
r^j -H f-f
o o o -< o IN o
HO O 00 LO 00 O
vD ^r 0*0 r^- fsi oo
i 4 t-H «0" i-H *3"
*o o CM
i4
oo o o> o oooo
Ol O O to O O Ol O
o^r OJ(NI^- oo to r--
f < \o i 01 i/> r
o o o
\D ^t ^
1 1
*d" O ^f
\O
tO t^ (N (S
f^
o o o o o 10
OO O Is* OO Ci vO
r- «o rj n*>
r^-to (NtN ^r^-t T±fM m in
OO tO K> OO .-ttO
(O \O i"H ^ 00 CN
LO o to ro «o o
LO to to r-- ~ ,-H
CN r-* r* '
r- or oo (N cc o
ro I-H
^^ '-* . ( O
00 .00
oo ~7"7 cio ^V^T oo
3t- 3r< .'3tH
J 1 £ J, 1 " J_ 1 «
LO ^C Iv
f: "* to
O O O O
^ d * d
(M O IM 0
rH rH rH LO
CM *O
ft
to o tv o
en o rH o
CO l-l tv O
LO rH
t 1 Ol
r~-t
Iv OOOO
Ol *£> O O O
Ol rH tO IV tO
rH Ol fO CM
LO en LO
en oooo o
rv o r-- o o o
tv O OO tO LO -*t
LO LO to \C
rv to en r-i
o o
in »
rvi o
CM
tv O tv
rv o >»
rH |v LO
rH
rv o O O
to to o o
CM CM O
00 rH Ol
\O O CM O
o oo en o
** rv yj o
LO Ol
rv
CMrH l-lrHOOtO tOrHCM
rH CM
CM
CM
O
^
00
to
M3
rH
LO
0
o
LO
CM
O
o
00
0
rH
|v
to
to
Ol
0
CM
^f
a-
o
fj
^
s'.
to
O
00
v£>
CM
\O
|v
to
o
o
^5-
\£)
0
o
-------
are in general less than the mean. Both the incidence and the accuracy of
process emission data are quite variable and no definitive conclusions can
be reached. Again, the more traditional point source pollutants, SO , TSP
and secondarily NO , show a more complete set of information. Examples of
A
the overall variability and unreliability of the data at an aggregated level
can be seen by examining the 1-digit figures at the bottom of the table.
For SIC 2 under SO , a value of 255 x 10 Ibs/employee-hour for fuel emis-
*- s
sions and 1570 x 10~ Ibs/ employee-hour for process emissions is found;
however, the sum of those, as exhibited in the Massachuestts total data, is
only 40 x 10~" Ibs/employee-hour. For SIC 3 the fuel emissions are 42 x 10"
Ibs/employee-hour and the process 10 x 10 Ibs/employee-hour whereas for
Massachusetts the aggregated number is 32.x 10 Ibs/employee-hour and for
those sources where separate proces.s emissions were available the mean was
142 x 10~ ibs/employee-hours. For SIC 2 the same kind of findings are ex-
hibited for TSP, whereas for SIC 3 Massachusetts values are of an order of
magnitude greater than the New Jersey ones.
Although reasonable results were not expected at such an aggregated
level, the variability that was found was quite disappointing. Moreover,
if one examines SIC 20 for S00 one sees that fuel emissions are 165 x 10
f\
Ibs/employee-hour from New Jersey data and total emissions 21.6 x 10 Ibs/
employee-hour Massachusetts data. In both cases the variances are greater
than the means. In general, there were too few 2-, 3- or 4-digit categories
that had enough information or showed similar findings to be able to make
either positive or negative conclusions about the accuracy or usefulness of
the different comparisons.
Accordingly, it was not fruitful to use this information to project
future emissions or to update the activity indices of the AQUIP System.
-------
PROJECTING FUTURE EMISSIONS
The previous AQUIP study pointed out that current data should be used
as much as possible to develop the future inventory as shown in Figure 2-8.
For the purpose of consistency, sources in the current inventory should be
carried forward to the future time period and only the most significant
new sources added as point sources; other sources should be added as area
sources. Regional and national protective data and "control totals" as
to fuel use, population, and employment should be used in conjunction v.'ith
the most reasonable activity indices. Many of these indices, such as the
heating demand per square foot, need not vary greatly from region to region,
except with variation in temperature. Others, such as propensity to use
different fuels, are highly a function of current uses in the particular
region. Fairly reasonable estimates can be made of the number of hours
of operation for each type of facility and for process heat for all land use
categories except industrial. Lack of information and tremendous variation
in this variable, as experienced in the point source inventory, affected
the results of the previous study. Finally, with uncertainty in inter-
national fuel supplies, even one to two years in the future, it was vir-
tually impossible to make reasonable estimates by land use category for
1990 as to fuel usage. In using the activity indices determined in the
previous AQUIP study, the planner is constrained by the national and
regional availability of fuel-use related data.
The projection of fuel consumption for 1990 made in the previous AQUIP
study was based largely on national trends. Little information is avail-
able on the different regional areas such as the New York metropolitan
area. Furthermore, it was beyond the scope of the previous AQUIP study to
undertake a detailed regional fuel projection analysis. Several nationwide
projections are available, the results of which are inconsistent with each
other. The majority of these projections were made before 1965 and all
projections make assumptions that are suspect.
An elaborate system was set up in the previous AQUIP study to project
percent process heating, schedule, fuel use propensity and process emissions
for existing New Jersey industrial sources to 1990. Indices derived from
current activity data for the individual source as well as data on current
erployces, enclosed spcice and gross pl;mt circa \;crc requested for each
55
-------
in
C
O
H
6
OJ
bO
c
H
4->
O
H
O
56
o
c
C) C
CD O
QCL
CO
0)
II
n >
U. .£
Sj»
c ^
O Q.
CL JE
-------
industrial source in the inventory. The data obtainable for a large number
of sources were the number of employees; therefore, this parameter was used
as the major projection variable. The availability of this parameter (and
the unavailability of other parameters) was confirmed in the AQUIP Exten-
sion Study.
For each point source the number of BTUs for space heat per hour per
employee was derived in the previous AQUIP study. It was assumed that this
parameter v/ould not vary significantly by industrial category; however,
when summaries were made by industrial category, wide variation was found
and no statistical conclusions could be drawn. This is, no doubt, due in
part to the inaccuracy in the percent process heat variable from which
the amount of space heating vs. process heating is derived. Again, similar
findings have occurred for the AQUIP Extension study, although the data
base has been larger.
Information was determined on the ratio of 1980 to 1969 employment
by county and SIC code from the New Jersey Bureau of Labor and Industry
in the previous AQUIP study. Many assumptions had to be made because of the
categories of SIC codes for which the data are available and the labor market
areas (cutting across county boundaries) for which information was assembled.
It was intended in the previous AQUIP study to project 1990 space heating
directly in BTUs per hour using the employment ratios and any assumed change
in the BTUs per hour and employee index. This would then be combined with a
new projection of percent process heat to yield total BTU heat demand for a
source for 1990. Accordingly, information on current percent process
heat was used to develop an index of percent process heat by SIC. This
parameter yielded two broad categories of industrial use. It was therefore
concluded in the previous AQUIP study that present information was not
sufficient to carry through the analysis as intended.
Initially, our intentions in the AQUIP Extension study were to improve
upon the data and to relate indices contained within the basic categories of
economic, geographic and demographic factors to emissions associated with
industrial activities. Intuitively, this approach is sound for residential
and commercial activity where heating of homes and transportation are the
principal activities which generate emissions. Here one could expect a
high decree of correlation bctueon such indices as "floor space," "number of
dialling units," "nu~Ver of occi'p''nt:s," "dc';ree-days," "passenger miles," etc.,
57
-------
and emission levels. In fact, suitable and accurate relationships have been
derived by applying regression techniques to actual observations for such
variables. In such cases the fuels burned, the technology of heating and
the technology of the combustion engine were fairly uniform throughout the
area and range of observations made. Furthermore, the technology of emis-
sion control associated with the technology of heating and transportation
were also fairly uniform.
Relative to industrial activity, one would expect that for any industry
(described by an SIC code) similar relationships between such indices as
"floor space" or "employment" and emissions might also pertain, provided
there was uniformity in:
1) Technology of supplying power and heat
2) Technology of the processes (manufacturing and production)
3) Technology of emission controls applied
Unfortunately, such a constraining condition is rare within any industry.
or group of industries. For example, if today one were to select any
industry that has the facility to supply its own power and heat, one will
find a vast' spectrum of fuels burned and applied technology. Figure 2-9
depicts schematically the range of possibility involved in this one instance.
Figure 2-10 illustrates in a similar way a limited range of possible process
variations in petroleum refining with just a small number of applied tech-
nologies. As a consequence, one would expect to find for any respectable
sample of facilities within any one industry a large variance between
activity indices which are purely demographic, geographic and economic in
character and emission levels. This is especially true when one takes into
consideration the vast array of emission-generating processes involved.
Indeed, observations associated with the analyses of variance recently
undertaken by others have vindicated this expectation.
As the aforementioned realizations become more apparent, the plan for
simply relating activity indices to emissions by regression analyses was
augmented to include constraints of power generation technology, process
technology (particular to a given industry) and applied emission control
technology. In accomplishing these investigations we would expect to:
1) Establish relations between economic productivity, demographic
and geographic variables and emissions for each industrial element
58
-------
x
bo
O
1-1
o
c
g
a>
E-
i-i
O
h
P
c
o
CJ
c
o
H
W)
V)
H
g
(X)
c
o
H
P
Bj
rH X
O O-Q
c o
0 i-H
CJ 0
fel
bO U
M a>
3 r-l
u. rt
o
CJ
o\°
in
o
V
r
r-l
O
rH
P
c
o
CJ
CM
o
CO
r-l
o r
M X
*> rH
C 0
O >
CJ 0
o
CM a>
O OS
CO
o
o :z;
VZ r
r e
o
4->
4->
O
oa
+J
O i-l CM Q
rH
O
!H
+-J
C
O
CJ
CM
O
co
^
X
rH
0
>
o
o
r- (
f
crj
i i
rH
B)
>
<
~~*^~~
BS
rH
o
H
W
>
0
0
0>
OS
o
g:
T3
C
rt
i/>
0
>
H
P
H
V
5
<
a
^j
rH
0
O
CJ
rH
H
O
T3
H
in
0
OS
o-.°
r-l
1
LO
O
A
rH
O
t.
H
4-J
C
o
CJ
01
O
CO
X
rH
0)
>
O
o
V
OS
X
13 rH
c c
rt o
10 1
a> o
> >
H -H
P -P
H -H
T3 X)
TJ XJ
<
c
o
CJ
>
rt
U/
rH \
Ct>
3
UH
rH
H
Tj t/)
H CO
W TJ O
O -H U,
Oi 10
O 4->
o\° OS rS
r-j u> tn
1 oS° -H rt
rH CM i-l CJ
J
B
o
I
4J
O
W
J
OH
LO
<
I
z
o
1I
H
2
w
w
CJ
OS
o
a,
P-,
o
co
O
\I
$
H
UJ
OH
a>
i
CM
W
OS
3
CJ
51
u.
0>
ca
59
-------
X
bO
O
f-H
o
c
o
0>
H
i-H
o
?H
4->
c
o
u
4J
c
CD
3
r~t
<*H
MH
UJ
\
o
H
t/1
I/)
H
g
UJ
r
bO
H
bO M
C ri
H i-H
M P-.
ri
i-H O
W- Z
V
/
+J
c
(H
4->
*o
x
i-H
H
o
1
I/)
ri
CD
V
fc
O
U
CD
05
CO
rr
V
o
o
(U
OS
CO
c
X
60
c:
H
bO XI
C 4:>
r-t 3
43 !-'
43 0
B U1
o c:
CO 0)
rt
4J
P
vi ri
o
^
(NO CD
K
4J
ri
CD
rH
+J
o
^
13
X
K
bO
H
C
o
iw
H
c
CD
4i
2
z
bo
c
H
^
o
l»H
CD
o:
o
H
M
x
i-H
ri
4->
ri
U
"2,
bO
C
H
to
in
CD
O
O
(H
a,
o
a,
j
4-)
sx
c5
C)
^ c>
«r
C)
*'
*s"
C
4->
H
.-«^_
W)
C
H
N
H
C
ri
43
4->
0)
CD
Q
/ *
ri
Column
in
rt
bo
i
|
3
O
rt
>
bo
C
H
H
H
§
P.
O
fH
PU
0)
Q
/ v
43
t/)
!i
O
4-1
0
rt
4-1
C
0
u
en
o
H
4-»
H
i-H
H
O
rt
U-.
bo
C
H
+J
ri
-H
X
^
r-H
<
i-H
O
O
>
ri
bo
i
e
3
0
rt
:>
o
z
c
0
H
4J
0
3
a
o
M
cu
i
X
43
CD
C
ri
X
V
K
\
to
JH
O
4-1
O
rt
4->
C
o
u
/
N
fi
o
H
+J
O
3
-a
o
!H
Pu
1
X
43
C
O
bo
O
fH
tJ
X
3C
bO
O
43
U
t/1
w
0)
o
o
tq
j
a,
CD
in
O
P
0)
I/I
I
s
z
II
U-,
UJ
oi
UJ
1-3
o
UJ
a.
i
tN
UJ
g
U-,
P;
H
4-'
re
ti
r-l
I/I
i-l
Cl
Wl
W!
Cti
a>
Ei
60
-------
(as codified by SIC) employing similar energy-generating and
process technology.
2) Establish relations between applied technology and emissions
for the industrial elements considered.
These observations would assist in establishing trends in the effects
certain technologies might have in either augmenting or alleviating
emissions and would form the quantitative basis upon which the forecasts of
future emissions might be derived by planners. Furthermore, such studies
would also entail predictions relative to the qualitative changes in emis-
sions that would be likely as a consequence of application of more advanced
technology.
Our intentions were to establish models that would employ available
economic and technologic information as well as an appropriate analytical
framework for such models. The analytical framework depends mainly on
a multiple regression approach wherever the data permit. For the purposes
of formulating structural models, the number of data observations were to
be developed on a geographical basis for use in the regression models. Since
consistent and comparable data on emissions and other related variables would
probably not be available for any time periods of significant duration, the
analysis to be carried out in this regard would be chiefly a cross-section
one on a spatial basis, rather than a time series against the two objectives
previously mentioned: (1) structural relations to explain the correlation
between emissions and related indices and technologies, and (2) prediction
of future emissions for use in planning formulations.
Development of forecasts would be carried out in two stages with
respect to quantity and to type. The latter would have to be developed
separately from the regression models since the models cannot be used in
predicting changes in composition or technology on a qualitative basis.
As previously mentioned, the analytical approach which was envisioned
involves fitting regression equations by the least-squares method of
estimation for different types of emissions. The success of this approach
depends upon the number of independent observations (at least ten) that
can be collected on a cross section basis for each type of emission within
the individual industries to be considered.
61
-------
A search for available informatian within various industry classifi-
cations that contained the distinction required relative to activity
indices, process technology, power technology and emission control tech-
nology, proved disappointing. All indications pointed toward a rather
extensive program of data-gathering by such means as interviews and question-
naire mailings since the existing data bases were so incomplete and incon-
clusive. Since neither time nor resources had been budgeted for this purpose,
investigations in this regard were curtailed. This direction appears to
hold considerable promise and such research should be supported.
Having worked with the 1970 Implementation Plan emissions data pro-
vided by the states has given the consultant some insight into the kinds
of data required to plan for urban and industrial air quality. While there
is sufficient information available to hypothesize planning procedures,
the quantity and quality of existing data are useful for only tentative
conclusions. It is suggested that future studies work within the framework
of this AQUIP Extension study, but that future results are now limited by
the availability of pertinent data. With this in mind, it is suggested
that the type of data inventory shown in Table 2-10 is necessary for further
studies of this nature.
Item 1
A point source code number is necessary for data handling since con-
fidentiality is important in dealing with emissions factors. Such a
code number would be used as a mechanism for storage and retrieval
by computer systems, of relevant air quality data. A numerical code
could be used at either the state or national level. A national cod-
ing of alphabetical and numerical figures would prove to be most
useful. For those states that already have their own coding systems
conversion to a national system would be relatively simple. For
mapping purposes x and y coordinates in some standardized coordinate
system would nlso be asked for.
Item 2
SIC codes are currently used in Implementation Plan data. Any
reasonable data file would incorporate SIC codes as a means of
grouping point sources by both land use and point source charac-
teristics. In future work, however, the use of SIC codes could
62
-------
TABLE 2-10
INVENTORY OF AIR QUALITY PLANNING DATA
FOR IMPLEMENTATION PLAN QUESTIONNAIRES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
so2
TSP
NOY
A
HC
CO
Point Source Code Number (assigned by air quality agency) MA- 1234
Standard Industrial Classification Code Number (s)
Year to Which Data Applies
Employment of the Point Source
Annual Hours of Operation
Floor Area of the Point Source
Annual Fuel Use
Percent of Fuel Used for Space
Degree Days at Point Source
Process Weight Rate (pounds per
(square ft.)
Coal (tons!
R. Oil (10, gal.)
D. Oil (10-5 gal.),
Nat. Gas (10. ft )
Heating
hour)
Solid Waste Rate (pounds per hour)
Percent Fuel for Incineration of Solid Waste
Space Heat
Emissions
Tons/Yr
a. 43.0
b. 12.0
c. 6.4
d. 8.2
e. 2.4
14. Separate
Process
Emissions
Tons/Yr
a. 0
b. 0
c. 2.1
d. 12.2
e. 3.1
2431
1970
215
6000
40,000
a. 0
b. 155
c. 0
d. 20
60
4000
100
0
0
15. Solid Waste
Emissions
Tons/Yr
a. 0
b. 0
c. 0
d. 0
e. 0
63
-------
be greatly expanded. In this particular study the inventory
of manufacturing point sources was relatively small. This made
it necessary to use 2-digit SIC codes to group industries with
simlar point sources. These 2-digit groupings are in large part
responsible for large variances because of the tremendous diversity
within 2-digit SIC groupings. Future work should be done with a
sufficient base of data to fully use the SIC coding system.
The SIC codes for industries could be used on a 2-, 3-, or 4-
digit basis depending on the size of the data base. The 2-digit
codes which have been used in this study (e.g., SIC 23) identify major
industry groups such as "manufacturers of food and kindred products."
At this stage there was only enough information to compile a marginal
statistical sample for most industrial groups. With an expanded
set of point sources and with increased data completeness, 3-digit
codes (e.g., SIC 231) which identify subgroups within an industry,
or even the 4-digit codes which specify industries by specific
products (e.g., SIC 2311 meat packing plant) could be used. As the
code became more detailed so should it be expected that air quality
planning parameters would become more accurate..
Item 3
The year to which data applies is merely a "bookkeeping" measure.
Since data will be solicited and checked periodically some means of
insuring consistency in time periods is required.
Item 4.
Employment is one of the two major units for quantifying activity
indices in the planning proces.s. Production rates, population,
and economic grov;th are all dependent upon employment to seme
measure, yet Implementation P.'.an data fails, to give employment by
point source. In the course of this study the emplo}inent figures
provided relatively major data management problems. It was neces-
sary to seek employment estimates from multiple sources. In some
cases questionnaires from the sources outlined the employment
figures. In other cases it v.as necessary to petition state agencies,
-------
such as the New Jersey Department of Commerce and Labor for employ-
ment figures.. These figures, when available, were of questionable
reliability for several reasons. First of all, the employment
may or may not correspond to the time period for which the Imple-
mentation Plan data is valid. Secondly, the employment may not
correspond directly to the point source (i.e., a manufacturer
may produce two or more products at different locations within the
state but the state agency will list employment by corporation,
rather than by point source). All the variables dependent upon
employment were affected by the quality of the employment figures
that had to be used in the study. Statistical analysis categories
that were affected by employment figures include the following:
1) "FUEL BTU/HR-EMPL" - the activity index which indicates
a propensity to consume energy. (This might also be phrased
as a unit demand for fuel in BTUs per manhour.)
2) "FUEL AMT/HR-EMPL" - is the activity index which describes
the unit propensity to emit pollutants in pounds of pollutant
per manhour for space heating.
3) "PROC AMT/HR-EMPL" - is the activity index which describes
the unit propensity to emit pollutants from industrial processes,
in pounds of pollutant per manhour.
Briefly, then, three major projection indices are directly affected
by the quality of employment data.
Item 5
Annual hours of operation are presently included in Implementation
Plan data. Operation hours are necessary to calculate unit time
figures for projection. The primary projection units (grams per
employee hour and grams per enclosed floor area hour) require some
means of time normalization. The annual hours of operation provide
this means.
65
-------
Item 6
Floor area is the only index, other than employment, that can
currently be used to quantify emissions into a systematic unit
grouping for planning calculations and projections. Implementation
Plan data are void of floor areas, however. In fact, preliminary
investigations show that there are no figures on floor area avail-
able from any source. This directly affects the remaining set of
activity indices.
1) "FUEL BTU/HR ENC" - is the fuel demand per unit time and
floor area.
2) "FUEL AiMT/HR ENC" - is the unit propensity to pollute per
unit time and floor space for space heating.
3) "PROC AMT/HR ENC" - is the unit propensity to emit industrial
process emissions per unit time and floor space.
Since no floor area data were available to input directly into
statistical analysis, rather than abandon consideration of floor
area as a useful tool in calculating and projecting emissions,
floor areas were synthesized from employment numbers. While this
provided some unit activity indices incorporating floor area, it
must be realized that all floor area data is limited by the quality
and availability of employment figures (since it is derived from
employment). Where employment is missing, floor area is missing;
where employment is not accurate, neither is floor area; but even
with reasonable employment date., floor area validity cannot be
assumed. Average floor height could also be obtained so that heat-
ing demand reflects building volume as well as floor space.
Item 7
Annual fuel use is currently included in Implementation Plan data.
Since fuel use is a major variable in pollutant emissions its
inclusion in data files is imperative. Current, detail in providing
fuel data is adequate, however,
66
-------
Items 8 and 12
The allocation of fuel use is particularly relevant to planning
for air quality. The knowledge of how a fuel is used helps project
the level of emissions by specifying the combustion characteristics
and also by providing a means for estimating related plant activi-
ties. Traditionally, air quality inventory data has included per-
cent process heat and percent space heat as the two means of fuel
use. It seems that probably three categories of fuel use may be
warranted.
1) Percent Space Heat Fuel has generally been estimated in the
past. The quantities of fuel required for space heating
should be fairly predictable and generally unrelated to
SIC code. Space heat is dependent upon enclosed floor
area and heated volume for the most part.
2) Percent Process Heat Fuel has generally been taken to be the
total annual fuel consumption less the space heating fuel.
This process fuel use has been the index for estimating
process heating emission levels. The process heating
emissions are those that may be expected to differ from
industry by SIC codes. The greater the fuel use the
greater the process emissions that are anticipated.
3) Percent Incineration Fuel is not currently a data file vari-
able. In many cases solid waste may play a large part in
total point source emissions, yet the use of fuel for incin-
eration is unquantified. While this may be unimportant on
the large scale it may be quite important for local considera-
tions where solid waste disposal is a significant element.
item 9
Degree days are not available from Implementation Plan data, yet
are a necessary element in projecting space heating requirements.
Given the number of degree days and the floor area of a plant, a
reasonable space heating fuel estimate can be made. The accumu-
lation of degree-day data would be relatively easy. By knowing
67
-------
the location of each point source, degree-days for each point source
could be determined using data from the weather service and from
local distributors.
Item 10
Process Weight Rate is currently included in Implementation Plan
data inventories. This information is useful in forecasting process
emissions. Its current use and availability is adequate.
Item 11
Solid Waste Rate is necessary for projecting solid waste incinera-
tion emissions. Assuming that certain similar products have
certain similar wastes, it can be seen that as process emissions
are characteristic of industries (by SIC code), so are solid waste
emissions by product. It must be realized, however, that not all
facilities with similar production activities can be expected to
have similar solid waste characteristics; some may have all wastes
trucked away, some may burn all wastes, and some will be somewhere
between the two extremes. Solid waste, then, becomes; an independent
variable for which data is required.
Items 15, 14 and 15
Space heating emissions and separate process emissions are currently
catalogued in Implementation Plan data. There is, however, no
current classification for solid waste emissions, or they are
recorded as process emissions. Where there are process emissions
there is no means for distinguishing which of the emissions are
due to separate process and which are due to solid waste burning
or incineration. Stack parameters - height, exit velocity, tem-
perature, etc. - would also be. useful information for diffusion
studies.
68
-------
REFERENCES FOR PART 2
1) Willis, B. H., et al, The Hackcnsack Meadowlands Air Pollution Study,
Summary Report and Task Reports, Tasks 1-5, prepared for the
State of New Jersey, Department of Environmental Protection,
Trenton, 1973.
2) See the following: Kircher, D. and D. Armstrong. An Interim Report
on Motor Vehicle Emissions Estimation, USEPA, Office of Air Pro-
grams, October 1972; and Compilation of Air Pollutant Emission
Factors, USEPA, Office of Air Programs, Publication No. AP-42,
February 1972 and April 1973.
3) See the reports produced for the Land Use Planning Branch, Office of
Air Programs, by Argonne National Laboratories, 1970-1973.
4) See Estimating Land and Floor Area Implicit in Employment Pro-'
jections, Vols. 1 and 2, U. S. Department of Transportation,
Federal Highway Administration, 1970.
5) Epstein, A. H., et al, A Guide for Considering Air Quality in
Urban Planning, prepared for USEPA, March 1974,
Contract No. 68-02-0567.
6) See the following reports by the Hudson Institute: "Study on the
Corporation Environment, 1975-1985," and "Energy and Energy Fuels,"
69
-------
PART 3
TASK 2 REPORT
COST-EFFECTIVE PLANNING FOR
ACCEPTABLE AIR QUALITY
by
Alan H. Epstein
70
-------
SUMMARY
This document presents a methodological approach to planning for and
evaluating the impact of land use on air quality. The planning process is
viewed as a series of sequential steps in which the economic implications
of planning decisions are evaluated in terms of their dollar value impact
on air quality. In this way postulated plans may be designed to be compat-
ible with both air quality criteria and various development preferences.
In addition, because running tallies of both benefits fostered by a given
plan and the resultant costs of air pollution control and damage are kept
for each planning option considered, alternative plans may be conveniently
compared and ranked according to how effectively, from a cost standpoint,
each utilizes the air resource.
The material presented is divided into three distinct but closely
interrelated chapters. The first chapter discusses the essential concepts with
which the methodology attempts to deal. The second chapter presents the methodology
itself, and the third chapter provides some application guidelines.
It is not intended that this document, of itself, be sufficiently com-
prehensive to enable the planner to implement the concepts presented. Rather,
it is expected that considerable effort will be required to fully research
individual areas of concern so that information sufficent to apply the
methodology is compiled. Furthermore, because specific applications of
the methodology are heavily dependent upon available site specific data,
the general technical and operational capabilities of the individual planner
or planning group, and the ever present constraints of time and money to do
detailed planning studies, it is anticipated that both the level of detail
and degree of sophistication of individual applications will be widely varied.
It is intended that this document, with its suggested analyses, be
used as a guide by the planning community in making and evaluating planning
decisions. It is hoped that sufficient ingenuity will be brought to bear in
the application of the concepts presented so that each use of this methodology
represents an accurate appraisal of actual physical and economic phenomena.
71
-------
INTRODUCTION TO PART 3
BACKGROUND AND PROBLEM DEFINITION
Because the urban planning process offers a fundamental means of
controlling long term air quality, it is necessary that the planner be able
to evaluate alternative land use plans in terms of their relative air
pollution impact.
The air quality of given regions or sub-regions depends upon two sets
of phenomena:
1. The assimilative capacity of the air environment for pollutant
material
2. Pollutant emission characteristics.
The capacity of a given air environment for atmospheric pollutants is
determined by ambient meteorological, topographical, climatological, chemical,
and biological conditions. In general, very little functional control can
be exercised over these phenomena on a regional scale. Consequently, it
does not appear practical to postulate a regulatory program of air quality
based upon the specification of desired conditions of naturally occurring
physical phenomena.
Pollutant emission characteristics include the quantity of pollutants
released to the atmosphere, the physical location of emission sites, and
source types (e.g., mobile, stationary, elevated, arid depressed sources).
These characteristics are determined from the specification of land use
type, level of activity or process rate, types and amounts of fuels used,
source controls, and activity schedules.
Because pollutant emission characteristics are directly derivative
from the specification of the mix, locations and intensities of land use,
it is evident that the urban and transportation planning process offers an
effective means of controlling long term air quality. Furthermore, both
analytical techniques and empirical data for estimating the relationships
between land use and air quality have reached the point of making air
quality impact-land use planning a practical tool in helping to manage
urban growth and development
For the planning process to effectively accommodate the requirements
of an expanding population within a limited air resource,, it is necessary
that the planner be able to differentiate between and evaluate alternative
plans in terms of their relative air pollution impact. The problem then
is how to apply existing analytical techniques and data for quantifying
the impact of land use on air quality to an evaluative methodology for
estimating how effectively the air resource is used.
72
-------
THE PLANNER
AND THE
PLANNING PROCESS
1
LAND USES
Mix
Intensities
Locations
ASSIMILATIVE
CAPACITY OF
THE AIR
ENVIRONMENT
POLLUTANT EMISSION
CHARACTERISTICS
Quantity
t Locations
Source Types
EMISSION
CONTROLS
AIR QUALITY
Figure 3-1: The Dependence of Air Quality on Land Use - Shows how
the planning process can contribute to the determination
of air quality and suggests that the planner needs a way
of evaluating the consequences of planning decisions in
terms of air pollution impact.
73
-------
THE ECONOMICS OF AIR POLLUTION
Cost implications provide the most tangible and immediate means of
evaluating the air pollution impact of alternative land use plans.
Air pollution may be viewed as the result of using the atmosphere as
a waste disposal medium for the urban processes supporting the economic
structure of a region. Because the air resource is not limitless, it is
apparent that a means of evaluating the effectiveness of 'spending1 this
resource is important to the decision making processes which prescribe
urban growth and development. The ideal measure of effectiveness would
be a single quantitative indicator which is generally applicable to the
spectrum of planning variables as ar optimization parameter. On this
basis, cost effectiveness is the obvious comparison of proposed alternative
land use strategies relative to their air pollution impact, as well as to
the other constraints within which the planner must operate.
There are three broad areas of economic concern identifiable within the
scope of a cost effective approach to evaluating regional air pollution
impact. They are:
1. The trade-off between air quality and those urban activities
which determine regional economic viability
2. The costs incurred in controlling pollutant emissions
3. The costs of damage resulting from expected air pollution levels.
Their interrelationship with the planner and the planning process may be
represented as shown in Figure 3-2.
represented as shown in Figure 3-2.
The first of these is important to the planning process because it defines
attainable limits for proposed land uses in terms of both air quality and what-
ever measure of economic viability the planner chooses. Federal and state air
quality standards and regulations have, in effect, placed constraints on the mix,
intensities, and locations of the various land use categories, particularly those
involving heavy industry and motor vehicular transportation. These constraints re-
quire the planner to reconcile the pressures for economic development, with their
anticipated impact on air quality, to a level of detail encompassing the spatial
allocation of both the sources and receptors of air pollutants.
Collectively, the cost of controlling pollutant emissions and the cost
of air pollution damage define the total cost of air pollution for a given
land use plan. Evaluation of total cost relative to the flow of benefits
inherent in the land use configuration which gives rise to these benefits
is the basis for determining the relative air quality impact cost effective-
ness of alternative plans.
Quantification of allowable Umits to preferred land uses and cost ef-
fective evaluations of alternative proposed land use plans generated within
these limits may be taken as the initial and final steps in a methodological
approach to cost effective air quality impact land use planning. Subsequent
sections of this chapter discuss individual aspects of air pollution economics
as they peitain to land use planning, and subsequent chapters outline the meth-
odology indicated above and present a set of guidelines for its application.
-------
1
TRADE OFF: AIR QUALITY VS. URBAN ACTIVITY
1
Planning Constraints
Air Quality
Economic Viability
I
The Planner and the Planning Process
Land Uses
Assimilative
Capacity of the
Air Environment
Pollutant
Emission
Characteristics
Emission
Controls
Air Quality
2
AIR POLLUTION
DAMAGES
I
3
THE TOTAL COST
OF AIR POLLUTION
THE COST OF
CONTROLLING
POLLUTANT
EMISSIONS
Figure 3-2.
The Economic Implications of Land Use Planning in Terms of Air
Pollution Impact - Shows how three basic economic concerns may
be combined to provide both constraints to and a means of
evaluating land uses in terms of air pollution impact.
75
-------
Trade-Off: Air Quality vs Economic Viability
Because land use activities which promote a healthy regional economy
tend to degrade regional air quality, there is a trade-off between these
two basic planning goals.
It has long been recognized that activities which are conducive to a
healthy economy can, and often do, result in unwanted side effects. Known
in economic linguistics as external diseconomies, these effects generally
demonstrate a positive correlation with the level of their causal activity.
Air pollution is an outstanding example of an external diseconomy because
it is the unwanted result of urban activities which ultimately exist to
enhance the quality of urban life. Further, it generally increases as
the levels or intensities of these activities increase. The result of this
cause and effect relationship on a regional scale is an overall decrease
in air quality as a function of increasing urban activity,
Given that some general inverse relationship exists between air quality
and economic potency, the planner must at the outset reconcile these opposing
planning goals. On the one hand, severely curtailing those activities which
are major contributors to air pollution may provide for good air quality
but may, at the same time, so severely limit both productivity and mobility
that the regional economic base may be eliminated. On the other hand,
vigorous pursuit of an effective and vital economic base may cause air
quality to be unacceptable. It is essential that the planner be able
to quantitatively establish these extremes at the outset so that both
good air quality and an effective economic base are 'built-in' to proposed
land use schemes.
Considering the reliance upon the combustion of fossil fuels as the
major source of energy, there will s.lways be some level of urban activity
beyond which acceptable air quality is simply unobtainable. It should be
recognized however, that emission controls offer considerable latitude in
specifying activity types and levels which meet air quality criteria. As
indicated in the accompanying figure, control strategies have the effect of
either increasing the capacity of a given environment for urban activity at
a given level of air quality (control strategy C^), or increasing the level
of air quality for a given level of urban activity (control strategy GS),
or some combination of the two (control strategy C ).
Unfortunately, there are some very real economic constraints associated
with the degree of emission controls; that a given land use strategy can
tolerate. For example, if a plan is; proposed which attempts a very high
level of emission control (in order to allow, for instance, for high industrial
employment), it may become economically impossible for certain of the
activities which are being relied upon as potential employers to survive
on a competitive basis. The next section discusses in further detail the
effects of emission controls and air pollution damage in terms of achieving
regional economic viability.
76
-------
JO / .9/I.97
rfwouooj /Duo/Bdy ijoddng 04
>%
>
tj
c
o
.a
i_
~D
»4
O
"a)
Figure 3-3.
The General Relationship of Level of Urban Activity to Air Quality
Shows the options available to the planner in terms of both air
quality and urban activity through the specifications of emission
control strategies.
77
-------
The Cost of Achieving Ecomomic Viability in Terms of Air Pollution Impact
For a given land use configuration, the cost optimum level of attainable
air quality may be determined from the combination of costs resulting from
the implementation of various control strategies and the corresponding
costs resulting from air pollution damage.
The cost of achieving regional economic viability in terms of air
pollution impact may be defined as the return received (e.g., in regional
employment) for each dollar spent on air pollution. Quantification
of this parameter for alternative proposed land use plans may be used as
the basis for performing cost effective evaluations and consequent ranking
of the alternatives in terms of their relative air pollution impact. It is
therefore essential to be able to determine the total cost of air pollution
for each of the proposed alternatives which have been postulated to provide
for both regional economic viability and acceptable air quality.
As indicated in the accompanying figure, the total cost of achieving a
given level of air quality for a gi^en land use plan is the sum of the costs
resulting from controlling pollutant emissions and the corresponding costs
resulting from air pollution damage. Ideally, the planner should attempt
to attain, for each of the proposed alternatives, that level of air quality
which results in the minimum total cost of air pollution. This would re-
quire specification of an emission control strategy which provides for the
cost optimum level of air quality. However, as a result of satisfying either
basic planning constraints or the requirements of legislated air quality reg-
ulations, it may well be that the planner has had to specify emission controls
as an integral part of one or more of the alternative plans. If this is the
case, and if the control strategy is such that it becomes impossible to
operate at the cost optimum air quality level for that land use plan, addi-
tional emission controls should not be specified. The point is that the
planner should always attempt to operate as close to the minimum total cost
as possible for each plan considered because the utility received is inherent
in the specification of the plan.
By the same token, a plan which does not operate at the minimum total
air pollution cost should not necessarily be abandoned. If the utility
received from the plan is high, the cost effectiveness evalution may
indicate that it is the most desirable alternative. Again, it is not the
total cost of air pollution which is the indicator of relative worth, but
rather the air quality cost effectiveness. This is the most critical
concept involved in the economic evaluation of air pollution impact and its
importance cannot be overstressed. Assuming that it is necessary to spend
air quality to buy economic viability, it is essential that the maximum
return be realized for the damage fiat is done.
78
-------
^Unacceptable..
Quality.]
Total Cost
Curve
Cost of
Control
o
o
Level of Air Quality
Figure 3-4. The Total Cost of Air Pollution - Shows how the curves of
control and damage costs for a given plan combine to form
a total cost curve the minimum of which determines the
cost optimum level of air quality.
o
CD
79
-------
The Cost of Controlling Pollutant Emissions
The cost of air pollution control is the economic indicator defining
the aggregate ability of the public, corporate, and ultimately the individual
sector to supply clean air.
Legislated air pollution control strategies and standards effectively
shift some of the direct cost burden of air pollution from the individual
to the corporate and public sectors. That is, rather than the individual
having to bear the direct burden of air pollution damage, public and cor-
porate facilities are required to assume the responsibility for air pol-
lution control. Other things being equal, internalization of air pollution
costs results in a decrease in net productivity. In turn, the ability to
shift this burden back to the individual will generally determine the via-
bility of public and corporate facilities within the structure of the re-
gional and national economy.
For the public sector, the cost of controlling air pollution results
from the combined costs of directly controlling pollutant emissions from
municipally owned sources (i.e. power plants, incinerators, public trans-
portation facilities, etc.) and from the costs of operating and maintaining
air pollution control and regulatory services and facilities (e.g. enforce-
ment, research and development, monitoring, information services, litigation,
etc.) In as much as these facilities are publicly owned, the cost of
pollution control is generally parsed on to the individual in the form of
increased taxes or usage rates.
Air pollution control strategies and regulations will have a signifi-
cant effect on the corporate sector as well. The final effect of pollution
control on quantity of output, facility location, profits, and consequently
prices will generally depend upon a) pricing policy of the industry in
which the individual firm is located, b) the direct cost of abatement
in terms of equipment required to meet emission standards and the operating
and maintenance costs of that equipment, c) the structure of the industry
(other than pricing policy), d) the demand elasticity for the firm's product
and, 3) the structure of the market. In order to counterbalance the effects
of internalizing pollution costs, individual firms will try to shift as much
of the burden resulting from the decrease in productivity as is possible to
the individual in the form of higher prices.
Ultimately of course, the largest burden of air pollution control costs
will be borne by the individual. While direct cost increases to the public
and corporate sectors do not, in general, translate dollar for dollar to
price and tax increases to the individual, for purposes of establishing a
relative system of accounts to perform cost effective analyses of alternative
land use plans, these cost increases are sufficient. The justifying assumption
for this statement is that the factors defining multiplier effects in the
regional economy will not be significantly altered by differences in growth
options. The cost of controlling pollutant emissions msiy then be defined in
terms of direct costs to the corporate and public sectors.
80
-------
LEGISLATED
Air Pollution Control
Regulations and Strategies
1
Cost to Public Sector
t Maintenance and
operation of
regulatory agencies
Emission controls
for public
facilities
1
Cost to Corporate Sector
Depends on
Pricing Policy
Abatement Costs
Industry Structure
Demand Elasticity
Factor Market Make Up
Indirect Costs
Higher Taxes
§ Usage Rates
.Direct
Costs
I
1
I
Direct Cost
of
Air Pollution Control
f
Indirect Costs
Higher Prices
I
Total Cost of Air Pollution
Control Borne by the
Individual Sector
Figure 3-5. The Cost of Controlling Pollutant Emissions - Shows how the
individual costs of pollution control may be considered for
purposes of defining the total cost. The ability of the
individual sector to support this burden may then be viewed
as the aggregate ability of society to supply clean air.
o
*
o
81
-------
The Damage Function
Air pollution damage functions define the extent of aggregate dollar
value damages attributable to polluted air.
The total nation-wide cost of air pollution damage for 1968 has been
estimated at approximately 16 billion dollars. This figure, while admittedly
a rough approximation, nevertheless, demonstrates that air pollution damages
are both quantifiable and significant. More recent studies have estimated
national damage costs ranging upwards of 300 billion dollars, indicating a
more complete understanding of the effects of air pollution as well as a
broader range of analytical capabilities with which to define its economic
consequences. As the body of knowledge concerning the effects of polluted
air continues to grow, it is expected that damage estimates will increase
even further.
Existing studies into the construction of air pollution damage functions
have postulated the following indicators as those best suited to establish
the deleterious effects of polluted air:
Health Costs
Materials Damage
Property Devaluations
Vegetation Damage
Soiling Costs
Animal Losses
Asthetic Effects
Litigation Expenses
Of these, sufficient data exists only for attempted quantification of
the first four. Even for these categories, data limitations have prevented
construction of reliable empirically based damage functions. Those postu-
lated to date are pro forma representations of indicated trends generated
through multivariate statistical analyses. In addition, it has not as yet
been possible to postulate an aggregate damage function of all pollutants.
Rather, damage functions are currently expressed by pollutant.
Data limitations notwithstanding, information currently available to
estimate per capita damage costs a:; a function of pollutant concentrations
is sufficient to indicate the comparative air pollution impact of alternative
land use schemes, particularly for the industrial pollutants, i.e., SCL
and particulates.
82
-------
° 200
Q.
O
CJ
i_
OJ
Q_
<**
V)
O
O
O
£
0 100
a
3
C
C
<
CL
O
O
k_
0)
Q_
ii
>
0
rS02=-$5.90+0.66 X
Particulates
= - $4. 7O + O.47X
100
200
X=Weighted Annual Average Pollutant Concentrations (ug/m )
Figure 3-6. Pro Forma Damage Functions for S02 and Particulates - Show
the effect of pollutant concentrations on annual per capita
damage costs.
VC
K5
O
CO
83
-------
LITERATURE SURVEY AND EVALUATION
Although a considerable amount of material is available concerning each
of the various aspects of air pollution economics, a comprehensive methodolog-
ical approach to cost effect air quality impact planning is not available on
a level commensurate with both the requirements and constraints of existing
planning practices.
Available literature relative to the economic implications of air pollu-
tion may be categorized into the three individual areas of concern indicated
previously:
1) The trade off of air quality with urban activity
2) The damage functions
3) The cost of controlling pollutant emissions.
Information relevant to the first area of concern is fairly abundant
and relatively easy to obtain. Unfortunately, much of the available litera-
ture is descriptive in nature and, for the most part, reflects efforts to
implement various control strategies for existing land use configurations.
The outstanding examples of implemented studies in this area, and those for
which the preceding generalization does not apply, are referenced in the
accompanying table.
In terms of documented studies attempting to establish empirically based
damage functions, very little has been accomplished to date. Conceptually,
this area has been relatively well defined; however, difficulties in compiling
data have as yet precluded the availability of anything but pro forma repre-
sentations. It is anticipated that extensive research efforts currently under
way will provide more suitable information.
Since the adoption of the Clean Air Act, considerable effort has been
spent on defining the costs to be incurred in controlling pollutant emissions.
Much data is already available and this is being supplemented on an almost
daily basis.
As discussed in the Summary, it is not expected that the application
of the concepts presented here be accomplished through consideration of
this document alone. Because of a multiplicity of factors involving site
specific concerns of individual areas, the capabilities and limitations of
various planning groups, the availability of representative data, and the
relative position that air quality occupies within the priority structure
of a given state, region, or municipality, it is anticipated that extensive
literature and agency surveys will be required of potential users. It is
therefore suggested that references cited in the accompanying table be
obtained and examined as a first step in performing such surveys.
-------
TABLE 3-1. REFERENCE MATERIAL
1. "The Arizona Environmental and Economic Trade-Off Model." Columbus,
Ohio: Battelle Laboratories, September 1972.
2. Barrett, L.B. and T. E. Waddell. Cost of Air Pollution Damage: A Status
Report. E.P.A. Publication No. AP-85, February 1973.
3. Benedict, H.M. and R.E. Olson. Economic Impact of Air Pollutants in
Plants, Annual Report, Vol. I. Irvine, California: Stanford
Research Institute, August 1970.
4. Crocker, Thomas D. Urban Air Pollution Damage Functions: Theory and
Measurement. Riverside, California: University of California,
June 1971.
5. "Demonstration of a Regional Air Pollution Cost/Benefit Model." TRW
Systems Group, EPA Contract No. PH 22-68-60, McLean, Va., July 1971.
6. "The Economics of Clean Air." Annual Report of the Administration of
EPA, U.S. Government Printing Office, Washington, D.C., 1972.
7. "Environmental Quality." The Third Annual Report of the CEQ, Washington;
Government Printing Office, 1972.
8. "Environmental Quality." The Fourth Annual Report of the CEQ, Washington:
Government Printing Office, 1973.
9. Fogel. "Comprehensive Economic Study of Air Pollution Control Costs for
Selected Industries and Regions."
10. Peckham, Brian W. "Bibliography of Literature Relating to the Economic
and Legal Aspects of Air Pollution." Chapel Hill, North Carolina:
University of North Carolina, September 1971.
11. Purdom, P. Walton. Environmental Health. New York: Academic Press, 1971.
12. Ridker, Ronald G. Economic Costs of Air Pollution. New York: Frederick A.
Praeger Publishers, 1967.
13. Williamson, Samuel J. Fundamentals of Air Pollution. Reading, Massachusetts:
Addison-Wesley, 1973.
14. Wolozin, Harold. The Economics of Air Pollution. New York: W. W. Norton
and Co., 1966.
85
-------
A METHODOLOGICAL APPROACH TO COST EFFECTIVE AIR QUALITY IMPACT
LAND USE PLANNING
i
SCOPE AND OBJECTIVES
The methodology presented here attempts to provide an analytical frame-
work for considering the economic implications of air quality impact land
use planning.
Current concern about the environment has fostered attempts to improve
urban air quality. The major focus of these attempts has been on emissions
control through fuel utilization and waste gas cleansing devices. Unfor-
tunately, this type of approach recognizes neither the problems of planning
for long term air quality, nor basic questions related to the economics of
air pollution. The methodology presented here combines some of the most
significant aspects of air pollution economics with an existing air quality
impact planning process to form a "bare-bones" procedural outline for:
1. Making preliminary planning decisions involving acceptable
amounts of heavily polluting land uses in terms of both
air quality and one or more planning goals considered rep-
resentative of economic viability
2. Evaluating alternative proposed land use plans in terms of
how effectively they use the air resource to obtain stated
planning goals.
In order that a cost effective planning and evaluation scheme for air
quality be of practical value to the planner, it must be both easily applicable
to a variety of planning situations and sensitive to time and budgetary con-
straints imposed on all planning exercises. Most especially it must be
flexible in scope, tolerant of a variety of operating variables, simple
to implement, and compatible with existing data and techniques for
relating land use to air quality.
The ultimate objective of this methodology is the improvement of the
decision making process concerned with the growth and development of urban
configurations in terms of air quality. The intention here is not to put
forth a dogmatic planning procedure but rather to provide a quasi-analytical
framework within which the planner can accomodate and evaluate the effects
of special characteristics of individual study areas. This is neither a
mechanism for producing ideal land use schemes nor a means of attaining
pollution free air. It is instead a planning aid and management tool for
conserving the air resource in the face of severe.pressures for economic
development and given the restrictions of a fossil fuel energy supply and
an imperfect emissions control technology.
Because this document is very conceptual and does not provide for the actual
application of the methodology presented, Table 3-2 may prove useful to those who
might wish to attempt an application. It presents the general objectives of the
methodology and indicates what these objectives translate to in terms of scope of
application. It represents, therefore, a concise statement of the intended util-
ity of the methodology. Additional comments on application may be found in
the third chapter.
86
-------
TABLE 3-2. SCOPE AND OBJECTIVES
Objectives of the Methodology
Implications to Scope
of the Methodology
1) To anticipate and evaluate the
air quality and economic implica-
tions of planning decisions.
1) Must recognize and include
quantification of air quality and
economic criteria.
2) To be compatible with existing
data and air quality and economic
analytical techniques.
2) Must allow for a range of
sophistication and detail.
3) To recognize and accommodate
the specific concerns of indi-
vidual planning areas.
3) Must be tolerant of a variety
of operating variables over a
range of physical scales .
4) To accommodate both the
operational capabilities and
limitations of individual planning
groups.
4) Must be sensitive to time and
budgetary constraints.
87
-------
CONCEPTUAL DESIGN
Conceptually, the methodology a;.ms at establishing the most effective
use of the air resource from among alternative proposed land use plans which
accomodate both air quality and economic development criteria.
In addition to the general problem of defining appropriate state and
regional planning control strategies in compliance with the requirements
of legislated air quality criteria, a very practical problem facing most
communities is planning for the future to accommodate anticipated or desired
growth. Government environmental regulations and citizen group activities
have caused many communities to re-evaluate the process by which growth is
achieved, so that environmental quality is now as much a. concern as the more
traditional problems of housing, employment, crime, and tax base. Consequently,
planning decisions involving the magiitude and direction of community and
regional growth must be cognizant of both economic and air quality constraints
and should be based upon the effectiveness of air resource utilization.
Generally, economic constraints to planning options will appear in the
form of minimum increases of industrial, commercial, and residential land
uses necessary to accommodate existing and anticipated or desired changes in
the economic, social, and political structure of a given study area. Factors
to be considered in establishing this 'lower bound to growth' should there-
fore include the existing land use configuration, indicated development trends,
physical characteristics of the area which either tend to promote or inhibit
certain types of development, specific political or legislative requirements,
and the development preferences of the existing population.
Air quality constraints to planning options are somewhat less difficult
to define than economic constraints. In most cases these air quality
constraints will be legislated federal or state standards. In addition to
absolute limits of individual pollutant concentrations, these may also
involve incremental limits, to either emission densities or pollutant
concentrations, over fixed values. In any event, these constraints will
translate to maximum amounts of industrial, commercial, and residential
land uses which can be tolerated within a given study area.
Together, the upper and lower bounds to growth, as established through
a consideration of air quality and economic constraints, provide a range of
land use types and intensities from which preliminary development options may
be specified. Designing within these limits will enable proposed plans to
effectively accommodate both economic and air quality requirements. However,
there are any number of combinations, of land use mixes and intensities which
may be permitted in a given area so that most often a series of alternative
configurations will be postulated. An evaluation of the alternatives, defining
how effectively each makes use of the air resource, provides a basis for indi-
cating the preferred direction and magnitude of area growth. This evaluation
is accomplished by performing a simplified relative cost effectiveness analysis.
The following section discusse:; the implementation of these concepts with-
in the planning process.
88
-------
Economic Constraints
on
Development
Air Quality
Constraints on
Development
Lower Bound to Growth
Upper Bound to Growth
Land Use Options
Mixes
Intensities
Locations
Cost Effectiveness Analyses
of
Various Options
Preferred Growth
Option
Figure 3-7. Conceptual Methodological Design - Shows how air quality
and economic development constraints are used to establish
preferred growth options.
89
-------
PROCEDURAL DESIGN
The procedural design of the methodology is effectively a step-wise
sifting process in which economic and air quality constraints are sequen-
tially applied to possible growth configurations and the residuals are
evaluated in terms of their relative air pollution impact cost effectiveness.
Translating the conceptual design of the methodology, as defined in the
previous section, into a practical, operational planning tool requires delin-
eation of a procedural set of working steps. Ideally, these steps are indi-
vidual operations which apply the current body of relevant information to
specific planning concerns and do so in a chronological sequence which eliminates
less promising plans early on and examines the residuals in further detail.
As indicated in the previous section, an Air Quality Impact-Land Use Plan-
ning Process has already been defined and documented. The process is outlined,
by its component steps, in the accompanying figure. It is the foundation upon
which the procedural design of the methodology is laid.
Establishing the upper and lower bounds to growth, as required by the
conceptual design, is accomplished by performing Steps 1 through 3 of the
Air Quality Impact-Land Use Planning Process (denoted by the dashed line).
These steps represent a 'quick and dirty1 means of defining the trade-off
between air quality and the pollution generating activities which generally
support the economic structure (i.e., industry and transportation). The
output of Step 3 is in the form of a series of preliminary alternative plans,
specified in terms of types and amounts of industry and transportation. Each
of these preliminary plans has been generated within the limits set by stated
air quality and economic criteria.
Steps 4 and 5 of the Air Quality Impact-Land Use Planning Process de-
velop and refine preliminary designs into 'final' land use plans and provide
a statement of expected pollutant concentrations and their relative spatial
distributions within the planning area. This information is applied to cen-
sus tract data (or its equivalent) to obtain per capita air pollution damages
from pollutant specific damage functions. The total cost of air pollution
damage for the planning area is then the sum, over all census tracts, of
the total damage in each tract.
The cost of controlling pollutant emissions is determined from the com-
bination of costs fostered by controls imposed to limit pollutant emissions
in excess of federal emission requirements and the cost of operating state,
regional, and/or municipal control and regulatory agencies. Together, damage
costs and control costs define the total cost of air pollution for a given
land use plan. By iterating on control strategies and working back through
the Air Quality Impact-Land Use Planning Process, it is possible to define,
for each land use configuration proposed, the cost optimum level of air quality.
Emission controls should be specified to obtain this level as far as pos-
sible.
The total cost of air pollution is then used as the denominator in a
ratio of the dollar value of benefits received from a given plan to its air
pollution costs. This ratio represents the relative cos;t effectiveness of
that plan. Comparison of this value with those derived for other plans
establishes a preferential ranking of all alternatives considered.
90
-------
The Air Quality Impact -
Land Use Planning Process
STEP 1
Establish the Air
Quality Baseline
STEP 2
Define the Tolerance of
the Planning Area to
Additional Pollutant
Emissions
STEP 3
Set Constraints on
Industry and
Transportation
STEP 4
Generate Comprehensive
Land Use Plan
STEP 5
Evaluate Air Quality
Impact
Basic Planning Goals
Acceptable Air Quality
Economic Viability
Upper Bound
to Growth
Lower Bound
to Growth
Utility of Plan
Total Cost of
Air Pollution
I I
Cost Effectiveness =
Utility
Cost
Figure 3-8. Procedural Methodological Design - Shows how the
implementation of the methodology interfaces with
the Air Quality Impact-Land Use Planning Process.
91
-------
APPLICATION GUIDELINES
The methodology presented is applicable to a variety of planning
situations over a range of physical .scales each subject in its turn to a
variety of operating variables.
As indicated previously, the methodology presented is a loosely
structured framework within which planning decisions are formulated and
evaluated in terms of how effectively the air resource is used to
obtain stated planning goals. Because planning exercises are invariably
site specific, and because the special concerns of individual planning
areas are extremely difficult to address in a generalized context,
application of the methodology to a specific problem requires that
preliminary decisions relative to physical scale, operating variables
defining utility, and data requirements be made. Generally, these
three concerns will dictate both the scope and levels of detail
of specific analyses required within the generalized procedural
framework.
The physical scale over which the methodology may be: applied is
considerable, ranging from macroscale studies involving several hun-
dreds of square miles to microscale studies of perhaps a square mile
or less. In the macroscale case the primary focal point of analytical
studies is the definition and evaluation of the aggregate effects of
all land uses within a given study area, in terms of both their economic
and air pollution impacts. By definition these are regional scale
planning exercises and are particularly effective for establishing
general development preferences in terms of over all land use types,
intensities, and relative locations. Microscale studies will usually
involve the design and placement of individual facilities within a
given study area. In many cases, it may be desirable to use such
studies as a tuning mechanism for regional scale exercises. For
example, after defining an acceptable regional configuration, alterna-
tive subregional and local development options may be examined and
evaluated within that regional context.
In choosing a physical scale which is appropriate for a given
study, it is important that the scope of impact of various planning
options be recognized. The environmental consequences of a regional
scale planning decision may be limited to the microscale. By the
same token, the economic implications of microscale development
options will often be regional in nature and might therefore require
an iteration on the macroscale scheme.
Once a physical scale has been chosen, it is necessary to structure
the work study programs defining utility (benefits) and non-utility (costs)
The first consideration in doing this should involve the specification of
operating variables.
92
-------
The number and types of operating variables available to represent
the utility of a given plan are as varied as the connotative meaning of
the word 'utility.' Any single planning goal or objective, or any
combination of goals which can be quantified, and to which a dollar
value assignation may be made, is fair game. These variables may be
either positive or negative, although any plan which demonstrates a
net negative utility in terms of a combination of variables defining
stated goals is, by definition, less than useless and should be summarily
rejected. Optimally, operating variables should be specified so as to
be mutually independent, and therefore directly additive on a dollar
value basis. It must be noted here that the evaluation of alternative
plans in terms of their relative air quality cost effectiveness is
meaningless unless the same operating variables are used to define
utility for each plan.
The variables which may be used to define non-utility, or costs,
appear in two basic sets:
1) Expressing the costs of air pollution damage
2) Expressing the costs of air pollution control.
The operational definition of the damage function is limited by
lack of extensive available data. As indicated in the section entitled The
Damage Function, the present definition would involve the summation of health,
vegetation, property devaluation, and soiling costs. The costs of
air pollution control are more varied and will be more site specific
in nature. For this reason, a bit more creativity can be brought
to bear in their operational definitions for a given study. In
either case, it will usually be the availability of empirical
data which will dictate the variables to be used.
Because the validity of any quantitative evaluation schene is
critically dependent on the validity of the data used, it is important
that comprehensive regional or locally specific data be used in defining
the operating variables. If this type of information is unavailable,
serious consideration should be given to a redefinition of utility
for that series of planning options. Where a redefinition cannot
retain the intended meaning of utility, pro forma data representations
may be used. However, even where relative evaluations will suffice,
results derived from these data should be closely scrutinized.
It should be painfully obvious at this point that an actual application
of the entire methodology will require a considerable effort. Furthermore,
there appears to be a general reluctance on the parts of both the planning
and economic communities to accept the validity of a cost effective approach
to air quality planning. However, it would be unreasonable to assume that a
cost effective approach to allocating perhaps our most precious resource can
be dismissed without further scrutiny.
93
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing]
1. REPORT NO.
EPA-450/3-74-020-a
2.
3. RECIPIENT'S ACCESSIOI*NO.
4. TITLE AND SUBTITLE
Air Quality For Urban and Industrial Planning
5. REPORT DATE
March 1974
6. PERFORMING ORGANIZATION CODE
C. Goodrich, Scott T. McCandless,
Michael J. Keefe, William P. Walsh, & Alan H. Epstein
8. PERFORMING ORGANIZATION REPORT NO.
ERT Document No. P-434
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Environmental Research and Technology, Inc.
429 Marrett Road
Lexington, Massachusetts 02173
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-0567
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYF'E OF REPORT AND PERIOD COVERED
Environmental Protection Agency
Office of Air Quality Planning & Standards
Research Triangle Park, N.C. 27711
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES Prepared in cooperation ^ £(-, the New jersey Department of
Environmental Protection, Office of the Commissioner, Labor and Industry Building,
Trpntnn. N..1.
16. ABSTRACT
This Final Report presents a summary of work undertaken, including
the proposed scope of work for each of the three major tasks, a sum-
mary of the actual work undertaken, and an explanation of any deviations
from the intended scope of work. The report also contains the findings
of Task 1, the development of improved emissions projection activity indices;
and Task 2, the development of a methodology for incorporating cost data
into the evaluation of the air pollution impact of land use plans. The
results of Task 3 are presented in a separate report entitled "A Guide
for Considering Air Quality in Urban Planning", EPA-450/3-74-020.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Land Use
Planning & Zoning
Local Governments
County Governments
Regional Governments
State Governments
Air Pollution Control
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
87
Unlimited
EPA Form 2220-1 (9-73)
94
20 SECURITY CLASS (Thispage)
Unclassified
22. PRICE
------- |