OCR error (C:\Conversion\JobRoot\00000AX3\tiff\2000XFBK.tif): Unspecified error
-------
EPA-600/4-81-034
May 1981
CASE STUDIES IN THE APPLICATION OF AIR QUALITY MODELING
IN ENVIRONMENTAL DECISION MAKING
Summary and Recommendations
by
Catherine G. Miller
John F. Kennedy School of Government
Harvard University
Cambridge, Massachusetts 02138
Grant No. R805558-01
Project Officer
Kenneth Demerjian
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
-------
ABSTRACT
Eleven case studies of the application of air quality models were under-
taken in order to examine the problems encountered when trying to use these
models in making environmental policy decisions. The case studies of air
pollution control decisions describe the decision process, the models used,
the critiques of the models, and the participation by outside interest
groups in the decisions process. The studies include two cases of federal
decisions, seven state decisions, one local decision, and a review of the
evolution of modeling requirements in the Clean Air legislation. The time
covered is from 1970 to the present and includes several cases for which a
final decision has not yet been made.
The results of this investigation show that indeed the well-known tech-
nical and political constraints exist but that unresolved policy issues,
the management of the decision process and conflicting institutional and
organizational interests also cause problems. Recommendations are made on
how to improve the technical planning and management of the decision process
so that the air quality models can become a better policy tool within the
state-of-the-art, political and organizational constraints.
This report was submitted in fulfillment of Grant No.R805558-01 by the
Kennedy School of Government, Harvard University, under the sponsorship of
the U.S. Environmental Protection Agency. This report covers the period
November 1, 1977 to January 5, 1981, and work was completed as of March 15,
1981.
Ill
-------
CONTENTS
Abstract iii
Tables vi
Acknowledgements vii
I. Introduction 1
The Case Studies 1
The Objective 5
The Approach 6
II. Uses of Models in a Policy Framework 7
Models Used for Problem Solving 8
Models Used for Conceptualization 11
Basic Research Leading to Application 12
Models Used for Political Advantage or to
Advance Self-interest 13
References 14
III. The Case Studies and Their Uses of Models 16
Clean Air Legislation 17
MATEP 21
Pittston 26
Anaconda 29
Westvaco 31
Massachusetts' Sulfur Regulations 34
Ohio's State Implementation Plan 36
Connecticut's Transportation Control Plan 39
The San Francisco Air Quality Maintenance
Plan 42
Ozone Standard Revision 47
New Source Performance Standards 49
Summary 55
IV. Conclusions 59
Technical and Political Constraints 62
Policy, Managerial and Organizational
Problems 67
Summary 78
V. Recommendations 80
Technical Recommendations 80
Managerial Recommendations 81
-------
TABLES
Number Page
1 Program Characteristics of the Case Studies 3
2 Modeling Characteristics of the Case Studies 18-19
3 Dichotomy between Modeler's Question and
Policy Question 61
4 Criteria Used in Choosing an Air Quality Model 71
vi
-------
ACKNOWLEDGEMENTS
The cases were prepared under the supervision of the author and
Professor Laurence E. Lynn, Jr., of the Kennedy School of Government.
Primary responsibility for the research and writing of the cases should
be credited as follows:
"Air Quality Modeling in Clean Air Legislation" by Catherine G. Miller and
Evie Verderber
"The MATEP Power Plant" by Christopher Allen and Evie Verderber
"Pittston-Eastport" by Christopher Allen
"The Anaconda Copper Smelter Case" by Christopher Allen
"The Maryland SIP: An Exemption for the Westvaco Corporation" by Evie
Verderber
"Massachusetts' Sulfur Regulations" by Deborah Wishner
"Ohio's State Implementation Plan" by Christopher Allen
"Connecticut's Transportation Control Plan" by Deborah Wishner
"Revision of the National Ambient Air Quality Standard for Ozone" by
Catherine G. Miller
"Balancing on a Tightrope: How the U.S. Environmental Protection Agency
Set New Source Performance Standards for New Coal-Fired Power Plants"
by Elizabeth H. Haskell
Additional support for this last case study was received from the U.S.
Public Health Service.
vii
-------
CASE STUDIES
This report briefly describes the case studies. For those interested
in a more detailed description, the case studies are available on an indi-
vidual basis from the
Technical Information Manager, MD-59
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Titles of the individual case studies are listed below.
AIR QUALITY MODELING IN CLEAN AIR LEGISLATION
THE MATEP POWER PLANT
PITTSTON-EASTPORT
THE ANACONDA COPPER SMELTER CASE
THE MARYLAND SIP: AN EXEMPTION FOR THE WESTVACO CORPORATION
MASSACHUSETTS' SULFUR REGULATIONS
OHIO'S STATE IMPLEMENTATION PLANS
CONNECTICUT'S TRANSPORTATION CONTROL PLAN
AIR QUALITY MODELING AND THE 1978 SAN FRANCISCO BAY AIR
QUALITY MAINTENANCE PLAN
REVISION OF THE NATIONAL AMBIENT AIR QUALITY STANDARD
FOR OZONE
BALANCING ON A TIGHTROPE: HOW THE U.S. ENVIRONMENTAL
PROTECTION AGENCY SET NEW SOURCE PERFORMANCE STANDARDS
FOR NEW COAL-FIRED PLANT PLANTS
viii
-------
SECTION I
INTRODUCTION
The U.S. Environmental Protection Agency as well as many academic and
other private institutions have developed a large repertoire of air quality
models, models which predict the impact of emissions on the quality, or
pollutant concentration levels, of the ambient air. The Clean Air Act
Amendments now require the application of these models when making decisions
about controlling air pollution. However, the cry is often heard that the
models are not used. This study grew out of the concern that many modelers
as well as potential users of models are dissatisfied: the modelers because
they are not listened to and the users because they do not hear much that
they want to listen to. Also users may apply the models in ways other than
those intended by the modelers.
To look at this conflict between the modelers and decision makers
(users), cases of specific applications of air quality models have been
studied. The focus of the case studies is the decision making process, i.e.
how policies and decisions are formed within the federal and state environ-
mental agencies and what role the air quality models played within that
process. Questions such as
- how were the models chosen,
- how did the industry models differ from the government agency
models,
- what technical information was actually presented to the
decision maker,
- what interest groups intervened in the process, and
- what policy or non-technical controversies arose and how were
they resolved
are examined in order to develop a basis for understanding the role of the
models, reasons for their success or failure and the critical attributes
necessary for their use in a given context.
THE CASE STUDIES
The eleven case studies completed for this research project were picked
to illustrate different levels of decision making (national to local) ,
-------
different institutional domains (administrative, judicial, and legislative)
as well as the various EPA programs which require air quality modeling
analyses. Table 1 lists the case studies and indicates which elements each
contains. It should be noted that the case studies are intended to illus-
trate the various roles that air quality models have been called upon to
serve but do not represent a scientific sample of all such cases.
A brief introduction to each case study follows in order to acquaint
the reader with the range of situations presented in the eleven case studies.
Clean Air Legislation
This case study outlines the history of the federal Clean Air legisla-
tion and why several provisions of the 1977 amendments include modeling
requirements. Some of these provisions are the focus of controversy in
subsequent case studies and this history is intended to provide a context
for these debates. It also points out the legislator's role in legitimizing
the use of these models without actually having applied the models or made
decisions requiring the use of the models.
MATEP
In 1972, Harvard University applied to the state of Massachusetts for
a permit to construct cogeneration facilities as part of a power plant (the
MATEP power plant) to be built near downtown Boston. The modeling analysis
showed that the facilities, which included diesel engines to generate both
steam heat and electricity, could meet the applicable national ambient
standards. However, public concern centered on the short-term NC>2 levels
which would be generated by the operation of the diesels and for which the
federal EPA had not yet set a standard. Thus, without a standard the state
had to review evidence on the health impacts of N02 in making its decision.
Pitts ton
In 1973, the Pittston Company applied for permission to build an oil
refinery in Eastport, Maine. Eastport is a rural community on the northeast
coast of Maine near the Canadian border. It is also near a national wild-
life refuge and an international park, thus making the application subject
to the prevention of significant deterioration (PSD) provisions of the Clean
Air Act. The increment under contention in this case was the 24-hour sulfur
dioxide PSD standard. Air quality modeling was done by both the company
and the state to determine what plant design would meet this standard.
Anaconda
In order to meet the national ambient air quality standards set by the
Clean Air Act Amendments of 1971, the Anaconda copper smelter located in
Anaconda, Montana, was required to install pollution control equipment.
This case study describes the extensive modeling done by EPA in order to
reach agreement with the company and the state on an 502 emission rate for
the smelter.
-------
^H O
i—3 O
M M
01
01 >
JcS
* 3 w
•r— * ^"^ CJ
5 O 4-1
e
Cfl
M
00
O
Pu<
«aj
O-i
O)
M
CJ
1
ON
iH
O
4-1
in
m
ON
rH
SI
•H
4-1
Cfl
rH
CO
•H
00
^J
CO
CO
M
00
°
\—S
iH
rH
<
C
o
•H
4-1
CO
iH
CO
"ob
3
-*
3
U —4
o
00
ON
iH
O
4-1
CM
r-
ON
rH
SI
•H tH
4J -cfl
CO -H
M 0
4-> -H
CO T3
•H 3
C *-)
S t^}
re3
Ol
4-1
cO
4J
en
X
o
2
4-1
•fj
P
£.4
Ol
a
(j) C
O ^3
3 "S
O "^
w "^
^
-»
O
3 Q"
o en
CO P-l
^
01
•z,
ON
P--
ON
rH
O
4J
CN
P-.
ON
rH
Ol
•H
4-1
CO
4-1
CO
•H
C
•H
6
T3
and
Office
0) rH
4J Cfl
Cfl C
4J O
en -H
00
tS
CM
O
en
^
rH
01 CQ
0 M
M 3
3 U
O ^— '
CO
Ol
oo u
c c
•H CO
4-> -H
CO rH
-H a.
X S
a o
CJ
ON
p-
ON
rH
o
4-1
^^
p^
ON
rH
01
•H
4-1
CO
i-t
4-1
co
•rH
G
'I—I
e
TD
Office
ngress
rH O
cfl ej
O T3
•H C
00 CO
cu
Pi
CM
o
en
^
rH
Q) Cfl
CJ r<
M O
3 M
o •^/
Cfl
Ol
00 U
C C
•H Cfl
4J -H
CO rH
•H Ou
X S
W 0
U
P-.
ON
rH
0
4-1
m
p-
ON
rH
ofl 01
Ol -H
^ 1J
•H CO
CO 4-1
rH CO
CO iH
•H C
00 -H
01 E
<
01
o
•H
T3 UH
cfl O
01 rH
4J Cfl
CO C
4-1 O
en -H
00
CO
prf
CM
o
en
^
'c
Ol cfl
CJ .Q
3 3
O ^
CO
01
oo a
•H CO
CO rH
n a
x s
w o
o
en
4-1
-i co
01 C
' '. -a o
— J 4-i 2
cfl 3 -H
4J -J3 *4 3
CO 3) — ( 30
U S 3 D
•„• ,3 s !s- '-2 3 ; z: ;/} =>s
o
oo
rH
0
4J
in
rH
01
•H rH
4J CO
CO iH
S-i U
4-1 -H
to 'a
•H 3
c *->
•H
T3
01
O
•H
-a UH
rf * 1 I
CO O
01 rH
4-1 CO
Cfl (3
4-1 O
en -H
00
01
*
CM
O
en
00 0)
C <•>
•H C
4J Cfl
W -H
•H rH
x a
cu G
0
-a o
cfl Ol
o
§ 3
rH O
P-i CO
CL!
r-t
en
o
•H
JT
ON
P-
rH
O
4-1
CO
p*»
1—1
•H
4-1
CO
S-i
4-1
co
•H
C
•H
G
-a
0)
a
•H
T3 UH
§ O
Ol rH
4-1 Cfl
CO C
4-1 O
en -H
00
Ol
&
X
o
G
cfl
rH
P-i
4J
3
CJ
•H
4-1
CJ
CU
2
M P-4
o y
CJ r-
ON
rH
O
4J
O
1—1
•H
4-1
cfl
)H
4-1
to .
•H
C
•H
e
•o
agency
tate
en
rH
ft) UH
CJ O
O
X
O
£^
rH
PL,
O
O
to
iH
^J
r*
^0
ON
rH
O
4J
\Q
P*»
ON
rH
•H
4-1
Cfl
M
4J
CO
•H
C
•^
0
•o
PH
W
rH
ca
V-i
0)
T3
roi
fe
X
O
""O
v^
cfl
C C
cfl O
4J -H
Cfl CO
•rH
C CU
0) M
•H
0
^
U
CO
ON
ON
rH
O
4J
p*^
p-
ON
rH
01
•H
4J
CO
j-i
4-1
co
•H
Cl
*H
S
T3
^E
^^
P*
W
rH
2
•3
cu
fa
CM
O
en
C
O
•H
4-1
cfl
tH
3
CO
01
-------
Westvaco
Maryland's State Implementation Plan (SIP) requires either the in-
stallation of a sulfur emissions control system on coal burning facilities
or that only coal with a sulfur content of less than 1% weight be burned.
In 1974, Maryland issued a notice of violation to the Westvaco Company for
its operations at their pulp mill located on the Maryland/West Virginia
border. Westvaco appealed the order to revoke its operating permit claiming
that no SC>2 ambient air quality standards had been violated using the
higher content sulfur coal. To substantiate this claim they submitted
monitoring and modeling evidence to the EPA regional office.
Massachusetts' Sulfur Regulations
Massachusetts had regulations for limiting the percentage content of
sulfur in fuel used in fuel-burning facilities in their state. For the
Boston metropolitan area, commencing October 1, 1971, oil-burning facilities
were restricted to 0.5% sulfur oil. In response to the 1973 constrictions
in Middle East oil, the Massachusetts' legislature passed Chapter 494, which
required the state environmental agency to relax its sulfur-in-fuel regu-
lation making it no stricter than necessary to meet the national ambient air
quality standards. To determine a new regulation on sulfur content in fuel,
air quality modeling was done by the state, by the EPA regional office,
and by the electrical utilities.
Ohio SIP
In 1972, the state of Ohio submitted a State Implementation Plan
to EPA without agreeing on that portion of the plan wich would control S02-
Three years later, with Ohio still unable to complete its SIP, EPA proposed
one for them. The EPA plan prescribed procedures for setting emission limits
for several large power plants near Lake Erie. When the utilities appealed
EPA's use of the rollback model, EPA's plan was rejected by the courts.
This case chronicles the subsequent modeling analyses done by EPA and the
utilities in an attempt to set emission limits for two of the Lake Erie
plants.
Connecticut TCP
The state of Connecticut, in preparing its State Implementation Plan,
needed a Transportation Control Plan (TCP) for the control of oxidants.
This case centers around the problem of transport: Connecticut wanted to
base its plan on the fact that the oxidant problem within its boundaries
was caused in part by pollutants emitted in New York and New Jersey and,
hence, should be controlled by them. EPA's policy did not at that time
allow for consideration of transport and their approved models did not take
this factor into account. Thus, the state attempted to develop its own
model to use in preparing a TCP.
San Francisco AQMP
The San Francisco Bay area, like Connecticut, needed a Transportatioa
-------
Control Plan for controlling oxidants, as part of an Air Quality Main-
tenance Plan (AQMP) required by the 1977 Clean Air Act Amendments. This
AQMP was also to be integrated into an Environmental Management Plan which
would coordinate governmental planning in the areas of air pollution, water
pollution, water supply and solid waste. This larger plan, to be done for
the San Francisco Bay area, was a first attempt by any state or region to
integrate these areas of concern. The exercise involved extensive air
quality model development as well as coordination of governmental, scien-
tific, industry and environmental groups.
Ozone Standard
In 1971, national ambient air quality standards were set by EPA for six
pollutants. The first of these standards to be revised (eight years later)
was the standard for ozone. The technical information developed for this
decision included a criteria document, presenting evidence on the health and
welfare effects of ozone but also containing a chapter describing the types
of air quality models available to predict ambient ozone levels. For the
standard revision decision, air quality models were used in preparing a
cost and economic impact report which estimated the costs, nationwide, of
achieving alternative levels of an ozone ambient air quality standard.
NSPS
In 1977, EPA announced that it would revise its New Source Performance
Standards (NSPS) for coal-fired power plants, which were first set in 1971.
This case study describes the economic modeling and other technical analyses
produced for the revision and how technical questions, which arose during the
decision process, were resolved. Although no model integrating environmental,
economic and energy considerations was available at the time, this case
illustrates the broad range of topics EPA faced and how an air quality
model and other methods of analysis in the appropriate format could have
been used and can be used for similar decisions in the future.
THE OBJECTIVE
The case studies are chronologies of what happened and what (if any)
decisions were made. They outline specifically the modeling analyses under-
taken and the results presented, who criticized them and what the criticisms
were, and the modeler's response. Each case study, through interviews with
persons who participated in the analyses and reviews of the public documents
produced, describes the technical information available to the decision maker.
In making environmental policy decisions, EPA and the states require a
sound scientific basis. Scientific research, including air quality models,
has become increasingly important as well as more sophisticated and more
widely used. However, as a useful input to the decision making process,
the research nas in many cases not met expectations. Critics often point
cut that the state-of-the-art of environmental research has not yet developed
to the point where it can supply information in the detail and accuracy
required to solve policy problems. Gaps certainly do exist in the available
-------
information; though, the reasons for these gaps and the constricted flow of
information between researchers and decision makers can also be traced to
the different professional and organizational settings within which" re-
searchers and decision makers work. Thus, the case studies also describe
the decision process, what role the models played within this political
context and how the modelers, given their professional and organizational
orientation, did or did not repond to the policy dimensions of the problem.
The studies focus on the problems that arise in using research tools
as policy tools. The policy maker must deal with poor information, uncer-
tainty about future events, imperfect legal and administrative structures,
and pressures exerted by various interest groups. The researcher is also
confronted by uncertainties, budget gamesmanship, political agendas, and the
reluctance to drop projects no longer strictly relevant to the policy
problem. Indeed, "the problem" may be constantly changing as the policy
mandate changes or as research leads to more mature formulations of appro-
priate solutions. Thus, the objective of these studies is to improve the
quality of the research information while recognizing the existence and
influence of these other problems, and to look at why the modeling—even
modeling specifically done for policy purposes—does not necessarily or
automatically conform to the decision maker's needs.
THE APPROACH
In this summary report I draw conclusions from the case studies but
do not try to chronicle the events of each one. As framework for analyzing
the case studies, Section II of this summary describes the uses of technical
information commonly made in the public political arena. Section III then
summarizes each case study, describes the uses of the models as they are
found in the individual case studies and what elements of the political,
professional, organizational and institutional environment of both the
modeler and the decision maker seem to have shaped the uses and the decision.
Section IV of this summary report draws conclusions about the common tech-
nical and managerial problems that arose during the decision making process,
how the problems were resolved, or why they have as yet eluded resolution.
Section V gives recommendations, based on the experiences outlined in the
case studies, on how these technical and managerial problems might be
alleviated or, at a minimum, better controlled.
One further qualification should be noted. In examining and analyzing
these case studies, it is not my purpose to say what the decision should be.
Making a normative judgment would be distracting to the arguments on how
the decision (as opposed to what decision) was made and how to improve the
presentation of the technical information.
-------
SECTION II
USES OF MODELS IN A POLICY FRAMEWORK
Environmental decision making is a complex process involving experience,
political insight and political pressures, as well as technical information.
The complexity of the decision process and the intrinsic uncertainties of
the scientific data base combine to make this study of model usage a wide-
ranging one. One manifestation of this complexity is the term the "use of
models" which can take on several meanings depending in part on the motives
and organizational constraints operating on the decision maker. Likewise,
the motives and constraints on the modeler developing the technical informa-
tion may determine what information is available to the decision maker.
Thus, this study of air quality models involves differentiating the uses
policymakers have for the kind of technical information provided by air
quality models and what elements of both the policy and research setting
restrict or promote a certain type of use.
Models within a policy framework can be divided into four categories:*
— models used for problem solving
— models used for conceptualization
— basic research leading to application
— models used for political advantage or to advance self-interest.
The case studies examine these different types of uses of models for a
particular decision. This section lays the groundwork for the analysis of
the studies by outlining, in theoretical terms, the various types of uses
and what motivational and organizational influences may determine how the
model is applied or what problems result from the attempts to employ the
models. The motivational factors include the background and professional
interests of both the researcher and the decision maker. The organizational
factors include both organizational interests within EPA and the various
institutional interests of other governmental and non-governmental agencies
and private and public lobbying groups in the larger political arena.
*
These categories overlap with those outlined by Weiss (1978)for social
science research. Similar uses are described by Greenberger ot_ al_. (1976)
and Brewer (1978).
-------
MODELS USED FOR PROBLEM SOLVING
The primary use of research by a decision maker is for problem solving:
the application of technical information to the problem at hand for the pur-
pose of delineating the consequences of possible solutions in order to decide
which solution is the most acceptable. Various motivational and organiza-
tional factors create problems in the use of models for this purpose.
Perhaps the specific information needed by the decision maker is not
available. For instance, as chemically reactive pollutants are transported
in the atmosphere downwind they are transformed into other substances. The
models currently available are limited to a range of about 50 kilometers
and cannot satisfactorily treat chemical transformation or removal processes.
Another example of this problem is transportation control planning.
Models are available which indicate the reduction in vehicle miles travelled
necessary for a given set of air quality standards. However, a publicly
elected official is not always able to institute programs to reduce car
travel locally and may prefer to depend on national emission standards or
higher fuel prices. Thus, the decision maker only has control over certain
variables. As an administrator for an environmental agency rather than a
transportation agency, he or she cannot design highway projects to reduce
vehicle miles travelled. Thus, what the modeler thinks is important may
not be what the decision maker can control or change, and, as a result, the
models are not relevant to the decision maker's needs.
This situation is exacerbated by institutional and organizational
arrangements. The model developer may work in a different institution than
the decision maker, for instance a consulting firm or a university. As
such, it may be logistically difficult to arrange meetings to discuss the
work. Also, the objectives of their institutions may differ. The univer-
sity's department or particular discipline may have more influence in shaping
the modeler's research agenda than the government agency. Research done
within the government agency may be more directly responsive to the policy-
maker's problem.
Another aspect of the institutional structure is the source of funding.
If the research is funded at the federal level, the needs at the state or
local level may not be clearly understood by the federal employees funding
the research. If the states are not able to make their needs clear and are
not able to fund it themselves, then the necessary technical information may
not be available to them.
Organizational arrangments can also aggravate the lack of common
interests between researcher and policymaker. The policy problem may re-
quire a combination of information which crosses academic disciplines. If
the organizational structure is such that researchers in each separate dis-
cipline report to separate administrators, then it may be necessary to create
a team which crosses organizational lines. Issues of budget and personnel
hours devoted to the team as opposed £:> the original discipline-oriented
jobs must be managed, and a clear definition of needed results must be
spelled out.
-------
In other situations, appropriate models may be available but the
decision maker does not employ them. Several difficulties might arise in
the course of the decision making process to cause this problem. The model
results may not be available on a timely basis, the necessary input to run
the model may not be available, or the user may not know how to use the
model's conclusions.
Another problem faced by a policymaker is that an issue may arise in
a political context and be perceived as a crisis. Fish in mountain lakes
have been dying. The newspapers call it acid rain. The information fueling
the crisis may be based on relatively few cases and the cause not really
known. The decision maker is faced with an aroused public and time is
not sufficient to design a survey, gather data and perform a careful analy-
sis.* Instead the decision maker searches for solutions and chooses the
first that is "feasible" or "realistic." This may not be the best choice,
but it serves the purpose of defusing the crisis. Problems are dealt with
in a piecemeal or fragmented manner. This incremental approach,** which
results in marginal adjustments to the system, minimizes uncertainty and
the amount of information needed; it, therefore, also minimizes the need
for research.***
In other cases the research may be prolonged as the modeler strives
for accuracy by refining the research methodology. However, the policymaker,
unable to directly control many variables of the process, may need perspec-
tive more than complete accuracy of the numbers. On the other side, the
policymaker responding to crises may find it difficult to engage in long
range planning which would have allowed the research to have started earlier
and provide results on a timely basis.
Institutional and organizational arrangements can cause further delay.
If research is not done in-house, procedures to procure and monitor grants
or contracts can be time consuming. If separate parts of the problem are
done by different organizational entities (e.g., modeling by one laboratory
and monitoring to collect data by another laboratory), delays can result
through the need to communicate up through the hierarchy and back down again.
*
Downs (1972) describes a cycle of how environmental crises come and go,
**
Lindblom (1959 and 1965) was one of the earliest expositors of the
notion of incrementalism, opposing the view of a rational decision maker
carefully reviewing all options. Others in this tradition include Simon
(1959) and Cyert and March (.1963).
***
Several authors have noted that the striking feature of the Clean Air
Act of 1970 is that it was not incremental (Ingram, 1978 and Jones, 1975)
and that this may have been due to the changed and increased role of the
public interest groups.
-------
Even if the research itself is relevant and timely, a frequent problem
is the lack of input data for the particular situation. Large, complicated
models often require detailed data which is only available on a case-by-case
basis.* This can be expensive and beyond the resources of the potential
user.** It might also be the case that the model requires the data to be
in a different form than the form in which the data was originally collected
and the available resources are not sufficient to change the data.***
This mismatch of input data can arise from the dissimilarities in
interests and goals of the researcher and the policymaker. The researcher
may be primarily concerned with furthering the field of research being
studied. The question of what type of data, what form, or if this is the
minimum amount of data that one can get away with and still solve the problem
may come second to the question of whether the theory is sound and the re-
search as accurate as possible. The policymaker with limited resources may
be willing to trade off accuracy for the chance to use the research.
An organizational structure separating the research and data collecting
efforts may create a communication problem which would tend to increase the
need for advanced planning. Within the federal EPA the routine data gather-
ing functions are carried out by an organizational group separate from the
group developing the models. Indeed, much meteorological data is compiled
by other federal agencies. Special efforts at communication must be ini-
tiated to either develop air quality models that conform to the available
data or, conversely, to develop a suitable data base for the models.
In some situations, research may be available which could provide rele-
vant information at reasonable cost, but the research is still not used
because the policymaker's education and experience may not have included
the form and techniques of scientific research. For example, the background
of the policymaker may be such that he or she is not experienced in using
quantitative information and the model results may not be summarized in a
fashion readily understandable.****
*
Pack and Pack (1977) and EPA (1974) show this is especially true for
environmental models.
**
Some investigators (Greenberger et al. (1976) for land-use models and
Roberts and Farrell (1978) for air quality models) have concluded that some
models may be beyond the capacity of the data in the sense that even a more
accurate model will not increase the accuracy of the results because the
input data is of such low quality.
Brewer (1978) discusses how this is a problem for demographic data for
example.
****
Ackeraan _e_t _ali. (1975) show how this led to a politically based deci-
sion even after a large modeling effort by the Delaware River Basin Commission.
10
-------
The modeler and the user, often coining from different academic back-
grounds, each have their own vocabularies, which can further complicate the
process. For example, in modeling the current form of the ambient air
quality standards, the standards are written into law as concentration levels
not to be exceeded more than once a year. Using past monitoring records this
is a simple determination and in the interests of protecting human health
seems suitably cautious. However, the many uncertainties in modeling make
such predictions unreliable. From a modeler's point of view a probabilistic
approach would be more appropriate, but such an approach is not easily
understood by a non-statistician.
Another problem for the decision maker may be the lack of documenta-
tion.* Even if the policymaker is experienced in using quantitative informa-
tion, without documentation detailing the assumptions and limitations of the
model, he or she may find it hard to determine whether the particular model
is applicable. Thus, the policymaker could use the model in the wrong
situation, or neglect to use an appropriate model.
The lack of validation has similar causes and effects. Sometimes a
model is not validated because it is a costly and time consuming process.
Also, models, as a representation of a system, can never be validated in the
sense of being fully supported by objective truth. Testing a model involves
looking for errors, trying to invalidate it. Thus, the process has no
uniform procedure or well-defined end. It is not surprising then that a
modeler may be more interested in starting a new study on a different aspect
of the problem than in doing numerous applications of one model with no
guarantee of success. This is especially true for models requiring large
amounts of input data.
On the other hand, when the modeler does explain in detail the assump-
tions and limitations of the model, the report of the model results may be
rewritten and summarized a number of times before it finally reaches the
decision maker, who is removed organizationally from the modeler. The final
report may alter the qualifications or leave them out altogether so that
the decision maker is unaware of the limitations and may end up using the
results incorrectly.
MODELS USED FOR CONCEPTUALIZATION
The use of research may not only be a consequence of the policymaker's
search for appropriate technical information in a pending decision. A more
general use is for conceptualization: planning for what issues are likely
*
Fromm _e_t _al. (1974) assess the extent of this problem for the U.S. gov-
ernment's domestic programs. They find that for many models a report on how
it was developed exists so the modeler thinks the documentation is adequate,
but lacking a detailed user's manual the potential user finds it inadequate.
-------
to come up next or developing a strategy to deal with issues.*
The use of research for conceptualization is not as direct an applica-
tion as for specific problem solving. However, this type of use may be
easier to undertake because the application is less specific. Thus, the
state-of-the-art is not so constraining and accuracy is not an overriding
concern. General trends may be sufficient, or new input data may not be
needed if a previous application was similar to the foreseeable future
problems on the planning agenda.
The obstacles to research utilization created by the institutional
structure may not be as serious in this case. If the need for technical
information is not too specific, then research done for academic purposes
may be useful to the government decision maker concerned with planning
strategies on how to analyze a situation or deal with a category of problems.
While availability of information may be less of a problem in this use
of research, the problems of organization and its resulting communications
barriers remain. Organizational divisions may hinder the creation of rele-
vant research as well as its communication to the decision maker through
the reinforcement of traditional discipline-oriented research and the separa-
tion within the hierarchy of modeler and user.
The problem of documentation also arises. If the decision maker is
unaware of research results, no matter how applicable, they will not be used.
BASIC RESEARCH LEADING TO APPLICATION
Basic research to expand the state-of-the-art may naturally lead to
practical applications and new policies. Examples of this type of use
generally come from the physical sciences: biochemical research often re-
sults in new drug uses unforeseen at the outset of the research; the develop-
ment of the transistor led to the portable radio and space satellites.
As a regulatory agency, EPA is primarily concerned with operational
programs for abating pollution. It focuses on the use of applied research
which in turn depends on basic research. The U.S. Congress is reluctant
to fund basic research for a regulatory agency which is responsible for
getting on with abating pollution.** Thus, the overall goal of EPA would
tend to limit the push for basic research.
A consequence of this reluctance is that the basic research to support
*
Gordon and Gordon (1972) discuss how it may be more difficult to use
mathematical models for this type of planning than for problem solving or
implementation.
**
See U.S. Congress (1979) and National Academy of Sciences (1977) for
the continuing debate on this subject.
12
-------
the applied research may not exist, or may not in practice be applicable to
policy issues. The lack of basic research may also result from the obstacles
to communication caused by the divergence of professional interests between
researchers and policymakers, as well as their organizational separation,
as has been described above.
MODELS USED FOR POLITICAL ADVANTAGE OR TO ADVANCE SELF-INTEREST
Decisions made in a political context require consideration of political
and social issues as well as technical information.* The tendency for the
modeler, following the scientific tradition, is to push for accuracy and,
failing that, to present research results with appropriate qualifications
and caveats. The reluctance to come to firm conclusions for policy purposes
leaves that task up to the policymaker who can then use the research in
justifying a decision made for political reasons, if the research does not
specifically say it is not applicable to the situation. The lack of documen-
tation can make such uses easier.
Uncertainty in the scientific data base can likewise be exploited by
advocates in the political process through the institutional structure.
The decisions of the environmental agencies in the U.S. are subject to a
review process involving the public, whereby evidence, including technical
information, is presented by persons representing various interest groups.
The lack of conclusive evidence allows the interest groups to be selective
in the evidence they present. If access to this review process is institu-
tionally determined—e.g., industrial lobbyists with more resources have
greater access—then the information available to the decision maker will
also be selective.
Organizational arrangements could also result in the use of models not
strictly relevant to the decision. If the modeler is organizationally
separated from the decision maker, the usual practice is to communicate the
result of the models through a written summary. At each level of the
hierarchy it may be necessary to further summarize. In the process the
qualifications may be disregarded and partial findings taken uncritically
at face value or misrepresented.
With these considerations in mind we now turn to the case studies and
examine the uses of air quality modeling research by EPA and state and local
environmental decision makers.
*
Greenburiter et _al. (1976) describe how these uses can build further
support for er.e wider use of models for problem solving.
13
-------
REFERENCES
Ackerman, B., S. Rose-Ackerman, J. Sawyer, Jr., and D. Henderson, The
Uncertain Search for Environmental Quality, The Free Press, New York,
1975.
Brewer G., "Operational Social Systems Modeling: Pitfalls and Prospectives,"
Policy Sciences, Vol. 10 (December 1978), pp. 157-169.
Cyert, R., and J. March, A Behavioral Theory of the Firm, Prentice-Hall, Inc.
Englewood Cliffs, N.J., 1963.
Downs, A., "Up and Down with Ecology: The Issue-Attention Cycle," Public
Interest, Vol. 28 (March 1972), pp. 38-50.
Environmental Protection Agency, A Guide to Models in Government Planning
and Operations, Washington, D.C., 1974.
Fromm, G., W. Hamilton and D. Hamilton, Federally Supported Mathematical
Models: Survey and Analysis, National Science Foundation, Washington,
D.C., June 1974.
Gordon M. and Marsha Gordon, Environmental Management; Science and Politics,
Allyn and Bacon, Inc., Boston, 1972.
Greenberger, M., M. Crenson and B. Crissey, Models in the Policy Process,
Russell Sage Foundation, New York, 1976.
Ingram, H., "The Political Rationality of Innovation: The Clean Air Act
Amendments of 1970," in Ann Friedlaender, editor, Approaches to
Controlling Air Pollution, The MIT Press, Cambridge, Mass., 1978,
pp. 12-56.
Kelleher, G., editor, The Challenge to Systems Analysis: Public Policy and
Social Change, John Wiley and Sons, Inc., New York, 1970.
Jones, C., Clean Air; The Policies and Politics of Pollution Control,
University of Pittsburgh Press, Pittsburgh, Penn., 1975.
Lindblom, C., The Intelligence of Democracy, The Free Press, New York, 1965.
Lindblom, C., "The Science of Muddling Through," Public Administration
Review, Vol. 19 (Spring 1959), pp. 79-88.
14
-------
National Academy of Sciences, Perspectives on Technical Information for
Environmental Protection, Washington, D.C. 1977.
Pack H. and J. Pack, "Urban Land Use Models," Policy Sciences, Vol. 8
(March 1977), pp. 79-101.
Roberts, M., and S. Farrell, "The Political Economy of Implementation:
The Clean Air Act and Stationary Sources," in Ann Friedlaender, editor,
Approaches to Controlling Air Pollution, The MIT Press, Cambridge,
Mass., 1978, pp. 152-181.
Simon, H., Models of Man: Social and Rational, John Wiley and Sons, Inc.,
New York, 1959.
U.S. Congress, 1979 Authorization for the Office of Research and Development,
Environmental Protection Agency, Washington, D.C., February 1978.
Weiss, C., "Improving the Linkage between Social Research and Public Policy,"
in Laurence E. Lynn, Jr., editor, Knowledge and Policy: The Uncertain
Connection, National Academy of Sciences, Washington, D.C., 1978.
15
-------
SECTION III
THE CASE STUDIES AND THEIR USES OF MODELS
In the general description of the possible uses a decision maker might
have for models, I have decribed what motivational and organizational
factors might influence what the actual uses are in any given situation.
In order to investigate how these factors currently affect the use of air
quality models by decision makers in governmental agencies responsible for
environmental affairs, eleven case studies have been conducted. This
section of the report describes the uses of the models in reference to the
framework of Section II.
The federal Clean Air Act provides for the establishment of ambient air
quality standards in order to protect public health. These are national
standards to be attained in all areas of the country. After the standards
have been determined by the federal EPA, each state must draw up a State
Implementation Plan (SIP) describing how it will control emissions from
pollution sources within its boundaries in order to meet these ambient
standards. For stationary sources the mechanism used is a permit (for new
sources) or a compliance schedule (for existing sources). For mobile
sources, such as automobiles, the Clean Air Act has mandated national emis-
sions standards. Thus, the states' methods to reduce pollution caused by
mobile sources include transportation control planning to reduce the number
of cars or to change how and where they are operated and enforcement of the
federal standards through inspection and maintenance programs or anti-
tampering laws.
Within this legal context and division of administrative responsibili-
ties the case studies illustrate the different types of decisions that must
be made. The first case study describes the history of the clean air
legislation, passed by the U.S. Congress over the last twenty-five years,
from the point of view of how air quality modeling came to be included
in its provisions and how its standard setting strategy influences how the
models are subsequently used. The next two case studies describe the
process whereby permits are issued for new stationary sources (one a co-
generation power plant, the other an oil refinery). The next four cases
involve regulating existing stationary sources (a copper smelter, a pulp
mill, and two power plants). The next two describe efforts by state
agencies to prepare implementation plans for controlling pollution from
mobile and stationary sources. The last two cases review the process by
which the federal EPA has recently revised two of its regulations: one an
ambient air quality standard and the other an emissions standard for new
stationary sources.
16
-------
This section focuses on how the results of the application of air
quality models were used in the decision process described by each case
study. Table 2 summarizes the models that were used, the decision that
was made, and the problems that were encountered for each case, as a
reference for the discussion to follow.
CLEAN AIR LEGISLATION
The Congress first passed legislation concerning air pollution in 1955.
This law restricted federal agencies to conducting research to aid state
programs. Over the next fifteen years, Congress prodded the states to take
action but it was not until 1971 that the federal EPA was required to, among
other things, set national uniform ambient air quality standards while the
states were required to develop programs to meet them.
The first federal legislation, however, viewed air pollution as a
local phenomenon, best dealt with by local and state governments. Thus,
the federal government was restricted to conducting research into the
health effects of air pollutants and to providing grant monies to assist
the local and state agencies. Out of this research grew the idea that
the federal government would develop "criteria," or the scientific descrip-
tion of the health effects that can occur at various pollutant concentration
levels in the ambient air. The criteria would then be used as guidelines
by the states in setting ambient air quality standards. These ambient
standards (concentration levels in the atmosphere not to be exceeded) were
in turn to be met by establishing emission standards for each pollutant
source in the geographic region for which the ambient standards applied.
This division of federal/state responsibility did not last long. The
automobile industry, a national industry, did not like the prospect of
having to meet fifty different state emissions standards. Also, there was
a growing recognition that the problem did not respect state boundaries
but that a source could have interstate impacts. But the factor that
may have been the most important in changing the pattern of responsibility
was the almost total lack of progress by the states and industry in develop-
ing programs and technology to control air pollution. Thus, the focus of
legislative debate shifted to how far the federal role should be expanded
and how much the economic feasibility of controls ought to be considered
in setting health-related standards.
In 1970, the legislators looked at the singular lack of progress,
jockeyed for popular support in the upcoming national elections, took note
of the widespread public demonstrations on Earth Day in April 1970 and the
vigorous lobbying by industry against any mandated controls. They con-
cluded that if they waited until the technology was available and control
was economically feasible, it would never happen. Instead, they agreed to
set into law the desirable controls based on health rather than available
technology. This type of technology-forcing legislation, i.e., requiring
technical fixes which the experts do net necessarily agree are available, has
permeated the clean air legislation since.
-------
CN O
Cd 00
rJ O
e
0)
rH
,0
0
M
P-i
>-,
a
•H
rH
O
PH
0
CU
rH
J3
O
M
PH
00
C
•H
rH
cu
*o
£
c
O
•iH
CO
•iH
a
£
M
0)
rH
CU
-o
o
2
i— 1
cu
•a
5£
i
3 4-1
rH C
rH 3
0 4J
PH
!
! <«
i M
S3
U
!
i
I!
II
1
1
I
1
1
CU
£
O
"Z
rH
rH
,£1 •
o cd
4J
•a cd
C -a
cd
cu
-a 4J
0) -H
•o co
CU 1
cu c
55 0
MH
O
c
1 O
0) -H
JH 4J rJ
cd cu
O rJ T3
4J CU C
a 3
rH O
cd co
CO « rH S
O O CU CU
CX -H CO -H
O rJ CU >
H 4J -H CU
PH CO T3 r4
CU
CU CU
0) 4-1 CU
CU 4-1 4-1
4-1 -H 4-1
4-1 § -rj
•H ft CU E
g(U 4J H
PH cd CU
CU ~-~ 4-1 PH
PH rH C/l ~^
-~- ^ <£
* 4J >
4J rH
rH Cd O
cd *o to
a -H ^
•H rH
oo cd a
0 > 0
rH -H
O rH 4J
SH cd 03
O 4-1 60
CU CO -rj
4J cd g
CU O 3
2 o MH
1
T3 71
cu "
4J H j"
c cu 3 •
3 4-» /-s
rJ Cd T3 -O
60? CU CU
•H rH
4-1 • c td
•H /->> CU CU
E CX3 t3 a
C r~ a.
cu -^ -u cd
a oo -H
rH S C
f-l-^ C 0
•H 00 CU i-l
< *~* a co
cu
CU
4-1
--. w
^-v -H
c S
OJ rJ
w cu
M D. -H .
C C ^ /->.
•H CU '•H T3
Cd rJ CU 0)
n cu o c
!H KH U -H
CU <4-l Cd
4-1 -H C W
T3 O J3
X CO 0
CU T3 )H v_^
rH CU CU
Q. T3 CX CO
E CU CO 4J
O CU -H C
U C 13 CU
CU rH
rH CU •
3 C T3 >< O
"O O O )H 00
CU -H S CU -».
^S CO CO
O CO MH -H rH
CO -H O «W
E CU C
cu cu C Pi 5
o o o
C J2 -H • T3
cd u 4-i co
•H cd cd OO T3
rH CU 4J CTv CU
& I-l iH rH CO
e e o
O O iH S^rH
O 4-1 rH -D U
"
S
r-I
|
CN
O
CO
i
1 rJ
co o
CU 4J
U iH
0 ti
cd o cu
C S 4J
•H cd
• >
C T3 -H
0 C >-> •
cd PL, >,
CO rH 4J
4-1 C rJ
O CU O 0)
Cd rH &
CX ^3 00 O
g iH C H
M CO -H O,
1
• rH C
c cd o
O 3 O
OOrH -H C
C cd 4-1 C S
iH O Cd Cd H •
4-1 iH 60 0) CO
•H 60 iH 00 4-1 C
CO O S C 1 O
rH 3 iH 4J -iH
M O <4-4 4J JJ 4-1
O JH rJ O cd
4.) O • 0) J3 >-i
•H CU cd > CO 4-1
C 4J 4-1 C C
o cu cd o o cu
2 E 'Q o 4-1 a
00
c
•H
M
o
4-1
•H
C
«.'
(X cd
3 rJ
00
4J O
CU I-l
c/3 a.
•^^.
CU -X
CU rH
4-1 CU
4J T3
•H 0
S E
i-i
cu co
cx-
-^. M
-X CU
s e <
Q Cd PH
cn IH :d
P- CJ
CM
o
C/3
i j
i *
i i
! s ^ i g :
1 w , •* , •J
! U ' C ! S)
PH : 33 0 > !
U4 «->
' --1 , u
1 3 ! 2
•J
3
-------
CU
C
c
o
o
M
<:
H
B
cu
rH
rj
O
PH
^
CJ
•H
rH
O
Cu
B
CU
rH
r>
S
P-4
00
c
•H
rH
0)
T3
o
s
g
•H
CO
•H
a
fH
S-i
01
rH
01
T3
22
•^^
rH
01
•a
o
S
i "
rd ^
3 *"*
a.
CJ
0)
CJ
i
i
c
o
o
4-1
CO C
^ CU
CJ 4-1
Cfl 4-1
4-1 -H
co B co
ft rH
rH CU O
rH 4-1 S-I
CO C 4J
H M
gcb •
a -a
0) ft 01
(0 4-1 4-1 CU rH
01 ft 1 4-1 01
a CU 4J rH T3
ft > SH Cfl O
3 C 0 B
o o js c
CO U CO O CU
CU rQ
ft • o cu
CO 4-1 CU O
14-1 4-1 ft 4-1
O Cfl rH 00
H3 cfl cfl co
\^ 3 CU
CJ T3 C 4-1 >
Cfl C C O -H
rJ Cfl Cfl Z ^
1
•H
" C •
cu o co
" CJ 13 C
IH rH S-I O
O CU 3 O i-l
3 0 C 4J
8-3 <4-l CO Cfl
O rH
T3 C >> 4J O
01 -H XI -H
4J 4-1 >
ca ft co a
rH 3 CU O) T3
3 14-4 -H -r-l CU
oOrH ft x> r-i
01 3 cfl 3 O
pa co > co 4-1
cu
J— I
CO
4J CU
cfl o)
--, 4-1
/-v 4-1
C -H
o> B
co •< s-i
s-i pu cu
C3 Cd cu
t—4 ^- "^
^ * *
s s s
a rH Q
Q- CJ Cfl
< Cfl PL,
CNJ
0
Cfl
"co
CO
*s.
O 4J
cfl ft
4-1 O
(0 Cu
CO
rH C
rH CO
CO S-i
H H
1
cfl
« 00
00 -H
C B
•H 3
4J M-4
•H
W "
j:
ft CO
O cfl
4-1 S
•H C C
c S o
0 0 -H
S T3 4J
1
ca I xi
4-1 4-1 3 ft •
•H -H CO 0) CO
E B rC 4J
•H ft « 4J CO
rH 0) , 0) T3
ft > C
O -H CO
0) 4J
x: o co •
4J CO 4J 4-1
0) C S-i
<4-l ft Cfl O
O 4J Cu
4-1 3 CO
^ Cfl rH C
CJ CU rH Cfl
Cfl ft O S-i
r4 4J CU 4-1
1 CO •
01 ^~.
ft rH CT.
cfl r-.
PH CO -^~
M O r~-
Cfl Cu^— '
0) OI
4J 4-1
cfl cfl
0 4J
Cfl Cfl 01
^ * CO
a oi 4J
CO i-i Cfl
1—1 z
4J 01
CO SH
S-I
00 ft
01 01 •
4-1 *o
i
i
CO
o
r». -H
4-1 r-l
•H CO
iH >
•H
XI 00
•H C
X -H
O) >>
rH rH •
<4-l CU CO
CU rH
-a co cu
0) "t)
01 C 0
Z 'H B
C
O
4-1
C T3 CO
CU 01 4J
•H CO CJ
XI cfl 0)
S xi *w
Cfl U-4
TD 0)
T3 ft
oi cfl x:
x -a 4J
Cfl C rH
rH CO cfl
CU 4J CU
PA CO X!
^
PU
W «!
•~- Oi
>-) W
X < ^
•H qu O
•a cd cfl
CI "^*- XI
O> < rH
as -H
Cu t^ O
< Cd OS
X
0
1
1
*
(0
co C
01 O
•H -H
4-1 4-1
a a.
•H B
cfl 3
4J CO
SH CO
cu CB
a
as
CO
c
O 1
•H rH
W 3
co B
'e °
pj w
CU Cu
oi co
rH iH
XI O •
Cfl S-I TD
•H 4J 0)
ft C 4-J
Cfl O «)
> 0 00
CO
J O O)
en z T)
•H g
B • S
CU CO
•H >,
T3 Qj 4-1
C "O -H
Cfl O rH
e co
^ 3
oo en cr
SH S
01 O ft
-3 K T
-vj
c
i
-3
I-l
^
;J" 1 5 —
= 5: c r ^
3 ^ 3 •=• .P. P "3 c-
N -* v;
-^ T, i Z
•C
w
B
-------
This standard setting strategy depended, in part, on having air quality
models to predict the relationship between emissions and ambient air quality.
The federal EPA's research and technical assistance programs included, in
addition to the health research, an air quality model development program.
The models are developed by EPA but not in general used by the central EPA
offices. Instead they are used by the states and EPA regional offices in
setting emission standards. These agencies depend on the model documentation
and communications with the geographically-removed central EPA offices
when applications of the models or their review are necessary.
In several of the case studies we will see that this occasions delay
(Ohio SIP) and misunderstandings (Westvaco). But such organizational con-
fusion is not unique to the air quality modeling program as much of EPA's
legislation contains this dichotomy between the federal policymaker and
the state as operations manager.
It was not until 1977 that the clean air legislation required the use of
air quality models in specific situations. Such models had been used exten-
sively before this, but in the 1977 Amendments three new provisions explicit-
ly called for their use for the first time. The three sections are for
policies relating to prevention of significant deterioration (PSD) , visibili-
ty protection, and nonattainment.
The prevention of significant deterioration policy came about as a
result of litigation. It established increments in ambient pollutant concen-
trations that cannot be exceeded even if the region's air quality is cleaner
than the ambient air standards. The nonattainment policy allows for an off-
set program, a program whereby new sources can locate in an area that has
not yet attained the ambient standards only if the additional pollution can
be "offset" by reductions from existing pollution sources. The new concern
for impairment of visibility is restricted to those areas of the country
which already have the cleanest air. The legislation directs EPA to develop
a definition for visibility protection, rather than defining standards in
the law.
In each of these provisions air quality models are mentioned as a
tool for making policy and enforcement decisions under the law. However,
our investigation of how the models came to be included in the legislation
did not find any situation in which air quality models were analyzed for
what information they could provide regarding accuracy and range of pre-
dictions. Instead, the legislators considered the strategy of the clean
air legislation and concluded that the standard setting strategy depended
on having air quality models to predict the relationship between emissions
and air quality.
The starkest example is the visibility protection provisions. These
were introduced through concern for two pending power plant projects to be
located in the southwest. The environmentalists felt that even if the
plants did net violate the PSD increments, they would visibly degrade the
clean air of the area. The resulting legislation called for development of
20
-------
modeling techniques to predict visibility impairment without being able to
define or characterize what visibility impairment means.
As far as the PSD provisions were concerned, they follow the usual
form of a standard not to be violated. It was recognized that the more
stringent standards required more accurate prediction techniques. However,
the air quality modeling technique was seen as the best, indeed the only,
available alternative for assessing new sources since no new sources can
be licensed or built without a prediction of what its effects will be.
In the case of nonattainment the air quality models had been used by
EPA and the states in some instances, but it appeared that in developing
the offset policy only "best available control technology" requirements
were being applied. The models address the additional question of whether
the new source ought to be located at that particular site at all. Without
them, the decision is made on economic grounds alone. Thus, the modeling
requirement was added because the legislators felt EPA tended to ignore
the siting problem in its offset regulations.
Thus, where standards are specified in the law and models exist to
predict the impact of sources in relation to the standard, the ability of
the models to make such provisions with the degree of accuracy required
was discussed but generally ignored, because no alternatives were proposed.
The general agreement was that models should be only one policymaking tool
and that they could only be accurate to within a factor of two or more.
However, little attention was given to what other tools might be available
or needed. Attention was focused primarily on the political questions of
how to strike the balance between protection of health and economic feasi-
bility rather than how the legislation might be implemented. To the extent
that there was evidence that EPA might be tempted too far to the economic
side (nonattainment), the modeling requirement was inserted to focus efforts
on the standards.
The technology-forcing philosophy of the act and its standard-setting
strategy reinforced the tendency not to investigate too deeply whether
models to suit the new programs were available. If the health-based stan-
dards require a given level of control, then technology to achieve that
level will have to be developed. Similarly if the strategy of standards
implies the use of modeling (not if the models are available), then models
will be required.
Thus, the 1977 Clean Air Act Amendments were indeed amendments,
changes to portions of the legislation in response to perceived implementa-
tion problems and lack of progress toward the goals of the act. They were
not changes in the basic goals or strategy as established in 1970. The
modeling requirements were added to focus attention on the health-based
standards as the primary goal.
MATE?
The second case study examines the review af an application for a permit
-------
for a new stationary source in an urban area. Using the cogeneration con-
cept, the source is a power plant using diesel engines to generate both
steam heat and electricity. The controversial pollutant in this case was
nitrogen oxides and, specifically, short-term concentrations of N02.
The applicant, Harvard University, proposed building the MATEP plant
near downtown Boston, Massachusetts. They submitted an environmental impact
report (EIR) to the state in 1975, although planning for the complex of
medical institutions which would use the steam heat and electricity had
begun as early as 1968. At that time, Harvard had bought nearby housing,
which was to be torn down to make room to build the enlarged hospital facili-
ties. After several design changes to the proposed plant were made in order
to reduce pollutant emissions, the EIR was approved in 1977. However, before
they could build and operate the plant, Harvard needed an air quality permit
from the state's Department of Environmental Quality and Engineering (DEQE).
This case centers on this permit decision by DEQE.
Cogeneration facilities offer several advantages over traditional power
plants. The most significant is efficiency: by utilizing waste heat in
nearby buildings a cogeneration plant is 30% more efficient than a normal
fossil fuel plant. An ideal cogeneration site is therefore located in a
cluster of large buildings. This often means an urban site. Already
burdened with poor air quality, urban areas are difficult places to build
power plants. The highest degree of emissions control is necessary but
can still be insufficient, as when high background (existing) concentrations
themselves violate a standard. MATEP's close proximity to a medical ser-
vices complex calls for an extra degree of caution in designing and licensing
the plant. The needs of the community would be a prime consideration in
successfully designing this cogeneration facility.
MATEP got off to a good start. As designed, the plant would emit
less S02 and TSP (total suspended particulates) and would make less noise
than the plant it replaced. MATEP would provide an additional environmental
benefit by reducing the local utility's load by 73,000 kilowatts, perhaps
postponing construction of another power plant.
With the existing plant rapidly becoming obsolete, Harvard was anxious
to begin construction as soon as possible. Also, at this time cogeneration
was being praised as a forward looking, efficient source of power and, for
Massachusetts, a source which might provide an alternative to nuclear power.
This, combined with the fact that the proposal appeared to meet all appli-
cable environmental standards, led Harvard to request permission to start
construction even before the application for a permit had been submitted.
DEQE approved this request on a limited basis. In addition to starting
construction, Harvard made another decision, before getting the necessary
permits, which proved irreversible. Harvard purchased six custom-made
diesel engines for MATEP in 1975, almost two years before the possibility
of an SOo problem was raised. Later, the state unofficially asked Harvard
to look for alternatives to the diesels that would reduce NOX levels.
Bat any substitutions at this point would be more costly. Thus, Harvard,
unwilling tc change course after making such a commitment to diesels,
instead challenged claims of adverse health effects.
22
-------
The beginning of construction and the application for an air quality
permit turned out to be the spur to community groups to coordinate and
focus their opposition. Prior to 1977, several disparate groups had been
battling Harvard's expansion into their neighborhood, Mission Hill. Although
this opposition had been largely neutralized by an agreement to relocate
the MATEP plant to a site where additional housing did not have to be torn
down and to supply free steam heat to a nearby low-income housing project,
there was a legacy of bad feelings surrounding the institutional complex
which had slowly displaced residential property throughout the 1960's.
With the start of construction, as concrete evidence of Harvard's intentions
were seen, additional neighborhoods close to the plant but not immediately
adjacent also became concerned. Brookline, the center of the most active
of this new opposition, was a more affluent neighborhood and could pro-
vide professionals with the skills to interpret and critique the technical
evidence being supplied in support of the proposed plant. The combination
of these two interest groups (the Brookline people formed a coalition,
called NO-MATEP, with the Mission Hill groups) proved stronger than the
separate, individual groups because of their complementary areas of exper-
tise. Brookline supplied lawyers and scientists while Mission Hill had
experience in organizing community interests, participating in public
hearings and knowledge of Harvard's previous expansion projects. As the
name implies, their goal by this time was to stop the plant rather than
search for alternative sites or design configurations. Their tactics did
focus, however, on the scientific evidence and environmental standards as
well as legal maneuvers to delay the construction and influence the decision.
Up to this time (during the EIR review) the environmental issues
focused on the existing federal ambient standards, which were easily met
once the project was redesigned to exclude an incineration facility. Many
comments during the review criticized the assumptions and the input data
used for the modeling analysis done for the EIR. To counter these criti-
cisms, Harvard instituted monitoring at various sites near the plant and
expanded their analysis to consider the downwash question, for example.
This proved to be convincing to the DEQE who proposed to approve the plant
subject to an NOX emissions rate for the diesels.
However, Harvard's problems were just beginning. The exclusive focus
on existing ambient standards and cost efficiency resulted from the federal
air pollution legislation, which required EPA to set national ambient stan-
dards, leaving the state to enforce them and industry to try to meet them
in the most cost efficient manner. This scheme, however, is not designed
to uncover potential problems not yet regulated by EPA, and it was pre-
cisely such problems that the community groups opposing MATEP focused on
as their previous arguments based on the existing standards were dismissed.
Public concern soon centered on the short-term NC>2 levels which would
be generated by the operatic- cf the diesels. Because EPA had not yet
set a short-term standard for HG?, the state was forced to consider evidence
of health effects caused by NO^ and establish its own standard. The state
had initially been inclined to issue the permit, but the NTC>2 controversy
prompted it to solicit mere information. Resolution of the N'Oo question
was dependent on two issues: the definition of a reasonable short-tern
23
-------
N02 standard and the magnitude of short-term NC>2 concentrations MATEP would
cause. Thus, the state was interested in the opinion of health experts on
what concentration levels caused health effects, and it required the appli-
cant to do more detailed modeling in an effort to refine the estimates of
the emissions impact.
The health experts varied considerably in their opinions. The appli-
cant found experts who proposed a high concentration level, while the
citizens' groups found other experts who suggested caution and a low level.
The state solicited opinions from a panel of experts of their own choosing.
Though not ideal from a scientific perspective because the experts were not
chosen in a scientifically random or systematic way, from the point of
view of the alternative (choices of the applicant and the citizens' groups)
it was considered an improvement and the only feasible approach given the
time and resource constraints.
The evolution of the modeling analysis seems to reflect the history of
the process—the development or expansion of the issues as forced by third-
party interests. The new model analysis was more complex but it still
relied on the assumption that all NOX emissions from the plant would be
in the form of N0£ (all NOX would be converted to N02 in the atmosphere).
This was a conservative assumption because in cases where there are not high
levels of ozone in the atmosphere to form a chemical reaction with NO (the
primary constituent of NOX emissions) less NC>2 will be formed and, therefore,
a standard will be more easily met. Less than full conversion does not
appreciably affect attainment of long-term (annual average) N0£ standards.
Thus, an analysis of conversion of NO to N0£ had not been necessary as long
as there was no short-term standard. Indeed, as long as the hourly standard
being proposed was around 480 jug/m3, the modeling analysis ignored the ques-
tion of the ozone limitation because it showed the plant could meet such
a standard even if all of the NOX emissions were converted to N02- However,
as lower standards began to be considered, Harvard brought up the question
of conversion. None of the air quality models used could deal with reactive
pollutants, such as NOX, so that additional analysis was added onto the
original model estimates. Criticisms of this approach and the data con-
cerning actual ozone levels were, naturally, presented. But the major
tactic of the opposition was to focus on the fact of conversion. Because
100% conversion could take place given enough ambient ozone, they main-
tained that the most conservative, "worst case" assumption was in fact 100%
conversion. Thus, the question of how best to account for the reactive
nature of NOX was not systematically addressed because of the nature of
the process in which single point (worst case) standards are the goal.
The effect of such standards and the available models can also be seen
in the relative lack of discussion of the diesel emission rate. Although
the federal EPA sets emission rates for many kinds of sources (new source
performance standards), they have not yet done so for stationary diesel
engines. Thus, the state was once again called upon to set a standard.
In this case, once an ambient standard had been set they could work back-
wards, using the modeling analysis to determine what emission rate would be
necessary co enable the plan, tc meet the standard. The models required
data input of average monthly or annual emission rates to predict the
24
-------
short-term (hourly) ambient concentration levels. Thus, the question of
the impact of the high variability of emissions at start-up or on a short-
term basis was ignored by the model and, it turned out, also by the parties
in the dispute.
The new model analysis was more complex but, when the modelers were
asked what the level of accuracy was, they could say only that they were
accurate to within a factor of two. The result was that because the one-
hour health effects threshold or standard and the ambient air quality model
results were both estimated to be within the same range, the policymaker
did not have a clear picture as to which way the decision should go. The
lack of a federal standard seems to have pushed the decision makers into
examining more closely the accuracy and uncertainties of the model results.
The close proximity of the public interest groups to the decision maker also
helped to keep the controversy alive. The public comments seemed to have
been effective in raising questions about aspects of the model assumptions
which might otherwise have gone unnoticed by the government analysts who
were reviewing the applicant's modeling results.
The role of the citizens' groups or public interest lobbies points
out another use that was made of models in the MATEP case. Several citizens
have stated that they oppose the construction of any power plant in an urban
area and they specifically oppose the expansion of a large institution like
Harvard University into their residential neighborhood. Thus, all infor-
mation that can be used against the applicant, and only such information,
is presented. Of course, the applicant also presents only such information
as is advantageous to his position. But that is the purpose of the public
hearings, to hear both sides of the issue.
For the decision maker who is not familiar with the details of the
modeling, however, it becomes a question of how to choose among the experts.
Numerous lengthy hearings have been held on the matter to ascertain how
the experts arrived at their conclusions. The decision shifted several
times over this period as new issues were raised.
To summarize, the models in this case were used to redesign the project
to meet applicable standards. Neither the applicant nor the state environ-
mental agency looked beyond what the models could tell them or beyond the
strict requirements of the legislation until third-party interests forced
them to do so. However, che public interest groups were not motivated
either by a search for the best alternative. They were concerned with
stopping the institution behind the project, who they felt had ignored their
concerns in other areas for many years, and used the environmental arguments
as tactics in this battle. The resulting controversy stimulated the decision
maker to consider just what the assumptions and model's methodology and
accuracy impliea. The uncertainties revealed in the process helped shape
the final decision which requires monitoring of the plant's impact. When (if)
the monicoring results exceed a certain level (230 ug/m3), the diesels will
have to be snuc down and electricity purchased from the local utility.
-------
PI ITS TON
The third case involves the application by the Pittston Company for
permission to build an oil refinery in Eastport, Maine. Eastport is a rural
community on the northeast coast of Maine near the Canadian border. It
is also near a national wildlife refuge and an international park and the
project is subject to the prevention of significant deterioration (PSD)
provisions of the Clean Air Act. These provisions establish increments
in air quality which cannot be exceeded by any combination of new projects
in such areas. The increment under contention in this case was the 24-hour
sulfur dioxide standard of 5
Pittston had first sought approval from the state of Maine in 1973.
However, state regulations at that time did not cover air pollution programs.
Thus, EPA was the government agency responsible for reviewing the air quality
impacts of the proposed oil refinery. The Pittston Company submitted a
draft environmental impact statement (EIS) to EPA in 1976 and formally
applied for a PSD permit in the spring of 1977. EPA's initial review of
the application was positive. But, when the Clean Air Act Amendments were
passed in the summer of 1977 designating nearby national parklands as
Class I areas, areas to be protected against significant deterioration of
the air, Senator Muskie of Maine, as Commissioner of the park and as EPA's
prime supporter in Congress, objected to the proposed refinery. The
regional office then undertook its own analysis of the project and announced
it could not approve it based on this analysis.
Over the next year, in response to both EPA's and several public
interest groups' analyses, Pittston redesigned the project and was able to
win EPA's approval for an air permit in August 1978. Pittston, however,
has been denied a wastewater discharge permit by EPA so that it is not
expected to begin construction in the near future. This case study is con-
cerned with the air quality models that Pittston used to redesign the project
and that EPA used to review and approve the air permit.
EPA has made several air quality models available to the states or an
applicant by means of a computer program. The Pittston Company used one
of EPA's models to show that the oil refinery's emissions would cause in-
creased 24-hour S02 concentrations at the park of 4. 3 jug/m3 which is within
the incremental limit of 5.0 ;ug/m3.
The organizational unit within EPA responsible for evaluating the
application and approving or denying the permit was its regional office
located in Boston, Massachusetts. The regional office conducted its own
modeling analysis using different models and came up with numbers two to
three times as large as Pittston 's numbers, which, therefore, indicated
violations of the standard. During the next several months negotiations
were held with the applicant. They centered on how the available models
should be modified for the unusual site (coastal) characteristics, the
assumptions concerning proper methods for deriving 24-hour concentration
levels, and what constituted reasonable worst-case meteorology.
Two problems of significance to this study arose during the review of
26
-------
the permit application and remained unresolved. The first problem was that
the meteorologic data used for the analysis came from a site over 100 kilo-
meters away. Later, data from a site about 25 kilometers away was also
used. The unique meteorology of the coastal site generated much debate on
just how representative this data was. Although under the 1977 Amendments
to the Clean Air Act, EPA can require an applicant to collect meteorological
data, no attempt was made to require on-site monitoring in this case. The
amendments were passed in the middle of the permit review and the EPA
Region I office decided to "grandfather" the project and follow the regula-
tions under which the application had been originally submittted before the
new legislation was passed. Although the particular standard in question
was not changed by the 1977 amendments, several procedural matters and
some control requirements were, and they would require more time and re-
sources from both Pittston and the EPA regional office. The decision to
"grandfather" the project precipitated a law suit from several public
interest groups as well as leaving open the question of the suitability of
the meteorologic data.
The second unresolved problem concerned the results of a validation
study for the model used in the analysis. The one coastal site studied
showed that the model underpredicted, and this was pointed out in the
public comments to the EPA regional office. ERA's response was that under-
prediction had not been found in other studies. However, this evaded the
question because other validation studies of coastal sites had not been
done. Several other modeling issues were also discussed, such as fumigation
analysis and a 24-hour scaling factor. These issues, however, were resolved
by calling on recognized experts within EPA who were able to agree on one
technique.
After their review of the draft EIS and Pittston's permit application,
EPA proposed to disapprove the project based on their modeling analysis.
The company responded by making major design changes in the project in an
effort to reduce the impact of emissions to within the allowable limits.
For example, they eliminated an anchorage site for tanker traffic and
lowered the sulfur content of the fuel oil. Thus, the models were used as
a design tool to trade off economic efficiency and environmental impact.
However, certain emission parameters were also changed, including the
number of stacks and exit velocity from the stack. In this case the pro-
ject was designed around the model. Without changing the total amount of
the pollutants, the models' specificity was exploited. The models determine
ambient pollutant concentrations by relating them to emission rates and not
total emissions. The models were not designed to answer the question of
build-up of airborne pollutants in other media, such as water or land, or
long-term effects. Thus, both the applicant and the decision maker ignored
these questions because they did not have the means to address them.
The sequence of events which led up to the final approval of the permit
implies another use of the models. The decision maker, the EPA regional
office, was also a member of a governmental task force assembled to coordi-
nate the task of siting oil refineries in New England. Thus, one objective
cf the EPA regional office was to help in finding a suitable site for an
-------
oil refinery. Sites other than the proposed site in Eastport, Maine, were
discussed in the environmental impact statement but were rejected on economic
grounds. Negotiations were then conducted between the agency and the appli-
cant in an effort to design the project so that it would conform to the
environmental constraints. Thus, it was politically advantageous and in
the interest of EPA to approve this permit. The way was made easier because
of the way the models are constructed and used. The model results are
presented as single point estimates to be matched against a single standard
value. Worst case meteorology is used and no sensitivity analysis was done
to account for the possibility of underprediction. Moreover, the documen-
tation and guidelines for choosing assumptions are ambiguous. The EPA
regional office was able to pick one end of the suggested range for the
wind speed, for example, without further justification other than "it ap-
peared in the guidelines."
It was the job of the scientists in the EPA regional office to run the
models, justify the assumptions, and respond to public comments. Discussions
with the regional administrator, the ultimate decision maker, seem to have
been limited to the legal question of when, not whether, the permit should
be issued. Apparently they did not discuss the questions of how accurate
the model results were or what environmental impacts may not have been
measured. Although the allowable PSD increments at issue here appear to
be of the same type as the other standards EPA administers, they are easier
to violate and demand more precise measurement. Because the decision maker
was not familiar enough with the air quality models to make this distinction,
he had to rely on the analysts. Yet the analysts did not raise this issue
because they did not have other methods to draw on.
To summarize the findings in this case study, the models were appro-
priately used to redesign the project to reduce the impacts on air quality.
They may have been used inappropriately if long-term and other impacts, such
as acid rain, are considered, because the models are not equipped to deal
with such problems. The analysis also may have been used to justify a
decision taken on other grounds. This was facilitated by the lack of
familiarity by the decision maker with the models, the tendency within EPA
to emphasize worst case analysis, and the lack of validation studies and
documentation. The end result of these tendencies is one point estimate
compared with the standard level without reference to accuracy or other
uncertainties.
The role of the public in this case was somewhat circumscribed. The
main opponents were the local citizenry, who lived in or near Eastport and
were fishermen who felt their livelihood was threatened by the oil tanker
traffic, and federal government officials concerned with the air quality
impacts of the project on nearby federal park lands. Thus in this case,
as opposed to the MATEP case, the citizens' groups opposing the permit
were located in northern Maine and in Washington, D.C., while the decision
maker and applicant were negotiating in Boston. This caused lel-avs in
obtaining information so that often the public c?tnsien;s were ic I^r.ger rele-
vant by the time they reached Boston. It was alsc f rust ran .15 t'or c.ie
publi: interest groups because they could not participate ~n the negotia-
tions and the changes in the project design and modeling analyses would
28
-------
take them by surprise. They were constantly put in the position of request-
ing time extensions for the public comment periods so that they could
receive the new materials and do their own critique.
EPA was under some time pressure to come to a decision on the permit
application because it was grandfathered under the 1977 Clean Air Act
Amendments, and this exemption from the new regulations could not be extended
indefinitely. Thus, EPA was reluctant to grant the public comment exten-
sions except when requested by Senator Muskie.
The public interest groups did submit critiques of the modeling analysis
and in one instance were able to get a change in the project design (elimina-
tion of an anchorage site). However, many of their issues, such as the lack
of suitable meteorologic data and validation studies for determining appro-
priate input parameters, were not resolved. Indeed, EPA technical documents
recognized the problems but no policy discussions on the issues seem to have
been held. The reasons for this appear to be the pressure to make a decision
under the new amendments and the desirability of the site for purely economic
reasons, as well as the lack of analytic tools to handle the questions and
the geographical separation of the public interest groups from the EPA
regional office.
The relative importance of this factor can be seen in comparison to
the MATEP case. In both the Pittston and the MATEP cases the public asked
questions which involved additional analysis on the part of EPA (or the
state) and which did not appear on the agenda until the public interest
groups raised them. However, the proximity of the interest groups to the
decision making agency in the MATEP case seems to have been part of the
reason for a greater and more detailed response from the state to the
environmentalists' concerns.
ANACONDA
The Anaconda copper smelter is located in Anaconda, Montana. The
smelter and the adjoining copper mine are virtually the only source of employ-
ment for this area of Montana and the only source of S02, as well. The
operations provide close to 12% of the domestic U.S. copper production.
Montana established state air quality standards in 1967. It was not until
1971 with the establishment of federal standards that the company agreed
to install control devices to reduce emissions. Their proposal was to meet
the 90% reduction as required by the state by 1973.
Before an agreement could be reached, however, events at the federal
level caused Anaconda to retract its proposal and claim that such stringent
controls could not be required. The federal Office of Management and
Budget (OMB) had recommended that EPA delete the requirement for emissions
limitations from their SI? guidelines to the states and that economic
impacts be considered in cnocsing control measures. EPA revised its guide-
lines to reflect cne ,vM3 recommendations. Anaconda subsequently petitioned
the state to ra-'zse its SI? regulation, requiring 90% reduction in S02
emissions, citing the new SI? guidelines from EPA.
-------
Within the state government there was disagreement on what the SIP
regulations ought to be. The State Board of Health continued to push for
the 90% removal requirement, but the governor did not support it. Without
the governor's approval, EPA was forced to propose its own S02 regulations
for Montana. The EPA proposal for the Anaconda smelter turned out to be
based on faulty and very limited data—one 24-hour sample from a monitor
which was found to be malfunctioning and uncalibrated. Anaconda went to
court over several issues concerning EPA's proposal. Regarding the data
base the court reprimanded EPA for using what it felt to be an arbitrary
figure which EPA intended not as defendable, in fact, but as bait for
discussion in subsequent public hearings.
The court's opinion that EPA had done an inadequate job in proposing
a standard originally seems to have been the incentive needed for EPA to
commit vast resources to modeling this particular copper smelter. EPA had
learned that, while they may be given the benefit of the doubt on technical
issues, the court would not stand for no effort to compile a technical basis
for its decisions. Also, non-ferrous copper smelters were singled out
in the Clean Air Act as sources particularly difficult to control and,
hence, potentially able to receive exemptions from complying with standards.
EPA had developed the "Valley" model for this particular situation and wanted
to prove its worth.
Thus, EPA began to collect data near the smelter and develop a model
for analyzing the impact of copper smelters located in mountainous terrain.
EPA was faced with developing regulations for several such copper smelters
so the model was not just for this one unique situation. What was unique
was the use of helicopters to monitor actual impacts for validating the
model for the Anaconda plant. These were supplied by the EPA Las Vegas
laboratory, which was relatively close by and whose mission included working
to upgrade EPA's monitoring program, so that the Anaconda project fit with
its responsibilities as well.
Throughout this modeling effort, EPA proposed specific emissions
limitations based on the modeling results. The company would then criticize
the model and make its own proposal which was in turn generally supported
by the state. The state was reluctant to take action against such a large
employer in its state. Indeed, at one point the company threatened to close
the plant if EPA's proposal went into effect. Again, EPA's response was
to do its own analysis. In this case it hired a consultant who concluded
that, while economic impacts would be significant, it was unlikely that the
smelter would close. Thus, EPA continued to upgrade its technical analysis
as its response during the decision process.
By January 1980, all parties had agreed to a compliance schedule for
the Anaconda copper smelter which would mean an emission limitation of
142 tons per day (down from current emissions of 170 TPD) by 1988. At the
final hour an EPA agency policy (that smelters would be subject to a 6-hour
standard as well as a 24-hour standard'; was arbitrarily applied and raised
objections from the company. But on further reflection a compromise was
reached and the emissions limitation based on the model analysis stood.
30
-------
The case of the Anaconda copper smelter in Montana is an example of EPA
spending an unusual amount of resources to conduct and defend its own
modeling effort. The company criticized the modeling as it progressed
but did not do its own modeling. However, the company did install, over
the years, pollution control equipment which resulted in a total of almost
80% reduced emissions. Also, the final decision by EPA was based directly
on its modeling analysis, after it was revised to reflect the company's
comments and validated by extensive and expensive monitoring data.
The agreement, however, seems to be due to a change in the attitude
of both the governor of Montana and the company as much as it was due to
the extensive monitoring and validation efforts of the EPA Las Vegas
laboratory. Early in 1978, the Atlantic Richfield Company bought the
Anaconda Company. Their willingness to make this investment took into
account the pollution control problems, and they were willing to negotiate
with EPA where Anaconda had not been. Also, the outcome of the state elec-
tions was such that the governor and the State Board of Health were in accord
for the first time in this process so the state was willing to participate
in the negotiations.
The dependence of the eventual outcome upon the willingness of the
separate parties to view the agreement as in their self-interest becomes
clear in the events subsequent to the approval of the compliance schedule.
In October 1980, Atlantic Richfield announced the closing of the smelting
operations. Their decision, which led directly to the loss of jobs by two-
thirds of the inhabitants of the town of Anaconda, followed a nationwide
strike of copper workers. Striking workers settled and went back to their
jobs at all but the Anaconda smelter. Also at this time, Montana promul-
gated stricter air quality standards. The company chose to cite the
environmental and health and safety regulations as the cause of the plant
closing. The company has obtained a seven-year, multimilion dollar contract
with Japan to smelt the copper mined in Anaconda.
WESTVACO
Maryland's SIP requires all plants either to burn 1% sulfur-in-content
fuel or to install emission control equipment. Since 1975, the Westvaco
company has sought an exemption to these requirements for one of its pulp
mills claiming it could burn up to 2.7% sulfur fuel without violating ambient
air quality standards. In 1974, Westvaco had bought several nearby coal
mines but would not be able to use this coal in their mill if they were
required to burn 1% sulfur fuel. Over the years, they submitted monitoring
data and an air quality analysis in an effort to gain approval of their
exemption request.
The state of Maryland approved Westvaco's request. The company is
a large ainpioyer throughout the state, and state officials cited the energy
anc employment problems and the competitive disadvantage for them in not
exempting Westvaco as part of their concern in this matter. The EFA
rfcgi;nai office, however, did not support approval of the exemption. Their
review of ;he company's air quality analysis convinced them uhat violations
31
-------
of S02 standards would occur unless the lower sulfur coal was burned (or
other controls applied).
In 1976, Westvaco completed construction of a smokestack 600 feet high.
Under the existing legislation, this was legal and could be done without
EPA's approval. It was not until the 1977 Amendments to the Clean Air Act
that taller smokestacks had to be justified, as preventing a downwash
problem, so that they would not be used to export the pollution problem
downwind.
Also in 1976, Westvaco installed S02 monitors in an effort to show
that no violations occurred when the higher sulfur coal was burned. The
land where violations of the standards were expected to occur (to the south
and southwest) belonged to Westvaco. Westvaco chose not to monitor there,
and the federal EPA had not issued a policy for monitoring on private land.
Throughout this time they were reviewing the issue but had come to no
decision. Instead, Westvaco sited two monitors to the north and one to the
east at the bottom of the valley (600 feet below the top of the stack).
Safe in the knowledge that the monitors were not likely to record the highest
concentrations, Westvaco claimed that the monitoring evidence showed no
violations of SC>2 standards. If there were no violations monitored, they
reasoned, there was no barrier to granting the exemption. The EPA regional
office continued to point out that the monitors were not sited so as to
detect violations, but it turned out to be a battle of words. EPA was unable
to persuade Westvaco to change their monitors, and i-t lacked the resources
to do its own monitoring.
Another controversy concerned the area where EPA suspected that viola-
tions would occur (to the south of the mill). Westvaco claimed that this
area was inaccessible and uninhabited so that standards to protect public
health should not apply there. Even after EPA discovered a secondary road
and several homes there, the company did not drop its claim in its subse-
quent pleas to Congress.
A third claim by Westvaco was also not strictly according to the facts.
Westvaco claimed that low sulfur coal was not available locally. However,
the Sierra Club, in opposing the exemption request, showed that low sulfur
coal was available but it would cost more and it was not part of the coal
resources owned by Westvaco.
The EPA's regional office response was not without problems, however.
Because they lack the resources to do modeling for each permit application,
their policy is to do no modeling of their own but, rather, to criticize
the applicant's modeling in an effort to get changes in the assumptions or
input which they think will more closely approximate "worst case" concentra-
tions. For almost two years (November 1975 to August 1977) EPA outlined
its objections co Westvaco's model. The major problem with the modeling
analysis vas the complex or hilly terrain surrounding the plant. There
was not a node! available which would readily deal with this problem so
that the criticism dealt with how Westvaco had altered the available models.
EPA criticized the fumigation and trapping analysis, the lack jf on-sita
meteorological monitoring, the assumption of no 302 background levels, the
32
-------
terrain reflection factor and the sector averaging assumptions. These last
two assumptions, for example, are used to model the impact of elevated
terrain on the vertical and horizontal dispersion of the plume and had
a significant impact on the model results. Westvaco's numbers for these
factors produced results one-fourth as large as results using EPA's suggested
numbers "for the same factors.
The example of the terrain reflection factor (where Westvaco modelers
assumed total absorption and EPA assumed total reflectivity) is similar
to the disagreement over how to treat individual sources at the plant.
Westvaco modelers suggested averaging the sources and EPA said common
modeling practice treated separate plumes as additive (pointing out that
Westvaco's approach would tend to cause each additional source to decrease
rather than increase the total impact of the plant). Both of these examples
show the opposite extremes chosen by each side. Throughout the three-year
modeling process they were unable to reach a compromise on these and most
of the other areas of contention. This result seems to have been due, in
part, to EPA's approach of stating in the most technical of terms how it
should have been done but not providing any financial incentive for Westvaco
to comply, not providing model results for comparison with Westvaco's analy-
sis, and not addressing the broader issues of employment and energy that
Westvaco and the state raised.
Thus, Westvaco was unwilling to change either its modeling or its
monitors. In the face of this intransigence, EPA proposed to disapprove
the exemption request.
Westvaco's response was to appeal to Congress for support. Several
senators and representatives from the region held positions on various
committees which oversaw EPA's budget. It was to these congressmen that
Westvaco appealed. Westvaco emphasized the fact that no violations had
been monitored, that its models predicted no violations, and that EPA had
produced no modeling analysis of its own to show violations. The company
also stated its concern that EPA's disapproval of its exemption request
would have adverse consequences for the employment and energy situation of
the region. The congressmen, in turn, wrote EPA expressing their concern
that EPA was not giving due attention to the Westvaco request.
The culmination of these exchanges of letters was a meeting set up
by Senator Randolph where Westvaco consultants came prepared to fully defend
their model but EPA appeared unprepared, having misunderstood the purpose
of the meeting. EPA had assumed that a congressman was interested in the
policy issues rather than the technical issues, and did not have technical
presentations ready.
There generally seems to have been little communication between the
parties except on the narrow issues of the Westvaco model. However, after
the meeting the EPA did hire a consultant to do some modeling and did begin
to respond to the range of technical issues raised by the congressmen as
well as by Westvaco. But the agency was stilJ circumscribed by its plans
for proceeding with the SIP revision process. This is exemplified by its
response to the suggestion of a joint industry-EPA study of models. EPA
33
-------
refused on the basis that its modeling had already been "objective." It
did not see any need for, perhaps, establishing a legitimacy to the process
in the political arena beyond the legal procedural requirements.
In the end, the EPA model was used to help site a new monitoring net-
work which Westvaco is required to run during a three-year variance which
will allow them to burn higher sulfur fuel. Already it appears that the
new monitors do indeed show violations, as EPA has contended. But the more
appropriate monitoring network was established more than four years after
monitoring was first begun. The delay seems to have been in part due to
the lack of an EPA policy on monitoring but also due to EPA's conception of
how such decision processes ought to be managed. EPA expected the Westvaco
modelers to respond to its criticisms and continued to review only the
narrow issue of modeling even after the controversy had been taken to
Congress and expanded to include other issues such as employment. EPA,
supported by the courts, had taken the position that the Clean Air Act
requires a polluter to prove his case on his own time, i.e., that emissions
must be reduced to comply with SIP provisions during the time an exemption
is sought. However, EPA did not enforce this policy in the case of
Westvaco and, after the issue was appealed to Congress, its decision was
to allow the burning of higher sulfur coal during the three-year monitoring
program. Thus, Westvaco had no incentives to change its modeling or moni-
toring until after EPA did its own analyses that pointed out the faulty
monitor siting and required changes in Westvacofs monitoring network as a
condition of continued operation of the plant.
MASSACHUSETTS' SULFUR REGULATIONS
In 1970, the state of Massachusetts set an ambient SC>2 standard and
developed regulations for meeting this standard; the regulations included
a restriction on the sulfur content in fuel that could be burned in the
Boston area. The restriction to 1% sulfur fuel at first and then to 0.5%
sulfur fuel was based on technology (the fact that such fuel was available
even if more costly) rather than any link to ambient air quality. It was
not until the energy crisis (with rising fuel prices) that the public, and
the Massachusetts legislature, required a technical justification based
on the impacts on ambient air quality for these regulations.
By 1973 a few modeling efforts had been undertaken. The rollback
model was used by the state to justify its 0.5% limitation. A study by the
Harvard School of Public Health (whose project director would later be
hired by the utilities to perform modeling analyses) predicted that the
regulation could be relaxed to 1% without violating the state ambient stan-
dards. The regulations were indeed relaxed in response to the technical
analysis and widespread public support in the face of rising fuel prices.
This relaxation did not, however, solve the problem. In 1974, the
Massachusetts legislature passed a law that required the state standards
to be no stricter than the federal standards and for the standards to be
met in the most cost effective manner. In response, several interested
parties initiated modeling analyses. Each party (utilities, the state, and
34
-------
EPA) formulated the question to be answered by the model from their own
perspective. The utilities wanted to be able to burn higher sulfur fuel
as often as possible and proposed fuel switching plans. The state asked
what percentage sulfur fuel would be sufficient to assure that the national
standards were met; they wanted to change the content but not the form of
the existing regulations. EPA was concerned with enforcement of any fuel
switching plans so tended to favor permanent limitations but also was con-
cerned with the possibility of future growth so that just meeting the stan-
dard was not enough. Thus, although each interest used an air quality
model to predict ambient air quality, their different perspectives led to
different conclusions.
All of the models that were used were, however, an improvement over
the previous technical analysis. For instance, the state developed an
improved emissions inventory (still based on fuel usage rather than actual
emissions) which allowed for more accurate geographical distribution of
emissions. Also models which calculated the varying impact of certain
meteorological parameters such as different wind directions and speeds and
stability classes were used. Questions remained, however, about the tech-
niques used to estimate 24-hour concentrations from either hourly or annual
averages.
In all, four modeling studies were done: one by the state, two by the
utility interests and one by the EPA regional office. In theory, numerous
analyses of one problem might give the decision maker (in this case, the
state, but with EPA's approval) a good range of data with which to work.
However, in practice, there was very little effort made to coordinate, relate
or compare the various modeling efforts and the consequence was that the
state felt free to base its decision solely on its own model analysis and
its own perspective on the problem.
None of the parties felt that it was its responsibility to contrast
or compare the analyses. EPA criticized the state's modeling assumptions
and data but did not reach an agreement on what should be done. In its own
analysis, a different model was used but, further, EPA's consultant did not
analyze the alternative actually chosen by the state. Instead, EPA continued
to push its model as superior for answering its own particular questions.
The result was that the state made a decision based on its model and EPA
was forced to agree since no violations were being monitored and experience
supported the state's stand.
EPA had maintained that the state's analysis was not accurate enough
to regulate large point sources and plan for future growth. But the state's
situation was that a deadline for the regulation had been set, it did not
have additional resources to do more modeling, no standards violations were
predicted by its models or had been observed, and the legislation had clearly
mandated less restrictive erJ.ssi.ons limitations. Thus, the state proposed
as a trial to implement: its alternative for one year to see if violations
occurred. la this lignt, the SPA regional office agreed that the state's
modeling was better tnan could be expected under the circumstances.
It should oe noted that in the end much of the resistance to the fuel
25
-------
regulations (as opposed to the modeling) came from the EPA's headquarters
office rather than the regional office located in Boston. Several elements
of the state legislation which precipitated the fuel regulation review
appeared to be inconsistent with national policy. These included the mention
of taller stacks and fuel switching as possible solutions. From a national
perspective tall stacks can only be justified in cases of a downwash problem.
The state's modeling did not include stack extensions but this had to be
explicitly clarified in the discussions. EPA also objects to "intermittent
control systems," such as fuel switching based on meteorological conditions,
because of the enforcement problems involved. The utilities had requested
such a system but the state was proposing permanent limitations. Where
a plant's emissions impact was predicted to be near the standard these
sources were required to maintain monitoring systems and a short-term supply
of low sulfur fuel in case the models had underpredicted. Thus, the state
was able to meet EPA's objections when the discussions were focused on the
policy and implementation implications.
Generally, EPA objects to regulations that carry the presumption that
they may have to be tightened. Rather they work for regulations that are
conservative and may have to be relaxed in the future. The court upholds
this view of the SIP process. This view is based on the cost and time re-
quired to reach compliance when capital expenditures for control equipment
are required. However, in this case, over the years it was shown that plants
could readily change the type of fuel they used. Indeed, the utilities pro-
posed compliance through fuel switching. Thus, a trial period to test the
model's predictions without requiring the higher expenditures on low sulfur
fuel may be a good solution given the relatively quick response possible to
regulation changes. It was satisfactory from the state's perspective because
it felt that its monitoring network would indicate compliance or, if not,
that stricter regulations could be promulgated. EPA tended to resist this
view, since, from a national perspective, a conservative (stricter) approach
was necessary for those states reluctant to enforce Clean Air laws.
OHIO'S STATE IMPLEMENTATION PLAN
Ohio was the last large industrial state to propose a State Implementa-
tion Plan. However, it could not agree on what S0£ control plan to include
so EPA formulated one instead. The EPA Region V office first proposed a
plan based on the linear rollback technique. As a result of public hearings
where extensive criticisms of the plan were voiced, EPA replaced its analysis
with one using an air quality dispersion model called RAM.
The state had taken from 1972 to 1975 to propose an incomplete SIP.
EPA's S02 plan based on RAM was promulgated in 1976. It contained specific
emissions limitations for Ohio's power plants. The utilities took EPA to
court questioning EPA's procedure in setting emissions limits (replacing
linear rollback with RAM without additional public hearings) as well as
the capability of the RAM model itself. The state joined the utilities'
suit, questioning EPA's authority to impose a plan on the state. EPA
corrected the procedural problems, and the court upheld EPA's authority
and decisionmaking based on the RAM model. Through the court suits, however,
36
-------
the state and the utilities had gained considerable time. The court ruled
in favor of EPA in February 1978, six years after the legislative deadline.
But the court's ruling did not guarantee action. The utilities re- ,
sponded with requests for specific revisions for their plants. This case
study looks at the actions taken by the Cleveland Electric Illuminating
Company (CEI) for its plants near Lake Erie.
At the time CEI had joined the court suit against EPA's use of RAM, it
had also set up a monitoring network around its plants and began its own
modeling study. These actions by CEI effectively set the stage for the
ensuing dialogue between EPA and CEI over the requested SIP revision, a
dialogue which was to last for almost three years.
EPA did not take any steps to influence CEI's modeling efforts until
after the court upheld the SIP. At that time they documented their criti-
cisms of the monitoring network and CEI's use of the rural version of the
RAM model. EPA asked for corrections from CEI and the discussion came to
dwell exclusively on model assumptions and accuracy and the comparison of
EPA's application versus CEI's.
EPA had problems with much of the company's modeling analysis. The
rural RAM under-predicted at several of the monitoring locations; neither
the analysis nor the monitors were arranged so as to show maximum impacts;
their fumigation analysis only addressed one of the stacks; their monitoring
network lacked spatial distribution; and the company's analysis depended
on building taller stacks to avoid downwash but defended this action with
an old study based on different operating characteristics.
The EPA Regional Office could only ask for better modeling and threaten
CEI with disapproval of its SIP revision request. This narrow sphere of
responsibility caused two problems: (1) EPA was unable to resolve the
question of whether CEI could build additional height onto its stacks in lieu
of reducing emissions and (2) EPA's threat became a hollow one when a
statement by the federal Council on Wage and Price Stability (CWPS) effec-
tively eliminated all solutions except the non-compliance one. In the end,
the EPA Regional Office was left only a defense on technical grounds of a
model they admitted did not cover some important issues (downwash and fumiga-
tion) and had no means to require the company to provide better monitoring
data. Also, while CEI's model underpredicted, EPA's model (urban) RAM over-
predicted in some cases. But requests by EPA for CEI to use a better model
fell on deaf ears.
The company, of course, acted in its own self-interest. It sited its
monitors so that the model would come up with the "right" answers; it
selectively used data from an old study purporting to show a downwash problem
so as to justify the building of taller stacks; it went on to build the
taller stacks in the face of EPA's indecision on a policy for this issue;
it failed to do a fumigation study, preferring to claim it did not present
a problem though without data to back che claim; and it continued to submit
data only from the original monitors. All this served as delaying tactics
whila the Ohio utility industry sought help trom Congress and the White House.
37
-------
The Section 125 amendment to the Clean Air Act in August 1977 and the CWPS
statement in March 1979 provided that assistance.
Section 125 of the Clean Air Act (first proposed by the Senator from
Ohio) allowed the President to restrict a power plant to using only locally
available fuels if this was necessary to prevent local economic disruption
or unemployment due to a switch in fuel use in order to comply with a SIP
requirement. The CEI plants are the largest users of high sulfur Ohio coal.
CEI threatened to switch to out-of-state lower sulfur coal if forced to com-
ply with EPA's S02 plan. The governor of Ohio asked that Section 125 be
invoked. EPA proposed to do just that. This would leave CEI with three
options: (1) mix high and low sulfur coal (more costly than just high
sulfur coal), (2) install costly "scrubbers" to clean the stack emission, or
(3) not comply with EPA's S02 emissions limitations.
At this point the President stepped in, through his Council on Wage
and Price Stability. CWPS issued a statement saying that the economic
disruption would not be as severe as EPA had indicated but that scrubbers,
by raising electricity costs, would cause unemployment. The statement con-
cluded by saying that economic considerations would likely delay compliance
with the SIP. CWPS, thus, chose the non-compliance option.
The debate returned to the modeling analysis. EPA again requested
better monitoring and modeling and CEI responded with more data from its
original monitors. The final decision, breaking this stalemate, was appar-
ently made by the EPA headquarters' office. In May 1979, CEI was informed
that, while neither EPA's nor CEI's data seemed to be sufficient for a
modeling analysis, "status quo" emissions rates were being proposed because
of the severe economic hardships involved. This decision did, however,
require expanded monitoring and testing around the plants to improve the
analysis and possibly provide for restricted emissions in the future.
This case points out the consequences of the narrow perspective of
the EPA modelers. Their job appeared to them to be to produce the best
model analysis that they could, including critiquing the industry models.
This narrow focus on the models as the issue hid the true issues. For
instance, the issue of long-range transport of pollutants and acid rain from
the taller stacks was never really addressed. Only the question of whether
EPA wanted to impose a case-by-case demonstration of downwash or whether this
represented an "undesirable burden" on existing sources. The question of
what impact the taller stacks of the CEI plants had on out-of-state areas
downwind of the plants was not of concern to the state and utility officials
and, also, could not be answered by the available models. EPA headquarters
in the face of the uncertainty and opposition could not decide on a "tall
stacks" policy, and the company was free to build them although the final
decision did include provisions for further analysis of this problem.
A second example of an unaddressed issue is the full range of economic
consequences. The models do not deal with costs, and the Clean Air legis-
lation only allows cost-effective considerations in wha: controls to impose,
not on whezher to impose controls. However, the final recisijn stared that
economic considerations were paramount in this case. Ihe difficulty is,
38
-------
then, that no strict analysis based on economic impacts was done. Numerous
statements about costs were made in public hearings, for instance, but no
official decision or technical document compared the costs of the various
options and whether the costs of an option other than non-compliance might
be acceptable. Thus, EPA, seemingly restricted to analyzing air quality and
not cost impacts, eventually made the decision to delay compliance based
on economic grounds which were not fully analyzed.
A third issue not appropriately addressed was the monitoring and collec-
tion of input data. The lack of attention at the very beginning when the
company set up its monitors lost for EPA the ability to collect appropriate
input data. Also, by limiting their actions to criticisms and requests
for changes, EPA did not provide any incentives for the company to do a
better analysis. Instead, the company's response was to lobby the state,
Congress, and the White House for support. The EPA Regional Office is re-
stricted in what it can do, of course, and did try to get EPA headquarters
to issue a tall stacks policy but it was essentially alone in its opposition
to CEI. The active opposition by the state, Congress, and the White House
as well as the passive attitude of EPA headquarters undermined the work of
the Regional Office. They were supported by the court and the EPA Office
of Research and Development. However, the court can only act when asked
and ORD only spoke to the incomplete modeling analysis. Thus, the question
of tall stacks, long-range transport and acid rain was not addressed because
the discussion of issues was never expanded beyond the narrow local impacts
that the available models addressed while the spectre of economic hardship
(without technical analysis of the issue) won the day.
CONNECTICUT'S TRANSPORTATION CONTROL PLAN
The state of Connecticut had to develop a plan for controlling pollution
sources within its borders so that the national ambient air standards would
be met. The particular portion of the plan that this case investigates
is the transportation control plan (TCP) aimed at meeting the ozone standard.
When work on the implementation plan first began in Connecticut, EPA
required the use of a technique called Appendix J, a modified form of
rollback. The graph to be used by the states was based on data from such
cities as Los Angeles. Thus, Appendix J was criticized as being inapplicable
to other cities not included in the graph. The method for generating the
graph which would be specific to a particular city required five years of
data on ambient pollutant concentrations. Such data was not available in
most cases. Thus, many states pressured EPA to allow them to use the
linear rollback method. The only other alternative was complex photochemical
dispersion models which required even acre input data.
Apart from tne problems of data Connecticut also was beginning to dis-
cover the problem of transport, tne flux of ozone and ozone producing
pollutants into their state frcn ether, upwind states such as New York,
New Jersey and even midwestem states. Neither rollback or Appendix J,
nor EPA policy, reco-ciizad tbe transport factor. They treated all pollution
as having been produced, arci therefore subject to control, locally. The
39
-------
result was modeling analyses requiring 90 percent control of all sources.
A number considered by Connecticut as impossible and, because of transport,
unnecessary to achieve.
Thus, Connecticut decided to develop its own model to take the pollutant
transport problem into account. This model consisted of a fixed grid system
with an extensive inventory of source emissions superimposed on it, with
assumptions made about how the sources mixed and were dispersed within each
grid. It estimated the effects of transport and reduced emission reduction
requirements accordingly. Connecticut's model also attempted to portray
the non-linear relationship between ozone and ozone precursors (hydrocarbons
and nitrogen oxides which react in the presence of sunlight in the atmosphere
to produce ozone and other oxidants). The linear rollback and Appendix J
methods assumed an essentially linear relationship and, as a consequence,
required higher emissions reductions than a model based on a non-linear
relationship would.
The state environmental agency ran into opposition from several quarters.
Although the state transportation agency had cooperated in the development
of the state model and supplied traffic control data, when the model results
showed that large areas of the state would have tight restrictions on what
highways could be built, the agency quickly changed its position. It cri-
ticized the model and emphasized that transport, rather than in-state mobile
sources, was the problem. The state legislature joined the transportation
agency in opposing the plan based on the model. The legislators and the
governor did not speak to the technical merits of the model but opposed
any plan requiring controls as long as EPA ignored the transport problem.
Their attitude was that they would be willing to implement controls if the
other states in the region were also required to, but they would not do it
alone and clean up someone else's pollution. When concessions were made,
environmental lobbying groups concerned with public health and environmental
impacts also began to oppose the model results. Thus, the state environ-
mental agency was opposed on all sides within the state for primarily policy
reasons even though this was camouflaged by technical criticisms leveled
at the model.
When they turned to EPA for support, however,they found it lacking
there as well. The initial federal approach to the ozone problem had not
taken transport into account. Thus, while they were studying the transport
issue, the policy had not been changed. States, in order to meet the
national standards, still had to base their reduction estimates on existing
ambient air quality regardless of its jurisdictional source. Also, since
the job of evaluating state plans would be considerably harder if each
state developed its own model, all states were required to use Appendix J.
Thus, besides criticizing the theory and assumptions of the state's model,
fundamentally EPA wanted all states to use a uniform data base and method-
ology to make their review easier.
The primary problem facing the state agency, then, was to get the
federal EPA to recognize the transport problem. This fact, combined with
the opposition within the state re the model's results, led to the state's
model being abandoned. To remove one obstacle, the state confined its search
40
-------
for a new model to those acceptable to EPA. However, both Appendix J and
linear rollback were still unacceptable to the state, and the photochemical
dispersion models required money and data resources that were not available.
By this time, however, EPA had developed the model called EKMA.
The difficulties with the available methodologies for determining the
requisite reduction in the precursors to ozone had led EPA to develop this
new model. Their objective was to develop a technique for modeling the rela-
tionship between the precursors to ozone and ozone itself which would take
into account nitrogen oxides as well as hydrocarbons, would be based on the
physical and chemical nature of the pollutants, and would necessitate only
a limited amount of input data. The result of this effort was the EKMA
model, an approach based on smog chamber data and using isopleths to deter-
mine the necessary reduction. Although this model still did not treat the
transport problem, the state decided to use EKMA because it was a
methodology acceptable to EPA.
Thus, the state's environmental agency submitted another transportation
control plan to EPA, this one based on the EKMA model. However, the number
of opponents to the actions required by the plan did not appreciably change.
The plan set up a transportation plan review program whereby projects would,
in theory, have to be certified as consistent with the air quality plan
based on the model. In practice, because of the criticisms that could be
leveled at the model, the criteria for consistency became any improvement
in air quality (rather than a sufficient improvement to meet the standards).
The state transportation agency and the legislature continued to approve
projects in this manner while the environmental agency felt powerless to
oppose them because of the lack of support from EPA.
In this case, the EPA Region I office had been supporting the state's
efforts to gain recognition of the transport problem. A regional conference
of northeastern states had been held to obtain agreement that (1) transport
was a problem and (2) until models were developed as a predictive tool,
reasonable available control technology (RACT) would be required. Unfor-
tunately, the EPA headquarters' office did not endorse such a policy until
a year later and the national legislation did not recognize such interim
measures until two years later. Meanwhile, EPA had been threatening sanc-
tions against the state if it did not come up with a workable and enforcable
transportation control plan. However, since they never carried out their
threat, the state legislature felt free to ignore them. The
legislature was concerned with not doing anything "prematurely," before
other states in the region.
A cycle in the planning process developed. The state environmental
agency would fund a model application effort. Both the state transportation
agency and EPA would find fault with the model. The transportation agency
would suggest another study and lacking support for the environmental
agency's modeling efforts or a believable threat of sanctions from SPA,
the state legislature would agree co further study. Unless forced to imple-
ment a plan the governor and the legislature saw no reason for doing so.
This is the situation as it stands today. Most of the old issues remain
unresolved while new ones, also net handled well by existing models, continue
-------
to arise.
The case points out the problem in communication and divergence of
goals that exists in our federal-state system. The problem that the state
was attempting to model was different than the problem as defined by the
federal EPA. Much time and effort was spent in developing a model specific
to the state's definition of the problem. In the end they had to abandon
it because it did not meet the criteria of EPA.
EPA, on the other hand, views the process as linear, in one direction.
They develop the models, then the states use them. Because they are familiar
with the models and all states are treated the same, EPA's job is made easier
by not having to assess each state's model as well as their plan. However,
it does mean that they resist learning from the states and are slow to
perceive and consider new issues.
THE SAN FRANCISCO AIR QUALITY MAINTENANCE PLAN
The San Francisco Bay region has had a governmental agency concerned
with air pollution control since 1955. In 1961 the Association of Bay Area
Governments (ABAC) was formed to act as a coordinating institution for all
of the region's governmental programs, not just environmental ones. This
history of regional and joint perspective is perhaps longer than in any
other state and provided an important tradition of cooperation as a basis
for developing an air quality control plan.
However, the severity of the oxidant problem, the widespread dependence
on the automobile in California and the political unacceptability of trans-
portation control measures beyond automobile emissions standards tempted
the state to let EPA develop a plan since they could not develop a political-
ly acceptable one that would meet the national oxidant standard. But EPA's
plan, promulgated in 1973, contained the same politically unacceptable
strategies and was based on linear rollback, which the state felt was tech-
nically indefensible. Thus, the responsibility for developing a transporta-
tion control plan was back in the state's hands, or, for the particular
case of the San Francisco region, in the hands of the various regional
government agencies and ABAC. The issues were how to solve the modeling
problem and, more importantly, how to gain legitimacy for the plan in the
political sphere.
In addition to the EPA's transportation control plan, another incident
established the seriousness of the problem and the need for an acceptable
solution. During this same period under the New Source Review program, the
local air pollution control agency had proven its resolve to enforce its
regulations by denying a permit to Dow Chemical Company. Dow wanted to
build a large facility which would create new jobs and tax revenues but was
denied a permit because of the increase in air pollutants the facility would
cause.
Thus, the stage was set for a conflict between environmental quality
and the regional economy. The willingness of the environmental agencies to
42
-------
impose stringent control requirements put industry and other governmental
agencies on notice that the threats were real (as opposed to the case of
Connecticut's Transportation Control Plan where none of EPA's threats were
carried out). But the environmental agencies had also learned that they
would have to come up with a politically acceptable plan—that federal
actions would not bail them out.
Also during this time, the federal EPA had decided that the San
Francisco region was an appropriate area to experiment in integrated planning.
In 1975, ABAC was awarded a federal grant to compile an integrated air
quality, water quality and solid waste management plan. Thus, the presump-
tion that the type of planning was experimental plus the additional funding
helped lay the foundation for a more sophisticated and elaborate air
quality modeling exercise than was undertaken in most other states. As we
saw in the Connecticut Transportation Control Plan, the Appendix J model
and, later, the EKMA model were the models which EPA approved for use by
states for analysis of the oxidant problem; EPA suggested these models
particularly in cases where the modeling resources were limited. Connecticut
attempted to develop its own model but its methodology was not significantly
more advanced to overcome EPA's reluctance to review different or new models.
In the San Francisco case, however, funds were available and a photochemical
air quality dispersion model, called LIRAQ, was developed. It utilized
advanced methodology in analyzing both transport and chemical reactivity.
Also, it was evaluated by various technical modeling experts and EPA
approved the use of LIRAQ for the San Francisco plan.
Other factors also led to the choice and approval of this particular
photochemical air quality dispersion model. For one, ABAC was lucky; in
the San Francisco region the Lawrence Livermore Laboratory had been trying
to interest the Bay Area Air Pollution Control District (BAAPCD) in the idea
of developing an air quality model specific to the region and useable by
the agency. Their entrepreneurial efforts included raising the necessary
funds from outside sources. They had received a grant from the National
Science Foundation in 1970 and had an operational model for the San Francisco
region by 1975. The conditions of the grant included specific tasks to be
performed by the local environmental agency, the Bay Area Air Pollution
Control District. Thus, the modelers at Lawrence Livermore Labora-
tories, interested in advancing their own professional field, were able to
aid BAAPCD by recognizing early the need for a better model than linear
rollback and by gathering the necessary resources. By having to get sponsor-
ship from BAAPCD the modelers were aware from the beginning who would use
the model and what this meant for how the model should be developed. Also,
by the time the model was operational the BAAPCD staff was familiar with
how to run it on their own in-house computer.
Throughout the planning process other models were compared to the LIRAQ
model, the model developed by the Lawrence Livermore Laboratory. The long
time over which LIRAQ had been developed and verified specifically for the
San Francisco Bay region combined with ins relative sophistication when
compared with other oxidant models and its familiarity to the users helped
it remain the model of choice. These evaluations by technical experts in
the modeling field also playea a central ro .e in establishing the model's
-------
political legitimacy within the planning process as a whole. The early dis-
cussion and support of recognized experts helped to resolve technical ques-
tions before they became political questions.
Once the sponsoring government agency and the modeling experts were
identified the experience and history of how public policy was formed in the
San Francisco Bay region came into play. The failure of the previous trans-
portation control planning by EPA, an outsider, pointed out the folly of not
using the successful structure of joint effort and responsibility provided
by ABAC. This early recognition of the need to provide for political
acceptability of the technical methodology led to extensive use of joint task
forces. The elements of this structure, important to the success of the
planning effort, include:
—the formation of task forces at both the managerial and technical
staff level;
—the broad representation of interests on the task forces; and
—the formal imposition of responsibility for a successful solu-
tion on the task force rather than on one particular agency.
The primary aim of the planning process with regard to the technical
data was public acceptability. This was accomplished by creating an open
planning process which allowed all relevant interests to debate the technical
issues from the beginning. ABAC formed an Environmental Management Task
Force (EMTF) to coordinate the development of the overall plan. Each
pollution area (air, water, water supply and solid waste) had its own staff
and committees, but parallel work schedules, tasks and the technical baseline
data were rigidly enforced so that the final plan would indeed be an inte-
grated plan. Private citizen and industry participation was encouraged
throughout the process. A set time for citizen comments was reserved at
the beginning of each EMTF meeting. Also, representatives of the various
interest groups were members of the technical and advisory committees which
reported to the EMTF. Regional and local governmental agencies were also
not forgotten. The Metropolitan Transportation Committee and the state
Department of Transportation (CalTrans) sent staff representatives to the
Joint Technical Staff and Modeling subcommittees. These committees also
included modelers from the regional, local, state and federal environmental
agencies as well as private counsulting firms who were supported by private
businesses. In this way the concensus of the technical committees was broad-
ly based and subject to public scrutiny. In the end the. process provided a
defensible data base—a modeling analysis which most participants felt was
as good a job as was possible within state-of-the-art and resource con-
straints .
The formation of task forces at the managerial level had the effect of
putting the responsibility for a successful plan on the individual members
of the task force rather than on one particular agency. Thus, the members
worked reward a plan which each could sell to his cr her own agency. This,
ii effacr, aligned their self-interast vita developing a successful plan to
replace the apparent alternative, EPA's plan. Inclusion of industry and
-------
environmental representatives from the beginning had the same effect. These
interest groups could not criticize without also participating in finding
a solution. This can be contrasted to the case of the Ohio SIP where utili-
ties claimed in court that the RAM model was inadequate but, when asked for
something better, claimed proprietary interests forbade them from providing
that information. In this case, criticisms from both industry and the en-
vironmentalists were raised and considered early in the process while there
was still time to respond. This helped create confidence in the defensibility
and, hence, acceptability of the plan.
The last factor, responsibility of the task force rather than one agency,
points to a creative solution to a frequent problem. For example, in the
state of Connecticut, the Department of Transportation was able to defeat
the modeling efforts of the environmental agency because its only responsi-
bility in the transportation control planning effort was to criticize. Given
their mission of building and maintaining highways, their response was to
resist efforts which would delay or reduce these activities. In California,
however, the Department of Transportation, based on its traffic control
experience, was assigned a major role in the development of the transporta-
tion control data and membership on the technical committees. Thus, they
could not just sit back and criticize; they had to come up with solutions
acceptable to their agency as well as to the environmental agency. If they
chose not to participate, they lost the opportunity to influence the outcome
because the process was set up to function on the recommendations of the
joint committees rather than having to wait for separate action from each
agency. Also, because the question at this stage was a limited, technical
one (which of the available models is best?) and not a policy one (which
of the alternative model outcomes is best?), the criticisms had to be based
on technical factors not policy preferences. Indeed, there was resistance,
but the transportation agencies could not stop the process and did partici-
pate by providing certain baseline data.
The stated goals of this joint regional planning process were
—to improve air and water quality by the greatest possible amount,
—to reach compliance with federal and state standards at the
earliest possible date, and
— to be implementable.
Thus, the reality of the politically unacceptable transportation control
plan of EPA and the non-attainment of the oxidant standard in the Bay Area
shaped the plan developed by the task force. The task force members recog-
nized that certain limitations, including possible non-compliance with
current standards, existed.
Before ace-ally assessing alternative plans according to the above goals,
however, the objective #as to settle the technical questions. Thus, part
of the task of estaal^shing political legitimacy was to attain technical.
credibility. To a; this it vas necessary to gain a concensus arcr.g the
technical experts that the best job possible, given the state-of-tht'—art,
45
-------
was being done. Again, individuals were not allowed to criticize without
offering solutions that the task forces could accept. Also, recognized
experts from independent institutions (tied neither to government nor to
industry) were asked to pass judgment on the quality of the job being done.
These actions served to focus the early stages of the process on the
questions of a strategy for picking a model, explicit justification for the
choice, technical problems, such as input data, and what outputs would serve
the needs of the particular user. The consequence was that later in the
process the politically debated issues were the political issues, such as
land use controls, rather than the technical ones which had been debated by
the experts before the modeling results were known. Indeed, by recognizing
that issues would be politically debated the presentation of the model re-
sults was divided into long and short term policies, thereby relegating those
policies that could not be implemented to the long term and preventing them
from bringing down the whole plan.
The resulting plan eliminated completely highly unacceptable strategies
no matter how effective they might be in reaching the standards. For
example, land use controls were left to the jurisdiction of local agencies.
The so-called long term policies were policies that for some reason, in-
cluding their controversial nature, could not be implemented in the near
future. At the same time, they were not completely rejected; it was recog-
nized that long term policies needed more debate at the political level.
This approach was meant to allow for future flexibility. The agencies were
not locked into a plan without recourse if appropriate changes were needed.
The final plan, recognizing these limitations, requests a five-year
extension to 1987 for the attainment of the federal oxidant standard. Thus,
it attempts to improve air quality while taking into account the political
and economic realities. EPA, in its initial review of the plan, has in
turn questioned this political assessment by calling for further justifica-
tion for the absence of certain transportation control measures.
Thus, in the face of EPA's transportation control plan for California
the stage was set for action by the regional governmental agency. This
case parallels the case of Connecticut's transportation control plan but
ABAC succeeded in developing a model and technical basis for a plan where
Connecticut's attempt to develop a workable plan largely failed. The
elements that seem to be different for the San Francisco Bay Area agencies
include
—a history of air pollution control always ahead of and leading
the rest of the nation;
—a recognized team of experts, outside the agency, willing to
help raise funds and undertake the work to expand the state-of-
the-art; and
—a governmental structure and tradition organized toward coopera-
tion and communication as tools necessary to lend legitimacy
and credence to public policies.
46
-------
The process is not finished; EPA has not approved the plan and the air
quality standards have not all been attained. Also, the process was not
perfect. The data base is still of poor quality in several areas. The
question of whether or not a model (rather than which model) was not ad-
dressed. The transportation control measures incorporated into the plan are
limited. However, a plan tolerated by the various interests involved was
completed and no one agency or group was allowed a veto. This is in contrast
to both the situation in Connecticut with its transportation control plan
and the situation in California previously when the state was unable to
come up with a plan and EPA had promulgated a totally unacceptable plan.
OZONE STANDARD REVISION
In April 1971, EPA set national ambient air quality standards for six
pollutants. In 1976, it began the process of revising these standards.
The first such standard to be reviewed was the one for ozone. Because the
Clean Air legislation requires a standard based on health effects, EPA
publishes a criteria document which describes the available health effects
studies. By tradition, this document for the ozone standard revision also
contained the available scientific information on the control and formation
of ozone and a review of the air quality models.
The 1971 regulations establishing the ozone standard included the re-
quirement that the Appendix J model be used to show future compliance with
the standard. As we saw in the Connecticut TCP case, the states complained
about Appendix J and the EKMA model was developed in response. The revision
of the ozone standard provided the opportunity to change the regulations
regarding the use of models. The change was accomplished without controversy.
No one objected to replacing a method, which had been shown to be in error,
by the most flexible of approaches.
The process followed in revising the criteria document (both the health
effects and modeling chapters) included the compilation of studies by means
of a conference of recognized experts as well as external review of drafts
of the document. In general this review was highly controversial with regard
to the health effects data but almost non-controversial with regard to the
models. The reasons for this difference lie primarily in the fact that
the relevant decision, the level of the primary standard, was based exclu-
sively on the health effects data. The models were to be used much later,
in implementing the standard decision. The process for reviewing the
scientific information was an open one, with various experts called upon to
comment or various groups (National Academy of Sciences, American Petroleum
Institute, EPA Task Force) submitting their evaluations of the various models.
In this way, a concensus developed that, while air quality simulation
models might be better predictors in a given situation, their actual use
was severely limited by their large data and resource requirements.
The office within EPA responsible for the criteria document was the
Office of Research and Development (ORD). The office responsible for the
recommendation on a standard level and for overseeing the use of models in
che implementation of the standard decision vas :he Office of Air Quality
-------
Planning and Standards (OAQPS). These two offices had formed a joint task
force to develop a new model when the states were complaining about linear
rollback and Appendix J. The result of this cooperative effort was EKMA.
The criteria document, while reviewing the strengths and weaknesses of all
the types of models, concluded that EKMA was the model of choice when data
and resources were limited and not available for using with air quality
simulation models. OAQPS reflected this conclusion in revising the regula-
tions concerning model use. They recognized the limitations of the available
models and maintained maximum flexibility by no longer specifying a particu-
lar methodology.
OAQPS also used the recommended EKMA model when it came time for them
to prepare the cost and economic impact report. The cost and economic impact
report was another document produced for the standard revision decision.
However, umlike the criteria document, this report was an administrative
requirement of the President and not a requirement of the Clean Air Act.
Indeed, the Clean Air Act, as interpreted by EPA, prohibits the considera-
tion of anything but the health effects data. The cost data was instead
intended to be an educational tool to let the public know the costs of the
legislative requirements.
It was the job of OAQPS to produce this report and, over time, their
choice of models changed. The first version of the cost report was an
analysis of a sample of eight cities. The EKMA model, which is most suitable
to high density, urban areas, was used in the initial drafts of the report.
When the analysis was expanded to include all large cities, the input para-
meter which made the model results specific to each city, the HC/NOX ratio,
was not changed. A sensitivity analysis was done which showed that the
results were not sensitive to this parameter.
However, as the report began to be reviewed by others within and out-
side EPA, the models and their applications were changed. First, linear
rollback was introduced as another estimate in a range of estimates. This
was justified because city specific data was not available and nationwide
reduction estimates do not require a high degree of accuracy. These
estimates were considerably lower than those using the EKMA model.
Even though the cost estimates were not to be used in making the stan-
dard revision decision they became controversial as the total magnitude of
the costs became known. They ranged up to an annual cost of $12 billion.
Thus, although technical reasons were cited in the report for the choice of
models, the persons involved in the writing of the reports also cited the
pressure within EPA to make sure the costs were as low as possible. The
initial choice of EKMA seems to have been one of "what model am I familiar
with" and "what model is the latest new technique" rather than a complete
search for what was most appropriate. Indeed, the operations office (OAQPS)
did not consult with the developers of the model in the research office
about tneir choice of models. As pressure mounted to keep the cost estimates
lew, linear rollback was introduced. In the final version and in presenta-
tions to the EPA Administrator only the linear rollback figures were cited,
because it was felt that listing different ranges and estimates from the
/arious models was too confusing. While the choice of model may have been
48
-------
correct for the problem as defined by EPA, the choice was simplified by
political considerations.
In this case, we once again see the attempt to conduct an open decision
process. The decision process in this case concerned what kind of regulation
would be promulgated concerning the use of models by states in planning for
attainment of the oxidant standard. A joint task force was successful in
identifying the problem as one of data and resource limitations. As long
as the discussion remained on the scientific or technical level of evaluating
model methodologies within the structure of a joint task force, the coopera-
tive effort was successful. The task force participants had a common pro-
fessional background (in meteorology) and a circumscribed problem area.
After the review panels for the criteria document were concluded, EPA
returned to its separate organizational method of operation. ORD completed
the criteria document and OAQPS began work on its recommendation for a
standard level and its cost report. Without the structure of a joint task
force the use of the models (in estimating costs) was left entirely to OAQPS
without any review by ORD, the developer of the other models. Each organiza-
tion had returned to their own separate interests—ORD in advancing the
scientific foundations for the tecnical analysis and OAQPS in implementing
the regulations. The result was an emphasis on the details of the cost
assumptions by the economist in charge of the cost report and relatively
little evaluation of which air quality model might be the most appropriate.
Even though the process was a relatively open one with experts able to
recognize their influence on the final decision, the non-controversial
nature of the decision to change the ozone modeling requirements was also
due to
—the fact that the models did not provide information germane to
the immediate decision but would only be relevant in later de-
cisions,
—other information (on health effects) was highly controversial,
and
—the proposed regulation on model use was much more flexible than
the one it replaced.
NEW SOURCE PERFORMANCE STANDARDS
The technology-forcing strategy of the Clean Air Act occasions quite a
bit of debate and controversy within EPA and often results in varying inter-
pretations of the law. The case study of the new source performance stan-
dards for coal-fired power plants describes the process followed by EPA to
develop a regulation for new coal-fired power plants based on controversial
technology and for which EPA spent an unusual amount of resources for tech-
nical analysis.
In 1971, the original new source performance standard (NS?S) for
49
-------
coal-fired power plants had been set in the form of a ceiling, an emissions
limitation of 1.2 Ibs. of SC>2 per million Btu of heat input. No percentage
reduction of potential emissions was specified. As required by the 1970
Clean Air Act, this standard was based on an engineering judgment of what
constituted the best technological control system and no analysis of cost
or emissions impacts was done. The consequence of setting the standard
in this way was that the utility companies shifted to the use of low sulfur
coal whenever the costs of transporting the coal were less than the cost
to install the available control technology, flue gas desulfurization systems
also called wet scrubbers. These wet scrubber systems were considered
expensive and unreliable and also produced considerable wet sludge as a
waste by-product. The alternative, low sulfur coal, is mainly found in the
western part of the United States. However, the utilities, especially those
in the Middle West, found it cheaper to transport this coal from the West
than to buy and operate scrubbers. The result was a decline in demand for
the higher sulfur coal of Appalachia and the Middle West and a rise in unem-
ployment in these coal mining regions.
By 1976 the experience with the few scrubbers that had been installed
had led environmentalists to conclude that the technology was capable of 90%
removal of SC>2, at least when burning low sulfur coals on which scrubbers
were more reliable. This led the Sierra Club to petition EPA to revise
the NSPS on the grounds that scrubbers were now a proven technology and,
therefore, the performance standard should be upgraded to reflect this new
technology. In January 1977, EPA announced it would review the 1971 stan-
dard and studies were begun.
However, the environmentalists did not stop there. During this time,
amendments to the Clean Air Act and a National Energy Plan were being debated
in Congress. Thus, an unusual political alliance developed. Both environ-
mental and high sulfur coal interests began to push for an amendment
(Section 111) which would require a NSPS to include a percentage reduction
figure as well as an emissions ceiling. A percentage reduction figure would
not permit low sulfur coal to be burned without some controls and would,
thus, eliminate some of its advantages and encourage more use of high sulfur
coal. For the environmentalists, a uniform standard of 90% reduction of
potential S02 emissions, regardless of the sulfur content of the coal,
would reflect the best technology available and would require emissions re-
ductions even in the West. They saw this as an aid in preventing visibility
degradation in the West. The utilities were opposed to any amendment,
however, because of the costs and sludge disposal problems as well as what
they still considered to be the unreliability of scrubber systems. The new
Carter Administration and the Department of Energy (DOE) saw the use of
scrubbers as a way to increase the use of coal, the linchpin of their pro-
posed National Energy Plan, in an environmentally sound way.
The House passed the amended Section 111, but the Senate did not.
Senator Muskie considered it economic protectionist legislation rather than
clean air legislation. The cotnororjLse enacted into law retained the amended
Section 111, with its requirement of "the degree of emissions limitation and
the percentage reduction achievaoic through trie application of the best
technological [control] systam . , . taking into consideration the cost . . .
30
-------
and any non-air quality health and environmental impact and energy require-
ments," but included confusing language in the accompanying Conference
Report. The utilities were successful in lobbying for a paragraph in the
Conference Report which said that EPA could "set a range of pollution reduc-
tion that reflects varying fuel characteristics," while the environmentalists
and high sulfur coal interests obtained a warning that "any departure from
the uniform national percentage reduction requirement . . . must be accom-
panied by a finding that such a departure does not undermine the basic
purposes . . . of the Act, such as maximizing the use of locally available
fuels." This uncertainty about congressional intent was later to plague
EPA in its review of the standard, but the actual language of the Act as
well as the initial support of the White House and DOE led EPA to make the
presumption that the same, or uniform, percentage reduction would be required
of low as well as high sulfur coal unless costs or other impacts were un-
reasonable. This presumption of the EPA staff was fortified by the fact
that the EPA Assistant Administrator who would be making the new standard
recommendation had previously headed the air quality project of the Natural
Resources Defense Council, a group active in lobbying for the Section 111
amendment. Thus, the question posed by the EPA staff was what percentage of
SC>2 could reasonably be reduced from high sulfur coal.
The initial technical analyses examined the associated cost and energy
impacts of alternative uniform percentage reduction figures and emission
ceilings. It actually began as a small effort based on the type of economic
studies EPA had used in the past—cost and energy impacts and emissions
based on a single model plant plus extrapolation to a nationwide basis. It
was initiated by the program office responsible for developing the regula-
tion, the Office of Air Quality Planning and Standards (OAQPS). The analysis
showed small differences in costs with relatively large differences in
emissions levels for various alternatives and supported the initial EPA
assumption that a uniform percentage reduction or full scrubbing (scrubbers
on all plants regardless of what type of sulfur-content-in-coal was used)
would be required.
In December 1977, EPA went public with the results of its technical
analysis and indicated that they were considering a stringent standard of
90% reduction. This draft NSPS was presented by the staff of OAQPS at a
public hearing. Their strategy was to start with the toughest defensible
standard with the realization that a more moderate final standard would
probably be negotiated. Indeed, the draft standard did elicit debate from
groups both within and without EPA. As these other groups joined in the
decision process, both the preconceptions and the technical analysis radical-
ly changed.
Within EPA, the Office of Planning and Evaluation (OPE) was concerned
with cost effectiveness of the standard (whether different controls on
different coals might be justified considering the cost of controls in
light 3f resultant reduction in emissions) while it fait OAQPS was only
looking at the cost affordability (whether the industry could afford the
ccst of controls). The Office of Research and Development (ORD) was also
concerned with the 90% reduction figure but it emphasized that 90% right
not be technologically feasible.
-------
Outside EPA, opposition seems to have been stimulated by the public
release of the draft NSPS. The utility industry began to lobby actively
in Congress, the White House and the Department of Energy. With a concrete
proposal they were able to tie specific costs of the full scrubbing alterna-
tive to constituents (consumers of electricity) who would have to bear them.
DOE responded to this pressure; the Deputy Secretary and the Assistant
Secretary for the Environment were both to publicly state that full scrubbing
may be unreliable and unacceptable. The opponents to EPA's draft NSPS were
proposing a variable rather than uniform standard, a standard that would re-
flect varying fuel characteristics and would not require scrubbing on the
lowest sulfur coals.
As the debate intensified, OAQPS realized it would have to conduct
further technical analyses in order to resolve intra-agency disputes.
OPE objected to the model* OAQPS was using and had funded a different model-
ing analysis. In this case OAQPS and OPE reached an agreement to form a
joint task force for developing the technical analysis. The joint staff
agreed to use the OPE model because it was more detailed, and, being in a
later stage of development, seemed more "consistent and persuasive." Also,
it was a model that DOE had previously sponsored. Thus, the joint task force
was expanded to include DOE staff members. With the establishment of this
task force both the head of EPA and of DOE directed their staffs to secure
agreement on the modeling assumptions and modeling techniques, thus commit-
ting their agencies to resolving their differences on technical, if not
policy, issues. The major impetus for the commitment of these resources as
well as the time of the top management of EPA to this decision was the recog-
nition of the importance of this decision for EPA in the political arena.
The debates on the Clean Air Act Amendments, the push for regulatory reforms
from within the administration at this time, and their recent controversial
decision on the ozone standard underscored the political vulnerability of
EPA's regulations. Thus, EPA was determined to resolve as many of the tech-
nical issues as possible so that the debate would focus on the policy issues
—whether costs or emissions are more important—rather than the technical
issues—what will be the costs or the emissions.
This new modeling analysis was undertaken by DOE and EPA, but all
interested parties were invited to debate the assumptions, suggest alterna-
tives and inspect the results. The new analysis eventually convinced the EPA
decision makers that their original presumption of a uniform standard was
wrong if all environmental, economic, energy and political consequences
were weighed. The new model found an overall smaller difference in costs
between the full and variable scrubbing options than the original analysis.
It also produced a counterintuitive result—that stricter emissions limita-
tions would lead to more, rather than less, emissions. This result came from
the assumption that an increase in construction costs (of scrubbers) would
cause the owner of an old plant to keep it running longer (with higher
re; siiaissions, energy and economic impact model, not an air quality
aode1.
52
-------
emissions) rather than building a new source subject to the higher costs of
control.*
The final standard was a variable standard allowing from 70 to 90 per-
cent reduction of S02. The actual figures were chosen to allow for a tech-
nology called dry scrubbing. The dry scrubbing option actually became the
focus of attention relatively late in the process. It is a newer technology
so that ORD, while having suggested that it could save utilities money, was
not willing to attest to its reliability. But more importantly, as long
as the Assistant Administrator for air programs, who was pushing the full
scrubbing option, felt that the EPA Administrator still favored a tight rule
and the legislative presumption of full scrubbing, he preferred not to sug-
gest any variable scrubbing solution. Also, the industry was interested in
a relaxed standard requiring no scrubbing for some coals so that a new
option based on some scrubbing, even if less costly, did not interest them
for a mandatory standard. However, when the technical analysis showed the
large cost savings from a variable standard without large increases in
emissions, the discussion turned to an analysis which included the possibili-
ty of allowing the dry scrubbing control option for low sulfur coals, which
in turn established the 70% lower limit for the standard.
The strategy of pushing for agreement on technical issues also turned
out to be of importance in concluding the decision process with, at least,
a reduced level of controversy as compared with other recent environmental
and health regulations. Even though the White House and several Senators
continued to express their keen interest in the decision and the technical
information, the briefings by EPA focused on the policy issues and were struc-
tured so that the President could reserve the decision for himself. President
Carter left the decision to EPA after the opposition views had been aired.
EPA was sued by both the environmentalists and the industry over this
decision. The environmentalists contended that the last few weeks of hectic
meetings and model runs, conducted to meet a court ordered deadline, that
produced the dry scrubbing option amounted to "ex parte," or off-the-record,
communications with Congress and industry officials and that the decision was
made on the basis of new information, not part of the public record. They
still wanted a uniform scrubbing standard and were never reconciled to the
counterintuitive results of the model. The industry suit contended that
several provisions of the standard were too restrictive and were impossible
to meet. But EPA officials were generally pleased with the outcome. They
felt that large expenditures of time and resources had resulted in a tech-
nically defensible standard that did indeed balance environmental with energy
and other considerations. They were not surprised by the law suits given
the adversary tradition that has been set up by the clean air legislation
and its various interest groups.
*
The environmentalists objected to this assumption. They maintained
that political, financial and institutional constraints would alter this
cost-minimizing benavior.
-------
That this effort to produce a technical analysis was in large part suc-
cessful can be attributed to several factors, including:
—a model of sufficient complexity for the situation was available
and did not have to be developed,
—the model was flexible (some thirty-three alternatives were
eventually analyzed) and produced the required output format,
—the model itself, its assumptions and how it worked were
understandable to the decision makers, and
—EPA committed the level of resources and time needed to do
a full analysis.
In the end, EPA was able to produce agreed upon numbers and a coopera-
tive interagency process that focused on the policy issues and left the
ultimate decision in the hands of EPA. Other factors than the model, of
course, helped produce this outcome. They included:
—top EPA management involved early and actively in the process,
—interagency groups that attracted staff with similar perspectives
and backgrounds,
—EPA's solicitation of public participation from the beginning of
the process and their willingness to change their modeling analy-
sis in response to public criticisms, and
—the willingness of the utility industry, as a regulated monopoly,
to divulge information which other industries might regard as
trade secrets.
Thus, the development of this regulation contained all the potentially con-
troversial elements—opposing political interest groups, technical uncer-
tainties and organizational conflicts—of the other case studies. However,
the fortuitous existence of a relatively complex model, closely approximating
the situation, as well as the conscious management of the process so as to
defuse the opposition, at least on technical grouds, helped it to work well.
Clearly, in almost any EPA decision and especially in an important one like
this not every element of controversy can be avoided. In this case EPA
reduced the controversy even if they could not avoid it altogether.
The major technical qeustion that was not addressed was that of the
benefits of pollution reduction. A first step in such an analysis would be
an air quality model to relate emissions to ambient air quality. EPA has
not in general tried to conduct benefits analysis for its decisions in the
past. New interpretations and new legislative language may change this in
trie ifature and air quality models may be called upon to fit into and be com-
pacible with the larger analysis. Thus, this case study has been included
ir. chis report both to indicate how technical analysis can be used in the
EPA decision process in general and to show what expanded uses might be
54
-------
made of air quality models if they were available in the correct format in
the future.
SUMMARY
The case studies conducted for this report cover a variety of programs
(standard setting, implementation planning and implementation for both new
and existing sources) that EPA has jurisdiction over and which require some
form of technical or modeling analysis. Although the cases vary by geo-
graphical and political extent, these attempts to apply models and use the
results in solving the political as well as the technical problem reveal some
common threads that delay and confuse the decision process. The difficul-
ties arise from the model formulations but also from the process as laid
down in the legislation and as managed by EPA, its regional offices and the
states. Section II of this report outlined four possible types of uses of
models: for problem solving, for conceptualization, applying basic research,
and for political advantage or self-interest. We have seen examples of all
of these in the case studies, and they provide the basis for identifying
common difficulties linking the case studies.
Uses for Problem Solving
Air quality models provide the data which applicants for a permit to
build a new project can and do use to design their project so that it will
meet environmental constraints. Environmental agencies use such data to
plan for new pollution sources and for assessing compliance with standards
of existing sources. However, the models are far from perfect. The major
obstacles to the use of air quality models for problem solving will be
discussed in the next section of this report but, briefly, they include
technical problems such as:
—a lack of theory and validation for complex topographic situa-
tions ,
—a lack of input data (monitoring), and
—time and resource constraints.
Furthermore, there are managerial problems, such as:
—a lack of federal policy directives,
— the mismanagement of model critiques, and
—conflicting or narrow institutional and organizational perspec-
tives ,
which stem from not so much the technical problems of modeling but from the
legislative framework and the adversary structure for decision making that it
creates.
55
-------
Political Uses of Models
One example of the interaction of the political and technical frameworks
is the use of models or technical information by the interest groups. The
legislation has set up an adversary system of public hearings, administrative
hearings and court appeals. Such a system encourages the selective use of
technical information which in turn is bolstered by the uncertainties and
numerous assumptions of the models. By far the greatest use of the models
by environmental and industry groups in these cases was in opposing or defend-
ing the projects, often for non-environmental reasons, rather than in seeking
to resolve or even define the problem as a broad one of how to balance the
conflicting interests. Essentially the adversary system of opposing sides
sets up a narrow, limited sphere of concern—whether the project meets certain
standards and not whether this project at this time and place meets the broad
public interest.
However, in some cases the environmental agencies were able to use the
advocacy system of public hearings to advantage. Several times key issues
regarding the accuracy and appropriateness of the models were initially
brought up through the public hearing process. In a society of limited
resources this may be a fairly good way of making sure such issues are raised.
EPA or the state agencies cannot have the resources or the time to investi-
gate every aspect. The San Francisco AQMP shows how this system can be
used to advantage. A problem with this approach is that it tends to occur
late in the process so that the EPA modelers are put in the position of
justifying their conclusions. Although, as we saw in the NSPS case, when
the issues are raised in the early stages of the process, it is possible for
EPA to respond and correct the analysis. This problem may be exacerbated if
the distribution of resources or availability of information is unequal.
In the Pittston case the citizen advocates were at a disadvantage on both
counts. In the MATEP case the citizen groups had better access to the
information. In general, both the industry groups and the environmental groups
were predictable in their criticisms and responses, but the environmentalists,
in some cases, lacked the resources to play a significant role while there
was always clear industry opposition to controls.
Interest groups are not the only outside advocates in these case
studies. The legislative (Congress) and the executive (governors and the
White House) branches were also called upon to lend their influence. For
instance, in the Westvaco case the company pleaded their case to several
congressmen who in turn demanded technical justifications from EPA. Subse-
quent congressional inquiries led EPA to conduct its own technical analysis.
In the case of the Ohio SIP, the statement by the Council on Wage and Price
Stability foreclosed several options that EPA had been considering without
referencing specific technical information. But it was clear to EPA that
there was no appeal and another solution would have to be found. In the
Connecticut TCP case the Governor and legislature, agreed that when and if
other states established strict control programs then and only then would
Connecticut cic 30. This type of intervention by other governmental interests
was possible -sot^. because air quality modeling is not an exact science and
because ^eatrally the
-------
San Francisco AWMP and the NSPS cases being notable exceptions).
The state-of-the-art of air quality modeling and the management of
the decision process also result in "self interest" uses by the decision
maker. For instance, the lack of input data and the emphasis on conserva-
tive "worst case" analysis leads to the presentation of just one point
estimate. No sensitivity analysis or allowance for uncertainty in the data
or the methodology is presented to the decision maker. This is encouraged
by the decision maker, who is usually not a scientist and therefore has
difficulty assessing variable results. When only one number, which is below
the standard level, is presented, the decision is obvious, and the decision
maker does not have to question the accuracy or indeed the applicability of
the model. These questions did come out in the MATEP case because no
standard had been set. But even here, the short-term NOn effects were not
seen as an issue until public comments raised them.
This points out a major influence on how models are chosen and used.
Choices are made based on experience, but this experience is not that of
validation and documentation. Instead, the experience is what was used
last time. For the decision maker this means point estimates of concentra-
tion levels have become familiar. Whether the prevention of significant
deterioration provisions are substantively different and demand a different
approach or whether the problem should be expanded to include possible
future growth on employment impacts is not asked. If no standard exists,
the tendency is to avoid the issue. Indeed, even in the more successful
example of the NSPS case the analysis began with the familiar but was
expanded as the importance of and widespread interest in the regulation was
recognized.
The modeler also has "self interest" uses for the models. For the
modeler it is easier to apply a model previously used than to learn how to
apply a new one. Also, if the decision maker is satisfied with point esti-
mates, then the work of producing sensitivity analyses and educating the
decision maker as to what they mean is not worth the effort. Connecticut
also learned it is easier to use a model that is familiar to the federal
EPA than to fight the modeling issue as well as the transport issue with
both EPA and its own state agencies.
Uses for Conceptualization and the Use of Basic Research
As for the other two uses—conceptualization and the use of basic
research—fewer examples were found. The accumulation of experience in
trying to use Appendix J and linear rollback led to a search for a better
method by both San Francisco and Connecticut. But policies on transport, on
monitoring for permit applications or on the prevention of significant de-
terioration provision have not been developed yet, although models do exist
which could provide useful information for policy purposes in these areas.
Also, the available models are used but there is no strategy for how they
ought to be used.
Basic research to develop better, more accurate models has concentrated
-------
on the photochemical dispersion models. But they require data and computer
time beyond the resources of the state agencies, and therefore, are not
often used. It was because this research was not responsive to the decision
makers' needs for more timely and less expensive models that the EKMA. model
was developed. But generally the problems of data and computer resources
remain.
With this background on how the air quality models were used during the
decision process, I turn now to the difficulties (technical, managerial
and political) in trying to use them and how the difficulties helped shape
some of the political and "self interest" uses and hindered the use of the
models for problem solving and conceptualization.
58
-------
SECTION IV
CONCLUSIONS
The case studies describe what air quality models were applied in the
given situation and what technical, managerial, and political problems were
encountered in the attempted applications. Except for the bare skeleton
of identifying the project or industry and sites affected and ensuring com-
pliance with the public participation provisions of the law, these cases
reveal no standard or organized procedures to affect decisions in the envi-
ronmental area where air quality models are involved. Neither the responsi-
bilities of the various governmental agencies nor realistic expectations of
the interested parties are established at the outset, and they often are in
flux throughout the decision making process. The details of the case studies
show that
—the legislative and political framework,
—the limitations of the technical capability of the models and
the input data, and
—the management of the generation and use of the technical infor-
mation within this framework
together cause a narrow and sometimes contradictory definition of both the
problem to be resolved and the solution proposed.
The legislation and EPA's interpretation of it have established air
quality models as the tool of policy analysis. However, the models relate
only emissions to ambient air quality. The models can be used to redesign
a proposed project, assess compliance of existing sources and answer planning
questions about future growth. They do not by themselves evaluate impacts
other than air quality impacts or problems not directly related to existing
standards. The reliance on the models tends to narrow the definition of
the problem and the perspective of the decision maker.
The legislation also specifies a structure for the decision process.
It requires EPA to conduce an "open process" of public hearings. They are
meant to ensure that the decision maker has obtained all available technical
information. In practice, the public participation process is modeled after
the advocacy form of court proceedings, where opponents are identified and
information (testimony.) is strictly limited to what is in the interest of
the speaker as an advocate for one side of the issue. The problem for
environmental decision makers arises in the limited purpose of the public
59
-------
hearings and in EPA's management of the process. Generally, their purpose
is to establish a record for the technical information without the further
step of utilizing the technical experts to assess that information for its
relevance to the decision. In this way a fundamentally political decision
is structured as a technical decision. But the nature of the uncertainties
of the models and the dependence on technical analysis that is narrower than
the problem to be solved leads to delay and, frequently, no decision while
the essentially uncertain analysis is debated as if the uncertainties could
be resolved. The more successful cases show how structuring the problem
to separate the technical facts and uncertainties from the political forum
and using the technical analysis as an input, rather than the solution, to
the policy problem helps alleviate (but not necessarily eliminate) some of
the controversy and delay of the decision process.
The process as conducted in these case studies also is, by and large,
a one way process. The federal EPA has the resources to develop the models,
which are then made available to the states. Thus, seldom does a state
search for other options. Or, when it does, much effort has to be made to
convince EPA that the state developed model is acceptable. Moreover, EPA
is organizationally structured so that the research office (ORD) is separate
from the office (OAQPS) which helps the states in their model applications.
This separation of the model development and use activities both within the
EPA and between it and the states hinders communication of the states' ex-
periences. Officially the federal EPA provides technical assistance to the
states. Only a few administrators have realized that the states' experiences
may also be useful to the federal EPA. Thus, the process itself and the
experience of the states in using the models has had little impact within
EPA on broader issues of strategy and policy.
The overall result is a separation of modeler and decision maker during
the time the technical information is generated. The most obvious symptom
of the problems such a separation produces is seen in the contrast between
the questions posed and answered by the modeler and the policy questions
encountered in the given situation. As Table 3 shows, in a majority of
cases the appropriate question from the point of view of policymaking was
not asked or answered by the modelers. The process is designed so as to
require air quality models to assess a health-based standard. The conse-
quence of this strategy has been an application of the models, chosen for
their familiarity with relatively few resources expended for validation and
documentation, and model results presented as point estimates based on
"worst case" assumptions with little focus on educating the decision maker
with regard to the remaining technical uncertainties and impacts not as-
sessed by the models.
As discussed above, the reasons for this dichotomy between what the
modeler saw as the problem to be solved and the question that the decision
maker needed answered are based in the structure and management of the deci-
sion process as well as the technical capabilities of the models. The
analysis of the case studies in this se-txjn of Che report shows in detail
how these technical, political and onager I si factors produced the problems
enco'^r.tared in the case studies, ~cr purposes of organization and clarity
four areas have been identified for discussion:
60
-------
H
cn
O*
H
H
c/l
Cd
33
O
c
0
•H
4-1
CO
cu
3
Cf
^
O
•H
rH
O
I-i
CU
CU
T3
2J
^%
n
13
CU
)^
cu
3
co
B
B
O
•H
4-1
to
cu
3
o-
i i
MH ex
CO CO Cfl
TJ ,_|
CO C O <^
g cfl O CO
CO 4J rH
1-1 o cu
0 'g 0 0
rl 0 -H g
CX C rH
O O MH
rH 0 CX O
O CU
rl CU "O
4-J « rl Cfl C"
C rH CU CU g
O CO JS 4J CU
O 4J 4-J CO rH
fi C J3
C CU CU -H O
O g rl rl
ft C Cfl rl CX
4-1 O O
3 l-i •- CU
rH -H CO O JS
0 fi CO
a cu i-i c o
00 O W
rl M O -H
•H CU I-i 4-1 CU
Cfl JS CX -H 4-1
4J T3 cfl
O O >-, TJ -H
•a oo co i-i
4J I-l CX
3 0 cu c o
O CU C -H rl
3- 14-1 a) ^ CX
0)
B
O
25
c^-
cu
J3
-O
rl
CO
•a
c
CO
4-1
CO
CU
JS
4-1
TJ
rH
3
O
JS
CO
4-J
CO
3
C
O
•H
4-1
cfl
rl
4-1
fi
CU
O
fi e-
0 rl
CJ ft
CO
00
G 4J
•H fi
4-1 CU
rH -H
3 J3
CO g
5 -
CU
4-1
B -H
CO 33
U co
3 ~i
i
f
OJ ' G
cn c
?3 ' *«) — i '
^j .», «_j ' , — •
< - 5
JJ -r ^ t
JJ
•a
; -5 •- -jj ' -»
! - ~ > •— «•
- J; '•£ ;H
c***
^Q
cu
3
CO
Cfl
•H
01
•°
4-1
'3
O
4-1
T3
cu
T3
cu
cu
C
cu
u
cfl
33
B
O
•H
4-J
cfl
4-1
•H
g c--.
•H T3
rH rl
Cfl
CO "O
C C
O cO
•H 4-1
co ca
ca
•H CU
g JS
CU 4J
4-J 4-1
CO CU
XI CU
3 g
y*^
B
O
•H
4-1
CO
s
cr
^
CJ
•H
rH
O
CX
31
Cfl
i
tfl
> — f
j
to
13
3
1 3
'J
fl
-§
ca
rl
O
4-1
•H
C
o
g
CU
rl
cfl
• A
*Td
CU
4J
O
CU
ex
cu
cu «•••
rl g
CO CU
CO 4-1
c
o 3
•H O
4-J fi
CO CO
rH
O O
•H 4-1
>
13
4J CU
Cfl 4J
JS ft
5 3)
ca
M
O
4-1
•H
C
O
g
00
c
•H
4J
co
•H
X
CU
O
T3
co
fi
0
ft
4-1
CO
rH
O
•H
^
4-1 "3
03 O
JS JS
3 co
o
JS
4-1
^
o
rl
00
I-l
O
U-l
fi
cfl
rH
|X|
CO
B
O
ft
4-1
•H
*T3
fi
O
CJ
00
fi
•H
4-1
co
•H
X
cu
u
o
>4H
B
tfl
rH
CU
CO
fi
O
•^
4J
CO (3
4J rl
4-1 3
• n
ca
Cfl C"
CU CO
rl B
cfl O
•H
T3 4-1
B cfl
•H rH
3 0
C -H
3 >
0
•a 3
o
4-J (J
o co
cfl
CX O
g *->
•H
T3
co cu
4J 4-1
C -H
CO CO
rH
cx ca
rl
cu o
4-1 ~t
C
0 0
-a g
3 cu
0 rl
33 CO
CO
rl
O
4-1
•H
C
g
00
c
•H
4-1
CO
•H
X
cu
o
T3
CO
B
O
•H
4-1
CO
rH
O
•H
^
4J "3
cfl O
3 to
CU OC
co y i
3 as • aj
00
B
•H
T3
3
rH
CJ
6
•H
•*^s
4-1
ca
cu
J2
co «•-•
•H x^\
CO
00 r*
CU -H
4-1 4J
0) Cfl
S S
ca •
co ca
4-J
rH rH
CU 3
"O CO
i *"
O CU
•H TJ
3 S
CJ
H
u
3
O
T-l
cj js : *-> 4-1
*fl Cv l"i ' V3 -^
1
CU
rH
CX
•H
CU
B
cfl
O
4-1
cfl
j:
4-1
^
00
cu
4-J
cfl
rl
4-1
CO
4-1
CO
cu
in c*-
•H -a
cu
4-J 4-J
cfl fi
3 S
/^-^
C
0
•H
4-1
CO
cu
cr
>^
u
•H
rH
O
CX
ca
CO
1
CO
cn
„
S
o-
"^
14-1
CU O
4-J CO
4-1
00 CO
C 0
•H O
CX
O
J2 ft
00 4J
co B ffl
•H ft fi
c fi
4J fi CU
CO Cfl 4-1
JS rH rH
3 CX co
s**.
c
0
•H
4-1
CO
cu
cr
^
o
•H
rH
o
a
CO
cfl
ii
cfl
cn
•a
p ^
CO | -3
4-1
Cfl
JS
3
• *.
4-1
CJ
CO
P^
g
•H
ca
B
0
•H C"
co co
CO rH
ft O
g r.
CU 4-1
fi
•0 0
G CJ
cfl
4H
CO O
4-1
CO CO
0 4-1
O -H
*4H
ai cu
>-> fi
to cu
fi
4-1
Cfl CU
»c ^
3 cfl
c*-«
4-1
CJ
Cfl
CX
g
•H
CO
fi
O
•H
ca
ca
•H
g
cu
•a
c
ca
CO
4-J
CO
o
CJ
cu
I-l
cfl
4-J
Cfl
S
•j I 3 '
S3 C/3
> .-03! i SJ U. TJ
^j w -4*" ' c ' G i a ! z.o
V)
33 -i -i , a ; a ; c a,
3 ' S v3 ' "5 ~ ' v: C ! z
-------
—technical and political constraints,
—unresolved policy issues,
—management by EPA of the decision process, and
—conflicting institutional and organizational interests.
It will be emphasized, however, that the problem is a complex one with the
factors interacting to produce the confusion, delay and controversy found
in the case studies. No one factor stands out as the dominant cause or easy
target for reform. Solutions appear to be as complex as the problem.
TECHNICAL AND POLITICAL CONSTRAINTS
The first part of this discussion describes technical problems which
face the modeler and shape what questions he or she can answer. It is
pointed out, however, that the technical limitations do not necessarily deter-
mine what questions the modeler actually chooses to answer. Likewise, a
decision maker faces limitations in legal or administrative authority which
determine what problems appear on the political agenda but, as the cases
show, are not the sole determinants of what solutions are applied. Instead,
other factors (managerial and organizational) that are under the control of
the decision maker and the modeler significantly affect the decision process.
They are discussed separately and will be the focus of the recommendations in
the final section of this report.
Technical Problems
The major technical problems encountered in the case studies are
—state-of-the-art,
—lack of input data,
—lack of sensitivity analysis, and
—form of the output of the model.
Each case study gives specific details on the models and what particular
assumptions were criticized so that this summary report will discuss some
problems common to the cases and draw conclusions from the successful cases.
State-of-the-Art—
A major problem in all the case studies is a lack of theory for handling
different topographic situations (complex terrain, coastal areas, fumigation,
downwash), choosing values for specific meteorological parameters (wind speed,
sector averaging) and reactive pollutants (their formation and transport in
the ambient air). Lacking a definitive theory the modeler is forced to
make assumptions and EPA has published guidelines for choosing the assump-
tions and input parameters for their models. However, there is great lati-
tude in the choices and, as ve saw in the Massachusetts' Sulfur Regulations
and Ohio SIP cases, even the same uodel in the hands of different modelers
62
-------
will produce different results.
Compounding this problem is the lack of verification studies. Many more
resources have been spent developing new models than in verifying old ones.
The case studies point out that rarely is there sufficient data to even apply
the model much less to verify it with an independent set of data. Usually
the data is supplied by the applicant, hardly an independent data source.
The Anaconda and the San Francisco AQMP cases are the notable exceptions.
In these cases much larger amounts of time and resources were spent by EPA
and the San Francisco governmental agency to verify the models for the topo-
graphic and meteorological characteristics of the particular situation. In
general, however, the users (EPA or the state agency who is making the
decision) do not have available such resoruces nor is it seen as their respon-
sibility. Their job is to assess the model results not develop the model
itself.
Lack of Input Data—
Another critical problem in almost every case was the lack of input
data for the meteorological and topographic parameters needed to apply and
validate the model for the specific site. As the Anaconda, San Francisco
AQMP, and NSPS cases show, more time and resources help considerably in
solving this problem. However, time and resources may not be the binding
constraints. Table 1 shows the time span the decisions covered. Only three
cases took less than four years and all local or regional cases (except
Pitts ton) used on-site monitoring for part of the necessary data. The major
problem was that the monitoring was not, in general, of the type or location
needed to answer the questions of the decision maker.
EPA's policy, in light of the state-of-the-art and uncertainties of the
modeling analysis and the fact that the legislation calls for no violations
of a health-based standard, is to require a "worst case" analysis rather
than a statistically likely prediction. Thus, accurate monitoring and input
data is essential. However, when the permit applicants must spend the re-
sources to supply this data, they will do so only when it is in their inter-
est. In the MATEP case the applicant supplemented the available data with
new monitoring sites because EPA threatened to base their decision on data
from a "hot" spot, a site not representative of the sites impacted by the
proposed power plant. Since additional data was neither advantageous nor
required in the Pittston, Westvaco and Ohio SIP cases, no new, more appro-
priate input data was collected and the problem remained. In the Anaconda
case, EPA had to collect the necessary data. In the other cases, however,
EPA did not monitor and, instead, requested better monitoring sites from the
applicants, with a singular lack of response. That this was a major problem
can be seen by the number of decisions which were contingent on further
monitoring taking place (MATEP, Westvaco, Massachusetts' Sulfur Regulations,
and Ohio's SIP).
Lack of Sensitivity Analysis—
Another problem, the lack of sensitivity analysis, exacerbates che input
data problem. In only a few cases did the modelers carry out a sensitivity
63
-------
analysis in order to determine just how accurate their input data needed
to be or which assumptions significantly influenced the outcome. For example,
in applying the EKMA model in the ozone standard revision case sensitivity
analysis showed that a city-specific ratio, one of the input data items lack-
ing for many cities, did not have a significant effect on the decision on
how much hydrocarbons should be controlled nationwide. Only if the decision
maker was interested in a particular city, rather than an average for the
nation as a whole, was such a high degree of accuracy required. Thus, the
accuracy available from EKMA, as shown by the sensitivity analysis, was
sufficient since the particular decision concerned a national standard and
the standard was based on the health data rather than the cost data. Also,
in reaching agreement on the technical analysis in the NSPS case the Depart-
ment of Energy insisted on sensitivity analyses and dropped some of their
opposition based on the results.
However, in most of the case studies such an analysis was not done.
Instead, the modelers used the available input data but did not test what
the inaccuracies in the data implied. The emphasis on conservative "worst
case" analysis is one cause of this problem. The models are used to produce
one point estimate. The decision maker can then ask: is that estimate above
or below the standard. This appears to have taken into account the uncer-
tainties of the data. However, the variabilities in the models are quite
large so all that it really does is hide the inaccuracies. The Pittston
case is an example of this problem. The oil refinery project was designed
so that modeling analysis showed an impact of 4.9 ju/m^ when the standard
was 5.0 ja/m3 . As different models were applied, the refinery characteris-
tics were redesigned so that the impact always remained just below 5.0 ,u/m3 .
However, as one public interest group representative pointed out, if just
one parameter (the wind speed) was not accurate the results could have
reasonably been as high as 10 p/m^ . Relying on one case, said by the com-
pany to be the worst case but clearly not considered by everyone to be the
worst case, without doing sensitivity analysis to show the range of possible
results leaves the analysis open to wide and extensive debate. The purpose
of sensitivity analysis is to detail which values of the assumptions or
input parameters may significantly alter the results. But, for the policy
decision process, a second important result of sensitivity analysis is to
show which assumptions are not as important, and therefore, do not need ex-
tended debate. It should be noted that sensitivity analysis rightly comes
after the model has been evaluated and verified and is not a substitute for
those procedures, which, as discussed above, have also been lacking.
Form of Output of Model—
A fourth technical problem encountered in the case studies is a lack of
appropriate form for the model's output. While not a widespread problem
in the sense that the models do predict pollutant concentrations in appro-
priate averaging times at various geographical locations, the models do not
always serve to answer the decision maker's broader questions. For example,
the question of the impact of the decision on employment and energy was
often raised. The ncdels are not meant to answer such questions but neither
are they designed so that their output could then be used in furtner analysis
of these other issues. One exception was the ozone standard revision case
64
-------
where the organization applying the air quality models was primarily respon-
sible for the employment and energy questions, rather than the air quality
question, so that they did do the full analysis. The San Francisco AQMP
case gives another example; again the responsible agency was not limited to
environmental policy questions but was required to look at the broader prob-
lem.
A second manifestation of the lack of an appropriate form for the models'
output can be seen by contrasting the San Francisco AQMP case with the
Connecticut TCP case. The use of the LIRAQ model in conjunction with employ-
ment models provided the decision makers with the flexibility they needed
to produce a politically acceptable plan. The modelers provided an analysis
of various strategies and, when the policymakers decided to reject them on
the basis of their land-use implications, the modelers were able to provide
an analysis of other strategies whose land-use implications were acceptable.
In contrast, Connecticut's Administrator, if he had accepted the model re-
sults, would have rejected over half of the proposed construction projects.
Clearly, the situation in Connecticut could have been such that, no matter
what strategy was chosen, over half of the projects would be stopped. But
this is not certain since the modelers could only produce one analysis.
Their model was not flexible enough to analyze different strategies within
the time frame needed. This caused the rejection of the model and the
Administrator was left without a strategy for assessing proposed projects.
Probably the major factor causing this problem is the professional back-
grounds of the modelers. As meteorologists, they do not know how to model
the economy or are not familiar with energy or employment models so that they
could make air quality models compatible with them. However, the problem
also steins from the narrow interests of the environmental agencies and,
sometimes, departments within the agencies. The Connecticut TCP case gives
the example of two agencies, with two separate missions (environmental pro-
tection and transportation), which could not come to a compromise because
they were never required to expand their analysis beyond what would be the
impacts on their own programs. The San Francisco AQMP case shows how reor-
ganizing the job responsibilities into joint task forces where, instead of
competing solutions, the agencies are responsible for coming up with one
joint solution, led to an acceptable plan. The NSPS case also used the task
force method to advantage in this way.
Social and Political Constraints
Just as the modeler faces limitations due to the state-of-the-art of
modeling, the decision maker faces social and political constraints within
which he or she must make decisions. For instance, visibility was not a
concern in federal air pollution programs until the 1977 Clean Air Act
Amendments were passed, while emissions of hydrocarbons and carbon monoxide
from automobiles have beep regulated by some states since 1960. Thus, while
in the future the laws may change, for the present they set certain bounds
on the range cf problems and scluticns available to the decision maker.
Funding and time ;onstraincs affect the decision maker as well as the modeler;
even if the technical information exists, it may be too time consuming co
collect.
65
-------
Examples of the social aid political constraints encountered in the
case studies are discussed in this section. However, it should be noted
that "politics" may not be as binding as is usually believed. A common ex-
cuse for why models were not determinant is that the politics of the situa-
tion caused the model results to be ignored. But an examination of the case
studies shows that there are often other reasons, such as the organizational
structure or the management of the decision process, which fueled the
controversies.
Energy and Employment Impacts—
As an example, we turn to the problem of energy and employment impacts.
Energy was a pervasive theme in two cases where the outcome was different.
The state legislature called for a revision of the Massachusetts' sulfur
regulations because of the rising price of oil. Originally, the regula-
tions concerning sulfur content in fuel had been set without an analysis by
air quality modeling. They were relaxed in the same manner under the emer-
gency legislation. Several modeling analyses were done to support (and
refute) the regulations after the fact and the relaxation was upheld. In
contrast, the MATEP cogeneration power plant was aimed at conserving energy,
but the debate remained centered on air quality and neighborhood land-use
issues.
A similar diversity of outcomes can be seen in the different treatment
of the issue of employment in the Westvaco and the Anaconda cases. In the
Anaconda case, when the company raised the spectre of lost jobs, EPA hired
a consultant to investigate and the issue did not delay the decision. In
the Westvaco case, the issue remained controversial after the company raised
it because EPA did no study of the problem. In this case, the EPA regional
office did very little technical analysis of any kind, including the "jobs"
questions, while in the Anaconda case massive resources had been allocated
to the air quality modeling, and the contract to study employment impacts
was not a large additional item of analysis.
Intervention of Other Interested Parties—
The problems of energy and employment can be foreseen and, if necessary,
studied. The problem of intervention by other political interests is not
so easily dealt with, however. In several cases a party not apparently
connected with the decision intervened with significant consequences. For
example, the Council on Wage and Price Stability issued a statement under-
mining EPA's position on Section 125 and its application to the Ohio SIP.
Also, the Office of Management and Budget required revisions to EPA's SIP
regulations so that the Anaconda company withdrew their proposal to reduce
emissions by 90% and instead insisted that reasonable available control
technology was all that was required. It took EPA (and a change of governors
and company ownership) over seven years to again approach the 90% reduction
level.
The Anaconda case also points out how the individuals who are directly
involved in reaching tne decision car. play a decisive role. A new governor
and the new owner of the aiine were acre amenable to reaching an agreement and
66
-------
the permit decision proceeded without further delay. In contrast, the
Connecticut Transportation Control Plan seems to have proceeded slowly or
been delayed because a new governor needed time to test the political waters.
Summary
These technical and political problems are not new. EPA has a moni-
toring program; it has underway a validation program; more time and resources
for technical analysis of all kinds is always sought; and outside political
issues and interests are always there. However, what I have tried to point
out by the examples cited in this section is that in several ways, these
technical problems are not caused by the state-of-the-art or by resource
and political constraints alone. Instead, policy, managerial and organiza-
tional elements add to the problems and extend the controversies in many
cases. It is these factors which are now discussed and which the recommenda-
tions of this report will focus on because, even if the technical problems
are solved, the controversies will not end and, even without solving the
technical problems, the decision process can be improved and better technical
analysis done if the motivational and organizational structure is changed.
POLICY, MANAGERIAL AND ORGANIZATIONAL PROBLEMS
The examples of the political constraints have been cited to illustrate
the various influences, outside the technical sphere of air quality modeling,
that have an impact on the decision making agency in a manner similar to
the technical problems' impact on the modeling analysis. The examples of
these technical and political constraints also serve to show that their
impact on the decision and the decision making process varies from case
to case. Now, however, I turn to a discussion of other factors that also
serve as constraints and significantly influence the decision making process
but can be more directly controlled by EPA or the decision making agency.
For convenience these factors are divided into three categories:
—policies of the federal EPA,
—management of the decision process, and
—institutional and organizational interests.
But they do overlap and actions taken to correct one problem may have con-
sequences in the other areas.
Policies of the Federal EPA
The decisions requiring the application of air quality models are, for
the most part, taken by the individual state or (if the stats lacks a SIP)
the EPA regional office. The major role of the federal EPA is to set
national policies within which the regional and local decisions ara made.
In a majority of our cases the lack of EPA policy on. a particular issue led
to considerable controversy and delay (see Table 2).
'The missing policies encompassed both political and technical areas.
Three political issues raised in our case studies and addressed in the 1977
-------
Clean Air Act Amendments, but not yet resolved by EPA are:
—transport (ozone, tall stacks and acid rain),
—NOX short-term standard, and
—prevention of significant deterioration (PSD) and growth.
A more technical issue for which EPA has not established guidelines and
which added to the controversy in several of our case studies is that of
monitoring,
—in what instances can EPA require an applicant for a permit
to set up a monitoring network on private property, and
—should monitoring be done in areas inaccessible to humans.
A third area
—what happens when national and regional policies conflict
(as in the case of the use of scrubbers or low-sulfur coal),
illuminates how federal policies do impact on regional and local decisions
and, to a limited extent, why the federal EPA may be reluctant to establish
such policies.
Political Issues—
The transport issue arose in several cases. For a state trying to
devise a State Implementation Plan the EPA policy of ignoring that the prob-
lem exists leads to the need for extensive controls to counterbalance the
pollution arriving from an upwind state. Connecticut was unwilling to impose
what they saw as extra controls. The EPA regional office and other states
in the Northeast agreed and together devised an alternative policy. But
this was rejected by the federal EPA as not consistent with national policy
and Connecticut's TCP was stalemated.
A more satisfactory solution was reached in the San Francisco AQMP case.
Their modeling considered the problem of the formation of ozone from upwind
sources. They were able to do this because the geographical area covered
by the plan did not cross state boundaries and because they had the resources
to employ a sophisticated photochemical dispersion model (LIRAQ) to make
their case.
The lack of a policy on transport also translates into specific conse-
quences when permit applications are considered. In the case of the Ohio
power plants, the company proceeded to build taller stacks without the EPA
regional office being able to require them to substantiate the need for them
and before any permit was issued. The company felt free to do so in the face
of EPA's confusion on the issue.
EPA has not yet promulgated a short-term standard for NC^. The lack
of such a standard in the MATEP case forced the state into doing an analysis
of health effects studies, duplicating analysis that the federal EPA has
been and is now conducting. It also put the state into a position of
68
-------
potential conflict; if the federal standard turns out to be less strict
than the one established by the state for MATEP--a state law requires that
all state standards be no more stringent than the federal ones—the standard
for MATEP will have to be relaxed.
The third political issue, PSD and growth, arose in the Pittston case.
Officially EPA ignores the problem of how and if additional pollution sources,
beyond the one under consideration, should influence the decision at hand.
The result is an application of the usual air quality models to one source
without recognizing that PSD increments, being small, may require more
accuracy than if the standard was a national ambient air quality standard
and without requiring an investigation of the impact of potential future
sources.
Technical Issue of Monitoring—
The more technical issue of monitoring also shapes the information avail-
able to decision makers. The Clean Air Act Amendments of 1977 allow EPA
to require monitoring on private land but EPA has been reluctant to impose
this requirement and has no guidelines for when it should be imposed. Two
of our cases demonstrate the problem caused when monitoring is not required:
Westvaco and the Ohio SIP. In each case the company set up monitors where
the EPA regional office's analysis indicated violations were not likely to
occur but did not locate monitors where violations were expected to occur.
Subsequently, the companies tried to focus the debate on the fact that their
monitors were not located so as to detect violations. In the absence of a
federal policy the companies successfully delayed for many years, installing
monitors which conceivably would have shown the need for them to install
addition control equipment.
Appropriate monitoring was conducted in two permit decision cases:
MATEP and Anaconda. In these cases, however, EPA did the monitoring and the
burden of proof that the monitoring was wrong was on the applicant instead.
National versus Regional Policies—
National policy which can be applied to all local situations is not
easy to develop. The Ohio SIP case highlights this problem. The effect
of the NSPS for coal-fired power plants is to have new plants install
scrubbers and use high-sulfur coal so that low-sulfur coal can be saved for
existing plants. However, in Ohio the existing plants prefer to use the
high-sulfur coal because it is locally available.
Another example is the Connecticut TCP case where the state was joined
by the EPA Region I office in sponsoring a conference on the transport issue
and in pressing for a resolution satisfactory to the nineteen states of the
north and east which sent representatives. Transport was one of the major
problems in this area of the country and had been delaying completion of
Connecticut's TCP. The regional solution was co apply reasonable available
control techniques (RACT) until more accurate and sophisticated modeling
was available to pinpoint the concribur.Ion of specific sources. From a
national point of view, however, such a solution would imply RACT was
69
-------
sufficient for a host of other states not experiencing the transport problem
and these states would want to revise their SIPs to require fewer controls.
Therefore, EPA headquarters stood firm in demanding the use of the available
oxidant air quality models.
There are always exceptions to policies whose consequences have to
be dealt with. However, as we will see in the next section, a lack of clear
and consistent policy, combined with a weak management process, leads to
long time delays without an efficient development of information.
Management of the Decision Process
Most of our case studies involved long delays and unresolved controver-
sies. As we saw in Table 3, this often resulted from the modeler's having
asked and answered the wrong question. Part of the reason for why this
happened can be traced to the management of the decision process. Three
problems in particular are found in the case studies as contributing to
this problem:
—no systematic effort at choosing the most appropriate model
was made or, at a minimum, was made clear to the public;
—no effort was made to establish the legitimacy of the deci-
sion process; and
—EPA's method of criticizing others' model applications was
ineffective.
Choice of Model—
The earliest technique for estimating the degree of control of emissions
needed to meet air quality standards was the linear rollback model. This
technique, as an adequate policy tool, pleased almost no one and in all of
our case studies other models were used. However, as Table 4 shows, the
choice of these other models was not subject to an exhaustive search for
the best model for the particular topographic and meteorological situation.
Even when appropriate constraints such as cost, data availability and user
needs are added to the list of criteria for choosing a model we still see
that in only a few cases were the models chosen for these reasons.
Instead, the models were chosen because that was what the modeler was
familiar with or because it was the latest developed model. In some cases,
aociels were rejected and others applied because the policymakers did not like
the results of the model. Also, in several instances a consultant was chosen,
often also for reasons of familiarity, who then applied the particular
model that had been developed by that consulting firm. The state or regional
office, then, relied on the consultant's model without evaluating it against
alternative models.
This procedure can be contrasted with the procedure followed by the Asso-
ciation of Bay Area Governments (A3AG) in its "he ice o~ the 1IRAQ model for
developing the San Francisco Bay Area's Air Quality Maintenance Plan.
-------
w
53
w
Q
g
Pi
O
3
H
cn
o
o
K
o
s
M
co
3
a:
w
H
M
OS
CO
1 O
H H
3 03 0)
(0 - TJ
C 4-1 O
0 C S
U (0
4J
-u .
s3
««
1 JS
M 4-1 rH
«d iH 0)
•H 3 T3
iH O
•H >,S
a 4-1
«8 -H
fe
>•>
1 43 CO
SJ 4-1
01 T> iH
4-J CU 3
0) C M
Q -,-j CD
6 PS
01
y
M C to
a; o> TJ
CO -H 01
C3 -U 01
3 a
<5
4J
i C
« i -H
4-1 5 fl
CO 0 M
Q " JJ
CO
1 d
0 ° S
CJ ° 4J
co
U-l
O 4J
M
oj <:
CQ 01
CO 4J
•o
01 1
4-i a w
n c o)
3 U >
— a -^i
cfl *J -i
> —
UJ <
X
X
X
|
j
i
i
M
^
^~ i
2
X
X
X
j
1
I—
o
^
J)
^ '
•^ 1
X
I
•rt
•o
* '
o
-J
iH
j*
X
1
o
U
3}
>
4J
M !
0)
X
X
X
X
_
CO
_)
4-1 CO
•D a
CO O
3 -J
— -> ,
•j u a i
3 3 -i '
CO '4- 3
a -( :« .
_2 3 u
X
1
1
04
— 1
V3
^
.^4
^
X
X
X
X
x
,
4-1 '
3
'^
•*4
4_
•j 1
o
H &,
C. V
X
X
X
w
CJ
".0
— <
o
2
T3
L. '
^
«j
•— 2v
w D* i
X
X
t
!
;
-,
fl
I/ ~3
C ^
-0 «, '•
; <
71
-------
ABAC had been faced with a plan, developed by EPA utilizing linear rollback,
which was totally unacceptable in its implications for land use and trans-
portation restrictions. In this political climate and under the initiative
of the Lawrence Livermore Laboratory, LIRAQ was developed. The first phase
of the new plan development consisted of drawing up criteria to which the
model had to conform. These criteria included:
—technical acceptability,
—reasonable cost,
—user-oriented, and
—consistency with other models to be used.
The government agency was willing to participate in the coordination of the
development of a data base and the state-of-the-art in order to assure con-
formance with these criteria. However, ABAC did not stop there. Throughout
the modeling process they were careful to evaluate alternative models and
maintain open avenues for public criticism of the model. The combination of
these actions ensured public acceptability of the results and, thus, in-
creased the chances that the plan could be implemented.
Legitimacy—
The ABAC process was managed from the point of view of establishing
legitimacy for the process. They realized that technical defensibility of
the model was not enough; they knew they had the best model given the particu-
lar circumstances, but they also knew that they had to prove it was the best
to the public. They did this by involving interested parties on all sides
of the issue from the beginning. This strategy was also followed in the
NSPS case and, clearly, it involves extensive meetings, time and resources.
However, other case studies show that a less open process generates
controversy over models (rather than over policy) and delay. EPA is required
to hold meetings for soliciting public comments and consider them in its
decisions. The problem with the process as followed in the other case
studies is that the public hearings are not held until quite late in the
process and the modelers are put in the position of justifying their already-
reached conclusions rather than changing their analysis in response to appro-
priate criticisms. Thus, the process as managed by EPA is an "open process"
in the sense that public hearings are held. However, it is not open in
the sense of encouraging all interested parties to bear some responsibility
for assessing the technical data and the possible solutions.
Involving environmental, business and other government agency represen-
tatives from the beginning gives these outside interests a stake in the
outcome. For example, ABAC included the California Department of Transporta-
tion as a participant and made the transportation representative responsible
for explaining and obtaining his department's concurrence as the work pro-
ceeded. In the NSPS case, EPA formed a joint task force which included
potential opponents from ether government agencies. In Connecticut, however,
t.ie state's Department of Transportation remained a critic rather than a
participant and in tne end helped sabotage the state's model by withdrawing
its support. Also, if environmentalists and industrialists are members of
72
-------
the working groups, it is harder for them to raise unimportant issues as
if they were important without participating in their resolution. For
instance, the Westvaco company insisted that no violations had been monitored
without admitting that the monitors were not sited so as to reveal violations.
Or, in the Ohio SIP case the companies assailed the assumptions of EPA's
models without admitting that the state-of-the-art allowed for no better
analysis.
Role of Criticism—
The method by which the state or EPA regional offices managed the process
of criticism in making permit decisions also had pitfalls. Generally, the
company applied for the permit by submitting modeling analyses which EPA or
the state would then criticize. By this point in the process, though, the
company had already set up any monitoring they were going to do and chosen
the model (or consulting firm to do the modeling). Thus, their response
to EPA criticisms would be minimal, perhaps including more recent monitoring
data but rarely changing their assumptions, model, or monitoring network.
This method of criticism, insistence on doing it EPA's way of "worst
case" analysis, maximizes the adversarial positions of the two parties and
does not appear to be an effective way of improving the modeling analysis,
based on the evidence of the case studies. The exchange of letters and lack
of progress, for instance, lasted over five years in the Westvaco case and
is a fairly accurate description of the actions of industry in the
Massachusetts' Sulfur Regulations and Ohio SIP cases. In the cases of MATEP
and Pittston, EPA asked for and received more refined modeling but not better
input data.
When the modeling analysis is being conducted strictly within government
agencies (e.g., in the San Francisco AQMP, Connecticut TCP, and the NSPS
cases), critiques of the models seem to have more effect. In San Francisco
and for the NSPS, criticisms were actively solicited and responses made;
Connecticut adopted EPA's model, dropping their own, at EPA's insistence.
However, counter examples also exist. Massachusetts ignored EPA's analysis
of their sulfur regulations because the EPA model did not analyze their
preferred strategy. Also, in the ozone standard revision case EPA ignored
CEQ's criticisms unless they happened to coincide with their own preference.
The deciding factor in these last three examples seems to be whether the
critic held the authority to enforce his preference. In the Connecticut
case EPA did; in the others the critic did not. For industry permits, EPA,
in theory, has the authority; but our case studies show they are reluctant
to apply it because such action leads to court and further delays.
Institutional and Organizational Interests
Each case study involves many actors who operate within the context of
their own interests and experience. As ve saw above, the modelers often
answered questions of their own devising rather than questions relevant to
tae decision maker's need (see Table 2" . The information they produced was
shaped by what they were able to provide bur also by what was of scientific
interest to them. Similarly, if we look, at the various institutions
-------
involved (EPA headquarters, EPA regional offices, state agencies, public
and private interest groups), we also see divergent interests shaping the
debate.
EPA Headquarters versus EPA Regional Offices—
The federal EPA's major responsibility in air pollution control is to
set the national ambient air quality standards and other general policies
called for in the Clean Air legislation. Examples are performance standards
for new sources, regulations on what a State Implementation Plan should
include, transport or tall stacks policy and other areas where actions impact
beyond state boundaries or there is a desire for all those affected to be
treated the same. The state is then expected to apply these policies for
their particular circumstances. When states have been delegated this au-
thority, the regional EPA office acts as the agent for EPA headquarters by
overseeing compliance with national policies (e.g., MATEP and the Massachu-
setts' Sulfur Regulations cases). However, when the state does not yet have
an approved plan for issuing permits or when it chooses not to act,then the
EPA regional office will act in its stead (e.g., Pittston and the Ohio SIP
cases). In this way the regional office's interests are shaped by regional
concerns which sometimes brings it into conflict with EPA headquarters'
offices.
This conflict between national and regional perspectives shapes the
actions of the various organizational levels within EPA. One example is the
Massachusetts' Sulfur Regulations case. The Massachusetts law requiring
the regulation revision mentioned the use of tall stacks as one technique
for ensuring compliance with ambient standards. While EPA does not yet
have a specific guideline for when tall stacks are appropriate, they wanted
to express their disapproval of the state's proposed policy. The EPA
regional office, however, having had experience with such situations where
EPA lacked specific policies and knowing that the likely consequence was
delay and controversy, rather than a resolution of the issue, instead urged
a speedy resolution of the specific regulation issues as proposed (i.e.,
limitations on sulfur content in fuel) and not the more hypothetical issues
of the law.
In areas not concerned with deciding general policies, however, EPA
headquarters supported the regional offices (and states). For example, in
the Ohio SIP case industry tried an end-run around the Region V office by
asking for a review by the EPA headquarters office located in Research
Triangle Park of the modeling done by the regional office. The reply was
that the modeling was adequate under the circumstances. In several other
cases the rsgional modelers requested and obtained assistance from head-
quarters on particular modeling problems (e.g., Pittston on the fumigation
issue and Anaconda on validation monitoring).
It should, also be pointed out that within one agency there will be
organizational units with divergent interests as well. Tha ozone standard
rdvisica case illustrates how different organizational interests will lead
ico members to view the same information differently. The Office of Research
and Development (ORD), charged with the development of air quality models,
74
-------
presented the various alternative techniques in the criteria document.
However, the choice of a model for obtaining cost estimates was left to the
Office of Air Quality Planning and Standards. In this case the initial
choice of the EKMA model seems to have been one of "what model am I familiar
with" rather than a complete search for what was most appropriate, the
assumption behind the modeling chapter in ORD's criteria document.
EPA versus the States—
The case studies describe to one extent or another the interaction be-
tween EPA (headquarters and regional offices) and the states, and it is at
this level that regional versus national perspectives come into sharpest
focus. In the Anaconda and Ohio SIP cases, for instance, the states refused
to act because they were not willing to risk having the companies close
down or cut back in employment in their state in preference for a state
with less strict emission standards. Thus, EPA had to conduct the modeling
analyses and make the decisions. Connecticut's governor and legislature also
refused to pass a strict transportation control plan unless and until other
states had such programs.
Of course, not all the states are the same. In the Westvaco case,
Maryland was pushing for strict emission standards while West Virginia was
considering relaxing its standards. But generally there is a regression
towards the mean; the states are indeed applying more uniform standards.
The Massachusetts' sulfur regulation changes illustrate this. The state
environmental agency was charged with relaxing its regulation of the sulfur
content in fuel so that it was no stricter than was necessary to meet the
national ambient standard. EPA actually encouraged such action by Connecti-
cut's environmental agency when they would settle only for one of their own
models. They rejected the model developed by the state at least in part on
the grounds that the SIPs had to be uniform in their technical bases and
this included consistency in data base and model. Such uniformity, it was
said, made EPA's review job easier.
EPA versus Other Government Agencies—
Just as states do not present a united front, neither do all government
agencies. For instance, the U.S. Departments of the Interior and Commerce
opposed EPA's permit allowing Pitts ton to build an oil refinery in northern
Maine. Their concern was with endangered species as well as air quality.
(Indeed, within EPA the same project that received an air permit was denied
a water permit.) Another, more predictable opponent of EPA decisions is
the Department of Transportation (DOT). In the Connecticut TCP case, the
state DOT supported the development of the analysis but continuously and
effectively opposed the results when they called for reducing highway con-
struction projects. In other words, DOT was opposing policies whose effect
would be to reduce its own programs.
In the San Francisco AQMP case the environmental program managers
foresaw this problem and tcck steps to neutralize CalTrans' opposition by
including the transportation agency on the governmental task force responsi-
ble for coming up with a solution. Thus, CalTrans was net allowed to get
75
-------
away with simply criticizing. They also had to make constructive sugges-
tions and participate in finding a solution acceptable to their agency as
well as to the environmental agency. Clearly, the resulting plan was not
all EPA could have wished for, but when compared to the almost total lack of
a plan in Connecticut, it is a more satisfactory solution. In a similar
manner EPA foresaw opposition from the Department of Energy and others in
the administration to their proposed NSPS for coal-fired power plants. In
this case, in addition to joint task forces, they scheduled many meetings
to inform the potential opponents about the analysis in order to raise and
resolve the technical issues.
Business Interests—
Industry is the most obvious adversary of EPA in the process of making
the type of environmental decisions found in these case studies. Generally,
the interest of the company applying for a permit or who will have to comply
with the state's or EPA's regulations is to minimize production costs and
this can usually be directly translated into minimizing expenditures on
pollution control equipment.* Several consequences arise from this diver-
gence of interests.
First, while the company is required to submit evidence of air quality
impacts of its proposed project, once it has picked an air quality model,
assumptions and the monitoring data, the tendency is to stick with it,
defend it if necessary, but not to respond to EPA's criticisms by either
changing the model or the monitoring. The advantage in this tactic is that,
at a minimum, it results in delay, especially if the matter ends up in court;
and delay means delay in incurring the costs of installing pollution control
equipment. We can see the importance of the capital cost factor by contrast-
ing the cases involving large construction projects with their long delays
and frequent court cases with the case of the Massachusetts' Sulfur Regula-
tions. While this was a case of relaxing standards, the utility industry
was lobbying for an even greater relaxation. However, there was not, in the
end, a court case or protracted delay in part because the costs involved
the cost of fuel not capital expenditures. These operating costs could be
passed on directly to the customer and no large fixed costs would be for-
feited if later the regulation was again changed. For similar reasons,
the utility in the Ohio SIP case delayed and went to court as often as
possible because they would not be able to pass on the costs of installing
the pollution control equipment directly to the customers.
A second consequence of the adversarial system is that EPA loses from
the beginning any ability it has to obtain information that the company may
not want to supply. Industry's use of the courts and delay led in several
cases to the inability of EPA to obtain adequate monitoring data. In both
the Westvaco case and the Ohio SIP case the company had sited monitors at
the beginning of the process .and refused to change them when EPA showed that
in the water pollution area production processes can often be changed
to reduce toxic effluents and at the s-atue tine save money. But this has not
often been done in the air pollution area.
75
-------
they were located at the wrong sites if maximum concentrations were to be
recorded. Of course, EPA can do its own monitoring, as in the Anaconda case,
but this is expensive, and, without a policy allowing them access to private
property, difficult if the company wishes to oppose them.
The third result of business interests in the adversarial system of
environmental decision making is the inclusion in the debate of issues
broader than just environmental issues to which EPA is not prepared to re-
spond. For instance, the concerns of energy and employment impacts were
often raised in the case studies. The problem is that, while these are legi-
timate public policy issues, they may or may not be important concerns in a
specific situation. However, in the adversarial system of environmental
decision making, business can raise the question without any attempt to re-
solve it. If EPA is unprepared with answers, the issue may remain controver-
sial without anyone knowing if it is important or not. The Westvaco case
is perhaps the clearest example of the problems that occur when EPA is
unprepared to respond to all the issues. In this case EPA did not do model-
ing of their own until fairly late in the process so that they were not able
to defend their assertion that violations would occur. Further, the company
raised the spectre of jobs and plant closings and had many individual workers
testify on the hardships this would cause. In many instances, EPA has
found the threat of plant closings to be just that, only a threat, but no
independent information on this issue was presented in this case and the
controversy remained.
The case of the Anaconda copper smelter shows how EPA can counter
these business interests. In this case they conducted their own monitoring
and modeling and hired a consulting firm to investigate the employment issue.
However, a fortuitous circumstance may also have helped in the resolution
of this case. The smelter was sold to another company, Atlantic Richfield,
in the middle of the negotiations. Atlantic Richfield, then, knew of the
environmental problems and presumably factored in the costs of pollution
control when they agreed to buy the company.
Environmental Interest Groups—
One other organization also has an interest in the type of decisions we
are looking at in these case studies: the public interest groups of environ-
mentalists. However, in terms of resources for supplying information they
are the odd man out. Generally, these groups do not have access to the re-
sources and expertise necessary to do their own modeling, although in several
cases they did do technical critiques of the modeling done by EPA and the
industry.
The problem is that their role is restricted to public hearings where
they voice criticisms and present evidence. Over time their positions be-
cooe predictable and tend to be discounted as a presentation of just: one side
of tne issae, just like business when it raises an issue but does r.ot offer
a solution or an explanation of its significance. Also, the environmental-
ists will raise the hard issues, for example, the short-tera NOX impacts in
the MATEP case or why did Connecticut not sue EFA in court over the
-------
transport issue, when the analysts or modelers had not raised them because
they did not have the answers.
A factor that often exacerbates this adversarial relationship between
EPA and the environmentalists is the fact that the public hearings usually
come late in the decision process. This means that the technical analysis
has been done and the modelers would rather defend what they have than change
it.
The San Francisco AQMP case shows the advantages of changing the pro-
cess so as to encourage public participation as early in the process as
possible. By having representatives from the environmental groups on the
original task forces assigned to conduct the technical analyses the environ-
mentalists could ask their hard questions early with a chance that answers
could be found. They also became allies with a stake in helping to find a
solution rather than remaining a critical spectator. In this way, the final
debate was focused on the policy issues by obtaining the agreement of all
interests on the technical issues in the earlier task force meetings.
SUMMARY
These case studies have investigated technical, political, policy,
managerial and organizational aspects of the use of air quality models by
EPA and the states. The technical problems which arose in these cases
included:
—lack of verification studies, especially for complex topo-
graphic situations;
—lack of input data, especially if industry supplied the data;
—lack of sensitivity analysis due in large part to EPA's "worst
case" requirements; and
—output formats that do not take into account the larger picture
of how air pollution fits into the energy and employment prob-
lems.
The political issues which appeared were:
— the analyses tend to ignore the larger issues because air
quality models do not address them, and
—other interested parties intervene to try to influence EPA's
decisions.
The policy problem centered on:
—a lack D£ directives because the focus was on the technical
rather than the policy implications.
78
-------
Several aspects of the management of the decision process led to:
—an isolation of the modeling analysis in the early stages of
the process, and
—a lack of legitimacy or credibility for the analysis.
The investigation of the institutional and organizational activities in the
cases points out that:
—institutions and organizations have separate and conflicting
interests in the impacts of decisions, which tend to fuel the
controversies rather than resolve them.
Clearly, not all of these factors will be under the control of the
decision maker or the modeler. The recommendations in the next section focus
on those that can be controlled or changed (mainly the technical and mana-
gerial ones). They recognize and try to work within the others as limita-
tions or constraints on the decision process.
79
-------
SECTION V
RECOMMENDATIONS
There is no easy methodology or approach to using an air quality model
in an environmental contrdl decision. The decision based on the models,
given the state-of-the-art of the modeling, will continue to be controversial.
However, several steps can be taken to make the available information more
useful to the decision maker.
The first three recommendations recognize several of the technical
problems that the EPA Office of Research and Development (ORD) could address
alone. The other recommendations focus on managerial problems. These
recommendations require ORD to act in concert with other organizations within
EPA and the states. Thus, they will require additional planning and coopera-
tion across organizational and institutional lines which recognizes the con-
flicting interests involved.
TECHNICAL RECOMMENDATIONS
Several unresolved technical issues (e.g., downwash, fumigation, reac-
tive pollutants, transport) result from the state-of-the-art of air quality
modeling. EPA's research programs are designed to discover and investigate
this type of problem and it is not my purpose here nor was it the primary
purpose of the project to indicate the priorities of the research agenda.
Instead, the recommendations of this section focus on technical planning
and analysis that can be done within the existing state-of-the-art.
1. Plan and require monitoring.
The lack of monitoring and input data created problems in many of the
case studies. Often the decision was to require a better monitoring network,
but this decision was made after a delay of several years and controversial
court cases. In most of the cases it was possible to define a better
monitoring network. Thus, an analysis of what monitoring would be appropri-
ate should be done or required by EPA as a first step in the decision process.
2. Detail guidelines for sensitivity analysis and output format when
developing a model.
In order to focus the modeler's attention on the policy issues that the
80
-------
decision maker will have to face when using a model one requirement for docu-
mentation of the model ought to be how a sensitivity analysis should be
done and what different forms the output can take, depending on the policy
question asked. The inclusion of such topics in model development rather
than in model application will make sensitivity analysis easier and the model
output more flexible but, more importantly, it will force the modeler to
develop a model which is more likely to be relevant to the policy (as opposed
to research) setting.
3. Focus verification studies on actual needs.
EPA's programs to verify the models should be based on what applications
and models have in the past shown to be problems. A systematic attempt to
discover these problems could be made through a questionnaire to the states
and other model users each year during the budgeting cycle. It is important
to ask the user, not the modeler, what problems the models are and will be
called upon to address.
MANAGERIAL RECOMMENDATIONS
Technical issues generally comprised only a small portion of the prob-
lems which arose in the case studies. The management of the decision process
also caused delay and controversy. The recommendations of this section focus
on the managerial aspects in an effort to increase the interaction and
communication between the modelers, the users and the other interested
parties. The purpose of the changes recommended here is to focus the decision
on the policy issues rather than on technicalities which cannot be resolved
and to reduce the time delays in an effort to increase the credibility of
the technical analysis and establish the legitimacy of the decision process
based on the models.
It should be noted that neither these recommendations nor the project
as a whole considers changes in the legislation. The legislative framework
is taken as a given and the managerial changes focus on improvements in the
development and use of technical information by EPA that can take place
within the existing law.
4. Include all interested parties from the beginning.
Non-governmental interest groups as well as non-EPA government agencies
should be included in the process from the beginning. This could take the
form of early fact-finding public hearings or it could include the formation
of task forces. This recommendation appears to require resources and time
which may not seem to be available. However, the case studies show that
decisions take years and, more importantly, that it is costly and time
consuming to have an issue raised late in the process when the issue could
and should have beer, dealt with earlier.
Several of the managerial recommendations suggest the formation of a
-------
task force or other cooperative group which crosses organizational and
institutional lines. In assessing the need for and appropriate form of such
a task force, attention should be given to the assignment of responsibilities
for the development of a solution that can be endorsed by the organiza-
tions represented in the group. Without commitments to the solution, even
if the solution is limited, say, to certain technical issues, such task
forces would be little more than another layer of meetings and incentive
to delay rather than a serious attempt to assess the importance of issues
and negotiate and balance interests.
5. Establish criteria and boundaries for the technical analysis for each
decision.
Criteria, such as what the broad issues are, what is the range of policy
options, if and how modeling will be done, and how the data will be used,
should be established before choosing which model, which data or which
assumptions to use. The current legislation, by mentioning air quality
models as a technical analysis tool, presumes they should and ought to be
used in a variety of situations. This in turn has tended to narrow the range
of issues considered and the options for technical analysis investigated.
If such questions as how the energy and employment issues will be
analyzed and whether monitoring rather than modeling is more appropriate for
the particular situation are addressed before any technical results are
presented, it is more likely that the decision will not founder on state-of-
the-art questions but instead can be focused on the policy issues.
6. Establish agreement on state-of-the-art and assumptions before the
technical results are presented.
Early in the process the discussion should focus on what is the appro-
priate model given the available resources. In this way it is harder to
choose a model because of its results rather than because it is appropriate
for the particular situation. How to negotiate such agreements will have
to come through practice as there are very few examples of such attempts,
much less successful outcomes. Several methods could be tried in various
situations, including:
—joint task forces,
—industry application requiring approval of model(s) before
analysis is undertaken, or
—independent scientific review body to set criteria.
Which method is chosen may depend on the size and expected impacts of the
project, the willingness of the parties to negotiate or, indeed, the incen-
tives for the parties to cooperate.
82
-------
7. Establish joint industry/EPA modeling studies.
Presently, a firm applying for an air quality permit will conduct a
modeling study as part of its application. EPA (or the state) then criti-
cizes the modeling analysis and sometimes conducts its own. In the spirit
of recommendations #4, #5, and #6, consideration should be given to under-
taking a joint modeling study for the decision, rather than separate ones.
Other interests should also be included but the major resources would be
from industry and EPA. The idea is to force agreement early on the technical
assumptions. Clearly, agreement will not always be possible but disagree-
ments would be investigated and their impacts detailed for the decision maker
without going to court.
8. Explain decision and alternatives considered in detail.
As decisions are made throughout the process, the alternatives con-
sidered should be explained in detail. Presently, the modeling analysis is
usually described, in more or less detail, but the alternatives such as other
models, other data or other issues, are not. To establish legitimacy for the
process and to counter the technical and political debate the basis for the
decision has to be made public. This also, of course, encourages the broader
consideration of alternatives in the decision.
83
-------
TECHNICAL REPORT DATA
(Please read /xuructions on the reverse before completing)
1. REPORT NO.
EPA-600/4-81-034
3. RECIPIENT'S ACCESSIOWNO.
4. TITLE AND SUBTITLE
CASE STUDIES IN THE APPLICATION OF AIR QUALITY
MODELING IN ENVIRONMENTAL DECISION MAKING
Summary and Recommendations
6. REPORT DATE
Mav 1981
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
C.G. Miller
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
John F. Kennedy School of Government
Harvard University
Cambridge, Massachusetts 02138
10. PROGRAM ELEMENT NO.
CDTA1D/03-0257 (FY-81)
11. CONTRACT/GRANT NO.
805558
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory—RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT.
Eleven case studies of the application of air quality models were undertaken
in order to examine the problems encountered when trying to use these models in making
environmental policy decisions. The case studies of air pollution control decisions
describe the decision process, the models used, the critiques of the models, and the
participation by outside interest groups in the decision process. The studies in-
clude two cases of federal decisions, seven state decisions, one local decision, and
a review of the evolution of modeling requirements in the Clean Air legislation. The
time covered is from 1970 to the present and includes several cases for which a final
decision has not yet been made.
The results of this investigation show that indeed the well-known technical
and political constraints exist but that unresolved policy issues, the management of
the decision process and conflicting institutional and organizational interests also
cause problems. Recommendations are made on how to improve the technical planning
and management of the decision process so that the air quality models can become a
better policy tool within the state-of-the-art, political and organizational con-
straints.
7,
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lOENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
9. DlSra'gvr.Ov a'-AT£M£NT
RELEASE TC °UBLIC
19. SEC'-'S TV CLASS (This Report/
UNCLASSIFIED
21. NO. OF PAGES
92
20 ScC^AITY CLASS (This pagr,
UNCLASSIFIED
22. PRICE
«»A fat".
84
------- |