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  3—1   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)71.1535,2•4376*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*

-------