June 1971
St. Louis Case Study
*. t *
Environmental Protection Agency
Office of Air Programs * -
Washington, D.C. ~, ,
Contract No. PH 22-68-60
TRW]
-------
11130-W002-RO-00
CONTROLLING AIR QUALITY;
ST. LOUIS CASE STUDY
S. E. Plotkin
D. H. Lewis
June 1971
Prepared for
Environmental Protection Agency
Air Pollution Control Office
Contract No. PH 22-68-60
TRW SYSTEMS GROUP
7600 Colshire Drive,
-------
The work upon which this
publication is based was performed
pursuant to Contract No. PH 22-68-60
with the Air Pollution Control Office,
Environmental Protection Agency.
-------
TABLE OF CONTENTS
Page
1.0 INTRODUCTION 1
1.1 EMISSION CONTROL STRATEGIES 2
1.2 EFFECTS OF CHANGING LAND USE 4
2.0 EMISSION CONTROL STRATEGIES 7
2.1 DISCUSSION 7
2.1.1 Conventional Source Category Strategy 11
2.1.2 Rollback 13
2.1.3 Least-Cost Strategy 1?
2.2 METHODOLOGY 19
2.2.1 Models 19
2.2.1.1 Implementation Planning Program 19
2.2.1.2 Least Cost Model 22
2.2.2 Setting Up The Diffusion Model 29
2.2.3 The Three Strategies 31
2.2.3.1 Conventional Source Category Strategy 31
2.2.3.2 Rollback Strategy 37
2.2.3.3 Least-Cost Control Strategy 41
2.3 RESULTS 44
2.3.1 The St. Louis AQCR Today 44
2.3.2 Conventional Source Category Strategy 48
2.3.3 Rollback 52
2.3.4 Cost/Benefit Comparison of Rollback and
Conventional Strategies 56
2.3.5 Least-Cost Control Strategy 56
2.4 CONCLUSIONS 62
-------
TABLE OF CONTENTS (CONT'D)
Page
2.4.1 Rollback Effectiveness 62
2.4.2 Uniform Application of Emission Standards
Versus Least-Cost Strategy 64
2.5 RECOMMENDATIONS 65
2.5.1 Rollback and Air Quality 65
2.5.2 Least-Cost Control 68
3.0 EFFECTS OF CHANGING LAND USE 70
3.1 DISCUSSION 70
3.2 METHODOLOGY 74
3.2.1 Review of Modeling Procedure 74
3.2.2 Basis for Procedure 75
3.2.2.1 Multiplicative Property of Diffusion
Model 75
3.2.2.2 Additive Property of Diffusion Model 76
3.2.3 Further Details of Procedure 77
3.2.4 Construction of the Emission Source File 78
3.2.5 Setting Up the Diffusion Model 83
3.2.6 Scenarios 85
3.2.7 Model Shortcomings 87
3.3 RESULTS 90
3.3.1 Where to Locate a New Power Plant 90
3.3.2 Dispersal of Industry 89
3.3.3 Comparison of Diffusion Model Results With
Those of Section 2 95
-------
TABLE OF CONTENTS (CONT'D)
Page
3 . 4 CONCLUSIONS ............................................... 97
3.4.1 Where to Locate a New Powerplant ................... 97
3.4.2 Dispersal of Industry .............................. 99
3 . 5 RECOMMENDATIONS ........................................... 100
4.0 REFERENCES ...............................................
-------
TABLES
Page
2-1 Constraints on Source Emission Control Levels 27
2-2 Breakdown of Area Source Emissions, St. Louis AQCR 36
2-3 Input Data for Sources Controlled Under the Least-Cost
Strategy 40
2-4 St. Louis AQCR - Existing Conditions 45
2-5 St. Louis AQCR - Results of Conventional Source
Category Strategy 49
2-6 St. Louis AQCR - Results of Rollback Strategy 53
2-7 Least-Cost Strategy Impact on Controlled Particulate Sources.. 60
2-8 Least-Cost Strategy Impact on Air Quality 61
3-1 Dummy Source File 79
3-2 Emission Sources to be Relocated 86
3-3 Strategy 10 - Maximum Dispersal of Point Sources , 88
3-4 Where to Locate a New Power Plant; Air Quality Results 91
3-5 Where to Locate a New Power Plant; Number of Receptors
in Different Ranges of Air Quality 92
3-6 Dispersal of Industry; Air Quality Results 93
3-7 Dispersal of Industry; Number of Receptors in Different
Ranges of Air Quality 94
3-8 Comparison of the Two Diffusion Models 96
-------
FIGURES
Page
2-1 St. Louis Air Quality Control Region 9
2-2 IPP Flow Chart 20
2-3 Diffusion Model Receptor Grid 30
2-4 Allowable Particulate Emissions Based on Input Heat Capacity.. 32
2-5 Allowable Particulate Emissions Based on Industrial Process
Weight 33
2-6 Potential Emissions Standard 34
2-7 Receptor Locations for the Least-Cost Model 43
2-8 St. Louis AQCR Existing SO. Ground Level Concentrations 46
2-9 St. Louis AQCR Existing Particulate Ground Level
Concentrations 47
2-10 St. Louis AQCR - SO Ground Level Concentrations After
Imposition of Conventional Source Category Strategy 50
2-11 St. Louis AQCR - Particulate Ground Level Concentrations
After Imposition of Conventional Source Category Strategy 51
2-12 St. Louis AQCR Particulate Ground Level Concentrations
After Imposition of Rollback Strategy 54
2-13 St. Louis AQCR - SO Ground Level Concentrations
After Imposition of Rollback Strategy 55
2-14 SO Cost and Benefit Curves 58
2-15 Particulate Cost and Benefit Curves 59
3-1 St. Louis Study Area r 73
3-2 Prediction Model Flow Chart 77
3-3 Location of Dummy Sources 82
3-4 Diffusion Model Receptor Net 84
3-5 Diffusion of Pollutants from a Point Source 98
-------
1.0 INTRODUCTION
This report addresses two major questions in air pollution control:
What emission control strategy should be used
by the states to achieve their air quality goals?
How can air quality effects of changing land
use patterns be predicted?
The report presents a comparison of three alternate emission control
strategies as applied to the St. Louis Air Quality Control Region. The
strategies are:
A conventional set of emission source-category
standards.
A Rollback strategy
A Least-Cost strategy.
The conventional strategy is used as a control from which to evaluate the
Rollback and Least-Cost strategies. Study results include regional costs,
air quality achieved, emission reductions, plots of pollutant concentration
levels (isopleths), and a measure of "benefit."
In addition, the report presents a description and brief analysis
of a s,imple procedure by which a diffusion model can be used to predict
the (air quality) consequences of shifting land use, without incurring
the considerable expense of continually re-running the entire model. The
procedure is used to analyze the effects of two scenarios in the St.
-------
A large new powerplant is added to the region.
Industry is dispersed to the suburbs.
1.1 EMISSION CONTROL STRATEGIES
Volume 36, Number 67, of the Federal Register (April 17, 1971)
proposes that, in order to comply with the Clean Air Act, each state
must submit to the Environmental Protection Agency a control strategy
for each national ambient air quality standard, and must demonstrate
that the strategy is adequate for attainment of each standard. The
criteria for "demonstration of adequacy" is the use of either a diffusion
model or a proportional ("Rollback") model.
The Rollback model is a means of defining regional emission control
needs in the absence of diffusion modeling, or when attempts at correlating
model predictions and actual air quality measurements fail. The model
defines a required percentage reduction (rollback) in total regional
emissions as the basis for achieving a desired air quality goal; the
magnitude of the reduction is based on the difference between the air
quality goal and the current air quality as detected by air pollution
measuring stations. As discussed in Section 2.1, such a reduction in
emissions does not guarantee attainment of the air quality standard,
because the model does not specify how the emission reduction is to be
attained. Nevertheless, it may be expected that a large number of states
will select a Rollback strategy. This report therefore attempts to show
whether an appropriately constructed Rollback strategy will achieve the
desired air quality. A discussion of different available strategies is
presented, and an "ideal" strategy is selected and implemented. The
-------
conventional control strategy which achieves the defined air quality
standard.
The conventional and Rollback control strategies presented in the
report are both based on the premise that it is inequitable and politi-
cally infeasible to apply emission standards which vary with plant
location within a political jurisdiction (state, Air Quality Control
Region, county, etc.)- Thus, both of these strategies control industry
located in areas whose air quality is above the standard as stringently
as those located in air quality trouble spots. However, it should be
clear that the price of this uniformity is an added cost which does not
contribute to attaining the air quality standard. These costs are de-
fined in this report by comparing the conventional strategy to a Minimum
Regional Cost strategy achieved by utilizing a Linear Programming Model.
In this strategy, emission sources are controlled only when they strongly
contribute to a violation of the air quality standard. In addition,
control is optimized so that the least cost is imposed on the region.
The "least-cost" strategy defined by the Linear Programming Model
attains the desired air quality standard at a considerable savings in
control costs to the region's industries. However, the patterns of
pollutant concentration throughout the region "flatten out," that is,
there are more areas where air quality is just at or slightly below
the standard than would be the case in a uniformly applied emission
control strategy. The "overcontrol" which is achieved in some areas
by a uniform strategy may be judged desirable because it leaves the
region with greater flexibility for continued industrial growth. Thus,
the savings of a "least-cost" strategy must be balanced by its added
-------
1.2 EFFECTS OF CHANGING LAND USE
Diffusion modeling is normally used to predict the existing air
quality in a region given the existing emission sources and their
locations. The nature of the diffusion model allows it to be used as
a predictive tool also, since the emission source file used in the
model can be altered to reflect the shutdown or alteration of sources,
the shifting of their locations, or the addition of entirely new
sources*. Since a full scale diffusion model run requires a very
substantial amount of computer time (normally several hours on an
IBM 360-40) and is thus extremely expensive, an analysis investigating
several land use alternatives becomes somewhat impractical if the
model must be rerun for every alternative.
It is extremely important, however, to be able to predict the
effect on air quality of changing land use. Although pollution control
strategies being promulgated now should reduce ambient air quality to
acceptable standards, continued economic growth can cause concentration
levels to rise back above these levels (even with National Emission
Standards for new industrial plants). Predictive tools are needed to
place new plants in areas which can sustain them without violating
*However, any changes of location of sources will result in some
degradation of the calibration of the model, since the calibration,
which corrects for topography in relation to plant locations is based
-------
standards, and to evaluate different land use plans so as to minimize future
pollution levels.
The procedure presented here for using the diffusion model as a con-
venient predictive tool utilizes the linear qualities of the diffusion model.
The diffusion model is run once with an emission source file consisting of
all the sources presently existing in the region, plus a number of "dummy
sources"area and point sources with miniscule emissions. A new air quality
"map" of the region can then be reproduced, without rerunning the diffusion
model, while scaling any source's emission up or down. Sources can therefore
be made to disappear, or appear (if they were "dummies" in the first run), or
grow...utilizing a simple program which requires only a few minutes of com-
puter time. The model used in this study was the Control Strategies Segment
of the Implementation Planning Program (IPP) and is thus more complex than
is necessary given a separately developed program.
The purpose of this report is three-fold:
To present the prediction procedure.
To present the results of two "scenarios" produced
by the procedure.
To describe the shortcomings of the procedure and
define what can be done to overcome them.
The limited nature of the model demonstration prevented any conclusive
estimate of the efficacy of the prediction procedure to be made at this time.
The scenario results indicate that the addition to a region of a powerplant
with a very tall stack does not make a strong local impact on an average
annual" basis, a conclusion which agrees with expectations. The "dispersal
of industry" scenario illuminated some mild possibilities for air quality
improvements by shifting emission source locations, but the results were
definitely not clear cut and deserve further study. Recommendations are
made in Section 3.5 for investigating the accuracy of the defined air quality
prediction procedure; it is felt that the potential value of such a procedure
-------
2.0 EMISSION CONTROL STRATEGIES
2.1 DISCUSSION
The purpose of this section is to examine three kinds of strategies
for achieving a region's air quality goals. These strategies are:
A conventional set of emission source-category standards
A Rollback strategy
A least-cost strategy
The examination is designed to answer two questions:
Will a well-constructed Rollback strategy achieve its air
quality goal?
What price does a region pay for applying emission standards
uniformly, without regard to plant locations?
As noted in the Introduction, it is highly probable that many states
will select Rollback strategies for their air quality implementation plans.
The use of such strategies does not guarantee that the designated air
quality standards will be met, because the reduction in total regional
emissions specified by Rollback will not necessarily achieve the same
reduction in ground level pollutant concentrations (air quality).
The failure of a Rollback strategy could have serious consequences
for a state. A further stiffening of the emission standards could be more
expensive than the accompanying reduction in emissions and improvement in
air'quality would warrant ... because the industrial plants, incinerators,
-------
pollution control devices which might have to be removed and replaced with
more efficient devices, at a total cost possibly far in excess of what
would have been spent installing the more efficient equipment in the first
place. It is therefore important that Rollback be able to achieve the
stated air quality goal without a process of trial and error.
This study attempts to show whether or not a Rollback strategy
will achieve its stated air quality goal given:
A goal which is known to be reasonable
A set of emission standards which are equitable to the
controlled industries and which will cause a reasonably
uniform reduction in emissions throughout the region.
The St. Louis Air Quality Control Region (Figure 2-1) is used as
a test area. Pollutants examined are SO,, and particulates. A set of
source-category emission standards similar to those in the 305(a) Cost
of Clean Air Report to Congress is applied to the region in order to
establish a control case with which to compare Rollback strategy. The
air quality goals set for the strategy are those achieved by the control,
thus assuring that the goals are attainable (the method used in arriving
at an air quality goal has nothing in common with the conventional pro-
cedure, which is to base such goals on the known effects of different
air quality levels). The modeling tool used is the Implementation Planning
Program, which predicts the air quality, emission reductions, and costs
resulting from the application of emission control standards. Besides
comparing figures of merit produced by IPP (air quality, cost-effectiveness,
total cost), a "cost/benefit" comparison of the two strategies is made using
a Regional Cost/Benefit Model (Reference 3) developed in parallel with
this study. The model utilizes a pro forma linear damage function which
-------
KILOMETERS
Figure 2-1.
-------
relates total damages per capita from all direct effects of pollution to
regional air quality.
Neither the Rollback nor the 305(a)-based source-category emission
strategies consider source location as a determinant of allowable emission
levels for industrial plants. Given two identical plants in the same
political jurisdiction, one of which is located in the urban core area and
is contributing heavily to an air pollution problem, the other located out-
side the core area in a "clean air" district...a source-category emission
standard requires both plants to control to identical levels. If the
attainment of an air quality standard is defined as that situation where
no location in the region has a ground level pollutant concentration above
the specified limit, then obviously this uniform method of control is not
the most efficient way to achieve "air quality"...the most efficient, or
least-cost method would be to vary control requirements so as to impose
the heaviest controls on those plants most affecting concentrations at
locations where the standard is violated, while allowing those plants which
do not contribute to air quality violations to remain uncontrolled.
In this analysis, a least regional cost strategy for particulate
control is identified. The strategy is based on the selective control of
the 27 largest emission sources in the region according to their relative
contribution to ground level concentrations at receptors where air quality
standards are violated. A linear programming model is used to apply con-
trol devices to these sources to attain an air quality standard at selected
receptors identical to that achieved by the source-category strategy,
allowing a direct comparison between the alternative control schemes to be
made.
-------
2.1.1 Conventional Source Category Strategy
The first of the three emission control strategies compared in this
report, the Conventional Source Category Strategy, is used essentially as
a control from which to evaluate the Rollback and Least Regional Cost
Strategies. The air quality achieved by the "conventional" strategy, as
measured by the atmospheric diffusion model of the Implementation Planning
Program (Section 2.2.1.1), is used as the "goal" for the Rollback and Least
Cost Strategies so as to provide a clear basis for comparison of the
strategies.
The Conventional strategy is quite similar to that used in the
305(a) Cost of Clean Air Report to Congress; it consists of the following
emission standards:
Particulate Fuel Combustion Sources - HEAT INPUT STANDARD
Particulate Industrial Process Sources - PROCESS WEIGHT
STANDARD
Particulate Solid Waste Disposal Sources - POTENTIAL EMISSION
STANDARD
S02 Fuel Combustion Sources - EQUIVALENT FUEL SULFUR LIMIT
S02 Industrial Process Sources - EXHAUST CONCENTRATION
STANDARD
Although the Process Weight Standard and Exhaust Concentration
Standard are common control measures, they both have serious
shortcomings. The Process Weight Standard penalizes industries and
processes which are conservative of raw materials, and inversely rewards
I
those which use large quantities of raw materials, by allowing higher
emission rates for higher process weights with no regard to the actual
-------
physical output, in finished goods, of the plant. The Exhaust Concen-
tration Standard disregards the fact that the only sensible measure of
emissions is the actual amount of pollutant leaving the stack, and not
the relative dilution of that pollutant in the exhaust gas. Since some
types of processes naturally produce more exhaust gas than others, the
Concentration Standard favors these sources over those which produce
similar amounts of pollutants but have lower exhaust gas production.*
* A justification for this "favoritism" is that the cost of control
devices varies directly with exhaust gas rate, so that the high (gas)
volume plant would incur far greater expense to control to the same
efficiency as the low-volume plant. However, this variation of
control device cost is certainly not accounted for by an allowable
concentration which is the same regardless of gas rate; at the least,
a concentration which varies with gas rate might be used.
-------
2.1.2 Rollback
Rollback is a means of defining regional emission control needs in
the absence of diffusion modeling, or when attempts at correlating model
predictions and actual air quality measurements fail. Rollback defines
the net reduction in total regional emissions needed to satisfy a given
air quality standard; the reduction R is calculated by the formula:
X X
_ max standard
max 'oackground
where X = ground level concentration ("air quality") of a
given pollutant
X = existing air quality at the location having
the highest measured or estimated concentration
in the region
X . , , = air quality standard
standard
X, , , = background concentration
Background
Although it is implicitly assumed that the reduction R will achieve the
desired air quality level, the actual resulting air quality may be con-
siderably better or worse than the standard, depending upon the means
chosen to implement Rollback. For instance, it is possible to concentrate
on reducing emissions from an area's powerplants (which traditionally pro-
duce a significant portion of total regional emissions) yet not affect air
quality in the urban core areas where the major problems exist. On the
other hand, if reductions are concentrated geographically in and around
the pollution "peak" areas, Rollback requires a more severe reduction than
is really necessary.
-------
A desirable rollback strategy should have the following characteris-
tics:
1) It should be equitable. Industries should not be penalized
for prior attempts at controlling emissions, nor should
certain emission sources be controlled severely while others
escape control.
2) If the level of control is varied geographically,
the areas of maximum severity should be those which
have an air quality problem. Otherwise, severity of
control should be uniform throughout the area.
The conceptually simplest method of applying rollback is to require
all pollution sources in the region to reduce their emissions by the factor
R. Although this strategy is certain to achieve the desired air quality*,
it is not used because of its gross inequity. Industrial facilities which
have taken steps to control their pollution prior to any legal require-
ments are penalized for this action, since they must reduce their
already controlled emissions by the same percentage that is applied to the
uncontrolled polluter...and cost-effectiveness of pollution control devices
decreases as the total degree of control increases. Furthermore, plants
which are already utilizing extremely high efficiency devices will not be
able to comply with added reduction requirements, forcing legal penalties
upon the most (rather than the least) conscientious industries. One con-
cludes from this example that a good strategy would give credit to:
The use of emission control devices.
The burning of "clean" fuels.
*Because such a uniform reduction will automatically reduce ground-level
concentrations (over the background level) at EVERY POINT IN THE REGION
by the same factor R.
-------
An initial investment in a "clean" process or piece
of equipment.
Certain types of emission standards satisfy this condition very well
and are particularly suited for rollback applications. For instance, a
Potential Emissions Standard bases the emission rate a plant is allowed
not on its present emission rate, but instead on that rate it would have
if its controls were removed. As an example, Plant A and Plant B are
identical except that Plant A has installed an electrostatic precipitator
with efficiency of .90 to control its particulate emissions, while Plant
B's emissions are uncontrolled; a standard which requires 85 percent
control of potential emissions is applied to both plants. The allowable
emissions from the two plants are the same; however, Plant A is within
the law, since it already controls its potential emissions by 90 percent,
and thus it incurs no additional expense; Plant B, on the other hand,
must purchase a control device of at least 85 percent efficiency.
Although the Potential Emission Standard (PES) is a generally
equitable* standard for industrial process emission sources, it is not
satisfactory,in its present form, for application to fuel combustion sources
(boilers). In the case of sulfur dioxide emissions, a PES does not account
for those sources burning low sulfur fuels, i.e., a plant burning low sulfur
fuel would be allocated a smaller allowable S0_ emission than an identical
plant burning high sulfur fuel. A more equitable emissions standard would
be an "Equivalent Low Sulfur Fuel Standard," which requires either the
use of fuels containing less than a specified percent of sulfur by weight,
or else the installation of a flue gas desulfurization device yielding an
equivalent controlled SO emission rate. PES's have the same limitation
*An exception: When two plants of identical capacity use different
processes, one of which is "cleaner" than the other.
-------
with respect to low ash-content fuels for particulate emission controls.
In addition, the standards do not account for the wide range of particulate
emission factors from coal combustion. For example, potential emissions
from a wet bottom boiler with reinjection are nearly twice as high as those
from a similar type boiler without reinjection; thus, the operator who
initially chose the cleaner equipment would be required by a PES to reduce
his emission rate to a considerably lower level than that required of the
"dirtier" operator.
One means of giving "credit" to the operator of a clean plant is to
calculate potential emissions on the basis of an "average plant" rather
than the actual plant, using some measure of plant size such as kilowatts/
hour produced (for powerplants), etc. Thus, the allowable emission rate
depends only on plant size and not upon previously installed controls,
fuel types, or boiler types. A commonly used emission standard which
duplicates the effect of such an "improved" PES is the Heat Input Standard;
this standard specifies an allowable emission rate on the basis of the
maximum BTU input to a fuel combustion plant. Since specification of a
model plant would require the definition of a relationship between poten-
tial emissions and heat input (or some other measure of plant size), the
"x axis" of the Heat Input curve could easily be changed from BTU/hour to
Potential Emissions; thus, the two standards are interchangeable.
In summary, we may define a "Rollback Strategy" for controlling SO^
and particulates as follows:
Particulate Fuel Combustion Sources - HEAT INPUT STANDARD
Particulate Industrial Process Sources - POTENTIAL
EMISSIONS STANDARD
-------
Particulate Solid Waste Disposal Sources - POTENTIAL
MISSIONS STANDARD
SO Fuel Combustion Sources - EQUIVALENT FUEL
SULFUR LIMIT
SO Industrial Process Sources - POTENTIAL
EMISSIONS STANDARD
This strategy satisfies the above definition of "equity". It
remains to be shown whether or not the strategy will achieve the air
quality standard selected for the region.
2.1.3 Least-Cost Strategy
Both the Conventional source category strategy and the Rollback
strategy described in the preceeding sections apply emission standards to
each of three categories of emission sources: fuel combustion, industrial
process, and solid waste disposal sources. Smaller plants are typically
given a break when these emission standards are designed, but all plants
of given size and type are treated equally. In other words, an integrated
iron and steel plant in the central business district of a given region
would be required to control to the same level as a steel plant of the same
size located in the outskirts of that region. This ignores the fact that
the suburban plant is not likely to be contributing to an air quality
violation to the same extent as the centrally located plant.
The Least-Cost Control Strategy tries to overcome this deficiency by
recognizing the dependence of air quality on plant locations (i.e., on
meteorology and topography). Individual point sources are controlled to a
level which depends upon how much they contribute to pollutant concentrations
-------
above the air quality standard. Intuitively, this should be a cheaper
control technique, at least from the regional point of view. The dif-
ference between the cost associated with the conventional or rollback
source category control strategies and the least-cost control strategy is
an indication of what it is costing the region to maintain the equity
treatment of plants implicit in a source category strategy.
It should be noted that this additional "equity" cost allows the main-
tenance of a level of air quality in certain areas which is considerably better
than that which would be obtained with the least-cost strategy. The
"over control" caused by the source category strategies provides a cushion
for further industrial and residential development. If the region is con-
trolled only to where the air quality standard is barely met at all
points, then the addition of any new emission sources will cause an air
quality violation; thus, the least-cost strategy might restrict a region's
flexibility as far as locating new development is concerned.
The strategy that is developed in this study controls particulate
emissions from 27 major point sources in the St. Louis AQCR. The air
quality standard used as a constraint is that attained by the conventional
source category strategy, thus allowing a clear comparison between these
two different strategy types.
-------
2.2 METHODOLOGY
2.2.1 Models
2.2.1.1 Implementation Planning Program
The Implementation Planning Program (IPP) is a .series of models
developed by TRW under contract to APCO which help in the evaluation of
alternative strategies for control of air pollution sources. A flow
chart of the program is illustrated in Figure 2-2.
The heart of IPP is the atmospheric diffusion model, which predicts
expected regional ambient concentrations of pollutants by mathematically
simulating the dispersion of these pollutants throughout the region. The
inputs to this model are a detailed emission inventory, various meteoro-
logical data, and measured pollutant concentration data. The emission
inventory lists individually the major sources of pollutants (power plants,
incinerators, etc.) and describes in detail those parameters which
characterize the sources and their emissions. The inventory also
characterizes those emission sources which are too small to be identified
individually by aggregating them to form "area sources." The meteorological
data includes wind speed and direction, mixing depth, and other phenomena
which describe the transport mechanism which carries the pollutants from
the sources throughout the region. The measured concentration data is
used to calibrate the theoretical model, in order to account for inac-
curacies in the diffusion equation, inaccurate source emissions and
meteorological data, irregularities in the area's topography (the dif-
fusion equation assumes a flat plain), and other errors.
Besides predicting present air quality conditions, IPP can predict
the effects of a pollutant control "strategy" (a series of emission con-
trol standards which apply to all major sources in the area) on the area's
-------
Figure 2-2 IPP Flow Chart
FOR CALIBRATION
Air Quality Data
Meteorological Data
Emission Inventory
Emission Standards
ATMOSPHERIC DIFFUSION MODEL
CONTROL COST MODEL
EXISTING AI
QUALITY
DISPLAY
CONTROL STANDARDS MODEL
POST-CONTROL
AIR QUALITY DISPLAY
POINT SOURCE
CONTROL COSTS
-------
air quality and measure the resulting pollution control device demand and
cost. This is accomplished by a control cost, a control standards, and a
control strategy model.
The control cost model assigns to each major source all those
control devices which may reasonably be used for reducing emissions. Devices
for the control of particulates are generally applied to the outlets of
the polluting process (usually the stacks). Sulfur dioxide control is
usually accomplished by switching to low sulfur fuels or by the use of
flue gas desulfurization techniques. The model's output consists of lists
of device names, their efficiencies and costs, and their effects on
pollutant emissions, for each major emission source.
The control standards model applies a series of emission standards
to the three categories of emission sources: fuel combustion, industrial
process, and solid waste disposal sources. Output consists of a list,
for each major source, of the applicable standards, their prescribed
allowable emissions, suitable control devices selected on the basis of
effectiveness and least cost (one device for each standard), and the cost
and effect on emission of the devices (obtained from the control cost
model).
The control strategy model calculates the effects of applying
selected sets of three emissions standards (two in the case of SO.) to
every political jurisdiction in the control area. Selecting the applicable
results from the control standards model, the model develops a picture of
the change in emissions (and costs) resulting from a realizable pollution
control alternative. Using the output of the atmospheric diffusion model,
the strategy model recomputes the pollutant concentration distribution
resulting from the new emission pattern, thus allowing a decision to be
made on the effectiveness of the strategy based on both the resulting
costs (on a regional, industry-by-industry. political jurisdiction-by-
political jurisdiction, or source-by-source basis) and the actual air
quality produced by implementation of the standards.
-------
2.2.1.2 Least Cost Model
The purpose of the Least Cost Model is to determine a set of
point source emission reductions which will achieve a stated air
quality standard at the least total control cost to the region.
Basically, the model relates, via a set of transfer coefficients
obtained as an output from the IPP Air Pollutant Concentration Model
(Diffusion Model), the emissions at each source in the region to the
ground-level concentration (air quality) at a set of receptors dis-
persed throughout the region. With this approach, the control of
localized pollutant "hot-spots" is achieved, as well as control of
the central area of the region. This is particularly important
for air quality control regions which include a number of geographically
separated centers and industrial areas.
Since source emissions and source-receptor transfer coefficients
determine ground-level concentration, the model contains a set of
inequality constraints which restrain the ground-level concentra-
tion at each receptor location to be less than the value selected as
an air quality standard for each geographic subdivision of the region.
Long-term (annual or seasonal) average pollutant ground-level con-
centrations are used, although shorter term averages could be used with
the model as formulated, if the appropriate set of transfer coefficients
were available.
-------
The model considers the relationship between source control
technology, control costs, and control effectiveness at each pollu-
tant source in the region. The IPP Control Cost Segment is used
to determine the least-cost source control equipment mix required
to obtain given levels of reduction of pollutant emissions, with
the assumption that control costs are approximately piecewise-
linear. The amount of control to be applied to each source is
constrainted to be less than or equal to that which is technologi-
cally possible at this time. Additional constraints on the sources
may be imposed to insure that a given source achieves a minimum
level of emission reduction.
Thus, with the previously mentioned constraints, the model
minimizes total direct control costs for the region in the fol-
lowing manner :
Let f . ., (E..) be the transfer function between
the average ground-level concentration (g.l.c.)
of the j pollutant at the k receptor from
the i source, and let n . , n , n, be the total
number of sources, pollutants and receptors to be
considered, respectively. Values of this transfer
function are currently computed by the Air Pol-
lutant Concentration Segment for annual
averages. The ground-level concentration
-------
Is then
If XjV and X-ii, are the 8-1-c. of the jth pollutant at the.kth
jit jfc
receptor before and after control of this source, and E?. and E..
jit JK
are the source emission levels of the jth pollutant at the ith
source before and after control, then
If the required reduction is defined as
and the usual linearity assumption is applied, i.e.,
f k -f±jk(A) +f(B), (4)
then
-------
This cost function represents the minimum total cost of achieving a
reduction of an amount X . in the emissions of the j pollutant
from the i source. Thus, for a given set of source control measures
[X ] , the total direct cost to the region will be given by
ni ni
CT *
1=1 j=l
and the corresponding set of ground-level concentration reductions,
[X?i, ~ X.v]> will be given by equation (3).
JK J K.
At this point, the control strategy, or means of selecting the
preferred set of control measures [X..], is introduced. In
general, there are two basic approaches. The required set of
control measures may be computed by an input scheme, such as
applying an equiproportional emission standard to the relevant
sources. For each source requiring a maximum allowable emission
level of E3. , the reduction is expressed by
° max
Alternatively, a different set of x..'s could be computed by
some preference criteria. The most widely utilized preference
criterion is an economic one, the so-called least total cost
to achieve a given air quality standard. Although air quality standards
are generally stated in terms of concentration and relative frequency of
occurrence, an arithmetic average g.l.c. requirement is always implicit.
-------
Denoting the average g.l.c. required by the standard for the j
th
pollutant as S , and requiring that no receptor within the region
have a g.l.c. greater than the given standard, there are con-
straints on both the least-cost and equiproportional strategy so
that
Xjk
-------
TABLE 2-1. CONSTRAINTS ON SOURCE EMISSION CONTROL LEVELS
G.L.C. ] 1
Before | Source | Source
Control | 1 | 2
i |
o _ r ,
X12 " [a112Xll + a212X21
Q
*** I 1 1\jT^l 1 1 1 Vr^l 1
11M 11 /J_M 21
O r ,
X21 ~ la!21X12 3221X22
Of ,
X2M ~ la!2MX12 322MX22
X31 " Ia131X13 + a23X23
O _ r .
A Ou I 1 IW^I O *>1U^OO
Jn ±Jti 13 23rT 23
] ] IG.L.C.
1 j Source 1 After
j N j Control
+ ' ' ' + ^l^Nl1 - Sl
+ " + aN12XNl] <_ BI
+ " + aNlMXNl] - Sl
-»- ... + aNnxN2] <_ s2
+ " + 3N2MXN21 - 82
"*" " + aN31XN3] <_ s3
+ " * aN3MXN31 - S3
POLLUTANT 1, RECEPTOR 1
POLLUTANT 1, RECEPTOR 2
POLLUTANT 1, RECEPTOR M
POLLUTANT 2, RECEPTOR 1
POLLUTANT 2, RECEPTOR M
POLLUTANT 3, RECEPTOR 1
-------
the problem is amenable to solution by linear programming techniques.
The source deck listing for the Least-Cost Model is presented in
Appendix A. This Model was developed on the TRW Timeshare System
(based on a CDC 6500 computer) in F0RTRAN IV. The simplex technique
is used in the solution algorithm. With only minor changes in the
read and write statements, the program can be used on any system with
a F0RTRAN IV compiler. As currently dimensioned, the following con-
straints on problem size must be observed:
Ns + NR < 45
NS <_ 30
VNS + NR + 3(V + 2°] - 4°°°
where
N = number of sources
N = number of receptors
K
N = number of straight line segments
used to represent the cost function.
The array sizes are easily changed to handle larger problems, if
desired, i.e., problem size is limited only by available storage capacity.
As previously stated, the model formulation is compatible with
analysis of tactical, short-term control measures, as well as strategic,
long-term control. In the former situation, the source-receptor transfer
functions, f.,.v ^E^4^» are replaced by the functions applicable to the
short-term control measures being considered. The normal long-term cost
functions are replaced with incremental costs of short-term control
measures available for each source. The constraints, S , are replaced by
a set of g.l.c.'s, which represent the upper limit tolerable in an acute
situation. The model, as formulated, then computes a minimum-cost set of
control measures for this situation.
-------
2.2.2 Setting Up the Diffusion Model
The diffusion model receptor net constructed for this study con-
sists of a widely dispersed grid of 49 receptors spaced at 15 kilometer
intervals covering the entire area, and 136 additional receptors at closer
intervals covering the more industrialized portions of the area. Figure
2-3 shows the receptor net.
Parameters input to the model are:
SO Particulates
Ambient pressure, millibars 997.29 997.29
Ambient temperature, °Kelvin 285.5 285.5
Mixing height, meters 1387.0 1387.0
Half life, hours 3.0 Infinite
The model was calibrated using 1968 data from 23 S02 and 15 particulate
measuring stations. Regression parameters were as follows:
SO Particulates
y-intercept , vg/m3 36.97 62.11
slope .2711 .6039
regression coefficient .692 .623
regression coefficient of
5% confidence level .413 .514
The particulate y-intercept is considerably higher than is normally
encountered and calls into question the accuracy of the absolute value
of the particulate air quality results obtained. The air quality con-
clusions drawn from diffusion model results should not be seriously
affected by this high y-intercept value since they are comparative in
nature.
-------
Figure 2-3. Diffusion Model Receptor Grid
-------
2.2.3 The Three Strategies
2,2.3.1 Conventional Source Category Strategy
The emission standards used in the Source Category Strategy are
as follows:
Particulate Fuel Combustion
HEAT INPUT STANDARD
Figure 2-4 is used to define a design allowable emission
rate, pounds per 10 BTU. The actual allowable emission
rate reflects actual source operating practice as follows:
Allowable Emission Rate = (Design Allowable Emission Rate)
* Actual Heat Input, BTU/Hr
Rated Capacity, BTU/Hr
Particulate Industrial Process
PROCESS WEIGHT STANDARD
Figure 2-5 is used to define a design allowable emission
rate, pounds per hour, based on maximum process weight.
This value is divided by the "use factor," which is the
ratio of maximum to actual process weight. In reality, the
use factor in this analysis was uniformly considered to be
1.0 due to lack of data, and process weight entered was
therefore "average" process weight.
Particulate Solid Waste Disposal
POTENTIAL EMISSIONS STANDARD
Figure 2-6 defines the design allowable emission rate based
on the uncontrolled (potential) emission rate for each
source operating at maximum capacity. This rate is calcu-
lated as:
-------
10
u>
vo
o
c
o
0.6
0.2
Ul
a> 0.11
0.01
10
100 1000
Heat Input (10° BTU/hr)
10,000
100,000
-------
iuo
JS
J3
e
o
H
00
09
1
0.1
1000
10,000 100,000
Process Weight (Ib/hr)
1,000,000
10,000,000
-------
OJ
1000-
o
i '»
I
1.0
0.3
0.3
0.1
1.0
10 100
POTENTIAL EMISSIONS, LB/HR
10000
-------
(Existing Emission Rate) (Use Factor)
1 - Existing Control Efficiency
, Fuel Combustion
EQUIVALENT FUEL SULFUR CONTENT RESTRICTION:
1% SULFUR COAL, 1.38% SULFUR OIL
This standard applies a restriction on the sulfur level
of fuels used, or else demands an equivalent reduction in
S0_ emissions via flue gas desulfurization techniques.
SO. Industrial Process
EXHAUST CONCENTRATION STANDARD:
500 PARTS PER MILLION
Area source emissions were appropriately scaled by constructing
Table 2-2 from emission and fuel consumption data, and then applying
the emission standards to scale down appropriate segments of the total
emissions (percent reductions were calculated by using average values
for existing fuel sulfur and ash levels and BTU contents). Scale
factors calculated for this strategy were:
S02 - .53
Particulates - .60
-------
Table 2-2 . Breakdown of Area Source Emissions, St. Louis AQCR
Category
Transportation
Residential
Commercial/
Institutional
Refuse Disposal
Incinerators
Open Burning
so2
Tons/Year
7272
12425
4720
716
500
216
Percent
of Total
29
50
19
3
2
1
Particulates
Tons/Year
8784
3041
1540
3456
0
3456
Percent
of Total
52
18
9
21
0
21
-------
2.2.3.2 Rollback Strategy
As discussed in Section 2.1.2, the application of a Rollback Strategy
requires a knowledge of existing air quality and a definition of an air
quality standard. Under normal circumstances, the maximum concentration
in the Rollback reduction formula is a measured value, since Rollback is
used primarily in the absence of diffusion modeling. However, this study
attempts to maintain strict consistency between the three strategy types
by keeping all input data within the same model framework. Thus,
THE "MAXIMUM CONCENTRATION" USED IN THE ROLLBACK EQUATION
IS DEFINED AS THE HIGHEST COMPUTED VALUE OF EXISTING GROUND LEVEL
CONCENTRATION FOUND IN THE DIFFUSION MODEL RECEPTOR SYSTEM.
For this study, these values are:
S02 - 144 yg/m3
~ Maximum concentrations
Particulates - 171 yg/m
Also, the air quality standard used for both Rollback and Least Cost
Strategies will be that attained by the Conventional Source Category
Strategy, thus allowing a strict comparison to be made between the three.
These standards are:
S02 - 64 yg/m3 j
f Air Quality Standards
Particulates - 96 yg/m '
Finally, background concentration levels are as follows:
S02 - 37 yg/m3 [
r Background concentrations
Particulates - 62 yg/m '
-------
Applying these values to the Rollback equation, one finds that:
R (S02) = .75
R (particulates) = ,69
In other words, total regional S02 emissions must be cut by 75 percent,
and particulate emissions by 69 percent.
According to Table 2-2, 52 percent of particulate area source emis-
sions are caused by transportation vehicles, which cannot be controlled by
"stationary source" emission standards. Thus, 69 percent control of par-
ticulate area source emissions is not possible using such standards. Using
the same set of emission standards on both point and area sources, a 69
percent reduction in total regional emissions is attained by applying stan-
dards of sufficient stringency to attain a 77 percent reduction in point
source particulate emissions; the same standards will achieve an approxi-
mately 40 percent reduction in area source emissions (which comprise
slightly more than 20 percent of total regional particulate emissions).
Thus, a 75 percent reduction in S02 and a 77 percent reduction in particu-
late point source emissions is required because of the inability to control
a portion of the area source emissions.
As discussed in Section 2.1.2, the strategy selected to accomplish
this reduction consists of potential emission, heat input and equivalent
sulfur content standards. For the sake of simplicity and ease of calcula-
tion, the potential emissions curves selected are straight lines passing
through the origin (compare to Figure 2-6) and the heat input curve is a
straight line parallel to the "heat input" axis (compare to Figure 2-4)...
in other words, the curves are specified by constant values of (allowable
-------
emissions, Ib/hr/potential emissions, Ib/hr) and (allowable emissions,
Ib/heat input, 10 BTU). A sample derivation of an emission standard is
as follows:
Particulate fuel combustion standard: heat input standard
Total heat input = 5.5321 x 10 BTU/hr.
Existing emissions 27290 pounds/hr.
Emissions after 75 percent reduction = 6822 pounds/hr.
__. . , , 6822 pounds/hr allowable emission
Emissions standard = - c - r-r -
5.5321*10 BTU/hr heat input
- .123 pounds allowable emissions=
106 BTU
Following this procedure for all of the emission standards results
in a Rollback Strategy as follows:
Particulate fuel 6
combustion sources ........ - .123 pounds/10 BTU
Particulate industrial / »n ui * j iv/v \
, .,., / Allowable emissions, Ib/hr \
process sources ........... - .1171=- r~i - 3 - : - itTTtT" I
v ions, Ib/hr j
Particulate solid waste /MI ui j j iu/u \
/Allowable emissions, Ib/hr \
1 ' I
yPotential emissions,
yPotential emissions,Ib/hr /
SO- fuel combustion sources - .8% sulfur coal or equivalent
S02 industrial process / \
I Allowable emissions,Ib/hr \
sources '" ypotential emissions,Ib/hr)
Each emission standard is designed to control 77 percent (first three stan-
dards) or 75 percent (last two) of emissions in its category. Because it is
difficult to calculate exactly how effective (in terms of emission reduc-
tions) each standard will be, a second particulate strategy with less
severe standards was also run. However, the first strategy was very
successful, yielding a total emission reduction of 69 percent.
-------
Table 2-3. Input Data for Sources Controlled Under the Least-Cost Strategy
Control Cost Data
Source
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Standard Industrial
Classification
2011; Meat Packing, Boiler
2041; Feed and Grain Mill
2041; Feed and Grain Mill
2041; Feed and Grain Mill
2041; Feed and Grain Mill
2046; Wet Corn Milling, Boiler
2082; Brewery, Boiler
2082; Brewery, Boiler
2600; Paper Products, Boiler
2800; Chemical PH., Boiler
2816; Inorg. Pigments, Boiler
2819; Inorg. Ind. Chem Plant
2819; Inorg. Ind. Chem Pit., Boiler
2911; Petro. Refinery
2911; Petro. Refinery
2952; Asphalt Batch., Boiler
3241; Cement Plant, Dry Process
3241; Cement Plant, Dry Process
4911; Powerplant
4911; Powerplant
4911; Powerplant
4911; Powerplant
4911; Powerplant
4911; Powerplant
4911; Powerplant
4911; Powerplant
4911; Powerplant
Site
No.
001
001
009
010
012
001
001
002
002
002
002
003
007
001
002
001
001
003
001
002
003
004
005
006
008
009
010
Existing
Emission
Rate(T/D)
6.25
5.70
11.37
17.15
5.09
4.21
2.95
2.67
21.22
3.42
7.30
6.00
10.70
6.00
4.72
2.90
3.28
3.68
3.72
7.60
5.00
5.10
11.90
80.00
6.90
32.50
5.60
First Node
Cost,
$/Ton
16.
16.
11.
15.
341.
19.
13.
600.
4.
34.
63.
32.
4.
128.
58.
15.
2.
118.
214.
251.
86.
909.
75.
5.
104.
39.
240.
Emission
Reduction
75.
80.
75.
75.
52.
75.
75.
76.
75.
75.
52.
68.
75.
75.
52.
75.
75.
97.
93.
63.
66.
74.
81.
75.
'75.
75.
93.
Second Node
Cost,
$/Ton
30.
24.
53.
79.
1048.
38.
20.
929.
8.
310.
71.
57.
11.
355.
65.
91.
4.
125.
270.
273.
115.
1353.
98.
8.
799.
53.
305.
Emission
Reduction
99.
99.
99.
99
j y
99.
99.
99
.7 J
97.1
99.
99.
99.
99.
99.
99.
99.
99.7
99
J ?
99
J s
99
? y m
99.
92 4
./ A. *~t
99.
99.
89 7
\j y * £.
99.
99
-/..'
-------
2.2.3.3 Least-Cost Control Strategy
In Table 2-3, input data for the 27 major point sources of
particulates which were controlled under the least-cost strategy are
described. Emissions from the remaining point sources and all area
sources were allowed to remain at the existing levels. The first column
gives the source number which is listed here for convenience in later
reference. The second column is an indication of the type of point
source being controlled, and both the Standard Industrial Classification
code (SIC) and the descriptive name of the point sources are given. The
site number in column three allows the entries in this table to be cor-
related with those in the St. Louis emission inventory which has been used
throughout the studies described in this report. The existing particulate
emission rate (tons per day) is given in the fourth column; all point
sources over 2.5 tons per day were selected from the basic St. Louis
emission inventory. It is interesting to note that these 27 sources con-
tribute 79% of the total particulate emissions from point sources in the
region. The last four columns in this figure represent data taken from
the IPP Control Cost Model. The data defines two of the three "nodes"
(end points) of the piecewise linear curves of control efficiency versus
annual cost; the third "node" is the origin. The sketch below shows a
representative curve.
Control
Cost
($)
Second Node
First Node
Control Efficiency
-------
The emission reduction at the second node represents the maximum possible
emission reduction at that particular source, given the state-of-the-art
of control technology. This data was obtained by examining the IPP Control
Cost Model output, plotting points on the cost versus control efficiency
curve for each of the applicable control devices, and then drawing that
two-segment piece-wise linear curve which best represented a lower envelope
for these points.
The receptors selected for use in the Least-Cost Model (Figure 2-7)
are a subset of those used for the rollback and conventional source cate-
gory strategies. This was done so that the source contribution tape written
by the diffusion model could be used for all three strategies. The air
quality standard achieved in the conventional strategy, i.e., the maximum
ground level concentration at any receptor in the region, was 96 micrograms
per cubic meter. The maximum for the nine receptors selected for use in
the Least-Cost Model was 85 micrograms per cubic meter, i.e., none of the
nine receptors used here reached the 96 yg/m3 maximum for the region. To
achieve results which would allow a direct comparison between the least-cost
and the conventional approaches, the linear programming algorithm used in
the Least-Cost Model was constrained to achieve the same air quality as was
achieved by the conventional source category over this set of nine receptors
-------
Figure 2-7. Receptor Locations For The Least-Cost Model
-------
2.3 RESULTS
2.3.1 The St. Louis AQCR Today
According to the emission source inventory used in this study, the
St. Louis AQCR has total emissions of 465.9 Tons/Day of particulates and
1685.4 Tons/Day of S0_. Table 2-4 illustrates the breakdown of point
and area source emissions. Present air quality, as noted in Section
3 3
2.2.3.2, is 144 yg/m of SO- and 171 yg/m of particulates at the worst
receptor (as computed by the diffusion model). According -to the plots
of ground level concentrations (isopleths) in Figures 2-^8 and 2-<-9 ,
however, the level of particulate air quality, on the average, is far
worse than that of SO ... in the core area, particulate ground level con-
centration is an average of about 40 to 50 yg/m higher than the SO. con-
centration.
-------
Table 2-4. St. Louis AQCR - Existing Conditions
Existing Point Source Emissions 1639.2
Existing Area Source Emissions 46.2
Total Existing Emissions 1685.4
Air Quality (Highest Concentra- 144.2
tration)
Average Concentration At 20 85.9
Highest Receptors
Particulates
358.8 Tons/Day
107.1 Tons/Day
465.9 Tons/Day
170.6 yg/m3
138.2 Mg/m3
-------
SO ~ _1_ yg/nf
100
Figure 2-8.
St. Louis AQCR Existing S0? Ground Level Concentrations
-------
Particulate~ I yg/m"
100
Figure 2-9.
St. Louis AQCR Existing Particulate Ground Level Concentrations
-------
2.3.2 Conventional Source Category Strategy
The Conventional Source Category Strategy described in Section
2.2.2.1 achieves a 69 percent reduction in both SO and particulate emis-
sions. Table 2-5 illustrates the breakdown of point and area source re-
ductions, plus other significant strategy results. Figures 2-^10 and 2-11
present the isopleths for SO and particulate ground level concentrations.
A comparison between the pre-control and post-control isopleths indicates
that emission reduction appears to be fairly uniform throughout the region,
since the isopleths are fairly similar in shape ... though of course the
concentration values are considerably lower in the latter two figures.
The fact that this relatively strict set of emission standards did
3
not lower maximum particulate air quality levels to below 90 yg/m should
not be greeted with dismay. It should be noted that the particulate
"background" of 62 yg/m means that a total shutdown of every particulate
emission source in the entire AQCR will lower computed air quality to
3
62 yg/m and no lower. A background concentration level of this magnitude
is almost certainly due to a poor model calibration resulting from inade-
quate input air quality data, especially in the "clean air" portions of
the region. The discrepancy between the collection dates for the air qua-
lity and emission data - 1968 and 1970, respectively - undoubtedly plays a
significant role in this probable error.
-------
Table 2-5
St. Louis AQCR - Results of Conventional Source Category Strategy
Post-Control Point Source Emissions
Post-Control Area Source Emissions
Total Post-Control Emissions
Air Quality Achieved (Highest
Concentration)
Average Concentration at 20 Highest
Receptors
Average Reduction in Concentration
Total Cost of Strategy
Cost-Effectiveness*
so2
501.6
24.5
526.1
63.7
52.5
9.5
17,949,000
1,880,500
Particulates
81.1 Tons/Day
64.2 Tons/Day
145.3 Tons/Day
95.6 yg/m3
87.4 yg/m3
18.7 yg/m3
$10,371,000/Year
$551, 500/ (yg/m3)
* Cost effectiveness is total cost
divided by average reduction in
-------
SO ~_±_ yg/nf
100
Figure 2-10.
St. Louis AQCR - SO Ground Level
Concentrations After Imposition Of
Conventional Source Category Strategy
-------
Partlculate
100
Ug/m"
Figure 2-11.
St. Louis AQCR - Particulate Ground Level
Concentrations After Imposition of Con-
ventional Source Category Strategy
-------
2.3.3 Rollback
The Rollback Strategies defined in Section 2.2.3,2 achieve a 74
percent reduction in SO- emissions and a 69 percent reduction in particu-
late emissions. Table 2-6 illustrates the breakdown of point and area
source emission reductions, air quality achieved, and other strategy
results. Figures 2-12 and 2-13 present the isopleths for S02 and parti-
culate ground level concentrations. The Rollback and Conventional Source
Category isopleths are very similar, although the Rollback SO strategy
shows greater reductions in concentration levels.
-------
Table 2^6. St. Louis AQCR - Results of Rollback Strategy
Post-control point source
emissions
Post-control area source
emissions
Total post-control
emissions
Air quality achieved
(highest concentration)
Average concentration at
20 highest receptors
Average reduction in
concentration
Total cost of strategy
Cost-effectiveness
SO,
421.6
21.7
443.3
60.3
48.9
10.5
28,462,000
2,704,200
Particulates
78.9 tons/day
64.2 tons/day
143.1 tons/day
95.3 Mg/m3
86.2
18.8 pg/m
$9,551,000/year
$ 509,200/(yg/m3)
-------
Particulates~ 1 yg/m"
100
Figure 2-12. St. Louis AQCR Particulate Ground Level
Concentrations After Imposition of
Rollback Strategy
-------
SO ~
100
Figure 2^13. St. Louis AQCR - SO Ground Level Concentrations
After Imposition of Rollback Strategy
-------
2.3.4 Cost/Benefit Comparison of Rollback and Conventional Strategies-
Figures 2-14 and 2-15 depict the costs and benefits of the Rollback
and Conventional Source Category Strategies for S02 and participates,
respectively. The optimum air quality, that point where marginal costs are
3
equal to marginal benefits, is at the 48 yg/m air quality level (where
"air quality" is weighted with respect to population). For SO , the
actual weighted air qualities achieved by the SCL strategies are:
3
Existing 65.2 yg/m
Rollback 44.1 yg/m
3
Conventional 45.9 yg/m
The particulate cost curve is drawn somewhat arbitrarily, because two of
the three data points were almost on top of each other. The shape of the
3
curve was determined by assuming an asymptote at the 62 yg/m particulate
background. The optimum air quality is at approximately 82 yg/m3. The
weighted air qualities achieved by the strategies are:
3
Existing 112.01 yg/m
Rollback 78.47 yg/m3
3
Conventional 78.99 yg/m
2.3.5 Least-Cost Control Strategy
Table 2-7 presents the effects of the Least-Cost Strategy on the 27
major particulate point sources. The source numbers correspond to those in
Table 2-3. Most of the 27 sources are controlled to the maximum extent;
those sources which are not have been marked with an asterisk. A comparison
with Table 2-5 reveals that the least-cost approach achieves an emission
reduction which is 27 tons/day less than that of the conventional approach
(81 tons reduction for least-cost versus 108 tons reduction for conventional)
-------
Thus, the least-cost approach is able to achieve the same air quality stan-
dard as the conventional approach while allowing total emissions to be 25
percent greater. The total regional cost of the least-cost strategy is 6
million dollars, which is a little more than half that incurred in applying
the conventional strategy.
The air quality impact of the least-cost strategy is shown in Table
2-8. Ground level concentrations for the least-cost and conventional
strategies are compared at each of the nine receptors. An indication of
the cost which would be incurred in lowering the air quality standard is
given by the marginal cost presented in column four. Since only two of
the receptors have air quality which is equal to the existing standard,
they are the only receptors for which cost would be incurred to lower the
standard, The largest marginal cost would be incurred at receptor number 5
where an expenditure of 2.8 million dollars would be required for a 1
yg/m3 reduction in the air quality standard.
-------
oo
30 -
X
CO-
20
CO
H
CO
O
10 -
O
O
1 CONVENTIONAL
2 ROLLBACK
BENEFITS
70
OPTIMUM AIR QUALITY
50
AIR QUALITY WEIGHTED WITH RESPECT TO POPULATION, pg/m~
-------
Ln
vo
vD
O
W
a
CQ
CO
H
CO
O
§
H
Z
O
u
60 J
50 J
40 J
30 J
20 H
10-4
1 CONVENTIONAL
2 ROLLBACK
BENEFITS
OPTIMUM AIR QUALITY
110
100
60
AIR QUALITY WEIGHTED WITH RESPECT TO POPULATION,yg/nf
-------
Table 2-7. Least-Cost Strategy Impact on Controlled Particulate Sources
Power-
plants
Source
No.
1
2
3
4
5*
6
7
8
9
10
11
12
13
14*
15
16
17
18*
19*
20*
21
22*
23*
24
25*
26
27*
Existing
Emission Rate(T/D)
6.25
5.70
11.37
17.15
5.09
4.21
2.95
2.67
21.22
,42
30
00
10.70
6.00
4
2
3
3
3.72
7
5
72
90
28
68
,60
,00
5.10
11.90
80.00
6.90
32.50
5.60
Controlled
Emission Rate(T/D)
.06
.06
.11
.17
2.83
.04
.03
.08
.21
.03
.07
.06
.11
1.50
.05
.01
.03
3.68
3.72
7.60
.05
1.54
2.26
.80
1.73
.32
5.60
Control
Level
99.0
99.0
99.0
99.0
44.5
99.0
99.0
97.1
99.0
99.0
99.0
99.0
99.0
75.0
99.0
99.7
99.0
0.
0.
0.
99.0
69.7
81.0
99.0
75.0
99.0
0.
Annual ,
Cost($xlO )
.07
.05
.22
.49
.28
.06
.01
.88
.06
.38
.19
.12
.04
.21
.11
.10
.01
0.
0.
0.
.21
1.18
.26
.23
.20
.62
0.
Rate($/T)
73.8
57.7
184.2
-279.0
341.0
97.4
41.9
2114.0
20.5
1172.5
79.8
111.8
32.9
1064.4
72.7
321.8
10.2
118.0
214.0
251.0
173.0
909.0
201.5
17.4
4469.8
96.7
240.0
282.93
32.75
-------
Table 2-8. Least-Cost Strategy Impact on Air Quality
Ground Level
Concentration
After
Ground Level
Concentration
After
Marginal Cost
Receptor
Number
1
2
3
4
5
6
7
8
9
Least-Cost
Control, pg/m3
69.
75.
76.
71.
85.
85.
79.
80.
72.
5
5
6
5
5
9
Conventional
Strategy, ug/m3
67.
72.
73.
70.
85.
77.
68.
80.
73.
8
0
4
0
0
7
2
0
0
To Reduce AQ
Standard ($x!06)
0.
0.
0.
0.
2.
0.
0.
0.
79
61
-------
2.4 CONCLUSIONS
2.4.1 Rollback Effectiveness
The discussion in Section 2.1.2 indicates that, although a Rollback
control strategy is designed to meet a specific air quality standard,
actually there is no guarantee beforehand that it will do so. The SO
and particulate Rollback strategies used in this study (defined in
Section 2.2.3.2) were successful in meeting the specified emission
reduction almost exactly:
S0~: R = .75 vs. actual reduction = 74 percent
Particulates: R = .69 vs. actual reduction = 69 percent
2
The air quality achieved by the particulate strategy was within 1 pg/m
of the target; on the other hand, the S02 Rollback strategy overshot its
3
target by 3 Mg/m , a small error by diffusion modeling standards. This
small additional increment of air quality improvement is purchased at an
additional cost of 10.5 million dollars per year, which represents a greater
than 50 percent increase over the cost of the conventional source category
strategy.
Although the Rollback strategy is defined in relationship to only
one concentration value in the region.... the maximum value calculated
at a receptor (or, in real-life practice, the maximum measured at a
pollution measuring station), a true comparison of the Rollback and
conventional source category strategies should include more than a com-
parison of the total cost and "air quality" (maximum concentration)
achieved. A look at the isopleths generated by the diffusion model
-------
indicates that there is often little relationship between the "air quality"
as measured at one point and the "air quality" as indicated by the contour
lines extending throughout the region. A better measure of the success of
a strategy might well include the average ground level concentration
throughout the region or else (if attention is to be kept on the worst
parts of the region from an air quality standpoint) the average concentra-
tion at some number of the "worst" receptors. If the impact of the air
quality on a per capita basis is desired, the concentrations at the
receptors can be weighted by the population in zones of influence around
the receptors; cost/benefit results using this procedure provide yet
another measure of "air quality."
Comparing the particulate strategies on these bases, the average
3
concentration of the Rollback strategy is .1 yg/m lower than that of
the source category strategy, at a cost which is $820,000/year less.
3
The average concentration at the 20 "worst" receptors is 86.2 yg/m for
3
Rollback versus 87.4 yg/m for the source category strategy. The weighted
(with respect to population) air qualities attained by the two strategies
are nearly identical (Rollback = 78.47 yg/m , conventional = 78.99
3 3
yg/m ), and are quite near the 82 yg/m optimum (marginal costs equal
marginal benefits; see Figure 2-15). It can be concluded from all of the
above measures that the Rollback particulate strategy as applied here has
been a remarkable success, achieving almost precistly its target air
quality in a comparatively efficient manner.
-------
Continuing in the same manner, the "area^wide average," "average
at the 20 worst receptors," and "weighted (with respect to population)
average" concentrations produced by the Rollback SCL strategy are lower
than those of the conventional source category strategy by 1.0, 3.6, and
3
1.8 yg/rn respectively. Thus, the great additional cost of the Rollback
SO strategy is buying a small 3 to 4 yg/m improvement in the core area
and essentially no improvements in the suburbs. A comparison of the cost
effectiveness of the two strategies bears out the impression that the
Rollback strategy is considerably less efficient than the conventional
source category strategy:
3
Rollback cost-effectiveness = $2,704,200/(Mg/m )
3
Source category cost-effectiveness = $l,880,000/(Mg/m )
Going back to Figure 2-14, it can be seen that the Rollback strategy is
in a very unfavorable position on the cost curve; even though the
strategy's weighted air quality is within 4 pg/m of the optimum
air quality, the steep slope of the curve ensures considerable additional
cost for the small air quality improvement.
In conclusion, while the Rollback strategy investigated in this
study has achieved excellent success with respect to attaining the
particulate air quality standard, it has forced an expensive over-control
in attempting to attain the SO^ air quality standard.
2.4.2 Uniform Application of Emission Standards Versus Least-Cost
Strategy
The conventional source category strategy represents a uniform
application of emission standards which disregards an emission source's
location, ignoring the relative importance of its contribution to total
ground level concentrations. Thus, plants in relatively "clean" areas
-------
are controlled to the same severity as identical plants in high
pollutant concentration areas. The Least-Cost Strategy discussed in
Sections 2.1.3, 2.2.3.3, and 2.3.5 achieves a minimum cost of control by
applying severe controls to those plants contributing the most to air
quality violations, and applying more lenient controls to plants in clean
air areas. The strategy was applied to particulate control only.
The cost of the Least-Cost Strategy is 6 million dollars per year,
versus 10.4 million dollars per year for the conventional source category
strategy. Thus, the Least-Cost strategy achieves an air quality (as
measured by the 9 receptors in Figure 2-7) identical to that attained by
the conventional strategy at a cost which is 42 percent less. An inspection
of Table 2-8, however, indicates that the Least-Cost Strategy gives a
flatter plateau of air quality than the conventional strategy gives; that
is, in most cases the ground-level concentrations under the least-cost
approach will be equal to or greater than those which exist after the
application of the conventional strategy. Thus, it may be expected that
the "benefits" as measured by a cost/benefit model will be greatest for
the conventional strategy; it remains to be seen whether the additional
benefits outweigh the lesser cost of the Least-Cost strategy.
2.5 RECOMMENDATIONS
2.5.1 Rollback and Air Quality
The dependence of "air quality" on the concentration measured at
a single* maximum receptor ignores the extreme sensitivity of ground level
pollutant concentration measurements to small variations in receptor or
* I.e., if a region has 5 receptors measuring 64, 72, 41, 110, and 79
yg/m3, the region is said to satisfy an air quality standard of 110
Mg/m3.
-------
measuring station location. The differences in maximum concentration
levels detected (see Section 3.3.3) in the diffusion models run - for the
control strategy comparison in this section and the land use model in
the following one - emphasizes the fact that altering the diffusion model
receptor grid can seriously alter the maximum concentration detected while
leaving the isopleth patterns relatively unchanged.
Although the overall air quality of a region is sometimes thought
of in terms of a smooth contour surface that can adequately be described
by isopleths drawn using a uniform grid of receptors, in fact the sur-
face of the contour is often broken by spikes of high concentration at
points near a few very large sources. Isopleths will not display these
spikes unless a substantial number of receptors have deliberately been
placed near these sources. If the modeler is unlucky, small changes in
receptor location will cause a receptor to move from outside to inside
the spike, or vice versa, causing a considerable distortion in the
model results. The same is, of course, also true with respect to the
location of pollution measuring stations.
The importance of the Rollback technique, which is dependent on a
single maximum concentration for determining the required reduction in
regional emissions, is one reason why this spiking problem is worth
-------
further investigation. Another is the interpretation of an "air
quality standard" which demands that every point in the region be below
a given concentration level.
Under the present Rollback definition, the stringency of the
emission standards required will be dependent on the precise location of
the pollutant measuring stations. Furthermore, since it is normal to
discard some percentage of the measuring station data as inadequate, a
considerable opportunity exists for some judicious juggling of results.
When modeling is used, and emission reduction is no longer de-
pendent upon the Rollback reduction factor, the emission standards are
still subject to this locational sensitivity. Most control strategies
are designed to insure that every receptor in the region measures a
concentration less than the air quality standard. One receptor grid
may result in considerably different control requirements than another,
because of the possibility of sliding into or out of a concentration
spike.
As a partial solution to this problem, precise guidelines should
be formulated and issued on such matters as the location of pollutant
measuring stations, required receptor spacing around major sources, and
acceptance/rejection of measured data. If a requirement for sharply
decreased receptor spacing around potential spikes is formulated, then
the question of the meaning of "satisfying an air quality standard"
should be reopened....since such a receptor pattern will degrade a
region's calculated air quality.
Finally, the reliance of Rollback on the concentration at a single
-------
point should be reconsidered in the light of the sensitivity of the
measured "maximum concentration" to location of the measuring stations.
In those cases where the number of stations is ample enough to permit
it, an averaging technique employing a percentage of the highest
measured concentrations might be considered. To compensate for the
lower "maximum" this will produce, the reduction factor R might be
calculated in a more severe fashion.
2.5.2 Least-Cost Control
The potential cost saving of a least-cost air pollution control
strategy which is implied by this study is substantial enough to justify
further investigation of this means of control. The present study has
the following shortcomings:
The "conventional source-category strategy" used for
comparison with the least-cost strategy does not
represent the lowest-cost "uniformly applied" strategy
able to achieve the selected air quality standard.
The study covered particulate control only.
Area source control costs incurred by the conventional
strategy were not measured (the least-cost strategy
incurred zero area source control costs).
The least-cost strategy does not include control of all
point and area sources and thus does not necessarily
represent an absolute minimum cost solution.
The receptor grid used by the least-cost model did not
include a number of high concentration locations which,
if included, could theoretically alter the results.
-------
Control cost versus percentage of control was
represented by a piece-wise linear function using
only two line segments.
Several of the objections could be removed by increasing the
capacity of the Linear Programming Model to include more receptors and
sources and a better representation of control costs. Area source control
costs could be approximated using available data. IPP could be used to
search for an (approximately) least-cost version of a source-category
strategy. Finally, since the controlled emissions of the industrial
plants are known (they are calculated by the linear programming model), the
ability to scale point source emissions in IPP would allow the generation
of an "after least-cost control" diffusion model run which could be
directly compared to competing strategies.
If sufficient interest is generated in a least-cost solution to the
control of air quality, it is recommended that the above steps be taken
to clarify the advantages and disadvantages of this method of control
relative to the more common uniform application of emission standards.
-------
3.0 EFFECTS OF CHANGING LAND USE
3.1 DISCUSSION
The purpose of this section is to present a method by which a
diffusion model can be used as a tool for predicting the air quality
effects of changing land use. Use of a diffusion model under standard
operating procedures would normally be too expensive to justify the
continuous re-running of the model to investigate many alternate land
use changes. However, the diffusion model can be run with different
possibilities of land use development "built into" the same run; this mode
of operation is discussed in Sections 1.2 and 3.2.
The value of such a procedure is unquestionable. Many urban areas
are beginning to ask whether good air quality and continued industrial
and residential development, at increasing intensities, are compatible.
Strict emission standards are currently being enforced or are being
formulated which will reduce pollutant concentrations below levels
specified by law. However, continued growth will cause air quality to
degrade unless land use patterns are strictly controlled and the con-
sequences of growth are thoroughly understood.
The method of air quality prediction presented here has all the
problems of the standard diffusion model - lack of adequate source data,
poor calibration due to measuring station inaccuracies, etc....plus a
few of its own, such as degraded calibration due to the establishment
of sources in new locations and the inability to exactly simulate
effective stack height (except in a few restricted applications). It
is felt, however, that the results will be useful enough to the land
use planner and air pollution control agency to warrant further study.
-------
The "Land Use Prediction Model" is applied to 10 scenarios in the
St. Louis area defined in Figure 3-1 (this area is somewhat smaller* than
the St. Louis Air Quality Control Region, and consists of that area
within the "cordon line" of the East-West Gateway Transportation Study).
The first 7 scenarios depict the addition of a major new powerplant to
the region at 7 alternate sites. The remaining scenarios depict a
dispersal of sources from the central area to the periphery of the region.
* However, the emission source file used in the analysis is the same
as that used for the three strategies discussed in Section 2.0, and
therefore the diffusion model can include areas up to that enclosed
by the Air Quality Control Region.
-------
ST. LOUIS METROPOLITAN AREA TRANSPORTATION ZONES
Figure 3-1 St.Louis Study Area
-------
3.2 METHODOLOGY
3.2.1 Review of Modeling Procedure
The procedure for utilizing the diffusion model as a predictive tool
is an extremely simple one. As discussed in Section 1.2, the diffusion
model is run with a source file consisting of all existing point and area
sources plus an assortment of "dummy" point and area sources which have
emissions on the order of .001 tons per day or less. These dummy sources
should be placed in locations where growth is postulated (according to
growth projections, plant relocation schemes, land use plans, etc.). By
using the Implementation Planning Program Strategy Model (in a "Null Stan-
dard" mode with point and area source scaling), or else a program specifi-
cally designed for the purpose, the diffusion model results may be manipu-
lated so as to scale any source's emissions up or down. Sources can thus
be "created" by scaling a dummy source up to a significant emission level,
or "destroyed" by scaling downwards. Each dummy source has a specified
"Effective Stack Height" (real stack height plus plume rise) assigned to
it; any potential growth location can accommodate a range of potential
plant types by placing several dummy sources one on top of the other, each
with a different height. One source at a time can be "activated" to
duplicate a large powerplant, medium sized industrial plant, etc. A real
source can then be "relocated" to a new position by scaling it down to
zero emissions and scaling up the appropriate dummy to duplicate its
original emissions.
-------
3.2.2 Basis for Procedure
The atmospheric diffusion model used in this study is based upon a
diffusion model developed by Martin and Tikvart (1968). The basic output
of the model is in the form of calculated long term average pollutant con-
centrations at ground level. The model calculates concentrations downwind
from a set of point and area sources on the basis of the Pasquill (1962)
point source formulation; the plume rise equation used in the model is due
to Holland (1953).
The uncalibrated diffusion model has multiplicative and additive
properties which allow its use in this analysis:
3.2.2.1 Multiplicative Property of Diffusion Model
If one assumes that all stack parameters - stack diameter and height,
gas temperature and velocity - are kept constant
If a source Q at location L with emissions E produces an increment
of ground level concentration C. at the i receptor (in other words, if
Q is the only source in the region, receptor i will measure a ground level
concentration of C. ),
Then if source Q increases in size so that it has emissions N*E, it
will produce an increment of ground level concentration N*C.
In other words,
SCALING THE EMISSIONS OF A SOURCE UP OR DOWN SCALES THE
RECEPTOR CONTRIBUTIONS OF THAT SOURCE BY THE SAME FACTOR.
-------
3.2.2.2 Additive Property of Diffusion Model
If a diffusion model is run twice-keeping the receptor grid the same-
for Nl sources the first time, and for N2 sources the second.....
And assuming that the ground level concentrations computed at the
i receptor are C.(l) and C.(2), respectively
Then the diffusion model run for the (Nl + N2) sources will yield
concentrations at the i receptor of
. C±(2)
The additive character of the diffusion model allows any portion of
the emission source file which the modeler wishes to leave unchanged to be
run separately from the "variable" portion of the file, i.e., that portion
which will be scaled. Thus, the program which accomplishes the scaling
need not handle those sources which remain constant. Under certain circum-
stances,* this separation will considerably shorten the running time of the
secondary scaling program.
3.2.3 Further Details of Procedure
Figure 3-2 presents a flow diagram of the modeling procedure used
for this study. The additive property of the diffusion model, as described
above, is used to create a matrix of "base" concentrations and one of
"variable" concentrations which can be added to produce the actual air
quality matrix resulting from a given strategy being imposed on the
"variable" emission sources. The calibration of the concentrations is
conducted after addition of the two matrices.
*For instance, when the IPP "Strategy Model" is used for scaling.
-------
ORIGINAL
EMISSION
SOURCE
FILE
DUMMY
SOURCES PLUS
/
/ VARIABLE
/SOURCES IN
ORIGINAL
7
DIFFUSION
MODEL
DIFFUSION
EXISTING7
'GROUND LEVEL/A
CONCENTRA- /-^
TIONS
/
:ONCENTRA-
TIONS CON-
RIBUTED BY
"VARIABLE
FILE"
SUBTRACT
B FROM A
SCALING RUNS
USING IPP
STRATEGY
MODEL
/CONCENTRA-
/TION CON-
"CONSTANT"
PORTION OF
FILE
/CONCENTRA-
'TION FROM
'VARIABLE
FILE" AFTER
LAND USE
SHIFT
QUALITY
'AFTER LAND
USE SHIFT
ADD C & D,
CALIBRATE
RESULTS
-------
The reason for separating the source file in this manner is that a
considerable portion of the point and area sources in the original file may
be considered as too small to be included as variables in a (necessarily
coarse) land use projection study, or else may have stabilized to a suffi-
cient degree so that they may be considered constant during a modest time
increment. It is convenient to remove these sources from the scaling runs
so as to minimize the storage needs of the scaling program and to reduce
run time.
3.2.4 Construction of the Emission Source File
The basic emission source file used in this study is the same as that
used for the Emission Control Strategies Study discussed previously
the St. Louis AQCR emission inventory provided by the Office of Air Programs
and subsequently modified by TRW. To it have been added 55 dummy sources...
43 point sources and 12 area sources at 18 locations.
In order to decide where to place the dummy sources, socio-economic
data for 1966 and projections for 1990 (supplied by the East-West Gateway
Coordinating Council, St. Louis, Illinois) on industrial employment and
residential population were investigated. For the most part, dummy sources
were placed in sparsely populated areas and/or areas where substantial new
industrial growth was expected to occur obviously, if scenarios differ-
ent from those investigated had been used, different locations might have
been chosen.
Table 31 lists the dummy sources added to the file. A computation
of the effective stack heights of every large point source in the AQCR indi-
cated that the large majority of sources could be accommodated by dummy
-------
Table 3-1 . Dummy Source File
No.
Location
LARGE POWER PLANTS
1
2
3
4
5
6
7
(154,210)
(155,245)
(151.5,216)
(113,215)
(115,237)
(152,199)
(179.5,238)
Characteristic of Location
Growing Industrial Area
Indus trial/ Commercial
High Density Industrial
Future Industrial Area
Farmland
Farmland
Farmland
INDUSTRIAL PARKS (5 KM2 AREA SOURCES)
8
9
10
11
12
13
14
15
16
17
18
19
(110,224)
(110,224)
(152,199)
(152,199)
(115,237)
(115,237)
(172,249)
(172,249)
(171,198)
(171,198)
(164,241)
(164,241)
^Underdeveloped )
/Underdeveloped \
(Farmland
( Farmland
(Farmland |
(Farmland )
(Farmland )
(Farmland \
(Farmland )
(Farmland f
Farmland , Near /
Secondary Industrial Center \
INDUSTRIAL POINT SOURCES
20
21
22
23
24
25
26
(110,224)
(110,224)
(110,224)
(152,199)
(152,199)
(115,237)
(115,237)
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Effective
Stack Height*
_**
-
-
-
-
-
-
75 meters
140 m
75 m
140 m
75 m
140 m
75 m
140 m
75 m
140 m
75 m
140 m
75 m
140 m
280 m
75 m
140 m
75 m
140 m
(contd. )
* Based on plume rise due to Holland (1953) .
**Actual stack parameters were used in file.
-------
Table 3-1. Dummy Source File (contd.)
No.
Location
Characteristic of Location
INDUSTRIAL POINT SOURCES (contd.)
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
(125,244)
(125,244)
(125,244)
(172,249)
(172,249)
(179.5,238)
(179.5,238)
(171,198)
(171,198)
(171,198)
(181,204)
(181,204)
(113,215)
(113,215)
(119,214)
(119,214)
(119,214)
(116,208)
(116,208)
(154,210)
(154,210)
(150,230)
(150,230)
(130,190)
(130,190)
(138,245)
(138,245)
(164,241)
(164,241)
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
Farmland
(Undeveloped/farmland, |
/future industrial area.)
Supposed to gain
substantial industry
by 1990.
Undeveloped
Undeveloped
Industrial
Industrial
Industrial
Industrial
Largely Undeveloped
Largely Undeveloped
Farmland
Farmland
(Farmland, near secondary}
(industrial center. \
Effective
Stack Height
75 meters
140 m
280 m
75 m
140 m
75 m
140 m
75 m
140 m
280 m
75 m
140 m
75 m
140 m
75 m
140 m
280 m
75 m
140 m
75 m
140 m
75 m
140 m
75 m
140 m
75 m
140 m
75 m
140 m
-------
sources with heights of 75, 140, or 280 meters without incurring errors
of more than 20 meters. Sources 1 through 7, the new powerplants, were
modeled more precisely:
Actual stack height = 125 meters
Temperature = 450°K
Stack diameter = 8 meters
Gas velocity = 20 meters/second
After scaling, these powerplant's emissions will be 250 and 75 Tons/Day
for SO- and particulates, respectively.
Figure 3-3 shows the locations of the dummy sources.
-------
260
250
240
a 230
H
VI
o
2 22°
CO
n
OJ
o
,H
rl
^
t
>i
210 -
200
190
100
110 120 130
140
150 160
170 180
X, Kilometers from Origin
Figure 3-3. Location of Dummy Sources
-------
3.2.5 Setting Up the Diffusion Model
The selection of a receptor net for the diffusion model should be
based on the "scenarios" to be investigated, the nature of the source file
and location of the major sources, and the method by which the air qualities
resulting from the scenarios are to be compared. Ordinarily, receptor nets
are set up so that spacing between receptors is great in those regions where
few sources are located, and small in those areas of highest industrial
and/or residential acitvity. If resulting air quality matrices are to be
compared using some kind of cost/benefit model, or by a comparison of
[(people within receptor "zone")*(concentration)] factors, a receptor net
of this type is useful. However, if the comparison is to be made on a
simplified basis, perhaps by counting the number of receptors measuring in
various ranges of concentration, then a net of evenly spaced receptors
might be selected. In any case, the problem of "scoring" the results of a
diffusion model run which is essentially the same problem being addressed
by cost/benefit models of air pollution is an agonizing one, and one
which is often avoided by ignoring everything but the maximum concentration
and calling that concentration the "air quality" achieved.
The receptor net selected for this study consists of a uniformly
spaced grid whose corners are (x, y) = (100, 170), (LOO, 261), (191, 261),
(191,170) (see Figure 3-4 ). There are 196 receptors in all, spaced 7
kilometers apart. The grid was selected because it covers the area in
question reasonably effectively with a minimum number of receptors, thus
minimizing computer time. Uniform spacing was used to allow a fair
-------
260
100 110 120 130 140 150 160
X, Kilometers from Origin
Figure 3-4. Diffusion Model Receptor Net
170 180
-------
comparison of air quality results according to the number of receptors
in each concentration range, as discussed above. For the purposes of
a more sophisticated scoring system, or for the use of a few different
types of scoring, a receptor net could be constructed which combined
a uniformly spaced grid with additional receptors at selected locations.
The different scoring systems would use only those receptor measurements
which were applicable and would ignore the rest.
3.2.6 Scenarios
As discussed previously, the scenarios investigated in this
study are:
Adding a large new power plant to the region.
Dispersing industry to the suburbs.
The first seven model runs involve the placing of the new plant in
each of the "Large Powerplant" locations noted in Table 3-1, one at a time.
This is accomplished as previously described, by scaling up the appropriate
dummy point source and leaving the others unchanged. Four of the seven
areas are very sparsely populated and thus should have no present pollu-
tion problem. The remaining areas are all developed to some extent, with
one being in the core area.
The remaining three runs enact the "dispersal of industry." A number of
large point sources in the region's central area are scaled down to simu-
late their closing, while dummy sources on the outskirts of the area
are scaled up to duplicate the original sources. Table 3-2 lists the
sources involved in the "relocation." These sources include 573.89
-------
Table 3-2 . Emission Sources to be Relocated
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Location
150,223
150,223.4
156,218
152.1,214.5
150,215
161,241
161,241
152.5,214
156,225
151.2,218.6
150.6,215
151.5,222
155.5,244
155.1,244
155.1,244
SIC Code
(Process)*
2041(1)
2041(1)
2816(0)
2819(0)
2819(0)
2911(1)
2911(0)
2911(2)
3312(0)
4911(0)
4911(0)
4911(0)
4911(0)
4911(0)
4911(0)
Emissions (SO /Part)
Tons /Day
0/11.37
0/17.15
2.56/9.67
32.75/6.00
32.80/10.70
6.85/6.00
17.67/4.72
19.00/.42
1.12/7.86
46.20/11.45
40.30/4.68
115.00/36.61
62.13/80.00
45.60/6.92
142.00/32.50
Effective Stack
Height, Meters
69.8
155.2
79.5
74.7
45.8
91.6
89.5
50.3
66.6
145.4
144.3
279.8
99.9
120.0
120.3
* 2041(1) Feed and Grain Mill Products
2816(0) Inorganic Pigments-Boiler
2819(0) Inorganic Industrial Chemicals-Boiler
2911(0) Petroleum Refinery-Boiler
2911(1) Petroleum Refinery-Fluid Catalyst
2911(2) Petroleum Refinery-Moving Bed Catalyst
3312(0) Iron and Steel-Boiler
4911(0) Power Plant
-------
tons/day of SO and 246.05 tons/day of particulates (respectively about
35 percent and 50 percent of total emissions).
Runs 8 and 9 place all the sources to be relocated in industrial
groupings or parks. Referring back to table 3-1,
Run 8 relocates all sources to 3 locations:
1. Dummy sources 8, 9, and 22
2. Dummy sources 16 and 17
3. Dummy sources 14 and 15
Run 9 relocates all sources to 4 locations:
1. Dummy sources 18 and 19
2. Dummy sources 10 and 11
3. Dummy sources 12 and 13
4. Dummy source 36.
Run 10 establishes a nearly 1 to 1 relationship of original point
sources and dummy point sources. The object is a maximum dispersal of
point sources throughout the region. Table 3-^3 presents the corres-
pondence between the two source files.
3.2.7 Model Shortcomings
As briefly noted in Section 3.1, the procedure described in this
section is subject to all the inaccuracies of the diffusion model it is
based on, as well as to a few additional inaccuracies resulting from the
prediction procedure.
Some of the major deficiencies of the diffusion model are:
Accurate emission and stack data are difficult to
acquire.
The calibration procedure has only two degrees of
freedom.
-------
Table 3-3. Strategy 10 - Maximum Dispersal of Point Sources
Corresponding
Original Source No. Dummy Point Source
(from Table 3^2 ) (from Table 3-1 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
23
24
44
25
39
50
50
27
37
33
35
22
52
31
31
-------
Measured air quality data for calibration is often
of limited accuracy.
The diffusion equations do not fully explain the
physical phenomena of diffusion.
The model only accounts for average annual conditions
(short range conditions .can be predicted only
statistically).
The prediction procedure adds the following inaccuracies (some of
which may be corrected by increasing the scope of the procedure; an
asterisk indicates this possibility):
Since the calibration procedure accounts (somewhat) for
topographical features of a region (the theoretical model
assumes a flat plain), the calibration constants are tied
to the locations of the emission sources. Adding new
plant locations should change the calibration constants,
but this effect cannot be accounted for in the procedure.
The use of dummy sources introduces an error due to the
difference between the dummy's effective stack height
and the actual effective stack height of the source the
dummy is to replace (in a plant relocation scheme).
* The use of effective stack heights forces the model to
average diffusion parameters over all atmospheric stability
classes.
* Relocation of industry will certainly be accompanied by
shifts in traffic and residential development, neither of
which are accounted for in the plant relocation "scenarios."
-------
3.3 RESULTS
Three figures of merit are utilized to measure the effect of the
10 Prediction Model scenarios:
Maximum concentration measured in the region.
Average concentration of the "worst" 20 receptors.
Number of receptors in different ranges of concentration
(60-70 Mg/m3, 70-80 Mg/m3, etc.).
Although the maximum concentration is often used to denote "air quality,1
the average value should give a more meaningful idea of the true effect
of the scenarios.
3.3.1 Where to Locate a New Power Plant
Tables 3-4 and 3-5 present the results of the first seven runs of
the model, each of which represents the placing of a major new power
plant with a large effective stack height on a new site. As was
inevitable, the results uniformly show a degradation of air quality,
although the effects are quite mild.
3.3.2 Dispersal of Industry
Tables 3-6 and 3-7 present the results of the latter three runs
of the model; runs ("scenarios") 8 and 9 represent the relocation of
several centrally-located point sources to a few industrial parks on
the periphery of the region; run 10 represents a "maximum dispersal"
scenario where the same point sources are relocated separately to sites
scattered around the periphery. The results are unusual in that the
average concentration at the 20 "worst" receptors was higher, for
particulates, than was the average before dispersal.
-------
Table 3-4 . Where to Locate a New Power Plant; Air Quality Results
STRATEGY
(so2>
EXISTING
1
2
3
4
5
6
7
(Particulates)
EXISTING
1
2
3
4
5*
6
7
CONCENTRATION AT WORST
RECEPTOR, Mg/m3
85.94
86.45
86.15
86.27
86.28
86.33
86.60
86.04
163.85
164.24
164.10
164.26
164.15
164.24
163.96
AVERAGE* CONCENTRATION AT
20 WORST RECEPTORS, ^g/m3
57.63
61.80
61.66
61.78
61.64
61.68
61.77
61.52
112.09
112.43
112.28
112.41
112.32
112.42
112.22
* Simple arithmetic average of 20 highest concentrations
in strategy diffusion output.
** Incut error disqualified results.
-------
Table 3-5. Where to Locate a New Power Plant; Number of
Receptors in Different Ranges of Air Quality
so2
STRATEGY J^ "
^^-^fiANGE
^^(yg/in3)
EXISTING
1
2
3
4
5
6
7
30-40
135
132
134
132
131
132
133
133
40-50
41
44
43
45
45
43
43
43
50-60
11
11
11
11
11
11
11
11
60-70
5
5
5
4
5
5
4
5
70-80
2
2
2
3
2
2
2
2
80-90
2
2
2
2
2
2
2
2
Particulates
STRATEGY*^""'
^^-^ANGE
^-"(yg/in3)
EXISTING
1
2
3
4
5*
6
7
60-70
75
74
75
74
74
74
75
70-80
68
69
68
69
69
69
68
80-90
28
28
28
28
28
28
28
90-100
12
11
11
11
11
11
11
100-110
6
7
7
7
7
7
7
110-120
1
1
1
1
1
1
1
120-13C
2
2
2
2
2
2
2
>130
4
4
4
4
4
4
4
* Input error disqualified results.
-------
Table 3-6. Dispersal of Industry; Air Quality Results
STRATEGY
(so2)
NULL
8
9
10
(Particulates)
NULL
8
9
10
CONCENTRATION AT
WORST RECEPTOR f
ug/m
85.94
72.10
73.94
70.84
163.87
155.44
156.72
155.09
AVERAGE CONCENTRATION AT 20
WORST RECEPTORS,
jjg/m^
61.38
54.59
56.67
55.56
112.09
114.81
116.93
115.79
-------
Table 3-7. Dispersal of Industry; Number of Receptors
in Different Ranges of Air Quality
so.
2
STRATEGY #^-^
^^ow^r
EXISTING
8
9
10
30-40
135
115
123
107
40-50
41
64
54
73
50-60
11
12
11
11
60-70
5
5
6
4
70-80
2
1
2
1
80-90
2
0
0
0
Particulates
STRATEGYtf^-"'^
^^M3)E
EXISTING
8
9
10
60-70
75
34
34
41
70-80
68
71
75
57
80-90
28
49
44
59
90-100
12
24
24
20
100-110
6
7
6
9
L10-120
1
3
5
5
120-130
2
5
6
2
f>130
4
2
2
3
-------
3.3.3 Comparison of Diffusion Model Results With Those of Section 2
The use of separate diffusion model runs (with different receptor
grids) for essentially the same area presented an opportunity for an
interesting comparison. A comparison of two of the three "figures of
merit", the maximum concentration and the average at the 20 "worst"
receptors, indicates that both values differ significantly (Table 3-8).
This difference is certainly a function of the grid spacing and location;
for instance, the grid used in Section 2.0 has a greater number of
receptors in areas of high concentration, and therefore may be expected
to have a larger average concentration at the 20 "worst" receptors.
-------
Table 3-8. Comparison of the Two Diffusion Models
Section 2.0 - Emission Control Strategies
Comparison.
Section 3.0 - Land Use Model.
so.
Maximum concentration, Mg/nT
Average concentration at 20
"worst" receptors,
Particulates
Maximum concentration,
Average concentration at 20
SECTION
2.0
144.2
85.9
"worst" receptors,
170.6
138.2
SECTION
3.0
85.9
61.4
163.9
112.1
-------
3.4 CONCLUSIONS
The analyses carried out so far indicate that changes in air
quality patterns caused by the two sets of scenarios would be quite mild.
3.4.1 Where To Locate A New Powerplant
The results for the addition of a powerplant to the region suggest
that the precise location of the particular powerplant investigated may
not be extremely important from a regional air quality viewpoint. This
conclusion is undoubtedly due to the very large effective stack height
(about 400 meters), which was selected for analytical purposes as being
representative of the larger power plants in the region.
A sensitivity analysis conducted in parallel with this study
produced Figure 3-5, which shows the distribution of pollutants from a
314 meter (effective) stack height, the largest available from the
analysis. A 250 Ton/Day emission, typical of the SO- emissions of the
larger St. Louis powerplants, would have a maximum incremental effect
n
of about 5 yg/m between 5 and 10 kilometers from the source. The
effect of a 400 meter (effective) stack would be somewhat less.
Furthermore, use of the calibration coefficients of the St. Louis
diffusion model requires that this increment be multiplied by
.3 for SO. and .6 for particulates to determine the true effects of the
plant. Thus, the maximum incremental effect of the power plant used in
3 3
the analysis would be less than .9 ug/m for particulates and 1.5 Mg/m
for S0_ (daily SO- emissions = 250 Tons, daily particulate emissions =
75 Tons). In order to differentiate between alternate power plant
scenarios, the model would have to be rerun with a lower effective stack
-------
.1 _
X
.01
EFFECTIVE STACK
HEIGHT = 314 METERS
i I I I I I I
10
X, Distance from Emission Source
100
X = Maximum incremental concentration, yg/m
Q = Emission rate, Tons/Day
Figure 3-5. Diffusion of Pollutants From a Point Source
-------
height; time constraints precluded this rerun under the existing con-
tract.
Any conclusions drawn from these results about the use of tall
stacks should be tempered with the following two considerations:
Air quality results given by the modeling process
are in terms of average annual concentrations. Air
pollution incidents in the past have shown that large
amounts of pollutants emitted by tall stacks can cause
very high short-term impacts on ground level concentrations
during periods of atmospheric stagnation.
Although the long-term impact of any one tall stack
powerplant is not high at any single location, global
considerations accounting for all emission sources pre-
clude the use of tall stacks as a sole solution to air
pollution problems. The additive nature of the pollutant
contributions from each source in an area demands the use
of emission control to achieve acceptable air quality
levels.
3.4.2 Dispersal of Industry
All three industry dispersal scenarios succeeded in lowering the
maximum concentrations of both SO and particulates measured in the region.
The "maximum dispersal" scenario (#10) achieved the lowest concentrations
for both pollutants. In addition, significant reductions in the average
SO- concentrations of the 20 "worst" receptors were achieved in all three
scenarios. However, these average concentrations were higher for the
particulate scenario runs, an occurrence not easily explained by the
nature of the locational shifts made. The wide receptor spacing of this
-------
preliminary model makes it rather susceptible to the kind of error
described in Section 3.4.3, where a shift in location of the receptors
or sources can cause the receptors to slide into or out of a concentration
peak. It is possible that the receptor grid is missing several such peaks
in the "null" scenario (sources in their original positions).
3.5 RECOMMENDATION
Before the air quality prediction procedure outlined in this section
can be accepted as a useful planning tool, certain questions about its
accuracy must be answered.
Of primary importance is the sensitivity of the predictions to the
approximations inherent in the use of the dummy sources. One important
approximation is the use of a few effective stack heights to represent
the entire spectrum. Another is the error caused by the use of these
heights instead of using actual stack parameters. The sensitivity
analysis noted in Section 3.4 may provide useful data for analyzing the
degree of error represented by these approximations, and defining pro-
cedures to overcome or minimize these errors.
Another issue involves the sensitivity of the diffusion model
calibration constants to shifts in major plant locations. At first
glance, it seems that an analysis which looks at two time periods in the
history of a region would be necessary to establish an order of magnitude
sensitivity of calibration to plant location. It is possible that
sufficient data may not be available for such a study, and it is suggested
that a theoretical basis for establishing such sensitivity may be
available from the work being done on diffusion models which can account
for topography.
-------
Finally, it would be useful to construct an actual model (rather
than using the "wired-together" procedure that was necessary for this
brief analysis) and to run some more detailed scenarios in order to iron
out correct modeling procedures and to better judge the model's usefulness.
In regard to the latter point, an accurate comparison could be made of
actual computer time using the model and using the laborious procedure
of running a new diffusion model for every scenario.
-------
4.0 REFERENCES
1. TRW Systems Group, Air Quality Implementation Planning Program,
November 1970.
2. Dickerson, William D., Sensitivity Analysis of Selected Air
Quality Implementation Planning Program Input Parameters, TRW
Systems Group, June 1971.
3. Diamante, John and Goldstein, Burton, Demonstration of a Regional
Air Pollution Cost/Benefit Model, TRW Systems Group, June 1971.
4. Martin, Delance 0. and Joseph A. Tikvart, "A General Atmospheric
Diffusion Model for Estimating the Effects on Air Quality of One
or More Sources," APCA Journal (June 1968), pp. 68-148.
5. Pasquill, F., "The Estimation of the Dispersion of Windborne
Material," Meteorol Magazine (1961), 90, 1063, pp. 33-49.
6. Holland, J. Z., "A Meteorological Survey of the Oak Ridge Area,"
Atomic Energy Commission Report ORO-99 (1953), pp. 554-559.
7. The Cost of Clean Air, Report to the 91st Congress, Document No.
91-65, U. S. Government Printing Office, Washington, D. C.,
March 1970.
8. Boudreaux, A. D. and Weidemann, W. E., Forecast of Socio-Economic
Characteristics for the St. Louis Metropolitan Region, East-West
Coordinating Council, January 1970.
-------
APPENDIX A
-------
APPENDIX A
SOURCE DECK LISTING FOR THE LEAST-COST MODEL
00001 PR0GRAM OPT(INPUT*0UTPUT*TAPE6=0UTPUT)
00004 DIMENSION MI(45),ISTATEC30)*NSTATE(30)*REQ(30)
00006 DIMENSION BACKC30)*V(75)*X(45)*C0C30)*EX(45)
00007 DIMENSION 0UTPC80)
00008 DIMENSION DLEVELC15)*RATE(30*4)
00010 DIMENSION C(45*30)*C0ST(30*4)*QUANC30*4)*MR(75)* MIC(30)
00011 C SET QUAN(I*1):N0 CONTROL TONS/DAY OF ITH SOURCE
00020 DATA CQUAN(I)*1 = 1 *9)/6.25*5.7*11.37*17.15*5.09*4.21
00025 + *2.95*2.67*21»22/
0030 DATA CQUAN(I)*I=10*18)/3.42*7.3*6.* 10.7*6.* 4.72*
00035 +2.9*3.28*3.68/
00040 DATA
-------
.4491*.3298*
1187,.0896.
.4204,.1970,
,5079,.5582,
5697.
,4445,
.6937,2.0823:
00195 C SET INITIAL CCJ,I): ITH S0URCE C0NTRIBUT. T0 JTH DETECT0R
00200 DATA CCCI),1=1,9)7.2474,.4390,.5946*.3839*
00205 +.8218,1.3172,.3694,2.0697,.84937
00210 DATA CCCI),1=46,54)7.5665,1.0887,1.1673,.9735,
00215 +3.3816,1.5643,.6208,4.6410,.99897
00220 DATA CCCI),1=91,99>/«1704,.2585,.4871,.3496,
00225 +.8201,2.3927,.3170,1.3263,1.04417
00230 DATA CCCI),1=136,144)/.1520,.2511,
00235 +.5745,1.6024,.2899,1.0230,.72417
00240 DATA (CCI),1=181,189)7.0409,.0659,
00245 +.1379,.4082,.0799,«2714*.18677
00250 DATA CCCI),1=226*234)7.1466,.1613,
00255 +.4169,1.3011,. 19.02*.4576,1.37647
00260 DATA .CCCI),I=271,279)7.1364,.1788,.1606,.2064,
00265 +.2961,.2156,.4644,.6927,.29987
00270 DATA CCCI),1=316,324)7.3129,.6614,
00275 +4.6555,.7306,.3573*1»6288,.38667
00280 DATA CCCI),1=361,369)7.4851,.6668,.8131,
00285 +.8380,1.5646,.5512,.8457,1.55457.
00290 DATA CCCI),1=406,414)7.3190,.4439,.8794,
00295 +1.3501,1.6409,.2556,1.0757,.9003/
00300 DATA CCCI),1=451,459)7.5538,1.2900,
00305 +3.2759,1.0177,1.1418,.9075,.41817
00310 DATA CCCI),1=496,504)7.1174,.2471,.2457,.2412,
00315 +1.0650,.5721,.3518,1.7311,.23587
00320 DATA CCCI)*1=541,549)71.1535,24376*2.1218,2.0666,
00325 +11.7449,2.8054,1.3655,8.8684,1.67117
00330 DATA CCCI),1=586*594)7.3277*.3916*.4468*.3793*
00335 +.5778,.8371,.3282,.5581,1.57077
00340 DATA CCCI)* I=631,639)7 .0414,.0525,.0737,.0518,
00345 +.0317,.1353,.0867,.1535,.43347
00350 DATA CCCI),I=676,684)7.3561,.4893,.5351,.4770,
00355 +1.2980,.7596,.3337*3-7392,.43307
00360 DATA CCCI)., I = 721,729) 7.28 73, 4073, .8943, .291 8,
00365 +.7211,.4216,.1726,.4516,1.81537
00370 DATA CCCI), 1 = 766,774.)7 .0543, .0370, .0275, .0408,
00375 +.031 1, .0244, .0204, .0125, .01 137.. _
00380 DATA CCCI),1=811,819)7.1240,.0949,.0663,.0822,
00385 +.0604,.0528,.0659,.0334,.02307
00390 DATA CCCI),1=856,864)7.0557,.0766,.0703,.0562,
00395 +.0500,.0564,.0559,.0321,.02567
00400 DATA CCCI),1=901,909)7.0463,.1059,.1183,.1005,
004Q5 +.1897,.3325,.1157,.5858,.14807
00410 DATA CCCI), 1 = 946,954)7.06.44,.1592, .1367, .1337,
00415 +.5420,.2957,.1813*.6209,.12927
00420 DATA CCCI),1=991,999)7.0507,.0829,.1270,.1021,
00425 + .0996, .2920*.1182*«3253*.l7Q4/
00430 DATA CCCI),1=1036,1044)7.5511,.6827,1.0907,.8398,
00435 +1.1527,2.1001,1.0032,1.6938,3.70687 _
00440 DATA _CC.C I.)., 1 = 1 OS 1*1 089)7-0446, .0547, .0841, .0684,
00445 +.0929* . 161 7* .Q81 1*..1338* .27487
00450 DATA CCCI),1=1126*1134)7.2252*.2817,.4498,.3437,
00455 +.4791,.8643,.4014*.6682,1.36857
00460 DATA CC C I ), 1 = 1 171 * 1 1 79)7 .01 3.3*-0125* .0128* .01 75*
00465 +.0203*.0231,.0137*.0182,.02717
-------
00470
00480
00490
00495
00500
00510
00520
00530
00550
00560
00570
00580
00590
00600
00610
00620
00630
00635
00637
00640
00650
00660
00670
00680
00690
00700
00705
00707
00710
00720
00725
00730
00740
00750
00760
00764
00766
00768
00769
00770
00775
00780
00790
00795
00800
00810
008.11
00812
00815
00820
00830
00840
00850
C0 NVERT C(J*I) T0 INFLUENCE C0EFFICIENTS
D0 .177 1=1*NS
D0 177 J=1,ND
177 C(JjI)=C(J*I)/QUAN(I*l)
279 F0RMAT C9F7.4)
DISPLAY *INPUT REQ. LEVELS*
ACCEPT CREQ(I)*I=1*ND)
SET C0NSTANTS
DISPLAY *REDUCED PRINT? (0 0R 1):*
ACCEPT IPRT
NS1=NS+1
NE1=NS+ND
NS2=NE1+1
NE2=NE1+NS
SET INITIAL VALUES
DLEVEL:N0 C0NTR0L LEVELS REQ:REQUIRED LEVELS
ISTATE:CURRENT C0ST CURVE SEGMENT
D0 1 1=1,ND
DLEVELCI)=BACKCI)
1 XCI)=REQCI)-BACKCI)
D0 2 J=1,NS
NST=NSTATECJ)+1
ISTATE(J)=NST
PRESET RESIDUAL 0UTPUT
D0 10 1=2,NST
RATECJ,NST-I+2)=RATE(J,NST-I+1). .
10 QUAN(J>I)=QUANCJ,1)*C1.-QUANCJ, I)/l 00.)
FACT0R=1.
D0 3 I=1*ND
CCI,J>=CC!,J)*FACT0R
DLEVELCI)=DLEVELCI>+CNST)+QUAN(J*NST-1)
C0STCJ*!)=0.
D0 9 I=2*NST
9 C0ST(J*I)=C0ST(J*I-1)+RATECJ*I)*(QUAN(J,I-1)-QUAN
-------
00860
00865
00870
00880
00890
00900
00910
00920
00930
00940
00942
00943
00944
00945
00946
r\ n o A i
uuyH f
00950
00951
00952
00952
00954
00955
00956
00957
00958
00959
00960
00970
00971
00972
00973
00974
00975
00977
00978
00979
00980
00985
OQ990
01000
C
C
C
C
C
C
C
C
C
C
C
C
C
ND1=ND+1
D0 14 I=ND1*NE1
14 CCI*J)=0.
2 CCJ+ND*J>=1 .
D0 4 I=NS1,NE2
MRCI)=-CI-NS)
MI(I-NS)=I
4 V(J)=0.
MEX=0
D0 990 I=1*NE1
IFCXCD.LT.O.) DISPLAY *UNFEASIBLE*, I *X< I )
990 IFCXCI) .LT.O.) MEX=1
IFCMEX.EQ.l ) 60 T0 99
BEGIN L00PSSTART WITH X, V,C* MR, MI ,MIC
X:STATE VECTOR V: VALUE VECT0R C:C0NSTRAINT MATRIX
MR?MAPPING 0RIGINAL VECT0R T0 CURRENT L0CATI0N
MI:MAPPING BASIS VECT0R T0 ORIGINAL VECT0R
MICtMAPPING C0NSTRAINT VECT0R T0 0RIGINAL VECT0R
FIRST NS VECT0RS IN 0RIGINAL SET ARE S0URCE P0LLUTI0N CT0NS)
AB0VE MINIMUM F0R CURRENT SEGMENT. NEXT ND VECT0RS ARE
EXCESS QUALITY AT DETECT0RS. NEXT NS VECT0RS ARE REMAINING
P0LLUTI0N 0UTPUT 0N CURRENT SEGMENT.
IT=0
140 CONTINUE
IFCMEX1 -EQ.O) G0 T0 179
D0 181 I=1*NE2
IJ=MR(I)
0UTPCI)=0.
181 IFCIJ.LT.Q) 0UTPCI)=XC-IJ)
WRITE (6>182) C0UTPCI),I=1,NE2)
182 F0RMATC10E7.1)
179 CONTINUE
IT=IT+1
IFCIT.EQ.100) DISPLAY *IT=1 00 tENTERIT*
IFCIT.GT.100) ACCEPT IT
-------
U1U.UD U
01010 C
01020
01.030
01040
01050
0.1.060
01070
01080
0.1.090
Oil 00
011.10
01120 C
01135 C
01.1.30
01 140
01150
01.160
01.170
01180
01190
012QO
01210
01220
01230
01240 C
01250
01260
01270 C
01280
01290
01300
01310
01320 C
01330
01340 C
01350
01360
01.370
01375
01380 C
FIND.VECT0R F0R INSERTION
X2=Q,
D0 19 K=1,NS
.. IC=MICCK)
19 C0(K)=V(IC)
D0 20 I=1*NE1
IB=MKI) .
IFCVCIBV.EQ.O.) G0 T0 20
D0 20 J=1*NS
IF(CCI*J).NE.OO C0 G0 T0 122
X2=C0(K)
KMX=K
122 IFCC0CK5.GT.OO 60 T0 21
IC=MIC(K)
IFCIC.GT.NE1) G0 T0 27
IFCIC.GT.NS) G0 T0 21
NST=ISTATECIC)
IF+1) G0 T0 21
X1=-C0CK)+VCIC)-RATECIC*NST-M)
IFCX1 .LE.X2) G0 T0 21
G0 T0 24
27 NST = ISTATECIC-NE1 ) ..
IFCNST.EQ.2) G0 T0 21
IC=IC-NE1
X1=-C0CK)-VCIC)+RATECIC*NST-1 )
IFCX1 .LE.X2) G0 T0 21
24 KMX=-K
X2=X1
21 C0NTINUE
IFCX2.LE.O.) G0 T0 110
-------
Ul JO1
01390
01400
01410
0.1.430
01430
01440
01450
01.460
01470
01480
01490
01500
01510
01520
01530
0.1540
01550
01560
0.1.570
01580
f\ t c on
Ul. J7U
01600
01610
01620
01625
01630
01640
0165Q
01651
01652
01660
01665
01670
01680
01690
01700
01710
01720
01.730
01733
01736
01738
01740
01750
01760
01770
01780
01790
01800
01810
01820
01830
01840
01850
0.1.860
01870
01880
L»
C
c
C
c
c
c
c
c
c
c
c
c
c
c
c
c
c
KMX IS ENTERING VECTOR I IF NEG THEN CHANGE SEGMT.
IFCKMX.LT.O) G0 T0 35
MD=0
FIND VECT0R FOR REMOVAL
MIN 0F XCJ)/CCJ,KMX)
KMN=0
D0 22 J=1,NE1
IF(CCJ,KMX>.LE.O.) G0 T0 22
T0T=XCJ>/C
K5=MIC(KMX)
K6=MICKMN>
C0 IS CONSTRAINT C00RD IN OLD BASIS VECTOR COMPONENT
DO 28 1=1, NS
28 C0CI)=CCKMN,I>
D0 25 1=1, NS
IFCCOCn.EQ.Q.) GO TO 25
D0 26 J=1,NE1
26 C=C(J,I>+C0*EX(J>
25 CONTINUE .
D0 207 J=1,NE1
207 CCJ,KMX)=EXCJ>
CCKMN,KMX)=EX(KMN>+1 .
UPDATE MAPPINGS
IC=MI(KMN)
MRCIC)=KMX
MICKMN)=MICCKMX)
MICCKMX)=IC
IC=MI(KMN)
MR(IC)=-KMN
NEW SOLUTION VECTOR
IFCMD.EQ.l) GO T0 140
X1=X(KMN)
D0 30 1=1, NE1
30 X+EXCI)*X1
L00P BACK
G0 TO 140
-------
01881
01882
01 890
01.90.0
01910
01.920
01930
01940
01950
01960
01970
01980
01990
02000
02010
02020
02030
02040
02050
02051
02060
02070
02074
02075
02080
02090
02091
02100
02105
02110
02120
021.30
02140
0215Q
02151
02160
021 70
02180
02190
022QO
02210
02220
02230
02240
02250
02260
02265
02270
02275
02280
02290
023QO
02310
C CHANGE C0ST CURVE SEGMENT
35 KMX=-KMX
MD = 1
KM=MICCKMX>
IFCKM.GT.NS) G0 T0 36
NST=ISTATECKM>
ISTATECKM)=NST+1
KMN=-MRCKM+NE1)
XCKMN)=-QUANCKM,NST+1) +QUANCKM,NST)
VCKM>=+RATECKM,NST+1)
G0 T0 37 .
36 IC=KM-NE1
KMN=-MR(IC)
NST=ISTATE(IC)
ISTATECIC)=NST-1
XCKMN>=-QUANCIC*NST-1)+QUANCIC*NST-2
VCIC)=RATE(IC*NST-1)
G0 T0 37
C
C ERR0R EXIT
120 DISPLAY *UNB0UNDED S0LUTI0N*
KMX=MIC(KMX)
DISPLAY.*EXIT VECT**KMX**ITER**IT
G0 T0 99
C
C
C
FINAL S0LUTI0N ATTAINED
110 C0NTINUE
DISPLAY IT**ITERATI0NS*
* *
*SUMMARY 0F RESULTS*
*S0URCE LEVELS*
*S0URCE # %CUT
476
204
201
DISPLAY
DISPLAY
DISPLAY
DISPLAY
+** RATECS/TN)*
D0LLAR=0.
D0 201 I=1*NS
ID=-MRCI>
NST = ISTATECI) ..
NSTM=NSTATECI)-H
Xl=0.
IFCID.GT.O) X1=-XCID)*V(I)
X1=C0STCI*NST)+X1
X2=0.
IFCID.GT.O).X2=XCID)
X2=X2+QUANCI*NST)
D0LLAR=D0LLAR+X1
X3=100.-100.*X2/QUANCI>1)
IFCIPRT*ID.LT.CNST-NSTM)*20Q>
WRITE C6*204) I>NST,X3*X2,X1*
CONTINUE
F0RMATC2I4.,F10.1*F8.2»2F10.1 )
C0NTINUE
RESIDUALCTN) C0STCS)*
G0 T0
VCD
476
-------
02320
02325
02330
02340
02350
02360
02370
02380
02390
024QO
02410
02420
02430
02440
02450
02460
02470
02480
02485
02490
206
205
99
DISPLAY * *
DISPLAY *RECEPT0RS:*
DISPLAY.*RECEPT0R PRE P0ST
D0 205 I=1*ND
Xl=0.
ID = -MRCI+NS) ....
IFUD.GT.O) X1=XCID)
Xi=REQ(I)-Xl
X2 =REQU)
X3=DLEVEL
X4=0.
IFCID.LT.O) X4=-C0C-ID)
WRITE (6*206) I*X3>XI*X2,X4
F0RMAT
C0NTINUE
DISPLAY * *
DISPLAY *T0TAL C0ST:**D0LLAR
C0NTINUE
END
REQ
MARG C0ST*
------- |