ANALYSIS OF THE RELIABILITY OF A
SUPPLEMENTARY CONTROL SYSTEM FOR S0}
EMISSIONS FROM A POINT SOURCE
ERT Project 0669
JUNE 1974
/OHAI P. GAER1NER
BRUCE K. EGAN
NORMAN E. GAUJ
IOHN 5. LAGUE
JHOMAS E. LAVERY
HARVEY S. ROSENBIUM
ERED C. SCHWEPPE
prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK,
NORTH CAROLINA 37711
ENVIRONMENTAL RESEARCH & TECHNOLOGY, INC.
429 MARRETT ROAD, LEXINGTON, MASSACHUSETTS 02173
-------
ANALYSIS OF THE RELIABILITY OF A
SUPPLEMENTARY CONTROL SYSTEM FOR S03
EMISSIONS FROM A POINT SOURCE
ERT Project 0669
JUNE 1974
IOHN P. GAERJNER
BRUCE A. EGAN
NORMAN E. GAUJ
IOHN 5. LAGUE
THOMAS F. LAVERV
HARVEY S. ROSENBIUM
ERED C. SCHWEPPE
prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK,
NORTH CAROLINA 17711
ENVIRONMENTAL RESEARCH & TECHNOLOGY, INC.
429 MARRETT ROAD, LEXINGTON, MASSACHUSETTS 02173
-------
SUMMARY
This document is a final report for a study involving the design and
analysis methodology to assess the reliability of SCS systems. The work
includes a case-study investigation of the performance of an existing SCS.
A brief summary of the program results are presented here.
Summary of the Development of Analysis Tools
A number of analytical tools were developed in the course of this
study. These tools are briefly summarized by the paragraphs below.
The analysis methodology to assess SCS reliability has required a prob-
abilistic definition of SCS reliability. The reliability is measured by
the probability of violation of air quality requirements within the region
of influence of the controlled source. The representation of frequency
distributions of maximum ambient concentrations is presented as a useful
method of quantifying SCS reliability. This approach is utilized throughout
the study.
To analyze the reliability of SCS systems, a computer program (PROBL)
is presented as a tool for generating the frequency distribution of maximum
ground-level concentrations. The method incorporates knowledge of meteor-
ological statistics for the area, emissions statistics, a measure of air
quality forecasting error, and a prescribed SCS plan. The resulting fre-
quency distribution provides a measure of the anticipated reliability of
the plan.
Another analysis tool was developed to identify the air quality
standards for which an SCS is a pertinent control procedure and to investi-
gate the impact of an SCS on the achievement of standards having various
averaging times. The product of the analysis scheme is a cumulative fre-
quency distribution for maximum ground-level concentrations for any desired
averaging time and employing any control procedure designed to adhere to a
single air quality standard (or threshold).
Another analysis tool assists in determining the best distribution of
monitors for SCS applications considering the practical and economic con-
straints on the network. The analysis considers the climatological statistics
of the region, a minimum concentration threshold of interest, some required
resolution of peak concentrations, and the minimum acceptable or maximum
affordable number of monitors to be used. The analysis provides the best
111
-------
monitor locations ranked by their probability of observing concentration
values exceeding the threshold of interest.
Finally, a method is proposed of separately measuring the effects on
SCS reliability of meteorological forecasting error; emissions forecasting
error, and air quality modeling error. The method requires that complete
records be maintained of all predicted and observed meteorological, emissions,
and model parameters for each forecast time and for each verification time
during SCS operation. Through a series of model predictions using various
combinations of observed and predicted parameters, the component errors can
be isolated.
Summary of the Performance of an Analysis to Assess the Reliability
of SCS Methods
A complete analysis of the reliability of an SCS requires that the
analysis tools be applied at three stages of the SCS; that is, during design
of the system, during real-time operation of the system for verification of
system performance, and during real-time operation of the system for updating
and improving the system.
This report contains valuable information for use in the design stage
of an SCS. The source must comply with the EPA guidelines for SCS as detailed
in the Federal Register. The requirements which are pertinent to this report
have been summarized in Section 2. Implicit throughout the analyses is
the requirement that any proposed SCS operational strategy must theoreti-
cally insure compliance with all air quality standard requirements. With ade-
quate controls available on all components of an SCS, only uncertainty in air
quality projections should be the cause of non-compliance with air quality
standards. Equivalently, where no uncertainty in forecasting exists, no vio-
lations of air quality standards should exist.
Section 4 examines each component of the SCS system to isolate sources
of uncertainty and to identify criteria by which the components can be
assessed. Each SCS is assumed to consist of four components: ambient
air quality monitoring, meteorological forecasting, emissions forecasting,
and air quality modeling.
Criteria were established for assessing the appropriateness of a moni-
toring network for SCS applications. Through the analysis tool developed
for assisting in monitor siting, a quantitative method of measuring monitor
distribution adequacy was possible.
IV
-------
Similarly, criteria for assessing the adequacy of each of the remaining
three SCS components are discussed. Methods of evaluating the reliability
of each of the SCS component parameters are discussed in terms of relating
the reliability of predicting parameters to the overall reliability of the
SCS. In harmony with the overall analytical framework developed, a method
is presented in Section 4.5 for separately measuring the effects on SCS
reliability of meteorological forepasting errors, emissions forecasting
errors, and air quality modeling errors.
Also of great value during the design of an SCS is the analytical
technique for assessing the effects of SCS operation on various averaging
times. The technique both determines what averaging times can be adequately
controlled by SCS methods and determines if the control protects all per-
tinent air quality standards.
During the performance of an SCS, it is beneficial for both the source
and the control agency to verify that the anticipated SCS reliability is
being achieved. The separation of forecast error into the effects of each
SCS component is an essential element in assessing performance of the
system. Continual analysis of these error components enables some sources
of uncertainty to be identified for possible improvement. The PROBL
program can then be used to calculate the frequency with which air quality
standards are expected to be violated through future operation of the SCS.
The same analysis techniques designed to verify the reliability of the
SCS can be used to update the SCS system both during the initial testing
of the SCS and periodically during operation of the system. The isolation
of modeling errors by the techniques described in Section 4.5 can also be
used to update the air quality prediction methods. Systematic modeling
errors will be used to develop calibration factors for the models which
should improve system reliability as system experience accumulates. Also,
isolation of the meteorological uncertainty is a valuable indicator of
improvements which might be made in the forecast system. Finally, the
probability of violating standards can be generated periodically as the
emission, background, and accuracy of the system change with time. This
periodic check on system performance assures that system reliability can
be maintained and, ideally, improved as operators gain experience with
the system.
-------
Performance of a Case Study
A case study of the reliability of an existing SCS system was made to
relate to the methodologies and concepts developed in this investigation. The
ASARCO plant in El Paso employs real-time operation of continuous S0_ moni-
toring stations as well as full time meteorological forecasting. The ASARCO
plant is also representative of an isolated source during the test period
being primarily responsible for SO air quality levels in the area. Data for
only a limited time period of about two months was used.
The case study had three major phases:
1. An analysis of the overall effectiveness of the ASARCO SCS was
made by hypothesizing the expected concentrations that would
have resulted in the area had no emission reduction been
employed. The results show that the ASARCO SCS has a major
beneficial effect on SO levels in the area.
2. A case by case analysis was made of the occurrences of the
highest observed concentrations in the data set. This analysis
involved studying the meteorological conditions that existed
at these times and, in conjunction with the observed air quality
data, the determination of likely reasons why the SCS was not
effective in maintaining air quality concentrations below the
desired levels. Our analysis points to the possibility that
a better understanding of the effect of terrain on plume tra-
jectories and the occurrences of terrain induced downwash
during various meteorological conditions could result in an
improvement of air quality forecast reliability.
3. A demonstration of the analytical techniques developed for
general SCS analysis was performed for the ASARCO system.
Ideally, SCS system records would provide direct raw data
information on forecasted air quality levels and actual
emission rates and an objective control strategy which
could be used to generate the necessary probability distributions
directly.
In the absence of the ideal data, it was necessary to hypothesize
what the air quality forecasts might have been and what the criteria
VI,
-------
for controlling emission rates might have been on the basis of
observed air quality levels and available records of emission
curtailment orders. A model of the ASARCO system was constructed
on the basis of a hypothesized control strategy and implied fore-
cast model and run for the test period. The results indicate
that the computer model can provide a very useful framework with
which to understand the effect of different control strategies
on the expected frequency of occurrence of high concentration
levels and the effect of improving air quality model accuracy
on overall system reliability.
VII
-------
TABLE OF CONTENTS
Page
SUMMARY ii;L
1. INTRODUCTION 1_1
2. BACKGROUND
2-1
2.1 Motivation for Supplementary Control Systems 2-1
2.2 Types of Supplementary Control Systems 2-2
2.3 Current Supplementary Control Systems 2-4
2.4 Federal Regulations Regarding Supplementary
Emission Controls 2-6
3. DESIGN OF ANALYSIS METHODS TO TEST THE RELIABILITY OF A
SUPPLEMENTARY CONTROL SYSTEM 3-1
3.1 Introduction 3-1
3.2 Definition of Reliability 3-2
3.3 Probability Concepts Relevant to an SCS Analysis 3-3
3.4 Design of an Analysis Scheme to Quantify the Reliability 3-7
of a Supplementary Control System
3.4.1 Assumptions and Definitions 3-7
3.4.2 The Mathematical Description of the SCS Analysis
Scheme 3-10
4. RELIABILITY OF THE COMPONENTS OF A SUPPLEMENTARY SO EMISSION
CONTROL SYSTEM 4-1
4.1 Introduction 4-1
4.2 Assessment of Air Quality Monitoring Reliability 4-1
4.2.1 Sampling and Information Transfer Errors 4-2
4.2.2 Example Analysis to Determine a Distribution of
Monitors Sufficient to Represent the Field of
Downwind Concentrations 4-3
4.3 Assessment of Meteorological Forecasting Reliability 4-20
4.3.1 Introduction 4-20
4.3.2 Criteria for Assessing Meteorological Forecasting 4-24
Reliability
4.4 Assessment of Air Quality Modeling Reliability 4-30
4.5 A Proposed Method of Quantitatively Separating SCS
Component Reliability 4-38
IX
-------
TABLE OF CONTENTS (Cont.)
Page
5. ANALYSES OF RELIABILITY FOR SEVERAL HYPOTHETICAL SCS SYSTEMS 5-1
5.1 Test Case Conditions and Assumptions 5-1
5.2 Dependence of SCS Reliability on Various Influencing Factors 5-3
5.3 Additional Capabilities of the Proposed Probability Analysis
of SCS Reliability 5"15
6. SIMULATION OF THE EFFECT OF A CONTROL STRATEGY ON CONCENTRATIONS
FROM VARIOUS AVERAGING TIMES 6-1
6.1 Introduction 6-1
6.2 Description of the Analysis Technique 6-2
6.3 Example Analyses 6-3
7. CASE STUDY EVALUATION OF AN EXISTING SCS 7-1
7.1 Introduction 7-1
7.2 Statistical Analysis of SCS Performance 7-4
7.3 Analysis of Individual Instances of High SO Concentrations 7-18
7.4 Application of the Proposed Analytical Techniques to the Case
Study Data 7-34
7.5 Discussion 7-38
8. ENFORCEMENT OF SUPPLEMENTARY CONTROL SYSTEMS 8~1
9. CONCLUSIONS AND RECOMMENDATIONS 9-1
9.1 Review of Major Accomplishments 9-1
9.2 Recommendations for Future Work 9-3
REFERENCES R-l
APPENDIX A The ERT Point Source Diffusion Model A-l
APPENDIX B Effects of Control Strategies on Ground-Level
Concentrations for Different Averaging Times B-l
-------
1. INTRODUCTION
This report describes the results of an investigation conducted by
Environmental Research & Technology, Inc. (ERT), under contract for the
United States Environmental Protection Agency (EPA) entitled, Analysis
of the Reliability of a Supplementary Control System for S02 Emissions
from a Point Source.
EPA has determined that an in-depth analysis of the reliability of
supplementary control systems (SCS) is necessary to decide if the SCS con-
cept is technically and administratively workable. Several states are finding
it difficult to achieve National Air Quality Standards without drastic and
expensive changes in fuel use, changes in industrial production rates, or
changes in plant design. As an alternative, an increasing number of govern-
mental agencies, air pollution sources, and air quality specialists are
proposing conscientiously operated SCS strategies to reduce the impact of
emissions from large installations. By reducing emissions during periods
when weather conditions are not conducive to adequate dispersion of the pol-
lutants, the impact of emissions would clearly be reduced. It is not obvious,
however, that emissions could be controlled sufficiently by this method to
insure that all applicable standards are met. Because of this uncertainty
and because of concern that an SCS would be difficult to enforce and monitor,
a contract for this reliability study was awarded,
EPA has requested that several tasks be explicitly performed as a part
of this reliability study. These tasks are described in the six items below:
1. Define the standards for which an SCS with its built-in time con-
straints is valid. Define the impact of various control strategy
time scales on the reliability of the SCS system in achieving com-
pliance over differing averaging times.
2. Estimate the frequency that air quality standards might be expected
to be violated with and without an SCS and explain the basis for
the estimate.
3. Identify the subjective elements of an SCS.
1-1
-------
4. Present criteria for an adequate air quality monitoring network
necessary to develop and operate an SCS, and for meteorological
instrumentation and services.
5. Identify the factors involved in adequately enforcing an SCS by
control officials.
6. In a case study of an SCS, evaluate the reliability of the example
system and describe the progress, or lack thereof, in attaining
air quality standards.
The analysis was required to be applicable to an isolated, large point
source of S0~. The study reported in this document considers each of
the above items and describes an analysis method to assess the reliability
of SCS methods in general. This report consists of nine sections, ordered
for purposes of logical presentation. Section 2 provides a general back-
ground discussion of SCS types, existing SCS operations and appropriate
Federal regulations. Sections 3 and 5 address item (2)" above in detail.
Section 3 considers the definition of reliability to be used, reviews
relevant theoretical concepts, and defines the framework of the analysis
methods employed in the study. A probability analysis of SCS methods is
applied in Section 5 to reveal the sensitivity of SCS .reliability to vari-
ations of the individual components. Section 4 primarily addresses item (3)
and (4) in the context of establishing the criteria for assessing the
reliability of individual components of an SCS system. Section 6 addresses
the work task of item (1) above and presents a methodology to assess the
performance over differing averaging times of an SCS established for a
particular controlled averaging time. Time delay factors associated with
meteorological forecasting operations are discussed additionally in
Section 4. A case study of the American Smelting and Refining Company
(ASARCO) SCS programs is presented in Section 7 in response to item (6)
above. This case study concentrates first on the statistical summary of
SO concentration data for a selected test period and then on five individual
cases where hourly violations have occurred. The application of the ERT
program PROBL to the case study data is also part of Section 7. Item (5)
above associated with the enforcement of SCS procedures is discussed in
Section 8. Finally, Section 9 reviews the major accomplishments of this
study and presents specific recommendations for future related work.
1-2
-------
2. BACKGROUND
2.1 Motivation for Supplementary Control Systems
Industry is facing a long-term problem of availability of fuel and a
short-term problem of using available fuels in such a way that EPA environ-
mental standards are met. There appears to be no easy solution to the
energy problem and no easy solution to the environmental problem. The pro-
spect that energy costs are going to rise in the future seems inescapable
regardless of what course of action this country chooses to take. These
costs will take different forms, but the goal clearly must be to minimize
them in the aggregate.
Up to now, as a nation, we have only begun to reconcile the conflicts
between environmental and economic concerns. A case in point is the debate
now proceeding on dynamic emission control systems for emissions, identified
within EPA as supplementary control systems (SCS). Supplementary control
systems are defined as (Federal Register, September 14, 1973) "systems
whereby the rate of emissions from a source is curtailed when meteorological
conditions conducive to high ground-level pollutant concentrations exist
or are anticipated." Economists hale supplementary control systems such as
that based on fuel switching for their potential to save industry and the
public millions of dollars in fuel costs and lowered electrical costs.
Environmentalists are concerned with enforcement problems of SCS systems
and unquantifiable effects of SO. emissions. Once an allowable impact on
air quality is defined, fuel switching is an optimization technique which
can take on many forms. It can as easily be optimized for minimum dollar
cost for fixed environmental impact (the normally-advocated form of this SCS)
or for minimum environmental impact for a fixed dollar cost (say the present
total cost of fuel). An SCS based upon fuel switching is, therefore, neither
a windfall for industry nor a calamity for the environmentalist but merely
a tool which can be used to better, utilize existing resources within a wide
variety of possible constraints. The same considerations apply to other
forms of SCS systems such as process curtailment and load switching.
Because of the very great expense and obvious nonoptimum nature of
present-day rollback procedures for sulfur emissions, there is growing
desire within the power industry and other concerned and competitive
industries to optimize the use of fuels for minimum cost with the constraint
2-1
-------
that air contamination be maintained below all applicable ambient standards.
When standards were promulgated they supposedly had built-in margins of
safety for health at the primary limits and for welfare and nuisance effects
at the secondary limits. Therefore, it is argued that as long as air quality
is maintained at concentration levels below those required by the most
stringent standard, then the method which promises lowest total costs should
be utilized. Clearly, the simplistic rollback and absolute emissions limits
now.being enforced do not allow for any sophisticated system to jointly
optimize costs and environmental impact. The concept of the SCS, however,
can be aimed exactly at that optimization.
A major question regarding the use of SCS systems is the expected
reliability of these systems with respect to maintaining air quality standards.
This report presents a methodology for evaluating the reliability of SCS
systems in terms of probabilities of exceeding air quality standards as a
function of emission factors, meteorological conditions, and control
strategy.
2.2 Types of Supplementary Control Systems
For power-generating stations and industrial processes, there are three
types of supplementary emission controls which can be adopted: fuel switching,
load shifting, and process curtailment. The general aspects of these control
measures are summarized below.
Fuel Switching involves the use of two or more fuels, with different
sulfur content, according to a plan designed to meet specific air pollution
control criteria. Because the fuels are usually differentiated with respect
to sulfur content they have different impacts on air quality. At present,
the principal control method for SCL and SO emissions is the use of low
sulfur fuels. An example of fuel switching would be a facility using 2%
or greater sulfur content fuel during times when local ventilation is good,
switching to 0.5% sulfur content fuel during times of poor dispersion.
This type of control results in a negligible reduction in plant production-
In practice, three time scales for fuel switching have been proposed
for individual facilities:
1. Normal operational use. This would constitute short period
switching in which a facility would be operated with switching on a daily
basis, or more frequently if desired.
2-2
-------
2. Seasonal use. Because SCL emissions related to space heating
demands do not occur in the summer months, power-generating facilities
might use a higher sulfur content fuel during these times, saving the
low sulfur fuel for use in the winter period.
3. Emergency use. A power-generating station might operate with a
limited reserve of low sulfur fuel, for use during a few days of each
year, when specific local pollution problems occur.
Fuel switching operations with oil are relatively simple compared
with switching between high and low sulfur coal. The coal switch requires
not only extra storage and handling facilities, but possibly two feed
bunkers instead of one;and a time delay of several hours could occur before
changes in SCL emission levels would be apparent in the stack.
Load Shifting involves the controlled operation of two or more genera-
ting stations which are interconnected by power transmission lines, and under
the control of a common dispatch center. Load shifting for air quality control
is accomplished when the dispatching practices developed for a group of stations
are modified to incorporate consideration of the total regional impact of the
interconnected stations. In addition to load shifting program, fuel switching
can be simultaneously practiced at individual stations within a network.
Load shifting for air pollution control purposes is possible only
when an interconnected group of generating stations has capacity in excess
of demand. Many of the power networks within the United States operate
near capacity at certain times of the day and year. However, almost all
networks have excess capacity some of the time; for example, during the
late night-early morning period (typically, from 10 PM to 6 AM local time).
Shepard (1970) has performed a case study analysis of a proposed load
shifting plan for the generating stations in the St. Louis area. He found
that significant reductions of population exposures to S02 are possible
with small increases (about 5%) in power-generating costs for the region.
Shepard's analysis treated only load shifting operations, without considera-
tion of fuel switching at single stations. Because of the current price
differentials between fuels with various sulfur contents, fuel switching
operations can result in even greater cost efficiencies than those cited
by Shepard for the load shifting case.
2-3
-------
Curtailment of an industrial process is a conceptually simple way to
achieve dynamic emission controls, and sometimes the most practical.
When air quality standards are violated some small fraction of the total time,
significant emissions reduction can be achieved by cutting back operations
when standards are threatened.
Other possible types of supplementary control methods include dynamic
controls of stack gas temperatures to enhance plume rise or intermittant
use of scrubbers or similar emission cleaning devices during adverse meteoro-
logical conditions.
2.3 Current Supplementary Control Systems
Tennessee Valley Authority
Load shifting is being demonstrated by the Tennessee Valley Authority
(TVA, 1973) as a feasible, reliable form of dynamic emissions control for
meeting air quality standards. In September 1969, TVA began operating an
S0? limitation program at their Paradise Steam Plant in west-central Kentucky.
Plant-generating load reductions reduce emissions whenever plume dispersion
is unfavorable. These conditions are identified by on-site meteorological
measurements which measure adverse atmospheric dispersion characteristics.
The TVA program includes extensive pre-operational field studies to monitor
maximum ground-level concentrations and collect added information about plume
characteristics through helicopter and mobile ground measurements. This infor-
mation is combined with output from a critical plume dispersion model to
establish nine meteorological and plume dispersion criteria to determine
when critical conditions threaten standards and require emissions reductions.
The TVA program has demonstrated the effectiveness of the emissions
control techniques. During the first 39 months of this program, there were
41 days requiring reduction of the generating load. The magnitude of the
reduction ranged from 26 to 960 M1V with an average of 454 MW. The average
duration of the reduction was 3.6 hours, with a minimum of 24 minutes and
a maximum of 5.8 hours. Frequency distributions of measured SO concentra-
tions before and after implementation of TVA's SC>2 emissions limitation
program at the Paradise Steam Plant demonstrate that Kentucky's ambient
standards were violated before, but not after the initiation of this pro-
gram. Significantly, emissions during the first 39 months of the program
2-4
-------
averaged 13.0 kg/sec, a 50% increase over the 8.65 kg/sec emission rate
applicable before the program (Montgomery, 1973).
TVA has emission limitation programs at two other plants and plans to
perform studies at six additional plants. They are fully committed to load
shifting as the only feasible emissions control program they can employ
to meet ambient standards.
American Smelting and Refining Company
The American Smelting and Refining Company (ASARCO) has demonstrated
the effectiveness of a program for emissions control by curtailing smelter
operations. ASARCO has installed a control system that includes 18 S02
monitoring stations at their El Paso smelting complex. The system enables
ASARCO to curtain operations to reduce or eliminate unacceptable S0? con-
centrations . A real-time method of monitoring and forecasting concentra-
tions has been developed to enable ASARCO to achieve earlier, preventive
curtailment and further reduce the time lag between detectipn and reversal
of an upward trend in S0? concentrations. ASARCO curtailment programs
for dynamic emissions control at their El Paso and Tacoma complexes have
substantially reduced their air quality violations.
Environmental Research § Technology, Inc.
Environmental Research & Technology, Inc. (ERT), has prepared and sub-
mitted to EPA fuel management plans for the New England Power Company,
the General Electric Company, Lynn, Massachusetts, and 28 diversified
industries comprising 38 individual emission sources in the Connecticut
River region (Pioneer Valley) of Massachusetts. The plan for the Pioneer
Valley called for control of almost one-half of all the sources emitting
SO- over this entire airshed.
ERT has been operating a real-time S02 monitoring and forecast system
in Massachusetts and New York State since December 1971. This system is
called AIRMAP^, an acronym which stands for Mr Monitoring Analysis and
Prediction. AIRMAP operations and data over more than 24 months demonstrate
the feasibility and reliability of optimizing air quality through dynamic
emissions control in the Boston area.
The AIRMAP program consists of field sensors, telemetry, analysis
software, and prediction techniques to monitor and predict ambient air
2-5
-------
quality and monitor stack emissions. It is an integrated software-hardware
system. The prediction component of the system, which has proved to be
highly accurate, provides forecasts of air quality a day in advance.
These forecasts form an integral part of the dynamic emission control
procedures. The reliability of some components of this system are discussed
elsewhere in this report.
The AIRMAP system will soon include more than 200 monitoring sites
measuring a total of more than 500 parameters (SCL, N09, NO , coefficient of
£ £ A.
haze, wind direction and wind speed, temperature, etc.)- Data is reported
back from the sites to the AIRMAP Centrals once every minute via dedicated
telephone lines.
AIRMAP presently consists of five major components:
1. Sensors and station facilities
2. A telecommunication system
3. Central facility computers and meteorological data acquisition
equipment
4. Data analysis software
5. Air quality prediction software.
These components could form the basis of a comprehensive SCS network.
2.4 Federal Regulations Regarding Supplementary Emission Control
On 31 May 1975, air quality throughout the United States must meet
primary ambient air quality standards for a series of pollutants.
The Clean Air Act and Amendments thereto establish national primary
ambient air quality standards to protect public health. These standards
specify limits on ground-level concentrations of various pollutants
including sulfur dioxide (SO-).
In the Federal Register of July 27; 1972, (37 FR 15095) the Environ-
mental Protection Agency set forth a position against the use of SCS as a
substitute for so-called permanent control systems. They stated that
although air quality is improved by SCS, violations of air quality standards
still occur. They stated, however, that the reliability of SCS might be
improved to a level acceptable for limited use.
2-6
-------
In the Federal Register of September 14, 1973, (38 FR 25698) the EPA
noted that due to "substantial and consistent improvement in the performance
of SCS in the recent past" their use could be acceptable for a small
number of large isolated sources. The proposed regulations impose stringent
limitations on the types of sources and situations for which use of SCS
may be considered acceptable.
The two principal considerations in determining SCS acceptability are:
The system must be only a supplement to the best available con-
stant emission controls. This restriction implies that the
source must actively investigate emission control technology and
must implement new control techniques as they become available.
The system must exhibit a high degree of reliability and must be
made legally enforceable.
This report addresses the latter consideration for SCS acceptability.
The Federal Register statement contains a detailed description of
SCS objectives, system components, roles of each SCS component, and
reporting procedures. The analysis techniques which are included in
this report are designed to apply to SCS systems which comply with these
EPA requirements. Some important elements of an SCS relevant to this study
are included below.
The SCS must have an SCL monitoring network with an array of
sampling points located where maximum ground-level concentrations
are most likely to occur. The number of monitors must also be sufficient
to calibrate a model for interpolation between monitors.
The SCS must have an operating model which relates meteorological
inputs, emission rates, source data, terrain, and location factors
to ambient air quality.
The SCS must have meteorological parameter observations and pre-
diction for use in air quality forecasts. Similarly, the source
must have continuous emissions monitoring and forecasting.
Objective rules for emission control must exist. These rules
relate air quality predictions to a controlled emissions schedule.
2-7
-------
The SCS source must conduct a comprehensive background study of
not less than 120 days. The results of this study must, among
other requirements, provide objective reliability tests to measure
the ability of the SCS to protect against violations of national
standards.
The source must define a program whereby they systematically evaluate
and improve the reliability of the supplementary control system.
The above list of SCS requirements will force a degree of uniformity
among acceptable SCS plans. The analysis which follows assumes only that
the preceeding SCS requirements are applicable; thus the reliability analysis
is adaptable to any SCS which satisfies the EPA guidelines.
2-8
-------
3. DESIGN OF ANALYSIS METHODS TO TEST THE RELIABILITY
OF A SUPPLEMENTARY CONTROL SYSTEM
3.1 Introduction
Because there is no rigid format to which all supplementary control
systems must conform, it is necessary to design analysis methods which can
be applied to a variety of control system designs. This section is devoted
to defining analysis criteria which are part of a general structure with
which the performance of each component of a proposed SCS can be assessed
and combined to determine the reliability of the entire SCS.
The suggested analysis methods presented in this section are designed
to conform with the latest EPA recommendations for supplementary control
systems. In particular, the proposed methods require only the meteorological
and monitoring support required for SCS operations in the Federal Register.
Furthermore, the results of the analysis tools are useful inputs for satis-
fying EPA requirements during the design, verification, and updating of the
SCS.
The complete analysis of SCS reliability requires a series of quantitative
and some qualitative examinations of the SCS from its planning stages through
periodic updating of its performance after implementation.
During the design stages of the SCS the components of the system must
be engineered to provide maximum system reliability. A breakdown of the
SCS into four generalized components appears most logical - the air quality
monitoring system, the meteorological forecasting system, the emissions pre-
diction system, and the air quality modeling system. For each component,
analysis criteria can be isolated which are pertinent to assessing the effect
of the component on overall system reliability. Methods of applying the
criteria to each component are presented. Where feasible, quantitative pro-
cedures are developed for applying particular criteria. Also during the
design stage of an SCS an estimate of the potential reliability of the
proposed system can be assessed. For this purpose it is necessary to identify
expected meteorological statistics, an expected emissions schedule, and
expected uncertainty in air quality forecasting for the proposed SCS
strategy. An analysis framework is proposed which uses these factors to
provide an estimate of expected SCS reliability. The analysis methods are
also applied to assess the sensitivity of the SCS to changes in SCS strategy.
3-1
-------
During the initial usage of the SCS, it is essential to measure the
true reliability of the system. A quantitative procedure is presented
for isolating the effect of individual system components on overall SCS
reliability. Using these techniques, an overall measure of uncertainty in
air quality forecasting can be developed. In addition, a measure of uncer-
tainty in the component elements of air pollution potential forecasting -
(weather forecasting errors), emission rate estimation, and uncertainty
in model calculations (air quality model errors)-can be independently deter-
mined. This requires measures of meteorological statistics, plant emissions
schedule, and actual uncertainties in air quality forecasting. After the
system has been in operation, the above procedures are necessary for verify-
ing continued SCS performance and for providing a basis for updating the
system.
The remainder of this section provides the definitions and theoretical
basis for the reliability analyses outlined above. In addition, the quanti-
tative analysis tools developed for this project are described.
3.2 Definition of Reliability
Throughout this study the reliability of an SCS will be expressed in
terms of the probability of violation of air quality standards anywhere within
the region of influence of the controlled source. This definition has several
advantages over other proposed definitions. First, the definition is consistent
with the proposed EPA definition in the Federal Register. Second, the definition
is consistent with the wording of many of the air quality standards which re-
quire violations "not to be exceeded more than once per year". Finally, the
definition is generally applicable for any source which satisfies the EPA re-
quirements for operation of an SCS.
Two other proposed definitions were rejected for this study. One possible
definition of reliability involved a measure of the size of the geographical
area over which violations will occur. This definition would be useful if the
magnitude of the effects of S0_ were related to area; e.g. plant damage or
other economic, loss. This definition is not generally applicable when con-
sidering air quality violations as detrimental to human health. Another
proposed definition of reliability is the frequency of violations at a repre-
sentative location. This definition requires the subjective decision of
3-2
-------
determining a representative location from which reliability is measured.
Both of these definitions were rejected as less useful than the proposed
definition.
The specification that an air quality standard not be violated "more
than once per year" has been interpreted in probabilistic terms to require
that the projected frequency of occurrence of violations of that standard
be less than once per year. This probabilistic interpretation permits a
quantitative analysis of reliability.
3.3 Probability Concepts Relevant to an SCS Analysis
Throughout this study, the concept of a frequency distribution will be
used. A frequency distribution is a representation of the fraction of the
time a variable quantity assumes each of the possible values in its range.
A frequency distribution of ground level concentrations downwind of a source
provides much information about the characteristics of the source emissions.
The analysis model which is presented in this section is .based upon
studies of the frequency distributions of air quality levels from point
sources whose contributions dominate the concentration fields. Figure
3-1A illustrates two typical concentration distributions. The first would be
the case of a single receptor where the wind is often blowing in a direction
other than from source to receptor so concentrations are generally near zero.
The second is representative of a distribution of highest concentrations
at any one of a network of receptors around a source. In the latter case,
maximum concentrations near zero are less likely.
The value C has been designated on the abscissa of the maximum concen-
tration graph to indicate the value of some air quality standard. The sum of
the frequencies of occurrence of all concentration categories greater than C
is the fraction of the time the air quality standard is expected to be exceeded
The value F on the ordinate of each graph has been designated to indicate
O
the permissible frequency of concentration values exceeding C . To satisfy
air quality standard the sum of the frequencies for values of concentration
to the right of C must be less than F . Thus, from a compliance view point,
a more useful distribution is the cumulative frequency distribution associated
with each of the frequency distributions discussed above. In this case, the
sum of the frequencies of all values greater than the abscissa value is
plotted as the ordinate. Figure 3-1B illustrates the cumulative frequency
distributions associated with the distributions of Figure 3-1A. The sum
3-3
-------
A-l)
Single Receptor
(A-2)
24 6 8 10 12 14 16
C
2 4 6 8 10 12 14 16
(B-l)
(A) Discrete Frequency Distributions
(B-2)
Single Receptor
Maximum Concentration
2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 16
C C Cs
(B) Discrete Cumulative Frequency Distributions
(C-l)
\
Single Receptor
(C-2)
246
8 10 12 14 16
C
"N,
--
\
\
1
\
\
\
\
\
Maximum Concentration
\
X.
^^
*---*.
i
1
-~«_
2 4 6 8 10 12 14 16
C °s
(C) Continuous Cumulative Frequency Distributions
Figure 3-1 Representative Frequency Distributions for Ambient Concentrations
from a Point Source. (Graphs A-l, B-l and C-l apply to a single
receptor; Graphs A-2, B-2 and C-2 apply to the maximum concentra-
tions from a network of receptors.)
3-4
-------
over the frequency distribution of occurrence for concentrations greater
than GS can now be read directly from the ordinate of the graph.
It is likely that a continuous range of concentration values is possible
and that the step function presentation of the cumulative frequency distribution
can be replaced by a smooth function as illustrated in Figure 3-1C. Using
the graphs of Figure 3-1C, it can be simply stated that the goal of any control
procedure is to reduce the locus of F at the abscissa value C below the
dashed line representing F . When it is obvious which frequency distribution of
of the three presented in Figure 3-1, is being discussed, the term distribu-
tion will be used for convenience.
The cumulative frequency distribution can be used to illustrate the
effects of any control procedure. Figure 3-2A represents a hypothetical
distribution of maximum ground level concentrations. Since the locus of F
is above the dashed line at C = C , the source is in violation of standards.
Assume the graph of F represents the uncontrolled conditions. Direct
application of a constant emission control which reduces emissions uniformly
by 50% (say, changing from 2% to 1% sulfur fuel or installing 50% efficient
removal devices) would move every value of F from the abscissa value C to the
abscissa value C/2 to yield the graph illustrated in Figure 3-2B. The graph
of F has been reduced, as required, below F to satisfy the air quality require-
ments. More generally, we could have effectively moved every value of F from C
to B»C where $ is any value less than 3 such that F(C /3 ) £ F . The resulting
value of F*(C) = F(C/g).
An alternative to a rigid emission control system such as discussed above
would be an SCS capable of changing the tail of the graph to F to reduce
F* to values below F for C ^ C . Figure 3-2C represents some emission
reduction B(C) which reduces emissions by exactly the amount required to
achieve the standard all the time. In practice, no system will be so re-
liable. It is more likely that some fraction of the attempts to eliminate
concentrations greater than C will be unsuccessful.
o
Figure 3-2D represents a realistic frequency distribution resulting from
the actual operation of an SCS. The curve of F* is not different from the
curve of F at the low concentration end of the graph. This is in contrast to
the constant emission reduction case of Figure 3-2B where concentrations
well below standards before controls are reduced. The tail of the F* curve
3-5
-------
\
Maximum Concentration
-Uncontrolled F
*
LL
2 4 6 8 10 12 14 16
2 4 6 8 10 12 14 16
a)
Supplementary Control
X
- Controlled F &
Uncontrolled F'
Controlled
2 4 6 8 10 12 14 16
C °s
Supplementary Control
Controlled F £»
Uncontrolled F '
^f^Confra
ontrolled
24 6 8 10 12 14 16
C) d)
Figure 3-2 Representative Cumulative Frequency Distributions for Maximum
Concentrations under Conditions of a) no emission controls,
b) absolutely reliable constant emission controls, c) absolutely
reliable supplementary emission controls, d) realistic
supplementary emission controls.
3-6
-------
of Figure 3-2D is below the dashed line for values of C > C as required.
The values of C > C would be the result of uncertainty (errors) in the
o
operation of the SCS, but their frequency is so low that a reliable SCS
is still maintained.
The analysis of SCS systems which follows is based upon consideration
of modificatipns to the frequency distributions as discussed in this section.
3.4 Design of an Analysis Scheme to Measure the Reliability of a
Supplementary Control System
3.4.1 Assumptions and Definitions
The following analysis requires that an SCS has been developed for the
source under study which is designed to be consistent with EPA regulations.
Therefore, it is assumed that a reliable monitoring system exists and that
control of all air quality standards is assured if each component of the
SCS were without error. Methods of examining monitor network reliability,
the affects of air quality standards, and other elements of SCS design are
discussed in other sections of this study. The purpose of the analysis
scheme developed below is to verify and update the performance of a control
strategy.
The Reliability Analysis of SCS Methods embodied in the ERT software
package PROBL generates the frequency distribution of maximum ground-level
concentrations due to the source under study withi'or without emission controls,
In particular, the program can generate the frequency distribution for a
wide variety of possible SCS schemes.
The main assumptions of the analysis scheme are the following:
A single source of S02 is responsible for observed concentration
levels;
Without an SCS, emissions are independent of meteorological
conditions;
With an SCS, emissions are controlled according to rules which
depend on predicted meteorological conditions;
Error in prediction (as defined later) is independent of meteoro-
logical conditions. (Section 5.3 discusses how this assumption
can be relaxed.)
3-7
-------
The ERT PROBL technique can be easily modified for application to some-
what less restrictive conditions, but this complication is beyond the inten-
tions of this reliability study.
Consider the following definitions relevant to understanding the
model:
c(x,t): concentration at time t and location x
C(t): max c(}c,t); maximum concentration over all x locations at time t
x_: downwind location of C(t)
C : air quality standard or SCS threshold
Q(t): emission rate without SCS
M(t) : meteorological function relating the maximum concentration C(t)
to source emission rate Q(t) which will include the effects of
stack height, wind conditions, mixing depths or any other
pertinent meteorological inputs
R(t) : the error ratio of concentration prediction defined as follows:
With or without an operating SCS, the observed maximum concentration C
o
is related to the actual emission rate Q through the meteorological
function M as follows:
CQ = Q M at time t
With an operating SCS, the corresponding maximum predicted concentra-
tion is related to the actual emission rate Q and the meteorological
function M (defined above) through the Error Ratio R as follows:
Cp = Q M R at time t
From the above ; the error ratio can be defined as
R = C /C
P o
The observed maximum concentration under SCS control is given by C = Q .
\~ C
where Q is the SCS controlled emission rate determined from the forecast
concentration C .
P
The value of Q depends on the SCS control strategy being used. Two
examples of possible control strategies are:
3-8
-------
1) Fuel Switching
n -/Q if
(Strategy 1)
where 3 is a constant (less than one) which depends on the nature of the
fuels. A switch from 2% sulfur fuel to 0.5% sulfur fuel means 3 = 0.25
The threshold parameter y is a function of the air quality levels attempted
to be maintained.
2) Process Curtailment
if Cp « Y (Strategy 2)
if C > Y
In general, the threshold Y is set below standards to provide a margin of
safety.
The functions Q, M, and R require more careful description. Q(t) can
be a well defined quantity, if emission monitors are used or if emission
rate is simply related to the production rate or to the load of the plant
under study. If the plant emissions are not monitored some engineering
estimates of the frequency distribution of Q can often be made from pro-
duction or process control information.
The function M(t) can be determined in various ways. For an operation
with an extended historical air quality monitoring record, M(t) could be
estimated from the ratios of measured maximum concentrations C (tj and
known emission rates Q(t). Where a shorter monitoring record exists with
an extended meteorological data bank, M could be the output of a statistical
model: e.g., M = a-.ni., * a0m0 + -- + a m where m , m , m are meteorologi-
0 1122 nn 1 i n
cal parameters such as stability or air mass characteristics and a , a^, a
are regression coefficients determined from the available air quality data.
By knowing the distributions of the m. 's, the statistical data of the shorter
monitoring period can be combined with the longer period meteorological dis-
tributions. This is frequently the case since meteorological data has been
collected at many locations by the National Weather Service for periods up
to a century. Finally, where little site monitoring data exists, outputs from
3-9
-------
gaussian-diffusion or other air pollution models can be used in conjunction
with the meteorological data to construct M.
The function R(t) may be determined from historical real-time monitoring
and forecasting data. Since the function R will depend upon the unique fore-
casting difficulties for each SCS scheme, initial estimates of R must be
evaluated during the design and initial testing of the SCS. The upgrading
of an SCS as time proceeds will involve periodic re-evaluations of this
function.
Given the functions Q(tJ, M(t) and R(t) over an appropriate averaging
time period, frequency of occurrence distributions can be readily derived
for the various magnitudes of the observed or determined values of Q, M and R.
Thus, the time history data is transformed into a frequency of occurrence
distribution. For purposes of future estimations, the frequency of occurence
distributions become expected probability density functions.
When the frequency distributions for Q, M and R as defined above are
determined, it will be shown in the next subsection that the analysis scheme
will be capable of generating the following basic information:
1. The number of violations to be expected without and with any
SCS scheme.
2. The percentage of production lost if the SCS scheme is a load
reduction program.
3. The percentage use of high and low sulfur fuel in a fuel switching
SCS program.
4. The dependence of the number of violations, production lost and/or
percentage use of high and low sulfur fuel upon the implementation
threshold (Y), model calibration, meteorological forecasting skill,
and/or the difference in sulfur content of the two switching fuels.
The exact mathematical reasoning leading to the above conclusions is given
below.
3.4.2 The Mathematical Description of the SCS Analysis Scheme
Define P as the probability density function for the variable X.
A.
Then the probability of the variable X having a value between a and b is
3-10
-------
Assume that there exist probability density functions for M and Q, and we
wish to generate a frequency distribution for C when no SCS is operating.
If A is any concentration value, e is a variable and Q and M are independent
of each other and random variables, then
PC(C = A) = [PQ ( Q = e) - PM (M = A/e) +
PQ ( Q = 2e) PM C(M = A/2e) +
PQ ( Q = nej . PM (M = A/ne) + 1 ,Ae
or, in the limit as Ae approaches 0
CO
P (C = A) = j PQ ( Q = £) PM CM= A/?) d?
or
pc
Expressing the operator above by *,
Pr = PM * Pn
L M Q
This equation states that the probability density function for maximum
ground-level concentrations can be derived from the convolution of the
probability density functions for M and Q. Therefore, the frequency dis-
tribution of ground-level concentrations for an uncontrolled plant can be
determined from determinations of M and Q.
Once Pf is known, the graphs corresponding to Figure 3-1B can be dis-
played, and the probability of violating standards is directly known.
Consider next the case when the SCS is operating. In this case >
C = Q M where subscript c denotes the functional value when the SCS
c c
is operating. Q is no longer independent of meteorology since the opera-
tion of the SCS depends on meteorological forecasting.
3-11
-------
Pn will, therefore, also be generally dependent upon PM and will vary
for different control strategies. For computer solutions to the correlated
integration, the dependence of these quantities upon each other can be
readily simulated.
Assuming that the error ratio R is independent of M and of Q and given
P , P , and P it is possible to use the control strategy rules for deter-
R M (^
mining Q to numerically evaluate Pf under the SCS control.
In this case C = M Q R. The value of Q is determined in each
case from the predicted value of concentration C and from the strategy
(e.g., strategy 1 or 2). From the resulting distribution of Q , the value
of Pr is obtained from the equation
VjC
PCc = PQc * PM
Note the parallel nature of this equation and the equation for Pr.
L»
This means that the existing frequency distribution of ground level con-
centrations for a plant can be determined from archived measurements of Q
and M, and from records of air quality forecasting accuracy during operational
use of the SCS to determine R.
In practice, the determination of Pr is achieved by assuming that the
C
values of Q and M are quantified so they can assume only a finite number of
values. Therefore, P.. and P represent probabilities rather than probability
densities. The integral is then replaced by a numerical scheme which is
solved on a digital computer. Similarly for the generation of Pp , the function
CiC
R is assumed to be quantified and the probabilities of the quantified values of
Pr are determined numerically. The numerical scheme employed by the ERT
1_
-------
4. RELIABILITY OF THE COMPONENTS OF A SUPPLEMENTARY
S02 EMISSION CONTROL SYSTEM
4.1 Introduction
The reliability of a supplementary control system depends in a complex
way on the reliability of the components of the system. It would be most
desirable to describe system reliability in a strictly quantitative manner
as a function of some measure of reliability of each SCS component.
Because components interact in a complex manner, however, a well defined
measure of component reliability is not generally possible.
Every SCS is considered to have four generalized components in which
uncertainty can exist. These components are: (1) air quality monitoring;
(2) meteorological forecasting; (3) emissions forecasting; and (4) air
quality modeling. The meaning of the first three components is self-evident.
By air quality modeling, we mean the algorithms and methodology which are
used to relate meteorological inputs, emission rates, source data, terrain,
and location factors to current and future air quality in the vicinity of the
source. Each of the SCS components identified above are considered individually
in this section with respect to their effect on overall SCS reliability.
4.2 Assessment of Air Quality Monitoring Reliability
Every supplementary control strategy must have a monitoring network
to verify that the required air quality is being maintained through the
operation of the SCS. Also, real-time air quality monitoring must be
available as one input to the decision to control emissions. In addition to
these uses of monitoring data, they are also used during the definition
phase of the SCS and later when the forecast models are calibrated and
periodically upgraded. It is essential, then, to understand reliability
with respect to the data produced by a monitoring network.
Monitoring network data will be employed during every important stage
of an SCS: (1) development of the system; (2) operation of the system;
and (3) historical review of the system. Reliability, however, may be
usefully defined without considering the specific application of the moni-
toring network.
4-1
-------
The following sources of uncertainty in a monitoring system will con-
tribute to a degradation of system reliability:
Instrumentation accuracy limits
Percentage data capture statistics
Information transfer errors
Insufficient and/or inappropriate sampling locations.
Section 4.2.1 provides a brief discussion of the first three items.
The last item is addressed in more detail in Section 4.2.2. In this section
a specific methodology was devised to permit identification and ranking
of the most important monitoring locations from an air pollution meteorology
standpoint for any given isolated source.
4.2.1 Sampling and Information Transfer Errors
Proper choice of SO monitoring instruments depends on many factors.
For purposes of evaluating a system for an SCS application, it is important
to consider sensitivity, lag time and response time, interferences, accuracy,
calibration drift, and maintenance requirements. A useful reference on
these factors is a recent review by Forrest and Newman (1973).
For SCS applications concerned with compliance with short-term and long-
term standards, the lowest detection limit should not detract from assessing
this compliance.
This general requirement translates into the specific requirement that
monitors must provide SO values over the proper averaging times. Certain
techniques such as flame photometry provide instantaneous SO readings from
which any longer time average can be constructed. Wet chemistry techniques
such as the West-Gaeke standard procedure usually require that time averages
of 6 to 24-hours be directly determined. Control of shorter time averages
must, in this case, be indirect and require knowledge of the statistical
relationships of these averaging times to those which are directly measured.
The lag time of a monitoring network is the time between the occurrence
of the concentration and the time that this value is displayed for use by
SCS personnel. With telemetered data, short term averages (say, 2 minute
averages or instantaneous concentration values less than 2 minutes old)
are usually available for examination before a 1-hour or 3-hour averaging
4-2
-------
period has transpired. In these cases, the lag time is no constraint on
the system. For systems which require analysis of strip charts, manual
data handling, or chemical analysis, the lag time between collection and
display of the data severely constrains the potential uses of the system.
The percentage of useful data capture depends upon the combined down-
time of the sensors and associated data capture and transmission components.
Sensor downtime includes time periods of instrumentation calibration, main-
tenance as well as identifiable datasets of inaccurate measurements. A well
designed system will attempt to minimize these sensor downtime contributions
by providing automatic instrument calibration, remote sensing of possible
instrument malfunctioning and generally, remote control of the instrumenta-
tion. Thus, real-time monitoring and telemetry of information provides
mechanisms for substantially enhancing data rates. If the system involves
telemetry such as telephone line usage, the data capture rate will depend
additionally upon the downtime of this telemetry system and the remote
recording devices. If the system requires any real-time data processing,
the downtime of the data processing equipment must also be considered.
Figure 4-1 is an illustration of the percentage of data capture achieved
through operation of a real-time ERT monitoring network, AIRMAP, with results
telemetered to a computer which averages and displays the data. Of particular
interest is the nearly steady improvement in uptime during the initial
few months of operation. This improvement was associated with a reduction
of sensor downtime by establishing continuously \improving procedures to
identify causes of data loss and methods to re-establish instrument calibra-
tions. Furthermore, note the step-up in system reliability in February,
1973. This improvement is attributed almost exclusively to an improved
service agreement for the dedicated telephone lines which relay the data
to the network central. Figure 4-2 breaks down the composite results of
Figure 4-1 into the reliability of each of the 4 networks which comprised
AIRMAP during the 22-month period.
4.2.2 Analysis to Determine a Distribution of Monitors Sufficient
to Represent the Field of Downwind Concentrations
The question of sufficient number and spacing of monitors is difficult
to assess in general because every site has peculiar meteorology, terrain,
and land use. Monitor locations should, in general, be chosen to monitor
4-3
-------
1972
1973
5"
"c
o
100
90
80
70
60
50
40
30
20
10
Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 4-1 ERT's Combined AIRMAP Networks Real-Time Data Retrieval
(Weighted by number of sensors in each network)
4-4
-------
1972
1973
i
on
"s.
9)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 4-2 ERT's AIRMAP Real-Time Data Retrieval for Each Network
-------
the highest concentrations in the vicinity of the source. They should not
be influenced by nearby obstructions of minor local sources. Some help
in the practical problem of field location is given by the Federal EPA in
that inaccessible areas need not be monitored.
The following analysis is designed to assist in determining the best
distribution of monitors for SCS applications. Furthermore, the analysis
provides an estimate of the percentage of air quality violations expected
to be directly observed by any monitoring network. Through this analysis,
the improvement in the monitor network anticipated by the addition of one
or more monitors can be assessed and the concomitant improvement in SCS
reliability can be weighed against the increased cost and effort.
The analysis requires a dispersion model appropriate to the point source
under study. The following example employs the ERT Point Source Diffusion
Model (ERT PSDM) as described in the Appendix. The source under investigation
includes one 612 foot stack and one 828 foot stack at a single location.
The emission rate and other source statistics are those of the two corresponding
stacks at the ASARCO smelter in El Paso, Texas, under full load conditions.
Emissions from the zinc stack or from other smaller sources are not included,
however; so the analysis is not proposed to apply directly to the ASARCO
operation. The frequency of occurrence of particular meteorological conditions
is derived from a stability-wind rose from the El Paso Airport. Since no
effort is being made to duplicate the ASARCO operation, terrain effects have
not been included in the analysis. Their inclusion, however, will not change
the basic approach.
As a first step in establishing monitor locations, it is instructive
to illustrate the maximum ground-level concentration as a function of down-
wind distance for the entire range of weather conditions in a single direction.
Figures 4-3 through 4-10 contain these illustrations. The meaning of the
weather condition classes is as discussed in the description of ERT PSDM in
the Appendix. Clearly, high concentrations are possible at downwind distances
from less than one kilometer to as far as 25 kilometers. It is also clear
that the widths of the peaks are narrow for some weather conditions (coning
plume) and quite broad for other conditions (inversion conditions and fumiga-
tion) . There is great latitude for monitor locations to observe the latter
peaks, while only a limited range of locations is possible .to observe the
former narrow peaks. If it is desired to observe only the several highest
4-6
-------
peaks, Figures 4-3 through 4-10 provide enough information for monitor
placement in a single direction from the source.
The simple analysis above, unfortunately, is insufficient for most
monitor siting efforts. Local climatology is responsible for variable fre-
quency of occurrence of each weather condition for each wind direction.
The highest concentration peak may be less important than a lower peak which
has a significantly higher probability of occurrence. Also, by strategically
locating a monitor it can detect peaks from more than one weather condition.
Since it is impossible to monitor the peaks for all possible weather condi-
tions, it is important to develop an analysis technique which can consider
the above complications.
First, the analysis should consider a concentration threshold; that is,
peaks which indicate concentrations below some minimum significant value should
be eliminated from consideration.
Second, perfect resolution of the concentration peak is meaningless
because the real world does not experience only a discrete number of possible
weather conditions. Each modeled peak represents just one of a range of
possible peak values for a single weather condition. Location of a monitor
at any location where the concentration is within 10% or possibly 20% of its
peak value is a more reasonable criteria for monitoring the weather condition
of interest. This desired resolution might vary for each application;
Third, the emphasis of the monitoring network will probably be to
observe with monitors the highest possible fraction of all violations
which occur.
Finally, there is probably a practical limit to the number of monitors
which can be employed in the network.
The analysis scheme which follows considers each of the four pre-
ceding points:
A concentration threshold is assumed;
The resolution of the peaks is specified;
The fraction of observed violations is determined for each
proposed network;
The minimum acceptable or maximum affordable number of monitors
can be specified, and an estimate of the optimum location of the
monitors can be inferred.
4-7
-------
ENVIRONMENTAL RESEARCH + TECHNOLOGY. INC.. LEXINGTON MA.*« ERT DIFFUSION STUDY
BOTH STACKS AT FULL LOAD STACK HT = 186.5 232.4 M
STACK ELEV * 0.0 H MSL FUEL RATE * 0.0 GPH* 0 « 0.96E 10 MICROGM / SEC
HEAT RATE * 5261188. CPS*SULFUR CONTENT -0.0 PERCENT* AMB. TEMP « 50.0 DEO f
HEAT CONTENT * 149000.0 BTUPG** WEATHER CONDITION 1 SECTOR 1
''°*°
0 IS
DOWNWIND DISTANCE
10
KM
14
Figure 4-3 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 1.
4-8
-------
««TU -,»*- IT «- *« TECHNOLOGY. INC.. LEXINGTON MA.«« ERT DIFFUSION STUDY
BOTH STACKS AT FULL LOAD STACK HT = 186.5 252.4 M
SJA5K«EbEV " 0.0 M MSL FUEL RATE - 0.0 GPH« 0 - 0.96E 10 MICROGM / SEC
"I'M 5^1^* 5261188. CPS«SULFUR CONTENT » 0.0 PERCENT* AMB. TEMP - 50.0 DEG F
HEAT_CONTENT - 149000.0 BTUPG'* WEATHER CONDITION 2 SECTOR 1
t«
DOWNWIND DISTANCE C KM
Figure 4-4 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 2.
4-9
-------
ENVIRONMENTAL RESEARCH + TECHNOLOGY. INC.. LEXINGTON MA.«« ERT DIFFUSION STUDY
BOTH STACKS AT FULL LOAD STACK HT = 186.5 292.4 M
STACK ELEV * 0.0 M HSL FUEL RATE - 0.0 GPH« 0 * 0.96E 10 MICROGM / SEC
HEAT RATE « 3261188. CPS«SULFUR CONTENT -0.0 PERCENT* AMB. TEMP « 50.0 DEC F
HEAT CONTENT * 149000.0 BTUPG«* WEATHER CONDITION 3 SECTOR 1
WIND
1* !
DOWNWIND DISTANCE C KM
t«
Figure 4-5 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 3.
4-10
-------
|NVIRONMENTAL RESEARCH + TECHNOLOGY. INC.. LEXINGTON MA.«« ERT DIFFUSION STUDY
BOTH STACKS AT FULL LOAD STACK HT = 186.9 252.4 M
STACK ELEV - 0.0 M MSL FUEL RATE - 0.0 GPH« 0 0.96E 10 MICROGH / SEC
HEAT RATE - 5261188. CPS«SULFUR CONTENT -0.0 PERCENT* AMB. TEMP - 50.0 DEC F
HEAT CONTENT « 143000.0 BTUPG»« WEATHER CONDITION 4 SECTOR 1
) -MINUTE AVERAGE SULFUR DIOXIDE CONCENTRATION CPPM3
0«««»0
§ 5 ! ! I if 5 ! !
(0
«
1 IS It M M
WIND
SPEED
LEGEND
CMPH)
* 2
= 4
a = 6
* = 8
i = 10
= 12
Figure 4-6 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 4.
4-11
-------
*.
ENVIRONMENTAL RESEARCH + TECHNOLOGY. INC.. LEXINGTON MA.«« ERT DIFFUSION STUDY
BOTH STACKS AT FULL LOAD STACK HT * 186.9 292.4 H
STACK ELEV * 0.0 M MSL FUEL RATE * 0.0 GPH* 0 - 0.96E 10 MICROGM / SEC
HEAT RATE « 5261188. CPS«SULFUR CONTENT -0.0 PERCENT* AMB. TEMP - 50.0 DE6 F
HEAT CONTENT * 149000.0 BTUPG** WEATHER CONDITION 5 SECTOR 1
WIND
SPEED
LEGEND
CMPHD
« 2
= 3
o = 4
* = 5
i = 6
= 7
DOWNWIND DISTANCE
Figure 4-7 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 5.
4-12
-------
ENVIRONMENTAL RESEARCH + TECHNOLOGY. INC.. LEXINGTON MA.«» ERT DIFFUSION STUDY
BOTH STACKS AT FULL LOAD STACK HT = 186.5 232.4 M
SJAQK«ELEV " 0.0 M HSL FUEL RATE - 0.0 GPH« 0 - O.S6E 10 MICROGM / SEC
HEAT RATE - 5261188. CPS«SULFUR CONTENT -0.0 PERCENT* AHB. TEMP * 50.0 OEG F
HEATJCONTENT - 149000.0 BTUPG" WEATHER CONDITION 6 SECTOR 1
t«
DOWNWIND DISTANCE C KM D
Figure 4-8 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 6.
4-13
-------
ENVIRONMENTAL RESEARCH + TECHNOLOGY. INC.. LEXINGTON MA.*« ERT DIFFUSION STUDY
BOTH STACKS AT FULL LOAD STACK HT « 186.9 292.4 M
STACK ELEV « 0.0 M MSL FUEL RATE « 0.0 GPH« Q - 0.96E 10 MICROGM / SEC
HEAT RATE * 5261188. CPS«SULFUR CONTENT » 0.0 PERCENT* AMB. TEMP * 90.0 DEC F
HEAT CONTENT * 149000.0 BTUPG** WEATHER CONDITION 7 SECTOR 1
'" WIND
It !
DOWNWIND DISTANCE C KM
TC
t«
Figure 4-9 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 7.
4-14
-------
, TECHNOLOGY. INC.. LEXINGTON MA.«« ERT DIFFUSION STUDY
AT FyLL LOAD STACK HT = I86-5 252.4 M
. ~.0<0 M "S1- FUEL RATE * °-° GPH* 0 0.96E 10 MICROGM / SEC
HEAT RATE » 5261188. CPS«SULFUR CONTENT =0.0 PERCENT* A MB. TEMP « 50.0 DEG F
HEATJCONTENT - 149000.0 BTUPG«* WEATHER CONDITION 8 SECTOR 1
It !
DOWNWIND DISTANCE C KM
t«
Figure 4-10 Three-hour Average Ground Level Concentrations for Various
Wind Conditions of Weather Condition 8.
4-15
-------
The example analysis which follows corresponds to the graphs displayed in
Figures 4-3 through 4-10.
We first consider monitor placement in just two wind directions:
north (#1) and west-southwest (#12). These are highly probable wind direc-
tions according to the El Paso wind climatology. For this analysis, we
consider a concentration threshold of 0.25 ppm and we consider that a value
within 20% of peak concentration is sufficient to resolve the peak.
Table 4-1 summarizes the 30 peaks (from a total of 48 possible weather
conditions) which must be monitored to capture all concentrations greater
than 0.25 ppm within 20% of the peak. The first column indicates the
stability and wind speed category respectively associated with the peak.
The second column indicates the best monitor location to observe this
peak based on 40 previously defined possible locations. Column 3 contains
the maximum concentration expected for this weather condition. The fourth
and fifth columns indicate the beginning and end of the optimum range of
monitor sites. Finally, column 6 contains the fraction of the total time
this peak is expected to exist. From Table 4-1 alone, it is clear that a
monitor at 8 km from the source should observe the highest concentration
*-CMAX = 1-4731)> a monitor at 4.5 km from the source should observe the peak
with the highest frequency (0.00162) of cases above the threshold, and that
a monitor between 2 km and 4 km from the source will observe the highest
number of peaks from all weather conditions.
Table 4-2 illustrates the same information for wind direction 12.
Because the terrain is assumed flat, only the final column of Table 4-2
differs from Table 4-1. For this wind direction, the monitor location to
observe the peak with the highest frequency is nearer the source than for
wind direction 1. These tables indicate that some peaks occur so unfre-
quently that no appreciable improvement in SCS reliability would be gained
by monitoring for these peaks. Similar tables could have been generated
for all 16 wind directions for use in monitor site determination.
If the monitor network goal were to observe the highest possible fraction
of all cases above 0.25 ppm, Table 4-3 provides a ranked listing of up to 25
monitor locations considering all wind directions for the previous example.
The location is defined by an azimuth (wind direction) and a radial distance
(distance downwind). The fraction of all violations of the 0.25 ppm threshold
observed by the single monitor is contained in column 4, and the improvement
in the fraction of observed violation is measured in the cumulative capture
4-16
-------
TABLE 4-1
Summary of Monitor Siting Criteria by Weather Condition for
Wind Direction 1.
RESOLUTION WITHIN 20 PERCENT
CONCENTRATION THRESHOLD =0.25 ppm
SUMMARY OF INDIVIDUAL WEATHER CONDITIONS, WIND
DIRECTION * 1
IST,J
1
1
1
1
1
2
2
2
2
5
5
5
5
5
5
7
T
7
S
8
8
8
8
8
,2
,3
,«
,5
,6
,3
»4
.5
,6
,1
,2
,3
,4
,5
,6
,1
,2
,3
»4
,5
fb
,«
.5
.6
,\
,2
,3
,4
,5
,6
LOCATION
4000,
3500.
3000.
2000.
1SOO.
4500.
4000,
3500,
3000,
6000.
4500.
3500.
3000.
2000.
1500,
12000,
6500,
4500.
4000,
3500,
3000.
14000.
12000,
11000,
8000.
7500.
9000.
11000.
13000.
14000,
CMAX
0,3031
0.3605
0,4024
0.4655
0.5101
0.2977
0.3284
0.3442
0.3533
0.4567
0.6110
0.7293
0.8142
0.9327
1.0204
0.2931
0.4871
0.5985
0.6615
0.6931
0.7099
0.2542
0.2709
0.2794
1.4731
1.1460
0.7811
0.5841
0.4573
0.3788
XI
2901.53
2102.81
1503.29
1245.64
1065.12
3205.46
2613.84
2146.56
1860,17
4200,52
2916.83
2141.68
1529,44
1248.49
1065.35
8330,66
4427,75
3218.43
2630.65
2165.56
1867.41
9504.83
8145.88
7262,86
7877.03
7373.02
8866.54
10717.78
12710.35
13708.61
X2
6386.11
5027.86
4237.49
3431.86
2934.46
7193.17
5861.71
5078.72
4528.68
9635.39
6699.80
5244,27
4409,16
3581.43
3055.86
19922.85
10414.48
7489.54
6074.84
5257.32
4689,50
23454.39
19942,64
17677,70
19632.16
15988.51
17556.64
19206.96
21191.44
22313.70
FREQUENCY
0.00075
0.00071
0.00066
0.00066
0.00126
0.00162
0.00137
0.00129
0.00096
0.00004
0.00002
0.00001
0.00001
0,00001
0.00003
0.00001
0.00002
0.00003
0.00003
0,00003
0,00002
0.00007
0.00009
0.00031
0.00019
0.00015
0.00017
0.00010
0.00007
0,00004
THE FREQUENCY OF MAXIMIM CONCENTRATIONS LESS THAN 0.25 IS 0.06686
4-17
-------
TABLE 4-2
Summary of Monitor Siting Criteria by Weather Condition for
Wind Direction 12.
RESOLUTION WITHIN 20 PERCENT
CONCENTRATION THRESHOLD - 0.25 ppm
SUMMARY OF INDIVIDUAL WEATHER CONDITIONS, HIND
DIRECTION #12
I3T,J
I
2
2
2
5
5
5
5
5
9
6
7
7
7
a
a
a
a
a
a
,2
,3
,4
.5
,*
,3
,4
,5
,6
,1
,2
,3
,4
,5
,6
.1
,z
,3
,4
.5
.6
,<
,5
»6
,1
,2
,3
,4
,5
,6
LOCATION
4000,
3500,
3000.
2000.
1500.
4500.
4000.
3500,
3000.
6000,
4500.
3500.
3000.
2000.
1500.
12000.
6500,
4500,
4000.
3500.
3000,
14000.
12000,
11000,
aooo.
7500.
9000,
11000,
13000,
10000,
CMAX
0.3031
0.3605
0.1024
0.4655
0.5101
0.2977
0.32BU
0.3442
0.3533
0.4567
0.6110
0,7293
0.8142
0.9127
1.0204
0.2931
0.4871
0,5985
0.6613
0,6931
0,7099
0.2542
0.2709
0,2794
1.4731
1.1460
0,7811
0.5841
0,4573
0,3788
XI
2901,53
2102.81
1503.29
1245.64
1065.12
3205.46
2613,84
2146.56
1860.17
4200.52
2916,83
2141,68
1529.44
1248.49
1065.35
8330.66
4427.75
3218.43
2630.65
2165.56
1867.41
9504,83
8145.88
7262.86
7877,03
7373,02
8866.54
10717.78
12710,35
13708,61
X2
6386.11
5027.86
4237.49
3431.86
2934.46
7193,17
5861.71
5078.72
4528.68
9635.39
6699,80
5244.27
4409.16
3581.43
3055.86
19922.85
10414.48
7489.54
6074.84
5257.32
4689.50
23454.39
19942.64
17677.70
19632.16
15988.51
17556.64
19206.96
21191.44
22313.70
FREQUENCY
0,00054
0.00063
0,00073
0.00073
0,00264
0.00082
0.00250
0,00305
0.00525
0.00003
0.00001
0.00001
0.00001
0.00001
0.00005
0.0
0.00001
0,00002
0.00005
0.00006
0.00011
0.00004
0.00005
0.00055
0.00006
0.00005
0,00006
0,00006
0.00006
0.00003
THE FREQUENCY OF MAXIMIM CONCENTRATIONS LESS THAN 0,25 IS 0.06578
4-18
-------
TABLE 4-3
Optimum Monitor Locations Ranked by Expected Frequency of
Monitored Values Exceeding the Concentration Threshold with
the Desired Resolution.
RESOLUTION WITHIN 20 PERCENT
CONCENTRATION THRESHOLD 0.25 ppm
OPTIMAL MONITORING LOCATIONS
MONITOR *
1
2
3
4
5
6
7
8
9
10
11
12
13
11
15
16
17
IB
19
20
21
22
23
24
25
WIND DIRECTION
9
6
11
12
10
7
6
13
2
5
3
14
1
4
IS
16
9
8
1
2
10
3
tt
1
f>
DISTANCE DOWNWIND
2916. S3
2916.63
2916,83
2916.83
2916.83
2916.83
2916.83
2916,83
3218.43
2916,83
3218.43
2916,83
3218.43
2916,83
2916.83
2916.83
4427,75
4427,75
4427,75
1065.35
4427,75
1065. J5
4427.75
1065,35
4427.75
TOTAL FREOUENCY OF MAXIMUM CONCENTRATIONS LESS THAN
FREOUENCY
0.02869
0,02065
0,01657
0.01638
0.01554
0.01463
n. 01 259
0.01207
0.01193
0.01031
0.00948
0,00i27
0,00818
0,00694
0.00476
0,00343
0.002b8
0.00209
0.001B9
0,00174
0.00160
0.00152
0,00134
0,00129
0.00129
0.21; TS 0.76710
CUMULATIVE FREQUENCY
0,02269
0.04134
0.0599J
0.07629
0.09183
0,10646
0.11905
0.13112
0.14305
0.153)6
0.16284
0.17111
0.17929
0.18623
0,19099
0.19442
0,19700
0.19909
0.20098
0.20272
0.20432
0.20584
0.20718
0,20847
0.20976
4-19
-------
frequency in column 5. For this example, the percent frequency of occurrence
of values above .25 ppm is 23.29%. Thus, with 25 receptors 90.06% (20.976/
23.29) of all concentration values greater than .25 ppm would be observed.
Clearly- the use of this many monitors would be a very expensive and yet not
especially reliable way of assuring compliance with standards. It is inter-
esting to note that by adding the 17th monitor to a network including the
best 16 locations, an order of magnitude less improvement in important data
capture is gained than when the second monitor was added to the first.
For comparison purposes, Table 4-4 provides a ranked ordering of up to
25 monitor locations for a network designed to observe the peak within 10%
for all concentrations greater than 0.25 ppm. Table 4-5 illustrates similar
rankings for a monitoring network attempting to observe all concentration
peaks above 0.1 ppm within 20% of the peak concentration. Finally- Table 4-6
ranks monitors set out to observe concentration peaks above 0.1 ppm within
10% of the peak.
From the previous examples, one can deduce several general results:
1. As the resolution is increased (i.e., within some percentage of
the maximum value) and the threshold for significant peak values
is decreased, the effectiveness of a monitoring network to observe
significant peak values decreases; or, alternatively, more sensors
are required to observe the same percentage of significant peak values,
2. For any reasonable number of sensors, resolution, and threshold
of significance, a relatively high percentage of significant
peak values will go unobserved.
3. Using monitoring alone as the only guide to decisions concerning
emission alterations in an SCS could result in a significant number
of undetected violations of the short-term standards given that
violations are expected to occur.
4.3 Assessment of Meteorological Forecasting Rteliability
4.3.1 Introduction
The purpose of every SCS is to reduce the occurrence of violations of
air quality standards by reducing emissions during periods when weather con-
ditions are not conducive to adequate dispersion of the pollutants.
4-20
-------
TABLE 4-4
Optimum Monitor Locations Ranked by Expected Frequency of
Monitored Values Exceeding the Concentration Threshold with
the Desired Resolution.
RESOLUTION WITHIN 10 PERCENT
CONCENTRATION THRESHOLD 0.25 ppm
OPTIMAL MQNIT'.JRING LOCATIONS
MONITOR *
1
Z
3
tt
5
6
r
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
21
24
25
TOTAL FRtQ'jENCY
WIND DIHECTION
V
8
12
11
10
7
2
6
13
5
3
9
1
'l4
8
-------
TABLE 4-5
Optimum Monitor Locations Ranked by Expected Frequency of
Monitored Values Exceeding the Concentration Threshold with
th« Desired Resolution.
RESOLUTION WITHIN 20 PERCENT
CONCENTRATION THRESHOLD =0.10 ppm
OPTIMA! MONIT'JRING L"C*TIONS
MONITOR *
1
2
1
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
25
24
25
OTAI. FREQUFNCY
HIND DIRECTION
12
11
14
1
11
2
9
8
8
5
6
9
11
12
10
7
1
4
2
10
6
7
15
11
5
DISTANCE DOWNWIND
9504.6)
9504.85
9504,85
9504.85
9504.8}
9504.83
2916.83
2916.81
9504. 8J
9504.85
9504,83
9504,81
2916.81
2916.81
2916.81
4200,52
9504,81
9504,81
4166,15
9504. HI
4166,15
9504,83
8866,54
?916,81
4166.15
HF MAXIMUM CONCFNTRATIONS Ltss THAN
FRE8UENCY
0.04986
0.04072
0,01462
0.02845
0.02412
0,02305
0,02269
0,02065
0.01BJ7
0.01801
0.016*4
0.01684
0.01637
0.01638
0.01554
0.01505
0.01362
0.01356
0.01316
0.01292
0,01280
0,01241
0.01221
0,01207
0.01126
0.10 IS 0.40145
CUMULATIVE FREQUENCY
0.04986
0.09058
0.12S20
0,15163
0.17775
0.20080
0.22349
0.24414
0.26251
0.28052
0.29746
0.31430
0.33067
0.34725
0.36279
0.37784
0.39146
0.40502
0.41818
0.43110
0.44390
0.45633
0.46854
0.48061
0.49187
4-22
-------
TABLE 4-6
Optimum Monitor Locations Ranked by Expected Frequency of
Monitored Values Exceeding the Concentration Threshold with
the Desired Resolution.
RESOLUTION WITHIN 10 PERCENT
CONCENTRATION THRESHOLD - 0.10 ppm
OPTIMAL MONITORING LOCATIONS
MONITOR »
1
2
3
u
5
6
7
8
9
10
11
12
13
14
15
16
17
IB
1«
20
21
22
23
24
25
TOTAL FREQUENCY
WIND DIRECTION
12
13
14
1
11
2
9
8
5
B
6
9
12
11
4
i
10
7
10
7
2
15
6
13
5
DISTANCE DOMNrilND
8200.12
6200.12
8200.12
10848.89
10858.89
10858.89
3606,05
10858,89
10858,89
3606.05
10858.89
10858.89
3606.05
1606,05
10858,89
10858,89
1606,05
3606,05
10858.89
10858,89
3606,05
8200.12
3606.05
3606.05
3606. OS
PF «*XIMUM CONCENTRATIONS LESS THAN
FREQUENCY
0,04586
O.OJ777
0,03196
0,02655
0.02214
0.02109
0.01700
0,01682
0,01621
0.01615
0.01972
0.01511
0,0137*
0.01285
0.01205
0,01191
0.01181.
0.01168
0.01141
0.01111
0.01106
0.01047
0.01010
0.00988
9.00872
0,10 IS 0,40145
CUMULATIVE FREQUENCY
0.04586
0,08363
0.11559
0.14214
0,16428
0.18537
0,20237
0.21919
0.23540
0,25155
0,26727
0.28238
0.29617
0,30902
0,32107
0,33298
0,34479
0.35647
0,36788
0,37899
0,39005
0.40052
0.41062
0,42050
0.42922
4-23
-------
Identification of these poor dispersion periods must be accomplished with some
advance notice since there exist practical limits to the speed with which
emission reduction orders can result in lower emissions at the top of the
stack. Furthermore, there is a significant "ventilation time" before the
emissions can travel from the stack to beyond the important influence distance
of the source. The requirement for advance warning of impending poor dis-
persive periods forces the SCS system to include some form of meteorological
forecasting. Some conditions, such as inversion breakup and high wind
downwash situations, particularly demand advance forecasting. Monitored air
quality levels alone would provide no advance warning that the fumigation or
downwash was impending, and their onset would foul the air before any curtail-
ment action would be effective.
The principal role of meteorological forecasting for an SCS is to provide
a basis for the appropriate SCS response to anticipated inadequate dispersion
conditions. After the SCS has been operative and meteorological forecasts
and observed concentration levels have been recorded, analysis of the data
can verify the meteorological conditions which accompany poor air quality,
and point to ways of improving the meteorological forecast system.
4.3.2 Criteria for Assessing Meteorological Forecasting Reliability
The following factors are essential to a meteorological forecasting
system and provide a framework with which to discuss the reliability of
this important SCS component.
The spatial and temporal scales of the forecasting procedures must
be appropriate for the requirements of the SCS.
The relationship between errors in meteorological parameter fore-
casts and errors in predicted concentration levels must be under-
stood.
Verification of all aspects of the meteorological forecasting
system must be a part of the SCS.
These criteria are considered in detail below.
4-24
-------
Spatial.and Temporal Scales of Forecasted Conditions
Meteorological forecasting for the estimation of air quality is cate-
gorized by space and time scales. Forecasting for time scales of 24 hours
or more requires the prediction of synoptic scale (hundreds to thousands of
kilometers) meteorological events. For example, it entails the prediction
of the movement and location of stagnating anticylcones with their associated
light winds and poor dispersion characteristics. Shorter time periods
require detailed forecasts on smaller spatial scales. In determining the
reliability of meteorological forecasting for an SCS, it is necessary to
consider forecasts on temporal scales of 1 to 24-hours and on spatial scales
of several thousand square miles to within a few thousand feet of the pollu-
tant source.
For some operations, the effects of an SCS decision to reduce emissions
can result in reduced downwind concentrations within 1-hour. If forecasting
personnel operate around-the-clock, then only 1-hour forecasts are required.
However, assume that a 6-hour lead time is essential prior to implementation
of the switch mode of the control system. This implies that meteorological
conditions must be forecast 30-hours in advance to control a 24-hour standard
the 6-hour implementation requirement plus the 24-hour forecast time. If
continuous meteorological forecasting support is available, then the forecast
meteorological parameters for the beginning of the 30-hour period are most
important as the forecast can be revised and the switch mode subsequently
implemented when non-predicted weather conditions are imminent. If only
one forecast is prepared daily, then the entire 30-hour prediction is
important.
The "weather" variables which the meteorologist must forecast are
those which influence the dilution capacity of the lower atmosphere.
As direct forecasts and measurements of the turbulent components of the
wind are frequently unavailable, related parameters become the forecast
requirement. These include wind speed and direction, static atmospheric
stability; cloudiness, precipitation, and mixing depth.
The reliability of any weather forecast is never perfect and in general
depends on several diverse and often highly variable factors. A 6-hour
forecast of cloud cover is usually more reliable than a 24-hour forecast of
the same event. Local effects are important. The onset of a sea breeze
circulation in coastal areas can be in opposition to the wind flow
4-25
-------
normally associated with a weather system where there is no large body of
water nearby. Similarly, the diurnal variation in temperature is different
in urban areas compared to rural areas.
Meteorological Forecasting Errors Vs. Air Quality Forecast Errors
The reliability of a forecast for air quality prediction is determined
by the reliability of forecasting the meteorological parameters which
dictate the ensuing air quality. First, consider wind speed and direction.
For an isolated source unaffected by terrain, the wind direction may be
unimportant as possible high pollutant concentrations may occur in any
downwind direction from the source. Two examples when wind direction is
important are plume downwash, which may occur with strong winds from
particular directions, and terrain modified winds which might produce high
concentrations at a particular critical location. The predictability of
wind direction is generally good, especially with well-defined synoptic
systems. The predictability decreases with time and is generally lower in
areas with complicated terrain features. When an anticyclone is over the
station, wind direction can be variable and, hence, difficult to predict.
Wind speed usually is more difficult to forecast. It varies diurnally
with high speeds during the day, when there is a transfer of momentum from
higher levels to the surface boundary layer, and with low speeds at night.
Wind speed depends on the intensity of the pressure gradient, insolation,
surface roughness, terrain channeling, and other local factors. The reli-
ability of wind speed forecasts also decreases with length of forecast
time.
If it is known that pollutant concentrations are particularly sensi-
tive to wind forecasts, then it is imperative to know the wind speed fore-
cast reliability when designing the conservatism of the SCS.
The stability of the lowest kilometer of the atmosphere broadly describes
its turbulent characteristics. An unstable atmosphere is characterized by
thermal convection, turbulence, and good mixing. A stable atmosphere is
characterized by weak turbulence and poor mixing. Atmospheric stability
depends on both the vertical wind and temperature structure. Thus, its
predictability depends on the predictability of both parameters.
Temperature measurements in the vertical are usually the best indicator
of stability- If continuous temperature measurements in the vertical (e.g.,
4-26
-------
on a meteorological tower) are available, then atmospheric stability is
usually quite predictable. Temperature typically varies diurnally, with
changing air mass and with local effects.
When a vertical temperature profile is unavailable, the the prediction
of stability is indirect and, hence, less reliabile. The stability, however,
may be estimated by the prediction of cloud cover, wind speed, type of air
mass over the region, ground cover (e.g., snow cover, proximity to large
bodies of water) and time of day. The reliability of these estimates of
atmospheric depends in part upon cloudiness forecasts. The prediction of
cloud cover varies widely with locale, time of year, time of day, and synoptic
weather patterns. Generally, cloud cover forecasts are unreliable for long
forecast times and are highly variable with active synoptic systems.
The atmospheric mixing depth is defined as the depth of the atmosphere
through which complete vertical mixing occurs. It is characterized by a
dry adiabatic temperature lapse rate. Its predictability depends on the
predictability of the maximum temperature, the vertical distribution of
temperature in the lowest few kilometers, and the presence and height of
subsidence inversions associated with synoptic scale anticyclones.
The afternoon mixing depth is the height of the intersection of the dry
adiabatic lapse from the surface maximum temperature with the observed
or predicted temperature lapse. Pollutants trapped in a shallow mixing
layer could result in high ground-level concentrations when the effluent
is thoroughly mixed throughout the layer.
The prediction of maximum temperature is routine and generally quite
reliable.. The reliability of a temperature forecast decreases with time
and is affected by cloud cover, wind speed and direction, time of year,
and local effects. Similar to atmospheric stability, the mixing depth
depends on the vertical temperature structure of the atmospheric boundary
layer. Local temperature sounding data provide an excellent basis for the
estimation of the mixing height. The mixing height is limited by the height
of the base of a subsidence inversion; the mixing height is intrinsically
lower than or equal to the inversion height. Forecasting of an inversion
height is determined in part by the reliability of predicting the movement
and location of anticyclones.
The determination of model input parameters is strongly related to the
predictability of synoptic scale weather systems. The prediction of the
growth and movement of cyclones and anticyclones is routinely performed
4-27
-------
by forecasters in the National Weather Service (NWS) and in private industry.
At present, forecasts of synoptic scale weather are made in a "man-machine
mix" mode. Completely objective numerical forecasts provide guidance to the
meteorologist who prepares the "best" forecast. Meteorological forecasts
can often be improved by providing more relevant real-time data. Pilot
balloons, radiosondes, and on-site wind and temperature sensors are possibly
important sources of forecast inputs. It is obvious that the mix of NWS
guidance, on-site data collection, and forecasting experience and skill are
important for SCS forecasting reliability. The reliability of these predic-
tions varies with geographical area, temporal length of forecast, and large-
scale weather patterns. It is difficult to assess the absolute reliability
of these predictions as one must consider the unique meteorological and
engineering needs of each proposed SCS. For example, the time requirements
necessary to change fuel usage or to reduce fuel load at a power plant will
dictate the minimum forecast time scale and, hence, will enter into the assess-
ment of forecast reliability. Additionally, the meteorological event associ-
ated with high pollutant concentrations may be more difficult to forecast at
one particular locale, e.g., effluent plume fumigation may be more difficult
to predict then plume downwash.
To assess the reliability of a forecasting capability in reference to
air quality, a basic understanding of the relationship between meteorological
conditions and pollutant concentrations is assumed. This knowledge should
be gained by an extensive diffusion analysis and meteorology-air quality
monitoring program in which their relationship is modeled and observed over
some long time period. An understanding of the relationship between
meteorological conditions and air quality is a primary prerequisite in any
SCS.
Verification of Meteorological Events
The control decisions of an SCS are actually determined by forecasts of
the pollutant concentrations, not the meteorological conditions per se.
The pollutant prediction procedure could include gaussian or numerical
dispersion models, statistical techniques, ranging from simple correlation
through multi-parameter regression equations. Whatever technique is used,
an evaluation of the accuracy of these forecasts will determine the basic
reliability of the system.
4-28
-------
That incorrect forecasting of meteorological parameters is not simply
related to SCS reliability is shown by the following example. Application
of a gaussian plume model (PSDM) for two very different meteorological
conditions can generate similar maximum concentration values. In particular,
high wind speeds and unstable conditions may generate a maximum concentration
of from 0.5 ppm to 0.6 ppm very near the source. Low wind speed and a
trapping inversion condition, could cause a maximum concentration of from
0.5 ppm to 0.6 ppm at a further distance downwind of the source. An SCS
control action based upon the forecast of the latter inversion condition
would adequately control air quality for the former high wind conditions.
Because air quality levels have varying sensitivity to meteorological para-
meters and because air quality might be adequately controlled despite
errors in weather parameters forecasting, an alternative to measuring the
specific reliability of forecasted parameters is desirable. An alternative
approach is presented in Section 4.5.
No SCS can be developed without a continuous review of the forecast
accuracy through verification of the predicted meteorological parameters.
Verification of these forecast meteorological parameters is relatively
simple. Predictions are compared to observations to determine forecast
accuracy. The verification program should indicate the average reliability
of meteorological forecasts for different time periods as well as the
reliability of predictions derived from similar initial conditions.
For example, at a certain pollutant source, plume downwash is expected with
northeast winds above a certain wind speed, U . When the wind speed ex-
c
ceeds U , high ground-level S0? concentrations are measured. The verification
program must then show the ability to forecast strong winds from the north-
east. If this pollution source were located in the Boston area the relative
frequency of northeast winds is low, about 8 percent; and, hence, the
predictability of this wind direction may be lower than the predictability
of more frequent wind directions. However, if this meteorological event
is the only weather condition with a high pollution potential, then the
verification program must emphasize the predictability of strong north-
east winds. Accurate prediction of other wind directions may be unimportant.
Forecasting light northeast winds may be very difficult but is also unim-
portant as related to air quality prediction since downwash or other air
quality problems causing situations do not occur.
4-29
-------
The air quality-meteorology relationship, specific to the locale under
consideration, must guide the verification program and must also guide the
assessment of forecasting reliability. The verification program, thus,
should emphasize the predictability of those meteorological conditions
which produce the highest air pollution potential. It is important to
ascertain the reliability of these predictions - especially the probability
of occurrence of adverse meteorological conditions that were not predicted.
It is less important to accurately predict meteorological conditions that
will not produce high pollutant concentrations.
NWS and U.S. Air Force aviation forecasters routinely predict winds,
sky cover, visibility, precipitation, and temperature for 24-hour period
at a specific location. Ordinarily, verification statistics for several
years are available for these "terminals." These statistics are indicative
of the relative reliability of forecasts at a particular locale and will
indicate the predictability of the meteorological parameters important to
air quality.
Of interest to the maintenance of air quality standards is an indica-
tion of the maximum "underprediction" observed during the time the verifica-
tion program has been in operation. For example, a predicted 24-hour SO
average of 0.05 ppm in comparison to an actual observation of 0.10 ppm
represents a marked underprediction and certainly should be included in the
design of the SCS.
4.4 Assessment of Air Quality Modeling Reliability
Several types of air quality models have been developed to predict
ambient pollution levels resulting from pollutant emission sources. These
models fall into two general categories: (1) deterministic-atmospheric
dispersion models which calculate concentrations based upon physical rela-
tions between emission and meteorological variables and effluent plume dis-
persion; and (2) statistical or empirical models based upon the determination
of statistical relations between emission rates, meteorological conditions,
etc., and air quality levels. Models based upon multiple applications of
a gaussian plume equation to calculate the pollutant concentration at a
receptor or models based upon the numerical solution of a conservation of
pollutant mass equation are of the former type and will be primarily addressed
in this section.
4 -30
-------
The reliability of a model is defined by its ability to predict ambient
pollutant concentrations from given meteorological conditions and emission
rates. The best method for the evaluation of prediction model accuracy is
a thorough analysis of the accuracy resulting from a large dataset of pre-
dictions with the model. With a sufficiently large dataset, the model reli-
ability can be assessed over all weather conditions and observed emission
rates. Such an evaluation procedure results in three benefits: (1) the
model is immediately useful for operational application; (2) the expected
accuracy of short-term forecasts can be evaluated within close limits; and
(3) threshold pollutant concentrations for the reliabile operation of an SCS
can be determined.
To assess the reliability of an atmospheric dispersion model for a
particular locale, i.e., an isolated SCS, a basic understanding of the
relationship between meteorology, emissions, and pollutant concentrations
must be established. This can be determined through a joint meteorology -
air quality monitoring program and a model validation program.
For some applications a comprehensive verification program may be
unnecessary. If it has been determined that high pollutant concentrations
rarely occur or occur only under certain well defined weather conditions,
then the model validation study need only concentrate on the occurrence of
those particular adverse weather conditions and source emissions which
cause high pollutant concentrations. The model reliability, then, must
most carefully be established for the emissions and meteorology which
will produce concentrations above a threshold level, Because of the differ-
ences in the characteristics of the models it is difficult to establish
general analysis criteria applicable to all model types. Therefore, each
type of model is considered individually below.
Gaussian Plume Models
The empirical plume equation most frequently used to estimate the down-
wind dispersion of a pollutant from an elevated continuous point source is
the double - Gaussian plume equation:
where C is the pollutant concentration at height, z
4-31
-------
Q is the source strength;
u is the mean horizontal wind speed;
a ,o are the standard deviations of the distributions of concentrations
y ^
in the y (cross-wind) and z (vertical) directions, and are functions
of downwind distance, x, and atmospheric stability, and
h is the effective source height.
The above equation forms the basis for more general gaussian plume
model - it is generally integrated and otherwise modified to represent more
complex relationship between emissions, meteorology, and concentrations.
The Gaussian form of a plume equation is convenient because of its
simple analytical form. However, the Gaussian approximation must adequately
describe the plume spread as a function of downwind distance and meteorological
parameters. Several characteristics and assumptions of the Gaussian plume
equation should be noted:
The equation requires use of plume shape parameters derived from
field experiments. It is necessary, therefore, to use parameters
from experiments most representative.of the conditions to be studied.
Thus, measurements conducted in open, flat land may not produce
parameters representative of urban conditions nor areas with hilly
terrain.
The plume equation represents time averaged concentrations, as
determined by the details of the various field experiments.
The equation is representative of steady-state conditions. Its use
is less valid when local wind speeds and directions and local tur-
bulence rates are changing rapidly.
Calculations off the centerline are assumed to decrease symmetrically
in the cross-plume and vertical directions, following Gaussian
distributions in both cases. The measures of the plume spread,
a and a^, are related to downwind distance and to atmospheric
stability.
4-32
-------
The effective height of emission, h, is defined as the height of
the plume centerline when the plume has reached neutral buoyancy
conditions. For emissions from smokestacks this height is the
sum of the physical height of the stack and an incremental height
related to the buoyancy and vertical momemtum of the plume.
For emissions from other sources, the plume centerline is often at
or near ground-level.
The equation cannot be used for calm conditions, for which u = 0.
This is a major deficiency in the use of effective point source
models for SCS systems; the risks of high air pollution episodes
is greatest when winds are light and variable.
The unmodified equation makes no allowance for the influences of
local topography. The effects of topography include, in some
areas, enhanced turbulent dispersion due to surface roughness
conditions, and in others, reduced dispersion because of "channeling"
of local winds, as in flows within valley regions.
The equation does not take account of the vertical variability
in the horizontal winds.
In summary, the assumptions implied in the Gaussian plume equations
represent idealized conditions seldom, if ever, realized in the real world.
The causes of errors in point source model calculations may be broadly
grouped into three categories: inaccuracies in the representation of the
atmospheric transport and dispersion process by the model, errors in the
emissions data and errors in estimating meteorological parameters.
The empirical parameters a and a which parameterize turbulent dilu-
tion in the Gaussian plume formulation were developed from field measure-
ments of the atmospheric diffusion of tracer substances. Most of these
experiments involved downwind travel distances between 100 meters and a
few kilometers from the sources and were performed in open flat land.
More recently dispersion parameters have been determined for a few urban
areas. In applying the Gaussian model to a particular point source the
most representative set of dispersion parameters must be applied and their
limitations must be recognized.
4-33
-------
The dispersion parameters a and a are functions of downwind distance
and atmospheric stability. It is assumed that the atmospheric stability is
constant throughout the area of interest. However, changes in atmospheric
mixing rates, which are parameterized by a and a are possible with
y z
changes in surface roughness and thermal characteristics. For example, a
pollutant plume traveling over a region may experience an increase in
spread over a small village as compared to surrounding rural terrain.
The actual determination of stability is frequently indirect, e.g., from
surface observations only. These stability estimates could dictate the
incorrect choice of the diffusion parameters and, in general, are not as
reliable as similar estimates derived from temperature sounding data.
The Gaussian point source model also assumes the wind speed and direc-
tion are constant throughout the area. Model calculations are particularly
sensitive to errors in wind direction as non-centerline pollutant concentra-
tions decrease exponentially away from the centerline. Wind direction
persistence information is especially important for estimating concentra-
tions over time periods of a few hours. As discussed in the previous section,
wind direction will vary with a changing synoptic meteorological situation
and also with terrain effects.
The Gaussian model is usually limited by the lack of treatment of trans-
formation and removal processes in the atmosphere in addition to uncer-
tainties in the emissions data.. As formulated by the Gaussian model, the
pollutant concentration is directly proportional to source emissions.
Hence, uncertainties in the source strength and the temporal variability
in emission rates will lead to uncertainties in the concentration calculation.
The effective height (h) of a stack, which determines the centerline
height in the Gaussian plume model, is computed as the sum of the physical
stack height and the plume rise due to the vertical momentum and the buoyancy
of the effluent. Plume rise is related to the dimensions of the stack, the
effluent composition as temperature and heat flux, the wind speed above the
stack, and atmospheric stability. Uncertainties in these parameters will
affect the plume rise calculation. Several formulae (see Briggs 1969 for
a review) have been developed to described plume rise. Deciding which equa-
tion is applicable to a particular source is extremely difficult and, at best,
uncertainty by a factor of two in estimates of the plume rise is likely on
any one occasion.
4-34
-------
The uncertainties in plume centerline heights can cause major errors
in model calculations, particularly for cases where the distance between
source and receptor is small. At large downwind distances the importance
of stack height decreases, thus reducing errors due to uncertainties in
plume rise.
If a stack is located on a building and its efflux velocity is low or
if the stack is short compared to the building height, plume downwash could
occur, resulting in high pollutant concentrations. The following parameters
are important for aerodynamic downwash: the strength of the undistrubed
wind, stack height, effluent exit velocity and buoyancy, and the dimensions
and spacing of local obstructions to the wind. These parameters will
determine the likelihood of downwash. In turn, their reliability will
affect the reliability of the pollutant concentration calculation.
To eliminate some of the uncertainties of the Gaussian plume model
a comprehensive model validation - calibration program should be instituted
for the point source of interest. The primary objective of the validation
effort is to assure that the model adequately predicts concentrations over
the time and space scales of interest and over the range of expected source
emissions.
For purposes of definition, a distinction should be made between
"calibrating" a model with observed data and "validating" a model with
observed data. The former operation can be as simple as determining "cali-
bration factors" defined as the ratio of observed values to predicted values
or determining a least-square regression line relating predicted to observed
values. If the calibration factors or regression line slopes have values
that are not near unity, or if intercept values imply a negative background
pollution level, there is a strong implication that the model is inadequately
representing some important phenomena.
The judgment of model performance, thus, needs to include consideration
of the intercept values and slope of the least-squares regression line.
Positive intercept values should have a physical interpretation as the con-
tribution of "background" levels of air quality associated with emissions
from all the other sources not incorporated in the model. Non-unity regres-
sion line slope should have a physical interpretation in terms of systematic
over or under prediction of some input parameters.
Validating a model implies a detailed investigation of the model results
and a comparison of those results with measured values in order to identify
4-35
-------
and evaluate discrepancies. If the model results compare well with the
observed data or if, for the applications to be made, simple correction
factors are deemed appropriate, the model may then be simply calibrated
with the observed data. On the other hand, if systematic discrepancies
are found, the investigation may suggest alterations of model parameters
or of the model mechanics which would improve the representativeness of
the model. A final calibration is generally required as the last stage
of the validation procedure to best adjust for remaining discrepancies
between observed and predicted results.
The procedures for validating models will differ somewhat from appli-
cation to application depending upon the nature and purpose of the study
and depending upon the quality of the available data. The validation pro-
cedure will normally require a thorough study of the implications of model
assumptions and the performance of "sensitivity" studies for various input
parameters.
The calculation of calibration factors has been usually made on an
annual or seasonal basis. Ideally, however, calibration factors should
be determined for individual weather conditions or emission rates depending
on the length of the data base. For example, the amount of overprediction
of SO for stable atmospheric stability may be greater than for neutral
stability and, hence, would require a lower calibration factor.
Winter emission rates associated with space heating needs may be positively
correlated with cool northerly winds. Neglect of this mechanism often
causes models to overpredict pollutant concentrations during other weather
conditions. Lack of specific mechanisms to incorporate such effects into the
model will necessitate the correction by the final calibration factor.
Numerical Simulation Models (Conservation of Mass Models)
Most Gaussian-type models in current use ignore spatial and temporal
variations in meteorological conditions by assuming that wind speed and
dispersion parameters are uniform in both the vertical and horizontal direc-
tions. Large spatial and time variations, however, are generally found in
nature and especially in areas with irregular terrain. Local topographic
and thermal effects will be more important to the flow dynamics and the
dispersion of contaminants under conditions of light synoptic winds and
clear skies. Recently, numerical dispersion models which are capable of
4-36
-------
studying air pollution phenomena associated with vertical and horizontal
variations in meteorological parameters and capable of responding to the
time changes in emission and meteorology have been developed. Examples of
models of this type may be found in the referenced works of Egan and
Mahoney (1972) and Shir and Shieh (1973).
The numerical advection - diffusion models are based on solutions to
a conservation of mass equation for a trace material in a continuum fluid.
The tracer equation may be written
rtC
= V (UC) + V KVC + Q
where
C is the concentration
U and K are wind velocity and turbulent diffusivity generally varying
with space and time, and
Q is the emission rate per unit volume.
For general wind and diffusivity fields the continuum equation must be
solved numerically by finite difference techniques.
In general, the sources of error in a numerical dispersion model are
four: (1) emissions data; (2) specification of the wind field; (3) specifi-
cation of the turbulent diffusivities; and (4) errors resulting from the
numerical approximations. The possible errors in emissions have been pre-
viously discussed in regard to Gaussian models. Although errors in the
winds are possible, a numerical model is considerably more versatile than
a Gaussian model in that winds can vary throughout the area of interest.
Vertical components of the wind can also be specified at individual grid
points. An example where spatially varying winds are important is plume
downwash. Here both wind components change rapidly in the general down-
wind direction; and if knowledge of the wind profiles is available (e.g.,
from field measurements or wind tunnel experiments), then these profiles
can be specified in the numerical model.
The estimation of turbulent diffusion rates near the ground as a
function of atmospheric conditions, height and local topographic features
remains a major research topic in micrometeorology. Turbulent diffusivities
are determined either from observations or from dynamical atmospheric
turbulence models. Observation of K profiles have been fairly well defined
4-37
-------
within 30 meters of the ground and over open, flat land. Extrapolations of
K profiles to heights of a few hundred meters necessary to calculate pol-
lutant diffusion from tall stacks have not been well validated.
The solution of the tracer equation is accomplished by finite difference
techniques. Many finite difference advection-diffusion schemes can be
used to evaluate the equation. However, finite-difference approximations
to the advective part of the equation. In particular, introduce errors
that can be order of magnitude larger than mixing by real atmospheric dif-
fusion. The "pseudo-diffusion" associated with numerical advection must
be reduced substantially to realistically represent the physics of the
atmospheric boundary layer. The work of Egan and Mahoney (1972) has resulted
in a method to virtually eliminate the pseudo diffusion problem.
Statistical Air Quality Prediction Models
If an adequate historical data bank of pollutant concentrations and
meteorological observations is available for a region, it is possible to
construct a statistical model relating observed concentrations to various
meteorological parameters. Because statistical models do not consider
changes in emissions parameters, they are only useful for the prediction
of concentrations for short time periods. For short period predictions
as is necessary for an SCS, such models, developed from a sufficient data
base, can provide the necessary predictions with a minimum of computation.
In common with Gaussian plume-models, statistical models rely upon
meteorological forecasts for short period predictions. The accuracy of
all types of models depend upon the accuracy of the meteorological forecasts.
4.5 A Proposed Method of Quantitatively Separating SCS Component Error
The preceding sections discussed the considerations in assessing the
reliability of the components of an SCS. The reliability of a monitoring
network was discussed quantitatively through the example design of an opti-
mum distribution of monitors. The remaining SCS components -- meteorological
forecasting, and air quality modeling have been discussed primarily in
qualitative terms. This section proposes a technique for separately deter-
mining the reliability of components of an SCS. The reliability of the per-
formance of each component can be assessed and combined to determine the
reliability of the entire SCS through the program ERT PROBL (see Section 3).
4-38
-------
ERT has developed a working procedure for evaluating the uncertainty
of meteorological forecasting and of air quality modeling during the opera-
tion of an SCS. The procedure requires that the following three concentra-
tion values be recorded for each forecasting time:
The model predicted concentration using predicted meteorological
parameters. (This concentration value is the basis of the SCS
control action which will affect concentrations T hours later,)
The model predicted concentration using observed meteorological
parameters.
The maximum concentration recorded by the monitoring network.
The procedure combines the above recorded data in a way which isolates
the error in air quality forecasting due to meteorological forecasting un-
certainty from the error due to model uncertainty.
Recall the formulation developed in Section 3 for ground level con-
centration:
The observed maximum concentration is assumed to be C = Q M
The predicted maximum concentration is assumed to be C = Q M R
where the Error Ratio R is the ratio
R = C /C
P °
R is a function which contains contributions from all sources of error and
uncertainty which prevent a perfect air quality forecast; that is, C = C .
These sources of uncertainty arise from each component of the SCS -- meteoro-
logical forecasting, emissions forecasting, and air quality modeling.
Consider the following formulation of R:
R = R - R R
q w m
where R , R , and R are the error ratios for emissions prediction, meteoro-
q w m
.logical (weather) forecasting, and air quality modeling respectively.
In the absence of emission source uncertainty,
R = R R .
w m
4-39
-------
Define the following model predicted concentration values. The pre-
dicted concentration using predicted meteorological parameters is given
by:
Cp (Q,M ) = Q Mp = Q . M R .
This concentration value is the basis of the SCS control action which will
affect concentrations T hours later. The predicted concentration using
observed meteorological parameters is given by:
cp CQ,Mo) = Q MQ
C is the maximum concentration recorded by the monitoring network.
Hence,
Cp (Q,Mp) Cp (Q,MQ) Cp (Q.Mp)
CP «.MO) co co
= R
That is, the error ratio R can be expressed as the product of
two ratios. The first ratio is the quotient of two model predictions using
the same model, the same source emissions and different meteorology. It is
an error ratio isolating the effect of meteorological forecasting on the net
error ratio, R. The second ratio is the quotient of two concentration values
based on the observed source emissions and observed meteorological parameters.
The numerator is a model prediction, and the denominator is a monitored
concentration value. It is an error ratio isolating the effects of model
accuracy on R. The first ratio satisfies the requirements for R and the
second ratio satisfies the description of R , so that
m
R = R R
w m
where
c (Q.MJ
"w sn "<> --
In practical operation of the SCS, Cp(Q,M ) will be computed or deter-
mined at every forecast time and recorded. Then T hours later the value
of C will be recorded. Simultaneously, it is a simple matter to determine
the value C_(Q,M ), using the observed meteorological parameters. If these
4-40
-------
three values are recorded at every forecast time and every forecast verifi-
cation time, the ratios R , R , and R can be routinely computed.
w m r
To consider the possibility of uncertainty in emissions forecasting,
define the following additional model predicted concentration values.
The predicted concentration using predicted meteorological parameters and
predicted source emissions is given by:
Cp (Qp,Mp) = Qp Mp = Q M - R
This concentration value is the basis of the SCS control action which will
affect concentrations T hours later. The predicted concentration using
predicted source emissions and observed meteorological parameters is given
by:
p p' o
C (Q ,M ) is the predicted concentration using observed source emis-
sions and observed meteorological parameters.
It is clear that,
C (0 ,M ) C (0 ,M ) C (Q ,M )
P V P . P T o . P xo QJ _ R
o poo o
That is the error ratio R can be expressed as the product of three ratios.
Similar to the case with no source error, the first and last ratios
represents R and R , respectively. The second ratio is a quotient of
concentration predictions using the same meteorological parameters and the
same model but different source emissions. It is an error ratio isolating
the effects of source emission uncertainty and errors on the Error Ratio R.
The presence of source emission errors require that four concentration
ratio values, R , R , R and R be determined for each time. The distribution
w m q
of R which is generated is available for verification and updating of the
SCS by means in the Reliability Analysis of SCS methods (ERT PROBL) described
in Section 3 and applied in Section 5.
In order to meet the EPA requirements for documentation of the SCS
reliability, it is necessary to undertake a detailed test program to con-
tinuously collect air quality data for purposes of establishing system
reliability.
4-41
-------
To assess system reliability, it will be necessary to measure the reli-
ability of the individual components of the SCS decision making process.
Records of the predicted meteorological parameters, predicted emission rates and
predicted SCL concentration levels will be continuously recorded. These data
in conjunction with observed meteorological data, emissions data, and measured
concentrations provide sufficient information to separate meteorological fore-
casting uncertainty, emissions forecasting uncertainty and air quality modeling
errors. The analysis scheme described in this section is designed for this
application. After the test period of real-time operation of the SCS, a study
of the magnitude of these individual uncertainties and a study of the con-
sistency of these errors will be a valuable measure of system reliability.
Since federal regulations consider reliability to be defined as the
probability that an SCS will not violate air quality standards, it is infor-
mative to generate the frequency distribution for maximum ground-level
concentrations. Using the analysis scheme developed in Section 3, the fre-
quency distributions of Q, M and R can be generated during the test period
to define the probability that air quality standards will not be violated.
In summary, then, the test program can consist of continuous real-time
operation of the system, recording of all predicted parameter values, record-
ing of all measured parameter values, and the determination of system reli-
ability based on the analysis programs of Section 3. These results in con-
junction with a detailed description of all elements of the SCS can be
submitted to the control agency as demonstration of the ability of the sys-
tem to meet standards.
EPA regulations authorize control agencies to deny continued use of the
SCS if the source fails to refine and upgrade the response of the SCS to
adverse dispersion conditions. The isolation of modeling errors by the
techniques described in this section can be used to update the air quality
prediction model. Systematic modeling errors can be used to develop calibra-
tion factors for the models which should improve system reliability as
system experience accumulates. Also, isolation of the meteorological un-
certainty is a valuable indicator of improvements which might be made in
the forecast system. Finally, the probability of violating standards
can be generated periodically as the emission, background, and accuracy
of the system changes with time. This periodic check on system performance
assures that system reliability can be maintained and, ideally, improved
as operators gain experience with the system.
4-42
-------
The distribution of the error ratio, R must be determined for each
SCS application. A preliminary estimate of the character of the ratio can
be made from data collected from an existing air quality monitoring/fore-
casting network. The ERT AIRMAP network for the Boston area consists of SO
monitors which continuously monitor and record air quality levels. Experi-
enced forecasters predict 24-hour average concentration levels at each
receptor for the time period beginning 12 hours after the time of the forecast
.A one year record of both observed concentrations and predicted concentra-
tions for each day of 1973 was available. Maximum predicted concentrations
and maximum observed concentrations were compared for each day to generate
the distribution 6f R for this potential SCS network.
Figure 4-lla illustrates the resulting distribution of R for the Boston
AIRMAP network. The median value of R is nearly 1.0 indicating no significant
bias in prediction. Also, the frequency of R values approaches zero as R
approaches either zero or values much larger than 1.0. Figure 4-lib is the
same distribution with R represented on a logarithmic scale. The distribution
is nearly log-normally distributed. For the example analyses which follow
in Sections 5, 6 and 7 a log-normal distribution of R will be assumed.
It should be noted that the R distribution derived from the Boston
AIRMAP system is for a metropolitan area and may not be applicable to a
region characterized by an isolated source. On the other hand, to our
knowledge, the AIRMAP dataset represents the best available information with
which to assess SCS system forecasting accuracy- The use of this specific R
distribution in the example application of this report is based on the assump-
tion that one could predict air quality levels resulting from isolated source
emissions with accuracies comparable to those associated with predicting
air quality levels at specific receptors in a multiple source region.
4-43
-------
0.15
0.10
o
c
CT
0)
0.05
Ln
I
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
Error Ratio R
(a)
0.15
0.10
O
c
0.05
d=
Jl
O.I 0.2 0.5 1.0 2.0 5.0 10.0 20.0
Error Ratio R (log scale)
00
Figure 4-11 Distribution of the Error Ratio R for the Boston AIRMAP Network
4-44
-------
5. ANALYSIS OF RELIABILITY FOR SEVERAL
HYPOTHETICAL SCS SYSTEMS
5.1 Test Case Conditions and Assumptions
The ERT Probability Analysis of a Supplementary Control System (ERT
PROBL) as defined in Chapter 3 is applied below to several hypothetical SCS
operations. The various examples are designed to illustrate the utility
of the probability analysis technique as well as to illustrate the dependence
of the reliability of an SCS upon the various independent parameters which
influence SCS reliability.
The source data and the meteorological inputs are appropriate to the
ASARCO smelter operation at El Paso, Texas. However, because we wish only
to isolate and illustrate the effect on reliability of changes in the SCS
structure, the inputs have been simplified. Assumed input information is
given as follows:
The source is a single 828-ft stack with the exit diameter,
stack exit temperature, flue gas rate, and maximum SO emission
rate the same as that for the ASARCO copper stack. The emissions
from the lead stack, the zinc stack, and fugitive emissions are
not considered.
The meteorological data is based on a 5-year stability-wind
rose from the El Paso Airport. Estimates of the frequency of
inversion conditions and of inversion break-up fumigations are
only approximate.
Terrain is essentially ignored except for its effect on the
stability-wind rose.
The above simplifications preclude application of the results directly
to the ASARCO smelter. A discussion of the exact applicability of the analysis
technique to the smelter is included in Section 7.
From the source data and the meteorological inputs detailed above, the
Meteorology Function M and the Emissions Function Q can be defined. At any
time, the Meteorology Function is predicted by the gaussian point-source
diffusion model ERT PSDM for unit SO emissions. The distribution of M is
5-1
-------
determined from the stability-wind rose for the El Paso Airport. The maximum
downwind PSDM model prediction for each weather condition is assigned the
corresponding frequency from the stability-wind rose and the frequency dis-
tribution of M is constructed. The Emissions Function Q is assumed for
these examples to be constant and equal to the full load emissions to simulate
conditions in the absence of an SCS. The frequency distribution of Q is,
therefore, very simple. The probability of full load is 1.0 and all other
values have probability zero.
Specification of the SCS is accomplished through several parameters.
A range of values for each parameter is considered in the examples below to
represent all likely SCS operations.
Two SCS types are examined: (1) a single option switch plan, and
(2) a continuous option switch plan. The single option plan is representa-
tive of fuel switching. Implementation of the SCS reduces emissions by the
fraction 3 in all cases. The continuous option switch plan is representa-
tive of process curtailment. Implementation of the SCS reduces emissions
by exactly that fraction necessary to reduce the maximum ground level con-
concentration below a prescribed threshold. These two SCS types were chosen
for study from an infinite set of SCS types which can be investigated by this
probability analysis technique.
The distribution of the Error Ratio R is assumed to be a log-normal dis-
tribution well specified by a geometric mean value W and by the standard
deviation of the distribution cr.
The characteristics of the distributions for R were chosen to be similar
to the observed distribution of R described for the AIRMAP network in
Section 4.5. The log-normal shape and the width of each sample distribution
are similar to those corresponding to this distribution. Error Ratio will
be carefully specified for each example below.
As described in Section 4, the Error Ratio R can be expressed as the
produce of three ratios, R , R , and R . The majority of examples below
T* ^ Mm m-
consider the total function R without considering the individual component
contributions separately. However, for one set of examples. R = R R
W Q
where both R and R has a significant probability of being different from
w q
1.0. In this case, the distributions of R and R are explicitly considered.
Each SCS, regardless of type, is assumed to have a switch threshold.
If the predicted value of the maximum ground-level concentration is projected
5-2
-------
to exceed the switch threshold, the SCS emission reduction action is initiated
Obviously, the highest acceptable switch threshold is the air quality
standard. Because of uncertainty in air quality forecasts, the switch
threshold, y» is usually some fraction of the standard.
5.2 Dependence of SCS Reliability on Various Influencing Factors
The following examples have been provided in order to isolate the effect
of changes in each of the pertinent variables which influence reliability.
These variables include: II, the geometric mean of the error ratio R; a, the
standard deviation of the error ratio distribution; y> the threshold value
of the predicted maximum concentration above which some operational process
adjustment is made; and 3, the ratio of the sulfur content of the low sulfur
fuel to the sulfur content of the high sulfur fuel used in a fuel switch SCS.
Example 1 - What is the effect on SCS reliability of changing the
value of a for the Error Ratio R?
Reduction of the value of 0 for the Error Ratio R is a desirable
objective of every SCS operation. If a could be made negligibly small,
the SCS could be perfectly reliable with a minimum loss of production or
fuel costs for the source. A non-zero value of a results from the presence
of unbiased errors in meteorological forecasting, estimation of emissions,
or modeling results. A reduction in o would be expected from any of the
following system improvements:
Additional or improved meteorological data used in predicting
the meteorological parameters which are input for air quality
forecasts. Unless R is very near 1.0 or the system is operating
W
near the predictability limit for each parameter, some improvement
through added meteorological support is expected. Among the possible
improvements in meteorological support might be atmospheric sounding
data, on-site wind measurements, NWS teletype or facsimile circuits,
a wind field generator model, a faster data reduction system, or
simply more frequent observations of important meteorological data.
More experienced or more capable meteorological personnel. Because
personnel gain experience as the system is operated, the a of the
system should become smaller with time.
5-3
-------
An improved model. As a forecasting model is updated through
system experience, a reduction in a is to be expected.
An improved emission schedule forecast system. This improvement
might be gained by more thorough production planning or it might
involve more careful fuel or materials analysis, better emissions
monitoring, or better plant process monitoring.
Table 5-1 summarizes the results of this example analysis. Note that
all SCS operating parameters are the same for each SCS option except that
a is varied. The first column in the summary table describes the six SCS
options and the NO SCS option (for comparison). The second column con-
tains the frequency of violations of a 1-hour standard of 0.5 ppm expected
to occur with the indicated control strategy. The third column contains
the fraction of low cost fuel (higher sulfur content) which can be used.
The remaining fraction of fuel must be more expensive (lower sulfur content)
fuel. The fourth column contains the fraction of the time that full produc-
tion is possible assuming that the SCS process curtailment is the only con-
straint .
Clearly, any of the six SCS plans reduces the frequency of violations by
at least a factor of 2 but,:interestingly, no more than a factor of 3.4.
By improving forecast accuracy for the fuel switching cases, SCS reliability is
noticeably improved. Since the fuel switching constant 3 = 0.25 is overly con-
servative in most cases, nearly every switch action results in concentrations
below standards. Therefore, improved accuracy of prediction (reduced a) results
in fewer potential violations escaping control. Since the switch threshold
Y = 0.5 is exactly the standard, there is no conservatism in the process
curtailment forecasts. Although improved forecast accuracy reduces the magni-
tude of violating concentrations, the number of violations remains the same.
These examples indicate that some conservatism is desirable for an efficient
SCS strategy. Ways of including conservatism are discussed later.
Improved forecast accuracy can have possible economic and social benefits
despite the probable added expense. For the fuel switching examples, use
of valuable low sulfur fuel is reduced from .25 to .23 and finally to .19
of the total fuel used as forecasting accuracy is improved. Meanwhile, SCS
reliability is also improved. Note that full production is assumed to
be possible regardless of fuel type. For the process curtailment cases,
5-4
-------
01
in
CM
CO
Table 5-1
Effects on SCS Reliability of Changing the Value of a for the
Error Ratio R
Each SCS plan below has the following parameter values:
Fuel switching fraction g = 0.25
Switching threshold y = 0.5
Geometric Mean of Error Ratio R~ = 1.0
Width of Error Ratio distribution a, is variable
SCS Control Strategy
NO SCS
SCS #1:
FUEL SWITCHING
a = 0.5
SCS #2:
FUEL SWITCHING
a = 0.4
SCS #3:
FUEL SWITCHING
a = 0.2
SCS #4:
PROCESS CURTAILMENT
a = 0.5
SCS #5:
PROCESS CURTAILMENT
a = 0.4
SCS #6:
PROCESS CURTAILMENT
a = 0.2
Total Frequency
of Violations
0.16432
0.06605
0.06291
0.04875
0.08194
0.08216
0.08216
Fraction of Low
Cost Fuel
1.000
0.754
0.768
0.808
1.000
1.000
1.000
Fraction of
Time at Full
Production
1.000
1.000
1.000
1.000
0.918
0.948
0.961
5-5
-------
the percentage of full production is increased as forecast accuracy improves.
Meanwhile, SCS reliability is maintained at the same level.
Example 2 - What is the Effect on SCS reliability of changing the value
of "R for the Error Ratio R?
The geometric value of the Error Ratio, R, is less than 1.0 if con-
centrations are characteristically underpredicted, greater than 1.0 if
concentrations are characteristically overpredicted, and 1.0 if there is
no systematic bias in prediction. It is easy for a system to achieve a
value of R = 1.0 by simply reducing each, forecast value by the required
amount to bring the mean of past values to 1.0. It is generally desirable,
however, to intentionally operate an SCS conservatively to prevent a high
frequency of violations which are near but higher than the standard. The
limits on reliability of a nonconservative SCS were illustrated in the pre-
vious example analysis. One method of operating a conservative SCS is to
maintain an Error Ratio mean R greater than 1.0.
An air quality forecast model which overpredicts provides a means of
achieving IT greater than 1.0. Most air quality models overpredict because
"worst case" conditions such as persistent meteorology and conservative
plume rise are assumed.
Similarly, meteorological and emission predictions used for air quality
projections are often chosen to be "worst case" forecasts. For example,
predicting fumigation conditions for all clear mornings would produce a
value of R greater than 1.0, but may be necessary to prevent contravention
of standards on those several mornings when inversion breakup is a problem.
The example analysis which follows is designed to investigate the effect
of changing R on SCS reliability, leaving all other SCS parameters unchanged.
Table 5-2 includes the results of the operation of six hypothetical SCS
schemes and the NO SCS case.
Again, each of the six SCS plans reduces the frequency of violations by
a considerable amount. The increased conservatism of air quality prediction,
manifested in increased values of R~, reduces the frequency of violations of
the standard for both fuel switching and process curtailment. For fuel
switching, 43 of every 44 violations can be eliminated using an SCS with
R = 2.0.
The economic penalty for the indicated improvements in air quality is
shown in the final two columns of Table 5-2. With R = 2.0, lower sulfur fuel
5-6
-------
Table 5-2
Effects on SCS Reliability of Changing the Value of R~ for the
Error Ratio R
Each SCS plan below has the following parameter values:
Fuel switching fraction = g = 0.25
Switching threshold = j = 0.S
Geometric mean of error ratio R is variable
Width of error ratio distribution a = 0.5
SCS Control Strategy
NO SCS
SCS #1:
FUEL SWITCHING
R = 1.0
SCS #7
FUEL SWITCHING
R = 1.5
SCS #8:
FUEL SWITCHING
R = 2.0
SCS #2:
PROCESS CURTAILMENT
R = 1.0
SCS #9:
PROCESS CURTAILMENT
R = 1.5
SCS #10:
PROCESS CURTAILMENT
R = 2.0
Total Frequency
of Violations
0.16432
0.06606
0.00876
0.00370
0.08194
0.01306
0.00000
Fraction of Low
Cost Fuel
1.000
0.754
0.521
0.413
1.000
1.000
1.000
Fraction of Full
Production
1.000
1.000
1.000
1.000
0.918
0.836
0.609
5-7
-------
is required 59% of the time for operation of the fuel switching plan.
For process curtailment, a negligible violation frequency is accomplished
by reducing maximum possible production by 39%. Unlike reducing a, increasing
R above 1.0 has no compensating economic savings.
Example 3 - What is the effect on SCS reliability of changing the value of
the switch threshold y ?
The previous example analysis investigated the improvement in SCS
reliability effected by conservative air quality forecasting. Another
method of improving SCS reliability is through the use of a switch threshold
less than the standard. Similar to making conservative predictions, this
control technique compensates for tendencies to underpredict since most
underprediction errors will result in "violations" of the threshold which
are still below the standard.
Table 5-3 displays the results of the example analysis for six hypotheti-
cal SCS plans with switch thresholds of varying value.
Systematic improvement in SCS reliability is evident for both the fuel
switching cases and the process curtailment cases as the switch threshold
is made a smaller fraction of the air quality standard.
Systematic reduction in economic benefit manifested in fractional fuel
usage data and fraction of full production data is also evident. Similar to
maintaining the value of R greater than 1.0, a conservative switch threshold
is a simple tool for improving SCS reliability; but an overall loss of
plant efficiency is a probable effect of the control strategy.
Example 4 - What is the effect on SCS reliability of changing the fufcl
switching fraction 3?
Although choice of a fuel switching fraction 3 is most likely determined
by the availability of fuel types, it is interesting to observe the effect
of changing the value of 3. One can hypothetically achieve any value of 3
by blending fuels of known sulfur content, but engineering problems prohibit
this generality in most cases.
Three SCS plans with values of 3 of 0.25, 0.30, and 0.40, respectively
were investigated. No appreciable change in SCS reliability or in plant
production was observed. Apparently, the value 3 used in all three cases
is very conservative; that is, each time a switch is implemented to a lower
-------
TABLE 5-3
EFFECTS ON SCS RELIABILITY OF CHANGING THE
VALUE OF THE SWITCH THRESHOLD y
Each SCS plan below has the following parameter values:
Fuel Switching Fraction 3 = 0.25
Switching Threshold Y is variable
Geometric Mean of Error Ratio 1* = 1.0
Width of Error Ratio Distribution a = 0.5
SCS CONTROL
STRATEGY
No SCS
SCS #1:
Fuel Switching
Y = 0.5
SCS #11:
Fuel Switching
Y = 0.4
SCS #12:
Fuel Switching
Y = 0.3
SCS #2;
Process Curtailment
Y = 0.5
SCS #13:
Process Curtailment
Y = 0.4
SCS #14:
Process Curtailment
Y = 0.3
TOTAL FREQUENCY
OF VIOLATIONS
0.16432
0.06606
0.03500
0.01562
0.08194
Q. 05641
0.02557
FRACTION OF
LOW COST FUEL
1.000
0.754
0.615
0.526
1.000
1.000
1.000
FRACTION OF
FULL PRODUCTION
1.000
1.000
1.000
1.000
0.918
0.873
0.801
5-9
-------
sulfur content fuel a greater than necessary reduction in concentration is
achieved. Therefore, increasing the value of 3 toward 1.0 has no effect on
violation frequency for values of (3 less than 0.5.
Example 5 - What is the effect on SCS reliability of maintaining a
conservative value of R for the error ratio R and changing
the value of y?
The preceding examples indicate that significant improvement in air
quality can be expected from any one of many reliable SCS plans. It is
not possible to define which SCS is both reliable enough for acceptance
by control agencies and economically practical enough for acceptance by
plant operators. It is likely that some combination of the preceding
example SCS systems would be optimum for most operations.
Furthermore, it is conceivable that an operating SCS will require
updating due to demands for more SCS reliability or due to demands for more
cost effective operation by the plant management. In this eventuality
it is likely that some combination of the preceding SCS changes would be
optimum for the particular operation.
It is, therefore, important and interesting to observe the effects
of more than one parameter change on SCS reliability. Table 5-4 includes
six example SCS plans which observe the effects of changing the switch threshold
y and employing a conservative mean value of the Error Ratio R.
Comparing SCS number 7 and the three fuel switching plans in Table 5-4,
it is clear that increased conservatism yields only small improvement in
reliability until the SCS reaches its limit of reliability under the fuel
switching plan. Further improvement would require the plant to cease
operations .37% of the time, or 32 hours per year. For process curtailment
a comparison of SCS number 9 and the three plans included in Table 5-4 indicate
continuous improvement in SCS reliability with decreasing value of the
switch threshold. Values of y less than 0.3 are unnecessary since only
a negligible frequency of violations is expected at a value of y = 0.3.
Note that SCS number 19 expects less than 0.1 violations per year
and achieves more than 66% of full production. Considering no other com-
plexities in evaluating SCS reliability, SCS number 19 accomplishes most
acceptable reliability with maximum plant production of all SCS plans
considered in these examples.
5-10
-------
TABLE 5-4
EFFECTS ON SCS RELIABILITY OF OPERATING WITH A CONSERVATIVE
VALUE OF R AND CHANGING Y
Each SCS plan below has the following parameter values:
Fuel Switching Fraction 3 = 0.25
Switching Threshold Y is variable
Geometric Mean of Error Ratio R = 1.5
Width of Error Ratio Distribution a = 0.5
SCS CONTROL
STRATEGY
No SCS
SCS #15:
Fuel Switching
Y = 0.4
SCS #16:
Fuel Switching
Y = 0.3
SCS #17:
Fuel Switching
Y = 0.2
SCS #18:
Process Curtailment
Y = 0.4
SCS #19:
Process Curtailment
Y = 0.3
SCS #20:
Process Curtailment
Y = 0.2
TOTAL FREQUENCY
OF VIOLATIONS
0.16432
0.00471
0.00370
0.00370
0.00353
0.00000
0.00000
FRACTION OF
LOW COST FUEL
1.000
0.450
0.395
0.372
1.000
1.000
1.000
FRACTION OF
FULL PRODUCTION
1.000
1.000
1.000
1.000
0.784
0.664
0.572
5-11
-------
Example 6 - How can emission error be incorporated into the analysis?
Each of the example analyses considered so far in this section considers
the Error Ratio R to be some hypothetical log-normally distributed function.
No attempt has been made to simulate the effects of each component of
SCS reliability. The following example analysis will consider SCS schemes
which have meteorological error distributed like the Error Ratios of the
preceding examples, but which also have emission errors. According to
the discussion in Section 4,
D D . D . D
R " "w Q ^4
For these examples, we assume R.. = 1.0, therefore
D D D
K ~ "w Q
We will assume that RW has a log-normal distribution with R,= 1.0 and
a =0.5. Furthermore, we will assume that
w '
that is, that the error in emission rate Q is measured simply by the ratio
of predicted Q to the observed Q for that time.
Then,
" = QO RW
Q
It is reasonable to expect that -r*- has either a normal or a "top-hat" distri-
bution. The example below considers both of those possibilities. The hypothe-
sized distributions for RW and Rn are combined to form a distribution for R.
Figure 5-1 illustrates the three distributions of ^*- used. They are desig-
nated as Q functions 1, 2 and 3. °
Table 5-5 summarizes the results of the analysis using the combined
Error Ratios. The frequency of violations for all six example SCS plans is
greater than the frequency of violations for the corresponding SCS with no
emissions error (SCS number 1 or SCS number 2). Improvement in SCS reli-
ability is achieved as a of the distribution is reduced. The "top-hat"
emission error distribution is associated with a reliability intermediate
between the two normally distributed emission error functions.
5-12
-------
Q FUNCTION #1
X
X
/
X
/
/
/*
^\
Q
Normal Distribution
\
\
<7
v
\
-=o.z
s
>
"^^
^
^
Q FUNCTION #2
0.6 0.7 0.8 0.9 1.0 I.I 1.2 1.3 1.4 1.5
Q
Normal Distribution
3er=O.2
Q FUNCTION #3
0.6 0.7 0.8 0.9 1.0 I.I 1.2 1.3 1.4 1.5
"TOP
-hat"
Q
Distn
'butio
n
^
0.6 0.7 0.8 0.9 1.0 I.I 1.2 1.3 1.4 1.5
Q
Figure 5-1 Three Frequency Distributions of the Ratio Q / QQ
5-13
-------
TABLE 5-5
INCORPORATION OF EMISSIONS ERROR INTO
THE RELIABILITY ANALYSIS
Each SCS plan below has the following parameter values:
Fuel Switching Fraction 3 = 0.25
Switching Threshold y = 0.5
Geometric Mean of Error Ratio R = 1.0
Width of Error Ratio 0=0.5
SCS CONTROL
STRATEGY
No SCS
SCS #21:
Fuel Switching
Q Error 1
SCS #22:
Fuel Switching
Q Error 2
SCS #23:
Fuel Switching
Q Error 3
SCS #24:
Process Curtailment
Q Error 1
SCS #25:
Process Curtailment
Q Error 2
SCS #26:
Process Curtailment
Q Error 3
TOTAL FREQUENCY
OF VIOLATIONS
0.16432
0.07079
0.06837
0.06943
0.08582
0.08397
0.08523
FRACTION OF
LOW COST FUEL
1.000
0.765
0.764
0.766
1.000
1.000
1.000
FRACTION OF
FULL PRODUCTION
1.000
1.000
1.000
1.000
0.920
0.921
0.921
5-14
-------
5.3 Further Applications of the Probability Analysis of SCS Reliability
It is likely that in many applications the Error Ratio R will not be
independent of the meteorology function M. If conditions of very predictable
strong winds are responsible for many high SO levels, for example, it would
be incorrect to use an Error Ratio R derived from more difficult to predict
light wind cases. A different Error Ratio for each of several meteorological
categories is likely. The ERT PROBL analysis should be performed separately
for each category; then the resulting frequency distributions can be added
together.
It is also possible that several process curtailment actions are avail-
able for use but that a continuous option plan is not practical. This would
be the case for an operation with integral units which can be either shut
down or operated at full capacity. Such an SCS is easily investigated by
the ERT PROBL analysis.
Generally, SCS plans would have M functions which are not independent
of the Q functions. The principal cause of this dependence is plume rise
which is determined by both wind and stability (meteorology) and by heat
flux at the stack exit (emissions). The ERT PROBL analysis would require
modifications to handle this interdependence of M and Q. For a given
emission source this would require the specification of different M functions
(incorporating the effects of differences in plume rise) for each significantly
different source emission rate category.
For the ASARCO plant and for the examples through Section 5, heat flux
through the stack has been assumed to be fairly constant regardless of load.
It is instructive to investigate in some detail the effect that changes
in emission rate can have on ground-level concentration.
For a 36 day period in March and April of 1971, S02 emission rates,
flue gas rates, and exit temperatures were compared for the 828-ft copper
stack at ASARCO. Exit temperatures are very constant, and flue gas rates
are not a strong function of emission rate. A linear regression analysis
to relate heat flux and emission rate for the ASARCO stacks indicates that
heat flux is reduced by just 8.0% when emissions are reduced by 50%. The cor-
relation coefficient for these two parameters was, however, only 0.22.
The meteorologists at ASARCO agree that this poor correlation exists.
Many processes contribute effluent to the stacks at ASARCO and effluent
characteristics vary according to the stage of each process. This variety
5-15
-------
of operating conditions could contribute to the low correlation between SC^
emission rate and heat flux.
For power plants, the flue gas rate is a strong function of emission
rate. Since exit temperatures vary by only about 10% over the range of
possible power plant loads, heat flux is also strongly related to emission
rate. In fact, the heat flux is nearly proportional to emission rate over
the range of possible loads.
Assuming the plume rise formulation of Briggs (1969), the effect of
changing emission rates on maximum ground-level concentrations can be
assessed. Using the ERT PSDM gaussian diffusion model, maximum short-term
concentrations were compared for the ASARCO plant under full load and half
load conditions. An 8.0% reduction in heat flux was assumed. Under each
weather condition, the maximum concentrations under half load conditions was
no more than 53% and no less than 50% of the concentrations, under full load
conditions. A linear "roll-back" of concentrations with emissions seems
appropriate in this case.
For a typical power plant, a 50% reduction in heat flux is expected to
accompany a 50% reduction in emissions of SO . Under each weather condition,
the maximum concentration predicted by ERT PSDM under half load conditions
was no more than 81% but no less than 61% of concentrations under full load
conditions. Csanady (1971) developed a generalized technique of comparing
emission reduction with maximum concentration reduction when heat rate is
linearly related to emission rate. The hypotheses of the technique are
most applicable to a point source under unstable atmospheric conditions.
Figure 5-2 illustrates the result of this technique when heat rate is
assumed proportional to emission rate. Maximum concentrations under half
load conditions are expected to be 63% of concentrations under full load
conditions. The effect of other fractional load reductions can be estimated
from Figure 5-2. Clearly a linear roll-back of concentrations with emissions
is not valid in this case.
The switching of fuel types is a control action which has an easily
predictable affect on ground-level concentrations. Reducing the sulfur
temperatures, BTU content of the fuels, and fuel to air ratios are not
altered by fuel switching.
Switching from one fuel to another fuel such as from coal to #6 residual
oil will generally alter the operating conditions of the facility. It is
5-16
-------
g
t5
k_
c
O
c
O
O
15
0>
a
c
1
CD
x
a
C..\J
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0,4
0.2
0
I
/
/
Load
/
/
Red
/
f
uctio
/
/
n
/
x
.X
X
Loa<
/
i Inc
X
^
reast
X
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Figure 5-2 ^ Effects of Emission Reduction on Maximum Ground Level
Concentrations when Heat Rate through the Stack is
Proportional to Emission Rate
00
0)
ID
5-17
-------
common for flue gas rates to differ -with fuel type. Volumetric flow rates
of 1.2 x 10 scfm/10 BTU is common for coal fired boilers while the flow rate
is more typically 1.4 x 10 scfm/10 BTU for oil fired boilers. Thus, a
facility which maintains constant production rate can expect a 17% increase in
flow rate upon switching from coal to oil. Also, exit temperatures are
typically lower for coal firing than oil firing. This fact further increases
the flow rate when switching from coal to oil.
The conclusion of the discussion above is that ground-level concentrations
are not directly related to changes in sulfur emissions. The ERT PROBL
analysis can be modified to include the complications imposed by changes in
heat flux associated with emission reduction. More important, however, is
that invalid use of proportional roll-back procedures be avoided during the
design of the SCS.
5-18
-------
6. SIMULATION OF THE EFFECT OF A CONTROL STRATEGY ON
CONCENTRATIONS FOR VARIOUS AVERAGING TIMES
6.1 Introduct i on
Detrimental effects of air pollution are closely related to the dura-
tion of high levels of the pollutant. This relationship is reflected in
the existence of air quality standards with averaging times ranging from
one hour to as long as one year. Adherence to a single standard, say a
3-hour standard, does not insure adherence to any other standard for a
longer or shorter averaging time. It is necessary, then, to consider
and identify the air quality standards for which an SCS is a pertinent con-
trol procedure.
Any SCS system will have inherent time delays of response associated
with: (1) time requirements to transmit and interpret input data and to
form a control strategy on the basis of air quality forecasts; (2) curtail-
ment lead time associated with the various steps to be taken to change fuels,
reduce production, etc., and; (3) the time scales of pollutant transport
and dispersion - the lead time required to have an emission reduction
result in an air quality reduction at the furthest monitoring location of
interest. Each SCS system will have different constraints on response time
associated with the above factors. Curtailment lead time is relatively
long for coal fired power plants; pollutant transport time factors will
be more important to many oil fuel switching operations or to quick-response
process curtailment control systems. The analysis scheme presented in this
section investigates the impact of an SCS designed to achieve an air quality
goal for one averaging time on the achievement of air quality standards for
other (longer or shorter) averaging times.
The product of the following analysis scheme is a cumulative frequency
distribution for maximum ground-level concentration for any desired averaging
time and employing any control procedure designed to adhere to a single
air quality standard (or threshold). Using this analysis tool, we can
hypothesize an SCS which reliably prevents violations of, say, a 24-hour
standard. The cumulative frequency distribution of 24-hour averages would
verify that the SCS is achieving its goal. The resulting cumulative fre-
quency distribution of 3-hour averages would indicate what impact the long-
term control was having on short-term air quality. Similarly, we can
6-1
-------
hypothesize an SCS which reliably prevents violations of the 3-hour
standard. The cumulative frequency distribution of 3-hour averages would
verify that the SCS is achieving its goal. The resulting cumulative fre-
quency distribution of 24-hour averages would indicate the impact of short-
term control on long-term air quality. Control of intermediate averaging
times - greater than 3 hours but less than 24-hours can also be examined
as well as the effects of controls on concentrations with averaging times
greater than 24 hours.
It should be emphasized that there is no need to require that an SCS
control strategy be designed with a single response time. A strategy could
very well have separate control procedures to maintain the 3 hour and 24 hour
Federal standards. For design of such a system, however, it is important
to evaluate how the lead time necessary to implement source emission changes
relates to possible violations of short-term standards and how assured
compliance with one standard relates to compliance of a standard having a
different averaging time.
6.2 Description of the Analysis Technique
The terms defined below will be used throughout the following section:
Sampling Time
Averaging Time
Time Period
Standard
the shortest duration over which concentrations
are measured and recorded.
a duration of time over which sampling times are
averaged.
a unique interval of time. Two 3-hour time periods
are different unless they are coincident.
the maximum acceptable concentration value refer-
enced to some averaging time.
Design Standard - the concentration value for which the SCS is designed
to prevent contravention.
Design Standard
Averaging Time
the averaging time associated with the design
standard.
6-2
-------
The procedure requires an air quality prediction model validated for the
application. The model is used to calculate concentrations for a dense grid
of receptors for each meteorological and emission condition which is expected
to occur at any time. The analysis also requires a continuous record of
meteorological parameters and uncontrolled emissions data for some period
of a year or more.
The analysis translates the meteorological and emission parameters for
each sampling period (a one hourly sampling period is desirable) into a
meteorological and emissions category which can be used by the air quality
model. The appropriate model predicted concentration is then assigned
to each receptor for each sampling period. From these concentration values,
running means for any averaging time can be computed, and maximum concentra-
tions can be determined for each running-average period. Finally, the re-
sults of these maximum concentration calculations can be used to create a
frequency distribution for that averaging time for the entire period of
meteorological record. In this, way, the effect of the uncontrolled plant
emissions on maximum concentrations for any averaging time can be assessed.
The analysis technique is also capable of simulating the effect of
a supplementary control procedure designed to force compliance with a
single air quality standard. In this case, the emissions are rolled back
the amount required to comply with the standard. After the corresponding
controlled emissions are determined for each sampling period, running means
for any averaging time can be computed, maximum concentrations can be deter-
mined for each running-average period, and a frequency distribution of
maximum concentration results.
Through the above analysis, the effect of the uncontrolled plant emis-
sions on maximum concentrations for any averaging time can be assessed.
Furthermore, one can decide which averaging time should be most carefully
controlled and for what maximum concentration the SCS should be designed.
The schematic diagram of Figure 6-1 describes the operation of the
analysis technique. Blocks 1 through 4 represent data preparation.
Blocks 7 and 9''define the SCS implementation criterion; that is, the system
is designed to decrease emissions to prevent concentrations exceeding Cg
for any time period of duration Tg. Blocks 5 through 13 modify the uncon-
trolled concentrations for each receptor and for each sampling time exactly
the amount required to satisfy the SCS design criterion. Blocks 14 through
18 simulate the effect of the SCS on concentrations for various averaging
6-3
-------
D
A
T
A
P
R
E
P
A
R
A
T
I
0
N
S
C
S
I
M
P
L
E
M
E
N
T
A
T
1
0
N
E
F
F
E
C
T
S
0
F
S
C
S
C
0
N
T
R
0
L
Chronological
Meteorology
Data
Chronological
Emissions
Data
Compute
Chronological
Model Input
Parameters
Compute
Chronological
Concentration
Values
7
f .SCS Design
Standard
Averaging Time
Compute
Concentrations
For Time Period
of Duration
Ts
-
Go To
Beginning of
Chronological
Concentration
Record
I
Isolate The
Maximum
Concentration
CMAX For This
Time Period
SCS Design
Standard
Cs
in J
Yes
i.
Reduce
Concentrations
At Each Receptor
to Satisfy
The SCS
No
Yes
=t
Averaging Times
& Standards
of Interest
Compute
The Number
of Violations
For One
Standard
16
Precede
to Next
Averaging
Time
Generate
Distribution
of Concentrations
For One
Averaging Time
A Time-series
of Concentration
Values for Each
Receptor for
Each Sampling
Period is Now
Available. These
are Uncontrolled
Concentration
Values.
A Time - series
of Concentration
Values for Each
Receptor for
Each Sampling
Period is Now
Available. These
are SCS Controlled
Values.
Figure 6-1 Schematic Representation of Analyses to Assess Effects of SCS
Operation on Concentration Values for Various Averaging Times.
6-4
-------
times. These averaging times can vary from the length of the sampling time
to the length of the entire data record. Block 14 represents the computa-
tion of the number of non-overlapping violations for the entire period of
record.
6.3 Example Analyses
The effect of several example SCS plans on concentrations for various
averaging times was investigated. As was done for the example reliability
analyses of Section 5, the examples are done for a hypothetical point source
similar to the ASARCO Copper Stack at El Paso, Texas. Again, full load
conditions for the 828-ft SO source were assumed for the uncontrolled
emissions data. Three-hourly meteorological observations from El Paso
Airport for the entire year of 1971 were used as the chronological meteoro-
logical record. These data were processed to provide necessary input para-
meters for the ERT Point Source Diffusion Model (PSDM). Uncontrolled
3-hourly concentration values were generated for 160 receptors at each of
the 2920 sampling times of 1971. No terrain effects were considered; that
is, the receptors are all assumed to be at the elevation of the stack base.
Table 6-1 summarizes the results of analyses of 12 different SCS
design goals uniquely defined by values of the design standard Cc and the
O
design standard averaging time Tg. Each column of Table 6-1 represents the
effect of a particular SCS design. Tc and C , respectively, are contained
O O
in the first row of each column.
For each SCS plan, the effect on 3-hourly; 9-hourly, 24-hourly, monthly
and annual averages is assessed in two ways. First the total number of
non-overlapping violations per year is evaluated. Actual 3-hourly, 24-hourly,
and annual standards are used to determine violations. For 9-hourly and
monthly values, arbitrary "standards" were assumed for comparison.
The first three SCS designs attempt to control air quality by preventing
violations of a 3-hourly threshold. As required, no violations of the 3-hour
threshold C are expected. As the design standard becomes more strict, vio-
O
lations of the 24-hour secondary standard of 0.1 become less frequent, but
even the strict control of 0.34 ppm for 3-hours permits 10 violations per
year (see Column 4). No annual average violations are expected.
The next two SCS designs attempt to control air quality by preventing
violations of a 6-hourly threshold. With a design standard of 0.24 ppm,
no 3-hour violations are expected and only one "slight" violation of the
6-5
-------
TABLE 6-1
EFFECTS OF CONTROL STRATEGIES ON GROUND-LEVEL CONCENTRATIONS
FOR DIFFERENT AVERAGING TIMES
Number of Violations Per Year and Maximum Concentrations (ppm)
AVERAGING TIME
AND STANDARD
3 HOUR
S, 0.5 ppm
9 HOUR
Sg « 0.1 ppm
24 HOUR
S24 " O-1 PP
1 MONTH
Su 0.02 ppm
H
1 YEAR
S» » 0.02 ppm
Violations
per year
Maximum ppm
Violations
per year
Maximum ppm
Violations
per year
Maximum ppm
Violations
per year
Maximum ppm
Violations
per year
Maximum ppa
CONTROL STRATEGY
NONE
113
1.155
494
.506
94
.212
8
.031
0
.013
3-HOURLY
<.SO ppm
0
.500
494
.337
41
.162
3
.023
0
.011
3-HOURLY
<.42 ppm
0
.420
494
.317
17
.151
3
. .021
0
.010
3-HOURLY
£.34 ppm
0
.340
494
.308
10
.133
2
.021
0
.010
6-HOURLY
<.24 ppm
0
.480
360
.244
1
.105
1
.022
0
.009
6-HOURLY
<,20 ppm
0
.400
359
.209
0
.098
1
.022
0
.009
9-HOURLY
£.20ppm
73
.600
481
.200
14
.125
3
.021
0
.011
9-HOURLY
£.14 ppm
0
.420
448
.140
0
.098
1
.021
0
.009
18-HOURLY
<.14 ppm
23
.840
362
.280
15
.140
1
.022
0
.009
18-HOURLY
£.10 ppm
20
.600
350
.200
5
.109
1
.022
0
.009
24-HOURLY
<.14 ppm
111
1.12
483
.373
93
.140
6
..030
0
.012
24-HOURLY
£.10 ppm
22
.800
361
.267
0
.100
1
.022
0
.009
24-HOURLY
£.08 ppm
20
.640
350
.213
0
.080
1
.022
0
.009
-------
24-hour standard is expected. If the design standard is reduced to
0.20 ppm, no violations of any actual standard is expected.
Similarly, control of the 9-hourly values at 0.20 ppm is unable to
satisfy either the 3-hour or the 24-hour standard. Reduction of the design
standard to 0.14 ppm for 9 hours eliminates violations of any applicable
standard.
Control of averaging times of 18-hours and 24-hours were not efficient
methods of controlling the short-term 3-hour average as is clearly seen in
Table 6-1.
The results of this analysis indicates that control of a 6-hourly
threshold value or a 9-hourly threshold value will insure compliance with
all relevant air quality standards. Control of the 3-hourly values to
insure compliance for all standards seems possible at a threshold con-
siderably below the standard. Control of averaging times of 18-hours and
more appears to be an unreliable control for protecting short-term averages.
The results for this single source using the climatology of El Paso
are not expected to be generally applicable to other sources in other
regions. Apparently, periods of markedly high concentrations have a dura-
tion on the order of 6 to 9 hours in the El Paso area. This is the approxi-
mate duration of the period from the morning fumigation through the time
of maximum afternoon stability. Other sources with different climatology
and source characteristics would have different relationships among con-
centration distributions for various averaging times.
Figures 6-2 through 6-5 indicate the effectiveness of a 3-hour control,
9-hour control, and 24-hour control of achieving compliance with the 3-hour
and the 24-hour air quality standards for the example above.
Figure 6-2 is a cumulative frequency distribution which demonstrates
the unsuccessful attempts to control the 24-hour values through 3-hour con-
trol. The following two figures illustrate the success of the 9-hour control
in complying with both the 3-hour and the 24-hour standards. Figure 6-5
demonstrates the unsuccessful attempts to control the 3-hour values through
24-hour control.
Tables B-2 through B-16 contain the cumulative frequency distributions
from which the summarized data of Table 6-1 was derived.
6-7
-------
.50 ppm Control
.42 ppm Control
.34ppm Control
90
80
70
60 50
% Observations
40
30
20
10
Figure 6-2 Effects on 24-Hourly Average Concentrations
of Using a 3-Hourly Design Standard
No Control
-.14ppm Control.
.10 ppm Control
.08ppm Control
90
80
70
60 50
% Observations
40
30
20
10
Figure 6-3 Effects on 3-Hourly Average Concentrations
of Using a 24-Hourly Average Design Standard
6-8
-------
Control
.14 ppm Control
70
60 SO
% Observations
40
30
20
10
Figure 6-4
Effects on 3-Hourly Average Concentrations
of Using a 9-Hourly Design Standard
0.3
- 0.2
-No Control
-, 20 ppm Control
». 14 ppm Control
o
o
100
90
80
TO
60 50
% Observations
40
30
20
Figure 6-5 Effects on 24-Hourly Average Concentrations
of Using a 9-Hourly Design Standard
6-9
-------
7. CASE STUDY - EVALUATION OF AN EXISTING SCS
7.1 Introduction
The purpose of this section is to analyze the performance of an
existing SCS. Determination of reliability and effectiveness can be useful
guides for future SCS's. For this study the process curtailment operation
at the American Smelting and Refining Company (ASARCO) plant in El Paso during
1971 was used. Essentially this system is geared toward reducing the level
of operation and, hence, the output of sulfur dioxide during the periods
of expected or existing high ground level concentration. Peters (1971), has
described the meteorological conditions in the El Paso area associated with
high SO- episodes. The most significant of these is morning fumigation, when
elevated plumes emanating from the tall ASARCO stacks mix rapidly downward to
ground level in response to daytime heating. The time and place of this occur-
rence can be forecast several hours in advance and the level of operation can
be reduced in anticipation of this condition. To detect other significant
air quality occurrences, ASARCO maintains constant monitoring of air quality
by a network of field instruments. When high concentrations are observed,
the load can be reduced 25% from full-load in a matter of minutes. Further
reductions require longer lead times.
Basically, the ASARCO system relies on two activities, forecasting and
monitoring. The effectiveness of the system lies in the accuracy of the
forecasts and the reaction time to impending high air pollution cases. The
best way to evaluate both is in terms of the final results -- the air quality
observations at ASARCO's monitoring stations in and around the El Paso area.
A map of the location of these stations as well as surface elevation con-
tours is given in Figure 7-1. Table 7-1 lists the bearing and distance from
ASARCO to the monitoring stations.
For this study the period of March 5 through April 9, 1971, was chosen
because a complete set of air quality observations and load conditions was
available. The load conditions for the copper and lead ASARCO stacks, with
heights of 828-ft and 612-ft respectively, were given individually. Both values
are given because load reductions were carried out independently for each
stack.
A method for determining the success or failure of an SCS is in relation
to the achievement of some set of air quality standards. Based on Peters (1971)
7-1
-------
Figure 7-1 Base Map for Case Study Analyses
-------
TABLE 7-1
DISTANCE AND BEARING
FROM
ASARCO PLANT TO MONITORING STATIONS
STATION
UTEP
CORONADO
ZORK
FARRELL
4 -SEASONS
MONTRIDGE
RICHMOND
MISSOURI
TURNEY
CHELMONT
VECK
RIVER
ROBINSON
RIM
HAWTHORNE
MARLIN
PARK HILL
BEARING
DEGREES
116.6
346.2
116.6
312.0
55.0
48.3
76.8
135.0
65.8
89.4
311.0
312.9
100.8
88.5
118.6
48.7
39.3
WIND DIRECTION
FOR ASARCO PLUME
TO REACH SITE
DEGREES
296.6
166.2
296.6
132.0
235.0
228.3
256.8
315.0
245.8
269.4
131.0
132.9
280,8
268.5
298.6
228.7
219.3
DISTANCE, KM
2.0
5.5
3.8
8.2
2.4
9.8
6.6
3.4
2.2
9.5
5.0
5.7
2.1
3.7
2.5
9.9
2.8
7-3
-------
the data from the monitors surrounding the ASARCO plant will be compared
with the annual Federal primary standard of .03 ppm and with a 1969 vari-
ance for 1-hour concentrations which requires values to be less than or
equal to 0.5 ppm 99.3% of the time. Peters has also suggested an absolute
1-hour standard of 0.85 ppm.
It should be noted that the ASARCO plant was the major contributor of
SO- in the El Paso area during the study period. There are few other heavy
industries and home heating is accomplished chiefly by natural gas.
Some industries use oil during period of unusually cold weather; however,
since the test period is in the spring, they, and home heating contributors
can be neglected as a source of SO-. A few minor sources of SO exist near
some of the monitoring stations but their contribution is easily identifiable.
Meteorological data was available in the form of standard National
Weather Service (NWS) surface and radiosonde observations at the El Paso
airport as well as wind and temperature sensors at the ASARCO plant.
Ground and 400-ft level wind instruments exist at ASARCO with temperature
sensors at 6, 175, 500 and 800-ft permitting determination of vertical
structure. Other details of important air pollution conditions in the
El Paso area are found in Peters (1971).
One further item on the terrain in the El Paso area is appropriate.
The spine of high terrain extending North-South just north of the downtime
area is the steep-sloping Franklin Mountain Range which on occasion acts as
a barrier to the ASARCO plume under southwesterly flow conditions. When this
occurs the stations at the base of the mountains such as Park Hill, Richmond
and Montridge are especially susceptible to high SO values. In addition,
the Sierra del Christo Rey peak, located west of ASARCO, rises to an eleva-
tion of 4,675-ft and can interfere with plume transport to locations NW of the
plant. This is most important at night when the plume barely clears the peak
and can get caught in a downwash effect on the lee side of the mountain.
The approach used in this case study first considers the gross statistical
properties of the complete dataset and then focuses on subsets of the data
of particular interest. An objective measure of system reliability is obtained
using the analytical techniques developed in this study.
7.2 Statistical Analysis of SCS Performance
Examination of the observed data reveals many of the system characteristics.
Table 7-2 shows the diurnal variation in concentration. It may be seen that
7-4
-------
TABLE 7-2
AVERAGES BY HOUR OF THE DAY [OBSERVED), PPM
. HOURS
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
^erage
RICHMOND
.0167
.0142
.0103
.0067
.0067
.0033
.0031
.0047
.0108
.0131
.0139
.0078
.0083
.0078
.0086
.0075
.0158
.0292 -
.0128
.0150
.0186
.0036
.0014
.0006
.0100
MISSOURI
.0194
.0145
.0170
.0252
.0229
.0294
.0356
.0300
.0275
.0494
.0389
.0445
.0218
.0290
.0146
.0163
.0139
.0094
.0079
.0144
.0165
.0094
.0109
.0124
jV022jT_
MONTRIDGE
.0089
.0097
.0103
.0100
.0112
.0082
.0094
.0103
.0130
.0121
.0127
.0145
.0140
.0140
.0137
.0137
.0134
.0133
.0150
.0119
.0123
.0097
.0120
.0117
,0119 _ .
FARRELL
.0258
.0197
.0156
.0139
.0117
.0103
.0142
.0142
.0244
.0519
.0392
.0286
.0344
.0325
.0286
.0264
.0267
.0297
.0228
.0200
.0161
.0156
.0319
.0397
..0247 _
ZORK
.0061
.0067
.0064
.0086
.0069
.0092
.0083
.0103
.0156
.0172
.0200
.0089
.0081
.0061
.0047
.0025
.0028
.0064
.0047
.0208
.0167
.0075
.0047
.0069
^0090;
VECK
.0186
.0175
.0156
.0150
.0203
.0164
.0117
.0167
.0183
.0377
.0426
.0331
.0389
.0403
.0378
.0281
.0292
.0386
.0178
.0161
.0142
.0178
.0289
.0297
.0250
PARK HILL
.0039
.0011
.0017
.0008
.0031
.0011
.0031
.0028
.0017
.0028
.0058
.0058
.0094
.0011
.0017
.0015
.0011
.0077
.0031
.0011
.0011
.0022
.0017
.0014
.0028"
CORONADO
.0061
.0058
.0061
.0053
.0056
.0047
.0042
.0064
.0086
.0125
.0161
.0161
.0081
.0072
.0089
.0106
.0094
.0122
.0089
.0078
.0078
.0069
.0078
.0081
^.0084 _
MARLIN
.0017
.0022
.0017
.0014
.0008
.0006
.0014
.0022
.0042
.0033
.0008
.0006
.0006
.0003
.0003
.0003
.0006
.0006
.0011
.0006
.0000
.0000
.0003
.0003
.0011 .
ROBINSON
.0119
.0128
.0122
.0122
.0192
.0194
.0228
.0242
.0225
.0475
.0572
.0589
.0372
.0614
.0650
.0708
.0894
.0881
.0792
.0356
.0192
.0189
.0206
.0108
_.0382
RIVER
.0058
.0136
.0147
.0189
.0289
.0169
.0500
.0494
.0456
.0409
.0625
.0531
.0326
.0317
.0250
.0181
.0311
.0333
.0208
.0072
.0139
.0136
.0086
.0094
,0226
HAWTHORNE
.0175
.0264
.0233
.0292
.0144
.0131
.0333
.0164
.0211
.0208
.0350
.0217
.0269
.0167
.0133
.0177
.0094
.0144
.0225
.0378
.0447
.0342
.0200
.0264
~0229 1
CHELMONT
0175
.0267
.0236
.0164
.0192
.0219
.0186
.0289
.0251
.0246
.0367
.0220
.0174
.0209
.0218
.0233
.0272
.0275
.0194
.0200
.0156
.0211
.0389
.0172
.0230 __,
UTEP
.0275
.0275
.0289
.0357
.0254
.0234
.0337
.0271
.0335
.0569
.0709
.0662
.0765
.0691
.0621
.0547
.0533
.0536
.0483
.0569
.0531
.0439
.0317
.0381
7p_456'
4-SEAS.
.0153
.0014
.0022
.0006
.0053
.0014
.0033
.0025
.0042
.0064
.0267
.0256
.0431
.0272
.0286
.0186
.0075
.0167
.0031
.0003
.0003
.0011
.0014
.0008
.0096
RIM
.0236
.0353
.0267
.0303
.0292
.0278
.0206
.0200
.0203
.0350
.0381
.0372
.0428
.0614
.0606
.0650
.0814
.0739
.0367
.0325
.0331
.0269
.0289
.0222
.0379.
TURKEY
.0189
.0140
.0129
.0223
.0174
.0114
.0140
.0160
.0154
.0211
.0380
.0366
.0546
.0437
.0650
.0506
.0419
.0464
.0217
.0206
.0172
.0192
.0172
.0164
-0212
ALL STATIONS
.0136
.0147
.0134
.0148
.0142
.0127
.0168
.0164
.0182
.0264
.0325
.0286
.0277
.0275
.0271
.0247
.0268
.0296
.0204
.0188
.0177
.0148
.0157
.0149
.... 0?03~ .
I
in
-------
ten of the seventeen monitors exhibit maximum average hourly values between
10 AM and 12 noon. Many of the individual hourly averages exceed the 0.03 ppm
Federal annual primary standard. In fact, in several cases the maximum
approaches 0.09 ppm. However, only three of the seventeen monitoring stations
considered exceeded .03 ppm for all hours combined. The spatial distribution
of test period means is given in Figure 7-2. The average concentrations over
the 36 day period for each station implies that the annual standard could be
violated at one or more stations. Extending the comparison to the 3-hour
and 24-hour Federal S02 standards, 0.5 and 0.14 ppm respectively, violations
of the former occurred at River, and of the latter at two stations, Missouri
and Robinson.
The frequency distribution of observations exceeding selected thresholds
is included in Table 7-3. The hourly standard suggested by Peters (1971) and
granted by variance was satisfied, since not more than 99.3% of the hourly
observations exceeded 0.5 ppm at any location. However, the absolute standard
of 0.85 ppm was exceeded twice with values of 0.92 ppm at Missouri and 1.07 ppm
at River. Both of these monitors are located in New Mexico, however, where
the 1-hour standard is not applicable.
The spatial character of the SO distribution resulting from the emissions
of the ASARCO plant is shown in Figures 7-3A and 7-3B which are plots of
the number of observations exceeding 0.1 and 0.5 ppm over 1-hour, respectively.
In both diagrams two maxima are evident, one located to the northwest of the
plant, the other downwind from the plant in the direction of prevailing
westerlies. Although the latter maximum, centered at Robinson, is almost
twice as great as the former in Figure 3A, the situation is reversed in 3B.
The use of the SCS was effective in reducing the severity of a large number
of cases for eastern sector locations but was less effective for the north-
west stations.
In order to evaluate the effect of using the SCS control procedures,
it is necessary to estimate air quality for full load conditions that is with-
out the load reductions carried out under the direction of the SCS decision
logic. An approximate method to do this is to simply divide the observed air
quality values at the monitoring sites by the fraction of full load at which
the plant is operating under the SCS control. That is, if concentration
7-6
-------
0
Figure 7-2 Mean Observed Concentrations for the Test Period in Units of ppm.
-------
8265
TABLE 7-3
DISTRIBUTION OF HOURLY CONCENTRATIONS, ACTUAL DATA
STATION
RICHMOND
MISSOURI
MONTRIDGE
FARRELL
ZORK
VECK
PARK HILL
CORONADO
MARL IN
ROBINSON
RIVER
HAWTHORNE
CHELMONT
UTEP
4 -SEASONS
RIM
TURNEY
TOTAL
% OF 'THE
TIME
EXCEEDING
NUMBER OF
OBSERVATIONS
864
778
828
864
864
859
857
864
864
864
861
864
857
842
864
864
850
14508
>0.1
PPM
26
43
3
38
13
38
6
4
0
101
65
59
31
95
19
74
46
660
4.5
>0.2
PPM
8
12
0
15
4
16
1
2
0
50
33
22
13
24
11
27
11
249
1.7
>0.3
PPM
3
3
0
6
0
11
0
0
0
19
16
11
3
8
5
8
5
98
.68
>0.4
PPM
2
3
0
3
0
3
0
0
0
4
10
6
0
4
5
0
2
41
.28
>0.5
PPM
0
3
0
2
0
2
0
0
0
0
6
2
0
0
1
0
1
16
.110
>0.6
PPM
0
1
0
1
0
1
0
0
0
0
4
0
0
0
0
0
0
7
051
>0.7
PPM
0
1
0
1
0
1
0
0
0
0
2
0
0
0
0
0
0
5
.036
>0.8
PPM
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
2
.014
>0.9
PPM
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
2
.014
>1.0
PPM
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
1
007
-------
is assumed to be proportional to emission rate so an hourly observation can
be increased to the value expected at full load by considering the following
simple proportionality.
C Q
C _ X
C ~ 0
o xo
where
C = concentration under SCS control
C = concentration expected under full load
Q = SCS emission rate
Q = emission rate at full load
In this case, 0 will be set to an emission rate for SCL of 22 tons (hr) ,
the full load condition.
In deriving the full-load dataset it was not immediately clear whether
to associate.observed concentrations with concurrent operating conditions
or insert some sort of time lag to account for the clearing time. Since the
load data is given in hourly increments, both concurrent and 1-hour lag
pairings are logical choices. The concurrent pairing would be valid so long
as the clearing time, defined as the time necessary for emissions at ASARCO
to affect a location,-is 1/2 hour or less. Since none of the monitors is
more than 10 km from the plant, a critical transport wind of at least
20 km hr~ or 12 kts is required. For lower wind velocities, the 1-hour
time lag pairing is appropriate down to 3 kts. Since the plume is trans-
ported by flow at and above stack height, on the order of 1000-ft above
ground-level, the low wind cases are considered to be not important.
Based upon the above considerations, two datasets were derived -- one
with no lag and the other with a 1-hour lag between reported emissions and
sensor observations. Since the purpose of this procedure is to examine the
effect of the SCS in the statistical sense, a comparison of the statistics of
both sets of pseudo full load air quality observations under SCS control and
between themselves is useful.
Summaries of both derived full load datasets is presented in Tables 7-4
and 7-5. They are equivalent to Table 7-2 for the actually observed, SCS
7-9
-------
25 RT
I
I'
o
10
0
0
10
15
Figure 7-3a Number of 1-Hourly Averages Greater Than or Equal to
0.1 ppm During the Test Period.
-------
Figure 7-3b Number of 1-Hourly Averages Greater Than or Equal to
0.5 ppm During the Test Period.
-------
TABLE 7-4
PROJECTED AVERAGES (NO LAG), PPM
HOURS||RICHMOND
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
'l600
1700
1800
1900
2000
2100
2200
2300
2400
Averag
.0294
.0239
.0156
.0072
.0067
.0033
.0031
.0053
.0186
.0306
.0456
.0258
.0233
.0183
.0194
.0161
.0253
.0531
.0164
.0194
.0325
.0083
.0022
.0006
e .0187
MISSOURI
.0194
.0145
.0170
.0252
.0229
.0294
.0359
.0426
.0533
.1556
.1395
.1824
.0596
.0687
.0333
.0318
.0317
.0276
.0097
.0150
.0209
.0097
.0121
.0124
.0446
MONTRIDGE
.0091
.0097
.0109
.0106
.0112
.0082
.0094
.0138
.0248
.0279
.0352
.0391
.0386
.0297
.0289
.0263
.0263
.0278
.0194
.0122
.0126
.0097
.0123
.0117
.0193
FARREU
.0258
.0197
.0158
.0142
.0117
.0103
.0142
.0178
.0411
.0942
.0861
.0569
.0656
.0544
.0522
.0461
.0492
.0567
.0294
.0200
.0178
.0164
.0325
.0397
.0370
ZORK
.0064
.0069
.0067
.0086
.0069
.0092
.0083
.0131
.0267
.0403
.0725
.0308
.0228
.0142
.0103
.0044
.0050
.0111
.0053
.0225
.0228
.0078
.0047
.0069
.0156
VECK
.0192
.0181
.0158
.0150
.0203
.0164
.0117
.0208
.0314
.0697
.1029
.0703
.0694
.0639
.0628
.0489
.0564
.0672
.0242
.0164
.0153
.0181
.0292
.0297
.0379
PARK HILL
.0039
.0011
.0017
.0008
.0031
.0011
.0031
.0036
.0033
.0042
.0178
.0194
.0509
.0020
.0043
.0032
.0026
.0280
.0089
.0011
.0011
.0022
.0017
.0014
.0070
CORONADO
.0061
.0058
.0061
.0053
.0056
.0047
.0042
.0081
.0147
.0297
.0441
.0497
.0236
.0136
.0158
.0192
.0178
.0239
.0106
.0078
.0081
.0069
.0081
.0081
.0145
MARLIN
.0019
.0022
.0019
.0014
.0008
.0006
.0014
.0036
.0106
.0075
.0025
.0022
.0006
.0003
.0003
.0003
.0008
.0008
.0017
.0006
.0000
.0000
.0003
.0003
.0018
ROBINSON
.0131
.0147
.0128
.0122
.0214
.0275
.0261
.0364
.0403
.1106
.0517
.1308
.1067
.1225
.1508
.1578
.1989
.1756
.0964
.0356
.0206
.0211
.0225
.0108
.0674
RIVER
.0058
.0144
.0156
.0189
.0239
.0169
.0500
.0586
.0636
.0600
.1031
.0914
.0549
.0489
.0447
.0297
.0536
.0567
.0253
.0072
.0139
.0136
.0086
.0094
.0369
HAWTHORNE
.0175
.0283
.0303
.0386
.0150
.0139
.0344
.0225
.0392
.0550
.1486
.0728
.0697
.0389
.0322
.0236
.0222
.0364
.0289
.0394
.0664
.0422
.0253
.0267
.0403
CHELMONT
.0194
.0297
.0239
.0164
.0192
.0219
.0186
.0400
.0571
.0534
.0983
.0683
.0463
.0411
.0459
.0469
.0564
.0533
.0258
.0203
.0175
.0233
.0486
.0175
.0378
UTEP
.0283
.0294
.0343
.0423
.0260
.0251
.0351
.0359
.0615
.1294
.2359
.1888
.2032
.1509
.1376
.1176
.1144
.1281
.0614
.0592
.0725
.0539
.0358
.0386
.0844
4-SEAS.
.0014
.0014
.0022
.0006
.0053
.0014
.0033
.0031
.0078
.0144
.1075
.0814
.1625
.0794
.0656
.0411
.0147
.0592
.0050
.0003
.0003
.0011
.0014
.0008
.0276
RIM
.0253
.0367
.0275
.0308
.0292
.0281
.0206
.0250
.0394
.0847
.1100
.1025
.1036
.1303
.1344
.1364
.1669
.1475
.0483
.0364
.0450
.0358
.0333
.0222
.0667
TURNEY
.0191
.0143
.0134
.0226
.0174
.0117
.0140
.0200
.0280
.0483
.1277
.1009
.1654
.1074
.1439
.1167
.0928
.1125
.0297
.0219
.0194
.0208
.0183
.0164
.0543
ALL STATIONS
.0148
.0160
.0148
.0158
.0144
.0134
.0172
.0216
.0328
.0591
.0953
.0766
.0742
.0576
.0578
.0516
.0515
.0628
.0263
.0197
.0228
.0172
.0175
.0149
.0361
-------
TABLE 7-5
PROJECTED AVERAGES (1 - HR TIME LAG)
HOURS|RICHMOND
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
Averag
.0294
.0219
.0197
.0094
.0075
.0033
.0031
.0047
.0125
.0228
.0294
.0261
.0217
.0217
.0206
.0181
.0267
.0592
.0256
.0178
.0228
.0072
.0028
.0006
s .0181
MISSOURI
.0197
.0145
.0170
.0252
.0229
.0294
.0356
.0303
.0384
.0191
.0912
.1552
.0687
.0814
.0318
.0314
.0300
.0229
.0158
.0174
.0165
.0109
.0118
.0136
.0385
MONTRIDGE
.0091
.0097
.0106
.0106
.0115
.0082
.0094
.0103
.0167
.0230
.0258
.0406
.0371
.0363
.0274
.0286
.0263
.0258
.0322
.0147
.0123
.0100
.0120
.0120
.0192
FARRELL
.0258
.0197
.0156
.0142
.0119
.0103
.0142
.0142
.0311
.0908
.0619
.0594
.0656
.0603
.0508
.0483
.0497
.0542
.0450
.0261
.0161
.0172
.0328
.0403
.0365
ZORK
.0064
.0069
.0067
.0089
.0069
.0092
.0083
.0103
.0203
.0272
.0447
.0322
.0269
.0175
.0097
.0044
.0058
.0122
.0075
.0242
.0169
.0081
.0047
.0069
.0139
VECK
.0192
.0181
.0161
.0153
.0203
.0164
.0117
.0167
.0226
.0654
.0634
.0729
.0674
.0714
.0631
.0508
.0525
.0681
.0356
.0228
.0144
.0189
.0292
.0300
.0366
PARK HILL
.0039
.0011
.0017
.0008
.0031
.0011
.0031
.0028
.0019
.0042
.0092
.0161
.0360
.0031
.0034
.0038
.0029
.0123
.0097
.0011
.0011
.0022
.0017
.0014
.0053
CORONADO
.0061
.0058
.0061
.0053
.0056
.0047
.0042
.0064
.0106
.0239
.0264
.0456
.0203
.0167
.0156
.0189
.0167
.0219
.0178
.0092
.0078
.0072
.0078
.0083
.0133
MARLIN
.0019
.0022
.0019
.0017
.0008
.0006
.0014
.0022
.0072
.0069
.0019
.0019
.0008
.0003
.0003
.0003
.0008
.0008
.0028
.0008
.0000
.0000
.0003
.0003
.0016
ROBINSON
.0131
.0144
.0133
.0125
.0258
.0217
.0308
.0280
.0306
.0847
.1183
.1278
.1067
.1686
.1439
.1722
.2086
.2006
.1444
.0433
.0192
.0200
.0225
.0111
.0743
RIVER
.0058
.0136
.0156
.0197
.0239
.0169
.0500
.0494
.0542
.0563
.0997
.0860
.0540
.0581
.0417
.0333
.0531
.0558
.0375
.0086
.0139
.0136
.0086
.0094
.0365
HAWTHORNE
.0175
.0264
.0247
.0378
.0158
.0133
.0342
.0164
.0286
.0422
.0833
.0800
.0861
.0444
.0317
.0225
.0217
.0333
.0400
.0461
.0456
.0383
.0247
.0300
.0369
CHELMONT
.0194
.0292
.0239
.0167
.0192
.0219
.0186
.0289
.0346
.0463
.0778
.0666
.0469
.0511
.0438
.0517
.0575
.0572
.0381
.0275
.0164
.0228
.0483
.0206
.0368
UTEP
.0286
.0283
.0309
.0420
.0274
.0240
.0366
.0274
.0429
.1049
.1603
.0897
.2259
.1847
.1347
.1209
.1144
.1183
.0906
.0697
.0547
.0542
.0364
.0425
.0822
4-SEAS .) RIM [TURKEY J| ALL STATIONS
.0014
.0014
.0022
.0006
.0053
.0014
.0033
.0025
.0053
.0117
.0694
.0661
.1350
.1064
.0728
.0325
.0158
.0272
.0086
.0003
.0003
.0011
.0014
.0008
.0239
.0253
.0364
.0278
.0322
.0297
.0278
.0206
.0200
.0258
.0694
.0867
.0903
.1061
.1564
.1336
.1351
.1706
.1478
.0744
.0411
.0367
.0356
.0353
.0239
.0622
.0191
.0143
.0131
.0226
.0177
.0114
.0143
.0160
.0200
.0380
.0914
.0949
.1491
.1351
.1453
.1133
.0933
.0825
.0436
.0278
.0181
.0211
.0189
.0175
.0517
.0148
.0156
.0145
.0161
.0149
.0129
.0175
.0168
.0236
.0472
.0669
.0730
.0734
.0710
.0572
.0574
.0558
.0590
.0395
.0235
.0184
.0170
.0176
.0159
.0350
-------
controlled values. The network mean exceeds the Federal annual standard
in both derived datasets. Thus, with no meteorological control at ASARCO
both long and short-term S0_ observations would have been dramatically
higher in the El Paso area.
The hourly averages for all stations combined in the derived full load
datasets show the same diurnal pattern as the observed data but with much
more amplitude. This is to be expected in light of the predominance of load
reductions during the early morning.
Another characteristic of the load-reduction system is evident upon
comparison of the two derived datasets. Both the mean and number of hourly
concentrations greater than 0.1 ppm are generally smaller in the 1-hour lag
derived data. This is due to a small reduction in load on the average before
the expected occurrence of a significant concentration in the 1-hour lag
derived data. In most cases, the load reduction condition had already been
in operation for several hours such that no difference exists between
the no-lag and 1-hour-lag. It is those few cases where .anticipation of high
SCL values in the following hour or so forced a sudden load reduction which
accounts for the difference in statistics. Some of these cases can be noted
in Table 7-6 where the maxima for the 1-hour lag data are sharply different
from the no lag, e.g., at Missouri, Hawthorne and Rim. Other than this,
both datasets exhibit similar properties, and therefore, only the "no time-
lag" derived data will be used hereafter for comparison with observed data.
Given both the full load and SCS controlled ambient air quality values,
the reduction in air quality values due to the use of the SCS can be cal-
culated. For instance, Figure 7-4 indicates the percent reduction in mean
SO concentration due to the use of the SCS at ASARCO. Similarly; in Table 7-7
the percent decrease in hourly concentrations due to the use of the SCS is
presented. The net effect of the control procedure on air quality values
may be easily seen. It is most effective for stations east and close to
ASARCO, particularly between the hours of 1100 and 1300 LST. At stations
and times for which the percentage reduction is not as large, with the
exception of Veck, River and Farrell, all to the northwest of the plant,
a serious problem^does not exist because concentrations for these cases are
safely below standards.
Another result of the SCS is a shifting of the frequency distributions
of concentration toward the lower end of the spectrum with the greatest
7-14
-------
TABLE 7-6
10
8
CO
ACTUAL DATA CONG. >0.10 ppm
By Station
STATION
RICHMOND
MISSOURI
MONTRIDGE
FARRELL
ZORK
VECK
PARK HILL
CORONADO
MARL IN
ROBINSON
RIVER
HAWTHORNE
CHELMONT
UTEP
4 -SEASONS
RIM
TURNEY
MAXIMUM OBSERVED
CONG, ppm
.48 ppm
.92 ppm
.16 ppm
.77 ppm
.24 ppm
.71 ppm
.21 ppm
.25 ppm
.07 ppm
.48 ppm
1.09 ppm
.59 ppm
.39 ppm
.45 ppm
.58 ppm
.36 ppm
.55 ppm
DERIVED DATA - NO TIME LAG
CONG. >0.10 ppm
STATION
! RICHMOND
MISSOURI
MONTRIDGE
FARRELL
ZORK
VECK
PARK HILL
CORONADO
MARLIN
ROBINSON
RIVER
HAWTHORNE
CHELMONT
UTEP
4 -SEASONS
RIM
TURNEY
MAXIMUM DERIVED
CONG, ppm
.94 ppm
3.49
ppm
. 37 ppm
1.47 ppm
1.27 ppm
1 . 72 ppm
.84 ppm
. 83 ppm
.16 ppm
1 .03 ppm
1.09 ppm
3.09 ppm
.57 ppm
2.36 ppm
3.08 ppm
1.32 ppm
2.25 ppm
7-15
-------
I
Figure 7-4 Percentage Reduction in S0n Concentrations Attributable
to SCS CurtaiImant.
-------
TABLE 7-7
PERCENT REDUCTION OF HOURLY CONCENTRATIONS
DUE TO SCS *
TIME
0100
0200
0300
0400
0500
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
ALL HOURS
PERCENT REDUCTION
8.1
8.1
9.5
6.3
1.4
5.2
2.3
24.0
44.5
55.3
65,9
62.7
62.7
52.3
53.1
52.1
48.0
52.9
22.4
4.6
18.0
14.0
10.3
0.7
43.8
"Individual hour reductions based on % decrease of mean hourly concentrations
for network due to SCS from Tables 7-2 and 7-4. Category "All Hours" is
based on % decrease of 36^-day test period network mean due to SCS.
7-17
-------
shifting occurring at the highest concentrations. Table 7-8 shows the 50 to
75% reduction in the number of cases with moderate concentration (0.1-0.5 ppm)
while the 0.5 ppm and higher categories experience reductions of 90% and
greater. To be effective the SCS must eliminate the occurrence very high
concentrations. Other than a total shutdown of operations practical load
reductions are limited to factors of 1/4 to 1/7. Under very adverse meteor-
ological conditions ambient SCL concentration in the moderate range but not
exceeding standards could still be expected.
The ASARCO System is found to reduce SO- concentrations significantly in
the El Paso area. If full-load operating conditions had existed at all times
during the test period, an intolerable number of violations of all short-term
S02 standards would have occurred. In this respect the SCS is effective; the
reliability of these actions can be measured by their ability to prevent viola-
tions. Since 17 monitored values greater than the 0.5 ppm one-hour standard
occurred, the necessary reliability has not been achieved. It is valuable to
investigate these cases individually in order to identify the meteorological
circumstances responsible for the violation. Remedial action can then be taken
to improve forecast accuracy by addressing these specific meteorological cases
and developing new forecast guidance.
7.3 Analysis of Individual Instances of High S0» Concentrations
The cases analyzed here are those in which the standard has been exceeded.
For reference, the 17 cases when 0.5 ppm was exceeded in an hourly average
were first selected. From this group 5 cases where either the absolute stand-
ard of 0.85 ppm was violated and/or more than one station exceeded 0.5 ppm
during an episode were chosen for an hour by hour analysis.
The purpose in analyzing these cases individually is to identify the
weaknesses of the ASARCO SCS. Once this is done, any potential improvement
can be evaluated in terms of increased accuracy and reliability gained in
overcoming these weaknesses. It is recommended that analysis of other SCS's
include a step such as this where the cases of difficulty can be isolated and
specific proposals be made if modifications in the SCS are needed.
7-18
-------
S268
TABLE 7-8
DISTRIBUTIONS OF HOURLY CONCENTRATIONS
Generated Data, no Lag
STATION
RICHMOND
MISSOURI
MONTRIDGE
FARRELL
ZORK
VECK
PARK HILL
CORONADO
MARL IN
ROBINSON
RIVER
HAWTHORNE
CHELMONT
UTEP
4 -SEASONS
RIM
TURNEY
TOTAL
NUMBER OF
OBSERVATIONS
864
778
828
864
864
859
857
864
864
864
861
864
857
842
864
864
850
14508
% REDUCTION
DUE TO SCS
ACTUAL CONDITIONS
>0.1
PPM
44
73
16
59
28
70
8
12
3
146
91
89
76
209
34
133
96
1187
44.4
>0.2
PPM
19
35
3
26
10
29
7
6
0
97
50
48
29
98
20
76
46
599
58.4
>0.3
PPM
12
21
2
12
7
14
4
3
0
70
27
29
14
52
14
50
29
360
72.8
>0.4
PPM
7
17
0
7
5
7
4
2
0
45
16
22
7
30
13
25
18
225
81.8
>0.5
PPM
5
16
0
3
2
5
4
2
0
30
8
15
4
12
12
12
12
142
88.7
>0.6
PPM
4
9
0
2
1
2
3
2
0
21
7
6
0
7
8
8
8
88
92.0
>0.7
PPM
3
6
0
2
1
2
3
1
0
14
4
4
0
6
7
4
5
62
91.9
>0.8
PPM
2
5
0
2
1
2
3
1
0
7
3
.1
0
4
7
3
4
45
95.6
>0.9
PPM
1
5
0
2
1
2
0
0
0
6
2
1
0
3
7
3
4
37
94.6
>1.0
PPM
0
5
0
2
1
2
0
0
0
4
2
1
0
3
7
2
3
31
97.0
-------
The five eases presented here cover the following time periods:
Case A 0900 to 1900 3/8
B 2200 to 1600 3/19-20
C 2200 to 1400 4/2-3
D 0600 to 0900 4/6
E 0900 to 1500 3/28
Cases A through D involve northwest sector stations, River, Farrell and
Veck while E describes a reading of 0.92 ppm at Missouri.
Figures 7-5 through 7-10 present plots of meteorological conditions and
concentration fields for Cases A and C. The format for the plots is as follows.
The monitoring stations have their hourly concentrations printed in parts per
hundred million (pphm). At ASARCO two wind observations were available, one
close to ground level and the other at 400 ft. For plotting purposes the
400-ft observation was used as it most closely represents transport conditions
at stack exit height. The wind at the plant and the NWS airport wind obser-
vation are plotted using conventional symbols. The airport wind is the 3-hourly
observation closest to plotting time. The number to the left of the wind
flag at ASARCO is the 175-800 ft temperature difference in degrees Fahrenheit
while the number on the right is output of sulfur in tons (hour) , 22 tons
(hr) representing the full load condition. With reference to dry adiabatic
conditions, the temperature difference for the 175-800 ft layer is -6.2°F.
Table 7-9 lists the 850 mb data, the height, (hhh) in meters, temperature
(tt), in °C, and wind direction and speed (dddff) in degrees and m sec
The 850 mb data is from twice daily NWS radiosonde measurements taken at
the airport. Most of the time 850 mb conditions are representative of a
level just above the maximum plume elevation and combined with the ASARCO
data can be useful in determining wind shear and stability conditions.
A) March 8 - 0900 to 1900
This case is of particular interest because of the occurrence of high
concentrations for two consecutive hours at two separate locations. The
significant observations were .71 ppm at Veck and .49 ppm at Farrell at
1100 and .50 at Four-Seasons at 1200. A wind shift from south and south-
east winds to NW is responsible for this curious event.
7-20
-------
I
to
Figure 7-5 Time: 0900 Date: 3-8-71
-------
25 FT:
I
ro
t-o
Figure 7-6 Time: 1000 Date: 5-8-71
-------
- J
I
I J
Figure 7-7 Time: 1100 Date: 5-8-71
-------
25
0
Figure 7-8 Time: 1200 Date: 3-8-71
-------
r j
en
0
10
Figure 7-9 Time: 0900 Date: 4-3-71
-------
0
Figure 7-10
Time: 1000
Date: 4-3-71
-------
TABLE 7-9
NATIONAL WEATHER SERVICE RADIOSONDE DATA
DAY
Case A
3/8
3/9
Case B
3/20
3/21
Case C
4/3
4/3
4/4
Case D
4/6
4/6
4/7
Case E
3/28
3/29
TIME
12Z
OOZ
12Z
OOZ
OOZ
12Z
OOZ
OOZ
12Z
OOZ
12Z
OOZ
HHH,M
1493
1467
1542
1466
1536
1502
1467
1554
1606
1584
1500
1494
TT, °C
8.3
19.0
8.2
18.1
17.8
8.5
6.4
14.2
0,8
12.5
18.0
24.7
ODD,
DEGREES
089
240
133
234
063
161
064
123
089
126
304
281
FF-1
M SEC *
02
05
08
04
02
04
01
01
12
05
06
06
7-27
-------
The wind shift was due to the passage of a weak upper level disturbance
with its surface reflection being a weak trough. The gradient wind is weak
before the shift having a general southerly component. The southerly wind
at ASARCO developed quickly at 1000 and was manifested by sharply increasing
concentrations at the NW sector stations. Despite a sharp decrease in load
from near full operating conditions to 9 tons (hr) , it was not enough to
prevent the morning fumigation from yielding high S0~ readings at 1100.
Wind velocities were light resulting in a long clearing time. This implies
that the load reduction took place too late. After the shift to light westerly
winds, the center of maximum concentration is advected southeastward to the
vicinity of Four-Seasons. The short-lived peak at Coronado is indicative of
this movement. It is likely that the maximum is sustained due to surface wind
convergence despite the fact that fumigation had already been in progress for
almost two hours and that the plume should have been well dispersed. Further
downward load reductions occur until 1300 when a minimum load of 3 tons (hr)
is reached. This load is increased slowly during the rest of the day and is
justified since concentrations remain well below 0.5 ppm.
Of extreme importance in the analysis of plume transport in the El
Paso region is the modifying effect on airflow due to the presence of
heterogeneous terrain. It is clear from the hourly maps that wind obser-
vations at the airport and the plant are often strikingly different both
in direction and velocity. Since any type of forecast relies in some
measures upon the mean transport wind, some knowledge of the relationship
between an observed wind and the effective advective flow is required. For
the ASARCO meteorologist the 400 ft wind at the plant is the most useful
measurement available.
The increase in concentration at the northwest stations after 1000 is
associated with persistent southerly flow at ASARCO. Based solely on the
ASARCO wind, the station most likely to receive significant concentrations
is Coronado. However, the plume centerline impacts at ground level near
Veck while Coronado "feels" the plume only during the wind shift. The
most likely trajectory of the ASARCO plume is a curved one; one that ini-
tially tracks northward but then turns to the west tracing the outline of
the Cristo Rey mountain.
The topographical effect can manifest itself in two ways, large vertical
wind shear may result from highly perturbed flow at low layers or horizontal
shear may result from perturbed flow at one locale. Most of the time the
7-28
-------
effects are combined. In this case the trajectory was affected by the wind
shear through some depth because the plume rise was substantial during the
light wind conditions. Under stable conditions, particularly at night, plume
rise is small and the plume is constrained to disperse in a horizontal plane.
When this occurs the elevated plume will trace out a curved path from the
plant. Morning fumigation can then take place at some unexpected location.
It has also been noted by ASARCO meteorologists that mechanical turbulence
and lee waves generated by Cristo Rey increase dispersion and can actually
fumigate the plume at night. Examples of these occurrences are given below
in cases B and D.
B) March 19 - 2200 to March 20 1600
The half-hourly data, given in Table 7-10 is presented in this case.
Synoptically. a vigorous southeasterly gradient flow becomes established
during the first part of the period but then relaxes and assumes a weaker,
westerly character. Since a uniform flow persists for 12 hours after 3/20
0200, minor wind fluctuations can be associated with corresponding responses
in the concentration field. It is significant that full and near-full
load conditions existed through most of the period and were lowered appreciably
only after the shift to west winds despite the occurrence of high concentra-
tions for several hours. Apparently there was little response to the initial
situation when hourly concentrations reached .66 ppm at 0500, .65 ppm at 0900
and .70 ppm at 1100 at River on the 20th.
The thermal stability was uniformly neutral throughout the past-midnight
period and the flow went from 7 to 17 knots at 400 ft just before sunrise.
Since the 850 mb flow at 0500 is from the SE, the likely plume trajectory is
a curved one through some depth. Note that the directional shear is opposite
to the one that would be expected in an Ekman layer, probably due to topo-
graphic influence. The sharp difference in wind observations between the
airport and ASARCO supports this view.
The first incidence of moderate concentration causes a load reduction
response at 0200 but by 0400 the plant returned to full load operations.
An increase in wind speed between 0400 and 0500 causes the first high
concentration at River. Since the stability remains constant the suspected
cause is mechanical turbulence on the lee side of Cristo Rey resulting in
7-29
-------
TABLE 7-10
METEOROLOGICAL CONDITIONS AND S02 CONCENTRATIONS
FOR MARCH 19 - 2200 TO MARCH 20 - 1600 CASE
TIME/DAY
2000 3/19
30
2100
30
2200
30
2300
30
2400
30
0100
30
0200
30
0300
30
0400
30
0500
30
0600
30
0700
30
0800
30
0900
30
1000
30
1100
30
1200
30
1300
30
1400
30
1SOO
30
1600
. 30
1700
30
1800
CONCENTRATIONS, ppra
FARRELL
.01
.01
.01
.01
.02
.06
.18
.21
.20
.09
.11
.07
.02
.02
.02
.03
.02
.01
.01
0
0
0
.02
0
0
.01
.01
.06
.02
.01
0
.02
.02
.06
.10
.20
.23
.17
.03
.01
.01
.01
.01
.01
0
RIVER
.01
0
0
0
.02
.03
.04
.03
0
.01
.03
.26
.35
.30
.25
.31
.31
.31
.85
.46
.29
.02
.26
.31
.37
.45
.54
.76
.63
.10
.40
1.00
.36
.56
.19
.12
.13
.05
.17
.03
0
0
0
0
0
VECK
.02
.02
.02
.03
.11
.18
.13
.12
.11
.07
.05
.02
.07
.01
.03
.04
.01
0
.01
0
.02
.03
.03
.01
0
0
.07
.07
.03
0
0
.03
.02
.06
.17
.33
.41
.42
.24
.04
.01
.01
.02
.02
.02
TUKNEY
0
0
0
.10
.10
.08
0
0
.09
MSG
MSG
.01
.05
.17
.13
.13
.02
0
0
CORONADO
.01
.01
.01
.01
.01
.01
.01
.01
.01
.02
.01
MSG
MSG
.01
.05
.03
.02
0
0
0
0
MET CONDITIONS
AIRPORT
dir[spd
360
330
030
020
360
140
160
290
06
03
03
07
09
13
07
03
ASARCO
400'
dirf spd
093
292
272
161
169
189
183
MSG
199
193
204
169
163
166
165
178
177
162
157
162
162
173
183
MSG
163
163
159
166
171
179
173
174
178
168
177
180
182
156
176
MSG
173
327
324
332
330
05
03
02
10
07
06
11
10
9
8
11
13
12
14
12
12
14
17
16
16
11
11
17
14
16
18
18
18
19
20
17
16
14
12
10
8
8
4
4
10
8
7
4T,°F
-3.6
-2.7
-0.2
-1.5
-2.2
-1.9
-3.9
MSG
-4.9
-3.8
-4.9
-4.1
-4.1
-4.2
-4.0
-4.0
-4.1
-3.7
-4.0
-3.9
-4.1
-4.2
-3.7
MSG
-4.4
-4.8
-5.2
-5.1
-5.1
-5.3
-5.3
-5.3
-5.6
-6.3
-6.1
-6.5
-7.1
-6.3
-6.4
MSG
-4.9
-4.4
-4.8
4.6
4.4
LOAD, TONS
(HRr1
22
22
22
22
22
22
20.1
19.9
22
22
22
21.3
18
18
18
18
18
18
18
7.1
6.0
12
7-30
-------
a "downwash" effect. The second occurrence at 0900 is associated with a
fumigation condition arising from strong low-level solar heating. The
third peak at 1130 is probably a combination of a looping, curved plume
(the wind at ASARCO is now more southerly) vacillating and fumigating at
the same time. The meteorological observations infer a highly turbulent
situation in which a complex interaction of terrain and atmospheric processes
take place.
The shift to west winds is immediately accompanied by a load reduction
of about 1/3. This prevents a high concentration at Turney but would have
been equally as effective several hours earlier.
C) 2200 April 2 to 1400 April 3
The significant concentrations occurred at 1000 of April 3 with .52 ppm
and .77 ppm being recorded at Veck and Farrell respectively. Data from River
was missing at this time. This case possesses many of the same characteristics
of the two previous ones: a well-established SE graident wind (note NE winds
at the airport again), the high probability of a curved plume trajectory from
the plant to the affected stations and an insufficient load response on the
part of ASARCO during the period of high concentration.
Thermal stability was somewhat larger during the nighttime hours inferring
a limited plume rise. Fumigation of the plume took place during the morning
hours when the atmospheric stability decreased markedly. Note the odd time
series of load: A sharp decrease at 0900 was appropriate but the 40% increase
an hour later was not warranted. Stable stability reduced the air pollution
hazard the rest of the day.
D) 0600 April 6 0900
A well-developed southerly flow at ASARCO of 20 kts. strength persisted
through the early morning hours when the peak concentration of the period,
1.09 ppm at River, occurred 0700. Since the wind at ASARCO could not have
transported the plume to the eventual site of ground impaction, the tra-
jectory of the plume was curved, heading northward at stack exist and curving
westward shortly thereafter. Since the flow at ASARCO was strong, a "down-
wash" effect is suspected as the cause of the highly localized maximum con-
centration. This occurs when mechanical turbulence induced by the presence
of a topographic feature in the downwind direction "fumigates" the plume.
7-31
-------
The load response was inadequate since full-load conditions existed through
the period of high concentration and was reduced only slightly at 0800.
E) 0900 March 28 1500
Like case D the peak hourly concentration of .92 ppm at Missouri at
1000 was singular in character. The ASARCO 400 ft wind does not indicate
transport toward Missouri but the 0500 850 mb data suggests the upper flow
is in the right direction. The stability was decreasing rapidly at the
time of peak concentration so classic fumigation is the likely cause for
this episode. The load was reduced from 12 to 6 tons hr at 1000 but
the decrease was obviously insufficient.
It appears that the SCS worked well in this case despite the violation.
One problem may have been the fact that Missouri was inoperative for the several
hours prior to 1000 so that the meteorologists had no advance information of
a concentration build-up. Since the concentration decreased to moderate levels
in the following hour, the fumigation was probably very short-lived despite
its intensity. There seems to be little possibility in eliminating cases
such as this from the list of troublesome cases. With the load being low to
start, the necessary fast response to sharp increases in concentration was
lacking due to the nature of the operations involved, e.g. turning off
roasters and ovens, see Peters (1971).
The most serious forecasting error involves southerly flow at ASARCO.
Under these conditions, the plume takes on a curved trajectory impacting at
ground level in a location not easily forecast in advance. Information on
vertical wind shear through the plume rise depth, horizontal wind variability
and generated turbulence is required for a proper judgment. The location
of a topographic feature, such as the Sierra de Cristo Rey peak west of ASARCO,
is of prime importance not only when the track of the plume goes over the
obstacle but also around it. Failure on the part of ASARCO to account for
this is evident because the corresponding SCS response is often insufficient
to prevent a violation.
The ability to forecast these two topographic effects would markedly
enhance the ASARCO SCS's effectiveness, since the other prime cause of
elevated concentrations, morning fumigation, is well-handled. Needed is a
model which would simulate the important effects present.
7-32
-------
In assessing the modeling tools available to accomplish this, the influence
of topography in the wind field was modeled based on a theory of Anderson (1971).
This model accomplished a two-dimensional solution of the equation:
720 = u Vh (7-1)
where:
0 = potential function
u = horizontal "synoptic-scale" wind
h = lower boundary height, i.e. elevation above sea level
2
7,7 = gradient and Laplacian operators
Assuming an impermeable lid at height H and a non-zero vertical velocity at
the lower boundary, equation 7-1 yields a unique solution for a given terrain
field. In essence, the terrain-induced wind is integrated through depth H, so
that the average effect is determined. Finally,
70 = u!
gives the resulting deviation, u', to the mean wind. If equation 7-1 is solved
for two unit orthogonal wind vectors, any arbitrary u can be solved for simply
by resolving u into orthogonal wind components and computing the resultant
wind deviant.
Equation 7-1 was solved using the topography from the El Paso area as
the lower boundary condition. Assuming H = 1 km, the deviation of an easterly
"synoptic-scale" wind in the ASARCO vicinity toward the north, at about 10%
of the flow velocity. In addition, ASARCO is embedded in a local maximum of
wind deviation. Similarly, the deviation of a southerly wind near ASARCO is
toward the west at somewhat less than 10% of flow velocity. These magnitudes
of the wind deviants is linearly proportional to H. It appears that the solution
yields information in concert with experience. Since the largest deviation is
apt to occur in the lowest part of depth H, the 400' wind at ASARCO should re-
flect a wind deviation larger than the model calculation. Thus, by considering
this effect qualitatively, the model predicts a curved trajectory for a plume
being emitted into an easterly or southeasterly mean flow. A more thorough
treatment would be to construct trajectories of plume travel under different
wind conditions so that the location of the plumes could be better known.
7-33
-------
Such information would have been especially valuable at ASARCO since more
than half of the incurred violations involved curved trajectory transport.
Given the success in predicting fumigation under non-curved trajectory
conditions, forecast performance would be improved significantly for north-
west sector stations. Thus, overall system reliability would be improved.
The greatest improvement here is the objectification of forecast pro-
cedures where none previously existed. Likewise, further refinement of
predictive guidance would improve the accuracy of fumigation forecasts.
Expected improvement in this area may boost system reliability but probably
by a lesser amount than by accounting for the curved plume trajectory.
The final category of high pollution potential is the mountain "down-
wash" effect. To investigate this problem adequately and develop appropriate
forecast aids would involve the solution of a three-dimensional air-flow
model which incorporates the important physical effects leading to the
generation of lee waves. This is beyond the scope of this study, but if
accomplished, would help eliminate cases such as Case D and improve system
reliability. Data from pilot balloon measurements or low level soundings
taken at the plant to supplement the NWS upper-level data would be helpful
in defining the three dimensional aspects of air flow in this area.
7.4 Application of the Proposed Analytical Techniques to the Case Study Data
This section deals with the application of the analytical techniques
developed in Sections 3 and 4 using the ASARCO El Paso data. Input of the
meteorological function M, the emissions function Q, and forecast error
ratio R are required for use in ERT PROBL. These functions are given in
the form of frequency distributions. Output from ERT PROBL yields expected
frequency of occurrence of concentration categories and consequently, fre-
quency of occurrence of concentration exceeding specified thresholds. If
a threshold is associated with a standard, ERT PROBL output can also give
the expected number of violations.
For proper use of the analysis tools developed, the data sets required
above must be available directly from records of forecasts and observed con-
centrations. In the absence of these data, a valid analysis using the ERT
PROBL methods cannot be made. The kind of data available from the ASARCO
SCS prevented a straightforward use of ERT PROBL from historical data.
The primary deficiency results from the fact that the air quality forecasts
7-34
-------
which precipitated the call for emission reduction were not recorded.
This prevents a direct comparison of forecasted air quality levels with
observed air quality levels to assess forecast accuracy.
For purposes of illustrating the power of the analytical methods de-
veloped, however, it is of interest to hypothesize possible sets of the
required statistics from the ASARCO data used in the previous section.
To fit the ERT PROBL analysis format, it was necessary to establish
the following items from available data:
A probability density function of the meteorology function M
for the data set available.
Air quality forecasts projections for the data set period.
An objective scheme for curtailing emissions on the basis of
the air quality forecasts.
A probability density function for the air quality forecast
error ratio R.
The meteorology function M(t) could, in principle, be determined with
the use of a calibrated atmospheric dispersion model using meteorological
inputs for the two month data set period. For purposes of this illustration,
however, we chose to estimate M from the available air quality and emissions
data using the relation M(t) = C (t)/Q (t) where
C (t) = the maximum observed concentration at time t, and
Q (t) = recorded emission rate*
Once M(t) is known, the corresponding probability density function, P ,
can be calculated for the time period as discussed in Chapter 3.
Because forecasted air quality levels were not recorded, it was necessary
to synthesize a forecast scheme from which objective forecasts of expected
maximum hourly concentrations could be derived. It was recognized that
practically all of the load curtailments occurred during the morning hours
and lasted until sunset. Thus two groups of data were defined, the full
load cases in which no load reduction occurred and the load reduction cases
*This data consisted of records of requested curtailments of load on
the basis of the ASARCO meteorologist's instructions. We understand that the
actual emission rates may often be substantially lower than these records
show when other factors determine curtailment.
7-35
-------
where trangressions of an air quality threshold were expected. When the
load was reduced it was hypothesized that a concentration value greater
than some threshold value, y> was forecasted under full load conditions
and that the reduction called for was sufficient to reduce the maximum ex-
pected ground level concentration to the threshold value. This synthesized
set of forecast.then allows the distribution of the error ratio R to be
calculated as the ratio of the assumed threshold value to the observed
concentrations.
Thus, if Q (t) was less than the full load emission rate 0~,
R(t) = Y/CQ(t)
If Q (t) was equal to the full load conditions and C (t) < Y> we assumed
that persistance of present meteorological conditions was forecasted and
that R = C (t)/C (t-1) where C (t-1) is the value of C observed during
the previous hour. If Q (t) equalled 0~ and C (t) > j, the air quality
levels were assumed to be underpredicted and again R = j/C (t). Having
synthesized the function R(t), the corresponding probability density
function, P can be calculated for the study period.
The hypothesized emission control strategy to be used in the ERT PROBL
computations was constructed after the proportional reduction control strat-
egy discussed in Section 5 as follows. The predicted maximum concentration,
C is calculated using M and R from C = M R Q~. The product of M and
R incorporates the expected errors in forecasting the meteorology function.
Then a controlled value of Q is determined by:
Q=/QfifcP . Y
Having synthesized M, and the control strategy defining Q for the test
period, ERT PROBL can then be run to calculate the expected maximum concen-
trations C = M Q, and the corresponding frequencies of occurrence of various
levels of concentrations.
If the real SCS system were working exactly as hypothesized in the ERT
PROBL simulation, we would expect that the predicted concentration levels
and the corresponding frequencies of occurrence of these levels would be the
same as those of the observed data set. Differences in the results would
point to inaccurate estimations of the various probability functions in
particular, to unrealistic representation of the actual process curtailment
7-36
-------
control strategy and to inadequate knowledge of actual emission rate. ERT
PROBL was run for the control strategy hypothesized above with a control
threshold of y = .5 ppm.
Inspection of the data indicated that there exists a subset of hours
for which the SCS was not in effect either due to a plume trajectory over
the uninhabited areas of Mexico or shut-down of the plant. It is not obvious
to us which is occurring at a given hour because only an estimate of emissions
and not actual plant load was available. An indication of this condition is
very low S02 concentrations during the hours of fumigation and large thermal
instability, i.e. late morning and early afternoon. This would indicate that
either the plume was transported away from the SO sensors or some non-meteor-
ological reason was responsible for curtailing operations. It was decided
that if C(t) < 0.1 ppm and Q (t) < Qf that
R = 1
M = C
These hours represent a third operating mode, the one in which non-meteorologi-
cal factors determine load conditions and should be accounted for.
In summary, three possible cases exist, full-load cases for which no
violations occurred, (case 1) cases which involved an SCS curtailment action
or which required an SCS curtailment action, (case 2) and cases for which
factors other than the SCS determined the operating conditions (case 3).
During the 36 day test period each occurred with the following frequency:
TYPE OF CASE No. OF HOURS PERCENT FREQ.
Case 1 364 42.2
Case 2 301 34.8
Case 3 199 23.0
864 100%
Note that Q/Qf = 1 during the input phase representing the full load con-
dition but is distributed between 0 and 1 in the output reflecting the load
reductions called for by the control strategy. Violations under the SCS
were defined as values of C > 0.5 ppm.
7-37
-------
The distribution of expected concentration, i.e., the results of ERT
PROBL is presented in Table 7-11. Note that the cumulative probability
density is given for concentrations and the probability density function is
given for emissions. For the hypothetical SCS described previously, ERT
PROBL predicts compliance with a y = 0.5 ppm standard, 99.61% of the time,
or violation during 3 or 4 of the 864 possible hours during the test period.
Similarly, compliance with other values of y can be gotten from Table 7-11.
The curious value of 0.49999 in class 20 of the emissions probability
function is coincided. It reflects the fact that ERT PROBL predicts an
emissions of between 95% and 100% of full load during 432 of the 864 hours
during the test period. This includes all 364 hours from Case 1 and, by
chance, 68 hours from Cases 2 and 3.
In light of the assumptions made to perform this calculation, we first
find the actual results of this comparison encouraging. It would be of
great interest to further refine the assumed probability density functions
with a more extensive data set (over say a minimum of a 4 month period) and
to better define an objective control strategy for use in ERT PROBL to repre-
sent present ASARCO operations. Once the present ASARCO operations are properly
simulated, ERT PROBL could be used to assess differences in expected air
quality levels resulting from alternative control strategies or from alterna-
tive air quality forecasting schemes.
7.5 Discussion
Selection for analysis of an existing SCS at the El Paso ASARCO plant
reveals many of the important characteristics that would be desired for
any proposed control system. The real-time operation of a large network
of continuously-monitoring observation stations as well as full-time meteor-
ological forecasting are imperative to any SCS. The assembling and processing
of historical data is necessary as a check on past performance and future
trends. The fact that meteorologists' forecasts are complied with as strictly
as they are gives the ASARCO operation good credibility. At the same time,
however, certain aspects are not consistent with Federal regulations con-
cerning the structure of an SCS strategy. Lack of objective forecast
criteria and well-documented forecasts make it difficult to predict future
performance, especially with regard to the analytical techniques presented
in Sections 3 and 4. These techniques allow extrapolation of performance
7-38
-------
TABLE 7-11
ERT PROBABILITY OUTPUT
CLASS*
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
CUM. PROB.
CONC.
.3208
.5960
.7433
.8363
.9055
.9544
.9713
.9862
.9933
.9961
.9977
.9987
.9993
.9994
.9997
.9998
.9998
.9999
.9999
1.0
PROB. FCN.
EMISSIONS
.00216
.00991
.01748
.02477
.03666
.03798
.03847
.02989
.03272
.04212
.01731
.04213
.01573
.03941
.01640
.01456
.05315
.01561
.01356
.49999
Class Intervals are 0.05 ppm for Concentration and .05 for Emissions,
where Emissions = 1.0 is Full Load.
Concentration Probability is Cumulative.
Emissions Function is Probability Density under SCS Conditions.
7-39
-------
results based on relatively short test periods and allow evaluation of the
impact of major operational changes before they take place. An accurate
emissions monitoring system is also required if any SCS is to be assessed
in light of ERT PROBL predictions. The lack of same in the ASARCO case
introduces severe uncertainty into the model results.
For EPA approval, a test period is necessary for the generation of
system statistics. The length of such as test period should be large
enough to cover more than one season but need not encompass a full year.
A minimum of a 120 day test period is required by EPA. This should take
place during the period of most serious air pollution episodes. A pre-
diction scheme should be prepared before the start of the test period and
be constantly updated to reflect the experience gained with operation. A
mathematical model which relates meteorological conditions and emissions
characteristics to ground-level concentrations is essential in objectifying
such a forecast scheme. The "man-machine" mix has worked well in other areas
of meteorological forecasting and should yield a system which is more reliable
than either of the two components above.
ERT PROBL can be used effectively if the input data is accurate and
consistent. Stratifying cases according to different meteorological regimes
and deriving separate M and R functions for each can be useful in considering
the potential strengths and weaknesses in an SCS. At ASARCO the predominant
meteorological conditions were daytime "coning" with unstable conditions,
nighttime inversion with some interference by topographic features, and
morning fumigation resulting in short-lived but excessive ground-level con-
centrations. The load reduction cases consist primarily of daytime and
morning fumigation conditions with the small subcategory of nighttime "down-
wash" effect cases appended. Improving meteorological forecasting by speci-
fied amounts can be readily interpreted as decreases, if any, in the frequency
of violations. If the desired decrease is not reached, it may be necessary
to increase the mean value of R, i.e., "overpredict" more thereby adding a
safety margin. Knowing the emissions accurately would have been useful at
ASARCO in pinning down M, the meteorological function. Only by developing
statistics accurately and considering the merits of subsets of the total
data base can credible use be made of the analytical tools developed. With
the aid of these tools, however, useful information is obtained on the ex-
pected future performance of an SCS.
7-40
-------
8. ENFORCEMENT OF SUPPLEMENTARY CONTROL SYSTEMS
The basis to the enforcement of supplementary control systems by
state regulatory agencies is implicit in the current EPA policy governing
criteria for acceptable state regulations authorizing the use of such
systems to attain national ambient air quality standards (40 CFR, Part 51,
Appendix; P, Section 3 - Federal Register of September 14, 1973). In effect,
the EPA guidelines require that workable enforcement procedures be esta-
blished by a control agency prior to its authorization of a supplementary
control system for any source within the agency's jurisdiction. Operational
details of enforcement programs will necessarily vary from case to case
according to the nature of specific sources and the constraints imposed
by the availability of funding and manpower resources within individual
control agencies. However, the present EPA posture regarding the use of
supplementary control systems is quite restrictive, limiting the number
of utilities and industries which qualify for this form of emission control
to very specific categories which comprise only a small percentage of
the nation's sulfur dioxide sources. Thus, most aspects of the enforce-
ment strategy outlined by EPA should be subject to fairly similar inter-
pretations throughout the country. In this section the key elements involved
in the adequate enforcement of a supplementary control system by state con-
trol officials are discussed. Special emphasis is placed upon the identi-
fication of problems likely to be encountered by control agencies in their
enforcement roles, with consideration given to means by which the agencies'
tasks of evaluating and monitoring supplementary control systems may be
simplified.
The proposed EPA regulations regarding state permits to authorize the
implementation of supplementary control systems define a rigorous set of
requirements which must be satisfactorily met by any source wishing to
use such techniques of emission control. State agencies must evaluate
and approve in advance both the justification for and the operational
details of these emission control strategies on a source-by-source basis.
Each source must demonstrate to the satisfaction of the appropriate agency
that it has investigated in good faith the availability and reliability
of constant emission reduction methods, and that it will continue to allocate
its resources to accelerate the development of constant emission reduction
technology applicable to the source. If permission to implement a supplementary
8-1
-------
control system is granted, the responsible agency is required to conduct
a formal review and reexamination of the permit at intervals of five
years or less. Such a review procedure must consider in detail all factors
bearing upon the justification for continued operation of a supplementary
control system by each source. The proposed EPA regulations also require
annual review of each operational supplementary control system by state
control agency officials. Progress toward the realization of constant
emission control methods should be considered during the yearly review
process.
Any acceptable state regulation authorizing supplementary control
techniques to attain ambient air quality standards must require each
source to establish and maintain a continuous monitoring program for mea-
suring (or estimating) pollutant emissions, air quality levels, and ap-
propriate meteorological variables. The source must agree to grant the
responsible enforcement agency continuous access to all monitoring data
collected by the source, and the authority to inspect, test, and calibrate
all measurement and recording equipment. In addition, the source must
agree to notify control officials whenever curtailment of emissions is
initiated and/or when air quality standards are exceeded.
Authorization by a state agency allowing a source to implement a
supplementary control system may be granted only upon its approval of a
two-phase plan and schedule, submitted by the source, and describing the
development, operation, and maintenance of the proposed emission control
program. This plan must include a comprehensive report describing a
thorough background study which demonstrates the capability of the supple-
mentary control system to attain all national air quality standards.
In addition, the source must present an operational manual which provides
a detailed description of the day-to-day implementation of the emission
control system, including the criteria which will be used to determine
the necessity for and the extent of emission curtailment under specified
meteorological conditions.
A careful appraisal of these preliminary documents by control agency
officials can greatly simplify the task of enforcing supplementary control
systems and minimize the degree of monitoring and surveillance which will
be required when such programs become operational. It is important that
before permission is granted to operate a supplementary control system, the
responsible agency is satisfied that: (1) the proposed program is capable
1-2
-------
of meeting the objective of maintaining air quality standards; (2) the
source is technically competent to implement the control strategy; and
(3) the source is committed to operate the proposed control system in good
faith and in the manner prescribed in the approved plan and operational
manual.
The proposed EPA regulations specify minimum requirements for the con-
tents of both the background study and the operational manual to be sub-
mitted by each eligible source. Agencies responsible fqr the authorization
of supplementary control systems should devote particular attention to
certain key information which should be contained in these documents to
avoid unnecessary problems of enforcement during operation of such systems.
The background study should enable the agency to adequately evaluate
the capability of the proposed control strategy to attain all air quality
standards. It should contain complete descriptions of:
1. The instrumentation which will be employed for in-stack moni-
toring, air quality and meteorological measurements, as well
as all recording and telemetry systems used to store or
transmit field data to a central facility (if applicable).
Reasons for the choice of all equipment in the proposed
monitoring network should be included. The methods used
in selecting field monitoring sites should be described in
detail, as well as the proposed procedures for maintenance
and calibration of all monitoring equipment.
2, The equipment, personnel and techniques which will be
employed to forecast meteorological conditions, the ex-
pected reliability of all elements of the forecasting
procedure, and the lag time which will be involved in
achieving effective emission reductions in response to
a forecast of adverse dispersion conditions.
3. The methodology used to identify the meteorological condi-
tions which are associated with poor dispersion, and sub-
sequent emission reductions. This should include analyses of
existing air quality and meteorological monitoring data, as
well as a full description of the techniques of diffusion
simulation (modeling) which have been employed to determine
8-3
-------
the conditions associated with high concentrations, and the
degree of control needed to maintain ambient air standards
under each meteorological situation.
4. The method chosen for varying the emission rate, the basis for
choosing this method, and information regarding the time re-
quired to effect sufficient emission reductions to prevent
contravention of ambient air standards.
5. The estimated frequency that emission rate reduction will be
required to attain all air quality standards.
The operational manual should provide the responsible agency with a
complete understanding of the day-to-day implementation of the proposed
supplementary control system by a given source. From the point of view
of enforcement the most important elements of the manual are:
1. A complete description and justification of the number, type,
and location of all field monitors.
2. Identification of the specific meteorological conditions and
monitor readings before and/or during which the emission rate
must be reduced to avoid exceeding short-term standards.
3. A description of techniques and criteria used to anticipate the
onset of weather conditions associated with ground-level con-
centrations in excess of standards.
4. A description of the criteria by which the source determines
the degree of emission control needed for each situation.
5. Identification of the specific actions to be taken for curtailing
emissions when critical meteorological conditions exist or
are forecasted, and/or when specified air quality levels occur.
6. Identification of company personnel responsible for initiating
and supervising such actions.
7. A description of the manner in which monitoring data are to
be transmitted to the control agency.
8-4
-------
8. A description of a program whereby the source systematically
evaluates and improves the reliability of the supplementary
control system.
9. Identification of a responsible and knowledgeable person (and
alternates) onsite, who are authorized to curtail emissions and
who can appraise the control agency on the status of the sup-
plementary control system at any time.
Control agencies should require that the indicated information in the
background study report and the operational manual be presented in detail.
Conservative diffusion modeling techniques and data interpretations should
be employed to ensure that decisions to reduce emissions will be made at
least as often as is required to maintain ambient standards. If a proposed
emission control system is well-defined by these preliminary documents, then
approval granted by a control agency amounts to an affirmation by the granting
agency that the system will meet the objective of attaining ambient air
quality standards, provided it is maintained in good faith according to
the procedures which have been approved. Clearly, this simplifies the task
of routine enforcement to (1) determining whether a source is in fact con-
ducting a supplementary control system according to the approved operational
manual, and (2) determining if an air quality standard has been violated
(as determined through an air quality monitoring network).
For example, a forecast observed concentration in excess of a designated
threshold and not accompanied by the prescribed SCS curtailment action would
constitute a violation of the SCS requirements. The only non-predetermined
parameter in such a scheme would be the weather forecast itself. Once that
had been made, the operation of the air quality model, would follow a well
defined methodology to reach a predicted concentration field which in turn
would automatically trigger the proper SCS control response. The reliability
of the weather forecast, as was previously shown in Section 4, can be objectively
evaluated a posteriori and adjustments can be made in the conservatism of
the control procedure, if necessary-
The degree of surveillance by a regulatory agency considered necessary
to make a determination of compliance may vary greatly from one state to
another. The proposed EPA regulations require the submission by each source
of a monthly report, which includes air quality analysis. The data presented
in these reports should be sufficient to allow the control agency to determine
8-5
-------
whether the supplementary control system is attaining ambient standards.
Spot checks of the operation, maintenance, and calibration of the air quality
and meteorological monitoring sensors should be conducted periodically by
the agency.
The most objective approach to enforcement, therefore, appears to be
one chiefly concerned with the source's compliance with the approved scheme
for emission rate reduction. Instrumentation capable of continuous in-stack
monitoring of pollutant releases is currently available and provides the
most direct information regarding emission rate. With the availability of
reliable stack sampling sensors, an enforcement program based upon emission
rate surveillance and ambient air quality monitoring is possible.
Such procedures are only slightly more complex than would be necessary if
constant emission control systems were required. The ambient air quality
data can be used for periodic reevaluation of the supplementary control
system's capability to achieve all ambient standards. Also, the simultaneous
monitoring of emission rate and ambient concentrations will provide an
extremely valuable dataset for evaluating and refining the diffusion model
techniques used to predict concentrations at points other than the monitoring
sites under all meteorological conditions.
From the above considerations, it would appear that a major emphasis
should be placed upon rigorous evaluation of a system before its approval,
to facilitate surveillance of the operational program. Also, since most
agencies are responsible for overseeing the air quality of an entire state,
it is not likely that the highly disproportionate enforcement procedures will
be applied to the very small number of sources which qualify for the use
of supplementary control systems. The monitoring programs and reporting
procedures required of these sources are far more rigid than those expected
of the great majority of sources within an agency's area of responsibility
and it appears that periodic reports and spot checks of the equipment and
personnel are all that can be expected in a compliance checking program
and, by careful evaluation before the SCS is implemented, all that should
be necessary.
8-6
-------
9. CONCLUSIONS AND RECOMMENDATIONS
This report describes the methodologies and results of a study involv-
ing the development of analytical procedures with which to evaluate the re-
liability of a supplementary control system for an isolated source of SO-
emissions. This section enumerates the most important end results of this
work effort. A number of recommendations are presented for future
work representing logical follow on activities which would further extend
and refine the analytical tools developed.
9.1 Review of Major Accomplishments
As a direct consequence of this study effort, a number of important
advances have been made in the development of procedures to analyze and
assess the reliability of SCS systems. Among the more significant accom-
plishments are the following:
1. An analytical framework has been developed to estimate the
expected frequencies of occurrence of high concentration
levels from meteorological data, air quality forecasting
accuracy data, emission rate accuracy data and as a function
of assumed objective control strategy criteria. The methodology
involves defining the interaction of the above elements in a
probabilistic manner and performing a computer simulation of the
SCS system operation. The analysis scheme is used to identify
the effectiveness of alternative methods of improving overall
system reliability.
2. A specific computer program was developed to assist in the
optimal placement of air quality monitors for an SCS system.
The program requires inputs of the expected frequency of
occurrence of meteorological conditions, and criteria defining
allowable differences between measured peak values and the
maximum expected values in the geographical area of interest.
The program can be used to evaluate the 'cost-benefit' of
information gained for alternative placements of monitors or
for increasing the number of monitors within a system.
9-1
-------
3. A methodology was developed to quantitatively investigate
the relationship between the time scales of emission reductions
associated with alternative control strategies and the likeli-
hood of compliance with various air quality standards. Thus
for example, the effectiveness of a control strategy oriented
toward controlling 24-hour SCL averages in simultaneously con-
trolling the 3-hour SCL averages can be evaluated. Several
example calculations are presented which indicate that reasonable
response times of SCS systems will result in reliable control
of ambient air quality levels.
4. The establishment of specific data elements for a proposed
or operating SCS system are recommended to comprehensively
analyze the reliability of the system. The data sets essen-
tially consist of independent documentation of the meteorological
forecasts, air quality model projections, emission projections
and actual meteorological and emission conditions, control de-
cisions as well as observed air quality levels. These data sets
can be used to develop estimates of the contributions of the various
components to the overall system reliability. Analysis of these
uncertainty contributions will point to methods to most effectively
improve and update system reliability-
5. A case study of the ASARCO SCS system in El Paso has employed
several of the methodologies developed in the course of this
work. Air quality and emission curtailment data over a two
month period has been used to evaluate the effectiveness of
the ASARCO SCS in controlling SO- levels in the region of in-
fluence of the operation. A computerized simulation of the
actual control decision strategy was constructed on the basis
of a hypothesized air quality forecast model and control
strategy derived from manipulations of the available data sets.
The model of the system can be run to calculate the expected
frequency of occurrence of high S0_ concentrations as a function
of various control strategies or forecast system accuracy.
9-2
-------
9.2 Recommendations for Future Work
The work undertaken and accomplished in this study represents an
important step forward in being able to analyze the reliability of SCS
systems. As a consequence of the study several areas for future work
are evident which were beyond the scope of the study assignment or due
to limitations of available data. Therefore it is appropriate to briefly
itemize the areas in which it is felt that additional study efforts would
be highly beneficial to advancing the state of understanding of SCS systems.
The recommendations fall into two categories: (a) the establishment
of an appropriate extensive data set geared to the specific needs of the
analytical framework developed in this studyj and (b) advancement of the
analytical methodologies for more comprehensive treatment of SCS systems.
1. The need for a data set completely documenting the operation
of an SCS system.
We recommend that a real-time monitoring and SCL forecasting system
be operated to generate a data set which includes sufficient records of
source data, meteorological data, and air quality predictions to carefully
apply the analysis tools developed in this project. It is not necessary
that an SCS be operated simultaneous with this test program. The analysis
tools require only knowledge of the proposed SCS design with the actual
source, meteorological, and air quality forecasting statistics. A test
period of approximately 120 days in accordance with the Federal Register
recommendations would be appropriate. The datasets would allow the analysis
tools to be fully tested and would, in addition, provide a base reliability-
analysis case for use to compare with the 120-day test program results of
organizations applying for SCS system approval.
2. The need to include economic-optimization considerations into
the analysis methodologies.
The intent of this study has been to examine the reliability of SCS
with respect to their ability to comply with air quality regulations.
Recognizing that SCS systems will be a substantial economic burden on source
operators, the design of an SCS should include economic cost considerations
simultaneously with its goal of maintaining a clean atmosphere. We recommend
that the economic ramifications for the operator of various SCS alternatives
versus constant emission control and no control be considered within the
general framework of assessing system reliability. The elementary analysis
-------
in Section 5 considered the "fraction of load reduction" and the "fraction
of dirty fuel" achieved for the hypothetical SCS schemes. These numbers
provide one measure of the relative costs of various SCS strategies. Other
factors include the cost of monitoring versus improved reliability, the level
of accuracy of air quality forecasting required versus reliability and the
effect of various time scales of curtailment on operating costs. The optimal
SCS design would meet the air quality reliability requirements at minimum
operating and capital costs. A study of these economic considerations
would enable the control agency to most intelligently assess a proposed SCS
and would enable the source to apply its available resources to improve air
quality in the most efficient manner.
3. Extension of the analytical techniques to two or more sources
of SO,, emissions.
The Federal Register recognizes that an SCS might be designed for more
than one source in a single region of influence. The analysis tools of
this project could be applied to such a multiple source SCS. We recommend
that the problems associated with assessing reliability of multiple source
SCS operations be considered within the framework of the analysis techniques
presented here. Some reliability considerations would, by necessity- be
inter-facility. That is, the reliability of each source would be independent
of other sources in the system. Other reliability considerations would be
intra-facility. That is, the reliability would depend upon the cooperation
of all sources. The overall system reliability would be affected by the
inter-facility and the intra-facility contributions in a complex manner.
The analysis methodology would involve inclusion of the EPA requirements
that an 'a priori' mechanism for assigning responsibility for contravention
of standards be established among the various SO sources. An objective
method of assessing the overall system reliability for multiple source SCS
operations should be of great value to control agencies in their role of
approving applications for such systems.
9-4
-------
REFERENCES
American Society of Mechanical Engineers, RECOMMENDED GUIDE FOR THE PREDICTION
OF THE DISPERSION OF AIRBORNE EFFLUENTS, M. Smith, Editor New York,
1968.
Anderson, G.E., "Mesoscale Influences on Wind Fields", JOURNAL OF APPLIED
METEOROLOGY, Vol. 10 (1971), pp. 377-381.
Atomic Energy Commission, METEOROLOGY AND ATOMIC ENERGY, 1968, David Slade,
Editor, 1968.
Briggs, G.A., PLUME RISE, Critical Review Series (TID-25075), Atomic Energy
Commission, Division of Technical Information, Oak Ridge, Tennessee,
November 1969.
Csanady, G.T., "Effect of Plume Rise on Ground Level Pollution", ATMOSPHERIC
ENVIRONMENT, Vol. 7 (1973), pp. 1-16.
Egan, B.A., and Mahoney, J.R., "Numerical Modeling of Advection and
Diffusion of Urban Area Source Pollutants", JOURNAL OF APPLIED
METEOROLOGY, 11, (1972), pp. 312-322.
Federal Register, Vol. 37, July 27, 1972, p. 15095.
Federal Register, Vol. 38, March 23, 1973, p. 7555.
Federal Register, Vol. 38, September 14, 1973, pp. 25698-25703.
Forrest, J., and Newman, L., "Ambient Air Monitoring for Sulfur Compounds",
JOURNAL OF AIR POLLUTION CONTROL ASSOCIATION, Vol. 23, No. 9 (1973),
pp. 761-768.
Montgomery, T.L., et. al., "Controlling Ambient SO ", JOURNAL OF METALS,
June 1973, pp. 35-41.
Peters, M., REPORT OF INVESTIGATION AT AMERICAN SMELTING AND REFINING
COMPANY, EL PASO, TEXAS, Air Pollution Control Services, 1971.
Shepard, D.S., "A Load Shifting Model for Air Pollution Control in the
Electric Power Industry", JOURNAL OF AIR POLLUTION CONTROL ASSOCIATION,
Vol. 20, No. 11 (1970), pp. 756-761.
Shir, C.C., and Shieh, C.J., "A Generalized Urban Air Pollution Model and
its Application to the Study of S02 Distributions in the St. Louis
Metropolitan Area", ENVIRONMENTAL SCIENCES, May, 1973.
TVA Press Release, 24 June 1973.
R-l
-------
APPENDIX A
THE ERT POINT SOURCE DIFFUSION MODEL
Throughout this study it has been necessary to generate concentration
fields for the various analyses. For consistency, an ERT gaussian plume
model has been used in each case. The following model description will be
useful in interpreting the model results as they are presented in our report.
The ERT PSDM (Point Source Diffusion Model program was developed
specifically for analysis of elevated and bouyant stack plumes under various
general meteorological conditions and for specific combinations of stability
regimes associated with the presence of upper level inversions. The model
utilizes gaussian plume spread statistics and incorporates modifications to
include the primary effect of topography on concentration values.
The PSDM model is specifically designed to provide estimates of the
downwind profile of short-term and annual concentrations resulting from the
emissions of elevated point sources. Ground-level concentrations along the
direction of the plume centerlines are computed for each of the 768 weather
conditions indicated in Table A-l, and the results define the weather condi-
tions which give rise to highest effluent levels at the ground.
PSDM predicts downwind concentrations for the normal spectrum of stabili-
ties (very unstable to stable) by the well-known gaussian diffusion formula
for an elevated continuously emitting point source as presented in the
Workbook of Atmospheric Diffusion Estimates (Turner, 1969).
In addition, two special conditions are also examined. The first con-
siders the result of the effluent plume being trapped below an elevated in-
version lid. The conditions beneath the elevated inversion can vary from
unstable to neutral. Lidded conditions are simulated by the trapping model
presented in the Workbook. The second special case accounts for fumigation
conditions during the breakup of the nocturnal inversion when it is dissi-
pated by the upward transfer of sensible heat from the ground. This situation
usually occurs during the morning after sunrise when the ground is being
warmed by solar radiation. Pollutants previously emitted into the stable
layer from elevated sources are rapidly mixed when they are reached by the
thermal eddies, and ground-level concentrations usually increase sharply
for a period of time (on the order of 15 minutes). After the height of the
mixed layer surpasses the plume, ground-level concentrations occur that are
A-l
-------
8605
TABLE A-l
SPECIFICATION OF ATMOSPHERIC STABILITY AND WIND SPEEDS
Stability Class
(Weather Conditions)
1
2
3
4
5
6
7
8
Description
Very Unstable
Unstable
Neutral
Stable
Inversion Lid - Very Unstable Below
Inversion Lid - Unstable Below
Inversion Lid - Neutral Below
Fumigation
Wind Speed Class
1
(mph)
2
2
4
2
2
2
2
2
2
(mph)
3
4
8
4
3
4
4
4
3
(mph)
4
6
12
6
4
6
6
6
4
(mph)
5
8
20
8
5
8
8
8
5
(mph)
6
10
27
10
6
10
10
10
6
(mph)
7
12
35
12
7
12
12
12
-------
more typical of an inversion lid condition. As this mixing process proceeds,
the inversion is usually completely destroyed so that typical daytime condi-
tions prevail.
Early morning fumigation is examined by a model developed at ERT.
Their experience has shown that previously used models over-estimate ambient
concentrations, particularly when they are assumed to be representative of
hourly integrated concentrations. The new model realistically predicts one-
hour integrated ambient concentrations during this transient phenomenon by
simulating the variety of conditions that occur during the one-hour of maxi-
mum concentration. The maximum ground-level concentration caused by the
mixing layer passing through the plume, as suggested by Turner, is used to
produce the average concentrations for the first 15 minutes of the period.
The lid is, therefore, set at the height of the plume during this quarter
hour. During the next 15 minutes the trapping model is used with the inver-
sion lid at twice the original plume height (i.e., twice the height assumed
by the plume during the stable conditions prior to the start of fumigation),
and the final 30 minutes is simulated by the trapping model with the inver-
sion lid at 1000 feet. If the plume height during stable conditions is
greater than 500 feet, during the final 30 minutes the model will place
the inversion lid at twice this height rather than at 1000 feet.
The horizontal and vertical plume expansion relationships as presented
in the ASME guide for one-hour integration periods are incorporated in
the ERT model for all conditions except the fumigation case. The Pasquill-
Gifford plume expansions as presented in Slade (1968) were used for this
special case because they are applicable to the shorter time periods that
are examined in the submodel.
All estimates of plume-rise follow the^recommendations of Briggs (1969).
In the lidded conditions (stabilities 5, 6 and 7), the height of the lid
is set at the height of the plume to estimate conservative ground-level
concentrations. If the lid were lower, the plume would punch through the
bottom of the inversion into the .stable layer and disperse only gradually
to the ground. If the lid were higher, the mixing layer would be deeper
and, hence, the volume of air into which the plume diffused would be greater.
In either case, the concentration at the ground would be lessened.
Consequently, the concentrations calculated for lidded conditions represent
the worst potential hourly values.
A-3
-------
The modification of the plume height induced by rising terrain is
treated by orienting the centerline so as to be intermediate between a path
parallel to the surface and one which remains at the effective stack height.
An additional restriction, which applies in the case of very sharp relief,
forces the centerline to remain at least half the distance above the surface
than it would maintain in the absence of intruding topography. Figure A-l
demonstrates the PSDM treatment of terrain-induced plume height modification.
This technique is compatible with a conservative interpretation of wind
tunnel experiment results, and with the basic principle of continuity of
mass flow. An additional element of conservatism in the calculations cor-
responding to elevated inversion cases stems from the model's placement
of the "lid" at precisely the plume centerline height, thus, assuring that
the resulting ground-level concentrations are maximum possible values.
That is particularly true over irregular terrain, since the simulation does
not consider the enhanced mixing of the effluent plume due to the mechanically-
induced eddies in the layer beneath the inversion.
A-4
-------
H0+h(x)-
h is physical stack height
Ah is plume rise
HO is effective stack height (physical stack height plus plume rise)
h(x) is terrain height above stack base elevation
h is plume centerline height above stack base elevation
Figure A-l Treatment of Terrain Influence Upon Plume Height Used in Model PSDM
-------
APPENDIX B
EFFECTS OF CONTROL STRATEGIES ON GROUND LEVEL
CONCENTRATIONS FOR DIFFERENT AVERAGING TIMES
-------
TABLE B-l
Effects of Control Strategies on 3-Hourly Average Concentrations.
Cumulative Frequency Distributions for 3-Hourly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
1.155
1,127
1,090
' 1.049
1.S40
1,111
0.962
0.951
0,974
0.845
0,167
O.H6
o,ao9
0,700
0,741
0,722
0.693
0,664
0.416
0.607
0.57B
0.544
0.520
0.441
0.462
4.431
t.40*
.176
0,3>7
e.ne
6,284
0,2*0
0,211
0,202
O.IM
0,144
l,lll>
0,097
0.05B
0.024
Percentage of
Values Less
Than Or Equal
to x
100,000
99.281
99.781
99.281
it. in
99,2ft
99,211
99. 2 H
47.568
97.568
97,560
47.566
97.568
97.5611
97.561
97,566
97.J97
97.3*7
97,16}
96.199
96,199
96.199
96.1-30
94.B97
96,4ia
91.016
92,671
92,671
6.815
81. 601
81.164
77,021
72,779
67, nil
65.022
61.097
66.726
14.861
11,161
14.69}
Cs = 0.5
Upper Limit x
of Concentra-
tion Category
(ppm)
0,500
0,487
0,475
0,462
0,450
0,*17
0.12S
4,a|-l
0,400
B.1B7
0,174
0,161
O.JSO
0.118
0,125
0,111
o.lon
0,28fl
0,275
0,261
0.250
0.21B
0,225
0.211
0.200
0,108
0,175
0.161
0,150
0,11*
0,125
0,111
0.100
0,6R6
0.075
0,061
0,050
0,010
0,025
0,011
Percentage of
Values Less
Than or Equal
to x
1PO,-000
90,09-7
94.097
94,410
91.B16
91. M6
92.671
92,671
2,671
92.671
92,671
92,276
06,815
06,815
86.164
81, 801
61.1(4
BI.I64
01. I.M
77.071
77.021
72.774
70.479
68.596
67.514
67.B66
65,622
64,097
64.197
64.726
64.726
64.726
41.511
34,861
14.661
16,1163
14,692
14.692
31.301
31, 3"!
Cs = 0.42
Upper Limit x
of Concentra-
tion Category
Cppm)
0.420
0.409
0.394
0.300
0.178
0,367
0,357
0.316
0,1)6
0,»5
C.115
0,306
9,29tt
0,204
0/271
0.262
0,252
0,242
0,211
0,220
0.210
0.200
0.189
0,171
0,166
0,150
0,147
0,1)7
0,126
0.116
0.105
0.095
0,064
0.074
0,363
0,051
0,94!
0,017
0,021
0,011
Percentage of
Values Less
Than Or Equal
to x
100,000
92.671
92.671
92.671
92.671
92.226
9] ,hlO
»6.«I5
86,81 S
86,160
M,ft01
81,507
81,164
81,164
81..1M
77,021
77,021
72.774
72,774
70.474
68.546
67.534
67.534
65,622
65,622
64.B97
66,697
64,776
64,726
.64.776
64,110
40, 518
I4.461
34.863
3«,«M
14,863
14,64!
14.692
11,301
11.101
Cs = 0.34
Upper Limit x
of Concentra-
tion Category
(ppm)
0.340
0.331
0.123
0,314
0,106
0,297
0,284
0,200
0,272
0,261
»,255
C.taJ
0,238
0.229
0,221
0,211
0,204
0.146
0,187
0.179
0.170
0,162
0.151
0.145
0,136
0.121
0,119
0,111
0,102
0,094
0.3B5
0.077
0.068
0,060
0,051
0,043
4.014
0,026
0,017
0,009
Percentage of
Values Less
Than Or Equal
to x
100,000
86,015
81.801
F1.80I
B1.801
81.164
81.164
81.164
81.164
77.091
77.021
71.286
72.774
72.774
70.479
6B.596
67.603
37.53C
67.466
65.812
65.622
64.897
46.647
66.647
64.726
64,726
64.776
66.726
41.541
S5.I03
34.863
14.861
34.863
14.461
34.692
14.692
14.692
11.101
11.101
ll.ltl
Design Standard Averaging Time, Ts = 3 HRS. Design Standard, Cs, is variable.
-------
TABLE B-2
Effects of Control Strategies on 3-Hourly Average Concentrations,
Cumulative Frequency Distributions for 9-Hourly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
O.SOft
0.19)
o.««o
o.a»R
O.aSS
O,fln2
0,«iO
O.M7
0,aoS
O.J92
0.379
O.J67
0.55«
O.Xfll
o.iai
0.1U
O.iOS
0.291
0.278
0.26S
C.25S
0,280
0,22!
0.2IS
0,202
0.190
0,177
0.1H
O.IS2
1,111
O.l?t
0,ll«
0.101
t.S'S
0,07k
S.Oll
O.OM
0.011
0.02S
1.011
Percentage of
Values Less
Than Or Equal
to x
100.000
99.93)
99.S52
9*. 309
99.280
99.260
99,07*
98.3SS
48.252
48.252
96.613
9i,B|3
46.7flo
96.710
*6,710
96.436
92.221
51.947
M.B7B
91.678
91, '"1
<1. SOI
96,660
89,690
87,206
at. 395
86.052
61.06?
61.900
eo.150
80,156
66.951
bO,A9l)
S?.7fl?
47. 430
3«.270
)0.'«Q
?9.8a9
1Z.O"
llfd^T
Cs = 0.5
Upper Limit x
of Concentra-
tion Category
Cppm)
0.1)7
P. 126
0. )?0
O.M1
0,391
P.?»5
D.?3*.
0,?T8
0,?b9
35.ftH6
31.3S7
J0.7«0
30,2?6
1S,«90
l?,097
11. "ST
4.670
Cs = 0.42
Upper Limit x
of Concentra-
tion Category
Cppm)
0,317
0,309
0,301
0.293
0.2HS
0,277
0.269
0,20?
0,25u
0,2U6
0,2*fl
0,230
0,??2
0.21Q
0,206
0,196
0.190
A. 10?
0,17(1
0,i<>6
0,159
0,151
O.IQ3
0.13S
0,127
0.119
0.111
9*103
0.045
U,OB7
0.079
0,071
O.Ot.3
0.1^5
0.000
0,000
0.012
0.02H
0.016
0.006
Percentage of
Values Less
Than Or Equal
to x
100.000
99,96b
99.661
99,79«
99,790
99,790
99,760
«9,6?3
99,«fl
«6,801
47,900
97,772
97.130
96,436
96,162
95.9?2
9S.166
48,722
4fl.sl*
43.86b
93.d?0
92,«95
92,255
BO. ,56
60.156
78,507
60.417
61.565
59,801
S?.7«2
7,430
3.215
34.270
31.357
S0.?9*
29.HI9
15,216
12,097
11.657
*.a7o
Cs = 0.34
Upper Limit x
of Concentra-
tion Category
Cppm)
0.106
0,300
0,24)
0.2BS
0,277
0.270
0,2b2
0.2SU
0.206
0.239
0.231
0.22)
0.216
0,206
0.200
0.193
0.1RS
0.177
0,t69
d,U2
O.lSa
O.lsfc
0,139
0.131
0.121
0,116
o.tga
0.100
0,092
0.065
0,077
0,0b9
0,062
o.fliy
0,oa6
a, 039
o.aii
0,023
0.01S
O.OOf)
Percentage of
Values Less
Than Or Equal
to X
IPO, oeo
99.9bb
49,9b6
99,9t>b
99,047
99,663
99.6^9
49,790
99,7<)q
99.7911
99.79Q
99.S50
98.9)8
97.601
47,513
96.710
4S.7R5
45,511 .
9fl.i8^i
9a.20B
43.557
92.700
*2.«?6
92.320
42.1S2
1.021
68.500
tO. 521
S4.iea
%2,570
«7,030
S7.971
33.122
X0,7«0
30,?9S
Z9.8Q9
12.132
12.097
I1.^«.T
9.1TO
Design Standard Averaging Time, Ts = 3 HRS. Design Standard, Cs, is variable.
-------
TABLE B-3
Effects of Control Strategies on 3-Hourly Average Concentrations.
Cumulative Frequency Distributions for 24-Hourly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
0,212
0.207
0.202
0,107
0,191
o, no
0.101
0.175
0.170
0,1 6S
0,lb9
0.154
0,189
O.l>]
O.I3B
0,111
0,127
0,122
0.117
0,112
0,106
0,101
0,094
0.010
O.OB5
o,o«o
0,07*
4.0*9
0.064
O.OW
0,053
0,OaS
0,042
0,037
0.01?
0,027
0,021
0.01.
0,011
0,001
Percentage of
Values Less
Than Or Equal
to x
100.000
?9.B«7
«9,ft«T
9*. 568
»9.i5tt
99.a85
f7,872
«7,?sa
97,013
«6,87fc
9S.709
43.237
93,?03
«,?6S
90.113
e».*9B
89,«6I
89.221
68,191
77.8SB
77.206
77,103
74.75*
75.263
72.880
70.11 I
4«.«0|
41.277
SB.»62
M.69S
B.|?«
».0*l
26, 136
1%.*«2
7.7?«
3.9H?
O.ftIT
0.2*0
0.0
0.0
Cs = 0.5
Upper Limit x
of Concentra-
tion Category
Cppm)
0,lb2
e.t^e
o.na
0.1SO
O.lftfr
0.1&2
0,118
O..UB
0,1 JO
D.125
0,121
0.117
0.113
0.104
0.10S
0 ,^1 0 1
0.097
0.093
0,089
0.00S
c.oei
0.077
0*073
0106Q
0,06$
0,061
O.OST
O.OSI
O.OQ9
0,105
0.040
O.Olb
0.032
d.024
0.0?Q
0.020
0.014
0.01?
0,004
0,000
Percentage of
Values Less
Than or Equal
to x
100,000
09.1(,ti
09,9hfc
«,«J1
99.A97
99.Kh3
9,063
99,B?6
90.4S7
99.0lb
99,059
9B.S9!
9B.001
«7.?«e
Ob.t?t
93.S4b
91.2110
90.5AO
B9.9U?
7.741
B5.239
3,110
79.5«t
77.309
7i.2«9
S3, 759
tO.<29
B.129
;.»a
l.ibh
JS.t7»
11. IH
7,7211
i.tfa
1.030
0.4H7
0,200
0,0
0,0
».o
Cs = 0.42
Upper Limit x
of Concentra-
tion Category
Cppm)
0,151
0,117
0,193
0,139
fl,1.3h
0,1 12
O.IM
0,129
0.12.1
0.117
0,113
A, 109
0,106
0,102
0,098
0,091
0.090
0,OB7
O.OB3
0.079
0,075
0,072
0,060
0,060
0,060
0.057
0.053
O.HQ9
O.OQ5
0,041
0,010
0,014
0,010
0,02k
0.02)
0,019
0,015
0,011
0,000
0.004
Percentage of
Values Less
Than Or. Equal
to x
too, o"n
99.966
99.966
99,96.6
99.197
99.461
99.B43
99.87B
97,657
99,550
«,798
98.310
98.181
97,637
97.018
9*, 395
93.821
91.103
BB,9flO
Si. 239.
11.261
80.810
78.613
7«,BD3
71.988
67.937
66. 561
<87
O.ZIO
0,0
0.0
0.0
Cs = 0.34
Upper Limit x
of Concentra-
tion Category
Cppm)
0.1U
o.no
0,U7
0,123
4,120
0,117
0,113
0.110
0,107
0.103
0,100
0.09T
0,093
0,0<90
D,0fl7
0,083
0,080
0.077
0,073
0,070
0,067
0,0b3
Q/ObO
0.057
O.OS1
i),OSO
0,0(17
O.OBJ .
0,000
0,017
0.033
0,030
0.027
0.023
0,020
0.017
0.013
0.010
0,057
0,093
Percentage of
Values Less
Than Or Equal
to x
100,000
99. 9*6
09.966
09,931
99.B63
99.691
99.657
99.657
99, ill
99,073
9B^«
41, ll«
9B.049
47,9110
97.490
9V.263
40,404
114,90?
*7. 470
B4.140
B1.B40
77.2K6
74.52B
71.132
6B.I14
65,534
61.277
57. BIO
12,176
15. 2M
11,672
i.451
I.9B2
0.75S
0,6*7
0.240
0,240
0.0
0.0
0,0
Design Standard Averaging Time, Ts = 3 HRS. Design Standard, Cs, is variable.
-------
TABLE B-4
Effects of Control Strategies on 6-Hourly Average Concentrations.
Cumulative Frequency Distributions for 3-Hourly Running Means
ro
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
1.15S
1.127
1.048
1,069
1.010
1*011
0.982
0,955
0,924
o.69s
0.667
0,618
0.809
0,760
0.751
0,722
0,b13
0,664
8,616
0,607
0,576
0.549
»,S?0
*,«9i
0.462
0.13)
0,404
0,376
0.347
0.31B
0.299
0,260
t.Zll
0.202
0.171
d.144
0.116
0,0*7
0.05S
9,029
Percentage of
Values Less
Than Or Equal
to x
100,000
99,20]
99.z*i
99,261
«9. 2M
99.261
99.2BI
99. 2M
47,568
97.568
17.568
97.568
97.568
97.568
97,568
97.569
97.397
97.397
97.363
96.199
96.199
96.199
96.130
94,697
94.416
43.634
92.671
92.671
06. MS
63.801
il,if>a
77,021
7J.77«
67,603
6^,822
**.S97
64.776
34.A63
S«.«h3
la.6«7
Cs = 0,24
Upper Limit x
of Concentra-
tion Category
Cppm)
0.4AO
0.169
0,if^6
dtn«a
0.432
O.U20
o.aoe
0,S«b
O.JSi
0,172
0,360
0,341
0,436
o.izfl
0.31?
0,300
0.28B
0.276
:.?to
0.252
0,?KO
e.22(t
0,216
0.204
0.112
0.1BO
0,168
.156
0,1m
0.132
0.120
4,108
[1,096.
0,080
0.072
O.QbO
O.OHH
0.016
0.02*1
0,01?
Percentage of
Values Less
Than Or Equal
to x
100.000
99.212
99.178
9.iqa
99,007
99.007
98.971
98.971
98.971
98.97J
96,?0«
94,62)
9a.5H9
94.5A9
92.466
90.205
89.B61
A9.863
16.199
8S.959
0.719
80.17!
T«.?«7
73.630
71.527
75.527
71,«0«
69, 760
69,726
49.6S8
69,62i
69.5B9
44,726
39.110
39.041
39,041
39.007
39.907
3V. 415
35.342
Cs = 0.2
Upper Limit x
of Concentra-
tion Category
(ppm)
0,000
0.390
0,380
0.370
0.360
O.lbO
O.Jao
i).330
0,324
0.310
0.^00
0,290
0.280
0.270
d,260
0,250
0.2UO
d.230
0,220
0,210
0,200
0,190
o.ieo
0.170
0.160
0,150
e.iao
4,130
0.120
0.110
4.100
0,090
0.060
0.070
0.060
0.050
0.4(10
0,030
a. o?o
0.010
Percentage of
Values Less
Than Or Equal
to X
100.000
99.007
99.007
90.93ft
90.938
90.795
90.760
90.760
92,637
92.637
90.qaS
90.ia5
90,911
90.1T7
66.9IB
Bfr.BBI
l.*52
61. Ml
79.281
76.507
7*.1«2
70,115
T3.7«7
71.05S
»*.663
69.621
»9,7AO
69.6*2
69.623
69.621
S.78B
39.3B4
S9.U«
39.110
19.001
19. m\
19.007
39.007
3S.a«3
35.342
Design Standard Averaging Time, Ts = 6 HRS. Design Standard, Cs, is variable.
-------
TABLE B-5
Effects of Control Strategies on 6-Hourly Average Concentrations.
Cumulative Frequency Distributions for 9-Hourly Running Means
00
i
in
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
0,506
0.143
O.oBO
«.«»»
0,195
0..142
0,UO
0,417
«.40S
0,342
0.3J4
0,367
0.151
0.341
o,3i0
0.116
0,103
0,291
0.278
0,265
1.253
0.240
0,228
0,215
o,?o?
0,1 to
0,177
0.166
0.152
.lit
0,li>6
O.llu
0,101
0,fl»S
0,07ft
0,0k3
0.051
0,059
0,025
0,015
Percentage of
Values Less
Than Or Equal
to x
100,00ft
99.9J1
99,052
»9.3a9
*9,?BO
«9(?&B
V.07S
9B.3SS
'B.?S?
VB.2S?
U^Bll
*t.B13
V6.7fl«
96.710
tft.TIO
*k,«lb
92.221
^I.WT "
01,870
91.87*
91,701
» 1.501
90.*Q«
8 9. MO
7.2B6
86. S^
-At. 0*2
85. 9 6?
«l.9flO
80.198
BO.ISB
6fl,9Sl
ftO.»99
S?.7«?
T.010
14,270
S0.7BO
29.814
12.097
ll.BST
Cs = 0,24
Upper Limit x
of Concentra-
tion Category
(ppm)
O.Z44
0,M>
0,232
O.P26
0,220
>,<!«
0,207
0,201
0.115
0,119
O.IM
0,177
0,171
«,!(>?
0.1S«
0,153
0,116
0.1BO
O,l3a
«,-l?A
0,122
O.IK,
0,110
0.10A
B,0«9
o.os?
0,105
0,07<
0,073
0,067
0.061
O.OS5
0,009
0,0(13
0,017
0,031
0,n?Q
o,ni«
0,012
0.006
Percentage of
Values Less
Than Or Equal
to x
100,000
<,<6t
IV.Ibb
*«,ft«7
-«9.,Bb3
11. en
*«.B29
H.7«>
«.7««
«9,7<)«
99,744
9»,79«
*9,760
«,726
»7,567
6,»IZ
46,059
45,490
»5,»19
45,716
45,271
11.7.22
BO, 706
T1.40S
'70,459
64,431
62,577
54,632
51.269
61.432
39,376
36,532
35.161
3d. 999
34.6M
15.294
I5.<«g
!S,2SO
15,113
12,151
Cs = 0.2
Upper Limit x
of Concentra-
tion Category
(ppm)
0,20*
0,204
0,149
0,194
O.lflA
11,1*3
O.,176
0,171
O.J6ft
0,162
0.157
0,192
0,147
0.141
0,136
0,13)
0,126
O.UO
0.115
0,110
0,1 05
0,149
0,044
0,099
0,091
0,074
0,073
0,068
0,063
0,058
0,052
0,007
0.042
0,037
0.031
0,026
0,021
0,016
0,010
0,005
Percentage of
Values Less
Than Or Equal
to x
100,000
44,431
49.947
44,79«
44.194
49,744
94.794
99,744
99.769
99.657
99,657
*4,58*
49.086
M.452
49,280
95,851
45.751
45.065
44.4K3
80,804
76.731
71.110
TO, 082
48.963
62.886
55.826
52, 60S
41.925
10.267
St. 600
35,248
35,127
34,940
34,681
18.437
15.284
15.284
15.113
15.115
12. «M
Design Standard Averaging Time, Ts = 6 HRS. Design Standard, Cs, is variable.
-------
TABLE B-6
Effects of Control Strategies on 6 hourly Average Concentrations.
Cumulative Frequency Distributions for 24-Hourly Running Means
w
cr\
NO SCS
Upper Limit x
of Concentra~
tion Category
(ppm)
0.712
0.207
0.202
0.1*7
0.191
0.1 St.
0,161
0.175
0.170
0.165
-0,159
0.154
0,199
O.iaj
0.138
o,MJ
a. 127
0,122
0,117
9,112
0,106
0,101
e,o»6
,090
O.Oi*
e.o»o
0.07'
O.C69
O.Oka
.ova
tf.OSJ
0,00ft
0,0*2
0.037
0,012
0.027
I. 021
0.016
0,011
0,00%
Percentage of
Values Less
Than Or Equal
to x
IPO. 000
99.897
99.R97
99, sea
99,554
99.S8S
9T.8T2
97,?1.*!
97.011
96.676
9S.709
9J.237
9J.?fi3
90,2(15
90.113
*9,59s
89,461
89.221
B8.19I
77.BS8
77,206
77,103
7k,75«
75.283
72. MO
70.511
M.401
61,277
58,462
S2.695
B.l?f
«.0i«
2 (..Mb
15.«1?
*.7?«
».*A?
O.A17
O.?«0
0,0
*.B
Cs = 0,24
Upper Limit x
of Concentra-
tion Category
Cppm)
1.105
0,102
0,100
0.097
0.095
0.092
0.0>«
O.OBT
o.osa
0.061
,079
0,07b
0.071
0,071
0.068
0.0»t
O.Ob)
0,060
o.ose
0,0*5
0.0b3
0,050
0.047
0,0a%
0,012
0,019
0,017
0,0}«
0,032
0,029
0,02fc
«,02>
0,021
O.OIt
0,01k
0,011
0,011
0,008
0.005
o.oos
Percentage of
Values Less
Than Or Equal
to x
100.000
99.161
99.96*
99.S6J
99.661
99.691
99.6J2
99.519
96.611
96.796
48.695
97.666
97.219
9*.70a
9A.292
95,606
tl.651
92.650
».llt
eo.jn
79.B71
77,769
75.160
TO, 921
"J.HJ
14.148
?S.?66
22.071
15,105
6,657
1.191
1.37!
1.167
1,115
1.111
0,927
0.0
0.0
0.0
0.0
Cs = 0.2
Upper Limit x
of Concentra-
tion Category
Cppm)
0.09B
0,A9b
0.0«S
0,091
0,069
0,086
0,083
4.081
0,076
0,076
0,074
0,071
0.069
0.066
o.Otn
0,061
0,059
0,056
0,05*
0.0V2
0,^09
0.687
0.004
0.042
0,939
0,0)7
0.0 Jd
0,032
0,029
0,027
0.025
0,022
0.020
0.017
o.ois
0.012
0,010
0,007
0.001
0.002
Percentage of
Values Less
Than Or Equal
to x
100.000
V9.M7
99.897
99.897
99,n;a
99.657
99,657
99,550
99,2)0
98, 5? «
97,738
96, (in;
96.*)]
96,12.
9S.469
94,061
93.718
92.036
90,971
9,124
77.034
T4.7C6
T0.95J
3. FBI
IS. 221
25,403
22. H?
I5.»?t
9,)*fc
4.614
1.613
1.167
1.111
1.113
0.996
0.216
0.0
0.0
0.0
0.0
Design Standard Averaging Time, Ts = 6 HRS. Design Standard, Cs, is variable.
-------
TABLE B-7
Effects of Control Strategies on 9-Hourly Average Concentrations.
Cumulative Frequency Distributions for 3-Hourly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
1. 1»
1.12!
1.015
1,169
1.010
IfOll
0.902
e.953
o.02«
0,095
o,»67
0,6»
0.809
0.760
0,751
0.72Z
0.693
0.660
0.616
0.60T
0,576
0.5»9
0.520
0,«1
0.962
0,«33
e,«oii
0.376
0.317
O.SU
0.219
0,280
0.211
0.202
0,171
O.IK
0,116
O.OA7
0,05«
0.020
Percentage of
Values Less
Than Or Equal
to x
100.000
99,2(1
99,2«1
99,281
99,J*|
99.2«l
«9,J«I
'V9.2M
7.5H
47.568
<7.96I
»7.5te
*7.$6B
*7.S6>
07.969
'7.56»
»7.3«7
*].**1
>7,.16J
96,199
96,199 :
96.199
96.130
91,097
9l,«ia
93.036
92.671
92.671
66. MS
03.001
Bl,16«
77,021
72.7711
67,603
6^.022
61.997
6«,726
39.K63
3B.063
39.692
Cs = 0.2
Upper Limit x
of Concentra-
tion Category
(ppm)
0.600
o.'.es
O.S70
0.5«
O.biO
0,525
0,510
O..H9S
0,800
0,<6i
O.ISO
O.I3S
0,020
0..05
0,390
0..375
0,360
0,305
0,330
0,315
0.3CO
0.2B5
0,270
0.255
0.200
e.ii-,
0.210
C.I9S
0,l>0
0.165
0,150
0,135
0,120
0,105
0,0^0
0.075
0,060
0,0«5
0,010
0.015
Percentage of
Values Less
Than Or Equal
to x
100.000
'T.9I1
»T,B7T
1T,«02
t7.T7«
17.7110
17.568
17aSflO
*b,taa
16,471
*fc.fl46
1S.1»3
15,«*B
K.8&S
la.iaa
*J,**.i
iJ.li*
Bft.tBS
er.*7i
«i,5fl«
B?.BOS
u.esa
?. ?is
7B.IS5
Tfl.oai
11.678
69.075
47.77*
67.671
fcs.est
65.000
6«.8?9
6l.7?b
4a.l7«
}5.a79
S&.103
Si. 966
lfl,7?6
34.69?
51.311
Cs = 0.14
Upper Limit x
of Concentra-
tion Category
(ppm)
D.420
0,010
0,599
0,188
0." P
0.368
0.1S7
0,107
ti,M6
Ojl2b
0.315
0.50S
0,Z9fl
0,t*fla
4.27S
0,263
0.?^>2
d.242
0,231
0,22i
0.210
0,200
0,184
0.179
O.U6
0,158
8,107
0,117
0,126
0.116
0,105
0.095
0,08A
0,071
0,063
e.o*s
0,0fl2
C.052
0,021
0.011
Percentage of
Values Less
Than Or Equal
to x
100.000
17.317
46.791
16,507
96.Q3B
«6.336 '
1S,«l5
»1.507
11.370
11.267
B9.?i2
K7.363
67.055
8fc,*78
B6.5TO
B2.671
81.952
77.979
77,363
T«,7?6
7?. 22*
70.137
69.178
67.055
66.712
64.753
65,377
65.068
64.M7
64.726
64.247
1.507
S5.B90
15.5A2
IS. 240
S5.I3T
14.79S
14.726
SI. 501
11*301
Design Standard Averaging Time, Ts = 9 HRS. Design Standard, Cs, is variable.
-------
TABLE B-8
Effects of Control Strategies on 9-Hourly Average Concentrations.
Cumulative Frequency Distributions for 9-Hourly Running Means
oo
00
NO SCS
Upper Limit x
of Concentra-
tion Category
Cppm)
0,506
0,091
0.480
O.flBB
i), 455
(),B92
O.H30
0.1.17
0,405
0,112
0.379
0,367
0.354
0.341
0.]2<)
0.1I6;
0,103
0.291
0.276
0.265
0.253
0,240
0.22B
0,215
0.202
o.i9o
0,177
0.164
o.isz
0,139
O.IJ6
0.114
0,101
0,0*0
0,076
0,0k]
0.051
0,03!
1,02%
0.013
Percentage of
Values Less
Than Or Equal
to x
100,000
99.911
99,452
99.149
99.280
99.280
99.07S
9B.3SS
98.25?
96.252
96.811
96.613
96.74H
96.710
96,710
«6,B36
92,221
»I,9«I
9|.«7»
91.J78
91.7S1
91. SOI
90.614
09.B90
7.286
06.39S
B6.0S2
.'3.962
M.9eo
ao.iss
80.118
68.9S1
60.V94
57, 7a?
7.91D
34.270
30,7«0
29.8*9
12.047
11.0^7
Cs = 0.2
Upper Limit x
of Concentra-
tion Category
Cppm)
0,200
o*m
0.190
0,105
0.1BO
O.IT^
0,170
O.US
0,160
0,1 SS
0,150
0.1*5
o.uo
o,us
0,130
0.125
0.120
e.ns
0,110
0(10S
0*100
0.095
a. o»o
0,005
0,060
0,075
0,070
d,0*.5
d.060
0.055
0,0*0
0,005
0,040
0.015
0,050
O.OZ5
6
6S.S04- .
64.681
B3.A53
B3.4B2
62.351
62.l«5
81,426
61.049
79.9S2
70.28B
69,774
66.276
61.143
60.075
It. 670
!Z,961
«7,e«i
45.613
43,352
IS. 058
31.740
31.341
30.774
30.129
10. m
29.472
12.166
12.112
12.097
11.857
II. 057
9.8TO
Cs = 0.14
Upper Limit x
of Concentra-
tion Category
(ppm)
0.1UO
0,1 it,
0.1JS
0,1?Q
O.l^b
o.iaz
0,110
0,115
0,112
o,ioa
0,105
0,101
o.ois
0,090
0.041
0,067
o,ofla
iJ.Oflf
0,077
O.OTs
0.070
0.067
O.Obl
O.Ot.0
0,OS6
0,053
0,009
O.CU6
O.ftOZ
0.039
0.015
0.032
0,071
0.025
0.021
o.oia
0,014
O.ftll
0.007
O.OOfl
Percentage of
Values Less
Than Or Equal
to x
100.000
85.71(1
M.05S
6S.681
fm.tSQ?
83.9?T
»2>9ti
7S.9S9
7S.509
73.338
70,082
bfc.OPQ
»5.0f5
6O,«?6
*3.3b5
57.06P
Sk.OS2
51.097
50,510
«7.«10
5,in^
56.395
35,?6Q
J?.79,S
S1.9ft5
Si. 151
10.877
1D.B32
10,529
F9.883
Z9.SQ1
15.520
I?.
-------
TABLE B-9
Effects of Control Strategies on 9-Hourly Average Concentrations.
Cumulative Frequency Distributions for 24-Hourly Running Means
w
to
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
0.212
0,207
0,202
C.H7
0,191
0.1B6
o.m
0.17%
0.170
0,165
0,159
0.154
0,149
0.14J
0.138
o,m
0,127
0,122
0,117
a. ill
0,106
0,101
0,09fc
0.090
O.OB5
o.oao
0,074
0.069
0.064
0.058
0,051
O.Oas
O.OU2
O.OIT
0,012
0,927
0,021
O.DIt
0.011
0.005
Percentage of
Values Less
Than Or Equal
to x
100.000
99.B97
99,897
99, sea
99,554
99.4B5
97.67?
97, ?«
97,015
96.876
91.709
91.217
9J.P01
90.285
90,113
89.598
89.461
B9.221
Bt.191
77.BSB
77.206
77,101
76.759
7S.2B1
72.BBO
70.S11
tl.AOl
tl.277
;a.<62
52,k9b
a. 129
aq.oia
2t.I>3l>
14,«H2
7,72«
3.9K2
O.H7
0.2<0
0.0
o.o
Cs = 0.2
Upper Limit x
of Concentra-
tion Category
(ppm)
0,125
0, 122
0.1I«
0,116
0,11).
0,104
0,106
0,103
0,100
0,097
0,094
0,011
0,08ft
0,064
0,081
0,078
0,075
0.0TZ
0,0 if
0.0 (.6
Q.ObS
0.059
0.0*»
O.OS3
0,050
9,017
0,040
0.001
0,038
0,OiU
0,011
0,02ft
0.02S
0.022
0.019
O.OIb
0,011
0.009
O.OOfc
0,001
Percentage of
Values Less
Than Or Equal
to x
100,000
49,931
f9,8kj
*9,794
99,T?S
99,t>??
99.S19
99,URS
99,31)2
99.10J
96,833
98,421
98, lab
97,631
97,391
96,6(17
95,198
t4,U70
67.9S9
*l.aQ9
58, »m
5a,i>S2
S0.9afl
«9,159
7,031
flfl.lM
?6,lHa
?fc,02t
lfc.033
12.S3D
T.b21
o.Sbfc
1 ,0fcfl
0,687
0,f>17
0,200
0,0
0,0
0,0
0.0
Cs = 0.14
Upper Limit x
of Concentra-
tion Category
(ppm)
0,096
0,096
0,093
0.091
o,o(>a
O.OA6
0,083
0,061
0,0?*
0,076
o.ora
0.071
0,069
O.Oftb
(1.06ft
i) .461
4.CS9
o.rst
o.c^c
O.cbZ
O.OJ9
0.0a7
0.000
0,0.1?
0,099
0.037
0.034
6,012
0,029
0.027
0,0?$
0.022
t),0?0
0.017
0,015
0,012
d.010
0,007
0,005
0,002
Percentage of
Values Less
Than Or Equal
to x
100.000
V9.B61
99.828
99,691
99,62?
99.451
99.34B
99,142
98,867
98.730
9.8,318
97,769
97.421
97.098
96,?58
95.228
94,404
93,169
91,727
30,669
8.163
45,912
29.763
18.150
22.58B
16.147
I1.S91
(.219
5.116
4.045
1.064
0.687
0.687
0.240
0.240
0,0
0,0
0,0
0,0
0,0
Design Standard Averaging Time, Ts = 9 HRS. Design Standard, Cs, is variable.
-------
TABLE B-10
Effects of Control Strategies on 18-Hourly Average Concentrations.
Cumulative Frequency Distributions for 3-Hourly Running Means
w
(-
o
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
1.155
1.127
I.09B
l,06«
1.010
1,01)
0.462
*.«3
0,9?fl
o,s9s
0,867
0,830
0.809
0,700
0.75.1
0.722
0,fc93
o,66a
0,6*6
0,607
6t57»
0,594
»,%20
o.»4i
0,«6?
0,«33
,04
.374
,J«7
1.31*
0.24*
.260
C.2U
«.ZO*
1.171
§,!
,116
0.867
0.65B
«,02»
Percentage of
Values Less
Than Or Equal
to x
100.000
49.261
94, 2M
49.201
99, 2M
49,281
49,281
'9. 2M
47.56*
97.568
47.568
47.548
47.568
47.566
47.569
47.568
47.397
47.347
47.363
46.199
46.199
46. 1 4V
96.130
4«,«47
4«,M8
43.136
2.671
42.671
6. MIS
B3.»01
Mt,t6«
77.021
72.774
67.6ftl
6S.822
61. 8«7
6«.7?6
31.H63
11.863
la. 64?
Cs = 0.14
Upper Limit x
of Concentra-
tion Category
(ppm)
0.010
o.tn
O.T90
0.7TT
O.Ti*
0.7JS
O.Ttfl
O.b93
0,6T2
0,t5l
0,630
0,604
O.SBA
O.S67
0,5«(.
0.5J5
O.%0«
«,flBS
0.462
O.«4l
0.420
t.111
0,17!
0.1)7
0,336
0,315
0,2«4
0,271
0.2S3
C.23I
0,210
0,111
0,118
0,117
0.126
0,10!
O.OB4
0,063
0.012
0.021
Percentage of
Values Less
Than Or Equal
to x
100.000
«.«»»
99.S63
9.B2?
*1.Tf>
99,726
99,692
99.55b
99.521
99,521
99.281
99,247
99,247 .
99,247
99,247
99.212
99.212
99.1«4
99.110
99,075
98,973
90.973
9B.9n«
96.767
94,075
91.712
9.349
09.349
15.205
79.966
74.110
73.459
71.404
69,726
69,623
4B.9D4
39,212
39.001
39,007
35,342
Cs = 0.10
Upper Limit x
of Concentra-
tion Category
(ppm)
0,600
o.ses
0,570
O.SS5
0.5x0
0.525
0.510
0.495
o.aeo
0,16$
O.DSO
0.43S
0,020
«,<05
0,3QO
0.375
0.3dO
0.3«5
0.130
0.315
0.300
0,285
0.270
0 , ?£5
0,240
0.22%
0.210
0.195
0,1*0
0,165
0.1SO
0.135
0.120
0,105
0,040
0,075
0.0*0
0.0«5
0.010
0.115
Percentage of
Values Less
Than Or Equal
to x
100.000
44,555
4 9, -.15
49.521
49,0Kb
99,)no
99.315
44.315
99. 2aT
49.212
44.]0q
49. tan
t9,007
48.973
48.971
»8.973
48,90*
4«.««6
4a.aie
42.245
B4.4fefr
9.623
"4.155
S.J42
80.271
77,53*
71.760
7».10«
75,471
7|,7«7
70. Ola
64,745
69.T60
64,178
O.Jo?
34.4.71
34,178
34.041
34,0ai
35.102
Design Standard Averaging Time, Ts = 18 HRS. Design Standard, Cs, is variable.
-------
TABLE B-ll
Effects of Control Strategies on 18-Hourly Average Concentrations.
Cumulative Frequency Distributions for 9-Hburly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
Cppm)
0,506
0.413
0.4BO
0.06R
0.055
0,492
0.410
0,«I7
0.405
0.392
0.319
0.367
0.150
O.Jul
0.321
0.1H
0.301
0,211
0.2FB
0.265
0.253
0.240
0.228
».2i5
0.202
0,110
0.17?
0.164
0,152
0,131
0,126
0,11*
0.101
0.0*8
0,076
0,063
0,051
0.038
0.02S
0.011
Percentage of
Values Less
Than Or Equal
to x
100,000
11.131
94, BS2
11,341
11.280
91.280
11.075
1B.3S5
IB, 25?
U.252
1t.»U
»b,»IJ
»S.7«0
V6.7IO
IB. 710
16,116
12.221
11,147
11.871
11.170
11,711
11.501
10.6A4
89.HO
87.2B6
>6,}15
B6.0S2
13.162
8.1,110
80.158
BO.I-.8
»8. 151
60,818
32.7H2
«T.O!0
11,270
10,790
Z1.8A1
12.011
11,8^7
Cs = 0.14
Upper Limit x
of Concentra-
tion Category
Cppm)
0,280
0,271
0,266
0,251
0,252
0,205
0,238
0,231
ti,22B
0.21J
0,210
0,203
0,116
0,181
0,192
0.175
0,16B
0,161
0,150
O.H7
0,100
0.133
0,126
*,111
il, 112
0,105
0,018
0,011
0.0&4
0,077
0,070
0,063
0,056
0,0*1
0,002
0,035
il.OJB
11.021
0.01Q
0,007
Percentage of
Values Less
Than Or Equal
to x
100.000
11.086
11.486
11.183
11.115
11.075
18,103
18.021
18.2H6
18.286
17.361
17.0P7
16.573
16.116
16.127
15.156
15.571
«.16B
10.170
13,450
13.077
12.975
12.735
11.398
80.321
76.01 1
70.211
69.210
61,160
54.626
«7.772
11.753
36.566
35.230
35.058
39.202
15,387
15.289
15,113
12,851
Cs = 0.10
Upper Limit x
of Concentra-
tion Category
(ppm)
0,200
0,115
0,110
0,183
0,180
0.175
0,170
0,165
0,160
0.155
0,150
0,195
0,190
0,135
0,130
0,125
0,120
0,115
0,110
0,105
0,100
0,095
0.010
0,085
o,oao
0,075
0,070
0.065
0.060
0,055
0,050
0,015
0,000
0,035
0,030
0,025
D,020
0,015
0,010
0.005
Percentage of
Values Less
Than Or Equal
to x
100.000
17,513
97,327
16.1BO
96.807
96.070
95.888
15,888
15.406
H.7H
14,106
13,100
93.557
91.281
13.112
11.112
12.255
81,311
00.770
76,831
71,487
70.528
60.166
62.171
55.175
51.851
46.429
41.227
11.113
16.K1
35.572
35.161
34.110
34.304
16.038
15,627
15.353
19,147
15.147
12,151
Design Standard Averaging Time, Ts = 18 HRS. Design Standard, Cs, is variable.
-------
TABLE B-12
Effects of Control Strategies on 18-Hourly Average Concentrations.
Cumulative Frequency Distributions for 24-Hourly Running Means
to
i
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
0,21!
0.20}
0,202
i, in
o.m
0.114
o.m
0,175
o.-uo
O.IM
0.159
0.151
o, 109
d.l«J
0.1)8
0,11}
0.127
0,122
0,117
0,112
0,10k
0,101
0,04*
0,090
0.005
0,080
0,07*
0,049
Ofrt
0.0*0
0,053
0,0ns
0.0«2
0,037
.012
.027
1.021
t.Olt
0.011
o.oof
Percentage of
Values Less
Than Or Equal
to x
100.000
99. 697
99.R97
91,568
99.55«
99.H05
97,87?
97. ?^a
97.0tl
96.876
95.709
95.237
93.703
90,?8S
«0. 113
l»9.5«»
09.46T
69.221
BS.I91
77.058
77.206
77.103
74.759
75,283
72. «*0
TO. Ml
44.t01
41.277
50.462
S?,»4S
8.179
M.Ofl*
?fc.S36
IS. -12
t.7?i
3. » ft 2
e.*M
0,?IO
»
1,0
Cs = 0.14
Upper Limit x
of Concentra-
tion Category
Cppm)
0,1*0
0.117
0,153
0,150
1.126
0.123
0.111
0.116
0.11?
0,109
O.IOS
0.102
0,090
0,095
0.091
0,066
O.OSfl
0.061
0,077
0.074
9,070
0,0*7
ft.ot*
d,960
0,05*
0,053
0.044
0.0'fc
0.042
0.039
0,035
O.OJ2
0.0*8
0,425
0.071
0,011
0,OU
0.011
0,007
O.OOfl
Percentage of
Values Less
Than Or Equal
to x
100,000
99.966
99.966
99.966
99. 9M
99,911
99.911
99. PM
99.79a
99,657
99.6S7
«7.a?S
97.«?5
»7.l«5
96,7011
94,950
««.370
«?,e25
»0.«60
09,015
et.JJT
85.S1B
e?,e)6
00,619
76.279
74.200
71.050
4V7.6?B
3.5iJ
31,966
21.?«1
15.379
5.630
1.411
l.l*T
1,133
0,994
s.e
o.o
9.0
Cs = 0.10.
Upper Limit x
of Concentra-
tion Category
(ppm)
0.109
0,107
0,104
0.101
0,098
0,096
0.093
0,090
0,067
0.005
0,082
0.079
0.077
0,074
a. 071
0,068
0.066
0,063
6,660
0,057
6. OS*
0,052
0,049
o,e«6
0.041
0.041
fl.03«
0.936
0.013
4.910
C.02T
0.02S
0.022
0.019
0.016
0,014
0.011
s.oot
9.00S
fl.OOl
Percentage of
Values Less
Than Or Equal
to x
100.000
99.9JI
99.097
99.79a
99.7'H
99.691
99.691
99,491
99.SIB
99.H2
99.3K2
««.?«>
9I.B33
90.S60
17.101
06.703
15.416
13.115
Bt.256
77.755
75.935
71. '51
72.331
49. '22
16.035
«Z.«30
X.329
K.040
I4.032
9.372
5.959
1.957
1.373
1.202
1.133
0.994
0.103
.069
0.0
0.0
Design Standard Averaging Time, Ts = 18 HRS. Design Standard, Cs, is variable.
-------
TABLE B-13
Effects of Control Strategies on 24-Hourly Average Concentrations.
Cumulative Frequency Distributions for 3-Hourly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
1.1M
t.127
1.04B
1,049
1,040
1,011
0,462
0,95!
0,«24
0,845
0,167
0,«1»
0,809
0.780
I. lit
ft, 122
0.693
0,464
9,416
0.607
6,578
0.544
0.520
0.441
9,462
0,411
0.404
0,376
0,147
.lit
0.284
0,240
0.211
0.202
0.171
0,144
0.116
0,097
0,058
0.029
Percentage of
Values Less
Than Or Equal
to x
100.000
9«. ZM
99,281
99.2B1
11, lf\
99.281
99.281
99.281
97.568
97.568
7.56B
7.568
7.568
7.568
47.568
7.S6B
7.197
«7,397
7.161
»t.l»«
>.!'
0,420
0.400
0,160
0,160
0,340
0,320
0,100
0.2BD
0,260
0,240
0,220
0,200
O.IBO
0,160
0,190
0.120
0,100
O.OQO
0.060
0,040
0,020
Percentage of
Values Less
Than Or Equal
to x
100,000
9.847
9.847
49.863
9.863
99.658
99.589
9.584
9.0(16
99.115
9.11i
94.315
99.281
99.2B1
9,247
94.207
99.I7B
99.104
9.110
.007
B.V71
8,«7S
6.870
4.178
I.B41
89.418
A4.41A
85.177
BO. 274
77.192
71.836
71.596
»«.«»(.
6«.726
69,621
46,101
39,115
39.041
39.007
IS. 342
Cs = 0.08
Upper Limit x
of Concentra-
tion Category
(ppm)
0,600
0,6?4
0,608
0.59J
0,576
0,560
0.544
0.528
8,512
0,496
0,9BQ
0,464
0,448
0,432
0,414
0,400
0,164
0,168
0.1S2
9,116
0.120
0,104
0,288
0,272
0,256
0,240
0,224
0.201
0,192
0,176
0,160
0,144
0,128
0,112
0,096
0,080
0.064
0.048
0,012
0.016
Percentage of
Values Less
Than Or Equal
to x
100,000
9,7?6
99.692
9,658
9.6V8
99.5B9
9.452
9.344
99. m
4. 1H
49.281
9.247
99.I7B
99.041
49.041
99.041
99,007
98.916
94.674
44.555
2.141
(9.966
89.79S
4.589
IS. 616
0.582
»I.«7»
74.673
74.281
72.021
70.342
70.014
64. 821
6V. 760
46.541
»4.«M
1».!15
H.075
19.041
35,377
Design Standard Averaging Time, Ts = 24 HRS. Design Standard, Cs, is variable.
-------
TABLE B-14
Effects of Control Strategies on 24-Hourly Average Concentrations.
Cumulative Frequency Distribution for 9-Hourly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
(ppm)
e.506
0.993
0,01
O.fiBfl
0,155
0.902
i),*30
O.c.17
0,<05
0,312
0.571
0.367
0.350
0.301
0,12V
0.31k
0.103
0,201
0.27«
0,2»5
1,253
0,290
0.22»
0,215
0,202
0.100
1T7
0.160
0,152
O.IJ9
0,17*
0,11"
0.101
o,o»e
0,076
0,063
1.051
0.03»
0,025
0.013
Percentage of
Values Less
Than Or Equal
to x
100.000
99.9J1
99.BS2
99,309
99,260
99.28(1
99.07«>
9H.35S
9R.2S2
98.25?
96.«13
96,611
9 6. To a
96,710
*6,710
96,«16
92.221
91.9-7
91,878
91.6T8
91,741
91.501
90.644
69.690
ST. 266
6.395
6.052
83. «62
81,980
BO. 158
so. iss
68.951
60.644
57,70?
7.010
14,270
10.700
29,««9
12.097
1 1.6S7
Cs = 0.14
Upper Limit x
of Concentra-
tion Category-
(ppm)
0.i7J
O.lfcU
0.555
0,105
0,H6
0.327
O.JtT
0,508
0,299
0,269
0,280
0,271
0,261
0,Zb2
0,249
0,213
0.224
0.215
0.205
0.196
0.167
0.177
0,168
0.1*4
O.ln9
0,110
0.131
«.1ZI
0.112
0.103
0,093
«.oeq
(1,075
0,065
B.056
9.047
0.037
A. 028
0.019
0,009
Percentage of
Values Less
Than or Equal
to x
100.000
«.OI2
?.«»»
7.772
7.533
»7.«H
f7.190
S.*3a
3.591
3.2B3
3.000
*2.<>1
2.255
I.'MI
»1.7«1
I.05A
O.A79
0.130
87.731
7.0RO
8fc.«9S
Kb. 29?
85.607
83.619
1.837
80.535
80,329
7».0
0.233
0,227
0,220
0.213
0,207
0,200
0,191
0,167
0,110
0.171
0.167
o.uo
O.IS1
0,197
0,100
0.131
0.127
0.120 .
0.111
0,107
0,100
0,091
0.007
O.OSO
0,071
0.067
0.060
0,051
0(097
0,090
0.011
0.027
0.020
0.011
0.007
Percentage of
Values Less
Than Or Equal
to x
100.000
*9.623
99,569
99,966
99.066
98.601
96.595
98.561
9A.M5
97.Q6B
97,127
96.6,3
96.. 036
96.367
96,093
«5. 716
95.202
90.170
91.669
93.209
93,003
92.900
92.015
00.569
74.1*8
70.611
69.S(,3
*2.*B3
$5.106
51,337
0.9)8
39. !!
15.401
35.161
Sa.990
19.01
15.559
15.244
15.113
12.651
Cs = 0.08
Upper Limit x
of Concentra-
tion Category
(ppm)
0,211
6,208
0,203
8.H7
0.192
0,167
0,101
0,176
0,111
0,165
0.160
0.1SS
0.199
0,190
0,139
0,133
0.128
0,121
0,117
0,112
0,197
0,101
0,096
0,091
0,0>5
0,080
0.075
0.069
0.069
0.059
0,053
0,098
0.01
0,037
0.012
0,027
0.021
0,016
0,011
0.005
Percentage of
Values Less
Than Or Equal
to x
100.000
<8.6?9
96.024
98,266
97,772
97.361
Ob.BM
96.070
96.093
96.059
95.7SI
95.065
94.345
-«3,729
91.55T
91.557
93.020
92.769
ft2.3I7
B1.1H6
77.142
71.931
71.105
70.010
63.1911
55.655
5)2.296
45,065
o.oaa
ST. 252
35.915
35.401
35.09]
3«.78«
19,705
Ik. 038
15.559
15,1*2
15.117
Design Standard Averaging Time, Ts = 24 HRS.
Design Standard, Cs, is variable.
-------
TABLE B-15
Effects of Control Strategies on 24-Hourly Average Concentrations.
Cumulative Frequency Distributions for 24-Hourly Running Means
NO SCS
Upper Limit x
of Concentra-
tion Category
Cppm)
0,212
0,207
0,202
o.m
o.m
0,186
0,161
0.174
0,170
0,165
0,159
0,150
0,149
0,l<]
0,116
0.133
0,121
0,122
0,117
0,112
4,106
0,101
0,096
,090
0.005
0.060
0,07«
0,06*
0.0«»
0.054
0,051
0.0«»
0,042
0,037
0.012
0.02T
9,021
0.01.
0.011
0,00*
Percentage of
Values Less
Than Or Equal
to x
100.000
99,897
»9.(I97
99,568
99. ««
99,BGi
97.87?
97,?Sa
97,015
96,87fc
9S.709
91.2S7
93.?P5
»0.2Si
90.1 11
M.SCft
B9...IV1
89.221
es.m
77.858
77,306
77,103
76.7S9
75.ZB3
7?. »«0
70.511
ta.aOl
61,277
58.462
52.69S
B.129
««.0«B
?6.S36
1*.0*2
7.7?a
1.9ft?
0,*.«7
0.210
0,0
fl,0
Cs = 0.14
Upper Limit x
of Concentra-
tion Category
Cppm)
0.1HO
0,136
0,133
0,129
o.ut.
0,122
0,119
0,115
0,112
O.lOfl
0.10S
0.101
0.098
0.090
0,091
0.0«7
0,081
O.OflO
0,077
0,074
0.070
0.067
O.Ofcl
0.060
O.OS6 .
O.OS3
0,OU9
O.OOfc
i),0«2
d,OA9
0.035
O.OS2
0.026
0,025
0,021
0,01ft
0,014
0,011
0.007
o.ooa
Percentage of
Values Less
Than or Equal
to x
100.000
»2.6'i«
"2.310
»1.J1S
VO.ti)
«4.a<*l
I1.S30
I."*?
RO.flS
7(».57a
7«.71»
77.789
'7,171
76. &b*
75,415
7i.tol
72.50J
70.7S2
».671
*i.ni
>l.39»
40.934
57.91 J
51.931
50.292
S.?W
5.657
42.677
26.962
21,250
13. 1««
1.067
.463
1.099
0.667
0.210
0.200
0.0
0.0
o.o
Cs = 0.10
Upper Limit x
of Concentra-
tion Category
Cppm)
0,100
0.09T
0,09!>
0,01?
0.090
0,067
0.06*
o.oe?
0.060
0,077
0.07S
0,072
0,070
0,067
0,065
0,062
0.060
O.O-il
O.ObS
O.OS2
0,0^0
D,0a7
O.OflS
0,002
o,oao
0.037
0,03S
0.053
d. 030
0,028
0,02S
0.023
0,020
0,018
0,015
0.013
0,010
6.00*
0,005
0.005
Percentage of
Values Less
Than Or Equal
to x
100.000
96.009
97.603
97.631
97.3P2
95.606
95,125
94.6*5
91.512
91.555
90.368
68.326
66.749
66.234
64.930
62,527
60.879
77.274
75.66T
7a,5?8
72.777.
70.166
66.426
3.907
36.595
26,227
22,726
16.066
06
5.669
1.6H2
1.236
1.113
1,153
1,133
0.461
0.137
0.0
0.0
0,0
Cs = 0.08
Upper Limit x
of Concentra-
tion Category
Cppm)
0,080
0,076
0,07t
0,074
0,072
0,070
0,066
0,0b6
0,0b4
0,062
0,060
o.ose
0,056
0,054
0,052
0,050
0.0*6
0.046
0,099
0,042
0,040
0,039
0.036
0.030
0,032
0,030
0,028
0.026
0.024
0,022
0,020
0,018
0,016
0,014
0.012
0.010
0,008
0,066
0,004
0,002
Percentage of
Values Less
Than Or Equal
to x
100.000
91.752
92.666
91.864
90.422
69.049
67.470
66.371
64.566
61.865
62.0S6
79.366
77.112
75.695
79.597
73,361
71.061
66.246
47.237
44.456
36.635
27.532
25.609
22.794
16.512
9.509
.317
i.399
1,751
1,270
1,236
1.131
1.131
0.996
0,461
0.275
e.117
0.0
«.»
0.0
Design Standard Averaging Time, Ts = 24 HRS. Design Standard, Cs, is variable.
------- |