United States Office ot February 1981
Environmental Protection Planning and Evaluation
Agencv Washington DC 20460
&EPA Characterization of Air
Pollution Control Equipment
Operation and Maintenance
Problems
-------
CHARACTERIZATION OP AIR POLLUTION
CONTROL EQUIPMENT OPERATION AND
MAINTENANCE PROBLEMS
Prepared by
Robert G. Mclnnes
Peter H. Anderson
Under
EPA Contract No. 68-01-4143
Technical Service Area No. 3
Task Order No. 67
EPA Project Officer
James S. Vickery
Program Evaluation Division
Office of Planning and Evaluation
U.S Environmental Protection Agency
Washington, D.C 20460
February 1981
-------
DISCLAIMER
This Report was furnished to the U.S. Environmental
Protection Agency by GCA Corporation, GCA/Technology
Division, Bedford, Massachusetts 01730, in fulfillment of
Contract No. 68-01-4143, Technical Service Area No. 3,
Task Order No. 67. It has been reviewed by the Program
Evaluation Division of the Office of Planning and Evaluation,
EPA, and approved for publication. The opinions, findings,
and conclusions expressed are those of the authors and not
necessarily those of the Environmental Protection Agency
or of cooperating agencies. Mention of company or product
names is not to be considered as an endorsement by the
Environmental Protection Agency or recommendation for use.
-------
EXECUTIVE SUMMARY
The Environmental Protection Agency (EPA) and the Council on Environmental
Quality (CEQ) sponsored five separate studies and EPA conducted an additional
four studies for a total of nine to determine the extent to which industrial
sources emit pollutants in excess of their applicable regulatory limit. These
studies were directed at better understanding the degree to which a source main-
tains continuous compliance over time. The frequency, duration, and magnitude
of individual "excess emissions" episodes were quantified as was the primary
cause of each emissions incident. These reports have been submitted to the EPA,
by the participating contractors.
This report presents the results of classifying, synthesizing and analyzing
the data compiled in these initial reports. A total of 169 individual source
surveys were investigated and quantified, from which 119 data points on excess
emissions incidents were extracted. GCA/Technology Division correlated this
excess emissions data with four causal factors and four major source and control
equipment parameters in order to answer three distinct questions:
• What is the extent of the excess emissions problem?
• Why is this problem occurring?
• Who is experiencing this problem?
A presentation of the data averages which were statistically adjusted to
eliminate the bias of unusally large values serves to define the extent of
the problem and the "typical" source investigated in this report.
Source Data
Annual uncontrolled emissions (tons) 13,341.5
Annualized allowable emissions (tons) 179.2
Annual excess emissions (tons) 22.0
Normalized excess (percent of allowed) 24.7
Annual emissions credits (tons) 100.1
Normalized credits (percent of allowed) 50.0
Excess Emissions Incidents Data
• Frequency (incidents/year) 12.3
• Duration (hours/incident) 23.7
• Magnitude (percent in excess of allowable/incident) 828-0
• Magnitude (tons/incident) 1.8
iii
-------
Control Equipment Data
• Capital cost (103$) 1,623.3
• Annual operating and maintenance cost (103$) 180.9
• Size of control device (103acfm) 105.4
• Age of control device (years) 5.1
Figure 1 graphically displays this data for the "average" source. With
reference to the figure, the average source experiences twelve excess emissions
incidents per year each lasting 23.7 hours with an emissions release of 1.8
tons which is equal to 8.3 times that allowed. On a yearly basis, the source
operates at a normal emission level equal to 50 percent of that allowed, and
emits excess emission resulting from the twelve incidents which is equivalent to
24.7 percent of that allowed.
The data for the "average" control device is graphically presented in
Figure 2. Referring to the figure, the control device has an inlet loading of
13,341.5 tons per year and a controlled outlet emission rate of 79.1 tons.
This "average" device, which is required to maintain a pollutant collection
efficiency of at least 98.7 percent in order to meet the regulatory limit, has
an actual collection efficiency of 99.4 percent. The device was installed in
mid 1972 at a 1977 cost equivalent of $1,623,300. and has a flow of 105,400
acfm. Annual operation and maintenance costs are $180,900.
To determine why an excess emissions problem occurs, we investigate its
causes. All factors leading to an excess emissions incident were defined by
one of the following four causal codes:
• Causal Code 1 - Design Related
• Causal Code 2 - Process Disruption Related
• Causal Code 3 - Control Equipment O&M Related
• Causal Code 4 - Unforeseen Event
The frequency duration and absolute magnitude of each incident was first
quantified and each incident was tabulated by its causal code. The pro-
portion of all frequencies, durations and absolute magnitude attributable
to each of the four causal codes was then calculated. A graphical rep-
resentation of these calculations is presented in Figure 3.
These charts display the relative contribution of each causal factor
to the primary excess emissions indicators of frequency, duration, and
absolute magnitude. They were derived by separately dividing the frequency,
duration and absolute magnitude of incidents with 'similar causal codes by
the total number of incident frequencies, the total hours of excess mode
operation and the absolute tonnage of all incidents in the excess mode,
respectively.
iv
-------
828%
AVERAGE
INCIDENT
III
>
Ul
to
CO
23.7 Nrt
12 INCIDENTS
TOTAL OP 22.0 toni
EMISSION CREDITS
100.1 ten*
ALLOWABLE
EMISSIONS
~I79.2 font
ACTUAL EMISSIONS
79.1 tens
.ALLOWED
EMISSION
LEVEL
ACTUAL
EMISSION
LEVEL
I YEAR
Figure 1. Source Profile
-------
79.1 tons
13,341.5 tons
Actual Control Efficiency = 99.4%
Control Efficiency
Required for Compliance =98.7
Control: Age: 5.1 years
Size: 105,400 ACFM
Capital cost: $1,623,300 (1977) = $15.40/ACFM
O&M cost: $180,900 (1977) - $1.72/ACFM
Figure 2. Control Equipment Profile
vi
-------
FREQUENCY
CAUSAL CODE
(UNFORSEEN)
3.9%
CAUSAL CODE I
(DESIGN)
8.7%
CAUSAL CODE 3
(CONTROL 0 8 M)
32.6%
CAUSAL CODE
(PROCESS)
54.8%
CAUSAL CODE 4
(UNFORSEEN)
13.8V*
DURATION
CAUSAL CODE 3
(CONTROL 0 8 M )
37.1%
CAUSAL CODE I
(DESIGN)
37.0%
CAUSAL CODE 2
' (PROCESS)
12.1%
EXCESS
CAUSAL CODE 4
(UNFORSEEN)
10.8%
CAUSAL CODE 3
(CONTROL 0 a M)
23.9%
CAUSAL CODE 2
(PROCESS)
7.4%
CAUSAL CODE I
(DESIGN)
57.9%
Figure 3. Percent contribution of each causal code by incident indicator,
vii
-------
With respect to "who" is experiencing the problems, the data from the
119 sources were analyzed by four major process and control related para-
meters. These were:
• Source type - industrial classification
• Source size - uncontrolled emission potential
• Control device type
• Control device size
Each parameter was broken down into specific categories to facilitate
data analysis. Excess emissions indicator data (frequency, duration,
magnitude) were then entered into each category according to the character-
istics of the source surveyed. Once tabulated, the data for each para-
meter was ranked by category according to the average frequency, duration,
and magnitude of occurrence. These individual indicator standings were
then integrated into one overall ranking which displayed the seriousness
of the excess emissions problems by category. The results of this integrat-
ed ranking technique are presented in Table 1. According to our established
convention, the lower the rank number, the greater the excess emissions
problem for that category.
The majority of this report concerns itself with the tabulation of
incident data along parameter and category lines, in an effort to profile the
problem sources. A total of nearly 3000 excess emissions incidents were
documented and synthesized to form the basis of our findings. We feel
that the true extent of excess emissions may be greater than our study has
shown. We also feel that a more thorough study aimed at specific industrial
categories will serve to more clearly delineate the problem.
viii
-------
TABLE 1. OVERALL RANKING OF PARAMETERS A THROUGH D.
Rank
I
2
3
tt
5
6
1
a
9
;.)
Ll
12
13
14
IS
16
Parameter A
Industrial category
Category
Cogent plant *
Surface coating operations
Grain handling operation*
^••iioleuB product* and handling
Incinerator*
Stone, clay and glai* plant*
Sloan generating plant*
Pulp and paper dill*
?ctrocheaical plant*
Lumber and wood plant*
Alpha It plants
Iron and *teel plant*
Other
Bra** and bronze plant*
Pood and drug plant*
Aluminum plant*
Total
Number
of camples
9
13
5
13
6
10
10
6
6
J
10
10
6
S
3
4
- 119
Parameter B Parameter C ParaaMter D
Source »i*e Control device type Control device sire
Category Number Nu>ber Category Number
(103 ton/yr) of (ample* w"««or» of (aple* (103 acfm) of sample*
100 -500 6 ESP 22 500 - 1000 6
0.1 - 0.5 31 Scrubber 36 100 - 500 23
10 - 100 11 Other 26 10 - SO 49
<0.l 24 Baghouie 3} <10 31
1-10 30 >1000 I
O.J - 1 17 50 - 100 9
~— ~ ~ ~
Total - 119 Total - 119 Total - 119
-------
CONTENTS
Executive Summary
Figures
Tables
1. Introduction .......................... 1
Background ........................ 1
Source data base ..................... 2
Definition of key terms .................. 5
Evaluation parameters ................... 6
Excess emission incident categorization .......... 8
Ranking of source excess emissions incidents ....... 18
Integration of rankings .................. 20
2. Discussion of Data ....................... 22
Parameter A. Excess emissions as a function of
industrial category ................. 22
Parameter B. Excess emissions as a function of
source size ..................... 37
Parameter C. Excess emissions as a function of
control device type ................. 43
Parameter D. Excess emissions as a function of
control device size .................. 48
Ranking of parameters/integration of rankings ....... 54
3. Conclusions and Recommendations ................ 98
Data review ........................ 98
Data modification ..................... 100
Problem area profiles ................... 102
Indicator profiles .................... 109
Recommendations ...................... 113
Appendix A. Raw Data Summary
-------
FIGURES
Number Page
1 Source profile v
2 Control equipment profile vi
3 Percent contribution of each causal code by incident indicator vii
4 Percent contribution of each causal code by all indicators viii
xi
-------
TABLES
Number
1 Overall Ranking of Parameters A Through D
Page
2 Parameter A - Industrial Classification
Data Summary 1°
3 Parameter B - Source Size
Data Summary 13
4 Parameter C - Control Device Type
Data Summary 15
5 Parameter D - Control Device Size
Data Summary 16
6 Excess/Credits
Data Summary 19
7 Parameter A - Industrial Classification
Category Rankings
Causal Code 1 (Design) 55
8 Parameter A - Industrial Classification
Category Rankings
Causal Code 2 (Process) 56
9 Parameter A - Industrial Classification
Category Rankings
Causal Code 3 (Control O&M) 58
10 Parameter A - Industrial Classification
Category Rankings
Causal Code 4 (Unforeseen) 59
11 Parameter A - Industrial Classification
Category Rankings
All Causal Codes 60
12 Parameter A - Industrial Classification
Integration of Rankings
Causal Code 1 (Design) 62
xli
-------
TABLES (continued)
Number Page
13 Parameter A - Industrial Classification
Integration of Ranking
Causal Code 2 (Process) 63
14 Parameter A - Industrial Classification
Integration of Rankings
Causal Code 3 (Control O&M) 65
15 Parameter A - Industrial Classification
Integration of Rankings
Causal Code 4 (Unforeseen) 66
16 Parameter A - Industrial Classification
Integration of Rankings
All Causal Codes 67
17 Parameter A - Industrial Classification
Ranking Normalized Credits and Excesses 69
18 Parameter B - Source Size Uncontrolled Emissions
Category Rankings
Causal Code 1 (Design) 71
19 Parameter B - Source Size Uncontrolled Emissions
Category Rankings
Causal Code 2 (Process) 72
20 Parameter B - Source Size Uncontrolled Emissions
Category Rankings
Causal Code 3 (Control O&M) 73
21 Parameter B - Source Size Uncontrolled Emissions
Category Rankings
Causal Code 4 (Unforeseen) 73
22 Parameter B - Source Size Uncontrolled Emissions
Category Rankings
All Causal Codes 74
23 Parameter B - Source Size Uncontrolled Emissions
Integration of Rankings
Causal Code 1 (Design) 75
24 Parameter B - Source Size Uncontrolled Emissions
Integration of Rankings
Causal Code 2 (Process) 76
xiii
-------
TABLES (continued)
Number PaEe
25 Parameter B - Source Size Uncontrolled Emissions
Integration of Rankings
Causal Code 3 (Control O&M) 76
26 Parameter B - Source Size Uncontrolled Emissions
Integration of Rankings
Causal Code 4 (Unforeseen) 77
27 Parameter B - Source Size Uncontrolled Emissions
Integration of Rankings
All Causal Codes 78
28 Parameter B - Source Size Uncontrolled Emissions
Ranking of Normalized Credits and Excesses. . . 79
29 Parameter C - Control Device Type
Category Ranking
Causal Code 1 (Design) 81
30 Parameter C - Control Device Type
Category Ranking
Causal Code 2 (Process) 81
31 Parameter C - Control Device Type
Category Ranking
Causal Code 3 (Control O&M) 83
32 Parameter C - Control Device Type
Category Ranking
Causal Code 3 (Unforeseen) 83
33 Parameter C - Control Device Type
Category Ranking
All Causal Codes 83
34 Parameter C - Control Device Type
Integration of Rankings
Causal Code 1 (Design) 85
35 Parameter C - Control Device Type
Integration of Rankings
Causal Code 2 (Process) 85
36 Parameter C - Control Device Type
Integration of Rankings
Causal Code 3 (Control O&M) 86
xiv
-------
TABLES (continued)
Number Page
37 Parameter C - Control Device Type
Integration of Rankings
Causal Code 4 (Unforeseen) 86
38 Parameter C - Control Device Type
Integration of Rankings
All Causal Codes 88
39 Parameter C - Control Device Type
Ranking of Normalized
Credits and Excesses 88
40 Parameter D - Control Device Size (acfm)
Category Ranking
Causal Code 1 (Design) 89
41 Parameter D - Control Device Size (acfm)
Category Ranking
Causal Code 2 (Process) 90
42 Parameter D - Control Device Size (acfm)
Category Ranking
Causal Code 3 (Control O&M) 91
43 Parameter D - Control Device Size (acfm)
Category Ranking
Causal Code 4 (Unforeseen) 92
44 Parameter D - Control Device Size (acfm)
Category Ranking
All Causal Codes 92
45 Parameter D - Control Device Size (acfm)
Integration of Rankings
Causal Code 1 (Design) 93
46 Parameter D - Control Device (acfm)
Integration of Rankings
Causal Code 2 (Process) 94
47 Parameter D - Control Device Size (acfm)
Integration of Rankings
Causal Code 3 (Control O&M) 95
48 Parameter D - Control Device Size (acfm)
Integration of Rankings
Causal Code 4 (Unforeseen) 95
xv
-------
TABLES (continued)
Number Page
49 Parameter D - Control Device Size (acfm)
Integration of Rankings
All Causal Codes 96
50 Parameter D - Control Device Size (acfm)
Ranking of Normalized
Credits and Excesses 97
51 Unmodified Data Summary 99
52 Modified Data Summary 101
53 Problem Area Profile of Causal Code 1 (Design) 103
54 Problem Area Profile of Causal Code 2 (Process) 104
55 Problem Area Profile of Causal Code 3 (Control O&M) 105
56 Problem Area Profile of Causal Code 4 (Unforeseen) 106
57 Problem Area Profile of All Causal Codes 107
58 Frequency Indicator Profile 110
59 Duration Indicator Profile Ill
60 Magnitude Indicator Profile 112
xvi
-------
SECTION 1
INTRODUCTION
BACKGROUND
The Clean Air Act of 1970 and subsequent amendments passed by Congress
in the summer of 1977 provided for the establishment of national ambient air
quality standards specifying the maximum levels of certain "criteria" pollu-
tants to be permitted in the ambient air. States were charged to develop and
implement approved state implementation plans (SIPs) designed to limit the
amount of air pollutants emitted from stationary and mobile sources in accor-
dance with the established standards. These SIPs defined specific emission
limitations which, when achieved by all pollutant sources, should lead to the
attainment of the national ambient air quality standards. With certain noted
exceptions, much of the United States has, as of this date, achieved the air
quality standards established by EPA. Further refinements and revisions to
the SIPs are currently underway, however, in an attempt to resolve those still
outstanding "nonattainment" situations.
The vast majority of all major stationary sources (those with an uncon-
trolled emission potential in excess of 100 tons of pollutant per year) are
currently considered in compliance with their applicable emission limitation.
This compliance determination, however, is often based on estimated uncontrolled
pollutant emission rates and control equipment design efficiencies, or on
specific one-time stack tests of the pollutant in question. The extent to
which a source maintains the process and control equipment parameters upon
which the compliance determination was made has never been fully investigated.
Perturbations to steady state operation due to changes in these parameters can
cause emissions incidents which are in excess of the applicable regulatory
standard. To fill the informational gaps that currently exist concerning
"continued" compliance of stationary sources, the EPA, in conjunction with the
Council on Environmental Quality (CEQ) sponsored nine separate studies, invol-
ving six state and three local air pollution control agencies. While these
individual studies varied in their overall scope and purpose, all were involved
in the collection and processing of specific pollutant information from sta-
tionary sources within the jurisdiction of each air pollution control agency
studied. This data gathering involved comprehensive onsite inspections of the
sources surveyed, collection of operating and maintenance data on selected pro-
cesses at each source, and the reduction of the emission data obtained into
frequency, duration and magnitude values associated with each "excess emission"
incident. In addition, data concerning uncontrolled and allowable emission
rates, emission credits and excesses, age, type, size and cost of control
equipment were gathered during the site visits.
-------
The contractors involved in these initial source inspections, their geo-
graphic areas of concern, and the number of facilities evaluated and reported,
were as follows:
• GCA/Technology Division - New York State and Iowa - 29 sources
• The Research Corporation of New England (TRC) - Connecticut,
New Jersey and Oregon - 52 sources
• PEDCo/Booz-Allen and Hamilton - North Carolina and Houston -
40 sources
• Pacific Environmental Services (PES) - San Diego and Chicago -
48 sources
The results of these source evaluations have been previously reported to
the EPA and CEQ in fulfillment of the initial contract requirements.
The purpose of this study is to review the individual source data
from all these earlier studies, eliminate those sources that have unusable
data or that do not meet, the minimum criteria of this project, collate the
usable data by various source and control equipment parameters and report the
results of the data tabulation. While the specific contractor reports should
be consulted to determine individual source evaluation methodology and problems
encountered in facility selection and onsite data gathering, the highlights of
those efforts will be reemphasized in this study.
SOURCE DATA BASE
The 169 facilities surveyed by the original contractors were initially
selected based on the following EPA/CEQ source selection criteria:
1. Near equal distribution of sources of the following pollutants:
TSP, SOX, HC.
2. All sources are data rich; i.e., have hard emission data
available.
3. All sources are willing to participate in study.
4. Near equal distribution of sources according to uncontrolled
emission levels in excess of 25 ton/yr.
5. Near equal distribution of sources according to size of
operation, but employing no fewer than 50 people.
6. Near equal distribution of sources across 16 defined industrial
categories.
7. Near equal distribution of sources according to ages of process
and control system equipment.
-------
8. Near equal distribution of sources according to types of
corporate structure; i.e., publicly held corporation,
private corporation, publicly owned utility.
9. Consideration for maximizing the variation in:
a. Types of control systems to be studied
b. Levels of required control; i.e., collection efficiency
c. Levels of capital and operating costs in order to achieve
desired control.
10. All sources must have been in compliance at sometime during
the study period.
An inspection of the individual source reports indicated that sufficient
data was available on specific process operations at these facilities to add
an additional 17 point sources, for a total of 186 potential data entries.
A detailed review of each source report was then conducted to ensure a
consistent interpretation of the selection criteria. Where data were unclear
or missing, the contractor responsible for the data was contracted and asked
to quantify the informational gaps, especially with respect to the frequency,
duration and magnitude (as expressed by percent in excess of the allowable
mass emission standard and by absolute tons of emissions per incident) of each
excess emission incident. This reveiw eliminated from consideration, a total
of 67 data points. This 36 percent reduction in available data entries is a
good indication of the difficulty encountered by each contractor in satisfying
the various selection criteria. The specific rationale applied in eliminating
data points was as follows:
• 18 entries (9.7 percent) - Insufficient data available to make
a reasonable estimate on the type and extent of excess
emissions incidents.
• 16 entries (8.6 percent) - Source was continually out of
compliance for the study period.
• 8 entries (4.3 percent) - Source operated uncontrolled without
any pollution control device.
• 8 entries (4.3 percent) - Source had no applicable mass emission
regulation.
• 8 entries (4.3 percent) - Source report dealt with a pollutant
not considered within the scope of the study (3 were NOX sources,
5 were odor sources).
• 3 entries (1.6 percent) - No emission control was required for
the source to meet compliance levels.
-------
• 3 entries (1.6 percent) - Emission control equipment was vented
within the plant and caused no ambient pollutant emission.
• 3 entries (1.6 percent) - Source achieved compliance by switching
to a conforming solvent, no excess emission was possible.
The remaining 119 (64 percent) point sources constituted the sample popu-
lation for this study. For each source, data of both a qualitative and quan-
titative nature, were gathered. The first dealt with actual excess emissions
incidents, their extent and their causes. It is important to note that no
source was considered in the study that did not have a quantitative estimate
made as to the occurrence or nonoccurrence of an excess emission. Where no
such estimate could be made for lack of data upon which to make an estimate,
that source was not included in the study population.
The second type of data dealt with process and control equipment param-
eters including emission and flow rates. These parameters were used to clas-
sify individual sources into applicable categories so as to compare sources on
a relative basis.
The raw data used for all entries in this report has been computerized
and a printout of that data appears in Appendix A of this report. This data
can be used by the reader to correlate potential relationships between source
and control equipment parameters and the occurrence or nonoccurrence of excess
emissions incidents.
In order to ensure the cooperation of the participating sources, it was
necessary to maintain their anonymity throughout the study. We have'accom-
plished this by codifying each source according to the major industrial cate-
gory of which it is a part. We have, therefore, 16 major industrial category
codes:
• Steam Generating Plants (Industrial and Utility)
• Pulp and Paper Mills
• Incinerators
• Petroleum Products and Handling
• Cement Plants
• Asphalt Plants
• Iron and Steel Plants
• Aluminum Plants
• Brass and Bronze Plants
• Petrochemical Plants
• Stone, Clay, Glass and Mineral Plants
• Grain Handling Operations
-------
• Lumber and Wood Products Plants
• Surface Coating Operations
• Food and Drug Products
• Other (Mineral Acid Plants and Fertilizer Plants)
Each facility evaluated within a given industrial category has its own indivi-
dual facility code; e.g., if surveys were conducted at three facilities in
category 1.0, they were codified as 1.1, 1.2, and 1.3. A further delineation
was made if multiple and duplicate sources were evaluated at any one facility;
e.g., four boilers reviewed at facility 1.1 were codified as 1.1.1, 1.1.2,
1.1.3, and 1.1.4. There are no direct references to the geographical location
of a given facility, its size or any process-specific parameters which would,
of themselves, "fingerprint" the facility.
In addition, each source is prefixed by a code letter indicating the con-
tractor responsible for the data development. These prefixes are: "B" for
Booz Allen/PEDCo; "G" for GCA/Technology Division; "P" for Pacific Environ-
mental Services; and "T" for The Research Corporation of New England. As will
be seen, this source coding system will be used throughout this report and
sources will be identified in each major source and process-related parameter
by its source number.
DEFINITION OF KEY TERMS
We have made several references throughout this report to certain terms
as they relate to excess emissions. Since these terms form the basis of our
findings and conclusions in this study, it is important to define them in a
clear, concise manner. We have, therefore, developed the following defini-
tions to provide consistency within our presentations throughout this report:
• Emissions - the absolute quantity of a given pollutant emitted
to the atmosphere (example: pounds, tons).
• Excess emissions - the absolute quantity of a given pollutant
emitted to the atmosphere in excess of a defined regulatory
limit (example: pounds, tons).
• Magnitude - the time rate of an emission of a given pollutant
discharged to the atmosphere during an emissions incident
(example: Ib/hr, ton/yr).
• Frequency - the number of excess emission incidents occurring
per year.
• Duration - the period of time attributable to a given emission
episode (example: minute, day).
• Allowable emissions - the absolute quantity of a given pollutant
that would be emitted to the atmosphere from a pollution control
device operating at the maximum emission rate permitted
by regulation over a specified time period.
-------
• Emission credits - the numerical difference between regulatory
emissions and actual facility emissions (not including excess
emissions) over a specified time period (includes both margin
of safety and overcontrol).
• Parameter - a process or control equipment related variable which
is used to describe the sample population (example: industrial
category, control device size).
• Category - the subdivision of a parameter which is provided to
more accurately describe the parameter (example: scrubber to
describe control device type; 1,000 to 10,000 ton/yr to describe
source size).
• Incident - one distinct and separate occurrence of excess emissions.
• Incident indicators - collective term referring to the frequency,
duration, and magnitude of a distinct incident.
• Causal Code - a convention used to describe the cause of an excess
emissions incident (example: Causal Code 1 - (design) - incident
was caused by an inherent design flaw in the control system).
• Normalized excess - the total annual excess emission from a source
divided by the annualized allowable emission limitation for that
source (includes both margin of safety and overcontrol).
• Normalized credits - the total annual credits for a source divided
by the annualized allowable emissions limitation for that source
(includes doth margin of safety and overcontrol).
We can now proceed to employ any of these definitions, singly or in com-
bination, to consistently characterize the nature of excess emission problems
and their association with the operation and maintenance of air pollution con-
trol systems.
EVALUATION PARAMETERS
All sources were classified according to four major process and control
equipment parameters. This technique was used to more clearly define those
sources which have an excess emission problem so that any subsequent regula-
tory action can be directed at specific problem areas. For the purpose of
this study, these parameters are:
• Industrial classification of the source
• Uncontrolled emission rate of the process studied (ex-
pressed in ton/yr of potential emissions)
• Type of air pollution control equipment used
• Size of the control equipment
-------
Additional information exists in the data base to provide further source clas-
sification according to control equipment age, control equipment capital cost
(expressed in 1977 dollars), and source operating and maintenance costs, if
desired.
To provide a further delineation of the classification scheme, each param-
eter was subdivided into categories. Individual sources were then entered by
their source code into the appropriate category within each parameter. Conse-
quently, each of the four parameters include data for all 119 sources surveyed.
An exact category-by-category summation of incidents data can then be made and
specific industrial and control equipment types and sizes can be identified.
These category breakdowns are as follows:
• Parameter A - Industrial Classification
- Steam Generating Plants (Industrial and Utility)
— Pulp and Paper Mills
— Incinerators
— Petroleum Products and Handling
Cement Plants
— Asphalt Plants
— Iron and Steel Plants
— Aluminum Plants
— Brass and Bronze Plants
— Petrochemical Plants
— Stone, Clay, Glass, and Mineral Plants
— Grain Handling Operations
— Lumber and Wood Products Plants
— Surface Coating Operations
— Food and Drug Products
- Other (including Mineral Acid Plants and Fertilizer Plants)
• Parameter B - Source Size (Uncontrolled Annual Emissions Potential)
— <100 ton/yr
- 100 to 500 ton/yr
- 500 to 1,000 ton/yr
— 1,000 to 10,000 ton/yr
- 10,000 to 100,000 ton/yr
- 100,000 to 500,000 ton/yr
-------
• Parameter C - Control Device Type
Electrostatic Precipitator (ESP)
— Scrubbers
— Fabric Filters (Baghouses)
- Other
The "other" category includes such devices as cyclones, afterburners, vapor
recovery units, absorbers, and mist eliminators.
• Parameter D - Control Device Size
- <10,000 acfm
- 10,000 to 50,000 acfm
- 50,000 to 100,000 acfm
- 100,000 to 500,000 acfm
- 500,000 to 1,000,000 acfm
- >1,000,000 acfm
EXCESS EMISSION INCIDENT CATEGORIZATION
All data collected relative to excess emissions incidents were defined
with respect to five inherent variables: cause of the incident, annual fre-
quency of occurrence, duration in hours of the individual incident, magnitude
of the excess as expressed by percent in excess of the applicable regulatory
limit and magnitude of the excess as expressed by absolute tons of pollutant
released per incident. Subdivisions were provided in the data array for each
category to isolate the specific cause of the excess emissions incident and
the frequency, duration and magnitude of that incident. This bookkeeping
system provided a means of tabulating average values for incident causal
factors, and the frequency, duration and magnitude of all incidents for any
category or an entire parameter, taken as a whole. In addition, it facili-
tated comparison between causal codes for individual categories within each
parameter.
The cause of each incident was defined by one of the following causal
codes:
• Code 1 - Design limitations of the control system. Poor
design relative to the size, materials of construction,
application, etc., was directly responsible for the
incident.
• Code 2 - Process. A change in the process operation or a
process upset which directly leads to an increase emission;
i.e., startup/shutdown, change in feed material, etc.
-------
• Code 3 - Control equipment. A malfunction or breakdown of
the control device or inadequate operation and maintenance
of the device which leads to an excess emission; i.e., loss
of a precipitator section, bag failure in a fabric filter,
pump failure in a scrubber, etc.
• Code 4 - Unforeseen occurrence. Acts that are beyond the
control of the equipment operator and lead to an excess
emission (i.e., power failure, natural gas shutoff), or a
mechanical, electrical failure that could not be foreseen
or prevented by normal maintenance practices.
This further refinement of the data allows us to attribute the reason for
any excess emissions incident to a specific cause. Any regulatory action that
is aimed at reducing the excess emission problem can then be directed to a
specific problem area; i.e., design problems, process problems, etc. Indivi-
dual excess emissions data was first entered by its specific causal code into
the applicable category data base. The summation of all four causal codes was
then tabulated for each source. The results of this tabulation for our four
selected parameters are found in Tables 2 through 5. It should be noted that
when summing individual incident data, weighted averages were developed for
the duration and magnitude of occurrences. Frequency sums are a straight
numerical average.
While most incidents clearly fell into one of these areas, some incidents
could be attributed to more than one cause and required the individual con-
tractor to select the single most relevant cause based on the above definitions
and his knowledge of the specific conditions at the facility.
The requirement that all excess emissions incidents be quantified by fre-
quency, duration and magnitude values was the single most difficult task faced
by the individual contractors. Every effort was made to obtain quality data
during the source visits. Operating and maintenance records were Inspected,
when available, and qualified source personnel Including maintenance foreman
and plant managers were Interview In depth. Hard emission data regarding
potential emissions were back calculated from information gathered during these
on-site interviews. Sound engineering judgement and objective analysis were
used in all emission calculations.
This study, therefore, reflects as accurately as possible the real world
situation. The frequency, duration and magnitude of excess emissions Incidents
cited in this study should be considered a reasonably accurate portrayal of
malfunctions that occur in industry today. Since not all excess emissions
incidents are noted in plant logs, and maintenance personnel tend to down
play the extent and seriousness of any equipment outage, this study may under-
estimate the extent of control equipment malfunctions. However, inasmuch as
great objectivity was used in assembling this data and the information included
was both complete and documented, this report does characterize the excess emis-
sion problem that exists today.
Individual Causal Code averages were computed using only those sources
which reported problems in that Causal Code. This procedure maximized the
impact of the computed averages and allowed us to address the severity of the
individual problems, when they existed. The number of sources which contri-
buted to each Causal Code average is nresenteH in the aoprooriate data summaries.
-------
TABLE 2. PARAMETER A - INDUSTRIAL CLASSIFICATION
DATA SUMMARY.
Category
Frequency
_ . Magnitude Number
Duration ,„ . c
( . (% above of
v ' allowable) samples
Causal Code 1 (design)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Industrial and utility boilers
Pulp and paper mills
Incinerators
Petroleum products and handling
Cement plants
Asphalt plants
Iron and steel plants
Aluminum plants
Brass and bronze plants
Petrochemical plants
Stone, clay and glass plants
Grain handling operations
Lumber and wood product plants
Surface coating operations
Food and drug plants
Other
Average
0.8
—
—
1.2
—
—
13.0
—
—
1.0
0.5
36.3
—
—
—
—
9.6
2381.7
—
—
1527.4
—
—
8.6
—
—
6000.0
264.0
18.4
—
—
—
—
158.2
985
—
—
729
—
—
139
—
—
83
108
1346
—
—
—
—
1003
2
0
0
3
0
0
2
0
0
1
1
2
0
0
0
0
Total =11
Causal Code 2 (process)
1
2
3
4
5
6
7
L
9
10
11
12
13
14
15
16
Industrial and utility boilers
Pulp and paper mills
Incinerators
Petroleum products and handling
Cement plants
Asphalt plants
Iron and steel plants
Aluminum plants
Brass and bronze plants
Petrochemical plants
Stone, clay and glass plants
Grain handling operations
Lumber and wood product plants
Surface coating operations
Food and drug plants
Other
Average
148.4
116.0
7.3
16.1
30.6
10.0
374.0
3.5
1.0
0.8
—
—
—
5.4
—
30.0
65.9
3.1
5.0-
0.2
8.5
46.5
0.5
0.4
0.1
1.0
520.0
—
—
—
85.3
—
3.8
6.6
162
180
542
88
3949
76
148
50
1150
255
—
—
—
679
—
122
421
6
2
2
4
5
2
2
1
1
2
0
0
0
4
0
3
Total = 34
(continued)
10
-------
TABLE 2 (continued)
Category Frequency
^ . Magnitude Number
Duration /. . e
x, x (% above of
Ur; allowable) samples
Causal Code 3 (control O&M)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Industrial and utility boilers
Pulp and paper mills
Incinerators
Petroleum products and handling
Cement plants
Asphalt plants
Iron and steel plants
Aluminum plants
Brass and bronze plants
Petrochemical plants
Stone, clay and glass plants
Grain handling operations
Lumber and wood product plants
Surface coating operations
Food and drug plants
Other
Average
—
—
2.8
5.9
20.3
14.8
6.9
4.0
2.0
1.0
6.2
5.1
28.8
2.5
1.5
5,3
8.6
—
—
17.1
168.3
6.1
3.9
11.4
24.0
2.0
2160.0
67.0
4.9
26.6
141.1
233.0
18.6
45.9
—
—
587
174
521
630
262
67
1150
50
297
361
34
3182
25
123
399
0
0
2
8
5
3
8
1
1
1
5
2
3
5
1
4
Total = 49
Causal Code 4 (unforeseen)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Industrial utility boilers
Pulp and paper mills
Incinerators
Petroleum products and handling
Cement plants
Asphalt plants
Iron and steel plants
Aluminum plants
Brass and bronze plants
Petrochemical plants
Stone, clay and glass plants
Grain handling operations
Lumber and wood product plants
Surface coating operations
Food and drug plants
Other
Average
0.5
—
0.5
1.4
4.3
—
3.3
—
—
—
3.5
0.1
—
0.6
—
0.8
1.9
336.0
—
4.0
160.3
47.1
—
15.1
—
—
—
8.7
168.0
—
672.6
—
115.2
83.9
12
—
2998
548
129
—
290
—
—
—
68
274
^~
1533
—
140
317
1
0
1
5
3
0
4
0
0
0
2
1
0
5
0
3
Total = 25
(continued)
11
-------
TABLE 2 (continued)
Category
V
Z-iall Causal Codes
1 Industrial and utility boilers
2 Pulp and paper mills
3 Incinerators
4 Petroleum products and handling
5 Cement plants
6 Asphalt plants
7 Iron and steel plants
8 Aluminum plants
9 Brass and bronze plants
10 Petrochemical plants
11 Stone, clay and glass plants
12 Grain handling operations
13 Lumber and wood product plants
14 Surface coating operations
15 Food and drug plants
16 Other
Average
Frequency
89.3
38.5
3.4
9.4
29.7
6.5
84.2
1.9
0.6
0.6
3.9
16.5
28.8
2.9
0.5
18.9
23.7
Duration
(hr)
7.2
5.0
45.9
122.6
31.2
2.8
1.6
12.8
1.7
2554.0
42.2
16.8
26.6
154.1
233.0
9.0
19.3
Magnitude
(% above
allowable)
163
180
614
167
2463
458
158
59
1150
147
253
1224
34
1566
25
123
438 Total
Number
of
samples
10
6
6
13
9
10
10
4
5
6
10
5
3
13
3
6
= 119
12
-------
TABLE 3. PARAMETER B - SOURCE SIZE
DATA SUMMARY.
Category _ Duration Ma8njtude Numb.er
(TPY x io3) Frequency (% above of
allowable) samples
Causal Code
<0.1
0.1 - 0.5
0.5 - 1.0
1.0 - 10
10 - 100
100 - 500
Average
Causal Code
<0.1
0.1 - 0.5
0.5 - 1.0
1.0 - 10
10 - 100
100 - 500
Average
Causal Code
<0.1
0.1 - 0.5
0.5 - 1.0
1.0-10
10 - 100
100 - 500
Average
Causal Code
<0.1
0.1 - 0.5
0.5 - 1.0
1.0 - 10
10 - 100
100 - 500
Average
1 (design)
0.5 264.0
1.5 2920.0
0.8 4912.7
0.8 400.8
24.8 37.5
— —
9.6 158.2
2 (process)
10.3 23.9
2.1 6.0
165.8 0.6
71.9 7.65
86.0 4.5
3.0 1106.0
65.9 6.6
3 (control O&M)
7.4 106.6
1.9 41.5
10.9 30.0
12.1 30.6
8.8 7.3
10.8 11.2
8.6 45.9
4 (unforeseen)
1.5 26.0
1.7 246.0
1.3 49.2
1.5 95.0
4.0 21.5
5.0 45.6
1.9 83.9
108
900
994
6,383
873
—
1,003
520
147
161
154
164
96,588
421
162
3,023
133
322
252
1,631
399
523
413
391
375
138
15
317
1
1
2
3
4
0
Total = 11
3
7
6
13
3
2
Total = 34
13
8
8
14
4
2
Total = 49
3
6
5
8
2
1
Total = 25
(continued)
13
-------
TABLE 3 (continued)
Category
(TPY x io3)
Frequency
Duration
(hr)
Magnitude
(% above
allowable)
Number
of
samples
Z*tf all Causal Codes
<0.1
0.1 - 0.5
0.5 - 1.0
1.0 - 10
10 - 100
100 - 500
Average
5.5
1.3
68.1
37.3
36.4
5.4
23.7
85.0
181.6
11.4
11.5
13.2
218.0
19.3
225
1,292
161
196
571
18,913
438 Total
24
31
17
30
11
6
= 119
14
-------
TABLE 4. PARAMETER C - CONTROL DEVICE TYPE
DATA SUMMARY.
Category Frequency
Causal Code 1
ESP
Scrubber
Baghouse
Other
Average
Causal Code 2
ESP
Scrubber
Baghouse
Other
Average
Causal Code 3
ESP
Scrubber
Baghouse
Other
Average
Causal Code 4
ESP
Scrubber
Baghouse
Other
Average
2^ all Causal
ESP
Scrubber
Baghouse
Other
Average
0.8
0.5
24.6
1.2
9.6
95.9
17.96
10.7
123.4
65.9
14.3
4.2
10.6
6.8
8.6
2.1
1.3
3.6
1.0
1.9
Codes
46.6
8.4
8.4
46.0
23.7
Duration ^tude Nu"*er
. . (% above of
inr; allowable) samples
1130.4
2765.0
15.8
3097.7
158.3
11.7
8.9
0.18
2.4
6.6
6.5
119.2
13.3
127.7
45.9
62.2
245.7
16.1
175.9
83.9
14.5
40.8
12.8
20.3
19.3
170
2817
1941
667
1003
815
160
88
180
421
438
348
312
535
399
162
672
104
746
317
751
212
518
208
438
3
1
4
3
Total = 11
9
13
3
9
Total = 34
10
14
14
11
Total = 49
8
6
4
7
Total = 25
22
36
35
26
Total = 119
15
-------
TABLE 5. PARAMETER D - CONTROL DEVICE SIZE
DATA SUMMARY.
Category
(acfm x 103)
Causal Code 1
<10
10 - 50
50 - 100
100 - 500
500 - 1000
>1000
Average
Causal Code 2
<10
10 - 50
50 - 100
100 - 500
500 - 1000
>1000
Average
Causal Code 3
<10
10 - 50
50 - 100
100 - 500
500 - 1000
>1000
Average
Causal Code 4
<10
10 - 50
50 - 100
100 - 500
500 - 1000
>1000
Average
Magnitude Number
Frequency Durat"n (% above of
inr; allowable) samples
(design)
1.2
0.5
—
31.0
1.0
6.0
9.6
(process)
14.5
22.6
17.7
146.0
60.0
—
65.9
(control
6.3
8.5
6.0
11.3
17.3
4.0
8.6
2097.7
1033.7
—
20.2
2190.0
24.0
158.2
4.5
10.6
10.9
5.8
4.0
—
6.6
O&M)
143.0
38.4
8.5
22.8
6.2
15.0
45.9
667
10,814
—
1,893
69
100
1,003
109
377
54
482
200
—
421
227
569
218
412
181
100
399
3
3
—
3
1
1
Total = 11
3
13
5
12
1
—
Total = 34
11
17
6
12
2
1
Total = 49
(unforeseen)
2.8
1.1
0.7
2.0
4.0
5.0
1.9
229.8
34.3
61.0
58.5
48.0
20.0
83.9
390
674
664
144
200
100
317
4
8
3
8
1
1
Total = 25
(continued)
16
-------
TABLE 5 (continued)
Category
(acfm x 103)
Frequency
Duration
(hr)
Magnitude
(% above
allowable)
Number
of
sanples
all Causal Codes
4.1
9.2
10 - 50
50 - 100
100 - 500
500 - 1000
>1000
Average
14,
86,
16,
15.0
23.7
184.3
24.4
12.0
8,
28,
20,
.1
.5
.3
214
406
111
468
192
100
31
49
9
23
6
1
19.3
438 Total =119
17
-------
Averages for the summation of all Causal Codes were computed using data
from the entire 119 source data base. Sources reporting no excess emissions
incidents were tabulated in these averages. In all, there were 41 sources
(34 percent) which reported no dlscernable excess emissions. Inclusion of
these sources tended to lower average frequency, annual excess emissions
and normalized excess emissions averages.
Total annual excess emissions attributable to all causes are listed in
Table 6, as are the total annual credits for each source. This data was ex-
tracted from the initial contractor's reports. The annualized allowable emis-
sions for each source is also presented. A straightforward presentation of
the annual excess emissions and credits data will result in averages for these
items which are unduly influenced by the larger sources which tend to have
greater absolute excess emissions and emission credits. Both total annual
excess and annual credits were divided by their respective annualized allow-
able emissions to derive "normalized excess" and "normalized credits." This
"normalization" technique serves to eliminate large source bias, as these
sources will typically have proportionally larger allowable emissions in
conjunction with the greater annual excess and greater credits, when compared
to the entire sample population. These normalized factors will be dimension-
less and will represent the percent of allowable of each item (excess and
credits); i.e., a normalized excess of 0.09 indicates that the annual excess
is 9 percent of that source's allowable emission rate. Numberical averages
of normalized excess and normalized credits were computed for each category.
The significance of this data will be discussed in a subsequent section of
this report.
RANKING OF SOURCE EXCESS EMISSIONS INCIDENTS
Once average incident indicator values (i.e., frequency, duration and
magnitude) were computed and correlated by Causal Code for each category, a
display of this data was made for each of the four parameters. The technique
used for this display is defined as a "rank ordering" of the frequency, dura-
tion and magnitude averages. In this technique, all categories within a
parameter are arranged in descending order with respect to the average fre-
quency, duration and magnitude of their incidents. This convention gives a
number one rank to the category with the highest frequency, longest duration
or largest magnitude percentage. The rank ordering was computed for each
individual Causal Code and for normalized excess and normalized credits,
within each parameter. Rank ordering allows us to quickly recognize problem
categories with respect to frequency, duration and magnitude of excess emis-
sions incidents, and will narrow the area of concern with regards to regula-
tory response to these incidents. By rank ordering, individual Causal Codes
as well as the summation of all causes we can detect the specific reasons
for emissions incidents for each category. Rank ordering of normalized
excess is accomplished in a slightly different fashion. For normalized
excess, the category with the largest decimal (largest percent of allow-
ables) is again ranked number one. Normalized credits, however, are ranked
with the smallest decimal (smallest percent of allowable) as number one.
18
-------
TABLE 6. EXCESS/CREDITS
DATA SUMMARY.
Total
Category
*»•-•
Parameter A
1 Industrial and utility bollere
2 Pulp and paper mills
3 Incineretom
4 Petroleum products end handling
S Cement plant*
6 Aaphalt plant*
7 Iron and eteel plants
8 Aluminum plante
9 Braee and bronia plant*
10 Petrochemical plants
11 Stone, clay and glee* plant*
12 Grain handling operation*
13 Lumbar and wood plant*
14 Surface coating operetlon*
15 Food and drug product*
16 Other
Parameter B
(TPY « 10s)
<0.1
0.1 - 0.5
0.5 - 1.0
i n - 10
I .U *V
10 - 100
100 - 500
72.1
0.9
0.2
45.4
38.0
2.5
7.2
0.02
0.2
11.5
0.7
123.5
0.4
10.1
0.1
4.7
0.43
9.0
27.5
34.6
64.5
6.5
165.9
275.0
26.4
196.9
123.5
14.3
145.9
16.3
18.9
25*i.5
33.5
180.2
6.7
68.3
8.1
159.6
3.77
23.9
87.6
238.6
233.8
207.0
2/9. i
iC0.9
'.0.2
145.3
121.2
64.4
221.1
2?.!
36:4
178.5
SJ.9
222.4
23.4
87.0
16. 1
258.9
8.14
50.3
159.5
39.7
416.6
376.8
Parameter C
ESP
Scrubber
Baghouee
Other
10 - 50
50 - 100
100 - 500
500 - 1000
>1000
Average for All Source*
42.2
17.2
20.0
16.6
14.1
10.0
3.28
69.1
19.68
15.0
22.5
203.5
81.1
83.1
111.5
16.2
69.3
77.5
189.0
317.6
1262.0
110.9
352.2
U6.8
140.0
197.2
Sh.5
122.J
162.9
391.9
508.9
1723. S
207.8
Normal Izrtl Normalised
. >i •••-- i rcdliH
(Z ... (I of
ollovablex) allowable*)
0.48
o.ou
0.020
0.(>2
O.IS
0.00071
0. IS
o.ocj'>
0.0019
0.75
0.0/9
0.82
0.0026
0.44
0.0012
0.014
0.11
0.35
0.18
0.28
0.26
0.15
0.24
0.27
0.14
0.33
O.i'i
0.14
0.024
0.38
0.076
O.OOS7
0.48
0.47
0.47
0.35
n.37
0.58
0.48
0.57
0.59
n.64
0.68
0.81
0.36
0.50
0.40
0.65
0.55
0.47
0.58
0.56
0.43
0.44
0.51
0.48
0.65
0.37
0.52
0.51
0.41
0.48
0.60
0.73
0.22
0.50
19
-------
This change ensures that the category with the smallest credits is ranked first
and maintains our established convention, that the smaller the rank, the greater
the problem, as sources with fewer credits operate closer to the allowable emis-
sion rate and are potentially a greater problem.
INTEGRATION OF RANKINGS
After individual category ranks have been established for frequency, dur-
ation and magnitude of excess emissions incidents, we narrowed our scope of
investigation to derive an overall worst category with respect to all factors.
Again a "ranking" procedure was used. For this we "integrated" the individual
rankings for frequency, duration and magnitude into one combined ranking of
categories. Once more we maintained the convention that the lower the rank
number, the greater the problem of excess emissions and continued source com-
pliance. While several techniques can be used for this integration, we have
chosen to arithmetically add the individual rankings of frequency, duration
and magnitude for each category, and then rank all categories by this overall
number. This procedure was used for all four parameters, and for all Causal
Codes within each parameter. While it may be argued that a greater weight
should be placed on the number of incidents, their duration or their magnitude,
we have decided, for this study, to weight all factors equally. The resultant
tables therefore, present a systematic ranking of each parameter due to inci-
dents within a specific Causal Code and indicate which categories appear to
have problems attributable to a specific cause.
A slightly different approach was used to integrate normalized credits
and normalized excess. As previously stated, the normalized excess were ranked
for each parameter by listing the categories in descending order with respect
to their numbers (largest value is first). Normalized credits were ranked
by listing the categories in ascending order with respect to their numbers
(smallest value is first). The numerical difference between these two normalized
numbers was then calculated (credits-excess) and these resultant numbers were
then ranked, again with the smallest number ranked first. The significance of
the difference between normalized values is that it points out the degree to
which a category has greater excess than credits, or vice versa. A negative
number for the computed difference indicated that this category had more excess
emissions than credits, and overall was emitting more pollutant into the ambient
air than is allowed by its applicable regulatory limit computed and summed on an
annual basis. As the difference becomes increasingly positive, the category has
more annual credits than annual excesses and operates further below the allowable
limits. Use of normalized values allows us to express the degree to which credits
exceed excesses in terms of percent of allowable emissions. Hence, when a source
reports a difference of +0.40 it means that emission credits completely offset
emission excesses and that the source operated with an annual credit surplus of
40 percent of its regulatory emission limit computed and summed on an annual
basis. Stated another way, the source emitted only 60 percent of the emissions
it was allowed to emit. Knowledge of this net difference between credits and
excess can potentially provide a useful planning tool when developing regula-
tory control concepts.
20
-------
To summarize the results of the parameter integrations, we have Included
tables of "problem area profiles." These tables, one for each Causal Code,
summarize the integration results on one page and provide the reader, at one
glance, the opportunity to visualize the extent of the excess emission problem.
Since certain control device types are not associated with some industries
(i.e., grain handling facilities typically do not use scrubbers) these profiles
should not be used out of context. They are included to essentially summarize
our findings with respect to those categories which may have greater problems
in maintaining continued source compliance.
21
-------
SECTION 2
DISCUSSION OF DATA
PARAMETER A. EXCESS EMISSIONS AS A FUNCTION OF INDUSTRIAL CATEGORY
The following subsections summarize the compilation of data supplied by
all four contractors relative to each of the major industrial categories
investigated. To maintain consistency within each subsection, our presenta-
tions contain the following elements:
• General comparative description of sources surveyed within the
category
• Discussion of source representativeness
• General comparative discussion of the nature of excess emissions
including average frequency, duration, and magnitude (percent
above allowable)
• Comparison of individual category averages to average
frequency, duration and magnitude (percent above
allowable) of all categories combined.
The information provided in some subsections is more thorough than others
because of the number of sources investigated in a given category and/or the
availability of data specific to a given facility. Causes of excess emission
incidents have been grouped into one of four causal factors. For comparison
purposes, we defined these causal factors according to whether they were
design, process, control device (operation and maintenance) related or unfore-
seen. Tables 2, 3, 4, 5 and 6 present a comparison of causal factor values
recorded for each category and the overall average for the 119 sources
evaluated.
Excess Emissions From Steam Generating Plants
On-site plant surveys were conducted at seven steam generating plants.
Since data for four boilers were available at one of the facilities, our total
population included 10 separate units. The size of the 10 units surveyed
ranged from 173 to 28,400 tons per year of uncontrolled emissions. Emissions
from 6 of the 10 plants were controlled by wet scrubbers with the remaining
4 being served by electrostatic precipitators or "other" devices. The size
of the control devices ranged from 25,000 to 550,000 acfm.
Most of the 10 units investigated were coal fired, with some wood fired.
A survey of utility and industrial steam generating units currently in use in
22
-------
the United States shows that there are 3,677 coal-fired units operating with
an average uncontrolled particulate emission rate of 12,771 tons per year.
Of the 10 steam units investigated, 9 (90 percent) reported excess emission
incidents, whereas for the sample population (119 sources), 78 (66 percent) had
problems, indicating that excess emission incidents are more common for steam
generating plants. For all causal codes combined, steam generating plants have
far more incidents per year than the average of the sample population with
almost four times as many. However, the duration of the incidents are shorter
(7 hours versus 19) and the relative magnitudes are less (163 versus 438) than
the average of the sample population. Although the average steam generating
plant has 3.3 times as many excess emissions as the average of the sample pop-
ulation their credits are also greater. Normalization of the data show that
excesses were completely offset by the credits resulting in no net gain or loss.
A total of 892.5 incidents were reported by the sources of this category.
Of these, 99.8 percent were process related, with the remaining 0.2 percent due
to design flaws and unforeseen events. It should be noted that of the 890
process related incidents found, 882 occurred at one plant which had a regularly
reoccurring problem.
No control device operation and maintenance related incidents were reported.
In addition, no cases of multiple causal factors occurred. That is, of- the
nine sources reporting problems none were due to more than one causal factor;
two (22 percent) had design problems, six (67 percent) were found to have pro-
cess related incidents and one (10 percent) reported an unforeseen event.
About twice as many steam generating units reported design related
incidents as the average of the sample population. Even though a greater num-
ber reported problems, each steam generating unit had fewer incidents per year
than the average and each incident had a higher relative magnitude and last
almost 15 times longer.
With respect to causal code 2, a larger number of steam generating plants
reported process (boiler) related problems than the average of the sample pop-
ulation. Six (60 percent) of the 10 sources identified reported process related
incidents. In terms of the average for the sample population, process problems
occur twice as often at steam generating plants. Although the frequency is
higher, the magnitude of these incidents are less than the average of all
sources surveyed.
Even though 41 percent of all the sources surveyed had operational and
maintenance problems, no problems due to this causal factor were reported for
steam generating plants. Only 1 plant of the 10 (10 percent) investigated had
an episode related to an unforeseen event.
Typically, the cyclical nature of the boiler operation is reflected in the
high process related frequency of upsets. Startups and shutdowns of boilers,
for example, can occur daily and each can result in an excess emission incident.
23
-------
Excess Emissions From Pulp and Paper Mills
Six of the 119 facilities surveyed were pulp and paper recovery boilers.
The size of these boilers ranged from 15,000 to 343,000 tons per year of uncon-
trolled emissions. Emissions from three of the boilers were controlled by
electrostatic precipitators, wet scrubbers were used on two of the units with
emissions from the sixth boiler controlled by an absorbing tower with Brink
mist eliminators. The control devices ranged from 23,000 to 540,000 acfm. The
results of a literature search conducted to determine source representativeness
found that there are approximately 120 pulp and paper recovery furnaces in the
country with the average unit having an uncontrolled emission rate of 24,257
tons per year.
Of the six boilers investigated, only two reported incidents of excess
emissions which were due to process related problems. There were no reported
incidents resulting from design, operation and maintenance or unforeseen
factors.
The number of pulp and paper mill sources reporting process related inci-
dents is slightly higher than the average of the sample population for causal
code 2 problems (33 versus 29 percent). The frequency with which these process
related incidents occurred was almost twice the average of the sample population
but the relative magnitude of the incidents were lower and the duration was
less.
Annual excess emissions from pulp and paper boilers were found to be 0.87
tons compared with 22 tons for all plants combined. Annual credits for this
industry were substantial, averaging 275 tons. This is more than twice the
average of the sample population. Normalization of these data elements show,
that on a yearly basis, when excesses were totally offset by credits, pulp and
paper boilers were found to operate with a credit surplus of 46 percent of the
allowable emission limitation.
Excess Emissions From Incinerators
Onsite surveys were conducted at six incinerator facilities. The facil-
ities were comprised of municipal, refuse, and industrial type incinerators.
The size of the sources investigated ranged from less than 100 to 2,260 tons
per year of uncontrolled emissions; two of the six incinerators fell into the
less than 100 tons per year category. With respect to control equipment, five
of the incinerators were served by wet scrubbers with the sixth unit employing
an electrostatic precipitator. Scrubber sizes ranged from 1,200 to 30,800 acfm.
The electrostatic precipitator operated at 750,000 acfm, an unusually high
flow rate for an incinerator with an annual uncontrolled emission rate of 522
tons.
Four of the six sources reported excess emissions incidents, with one of
the four identifying multiple causal factors which were operational and main-
tenance related and unforeseen. Based on the average of the sample population,
incinerator sources appear to have about the same number of incidents. None
of the sources investigated reported design problems and in only one (17 percent)
24
-------
case was an unforeseen related problem found. Two sources (33 percent) reported
process related incidents and two (33 percent) operation and maintenance inci-
dents were found at two other incinerators.
The frequency with which incidents occur at incinerators was far less than
the average of the sample population (3.4 versus 23.7), although the duration
and relative magnitude of each incident was greater. However, because the
incidents occur so infrequently, annual excess emissions for this industrial
category were only 0.22 tons, one one-hundredth of the 22 tons reported as the
average of the sample population sources. Credits for sources in this category
were well below the average sources of the sample population (26.4 versus
110.9 tons). Since the credits far exceeded excesses, the incinerators were
found to operate at only 53 percent of their allowable limit with the remaining
47 percent being credits. A total of 20.5 incidents were reported for all
causal cases. Of these, 70.7 percent were process related, 26.8 percent resulted
from control O&M problems and 2.4 percent were unforeseen.
Even though causal code 2 problems represent the highest number of incidents
occurring per year, incinerators have fewer process related incidents than other
plants. The duration of each incident is also less, even though the relative
magnitudes are higher. The average annual excess emissions due to process
related incidents for all sources are 100 times greater than the excess from
incinerators. This wide difference results from the fact that incinerator
process related incidents last only 10 minutes, and occur only seven times a
year.
Incidents related to control device operation and maintenance problems
occur at incinerators three times annually. This is less than half as many
times as the average of the sample population. The duration and relative
magnitude of these incinerator incidents are lower and higher, respectively,
than the average of the sample population. Average annual excess emissions
from incinerators due to O&M incidents are far less (0.39 tons) than the aver-
age of the sample population (10.6 tons).
Unforeseen incidents occur about one-third as often as incinerators as the
average of the sample population. The relative magnitude of the incidents is
high, but they last on the average only 4 hours causing a small amount (0.39
tons) of excess emissions. This is much less than the average of the sample
population (9.2 tons).
Excess Emissions From Petroleum Products and Handling
Investigations were made at 13 facilities falling into the petroleum
products and handling category. The sources surveyed ranged from gasoline
tank farms and loading racks to fluid catalytic cracking units. Source sizes
ranged from less than 100 to 2,160 tons per year of uncontrolled emissions.
Only one source was rated at less than 100 tons per year. Control systems
investigated include three electrostatic precipitators, three wet scrubbers and
seven "other" types including vapor recovery systems, afterburners, etc. Con-
trol equipment sizes ranged from 375 to 410,000 acfm.
As a whole, petroleum products and handling facilities had fewer inci-
dents (2.2) during the year than the average of the sample population (23.7)
25
-------
The average duration and relative magnitude of the incidents were found to last
123 hours at a rate 1.7 times the allowable standard, respectively. Annual
excess emissions (45 tons) were twofold higher than the average of the sample
population (22.5 tons). However, because average annualized allowable emission
limit is high, annual credits were also high (197 tons), greater than the
average of the sample population (110 tons). Consequently, sources surveyed
in this category were found tc operate with 37 percent of the allowables as
credits.
Ninety-two percent of the petroleum products and handling plants inves-
tigated were found to have at least one excess emissions incident. Only 1 of
the 13 sources studies did not report any incidents of excess emissions. Six
of the twelve sources with problems reported incidents due to more than one
causal factor. One source, a fluid catalytic cracking unit controlled by an
electrostatic precipitator, was found to have an excess incident related to
each of the four causal factors reviewed. Most of the sources' problems,
however, were related to O&M. Total annual excess emissions due to the O&M
incidents were 27.1 tons, well above the average (10.6 tons) for sources having
similar problems.
Of the .122.5 incidents reported for all causal codes, 52.7 percent resulted
from process upsets, 38.8 percent were due to control O&M problems, 5.7 were
unforeseen, and 2.9 resulted from design flaws.
Design related incidents were reported for 3 of the 13 (23 percent) sources
investigated, indicating that design problems occur at more sources in the
petroleum products and storage industry than the average of the sample popula-
tion (9.2 percent). When a design incident did occur, it lasted for a long time
(1,527 hours) and was of a high relative of magnitude (729 percent above the
allowable limit) resulting in a large amount of excess emissions (81 tons).
Even though more sources in this category were found to have incidents of
excess emissions than for other industries, the annual frequency of those
incidents was less than the average of the sample population (1.2 versus 9.5).
This accounts for the fact that annual excess emissions (97 tons) related to
design problems for petroleum products and handling plants are less than the
average of the sample population (118 tons) for this causal factor.
Although the number of sources in this category appear to have the same
percentage of process related problems as other industries, excess emission
incidents occur less frequently (16.1 versus 65.8). The duration of the
incidents is similar to the average of the sample population but the relative
magnitude is roughly one-fifth the average for all sources. Excess emissions
resulting from process problems at sources in this industry and the average
of the sample population are similar.
Control operation and maintenance incidents occur less frequently at
the sources of this category but they tend to last almost four times as long
with the relative magnitude of about half that of the average of the sample
population. Annual excess emissions are more than twice as high as that of
the average of the sample population.
26
-------
Unforeseen problems at sources in this category appear to result in
slightly more annual excess emissions (14 tons) than for the overall average
(9.2 tons) for this causal factor. This is due to the fact that when an
incident occurs at the facility, it lasts twice as long (160 versus 84 hours).
Excess Emissions From Cement Plants
On-site surveys were conducted at seven cement manufacturing plants, all
focusing on rotary kilns. Since data for three rotary kilns were available
at one of the facilities, our total population included nine separate units.
The kilns ranged in size from 792 to 200,000 tons per year of uncontrolled
particulate matter. Five of the units were controlled by electrostatic
precipitators, with the remaining four employing baghouse controls. Control
device sizes ranged from 100,000 to 525,000 acfm.
Based on a literature search, approximately 434 cement kilns are operated
in the United States. The average kiln size is 21,686 tons per /ear of uncon-
trolled emissions.
Of the kilns investigated, seven of the nine (77 percent) reported in-
cidents, showing that the probability of a cement plant having a problem is
greater than that of the sample population (66 percent). Four of the seven
were found to have multiple problems; i.e., incidents resulting from more
than one causal factor. None of the units reported design problems. Five
of the kilns (55 percent) were found to have process related incidents; five
(55 percent) had incidents resulting from O&M problems; and three (33 percent)
had incidents due to unforeseen events. Compared with the average of these
causal factors for all sources investigated, cement plants, specifically
rotary kilns, tend to have more excess emission incidents. A total of 267.5
incidents were reported at the sources surveyed. The breakdown of these
incidents by causal factors shows that 57.2 percent were process related, 37.9
resulted from control O&M problems and 4.9 percent were unforeseen.
Total annual excess emissions (38 tons) from rotary kilns were found to be
greater than of the sample population. In the same trend, the average emission
credits were 124 versus 110 tons. Normalization of these two values show that
credits offset excess emissions resulting in a net surplus of 22 percent of the
average allowable limitation. This 22 percent is slightly lower than the
28 percent surplus found for the average of the sample population.
With respect to process related problems (e.g., change in material feed
rate or temperature), cement plants have fewer incidents per year than the
average of the sample population but the duration and relative magnitude of
each were found to be higher. The higher duration, however, was caused by
one plant which had an incident last three-fourths of the year. Similarly,
the higher reported magnitude was attributed to an incident at a different
plant which lost total control. This resulted in a percent above allowable of
115,900. Annual excess emissions resulting from process related problems are
relatively high (30 tons) compared to the average of the sample population
(18 tons). Since cement plants require a high degree of control, whenever they
encounter a problem, the magnitude of emissions resulting is often very high.
27
-------
Operational and maintenance incidents were found to occur more often
at the rotary kilns that at the average of the sample population (20 versus
9). As expected for these large sources, the relative magnitude of the
incidents was higher than the average of the sample population but the dura-
tion of the incident was shorter. The combination of the larger number and
greater emissions resulted in more O&M related annual excess emissions (32
tons) released from cement plants that from the average of the sample popula-
tion (11 tons).
The number of unforeseen problems occurring at cement plants is almost
2.5 times greater than those for the average of the sample population.
Although they occur more often, the unforeseen incidents do not last as long
nor are they of as high a relative magnitude as the similar related incidents
found for the average of the sample population. Total annual excess emissions
are about the same (9.7 versus 9.2 tons).
Excess Emissions From Asphalt Plants
A total of 10 asphalt plants were investigated. Most of the sources
surveyed focused on rotary dryers. Source sizes ranged from less than 100 to
6,300 tons of uncontrolled particulate matter per year. Only one of the 10
was rated at less than 100 tons. Fabric filters (baghouses) were used to
control emissions at six of the sources, with wet scrubbers serving the
remaining four. The control devices ranged in size from 4,000 to 54,000 acfm.
Due to the ever growing number and construction projects in the country and
the increasing need for asphalt concrete, a reliable number for those plants
currently in operation could not be established. Data was available, however,
to estimate average source size, which would be 3600 tons per year of uncon-
trolled emissions.
The data reported show that 4 of the 10 sources investigated (40 percent)
had an excess emission incident. One of the sources reported incidents related
to both process and O&M problems. Compared to the sample population, asphalt
plants tend to have a fewer number of sources reporting problems. This is one
reason why annual excesses from sources in this category were one-tenth that
for the average of the sample population. When normalized and excesses are
offset, the credit surplus for sources in this industry were found to be 58
percent of the allowable limitation.
No design or unforeseen related incidents were reported for the asphalt
sources surveyed. With respect to process problems, 2 of the 10 sources had
excess emission incidents. The 20 percent reporting rate identified here
was slightly below the 29 percent rate for the sample population. Compared
with the average of the sample population, the frequency with which these
occur is less for rotary dryers (10 versus 66), as is the duration and
relative magnitude. This resulted, as would be expected, in lower annual
excesses due to process related problems for asphalt sources than other
industries.
28
-------
Control device operation and maintenance related problems were found to
occur more frequently at asphalt plants than for the average of the sample
population (14.8 versus 8.6). The duration of the incidents were shorter
than the average, but their relative magnitude was higher. These offsetting
findings resulted in similar annual excesses due to O&M problems for asphalt
sources and the sample population (8.2 and 10.6 tons, respectively).
Excess Emissions From Iron and Steel Plants
For this industrial category, 10 sources were included in the plant
evaluations. The sources surveyed were principally electric arc furnaces.
and gray iron cupolas. Source sizes ranged from less than 100 to 21,600 tons
per year of uncontrolled particulate matter, with 3 of the 10 rated at under
100 tons. The most common control devices used were wet scrubbers, operated
by 7 of the 10 sources. Baghouses were used by two sources with one facility
controlled by an electrostatic precipitator. Control device sizes ranged
from 8,200 to 1,700,000 acfm. The results of a literature survey conducted
to determine source representatives show that there are approximately 121
electric are shops and 1493 gray iron cupolas in the country today. The
average sizes of these two industrial sources, in terms of uncontrolled
emissions, are 3,760 and 120 tons per year.
For the 10 sources investigated, 9 reported excess emission incidents.
This finding implies that a greater number of iron and steel plants have more
problems controlling emissions than the average of the sample population
(90 versus 66 percent). Five of the nine sources reporting incidents were
found to have multiple problems.
In addition to having more sources reporting problems than the sample
population, incidents associated with iron and steel plants occur more fre-
quently (84 versus 24). However, because the duration and magnitude are less
than the average, annual excess emissions are also less. When the credits
and excess data are normalized and excesses are offset, iron and steel sources
operate within 68 percent of the allowable emission limitation. This means
that annual credits are 32 percent of their allowable limit.
Of the total 842 incidents reported, 88.8 percent resulted from process
upsets or changes, 6.5 were due to control O&M problems, 3.1 percent resulted
from design flaws and 1.5 percent were unforeseen. It should be noted that
93.5 percent (700 incidents) of the process related incidents reported
occurred at one source.
Design related problems were found at 2 of the 10 sources surveyed. This
is almost double the average number of sources which had similar problems for
of all 119 sources. The incidents (13 per year) occur more often than for the
average of the sample population (9.5 per year), although the duration and
relative magnitude are substantially less. The end result being that excesses
from iron and steel sources due to design-related problems are much less
(8.6 tons) than those (118 tons) for the average of the sample population.
29
-------
Incidents resulting f.rom process upsets or changes were found at 20 per-
cent of the iron and steel sources surveyed. Compared with the average of the
sample population (29 percent), iron and steel plants have fewer number of
sources reporting process-related problems. The frequency of these incidents,
however, are greater than the average of the sample population (374 versus 66).
It should be noted that the high average frequency reported for this causal
code resulted from one plant which was found to have 700 incidents a year.
The duration and relative magnitude of the process related incidents were
found to be less than those for the average of the sample population resulting
in lower than average annual excesses.
O&M related problems were found to cause incidents in 8 of the 10 sources
investigated. This is almost twice the number reported for the sample popu-
lation. The frequency of incidents reported for the iron and steel sources
was about the same for the average of the sample population (6.9 versus 8.6).
Incident durations and magnitudes were less than the average, causing annual
excesses for this industrial category (3.4 tons) to be lower than the
average of the sample population (10.6 tons).
Iron and steel sources have more unforeseen related problems than the
average of the sample population. Comparatively, 4 of the 10 iron and steel
plants had an unforeseen problem whereas for the average of the sample
population 2 sources out of 10 reported similar related problems. Among
the same line, the frequency of the unforeseen incidents is 1% times greater
for iron and steel sources than the average. The duration and magnitude
of the incidents were found to be less than the average of the sample popu-
lation. These latter two components help to keep annual excesses (4 tons)
lower than the average of the sample population (9.2 tons).
Excess Emissions From Aluminum Plants
On-site surveys were made at four aluminum plants. The sources investi-
gated were involved in the melting of aluminum scrap. Source sizes ranged
from less than 100 to 212 tons of uncontrolled emissions per year; two of
the sources were rated at less than 100 tons. Emissions from two of the
operations were controlled by baghouses, one was controlled by an electro-
static precipitator with the fourth employing afterburner control. The con-
trol devices ranged in size from 9,940 to 130,000 acfm.
Half of the sources surveyed reported excess emission incidents. This
is less than the number of sources in the sample population with problems.
The average frequency, duration, and magnitude reported for all causal codes
combined under this industrial category were less than the average for all
sources. As a result, annual excess emissions from aluminum plants were
found to be less than the average of the sample population (46 pounds versus
22 tons). With respect to an allowable emission limitation, aluminum plants
show a high degree of control. Average annual credits for the sources
surveyed are 16.3 tons compared with an allowable emission limit of 29.1 tons.
After normalizing the excesses and credits and offsetting the excesses, the
aluminum sources surveyed were found to operate with a credit surplus of 57
percent of the allowable limit.
30
-------
None of the sources surveyed reported design or unforeseen related inci-
dents. Only one source reported a process related incident which occurred
3*< times annually each lasting an average of 5 minutes at 50 percent above the
allowable limit. Aluminum plants are much better controlled than the average
of the sample population with this problem. Similarly, only one source was
found to have an O&M related incident. This incident occurred four times
annually lasting 24 hours at 67 percent above the allowable limit. Similar to
the process problem discussed above, aluminum sources have fewer excess emis-
sions resulting from O&M problems than the average of the sample population.
The process and O&M information presented should be used with discretion because
of the limited data base.
Excess Emissions From Brass and Bronze Plants
For this industrial category, five sources were surveyed. All sources
investigated relate to furnace melting operations. The size of the furnaces
ranged from less than 100 to 8,400 tons per year of uncontrolled emissions;
three of the five were rated at less than 100 tons. Fabric filters (baghouses)
were used to control emissions from all five furnaces. The size of the bag-
houses ranged from 6,000 to 450,000 acfm.
Only one of the five sources surveyed, a copper-melting cupola controlled
by two parallel baghouses, reported incidents. The three incidents reported
were due to process and control O&M problems. This 1 in 5 ratio is less than
the 3.3 in 5 ratio for the sample population. The average frequency (0.6)
of the incidents for the two causal factors combined was less than that for
the average of the sample population (23.7). Similarly, the duration was
shorter than average. The relative magnitude, however, was almost four times
higher than the average of the sample population due to total loss of control.
With respect to normalized data, once the excess has been offset, brass and
bronze plants operate with a credit surplus of 59 percent of the allowable
limit.
Process related problems reported for the cupola occurred when the cupola
was shut down for emergency maintenance. This occurred once per year lasting
10 minutes. Compared with the average of the sample population for this
causal code, brass and bronze sources have a smaller amount of excess emissions.
Control O&M related incidents occurred twice per year, both times due to
baghouse fires. With respect to the average of the sample population, brass
and bronze plants have fewer O&M related incidents which occur less often and
last for a shorter duration of time. The relative magnitude of the incidents
is higher in this category but overall these plants have fewer annual excess.
emissions than the average of the sample population.
Excess Emissions From Petrochemical Plants
For this industrial category, six sources surveyed were used in the data
analysis. They ranged in size from less than 100 to 8,600 tons of uncontrolled
emissions annually; two of the six were rated at less than 100 tons. Two of
31
-------
the sources were controlled by fabric filters (baghouses), one was served by
an electrostatic precipitator with the remaining three employing some "other"
type of control (e.g., afterburner opr absorber). Control device sizes ranged
from 5,500 to 25,000 acfm.
Of the six sources investigated, two were found to have incidents of
excess emissions. Both sources reported incidents resulting from more than one
causal factor. Based on the average of the sample population for all causal
codes, petrochemical plants have fewer excess emission problems, one plant
in three reporting a problem versus two in three overall. The frequency
and magnitude of the incidents at petrochemical sources are less than average
but the durations are longer. It should be pointed out that the long average
duration was attributed to two separate incidents occurring at two separate
plants for two different causal factors. One, a design related problem lasted
6,000 hours, while the other, a control O&M problem, continued for 2,160 hours.
Annual excess emissions from plants (11.5 tons) in this category are
about half those of the average of the sample population. Analysis of the
normalized annual credits and excesses show that petrochemical sources operate
with an annual credit surplus of 39 percent of the allowable emission limita-
tion. This mode of operation is higher than the average of the sample popula-
tion (28 percent).
Although only one of the six sources surveyed reported a design problem,
this ratio is higher than the average of the sample population. Even though
the duration was found to be higher, as discussed above, the frequency, magni-
tude and resulting annual excess emissions were found to be lower for petro-
chemical sources than the average of the sample population.
Process related incidents occurred at two of the six sources investigated.
This ratio is similar to that of the average of the sample population. The
frequency with which the process problems occur are much less than the sample
population average (0.75 times versus 65.8) and the magnitude is also less
(255 and 421, respectively). However, the duration of each incident is almost
79 times longer than the average of the sample population. With respect to
annual excesses, the lower frequency far outweighs the long duration resulting
in lower emissions (4.8 tons) than that based on the average of the sample
population (18 tons).
Problems related to control O&M were reported for only one of the six
(17 percent) sources evaluated. This is a lower percentage than that (41
percent) for the average source of the sample population. The incidents
resulting from O&M problems are similar in nature to the process related
incidents discussed above. They occur less often and are of smaller magnitude
but last longer than the average of the sample population resulting in fewer
annual excesses. No unforeseen related incidents were reported at the petro-
chemical sources investigated.
Excess Emissions From Stone, Clay, Glass and Mineral Plants
Data obtained from 10 on-site plant surveys were analyzed. The size of
the sources investigated ranged from less than 100 to 9,240 tons of uncontrolled
32
-------
emissions per year, 3 of the 10 were rated at less than 100 tons. Baghouses
were used to control emissions at five of the 10 survey sources. Electro-
static precipitators were operated by three of the sources with the remaining
two employing wet scrubbers. Control device sizes ranged from 230 to 23,000
acfm.
Six of the ten sources were found to have at least one type type of causal
factor related incident. Of these six, one reported problems resulting from
more than one causal factor. The number of sources within this industrial
category reporting problems is similar to the average of the sample population.
The average frequency with which the problems occur, however, is six times
less than the average of the sample population (3.9 versus 23.7). When an
incident does happen, it lasts twice as long but is of lower relative magnitude
than the average of the sample population.
Sources in this category tend to release small amounts (0.7 tons) of
excess emissions annually; well below the average of the sample population
(22.5 tons). This, along with the fact that these sources operate with a
normalized credit surplus of 60 percent of the allowable limitation
indicates that stone, clay, and glass plants are well controlled. A total of
38.7 incidents were reported. Most of these excursions (80.7 percent) resulted
from control O&M problems, with 18.1 percent due to unforeseen events and
1.3 percent caused by poor design.
Design related incidents were found at 1 of the 10 plants surveyed.
This is about the same as the average of the sample population. The frequency
of the problem occurred only once every two years, which is below the average
of the sample population (9.5 times a year). Similarly, the magnitude of
the incident was less than that for the average of the sample population having
the same causal factor-related problems. Although the incidents last longer,
total annual excesses were substantially lower than that of the average of the
sample population.
No process related incidents were reported for the sources surveyed.
With respect to control O&M problems, a greater number of clay and glass
sources tend to have problems resulting from this causal factor than other
industrial sources. Five of the ten facilities investigated reported O&M
related incidents. Based on the average of the sample population, the incidents
occur less often and are of lower relative magnitude but last longer. Average
excess emissions resulting from the combination of these factors are 14 tons
annually, more than that of the average of the sample population (10.6 tons).
Unforeseen problems were reported at two sources (20 percent). This is
about the same number as the average of the sample population reporting a
similar problem. Although these unforeseen incidents occur more often than
the average of the sample population, they do not last as long nor are they
as severe, resulting in fewer annual excess emissions than the average of the
sample population.
33
-------
Excess Emissions From Grain Handling Operations
For this industrial category, five sources were included in the data
analyses. Source sizes ranged from 1,960 to 19,700 tons per year of uncon-
trolled particulate matter. All sources were controlled by baghouses, which
ranged in size from 10,000 to 325,000 acfm. There are approximately 477 grain
terminal elevators in the country. Average uncontrolled emissions from loading
and transferring/conveying operations at these sources are 410 tons per year.
Of the five sources investigated, three reported excess emission incidents;
two were found to have multiple causal factor-related problems. Based on the
average of the sample population, the number of sources in the grain handling
category found to have problems are about the same, both reporting approxi-
mately 6 of 10 facilities with incidents. The frequency of the incidents and
durations are lower than the average of the sample population relative but
their magnitudes are substantially higher. Because of the latter, total
annual excess emissions (123.5 tons) from plants in this category are greater
than the average of the sample population (22.5 tons).
The study found that, of the sources surveyed, normalized excesses ex-
ceeded normalized credits indicating that on a yearly basis grain handling
facilities operate out of compliance. It should be noted, however, that one
source included in the data analysis operated totally uncontrolled for 1,296
hours of the year. This was a design related problem which resulted in
604.5 tons of excess emissions. Consequently, this one plant greatly affected
the average values reported.
Design related incidents appear to cause the most problems. Of the
total 82.7 incidents which were reported, 72.5 (88 percent) were design
related. Associated with the design problems were incidents with very high
magnitudes. One source reported a magnitude of 29,515 percent. What this
implies is that grain handling facilities are large sources of uncontrolled
emissions requiring a high degree of control and that, when a design problem
occurs, the source tends to lose total control.
No process related incidents were reported. With respect to control O&M
problems, two sources reported problems. This about the same number as the
average of the sample population. The average frequency, duration, magnitude,
and amount of excess emissions were found to be less than the average of the
sample population for this causal code. This implies that grain handling
sources have fewer and less severe (lower magnitude) O&M related problems than
other industries. This would be expected because baghouses, which were
operated by all sources, tend to have the fewest O&M problems.
Only one source reported an unforeseen problem, which is the reporting
frequency of the average of the sample population. Overall, the incident
indicators identified for this one source are less than those of the average
of the sample population.
34
-------
Excess Emissions From Lumber and Wood Products Plants
Data from three sources in this industrial category were included in the
data analyses. Two of the sources were rated at less than 100 tons/year with
the third having an uncontrolled emission rate of 650 tons. The largest
source was controlled by a baghouse and the other two by "other" devices, those
being a cyclone with water sprays and a sand bed filter. The control devices
ranged in size from 1,000 to 34,000 acfm.
All three sources reported incidents of excess emissions and each incident
related to control operation and maintenance problems. This 100 percent of
sources reporting problems is, obviously, greater than the average of the
sample population (66 percent). Also, greater than the average of those source
reporting a control operating and maintenance problem.
With respect to control O&M related problems, the lumber and wood sources
surveyed had seven times as many incidents as the average of the sample popula-
tion. It is important to note that one source, 'that controlled by the sand bed
filter, had 71 percent of the incidents reported. The average duration and
magnitude of the O&M incidents were less than that of the average of the sample
population. These latter two indicators explain why the average annual
excesses (0.4 tons) from these sources were found to be less than the average
of the sample population (10.6 tons). Based on the analyses of the normalized
credits and excesses, sources in this industrial category operate with a credit
surplus of 36 percent of the allowable emission limitation.
Excess Emissions From Surface Coating Operations
For this industrial category, data from 13 sources were analyzed. The
sources ranged in size from less than 100 to 2000 tons of uncontrolled emis-
sions per year; four being less than 100 tons. Eight sources used some "other '
device to control emissions, whereas two sources operated electrostatic pre-
cipitators, two were served by wet scrubbers, with the thirteenth source being
controlled by a baghouse. The control device sizes ranged from 3,500 to
80,700 acfm.
Of the nine sources reporting incidents, three had excess emission
excursions related to more than one causal code. The 69 percent (9 of 13 sour-
ces) reporting incidents is similar to the average of the sample population
(66 percent). Excess emission incidents occurring at surface coating sources
happen less frequently but last longer and are more severe than the average of
the sample population. These latter two incident indicators do not appear to
be as important as the frequency because average annual excesses from surface
coating sources is about half that of the average of the sample population
(10.1 versus 22.5 tons). Looking at the normalized data, sources in this
industrial category, on an annual basis, tend to operate close to their allow-
able emission limitation having only a 6 percent credit surplus.
No design related incidents were reported by the sources surveyed. Pro-
cess related problems, however, accounted for 58 percent of reported incidents,
with 4 of the 13 sources (31 percent) reporting excursions. Although the
35
-------
incidents did not occur as frequent as the average of the sample population,
the average duration and magnitude were greater. All things considered, sur-
face coating sources were found to have fewer excess emissions resulting from
process problens than the average of the sample population (£.8 tons versus
18.4 tons).
Control operation and maintenance related problems accounted for 34 per-
cent of the reported incidents, occurring at 5 of the 13 (38 percent) sources
surveyed. The 38 percent reporting incidents is similar to the 41 percent
based on the average of the sample population. The frequency with which
an incident occurs is less but the duration and magnitude are higher when one
compares surface coating sources with the average incident indicators of
the sample population. Because the incidents occur only 2.5 times per
year, excesses due to O&M problems are lower for surface coatings sources
than other industrial facilities. It should be noted that the average duration
and magnitude are high because one source reported an incident which lasted
an entire year (8,760 hours) while two sources reported magnitudes of over
2,000 percent (one at 30,456 percent).
Surface coating sources have a higher number of unforeseen problems
than the average of sample population. Approximately 39 percent (5 of 13)
of the sources surveyed reported unforeseen related incidents. Although more
sources reported incidents, these incidents occurred less frequently, roughly
once every 2 years. Annual excess emissions are higher than the average of
the sample population, apparently due to the severity and length of the
incidents (1,533 percent above the standard and 673 hours, respectively).
Similar to the O&M incident indicators discussed above, the high averages for
duration and magnitude were attributed to only one or two sources which had
major problems.
Excess Emissions From Food and Drug Plants
Data analysis was conducted on three food and drug sources. The sources
included pet food drying ovens, a meat smokehouse, and dry products handling
at a gum manufacturing plant. Source sizes were 9.2, 120, and 159 tons per
year of uncontrolled emissions. In the order presented above, source control
device types were a fume incinerator, wet scrubber, and baghouse. The size of
the control units ranged from 4,000 to 11,000 acfm.
Only one of the sources surveyed, the meat smokehouse, reported an
incident, which resulted from a control operation and maintenance problem.
Because only a few sources were surveyed in this category, with only one
reporting problems, the data presented should be used with discretion. The
O&M related incident is less severe and occurs less frequently than the average
of the sample population. However, the incident tends to last longer. For
this category, annual excess emission are totally offset by annual credits
resulting with the sources operating with a net surplus of 40 percent of the
allowable emission limitation.
36
-------
Excess Emissions From "Other" Plants
Data analyzed for the category, which includes mineral acid and ammonia
phosphate fertilizer plants, were obtained from six sources. The sources were
comprised of four fertilizer plants and two sulfuric acid plants. Plant sizes
ranged from 100 to 5,880 tons per year of uncontrolled emissions. Control
device types included three wet scrubbers, one baghouse and two "other" units.
The size of the control equipment ranged from 4,500 to 149,000 acfra. With
respect to source representativeness, there are 150 sulfuric acid and 41
ammonia phosphate plants in the United States. The average size of the plants
in these two industries are 283 and 3,591 tons per year of uncontrolled emis-
sions, respectively.
With regard to the sources surveyed, four of the six (66 percent) reported
an excess emission incident. This is the same number as the average of the
sample population. Overall average incident indicators, which include fre-
quency, duration, magnitude and annual excesses, were less than those found
for the average of the sample population. Consequently, it was not unusual
to learn that "other" sources generally operate with a credit surplus of 64
percent of their allowable limitation. This would indicate a high degree
of control with few problems. Of the 113.7 incidents reported, 79 percent were
process related, 19 percent control operation and maintenance related, and
2 percent were due to unforeseen problems.
Design related incidents were not found at any of the sources investigated.
Concerning process related problems, three (50 percent) of the six facilities
reported incidents, which is higher than the average of the sample population
(29 percent). Similar to the overall average of all causal factors discussed
above, process related incidents at "other" sources occur less frequently,
are shorter in duration and less severe than the average of the sample popula-
tion. This means that annual excess emissions resulting from these incidents
are also less than the average of the sample population.
Causal code 3 (control O&M) had the greatest number of sources reporting
problems. Of the six sources investigated four (66 percent) were found to
have O&M problems. This is more than the average of the sample population
reporting O&M problems. The average of the incident indicators for "other"
sources with O&M related problems are all less than those of the average
of the sample population.
Unforeseen problems were found at three of the six (50 percent) sources.
This is more than twice the average of the sample population. Although they
tended to last longer, the frequency, magnitude and excesses associated with
those incidents were less than the average of the sample population.
PARAMETER B. EXCESS EMISSIONS AS A FUNCTION OF SOURCE SIZE
To determine whether source size has any bearing on the amount of excess
emissions released from a source, we have summarized the average frequency,
duration and magnitude data for the incidents reported according to specific
37
-------
size ranges. The following size cutoffs were selected in an effort to distri-
bute the number of sources uniformly within each size category (numbers are in
tons of uncontrolled emissions per year):
• <100
• 100 to 500
• 500 to 1,000
• 1,000 to 10,000
• 10,000 to 100,000
• 100,000 to 500,000
A comparative discussion of the types of sources found in each size
category and the average incident indicators will be presented, starting with
the smaller sources and ending with the larger ones. Data relative to this
parameter is found in Tables 3 and 6.
Size Category - <100 tons per year
Sources falling into this category have typically been placed low on the
state and local agencies priority lists with respect to their compliance
status and emission control. Of the 119 sources evaluated, 24 or 20 percent
emitted less than 100 tons/yr of pollutant per year. The 24 sources were
represented by eleven different industrial categories, surface coating plants
having the greatest number (four sources). Of the incidents reported for the
sources of this category, 73 percent were control operation and maintenance
related, 23 percent were caused by process problems, 3 percent were due to
unforeseen, with the remainder (<1 percent) attributed to design flaws.
Based on an analysis of the normalized data, sources less than 100 tons per
year were found to operate with an annual credit surplus of 44 percent of
their allowable emission limitation. This is almost double that (28 percent)
found for the average of the sample population.
Focusing on the Control O&M and Process related incidents, 13 (54 percent)
of the 24 sources identified reported causal code 3 (O&M) problems, whereas
3 (12.5 percent) of the 24 were found to have causal code 2 (process) related
problems. Compared with the average of the sample population, O&N related
incidents were reported by a higher proportion of sources in the less than
100 tons per year size category. Conversely, the average of the sample popula-
tion has a larger number of sources reporting process related problems than
those sources less than 100 tons per year.
Process related incidents at these smaller sources occur less frequently
but have a higher relative magnitude and last longer than those at the average
of the sample population. Because they are small to begin with, total annual
excess emissions from these sources are minimal compared with those from other
sources.
38
-------
With respect to control O&M, the incidents do not occur as often and
they are of lower relative magnitude but again tend to last longer than the
incidents found for the average of the sample population.
Size Category - 100 to 500 tons per year
Data analysis was conducted on 31 sources which fell into this size
category. Eleven of the sixteen industrial categories are represented. Of the
31 sources identified, five are surface coating sources and four are steam
generating units which were located at the same plant. Causes of the incidents
reported at sources in this size category tend to be equally distributed between
process, control operation and maintenance and unforeseen related problems.
The percent of incidents attributed to each of these causal factors are 35.4,
36.6, and 24.4, respectively, with 3.5 percent due to design related problems.
Comparing the sum of all causal factors for this size category and the
average incident indicators for all sources of the sample population, an
excess emission incident occurs much less frequently but last substantially
longer and is of higher relative magnitude than the average of the sample
population. Because the incidents last longer and on a relative basis are
quite severe, 100 to 500 tons per year sources operate with the lowest annual
credit surplus (12 percnet of their allowable emission limitation). This is
roughly half the surplus (28 percent) found for the average of the sample
population.
Only one of the 31 sources identified reported a design related incident.
The incident occurs three times every 2 years, each one expected to last
2,920 hours resulting in emissions nine times the standard. Based on the
average of the sample population reporting design problems, 100 to 500 tons per
year sources have fewer incident per year and are of lower relative magnitude
but tend to last almost 20 times longer. All things considered, annual
excesses would be about the same for the average of the sample population and
the 100 to 500 tons per year source.
With respect to process related incidents, 7 of the 31 (22.6 percent)
100 to 500 tons per year sources reported a problem. This 22.6 percent is
close to the number of sources (29 percent) of the sample population process
related incidents. Concerning the average incident indicators, sources
reporting in the 100 to 500 tons per year category have fewer incidents per
year than the average source of the sample population each being of lower
relative magnitude and shorter duration.
Like the process related incidents, control operation and maintenance
related excursions do not occur as often nor for as long as the average of the
sample population, but they are of a much higher relative magnitude. This
latter case results basically from one source reporting an incident which was
30,456 percent above its allowable limit. The incident occurred only once,
lasting 5 minutes.
39
-------
Of the 31 sources identified in this size category, 6 (19.4 percent)
reported an unforeseen related incident. For the sample population, this is
about the same number of sources reporting a similar type of problem. The
average incident reported for 100 to 500 tons per year sources occurred about
as frequently and of nearly the same relative magnitude as the average of
the sample population but tended to last longer.
Source Category - 500 to 1000 tons per year
As shown in Table 3, 17 (14 percent) of the 119 sources investigated
had an uncontrolled emission rate of between 500 to 1000 tons per year. These
sources represent 10 industrial categories, with steam generating plants being
the most common occurring with three sources. Of the 1089.7 incidents reported
by these sources, 91 percent were process related, 8 percent were control
operation and maintenance related, with the remaining 1 percent made up by
design and unforeseen problems. With respect to the average incident indica-
tors for all causal factors, sources in the size category would have about as
many incidents which would last the same length of time but would be of lower
relative magnitude than the incidents reported for the average of the sample
population.
Analysis of the normalized data for sources in this size category reveals
that 500 to 1000 tons per year facilities were found to operate annually with
an average credit surplus of 40 percent of their allowable limitation.
Focusing only on causal codes 2 and 3, 500 to 1000 tons per year sources
were found to have more process related incidents but fewer control O&M related
problems than the average of the sample population. It. is important to note
that the higher number of process related incidents reported for the sources
in this category resulted from one source, a steam generating unit, which
was found to have 882 excess emission excursions due to grate cleaning. These
incidents represent 89 percent of all process related problems reported for
this size category and 81 percent of all incidents found to occur at 500 to
1000 tons per year sources. The average process related incident reported for
these sources was shorter and of lower relative magnitude than those calculated
for the average of the sample population. The percent (29.0) of sources
reporting process problems is the same as the percent (29.0) of sources of the
sample population reporting similar problems.
Concerning control O&M, 8 (47 percent) of the 17 sources investigated
reported problems. This is a slightly higher percent than that (41 percent)
found for the average of the sample population. Each incident occurred more
often but lasted for a shorter period of time and was of lower relative magni-
tude than a similar related incident for the average of the sample population.
Source Category - 1,000 to 10.000 tons per year
Of the 119 sources surveyed, 30 (25 percent) were rated at 1,000 to
10,000 tons per year of uncontrolled emissions. For the 16 industrial cate-
gories surveyed 12 were represented in this size category. Petroleum products
40
-------
and handling plants and grain handling operations had the greatest number of
sources identified with five and four facilities apiece. With respect to the
1,119 incidents reported, 83.5 percent were process related, 15.2 percent
were due to control O&M problems with the remaining 1.3 percent attributed
to design flaws and unforeseen events. A breakdown of each causal code shows
that 10 percent of the 30 sources reported a design related incident, 43.3
percent had process related problems during the year, 46.7 percent were found
to have had control O&M related incidents and 26.7 had some unforeseen event
occur. Approximately 70 percent of all sources reported at least one incident.
Comparing the computed numbers above with the corresponding values for the
sample population, (9.2, 29, 41, 21, and 66), plants in this size category
tend to have a greater number of sources experiencing problems. Based on the
average for all causal codes of the sample population, incidents reported by
the 1,000 to 10,000 ton category occur more often but are of lower relative
magnitude and last for a shorter period of time.
The following discussion will only cover incidents related to process and
control O&M problems which were found to occur with the greatest frequency.
Concerning process problems, sources in the 1,000 to 10,000 tons per year
range have more incidents per year than the average of the sample population.
However, one source, a basic oxygen furnace of the steel industry, reported a
process related incident which occurred 700 times in 1 year. This one source
thus has a significant impact on the average for this causal code. The average
duration of the process related incidents for all 1,000 to 10,000 tons per year
sources was 7.6 hours at 154 percent above the allowable limit. This is a
longer duration but lower relative magnitude than the average of the sample
population.
Incidents caused by control O&M problems tend to occur more often at
sources in the 1,000 to 10,000 tons per year range, but are shorter and of lower
relative magnitude than the average of the sample population.
Source Category - 10,000 to 100,000 tons per year
As shown in Table 3, 11 of the 119 sources analyzed fell into this size
category. These 11 sources represent 5 different industrial categories, with
5 of the 11 being pulp and paper recovery boilers. Of the 400 incidents
reported, 64.5 percent were process related (89.5 percent of which occurred at
two pulp and paper recovery boilers), 24.8 percent were due to design flaws,
8.8 percent resulted from control O&M problems, with the remaining 2 percent
due to unforeseen events. The breakdown of each causal code by number of
sources with problems is as follows: 36 percent of the sources were found to
have designs related incidents, 27.3 percent reported incidents resulting from
changes in operations, 36.4 percent had incidents related to control operation
and maintenance problems, and 18 percent reported incidents due to unforeseen
events. Combining all causal codes together, 72.7 percent of the sources had
an excess emission incident. Comparing these percentages with the corresponding
values for the sample population (9.2, 29, 41, 21 and 66), a greater number
of sources in the 10,000 to 100,000 tons per year size range would have design
and overall problems than the average but a fewer number would have process,
control O&M and unforeseen related incidents.
41
-------
With respect to normalized credits and excesses, sources in the 10,000
to 100,000 tons per year size range tend to operate with less of a credit
surplus than the average of the sample population (17 percent versus 28,
respectively). As expected, on the average annual excess emissions (64.5
tons) are greater from sources in this category than those (22.5 tons) from
the average of the sample population. Compared with this overall average,
10,000 to 100,000 tons per year sources have more incidents per year which have
a higher relative magnitude but do not last as long.
Incidents which resulted from design flaws occur more often but do not
last as long and are of lower relative magnitude than those of the average
of the sample population experiencing similar problems. Continuing on with
the other causal codes, process related incidents at 10,000 to 100,000 tons per
year sources occur more frequently (the reason for the higher frequency has
been discussed above), tend to be shorter and not quite as severe as those at
the average of the sample population. Control O&M related incidents occur
about as frequently as those of the average of the sample population but are
shorter and of lower relative magnitude. Incidents due to unforeseen events
occur more often at the sources in this category than the average of all
sources, but their duration and relative magnitudes are less.
Source Category - 100,000 to 500,000 tons per year
Of the 119 sources investigated, 6 were rated between 100,000 to 500,000
annual tons of uncontrolled emissions. Cement plants, specifically rotary
kilns, dominate this category representing 5 of the 6 sources. The sixth
source was a pulp and paper recovery boiler which did not have any excess
emission incidents. A total of 32.5 incidents were reported, 6 (18.5 percent)
resulted from process upsets, 21.5 (66.2 percent) were due to control opera-
tion and maintenance problems and 5 (15.4 percent) were due to unforeseen
events. No design related incidents were reported. Of the 6 sources classi-
fied in this category only 3 (50 percent) reported an incident. Two of these
3 reported incidents due to more than one causal factor. On a causal code by
causal code breakdown, 2 sources (33.3 percent) were found to have process
related problems, 2 sources (33.3 percent) reported incidents due to control
operation and maintenance problems, while only one (16.7 percent) source
reported an unforeseen event. On the whole, these larger sources tend to have
a fewer number of sources reporting incidents than the number reporting problems
from the sample population. Analysis of the normalized data reveals that the
sources in this category operate on an annual basis with a credit surplus of
29 percent of their allowable emission limitation.
Considering the sum of all causal codes as one, these large sources have
fewer incidents per year than the average of the sample population but they
last longer and have a much higher relative magnitude. This latter indicator
is common for large sources which require a high degree of control. When a
problem does occur, it results in a large quantity of emissions. It should be
noted that the average magnitude for all causal codes under this category is
biased high due to one source reporting incidents with a magnitude of 115,900
percent above the standard. These incidents, which occurred five times
annually and last only 10 minutes each, were process related.
42
-------
Average incident indicators for the sources reporting problems are shown
in Table 3. For the process related incidents, large sources have fewer per
year but they last much longer and are of higher relative magnitude than those
found for the average of the sample population. Referring to the same Table 3
above, control O&M related incidents occur more often and are of a higher magni-
tude than those of the average of the sample population but do not last as long.
With respect to unforeseen events, large sources were found to have more inci-
dents per year than the average but they did not last as long nor were they as
severe.
PARAMETER C. EXCESS EMISSIONS AS A FUNCTION OF CONTROL DEVICE TYPE
In the preceding discussion, we have summarized the causes, frequency,
duration, and relative magnitude of excess emissions on a source size basis.
This section is devoted to the presentation of our findings on the basis of the
generic types of air pollution control devices that were evaluated by the four
contractors. Such larger groupings provide us a more extensive population upon
which to develop specific and meaningful comparisons and correlations. The
following discussions consider only excess emissions associated with the sub-
ject control device and its auxiliary equipment.
We do recognize, however, that the potential of a process-derived bias on
the control device related excess emission incident data is still possible.
For example, fan blade failure is much more likely to occur for those cases
where the process exhaust stream is of a corrosive or abrasive nature than for
those which are not. The significance of such a bias lessens as the sample
population becomes larger and more diversified, however. By aggregating all
the data developed by the four contractors, the significance and validity of
the conclusions are strengthened and the impact of specific facility or in-
dustrial category bias is minimized.
A comparative description of the reported incidents by causal code for
each control device type will be presented in this section. Reference will be
made to the incident indicators of the average source, which evolved from the
data reported for all 119 sources surveyed. The generic control devices in-
vestigated are:
• Electrostatic Precipitator
t Scrubber
• Fabric Filter (Baghouse)
• Other (including cyclones, afterburners, vapor recovery
units, absorbers and mist eliminators
Refer to Tables 4 and 6 for data summaries pertaining to this parameter.
43
-------
Excess Emissions Attributable to Electrostatic Precipitator Related Causes
As shown in Table 4, emissions from 22 (18.5 percent) of the 119 sources
investigated are controlled by an electrostatic precipitator (ESP). A total
of nine industrial categories are represented by these sources, with cement
plant rotary kilns comprising 5 of the 22 sources identified. Considering all
causal codes, the number of ESP sources reporting problems (77.3 percent) is
higher than the number (66 percent) in the average of the sample population
reporting a problem. Moreover, 8 of the 17 reported incidents resulting from
more than one causal code. The frequency of incidents occurring at ESP
sources and their relative magnitude are higher than the average of the
sample population, although the duration of the incidents are not as long.
Of the 1025.3 incident reported for the four causal codes under this category,
84.2 percent were process related, 13.9 percent resulted from ESP operation
and maintenance problems, 1.6 percent were unforeseen related, with the re-
maining less than 1 percent due to design flaws. It should be noted that
81 percent (700 incidents) of the process related incidents were attributable
to one source, that being a basic oxygen furnace of the steel industry.
With respect to the normalized data, ESP sources had the second highest
credit surplus, operating on an annual basis with 27 percent of the allowable
emission limitation as credits. Normalized credits were 51 percent of the
allowable, whereas normalized excesses were 24 percent.
Only three (13.6 percent) of the 22 ESP sources identified reported a
design related incident. This is slightly higher than the number of sources
reporting a similar related problem from the sample population. This is not
surprising considering the sensitive nature of electrostatic precipitators with
respect to electrical tuning and their dependence on flue gas conditions.
Compared with the incident indicators for the average source, ESP sources have
far fewer design related incidents per year and they are of lesser relative
magnitude but tend to last quite a bit longer. The longer average duration was
attributed to one source that reported cm incident which lasted 2,190 hours.
Under process related problems, 9 of the 22 (40.9 percent) sources re-
ported an incident. This is higher than the average of the sample population.
The frequency of these indicators is higher at ESP sources than for the
average, as are the durations and relative magnitudes. The high frequency as
discussed above was due to one source reporting 700 incidents in 1 year.
Electrostatic precipitator operation and maintenance problems were re-
ported at 9 (40.9 percent) of the 22 sources classified in this category.
This is the same percent of sources reporting similar related problems in the
sample population. Operation and maintenance related incidents were found to
occur more often at the ESP sources than the average of the source population,
lasting a shorter period of time but are of higher relative magnitude.
With respect to unforeseen events, 36.4 percent (8 of 22) of the ESP
sources reported an incident. This is higher than the number in the sample
population having a similar related problem. ESP sources have a greater number
44
-------
of these unforeseen related incidents and they last longer but they are not
as severe as those for the average of the sample population.
Excess Emissions Attributable to Scrubber Related Causes
Of the 119 sources investigated, 36 (30.3 percent) were controlled by a
scrubber. A total of 11 industrial categories are represented, the two most
common are iron and steel plants and steam generating plants (seven and six
sources, respectivley). Considering all causal codes, 25 (69.4 percent)
sources reported an excess emission incident. This is about the same number
(66 percent) of sources as in the total sample population reporting an incident
regardless of causal code type. Data tabulation showed that 7 of the 25
sources reported incidents resulting from more than one causal code.
Normalization of the excess and credits data revealed that sources con-
trolled by scrubbers operate with an annual credit surplus of 21 percent of
their allowable emission limitation. This is slightly less than that (28
percent) found for the average of the sample population.
Of the 300.7 incidents reported, 77.7 percent were process related, 19.5
percent were due to scrubber O&M problems, 2.7 resulted from unforeseen events
with the remaining less than 1 percent due to design flaws. Since process
and O&M related incidents comprised most of the problems occurring at the
sources in this category the remaining discussion will be addressed to these
two causal codes.
Process related problems were reported at 13 (36 percent) of the 36
sources identified, which is slightly higher than the number (29 percent) in
the sample population reporting similar related problems. Sources in this
category have fewer process related incidents per year than the average of
the sample population. These incidents tend to last longer but are of lesser
relative magnitude.
With respect to O&M problems, 14 (38.9 percent) of the 36 sources con-
trolled by scrubbers reported an incident. This is slightly lower than the
number (41 percent) of sources in the sample population reporting O&M
problems. Scrubber sources were found to have fewer incidents and of lower
relative magnitude than the average of the sample population but they tended
to last longer.
Excess Emissions Attributable to Fabric Filter Related Causes
As shown in Table 4, emissions from 35 (29.4 percent) of the 119 sources
surveyed are controlled .by a fabric filter (baghouse). Of the 16 industries
surveyed, 13 are represented in this category. The most commonly occurring
sources are asphalt plants (six), brass and bronze plants (five), stone, clay
and glass plants (five), and grain handling facilities (five). Taking into
consideration all causal codes, the number of baghouse sources reporting
problems (48.6 percent) is lower than the number (66 percent) in the sample
45
-------
population which reported a problem. In addition, of the 17 sources reporting
excess emission related problems, six reported incidents caused by more than
one causal code.
The overall average frequency of occurrence and duration of an incident
at a baghouse source is less than the average of the sample population, whereas
the relative magnitude is higher. Of the 292.86 incidents reported for all
causal codes combined, 50.5 percent were attributed to O&M problems, 33.6
percent were caused by design flaws, 10.9 percent resulted from process
related problems and 5 percent were due to unforeseen events.
Analysis of the normalized data for sources classified in this category
show that baghouse controlled facilities, as expected, operate with the
greatest amount of credits. The annual credit surplus for these sources is
51 percent of the allowable emission limitation. Compared with the other con-
trol device type categories, baghouses had the lowest amount of normalized ex-
cess emissions (14 percent) and the greatest amount of normalized credits (65
percent).
Only 4 (11.4 percent) of the 35 baghouse sources identified reported
design related incidents. However, on a proportional basis this is a higher
percentage than the number (9.2 percent) of sources in the sample population
reporting a similar related incident. Design related incidents at baghouse
sources were found to occur more often than those of the average of the sample
population. They tended to be of lower duration but slightly higher relative
magnitude than the average. This latter case was attributable to one source
reporting an incident that occurred once every 2 years, which had a magnitude
of 29,515 percent.
With respect to process related problems, only 3 (8.5 percent) of the
35 sources surveyed were found to have an incident. Compared with the average
of the sample population, sources controlled by baghouses will have fewer
and less severe process related problems.
Operation and maintenance related incidents were found at 14 (40 percent)
of the 35 sources controlled by baghouses. This about the same number (41
percent) as the average of the sample population reporting a similar related
incident. These O&M incidents occur more often and last longer at a baghouse
controlled facility than the average but tend to be less severe.
Incidents caused by unforeseen events were found to occur at 4 (11.9
percent) of the 35 source surveyed, which is less than the number (21 percent)
of sources reporting a similar event from the sample population. Unforeseen
events occur more frequently at sources controlled by baghouses but they
do not last as long and are of lower relative magnitude.
Excess Emissions Attributable to "Other" Control Device Related Causes
As described earlier, "other" control devices include cyclones, afterburners,
vapor recovery units, absorbers, and mist eliminators. Of the 119 sources
investigated, 26 (21.8 percent) were controlled by one of the "other" control
46
-------
devices. A total of nine industries are represented in this category. The two
most prevalent are surface coating operations with eight sources and petroleum
products and handling operations with seven sources. Both of these industries
are comprised of sources of pollutants which tend not to employ the three more
conventional types of control devices as discussed in the previous subsections.
With respect to all causal codes, 19 (73.1 percent) of the 26 sources surveyed
reported an excess emission incident. This is slightly higher than that found
for the sample population, which had 66 percent of its sources reporting
excess emission related problems. In addition, 7 of the 19 sources reported «
incidents resulting from more than one causal code. Still making reference
to all causal codes, the sources of this category have more incidents per year
than the average of the sample population but they tend to last for a shorter
period of time and are of lower relative magnitude.
Analysis of the normalized data reported for the sources of this category
show that facilities controlled by these "other" devices, have, on a relative
basis, more problems controlling emissions than those sources controlled by the
more common devices. Normalized excesses for these "other" sources are the
highest for all control types surveyed, 33 percent of the allowable emission
limitation. Likewise, at 37 percent their normalized credits are the lowest
of all four control device types. Consequently, sources controlled by these
"other" control devices operate annually with a credit surplus of only 4
percent.
A total of 1196 incidents were reported for all four causal codes com-
bined. Of these, 92.9 percent were process related, 6.2 percent were due to
operation and maintenance problems, with the remaining less than 1 percent
split between design flaws and unforeseen events. For these latter causal
codes, three sources (11.5 percent) reported design related incidents and
seven (26.9 percent) reported incidents due to unforeseen events. The number
of sources reporting problems for each of these respective causal codes is
slightly higher than the number in the total sample population reporting
similar related problems. However, since the number of incidents related
to these two causal codes were so few compared with the number related to
process and O&M problems, no further mention of them will be made.
The number of sources reported process related problems (6 of 26 or 23.1
percent) is slightly lower than the number of sources in the sample population
having similar related problems. Comparing the average incident indicators,
sources controlled by "other" devices have a greater number of excess emis-
sions incidents per year but their duration and relative magnitudes are less
than average of the sample population. The high frequency of occurrence is
attributed to one steam generating unit which reported 882 incidents per year.
Incidents related to operation and maintenance problems, were reported
at 11 (42.3 percent) of the 26 "other" sources identified. This proportion
of sources reporting problems is similar to that of the sample population as
a whole. The number of O&M related incidents occurring at "other" sources
annually is less than the number expected for the average of the sample popula-
tion. However, the duration and relative magnitudes of each incident would
47
-------
be greater than the average. It should he noted that the high relative value
of these latter two incident indicators are attributed to two very large values;
one source reported an O&M related incident which lasted 1 year (8,760 hours),
while a different source had a similar related excess emission incident which
was 30,456 percent above the allowable emission limitation.
PARAMETER D. EXCESS EMISSIONS AS A FUNCTION OF CONTROL DEVICE SIZE
Data analyses were conducted on the 119 sources surveyed with respect to
control device size. Categorizing the sources by control device size provided
an additional dimension of analysis. The results of this classification
scheme will identify problem areas independent of industrial process and con-
trol device type. Again we realize that process-derived bias and specific
control device related problems will affect the data averages reported within
'each size category. By integrating the data collected by the four contractors,
the impact of biases due to specific control device types and industrial pro-
cesses are minimized.
The following size categories were selected in an effort to provide an
equal spread of the 119 data entries (values presented are in terms of gas
flow rate expressed in acfm).
• < 10,000
• 10,000 to 50,000
• 50,000 to 100,000
• 100,000 to 500,000
• 500,000 to 1,000,000
• > 1,000,000
In addition to describing the types of industries found in each category,
comparisons will be made between the average incident indicators for each
causal code for the sources of a given size range and the average values for
the 119 sources of the sample population. These analyses will show which con-
trol device sizes have the most excess emissions incidents, what causes the
incidents, and compares the extent of the incidents on a relative basis to
the average of the sample population. Data for this parameter is presented in
Tables 5 and 6.
Control Devices - < 10,000 acfm
As shown in Table 5, 31 (26.1 percent) of the 119 sources investigated
employed a control device that was rated at less than 10,000 acfm. A total
of 12 of the 16 industries categories surveyed are represented by the size
category. The most common sources are petroleum products and handling facili-
ties (seven), with petrochemical and stone, clay and glass plants having four
plants apiece. Excess emissions incidents were found to occur at 17 (54.8
percent) of the 31 sources classified in this category; four of the 17 reported
multiple problems. Compared with the average of all causal codes for the 119
48
-------
sources surveyed, the number of sources controlled by these relatively small
devices is less than the number (66 percent) in the sample population. Sources
with these small control devices have fewer incidents per year than the average
of the same population, and they are of lower relative magnitude but they tend
to last longer. The longer duration, however, is attributed to four sources
having reported incidents that lasted for more than 2,000 hours.
Normalization of the excess and credits data for sources in this category
shown that "< 10,000 acfm" facilities operate with the second lowest credit
surplus of the size categories delineated in this parameter. These sources
were found to have an annual surplus of 12 percent of the allowable emission
limitation. As shown in Table 5, sources employing control devices < 10,000
acfm operate with the highest amount of annual normalization excesses, that
being 40 percent of the allowable emission limit. It is possible that the
high amount of excesses are attributed to the operation of complex "other"
types of control devices such as vapor recovery units and carbon absorption
units which are used at petroleum products and handling plants and petrochem-
ical plants.
A total of 127.2 incidents were reported at the sources falling into this
size category. Control device operation and maintenance related problems
caused the greatest number of incidents (54.4 percent). The contribution of
the other causal factors are 34.2 percent due to process related problems,
8.7 percent resulting from unforeseen events and 2.8 percent due to design
flaws.
Design related incidents occurred at 3 (9.6 percent) of the 31 " < 10,000
acfm" sources surveyed. This compares favorably with the average of the sample
population. Roughly 9 percent of sources in the sample population reported a
design related problem. The frequency with which these design related problems
occur at the sources in this category is significantly less than the frequency
of incidents for the average of the sample population. However, the duration
and relative magnitude of these incidents are much greater than the average.
The relatively high values reported for these latter two indicators are due to
one source reporting one incident lasting 6000 hours, and two sources having
incidents that had magnitudes of 900 percent.
Incidents resulting from process related problems occurred at 3 (9.7 per-
cent) of the 31 sources classified in this category. This is a lower percent-
age of sources reporting problems than those (29 percent) of the sample popu-
lation. Sources in this size category were found to have fewer process
related incidents per year than the average of the sample population. The
incidents also did not last as long nor were they as severe.
Control operation and maintenance related problems occurred at 10 (32.2
percent) of the 31 sources surveyed which is fewer than the number (41 percent)
of the sources of the sample population reporting a similarly related problem.
Incidents resulting from this causal code did not occur as often and were of
lower relative magnitude than those of the average of the sample population,
although the duration of the incident tended to be longer than average.
49
-------
Of the 31 "< 10,000 acfm" sources surveyed, four (12.9 percent) were
found to have incidents resulting from unforeseen events, which is fewer than
the number in the sample population reporting unforeseen problems. Compared
with the overall average incident indicators for this causal code, sources in
this category have a greater number of unforeseen incidents occurring per year
which last longer and tend to be more severe.
Control Devices - 10,000 to 50,000 acfm
Of the 119 sources investigated, 49 (41.2 percent) were classified into
this size category. All of the 16 industries surveyed are represented in this
size category except cement plants which are typically controlled by larger
devices. Asphalt plants and surface coating operations dominate the category
with eight sources apiece. Steam generating plants and stone, clay, and glass
plants were next with the most facilities, each having six sources represented.
Excess emissions incidents were reported at 29 (59.2 percent) of the 49 sources
entered in this category. Compared with the average of the sample population,
fewer sources in this category reported excess emissions incidents. Of the
29 reporting problems, eight sources were found to have an incident related
to more than one causal code. Taken as a whole, sources classified in this
size category have fewer incidents per year, but they last about the same
length of time and are about as severe as those at the average of the
sample population.
Analysis of the normalized data reveals that sources controlled by devices
ranging from 10,000 to 50,000 acfm operate with an annual credit surplus of
43 percent of the allowable emission limitation. This resulted from these
sources having relatively few normalized excess emissions; only 2 percent of
the allowable standard.
Of the 449 incidents reported for all causal codes, 65.5 percent resulted
from process related problems, 32.2 percent were due to control O&M problems,
1.9 percent were unforeseen, with the remaining less than 1 percent due to
design flaws. Since most of the incidents reported were process and O&M
related, the following discussion will pertain to these two causal codes.
Incidents resulting from process related problems occurred at 12 (24.7
percent) of the 49 sources surveyed. This is fewer than the number of sources
of the sample population reporting a similarly related problem. With respect
to average incident indicators, sources in this category have fewer incidents
that are of lower relative magnitude but last longer than the average of the
sample population.
Concerning control device operation and maintenance related incidents,
17 sources (34.7 percent) reported an excess emission excursion due to this
causal factor. Compared with the average of the sample population, a fewer
number of sources in this category reported O&M problems. The incidents that
occur, happen as frequently as the average, but are shorter and of a higher
relative magnitude.
50
-------
Control Devices - 50.000 to 100.000 acfm
Only 9 of the 119 sources surveyed were controlled by devices that were
rated at 50,000 to 100,000 acfm. These nine sources represent six industrial
categories, surface coating operations comprising the majority with three
sources. All nine facilities reported at least one incident with three sources
reporting problems due to more than one causal code. Based on the average of
all causal codes combined, sources in this size category have fewer incidents
per year than the average of the sample population and these incidents tend
to be shorter and are of lower relative magnitude.
Normalization of the data revealed that the sources of this category
operate with an annual credit surplus of 39 percent of the allowable emission
limitation. Of the six control device size categories, the sources of this
category operated with the greatest amount of normalized credits, 41 percent
of their allowable limitation, thus being the reason for their high credit
surplus,
A total of 126.5 incidents were reported for all causal codes combined.
A causal code by causal code breakdown shows that 69.9 percent of the incidents
reported were process related, 28.5 percent resulted from control O&M problems
and 1,6 percent were unforeseen. No design related incidents were reported.
Since 98.4 percent of the incidents reported resulted from process and control
O&M problems, these two causal codes will be discussed.
Of the nine sources identified, five were found to have incidents related
to process changes or upsets. This percentage (55.5) of sources reporting
problems is much higher than the number C29 percent) of sources in the sample
population reporting a similarly related incident. Compared with the average
of the sample population, sources in this control device size range have fewer
process related incidents that tend to be less severe but last longer.
Control O&M related incidents were reported at six (66.6 percent) of the
nine sources surveyed. This is a higher number (41 percent) than those of the
sample population reporting an O&M related incident, The frequency, duration,
and magnitude of the O&M incidents are less than those of the average of the
sample population.
Control Devices «•. 100.000 to 500.000 acfm
As shown in Table 5, 23 of the 119 sources investigated were classified
in this control device size category. These sources were represented by eight
industries, cement plants, and petroleum products and handling operations domi-
nating with seven and five sources apiece, Incidents were reported at 19
(82.6 percent) of the 23 sources. This is a higher percentage of sources re-
porting incidents than that of the entire sample population. Of the 19
sources having problems, 10 reported incidents due to more than one causal
code, Regardless of causal code, when compared with the average of the sample
population, facilities in this size category have more incidents per year
which are of the same relative magnitude but would tend to last longer.
51
-------
The annual credit surplus (10 percent) associated with the sources in
this category was found to be the lowest of all the size ranges of this param-
eter. This resulted from these sources having the second highest normalized
excess and second lowest normalized credits.
Of the 1997.7 incidents reported for all causal codes, 87.7 percent were
process related, 6.8 percent resulted from control O&M problems, 4.7 percent
were due to design flaws, and 0.8 percent were unforeseen. Design related
problems occurred at 3 (13 percent) of the 23 sources identified. This is more
than the number found to report a similarly related problem in the entire
sample population. Considering the average incident indicators, sources within
this category have fewer incidents per year than the average of the sample
population, but each excursion lasts longer and is of higher relative magnitude.
Incidents resulting from process changes or upsets occurred at 12 (52.1
percent) of the 23 facilities classified into this category. The average num-
ber of causal code 2 (process) incidents found at these sources was slightly
more than twice the number occurring at the average of the sample population
(146 versus 65.9). The duration and relative magnitudes of these incidents
are very similar at the sources within this category and the average of all
sources.
With respect to control operation and maintenance related incidents, 12
(52.2 percent) sources reported excess emission excursions. Like the previous
two causal codes, this reporting percentage is higher than that found (41
percent) for the average of the sample population. Although the sources in
this category have more incidents per year that tend to be slightly more severe
than the average of the sample population, they are also last for a shorter
period of time.
Finally, unforeseen events were reported at eight (34.8 percent) sources.
This is a substantially higher percentage than the 21 percent of the sources
of the sample population reporting similarly related incidents. Comparing
the average incident indicator data derived for the sources of this category
with the average values for the sample population, sources in this size cate-
gory have slightly more unforeseen events per year that are shorter in duration
and of lower relative magnitude.
Control Devices - 500,000 to 1,000,000acfm
Of the 119 sources surveyed, six were controlled by control devices that
were rated at 500,000 to 1,000,000 acfm. These six sources are represented by
four industries, those being pulp and paper recovery boilers (three sources),
steam generating plants (one source), cement plants (one source), and incinera-
tors (one source). Only three of the six sources reported excess emissions
incidents. Of these three, one (a cement rotary kiln), reported incidents
related to more than one causal code, those being process, control 0&M( and
unforeseen events. This 50 percent reporting rate is less than the average
reporting rate (66 percent) for the sample population. With respect to the
average for all causal codes, sources of this category have fewer incidents
that are of lower relative magnitude than the average of the sample popula-
tions but they tend to last longer.
52
-------
The average credit surplus calculated for these larger sources was 52
percent of their allowable emission limitation. This relatively high percent-
age resulted from the fact that the sources of this category were found to
have normalized credits of 60 percent and excesses of only 8 percent, thus
indicating a high degree of continued control.
The breakdown of the 99.5 incidents reported for all causal codes is as
follows: 60.3 percent were process related, 34.7 percent resulted from control
operation and maintenance problems, 4 percent were unforeseen with the remain-
ing 1 percent due to design flaws. Concerning the various causal codes identi-
fied, the steam generating unit reported the only design-related incident.
The incinerator was found to have one of the two O&M-related incidents reported,
the cement plant having the other. In addition, the cement plant reported the
only process and unforeseen related incidents. None of the three pulp and
paper recovery boilers were found to have any excess emission excursions. The
average values of the incident indicators recorded for each of the four causal
codes and overall averages are presented in Table 5. Addressing only the
overall averages, these large sources have fewer incidents per year than the
average of the sample population, and they are lower in relative magnitude
but they tend to last longer.
Control Devices - Greater than 1,000,000 acfm
Only one source of the 119 surveyed had a control device that had a gas
flow rate of greater than 1,000,000 acfm. This source, an electric arc shop
controlled by a fabric filter (baghouse), reported a total of 15 incidents.
Of the incidents reported, six (40 percent) were due to design flaws, four
(26.7 percent) resulted from control O&M problems, and five (33.3 percent) were
unforeseen. The incident indicators for the three causal codes identified for
the source are shown in Table 5.
With the baghouse control device, the electric arc shop operated with an
annual credit surplus of 72 percent of the allowable emission limitation.
Normalized credits for the source were 73 percent and normalized excesses
were only 1 percent of the allowable emission limitation.
53
-------
RANKING OF PARAMETERS/INTEGRATION OF RANKINGS
A review of the ranking of each parameter is useful, as it points out
those categories which stand out with respect to high frequencies, long dura-
ations or large magnitudes. This review will logically follow the data presen-
tation and will proceed from the individual Causal Code rankings through
ranking of the summation of the Causal Codes and will conclude with the data
relative to normalized credits and normalized excess.
The relative severity of the overall excess emissions problem is more
clearly visible when we integrate the individual frequency, duration and mag-
nitude ranks. We will again proceed from individual Causal Codes to the sum-
mation of all causes and will highlight those areas that stand out. This inte-
gration will follow the ranking for each parameter.
Parameter A - Industrial Classification
Category Rankings—
Table 7 presents the rankings for causal code 1. Only 6 of the 16 cate-
gories encompassing 11 sources reported design problems. Virtually all (93
percent) of the frequencies in this code are found in two categories: grain
handling operations and iron and steel plants. Each of these categories reported
two sources with design problems, with one grain handling ductwork design prob-
lem accounting for 69 percent of all frequencies. Petrochemical plants are
the number one ranked category in the duration of design problems. One con-
tinuous emission incident at one plant led to the 6,000 hour upset in this
case. Steam generating plants and petroleum products facilities also reported
significant durations, with each category averaging more than 9 weeks of
continuous plant operation in the upset condition. Grain handling plants
also were ranked first with respect to magnitude of upset. The baghouses used
for particulate control at one grain elevator were partially bypassed during
certain process operations, resulting in a high rate of emissions that
greatly influenced the magnitude of incidents. In addition, steam generating
plants and petroleum products operations had incidents with emissions 7 to
10 times the allowable limit.
Process-related problems (Causal Code 2) were found to be more widespread,
with 12 separate categories involving 34 sources reporting problems. This data
is found on Table 8. Iron and steel plants ranked first in frequencies due to
a twice per day charging problem at one cupola operation. Steam generating
plants also had regular, recurring process upsets (> 2 per week) with one source
reporting an excess emission three times per day when it cleans the boiler
grates. This was the highest frequency due to any cause in the sample popula-
tion and accounted for 31 percent of all frequencies. Surface coating opera-
tions had the greatest episode duration, primarily due to one long term inci-
dent at one facility. Cement plants also had a significant average duration,
again resulting from one extended problem at one facility. The remainder of
54
-------
TABLE 7. PARAMETER A - INDUSTRIAL CLASSIFICATION
CATEGORY RANKINGS
CAUSAL CODE 1 (DESIGN).
in
in
Bank
I
2
3
4
5
6
Category
Frequency
Category
Duration
Cbr)
Category
Magnltudi
(X above
allowabli
Grain handling operation*
Iron and steel plants
PetroleuB products and handling
Petrochemical plants
Steaai generating plants
Stone, clay and glass plants
Pulp and paper Bills
Incinerators
Ceaent plants
Asphalt planta
AluatnuB plants
Brass and bronz* plants
Lumbar and wood plants
Surface coating operations
Food and drug plants
Other
36.3 Petrochemical plants
13.0 Steaa generating planta
1.2 PetroleuB products and handling
1.0 Stone, clay and glass planta
0.8 Grain handling operations
0.5 Iron and ateel plants
— Pulp and paper mills
— Incinerators
— Caaent plants
- Asphalt planta
— Aluadnun plants
- Brass and bronze plants
- Lumber and wood plants
- Surface coating operations
- Food and drug planta
Other
6000.0 Grain handling operations
2381.7 Steaa generating plants
1527.4 PetroleuB products and handling
264.0 Iron and steel planta
18.4 Stone, cley and glass plants
8.6 Patrochealcal plants
— Pulp and paper ailla
— Incinerators
— Ceaent planta
— Asphalt planta
— Alualnua planta
- Brass snd bronse plants
— Lumber and wood plants
— Surface coating operations
— Pood and drug planta
- Other
1346
985
729
139
108
83
-------
TABLE 8. PARAMETER A - INDUSTRIAL CLASSIFICATION
CATEGORY RANKINGS
CAUSAL CODE 2 (PROCESS).
Ul
Rank
Category
Frequency
Category
Duration
(hr)
Category
1 Iron and ateel plant! 374.0
2 Stean generating plant! 148.4
3 Pitlp and paper Bills 116.0
4 Ceaent plants 30.6
S Other 30.0
6 Petroleum product! and handling 16.1
7 Asphalt planta 10.0
8 Incinerators 7.3
9 Surface coating operations 5.4
10 AltmlnuB plants 3.5
11 Brass and bronze plants 1.0
12 Petrochemical plants 0.8
Stone, clay and glass plants -
Gi.iir handling operations —
Limber and wood plant* —
Pood and drug plant* -
Surface coating operations 85.3
Caatent plants 46.S
PetroleuB products and handling 8.5
Petrochemical planta 5.2
Pulp and paper Bill* 5.0
Other 3.8
Stean generating plants 3.1
Brass and bronze plants t.O
Asphalt plants 0.5
Iron and steel plants 0.4
Incineratora 0.2
Alualnua plants 0.1
Stone, cliv and gl- .•• plants -
Grain handling op.•• j: Jons -
LuBber «id wood pi.inia -
Food and drug plants -
Magnitude
(X above
allowable)
Cement planta 1949
Brass and bronze planta 11 SO
Surface coating operation* (-79
Incinerators 542
Petrochealcal plants 255
Pulp and paper Bills 180
Steaa generating plants 162
Iron and steel plenti US
Other 122
Petroleua products and hau.^.a^ 88
Asphalt plants 76
AluBlnua planta 50
Stone, clay and glass plants —
Grain handling operation* —
LuBber and wood plant! -
Pood and drug plane* -
-------
the facilities had similar length durations. Cement plants had the greatest
average magnitude of upset, primarily due to one of the five plants reporting
problems. The plant in reference lost total control when bypassed its control
system due to a potential explosive situation. The degree to which the cement
plant upsets exceeded regulatory limits (39.5 tiim>s the standard) was the
highest average for any causal code in Parameter A. Brass and bronze plants
also placed high in the magnitude ranking, again on the basis of one plant's
experience.
The greatest number of categories (14) reported control equipment problems
(Causal Code 3). Table 9 presents this data. Lumber and wood plants topped the
frequency ranking with cement plants and asphalt plants next in line. The one
petrochemical plant reporting control problems had one continuous episode which
gave it the highest rank for duration. This episode lasted approximately 3
months. Food and drug plants, petroleum products facilities and surface coating
operations all had significant average incident durations. It is interesting to
note that three of these four facilities are generally associated with hydro-
carbon emissions. Surface coating operations had the greatest incident magnitude,
and this reflects the complete loss of pollutant control at these sources when
a control problem occurs.
Unforeseen problems (Causal Code 4) occurred in nine separate categories
involving 25 sources. Refer to Table 10 for this data. No one industrial cate-
gory stands out relative to frequency of occurrence, and as might be expected
the average frequencies for this class of problems are the smallest of any of
the four Causal Codes. Cement plants had the highest average frequency, followed
closely by stone, clay and glass plants and iron and steel facilities. Surface
coating operations had the longest incident duration, with an average incident
lasting greater than 4 weeks. This was primarily due to natural gas curtail-
ments affecting several facilities with afterburners. Steam generating plants
(one source) also had durations lasting greater than 2 weeks. One sludge incin-
erator source reported a problem with its scrubber causing it to be ranked first
in incident magnitude. Outages at surface coating sources, again related to
gas curtailment, placed this category second in magnitude, with an excess emis-
sion rate which was 15 times the applicable standard.
When all Causal Codes are grouped together, we can distinguish those cate-
gories which are "incident prone." In this overview, we must remember that the
incident averages were derived using the data from all plants within the cate-
gory, including those sources which reported no excess emissions incidents. This
summary is found in Table 11.
Steam generating facilities and iron and steel plants are ranked first and
second with regard to frequency. Each of these categories have twice the number
of frequencies of any other category. Nine of the ten steam generating plants
reported some type of problem, although the high average frequency for this
category is overwhelmingly influenced by the 882 incidents reported by one source.
Nine of the ten iron and steel plants also reported some type of excess emission
and five of these plants had incidents attributable to more than one causal
factor. Iron and steel plant frequency data was also greatly influenced by one
cupola operation which reported 700 annual process related incidents.
57
-------
00
TABLE 9. PARAMETER A - INDUSTRIAL CLASSIFICATION
CATEGORY RANKINGS
CAUSAL CODE 3 (CONTROL O&M).
Bank
Category
Frequency
Category
Duration
(hr)
Category
Magnitude
(X above
allowable)
1 Luaber uod wood plants 28.8
2 Ceaent plants 20.3
3 Asphalt plants 14.8
4 I'on and steel plants 6.9
5 Stone, clay and glass plants 6.2
6 Petroleua products and handling 5.9
1 Other 5.3
8 Grain handling operations S.I
9 Aluainua plants 4.0
10 Incinerators 2.8
11 Surface coating operations 2.S
12 Brass and bronze plants ?.0
13 F»od and drug plantn . i
U PvtrocheBlcal plant 1 0
15 Steam generating pljnts
15 Pilp and paper Bills
Petrochemical plants 2160.0
Food and drug plants 233.0
Petroleusi products and handling 168.3
Surface coating operations 141.1
Stone, clay and glass plants 67.0
Aluainua planta 27.0
Luaber and wood plants 26.6
Other 18.6
Incinerators 17.1
Iron and steel plants 11.A
Ceaent plants 6.1
Grsln handling operations 4.9
Asphalt plants 3.9
Brass and bronze plants 2.0
Steam generating plants -
Pulp and paper mills —
Surface coating operations 318/
Brass and bronze plants 1150
Asphalt plants 63P
Incinerators 587
Ceaent planta 521
Grain handling operations 361
Stone, clay and glass plants 297
Iron and steel planta 76?
Petroleua products snd handling 174
Other Ui
Alualnun planta b7
Petrochemical plants 50
l.uaber and wood plants 3".
Food and drug plants 2.
Steaa generating plants -
Pulp and paper Bills -
-------
TABLE 10. PARAMETER A - INDUSTRIAL CLASSIFICATION
CATEGORY RANKINGS
CAUSAL CODE 4 (UNFORESEEN)
in
Rank
Category
Frequency
Category
Duration
(hr)
Category
Magnitude
(Z above
allowable)
1 Coaent plant* 4.3
2 Stone, clay and glass plants 3.5
3 Iron and ateel plants 3.3
4 PetroleuB products and handling 1.4
i Other 0.8
6 Surface coating operations 0.6
7 Steaa generating plants O.S
7 Incinerators O.S
9 Grains handling operations 0.1
Pulp and paper Bills -
Asphalt plant* *-
AluBlnuB plants —
10 ( Brass and bronze plants -
Petroehealcal plants -
Lunbei and wood plants —
\ Food and drug plants —
Surface coating operations 672.6
Steaa generating plants 336.0
Grain handling operations 168.0
PetroleuB products and handling 160.3
Other 115.2
Ceaent plants 47.1
Iron and steel plants 15.1
Stone, clay and glass plants 8.7
Incinerators 4.0
Pulp and paper Bills —
Asphalt plants -
Alum 1mm plants —
Bcass sod bronie plants —
Petrochealcsl plants —
Uiaber and wood plants -
Food and 'drug plants -
Incinerators 2998
Surface coating operations 1S33
Petroleiai product* and handling 548
Iron and steel plants 290
Oraln handling operations 274
Other 140
Ceownt plants 129
Stone, clay and glass plants 68
Steaa generating plants 12
Pulp and paper Bills ~
Asphalt planes -
AlualnuB plants ~
brass and bronze plants
PetrocheBlcsl plants -
LuBber and wood plants ~
Food and drug plants —
-------
TABLE 11. PARAMETER A - INDUSTRIAL CLASSIFICATION
CATEGORY RANKINGS
ALL CAUSAL CODES.
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
IS
16
Category
Frequency
Category
Duration
(hr)
Category
Stem generating plants
Iron and atcal planta
Pulp and paper ollla
i>nent planta
lumber and wood producta
Other
Grain handling operation!
Petroleum product* and handling
Asphalt plant*
Stone, clay and glass plant*
Incinerator*
Surface coating operation*
Alumlmra planta
Bras* and bronze plants
Petrochemical plant*
Pood and drug plant*
8».] Petrochemical planta 2554.0
S/..2 Food and drug plants 233.0
38.5 Surface coating operatlona 154.1
?•>.' Petroleum product* and handling 122.6
26 8 Incinerators 45.9
18.9 Stone, clay and glass plants 42.2
16.5 Cement plant* 31.2
9.4 Lumber and wood product* 26.6
6.5 Grain handling operation* 16.8
3.9 Aluminum plant* 12.8
3.4 Other 9.0
2.9 Steam generating planta 7.2
1.9 Pulp and paper mill* 4.9
0.6 Asphalt planta 2.8
0.6 Brass and bronze plante 1.7
0.5 Iron and steel planta 1.6
Magnitude
(Z above
allowable)
Cement planta 2463
Surface coating operatlona 1566
Grain handling operation* 1224
Brass and hronru planta 1150
Incinerators 614
Aaphalt'plant* 458
Stone, clay and glass planta 253
Pulp end paper mllla 180
Petroleum product* and handling 167
Steam generating planta 161
Iron and steel plant* 158
Petrochemical plant* 147
Other 123
Aluminum plants 59
Lumber and wood plants 34
Food and drug plants 25
-------
Petrochemical plants had incident durations which were 10 times those of
any other category. While only two of the six plants in this category had emis-
sions incidents (two per plant), each of the four reported excursions of greater
than 5 days, with the longest lasting 6000 hours. In addition, the four inci-
dents involved three separate Causal Codes so no trend was evident in this area.
Food and drug plants ranked second in duration of incident, however, this rank-
ing is based on only one incident at one facility. Of greater significance is
the third place ranking of surface coating operations. Nine of the thirteen
sources surveyed in this category experienced problems, with three of these
sources having multicausal incidents. If one includes the fourth ranked cate-
gory, petroleum products, then three of the four top sources are hydrocarbon
emitters. Since hydrocarbons are essentially invisible upon release, one may
surmise that the inability of affected area residents and regulatory plant
personnel to detect these emissions may play a part in the long duration of
each upset episode.
Cement plants ranked first with regards to magnitude of emission during an
incident. Seven of the nine plants surveyed had problems with four of these
plants experiencing multicausal incidents. The high degree of control required
to bring cement plants into compliance (in excess of 99.5 percent) relates to
the large magnitude of Incidents, as the complete loss of control equipment will
result in an emission which may be 500 to 1000 times the applicable regulation.
This was the case with one of the plants surveyed. Surface coating operations,
grain handling operations and brass and bronze plants all experienced incident
magnitudes in excess of 1000 percent, indicating that a problem with their con-
trol equipment typically resulted in complete bypassing of that control device.
Integration of Rankings—
The Causal Code 1 integration is provided in Table 12. As can be seen,
grain handling operations are ranked first due to their having the greatest
number of frequencies and the largest magnitude of emission. A further indi-
cation of the connection between poor design and grain handling pollution con-
trol is the fact that 40 percent of the grain handling sources surveyed (2 of 5)
indicated they experienced an emissions incident directly related to poor de-
sign. This is the highest design-related percent of any category. While the
magnitude of these incidents is relatively small, the high uncontrolled emission
rate of grain transfer leads to a high magnitude of emission release when a
problem does develop. Following grain handling operations, steam generating
plants and petroleum products facilities had identical integrated ranks. The
placement of steam generating facilities was due to the long duration and large
magnitude of their incidents while petroleum products was ranked third in each
of the individual categories. While petrochemical plants had the longest
duration of incident, they experienced relatively fewer problems and the
magnitude of emission was less than double the allowable standard.
The integrated ranking for Causal Code 2 is found in Table 13. Cement
plants were by far the number one ranked category with respect to process prob-
lems. This was due to the large magnitude of emissions released during an inci-
dent an
-------
to
TABLE 12. PARAMETER A - INDUSTRIAL CLASSIFICATION
INTEGRATION OF RANKINGS
CAUSAL CODE 1 (DESIGN).
Ni
Rank Category
SI
1 Grain handling operations
2 Steam generating plants
2 Petroleum products and handling
4 Petrochemical plants
5 Iron and steel plants
6 Stone, clay and glass plants
7 <
Pulp and paper mills
Incinerators
Cement plants
Asphalt plants
Aluminum plants
Brass and bronze plants
Lumber and wood plants
Surface coating operations
Food and drug plants
Other
Total -
Individual ranking
of nni-aUrm Magnitude
amples Frequency uu*ac*on (% above
1 ; allowable)
2151
252 2
3333
1416
2264
1645
0
0
0 - - -
0 - - -
0 - - -
0
0 - - -
0 - - -
0 - - -
0 - - -
11
Eall
ranking
7
9
9
11
12
15
—
—
—
—
—
—
—
—
—
—
-------
u>
TABLE 13. PARAMETER A - INDUSTRIAL CLASSIFICATION
INTEGRATION OF RANKING
CAUSAL CODE 2 (PROCESS).
— .
Rank
1
2
3
4
5
5
7
8
8
10
11
12
13
Category
Cement plants
Surface coating
operations
Pulp and paper mills
Steam generating plants
Petroleum products and handling
Iron and steel
Other
plants
Petrochemical plants
Brass and bronze plants
Incinerators
Asphalt plants
Aluminum plants
Stone, clay and
Grain handling
Lumber and wood
glass plants
operations
plants
Food and drug plants
Total
Number
of
samples
5
4
2
6
4
2
3
2
1
2
2
1
0
0
0
0
= 34
Individual ranking
Frequency
4
9
3
2
6
1
5
12
11
8
7
10
—
—
—
—
Duration
(hr)
2
1
5
7
3
10
6
4
8
11
9
12
—
—
—
—
Magnitude
(% above
allowable)
1
3
6
7
10
8
9
5
2
4
11
12
—
—
—
—
Zall
ranking
7
13
14
16
19
19
20
21
21
23
27
34
—
—
—
—
-------
these plants and the duration of a process upset can be attributed to the com-
plexity of chemical reactions that takes in a cement kiln and the time frames
that may be needed to sort out these problems. Surface coating operations, on
the strength of the long duration of their incidents, and pulp and paper mills,
due to their consistent high ranking in all categories, were integrated second
and third with respect to process-related upsets. While iron and steel plants
had the most frequent number of process upsets, these problems were of rela-
tively short duration and low magnitude and on the whole were not as severe as
in other categories.
The integration of control-related problems (Causal Code 3) is presented
in Table 14. No one category stands out in this area as having a significantly
greater problem than the rest. Surface coating operations is ranked first,
largely due to the high magnitude of its emissions incidents. This magnitude
rank is indicative of the go/no-go status of control equipment at these hydro-
carbon sources. It appears that when a problem develops with surface coating
controls, the equipment becomes completely inoperative and is bypassed. Stone,
clay and glass plants, petroleum products plants and cement plants all were
ranked closely behind surface coating operations, although none of these cate-
gories ranked first with respect to the incident descriptors. While lumber and
wood plants had the greatest incident frequency and petrochemical plants had the
longest duration, the ranking of these categories with respect to the other
descriptors was low and overall these categories did not have a significant
control device problem.
The unforeseen (Causal Code 4) problem integration is presented in Table
15. Surface coating operations lead the category rankings in this Causal Code,
due to their long incident duration and second ranking with respect to magni-
tude. As discussed earlier, both these factors are directly attributable to
natural gas shutoffs affecting afterburner performance. Petroleum products
ranked second overall, on the basis of a consistent high ranking with each in-
cident indicator. Cement plants placed third, due to their relatively high
frequency of unforeseen incident occurrence. While incinerators had the great-
est magnitude of emission, their problems were infrequent and relatively short
lasting so as to not be a significant problem.
The integration of the rankings for all Causal Codes combined is perhaps
the most informative table for each parameter as it not only sums the data for
all incidents, but also reflects the input of those sources which reported no
excess emissions problems. We must remember that only the average frequency
of incident occurrence will be affected by the addition of sources with no
reported problems, as the total number of incidents in a category will be now
divided by the total number of sources surveyed in that category and not just
those sources with problems. The average duration of each incident and the
magnitude of pollutant released during an incident are indicators which are
functions of the total number of incidents, and not the total number of sources
thus the affect of sources reporting no excess emissions incidents will be seen
in only one of the three indicator rankings which form our overall integration.
Table 16 presents the integration of rankings of all Causal Codes for Parameter
A. Cement plants are the industrial classification that have the greatest over-
all excess emissions incidents problem. Incidents at these facilities account
for 9.5 percent of all frequencies, 15.3 percent of all hourly durations, and
64
-------
a\
Ui
TABLE 14. PARAMETER A - INDUSTRIAL CLASSIFICATION
INTEGRATION OF RANKINGS
CAUSAL CODE 3 (CONTROL O&M).
Rank
1
2
3
3
5
6
7
8
9
9
11
12
13
14
15
15
Category
Surface coating operations
Stone, clay and glass plants
Petroleum products and handling
Cement plants
Asphalt plants
Lumber and wood plants
Iron and steel plants
Incinerators
Grain handling operations
Other
Aluminum plants
Petrochemical plants
Brass and bronze plants
Food and drug plants
Steam generating plants
Pulp and paper mills
Total
Number
of
samples
5
5
8
5
3
3
8
2
2
4
1
1
1
1
0
0
= 49
Individual ranking
Frequency
11
5
6
2
3
1
4
10
8
7
9
14
12
13
—
—
Duration
(hr)
4
5
3
11
13
7
10
9
11
8
6
1
14
2
—
—
Magnitude
(% above
allowable)
1
7
9
5
3
13
8
4
6
10
11
12
2
14
—
—
Zall
mnV in§
16
17
18
18
19
21
22
23
25
25
26
27
28
29
—
—
-------
TABLE 15. PARAMETER A - INDUSTRIAL CLASSIFICATION
INTEGRATION OF RANKINGS
CAUSAL CODE A (UNFORESEEN).
Rnnl'
1
2
3
3
5
6
6
8
8
10
Categorv
Surface coating
operations
Petroleum products and handling
Cement plants
Tron and steel plants
Other
Grain handling operations
Incinerators
Stone, clay and
Steam generating
glass plants
plants
Pulp and paper mills
Asphalt plants
Aluminum plants
Brass and bronze
plants
Petrochemical plants
Lumber and wood
plants
Food and drug plants
Total
Number
of
samples
5
5
3
A
3
1
1
2
1
0
0
0
0
0
0
0
- 25
Individual ranking
Frequency
6
A
1
3
5
9
7
2
7
—
—
—
—
—
—
—
Duration
(hr)
1
A
6
7
5
3
9
8
2
—
—
—
—
—
—
—
Magnitude
(% above
allowable)
2
3
7
A
6
5
1
8
9
—
—
—
—
—
—
—
Zall
ranking
9
11
1A
1A
16
17
17
18
18
—
—
—
—
—
—
—
-------
TABLE 16. PARAMETER A - INDUSTRIAL CLASSIFICATION
INTEGRATION OF RANKINGS
ALL CAUSAL CODES.
Rank
I
2
3
4
4
6
6
8
9
9
11
11
13
14
15
16
Category
Cement plants
Surface coating
Grain handling operations
Petroleum products and handling
Incinerators
Stone, clay and glass plants
Steam generating plants
Pulp and paper mills
Petrochemical plants
Lumber and wood plants
Asphalt plants
Iron and steel plants
Other
Brass and Bronze plants
Food and drug plants
Aluminum plants
Total
Number
of
samples
9
13
5
13
6
10
10
6
6
3
10
10
6
5
3
4
- 119
Individual ranking
Frequency
4
12
7
8
11
10
1
3
15
5
9
2
6
14
16
13
Duration
(hr)
7
3
9
4
5
6
12
13
1
8
14
16
11
15
2
10
Magnitude
(% above
allowable)
1
2
3
9
5
7
10
8
12
15
6
11
13
4
16
14
Zall
ranking
12
17
19
21
21
23
23
24
28
28
29
29
30
33
34
37
-------
53.5 percent of all percentage magnitudes in excess of the regulatory limits.
The high uncontrolled emission rate of these facilities is the prime reason for
this number one ranking, for even a relatively minor problem with the control
equipment can lead to an excess emission. Since 78 percent of those facilities
surveyed reported some type of incidents, this ranking cannot be attributed to
a random event or an unusual, atypical situation. The extent of the problem
at cement plants if further highlighted when compared to the second ranked cate-
gory, surface coating operations. As a group, these facilities account for only
1.3 percent of all frequencies, 10.5 percent of all hourly durations and 4.7 per-
cent of all magnitudes. Grain handling operations and petroleum products facil-
ities are ranked third and fourth, respectively. It is interesting to note that
only three industrial categories generally associated with hydrocarbon emissions
were surveyed for this study (petroleum products and handling, petrochemical
plants, surface coating operations) yet two of these three are among the top
four categories with respect to excess emissions problems. This observation
is substantiated by the fact that 69 percent of the surface coating operations
and 92 percent of the petroleum products facilities that were surveyed reported
emissions incidents. Steam generating facilities, which accounted for 31.7
percent of all frequencies was only ranked sixth overall due to relatively
short incident durations and small incident magnitudes. Similarly, petrochemical
plants, which ranked first with respect to incident duration and accounted for
16.4 percent of all hours in excess of the standard, ranked only ninth overall,
due to a low frequency and a relatively small magnitude. This low overall
ranking is confirmed by the fact that only two of the six plants surveyed in
this category reported emissions incidents.
Normalized Credits - Normalized Excess—
A final review of emissions data concerning Parameter A must also include
the rankings and integration of data concerning normalized credits and normalized
excess. This summary is presented in Table 17. The first column of this table
presents the normalized credit data. This column represents the extent to which
a category is operating below the applicable regulatory emission limitation (its
credits) and expresses that extent in terms of percent of allowable emissions.
Petroleum products facilities are ranked first in this area by virtue of the
fact they normally operate closest to the allowable limit, on an annual basis.
They are more likely to exceed the emission limit should any problem, however
minor, occur and thus have less margin of error in the operation and maintenance
of their control equipment if they are to avoid excess emissions incidents. At
the lower end of this ranking are grain handling facilities. Their normalized
credit value of 0.81 indicates that their credits are 81 percent of their allow-
ables, or conversely that they typically operate at an emission rate which is
only 19 percent (1.00 - 0.81) of the allowable limit. Ideally, these facili-
ties can afford to have minor upsets with process and/or control equipment and
still be continually in compliance. As we shall see, this is not always the
case.
The second column in Table 17 concerns normalized excess. Here again we
have taken the annual excess emissions for each facility and expressed them in
terms of percent of allowable emissions. Grain handling operations are the
worst category in this regards and rank number one. Excess emissions with these
facilities are equal to 82 percent of allowables on an annual basis. Ranked
last in this group are asphalt plants which have an excess emission of only
68
-------
VO
Bank
TABLE 17. PARAMETER A - INDUSTRIAL CLASSIFICATION
RANKING NORMALIZED CREDITS AND EXCESSES.
Category
Nonsal ised
credits
(Z of
•Ilovable)
Category
Normalised
exceta
(Z of
allowable)
1 Petroleum products and handling 0.3S
2 Limber and wood producta 0.36
-> Ceaent plant! 0.37
4 Food and drug plant* 0.40
5 Pulp and paper nllls 0.47
5 Incinerator! 0.47
7 Steaa generating planta 0.48
7 Iron and eteel planta 0.48
9 Surfece coating operations 0.50
10 Aluminum planta 0.57
11 Asphalt plant* 0.58
12 Brat* and bronse plants 0.59
12 Petrochemical plants 0.64
14 Other 0.65
IS Stone, clay and glass planta 0.68
16 Grain handling 0.81
Grain handling 0.82
Petroleum product* and handling 0.62
Staasi generating plants 0.48
Surface coating operations 0.44
Petrochenlcal plants 0.25
Iron and ateel planta 0.16
Cement plant* 0.15
Stone, clay and glaas planta 0.08
Incinerator* 0.02
Other 0.014
Pulp and paper mill* 0.01
Alumlnun planta 0.004
lumber and wood plants 0.003
Brass and bronze planta 0.002
Pood and drug plants 0.001
Aiphalt plant* <0.001
Category
Credits -
ox^eas
(Z of
allowable)
Petroleum products and handling -0.2~
Grain handling -0.01
Stean generating plants 0.00
Surface coating operation* •''.('*>
Ceoient plant* 0.22
Iron and ateel planta u.J2
LuBber and wood producta 0.36
Petrochemical planta 0.39
Pood and drug plants 0- 0
Incinerator* 0.45
Pulp and paper Bills 0.46
AluainuB plant* °-*7
Asphalt planta • .5n
Brasa and bronie plants u.59
Stone, clay and glass plants 9.60
Other n.6«i
-------
1/10 of 1 percent of allowables. This confirms the individual Causal Code
rankings and integrations which generally show that asphalt plants do not have
a major excess emissions problem.
The final column in Table 17 integrates the rankings of the first two col-
umns. Here we subtract the normalized excess from the normalized credits, and
arrive at a figure which shows the net annual credit or excess for an indus-
trial category. These resultant numbers are ranked with regard to severity
to the ambient air quality, a negative number indicating a greater excess than
credit and hence, a greater problem. To maintain our convention that the lower
the numerical rank, the greater the problem with continued source compliance,
the categories are ranked with the greatest negative number first and proceeding
to the highest positive number. Since we are expressing normalized data, the
results may be considered as expressing a percent of allowables. Thus petro-
leum products plants, with a credits minus excess value of -0.27 have an annual
excess emission which is equal to 27 percent of the average annual emission rate
for that category. It is interesting to note that only two of the 16 categories
had a negative difference, therefore only these two had annual excess emissions
which were not completely offset by emission credits. The second of these was
grain handling operations which had the best normalized credit ranking, yet more
than offset all of these credits with a high annual excess emission. Our overall
highest ranked category with respect to excess emissions incidents, cement plants,
demonstrated that all of its upsets were completely offset by credits, with an
additional credit equal to 22 percent of the allowable emissions to spare.
Parameter B - Source Size
The size of each source, as expressed by the uncontrolled emission rate
in tons per year (TPY) of pollutant, was investigated. This annual uncon-
trolled rate was calculated using the applicable hourly emission rate and the
hours per year that the specific process was operated. The uncontrolled emis-
sion rate of a source is a value that is reported on most comprehensive data
bases (NEDS, E.I.S., etc.). Since our study population covered a wide range
of industrial types and included the major industrial pollutant emitters, the
results of this parameter analysis may indicate trends that have universal
application. Large sources (those with uncontrolled emission rates in excess
of 100 TPY) were studied as were those minor sources with uncontrolled rates
of less than 100 tons per year. An attempt was made by GCA to divide this
parameter into easily identifiable size categories as well as categories size
cuts which divided the data base somewhat equally. This category division
was therefore subjectively and can be changed, using the raw data in Appendix
A, to suit the readers' specific needs.
Category Rankings—
The ranking of the six size categories for Causal Code 1 is presented in
Table 18. All categories contained at least one source with design-related
problems except for the largest size cut, 100,000 - 500,000 TPY. As the
table indicates, there is a wide range between the first and last ranked ca-
tegories for each indicator. The 10,000 - 100,000 TPY category far exceeds
the others with respect to frequency. Four of the 11 sources reporting de-
sign problems are within this category and 94 percent of all incidents are
in this source size range. Two categories stick out in the duration ranking.
70
-------
The 500 to 1000 TPY category ranked first, has an average duration that exceeds
205 days continuous operation and is based on data from two sources. The 100
to 500 TPY category ranks second based on the data of one source survey. Each
of these categories has an average duration that is seven times as great as
the third ranked category. The 10,000 to 100,000 TPY category is notable in
the relatively short duration of its emission incidents. The first ranked
category for magnitude of design problems is the 1,000 to 10,000 TPY size cut.
It far exceeds all other categories and has an average emission that is 64
times the applicable standard. The influence of one incident at a grain hand-
ling facility is strongly reflected in this average magnitude. The relatively
low magnitude of the smallest size category is noteworthy.
TABLE 18. PARAMETER B -
SOURCE SIZE UNCONTROLLED EMISSIONS
CATEGORY RANKINGS
CAUSAL CODE 1 (DESIGN).
Rank
1
2
3
4
5
6
Category
(103 ton/yr)
10 - 100
.1 - .5
1 - 10
.5 - 1
<.l
100 - 500
Frequency
24.8
1.5
0.8
0.8
0.5
—
Category
(103 ton/yr)
.5 - 1
.1 - .5
1 - 10
<.l
10 - 100
100 - 500
Duration
(hr)
4921.7
2920.0
400.8
264.0
37.5
—
Category
(103 ton/yr)
1 - 10
.5 - 1
.1 - .5
10 - 100
<.l
100 - 500
Magnitude
(% above
allowable)
6383
994
900
873
108
—
Causal code 2 rankings are presented in Table 19. Once again, the top
ranked categories for each indicator clearly stand out from the rest. All
categories include some sources with code 2 problems although the 1,000 to
10,000 size cut has by far the largest number, with 13 of the 34 sources that
reported process problems. The 500 to 1000 TPY category has the highest aver-
age frequency of occurrence, with almost twice as many as the number 2 rank.
The six sources in this size range that reported incidents included the single
highest frequency, 882, for any source in any category. Without this source,
this category would have an average of 22.5 incidents and would rank third.
The 10,000 to 100,000 TPY and 1,000 to 10,000 TPY size cuts also have signifi-
cant (> one per week) frequencies. An even greater difference is displayed by
the first and second ranked duration categories. The 100,000 to 500,000 TPY
category had an average incident duration that was equivalent to 46 days of
continuous process operation and was based on two source reports, one of which
was a year-long outage. It is interesting to note that the first ranked cate-
gory in frequency is the last rank in duration, implying that while there are
many episodes in this 500 to 1000 TPY size cut, each incident lasts only 1/2
hour. The greatest disparity between first and second place ranks for any
causal code and any parameter is found with the magnitude indicator for process
problems. Here, the 100,000 to 500,000 TPY category included five incidents
at one cement plant. Since this plant normally requires a control efficiency
in excess of 99.91 percent to meet the applicable standard, when it operates
71
-------
uncontrolled, the magnitude of the emissions exceed 100,000 percent, and this
enormous magnitude is reflected in the average for this size cut. The other
categories in this indicator have relatively small magnitudes, with only one
additional size cut, the smallest, having magnitudes that are greater than twice
Che standard.
TABLE 19. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
CATEGORY RANKINGS
CAUSAL CODE 2 (PROCESS).
Magnitude
Category _ Category Duration . Category
(I03ton/yr) requency (I03ton/yr) (hr)
1
2
3
4
5
6
.5
10
1
4
100
.1
- 1
- 100
- 10
c.l
- 500
- .5
165.8
86.0
71.9
10.3
3.0
2.1
100 -
<.]
1 -
.1 -
10 -
.5 -
500
L
10
.5
100
1
1106.0
23.9
7.65
6.0
4.5
0.6
100 -
<.]
10 -
.5 -
1 -
.1 -
500
L
100
1
10
.5
96,588
520
164
161
154
147
The category rankings for causal code 3 problems are found in Table 20.
Forty-nine sources reported control equipment-related problems, with the
smallest size category having the highest percentage of sources that reported
this type of problem (54 percent). All categories had similar average fre-
quencies with the 1,000 to 10,000 TPY size cut ranking first. The 100 to 500
TPY category was notable in this comparison in that it had relatively few inci-
dents (less than two per year). The smallest size category had the longest
average duration, with the 13 reporting sources having durations ranging from
0.5 to 5,112 hours. The average duration of this category was 2.5 times higher
than the second ranked size cut. It is interesting to note that the two cate-
gories with the longest durations were the same two with the lowest number of
frequencies. This inverse relationship constantly presented itself in the
data. If a source reported many frequencies, they very often were of short
duration, and the long duration episodes were often once-per-year events. The
magnitude of the 100 to 500 TPY category was sufficiently high to rank this
source first. All of the eight sources in this category reporting problems
had magnitudes in excess of 500 percent above the allowable standard. The
largest size category 100,000 to 500,000 TPY also had a significantly high
average magnitude, some 16 times the allowable standard.
The rankings for causal code 4 grobiems is presented in Table 21. The
25 sources that reported unforeseen problems are divided among all six cate-
gories, although the two largest size cuts reported on only one source each.
The frequencies for these two large size categories were sufficient to rank
them first and second. Very little separated the first and last ranked cate-
gories with respect to frequency. The 100 to 500 TPY category ranked first
72
-------
TABLE 20. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
CATEGORY RANKINGS
CAUSAL CODE 3 (CONTROL O&M).
Rank
1
2
3
4
5
6
Category
(103 ton/yr)
1-10
.5 - 1
100 - 500
10 - 100
<.l
.1 - .5
Frequency
12.1
10.9
10.8
8.8
7.4
1.9
Category
(103 ton/yr)
<.l
.1 - .5
1 - 10
.5 - 1
100 - 500
10 - 100
Duration
(hr)
106.6
41.5
30.6
30.0
11.2
7.3
Category
(103 ton/yr)
.1 - .5
100 - 500
1 - 10
10 - 100
<.l
.5 - 1
Magnitude
(% above
allowable)
3023
1631
322
252
162
133
TABLE 21.
PARAMETER B
- SOURCE SIZE
UNCONTROLLED EMISSIONS
CATEGORY RANKINGS
Rank
1
2
3
4
5
6
Category
(103 ton/yr)
100 - 500
10 - 100
.1 - .5
1 - 10
<.l
.5-1
Frequency
5.0
4.0
1.7
1.5
1.5
1.3
CAUSAL CODE
Category
(103 ton/yr)
.1 - .5
1 - 10
.5 - 1
100 - 500
<.l
10 - 100
4 (UNFORESEEN).
Duration
(hr)
246.4
95.0
49.2
45.6
26.0
21.5
Category
(103 ton/yr)
<.l
.1 - .5
.5 - 1
1-10
10 - 100
100 - 500
Magnitude
(% above
allowable)
523
413
391
375
138
15
73
-------
with respect to incident duration. Three of the six reporting problems in
this size cut had incident durations that exceeded 19 days continuous source
operation. No other category had noticeably long incidents. The magnitude
indicator for unforeseen problems is the only incident indicator for any causal
code in parameters B and D, which demonstrated a complete correlation between
category size and magnitude of incidents. The smaller size categories had the
highest magnitude and each succeeding size category had a lower magnitude, with
the largest size cut having a notably small average magnitude. There was little
difference between magnitudes of any size cut, with the exception of the largest
size category (100,000 to 500,000 TPY).
The ranking of categories for all causal codes is presented in Table 22.
Each indicator in this ranking can be divided into high and low ranges. For
frequencies, these ranges comprise the middle sized categories, all of which
have relatively high frequencies and the extreme categories, two low and one
high, which have relatively few episodes per year. The 500 to 1,000 TPY size
cut had by far the highest average frequency, followed by the 1,000 to 10,000
TPY and 10,000 to 100,000 TPY categories. Each of these size cuts has more
than seven times the number of frequencies of the fourth ranked group. The
ranking of incident duration has the 100,000 to 500,000 TPY category ranked
first, with the 100 to 500 TPY size cut a close second. The'former category
is based on data from three sources, while the latter had 19 sources with some
type of emissions problem. The smallest size category is ranked third and its
average duration was six times that of the other categories. The 100,000 to
500,000 TPY category also ranks first with respect to magnitude on the basis
of the extremely high emissions released during five incidents at a cement
plant. The other category that reported magnitudes in excess of 10 times the
applicable standard was the 100 to 500 TPY size cut.
TABLE 22. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
CATEGORY RANKINGS
ALL CAUSAL CODES.
Rank
1
2
3
4
5
6
Category
(103 ton/yr)
.5
1
10
4
100
.1
- 1
- 10
- 100
f.l
- 500
- .5
Frequency
68.1
37.3
36.4
5.5
5.4
1.3
Category
(103 ton/yr)
100
.1
4
10
1
.5
- 500
- .5
- 100
- 10
- 1
Duration
(hr)
218.0
181.6
85.0
13.2
11.5
11.4
Category
(103 ton/yr)
100
.1
10
4
1
.5
- 500
- .5
- 100
- 10
- 1
Magnitude
(% above
allowable)
18,913
1,292
571
225
196
161
74
-------
Integration of Rankings —
The integration of causal code 1 rankings is presented in Table 23. As
is shown in this table, a three-way tie existed in this integration between
the three middle sized categories. Nothing distinguishes any of these cate-
gories from the others, and all are ranked high for different reasons. The
500 to 1,000 TPY category had the longest duration and the second highest mag-
nitude but averaged less than one incident per year. The 100 to 500 TPY size
cut ranked second in frequency, even though the average occurrence was rela-
tively small, and had long durations and high magnitudes. Finally, the 1,000
to 10,000 TPY category had by far the highest magnitude, yet had a low fre-
quency and only an average relative duration.
TABLE 23. PARAMETER B -
SOURCE SIZE UNCONTROLLED EMISSIONS
INTEGRATION OF RANKINGS
CAUSAL CODE 1 (DESIGN).
Rank
Number
of
Individual ranking
, n .•
ton/yr) samples Frequency Duration
all
ranking
allowable)
1
1
1
4
5
6
.5
.1
1
10
100
- 1
- .5
- 10
- 100
.1
- 500
2
1
3
4
1
0
4
2
3
1
5
6
1
2
3
5
4
6
2
3
1
4
5
6
7
7
7
10
14
18
Total = 11
The integration of causal code 2 rankings is found in Table 24. The
largest category 100,000 to 500,000 TPY ranked first based on its number one
rank with respect to both duration and magnitude. While there were only three
incidents per year for this category, each incident was extremely long and had
emissions that were extremely high relative to standards. This ranking high-
lights the fact that large sources always have the potential for high excess
emissions, and should a problem occur which completely disables the control
equipment and this problem goes unsolved, a serious episode will follow.
Small facilities, with sizes less than 100 tons, ranked second overall, although
their average durations and magnitudes do not compare with the large sources.
While the 500 to 1,000 TPY category had the most incidents per year, each one
is short in duration and low in magnitude, and this category ranked fifth
overall.
Table 25 presents the integration for causal code 3 problems. For con-
trol-related operating and maintenance problems, the 1,000 to 10,000 TPY cate-
gory ranks first. This size cut barely ranked first in frequencies and had
75
-------
TABLE 24. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
INTEGRATION OF RANKINGS
CAUSAL CODE 2 (PROCESS).
Individual ranking
Rank
1
2
3
4
4
5
Category f
d03ton/yr) samples
100 - 500 2
<.l 3
10 - 100 3
1-10 13
.5-1 6
.1 - .5 7
Total = 34
Frequency
5
4
2
3
1
6
Duration
(hr)
1
2
5
3
6
4
Magnitude
(% above
allowable)
1
2
3
5
4
6
Zall
ranking
7
8
10
11
11
16
TABLE 25. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
INTEGRATION OF RANKINGS
CAUSAL CODE 3 (CONTROL O&M).
- -
Rank
1
2
3
4
5
6
_ t Number
Category
OO'ton/yr) samples
1 - 10
.1 - .5
100 - 500
<.l
.5 - 1
10 - 100
Total «
14
8
2
13
8
4
= 49
Individual ranging
Frequency
1
6
3
5
2
4
Duration
(hr)
3
2
5
1
4
6
Magnitude
(% above
allowable)
3
1
2
5
6
4
V* all
^ ranking
7
9
10
11
12
14
76
-------
only relatively average duration and magnitudes, yet was first overall due to
the lack of a consistent showing in all indicators by any one category. While
the 100 to 500 TPY category ranked first in magnitude, it ranked last in fre-
quency and had an average duration that ranked it second in this indicator.
Overall this category ranked second. Finally, the smallest size category,
which ranked first in duration, placed only next to last in both frequency and
magnitude and was ranked fourth overall. This entire causal code had no clear
cut problem category.
The integration of category rankings for causal code 4 are presented in
Table 26. This integration does have one standout category. The 100 to 500
TPY size cut was the only one that ranked in the top three for all indicators
and placed first overall. The only indicator in this causal code to have a
wide numerical separation between values was that for duration and the 100 to
500 TPY category ranked first here. A third place rank in frequencies and a
second place rank in magnitude ensured the overall number one ranking for
this category. The 100,000 to 500,000 TPY category barely ranked first for
frequency yet ranked fourth in duration and last in magnitude and was ranked
fourth overall. Similarly, the smallest category, sources less than 100 TPY,
ranked first in magnitude yet fifth in both frequency and duration and was
listed third overall.
TABLE 26. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
INTEGRATION OF RANKINGS
CAUSAL CODE 4 (UNFORESEEN).
Number
Individual ranking
Rank
Category .
(103 ton/yr) °f1 _ Duration Magnitude
J ' samples Frequency f^^ (% above
allowable)
all
ranking
1
2
3
3
4
5
.1
1
4
100
.5
10
- .5
- 10
- 500
- 1
- 100
6
8
3
1
5
1
3
4
5
1
6
2
1
2
5
4
3
6
2
4
1
6
3
5
6
10
11
11
12
13
Total = 25
An intepration of the rankinp.s for all causal codes is presented in Table
27. Again, there is a clear cut number one rank. The largest category,
100,000 to 500,000 TPY, is first, based on individual number one ranks in
the duration and magnitude indicators. Three of six sources in this category
reported problems, and the data from 6.5 incidents formed the averages upon which
the two number one ranks was based. It is interesting to note that the frequency
number one went to the 500 to 1.000 TPY category, yet that size cut ranked
last overall due to last placed rankings in duration and magnitude.
77
-------
TABLE 27. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
INTEGRATION OF RANKINGS
ALL CAUSAL CODES.
Individual ranking
Rank
Category
LO3 ton/yr)
Number
of
_. . . Magnitude /,
i r. Duration ,„, , *-^
samples Frequency /•UN '^ ab°ve
lhr; allowable)
all
ranking
1
2
2
4
5
6
100
.1
10
1
500
- 500
- .5
- 100
.1
- 10
- 1
6
31
11
24
30
17
5
6
3
4
2
1
1
2
4
3
5
6
1
2
3
4
5
6
7
10
10
11
12
13
Total = 119
Normalized Credits/Normalized Excess—
Data on normalized credits and normalized excess are presented in Table
28. All categories had normalized credits that exceeded 40 percent in the
allowable annual emission level. The two largest size categories had the
lowest normalized credits, and hence the greatest potential problems.
While all categories had excesses which were greater than 10 percent of
allowables, the ranking of the individual normalized excess values shifts
between different categories.
All categories had emission surpluses after excesses had been subtracted
from credits. The smallest size category had the greatest surplus while the
second smallest size cut had the least surplus. This data was also compared
to the integration ranking from all causal codes to see if some consistency
was apparent between problem categories. While the rankings were not similar,
two of the top three categories were identical. Both the 100 to 500 TPY and
the 10,000 to 100,000 TPY categories were ranked among the top three in both
tables. This suggests that these two categories are the main problem areas
with regard to specific excess emissions incidents and the degree to which
these incidents affect annual emissions surplus.
78
-------
TABLE 28. PARAMETER B - SOURCE SIZE UNCONTROLLED EMISSIONS
RANKING OF NORMALIZED CREDITS AND EXCESSES.
Rank
order
1
2
3
4
5
6
Category
(103 ton/yr)
10
100
.1
1
- 100
- 500
- .5
<.l
- 10
.5 - 1
Normalized
credits
(% of
allowable)
0.43
0.44
0.47
0.55
0.56
0.58
Category
(103 ton/yr)
.1
1
10
.5
100
- .5
- 10
- 100
- 1
- 500
<.l
Normalized
excess
(% of
allowable)
0.35
0.28
0.26
0.18
0.15
0.11
Category
(103 ton/yr)
.1
10
1
100
.5
- .5
- 100
- 10
- 500
- 1
<.l
Credits -
excess
(% of
allowable)
0.12
0.17
0.28
0.29
0.40
0.44
-------
Parameter C - Control Type
The general type of air pollution control device used at the sources
surveyed provides the next basis for comparison of excess emissions inci-
dents. We have limited the number of devices studied to four generic types:
• Electrostatic precipitators (ESPs)
• Wet scrubbers
• Fabric filters (baghouses)
• Other
Incorporated in the "other" section are conventional, though less common,
devices such as cyclones and afterburners as well as newer more complex con-
trols including mist eliminators and vapor recovery units. While combining
devices minimizes the specific conclusions that can be drawn from data on these
controls, we will benefit by having fewer total control categories. The bias
built into a data set which reports the results of only a few source surveys
will be minimized by limiting to four, the number of devices Investigated.
With our 119 source sample population, we will have roughly 30 sources in each
category, and extremes of performance will be reduced.
Category Rankings—
Causal Code 1, design data, is ranked on Table 29. Four sources, two
iron and steel plants and two grain handling operations reported design prob-
lems with baghouses. The repeated frequency of occurrence of these problems
caused baghouses to rank first in frequencies, with some 20 times more inci-
dents than the second category "other." Since this Causal Code had the fewest
data entries, 11, it can be unduly influenced by one source, and in fact, one
baghouse provided 73 percent of all incident frequencies. Even without this
source, however, baghouses would have ranked first for this incident indicator.
"Other" devices ranked first in duration of design related incidents, although
scrubber durations were a close second. Both categories had average durations
in excess of 115 days of continued source operation. Problems with a carbon
absorption system and two vapor recovery systems lead to the exceptionally long
duration experienced in the "other" category. Poor materials of construction
utilized on a utility boiler scrubber lead to one continuous outage and a
resultant long duration/high magnitude emission excursion for this control
category. Baghouses are notable in their relatively short incident duration
(only 1.4 percent the duration of the closest category). The one scrubber
outage had a magnitude which ranked it first for this incident indicator.
Process-related incident data is found in Table 30. The 34 sources
reporting Code 2 problems were fairly evenly divided among control categories
with the exception of baghouses which reported only three sources with pro-
cess incidents. Nine sources in the "other" category reported problems and
this category ranked first with respect to frequency of occurrence while ESPs
ranked second. If the one boiler source which reported 882 incidents is sub-
tracted from the "other" total, and the cupola problem of 700 incidents is
subtracted from the ESP total, our ranking for frequency would remain the same,
80
-------
TABLE 29. PARAMETER C - CONTROL DEVICE TYPE
CATEGORY RANKING
CAUSAL CODE 1 (DESIGN).
Rank
1
2
3
4
Category
Baghouse
Other
ESP
Scrubber
Frequency
24.6
1.2
0.8
0.5
Category
Other
Scrubber
ESP
Baghouse
Duration
(hr)
3097.7
2765.0
1130.4
15.8
Category
Scrubber
Baghouse
Other
ESP
Magnitude
(% above
allowable)
2817
1941
667
170
TABLE 30. PARAMETER C - CONTROL DEVICE TYPE
CATEGORY RANKING
CAUSAL CODE 2 (PROCESS).
Rank
1
2
3
4
Category
Other
ESP
Scrubber
Baghouse
Frequency
123.4
95.9
17.96
10.7
Category
ESP
Scrubber
Other
Baghouse
Duration
(hr)
11.7
8.9
2.4
0.18
Category
ESP
Other
Scrubber
Baghouse
Magnitude
(Z above
allowable)
815
180
160
88
81
-------
but all averages would be within 18 frequencies instead of the 113 frequencies
that currently separate the high and low values. ESPs ranked first in dura-
tion of incident as well as magnitude of emission based on nine separate source
reports. There is not a widespread difference between the categories with
regard to duration, and baghouses had an exceptionally short duration averaging
only 0.18 hours or 11 minutes per incident. The number one ranking for ESPs
in magnitude would be reduced to a number three ranking (magnitude of 144 per-
cent) if the five incidents of uncontrolled cement kiln were subtracted. The
baghouses again are ranked last for this indicator.
Control O&M data is found in Table 31. Forty-nine sources were surveyed
for this incident indicator, and were evenly distributed among control device
types. ESPs ranked first in the frequency of O&M problems, although all of the
categories were tightly grouped and no one device stood out. Such was not the
case for incident duration. Here, the "other" category and scrubbers both
averaged nine times longer incidents than the third ranked baghouses. Three
of the eleven "other" sources reported long (3 month) duration problems, al-
though none of the respective control devices were the same. Similarly, 2 of
the 14 scrubber sources reported average durations lasting greater than 3 months,
with one duration lasting an entire year. Incident magnitudes are all fairly
close for this Causal Code, with "other" ranked first. All categories averaged
3 to 5.4 times the applicable standard.
Twenty-five separate sources reported unforeseen problems and this data
summary is presented in Table 32. All categories had extremely low frequencies
of occurrence, with baghouses having the most and ranking first with 3.6 inci-
dents per year. Six scrubbers reporting problems had the longest incident
durations, averaging more than 11 days per outage. Durations at "other"
sources were similarly long (1 week) while the two remaining categories
averaged less than 3 days per incident. Finally, the "other" category ranked
first with the greatest magnitude, as each of the seven sources reporting a
problem had a magnitude in excess of 100 percent the allowable. Scrubbers
ranked a close second in magnitude, substantially ahead of the third ranked
ESPs.
The totals for all Causal Codes, summed collectively are presented in
Table 33. Inspection of the data for frequency reveals a two-tiered ranking,
with ESPs and "other" each having greater than 5 times the frequency of the
remaining categories. Elimination of our often discussed 882 and 700 fre-
quency sources would reduce the averages for ESPs and "other" to 15.5 and
12.6, respectively. Scrubber duration was twice that of any other control
device, with several of the sources reporting incidents lasting in excess of
60 days. ESPs ranked first with regard to magnitude. However, this figure
is misleading as it is greatly influenced by. five upsets at one cement plant.
If these five incidents are subtracted from the 1025 total incidents attribu-
table to precipitators, ESPs would have an average magnitude of 187 percent,
placing them last in this ranking. As reported, the high ESP ranking is in-
dicative of the large degree of control obtained with precipitators and the
concurrent potential for large magnitudes of excess emissions should a prolem
develop.
82
-------
TABLE 31. PARAMETER C - CONTROL DEVICE TYPE
CATEGORY RANKING
CAUSAL CODE 3 (CONTROL O&M).
Rank
1
2
3
4
Category
ESP
Baghouse
Other
Scrubber
Frequency
14.3
10.6
6.8
4.2
Category Du"£
-------
Integration of Rankings—
The integration of Parameter C data is especially informative due to the
small number of categories into which the sources are placed and the resultant
higher number of sources found in any one category. Data for this parameter
is therefore the most representative of those studied.
The integrated ranking of Causal Code 1 problems is found in Table 34.
"Other" ranks highest with respect to design problems, followed closely by
scrubbers and baghouses. Since the "other" category included data on several
types of control devices, we cannot draw conclusions for any one device within
this category. What may be inferred, however, from this overall number 1
integrated rank is that design problems are more likely to be a concern with
the uncommon and novel devices. The lack of long-term field experience with
devices such as vapor recovery systems is shown in this high ranking. The
second place rank of both scrubbers and baghouses seems to indicate that
these control devices may cause excess emissions if they are not properly
designed for a specific application. Use of improper materials of construction
was the most common reason cited for design problems with scrubbers and
baghouses.
Process-related problems (Causal Code 2) are integrated in Table 35. The
control device types are separated by relatively large margins in this inte-
gration with ESPs clearly ranked first. Since the process and not the control
device is associated with Causal Code 2 problems, this integration only points
out those devices which are associated with process problems. Precipitators
are generally more capital intensive than scrubbers or baghouses and are
typically installed on large facilities which require a high degree of control.
The number 1 rank of precipitators in this integration reemphasizes the con-
clusion of the Parameter B integration: that large sources have greater ex-
cess emissions problems.
From a control device standpoint, the integration of control-related prob-
lems is more significant. The integration for these Causal Code 3 problems is
presented in Table 36. In this causal code, "other" devices ranked first. This
indicates that the less conventional control devices have a greater problem
with excess emissions due to their actual operation and maintenance. When
used in conjunction with the design problem integration, it reemphasizes the
point that these less commonly found devices collectively have initial as well
as continued problems with their operation and maintenance. This ranking also
suggests that these devices may not be properly maintained. Control device
complexity or lack of training for maintenance personnel who work with these
controls may contribute to this rank.
Unforeseen problems are integrated in Table 37. Scrubbers are ranked
first here, closely followed by "other" devices. This integration indicates
those devices which are more susceptible to an unforeseen outage due to causes
beyond the control of the equipment operator. The number one rank of scrubbers
is perhaps indicative of the difficulty encountered by sources in keeping these
units on line, through no fault of their own. The nature of scrubber operation,
with constant movement of a caustic or acidic solution, makes constant sur-
veillance of these devices critical, as conscientious maintenance practices
alone may not be sufficient to eliminate problems.
84
-------
TABLE 34. PARAMETER C - CONTROL DEVICE TYPE
INTEGRATION OF RANKINGS
CAUSAL CODE 1 (DESIGN).
Individual ranking
Rank
1
2
2
3
Category
Other
Scrubber
Baghouse
ESP
Number
of
samples
3
1
4
3
Total = 11
Frequency
2
4
1
3
Duration
(hr)
1
2
4
3
Magnitude
(% above
allowable)
3
1
2
4
Zall
ranking
6
7
7
10
TABLE 35. PARAMETER C - CONTROL DEVICE TYPE
INTEGRATION OF RANKINGS
CAUSAL CODE 2 (PROCESS).
Rank
1
2
3
4
Category
ESP
Other
Scrubber
Baghouse
Total
Number
of
samples
9
9
13
3
= 34
Individual ranking
Frequency
2
1
3
4
Duration
(hr)
1
3
2
4
Magnitude
(% above
allowable)
1
2
3
4
Zall
ranking
4
6
8
12
85
-------
TABLE 36. PARAMETER C - CONTROL DEVICE TYPE
INTEGRATION OF RANKINGS
CAUSAL CODE 3 (CONTROL O&M).
Rank
1
2
3
3
Category
Other
ESP
Baghouse
Scrubber
Total
Number
of
samples
11
10
14
14
= 49
Individual ranking
Frequency
3
1
2
4
Duration
(hr)
1
4
3
2
i
Magnitude
(% above
allowable)
1
2
4
3
r* ail
£-*i ranking
5
7
9
9
TABLE 37. PARAMETER - CONTROL DEVICE TYPE
INTEGRATION OF RANKINGS
CAUSAL CODE 4 (UNFORESEEN).
Individual ranking
Rank
1
2
3
4
Category
Scrubber
Other
ESP
Baghouse
Total
Number
of
samples
6
7
8
4
= 25
Frequency
3
4
2
1
Duration
(hr)
1
2
3
4
Magnitude
(% above
allowable)
2
1
3
4
Lall
ranking
6
7
8
9
86
-------
The integration of the rankings for all causal code problems is found on
Table 38. As can be seen, ESP's rank first overall due to their number one
rankings with respect to both frequency and magnitude. This ranking, however,
is more indicative of the fact that ESPs are associated with problem areas,
than it is a statement concerning how well ESPs actually operate. This is
especially true concerning the individual magnitude rankings. The five inci-
dents of uncontrolled emission at a cement plant controlled by an ESP were of
such magnitude so as to rank these devices first. Without these incidents,
the average ESP duration would have been sufficiently low'to rank it last for
this indicator, and this change in magnitude ranks would have placed ESPs
last in the overall integration. Moreover, the cause of these five excursions
was process-related, and not due to some problem with the ESP. This situation
is presented to remind the reader of the cumulative effects of an uncommon
situation. As presented, scrubbers ranked second and baghouses ranked last,
as they had for two of the four individual causal code integrations.
Normalized Credits - Normalized Excess—
The rankings for normalized credits, and normalized excess and the inte-
gration of their difference is presented in Table 39. As can be seen, the
rankings for each of these items is exactly the same. In all cases "other"
has the highest rank due to its low normalized credits, its high normalized
excess and the small value of the difference. This indicates that these rel-
atively uncommon control devices normally operate closer to the allowable
emission rate, have relatively greater excesses, and have a net surplus of
credits minus excess which is smaller than the conventional control devices.
It is interesting to note that all control device categories had credits which
exceeded excesses. On an annual basis, none emitted more pollutant to the
ambient air than allowed by applicable mass emission regulations. On the lower
end of the rankings were baghouses. With their greater credits and lower
excesses, these devices produced a net annual surplus of emissions that far
exceeded any control device category.
87
-------
TABLE 38. PARAMETER C - CONTROL DEVICE TYPE
INTEGRATION OF RANKINGS
ALL CAUSAL CODES.
Individual ranking
Rank Category of
samples Frequency
1 ESP 22
2 Scrubber 36
3 Other 26
A Baghouse 35
Total = 119
1
3
2
3
Duration Ma.gnitude pranking
f . (% above
lhr' allowable)
3 1 5
1 3 7
2 A 8
A 2 9
TABLE 39.
PARAMETER C -
CONTROL DEVICE TYPE
RANKING OF NORMALIZED
CREDITS AND EXCESSES.
D , „ . Normalized
Rank Category credits
1 Other 0.37
2 Scrubber 0.48
3 ESP 0.51
A Baghouse 0.65
Rank Category
1 Other
2 Scrubber
3 ESP
A Baghouse
Normalized „ , „ . Credits -
^~~«,o Rank Category
excess ° ' excess
0.33 1 Other O.OA
0.27 2 Scrubber 0.21
0.2A 3 ESP 0.27
0.1A A Baghouse 0.51
88
-------
Parameter D - Control Device Size
Ranking of this parameter will serve to point out those control device
sizes, regardless of the type of control device, which are more likely to have
excess emissions incidents. Since larger sources are likely to have larger
control devices, these rankings will serve to complement the rankings pre-
sented for Parameter B, source size. Differences in the category rankings of
the two parameters are primarily due to sources which operate intermittently,
and have control devices sized to their maximum rated production capacity.
Category Rankings—
Causal Code 1 rankings are provided in Table 40. The 11 sources reporting
design problems are somewhat evenly divided among the six categories. The
50,000 to 100,000 acfm category contained no problem sources while the other
three categories below 500,000 acfm each reported three sources with incidents.
The larger size categories, 500,000 to 1,000,000 acfm and greater than 1,000,000
acfm each contained one source. The 100,000 to 500,000 acfm category had five
times the number of frequencies of any other category. The three sources re-
porting problems in this size range reported 89 percent of all design problems.
The one source which was investigated having a control device size in excess of
1 million acfm reported a repeated design-related problem and was ranked second
with respect to frequency. Three of the 31 sources in the size range less than
10,000 acfm, reported problems, and the duration of their reported incidents
ranked them first with respect to this indicator. Each of these three small
devices had incidents lasting in excess of 19 days continued operation. The
500,000 to 1,000,000 acfm and 10,000 to 50,000 acfm categories also had incident
durations which exceeded 43 days of continued device operation. The magnitude
ranking was led by the 10,000 to 50,000 acfm category. The average magnitude
of this category was more than five times greater than the next category. As
has been the case with all Causal Code 1 rankings, this large magnitude reflects
the input of one atypical, large event.
TABLE 40. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
CATEGORY RANKING
CAUSAL CODE 1 (DESIGN).
« «
Category „ Category Duration Category
Rank (103 Lfm) Frequency (103 Lfti) (hr) (10» acfm)
Magnitude
1
2
3
4
5
6
100 -
>1000
<10
500 -
10 -
50 -
500
1000
50
100
31
6
1
1
0
0
.0
.0
.2
.0
.5
<10
500 -
10 -
>1000
100 -
50 -
1000
50
500
100
3097
2190
1033
24
20
0
.7
.0
.7
.0
.2
10 -
100 -
<10
>1000
500 -
50 -
50
500
1
1000
100
10,814
1,893
667
100
69
0
89
-------
Rankings for Causal Code 2 are presented in Table 41. The 100,000 to
500,000 acfm category ranks first with respect to frequency, and this ranking
is significant for two reasons. The average frequency is more than twice that
of the second ranked category, and more than half (12 of 23) the sources in this
control device size range experienced process problems. Even if one were to
discount the 700 incidents at one source, this size range would be the highest
ranked category by greater than 50 percent. In general, it appears that the
larger sources have a higher incident frequency. Little variation appears
among categories with respect to average incident duration, the middle sized
categories stand out as having the longest durations, with the 50,000 to
100,000 acfm group ranking number one. No trend was evident with respect to
duration and category size. The 100,000 to 500,000 acfm category lead all
others with respect to incident magnitude. No one category stood out with
regard to this indicator and no correlation could be made between magnitude
and control device size.
TABLE 41. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
CATEGORY RANKING
CAUSAL CODE 2 (PROCESS).
0 , Category _ Category Duration Category ,v° .
Rank ,._a 6 ,'* Frequency /ir>3 * \ /-i. \ /i«3 * \ (* above
(103 acfm) n * (103 acfm) (hr) (103 acfm) ,, .
allowable}
1
2
3
4
5
6
100 -
500 -
10 -
50 -
<10
>1000
500
1000
50
100
146
60
22
17
14
0
.0
.0
.6
.7
.5
50 -
10 -
100 -
<10
500 -
>1000
100
50
500
1000
10
10
5
4
4
0
.85
.6
.78
.5
.0
100 -
10 -
500 -
<10
50 -
>1000
500
50
1000
100
482
377
200
109
54
0
Causal Code 3 problems are ranked in Table 42. All categories have simi-
lar average frequencies with the 500,000 to 100,000 group heading the list.
The average for this category is based on the two sources (out of six) which
reported problems. The general trend of these rankings indicates that the
larger sources had more problems per year. The duration indicator is lead
by the small size category; control devices with size less than 10,000 acfm.
Three of the eleven sources reporting operating and maintenance problems in
this size category had incidents lasting in excess of 89 days continuous
operation. These sources contributed to the high average duration of this
category which was almost four times as great as the second ranked category.
The 10,000 to 50,000 acfm category was ranked first with regards to Incident
magnitude. All categories were generally close in this indicator and no
clear-cut size category/incident magnitude relationship was evident.
90
-------
TABLE 42. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
CATEGORY RANKING
CAUSAL CODE 3 (CONTROL O&M).
Frequency
n
Du"tJOn
acfm) (hr)
,
acfm)
allowabie)
1
2
3
4
5
6
500 - 1000
100 - 500
10 - 50
<10
50 - 100
>1000
17.3
11.3
8.5
6.3
6.0
4.0
<10
10 - 50
100 - 500
>1000
50 - 100
500 - 1000
143.0
38.4
22.8
15.0
8.5
6.2
10 - 50
100 - 500
<10
50 - 100
500 - 1000
>1000
569
412
227
218
181
100
The rankings for Causal Code 4 are presented in Table 43. On the basis
of one source report for each of the two largest control size categories, they
are ranked first and second with regards to frequency of incidents. All cate-
gories are grouped closely together with respect to frequency, and no clear-cut
category size/frequency of incident relationship is evident. While only four
sources in the less than 10,000 acfm category cited unforeseen problems, two of
these incidents were in excess of 50 days continuous source opeatlon this rank-
ing this size category first with respect to incident duration. This duration
was almost four times as long as the next ranked category. Two size categories
stood out with respect to unforeseen incident magnitude. Seven of eight sources
with this type problem in the 10,000 to 50,000 acfm size category reported inci-
dents in excess of 300 percent of the applicable standard causing this category
to be ranked first. Three sources in the next largest size range, 50,000 to
100,000 acfm had large magnitude problems, and the average of this category
was more than six and one-half times the standard. Generally, the smaller size
categories had larger incident magnitudes.
The data for all causal codes considered together is presented in Table 44.
An examination of the frequency data reveals that the larger control devices
generally have more incidents than the smaller devices. Nineteen of the
twenty-three sources in the 100,000 to 500,000 acfm category reported problems
and this group ranked first with regards to the frequency of occurrence. Once
again this average is affected by the two large frequency sources, but even if
we eliminate the sources with 882 and 700 incidents, this category would still
rank first. Collectively, the small control devices had fewer incidents, and
the smallest category of all had the fewest. This smallest size category, less
than 10,000 acfm, had the longest duration as well, with 6 of the 17 facilities
reporting problems having average incident durations in excess of 73 days
continued source operation. The average duration for this category was 6.5
times higher than the second ranked size classification. No trend was evident
between control device size and excess incident duration. The magnitude indi-
by the 100,000 to 500,000 acfm category. With the 10,000 to
91
-------
TABLE 43. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
CATEGORY RANKING
CAUSAL CODE 4 (UNFORESEEN).
•-* (Wi "«"— >
1 >1000 5.0
2 500 - 1000 4.0
3 <10 2.8
4 100 - 500 2.3
5 10-50 1.1
6 50 - 100 0.7
Magnitude
Category Duration Category /v ,
7 3 ^ ^ \ \ ™ 8DOVS
(10J acfm) (hr) (103 acfm) _ii«.,,,ui«\
anowaDiej
<10
50 - 100
100 - 500
500 - 1000
10 - 50
>1000
229.8 10 - 50
61.0 50 - 100
58.5 <10
48.0 100 - 1000
34.3 100 - 500
10.0 >1000
674
664
390
200
144
100
TABLE 44. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
CATEGORY RANKING
ALL CAUSAL CODES.
Rank
1
2
3
4
5
6
Category
(103 acfm)
100 - 500
500 - 1000
>1000
50 - 100
10 - 50
<10
(103 acfm) (hr) (103 acfm) ,,a °Yf x
allowable)
86.9
16.6
15.0
14.1
9.2
4.1
<10
500 - 1000
10 - 50
>1000
50 - 100
100 - 500
184.3
28.5
24.4
20.3
12.0
8.1
100 - 500
10 - 50
<10
500 - 1000
50 - 100
>1000
468
406
214
192
111
100
92
-------
50,000 acfm group a close second. Both these categories had double the
magnitude of the third place size range, although none of the magnitudes
were greater than five times the allowable standard.
Integration of Rankings—
The integration of Causal Code 1 rankings is found in Table 45. The
smallest size category, < 10,000 acfm ranks first. This ranking is somewhat
surprising as one might expect the larger sources to have greater design
problems due to their size and complexity. The three sources upon which this
number one ranking was based consisted of two small vapor recovery systems
and one carbon absorption system which are relatively complex devices to
operate. The 100,000 to 500,000 acfm category had an integrated rank which
placed it second. While this size range had the greatest frequency, it had
relatively short incident durations which minimized the net ranking. Simi-
larly, the 10,000 to 50,000 acfm category ranked first in magnitude but had
relatively few frequencies and a mid-level average incident duration. There
was no identifiable correlation between integrated design rank and the size
range of the category.
TABLE 45. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
INTEGRATION OF RANKINGS
CAUSAL CODE 1 (DESIGN).
Individual ranking
Rank
1
2
3
4
5
6
Category
(103 acfm)
<10
100 - 500
10 - 50
>1000
500 - 1000
50 - 100
Number
of
samples
3
3
3
1
1
0
Frequency
3
1
5
2
4
6
Duration
(hr)
1
5
3
4
2
6
Magnitude
(% above
allowable)
3
2
1
4
5
6
Zall
ranking
7
8
9
10
11
18
Total = 11
The integration of Causal Code 2 is found in Table 46. Once again, the
overall top ranked category is the 100,000 to 500,000 acfm size range. Since
Causal Code 2 problems are process related, this integration is solely an in-
dicator of equipment sizes which are associated with process problems. It
cannot be used in itself, to pinpoint who is having a problem or where that
problem is. This overall first ranked category was also the number one ranked
category for both frequency and magnitude and should not be overly biased as
it is based on the input of 12 separate sources. The second ranked category,
10,000 to 50,000 acfm, had consistently high rankings for all indicators and
93
-------
was also based on a relatively large sample (13). The integrated ranking
does not appear to be related to size category in any discernable manner.
TABLE 46. PARAMETER D - CONTROL DEVICE (ACFM)
INTEGRATION OF RANKINGS
CAUSAL CODE 2 (PROCESS).
Rank
1
2
3
3
5
6
Category
(103 acfm)
100 -
10 -
500 -
50 -
<10
>1000
500
500
1000
100
I
Number
of
samples
12
13
1
5
3
0
Individual ranking
Frequency
1
3
2
4
5
6
Duration
hr
3
2
5
1
4
6
Magnitude
(% above
allowable)
1
2
3
5
4
6
Eall
ranking
5
7
10
10
13
18
Total - 34
The Causal Code 3 integration is presented in Table 47. More sources,
49, reported control equipment operating and maintenance related problems
than any other causal code incident. The greatest number of sources in any
one category reporting these problems was found in the highest ranked grouping,
10,000 to 50,000 acfm. This category ranked first in magnitude and was close
to the top in frequency and duration. The smallest category, less than 10,000
acfm ranked first with respect to duration, but had relatively fewer episodes
and ranked third overall behind the 100,000 to 500,000 acfm group, which
was consistently high in all indicator ranks. While the large 500,000 to
1,000,000 acfm category had the most frequencies, both the duration and mag-
nitude of these occurrences were minimal. Again no apparent correlation exists
between the size of category and integrated rank for this causal code.
The integration for Causal Code 4 data is presented in Table 48. As can
be seen, sources in the smallest category, less than 10,000 acfm, appear to
have the greatest unforeseen problems. This category ranked first with respect
to incident duration and was the only source to consistently rank high in all
three incident indicators. The top ranked category in magnitude, 10,000 to
50,000 acfm, had relatively few short incidents while the first rank in fre-
quency, the greater than 1,000,000 category, had the shortest average duration
and smallest magnitude ranked last overall.
94
-------
TABLE 47. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
INTEGRATION OF RANKINGS
CAUSAL CODE 3 (CONTROL O&M).
Rank
1
2
3
4
5
6
„ _ Number
Category
(103 acfm) °E.
samples
10 - 50 17
100 - 500 12
<10 11
500 - 1000 2
50 - 100 6
> 1000 1
Total = 49
Individual ranking
Frequency
3
2
4
1
5
6
Duration
(hr)
2
3
1
6
5
4
Magnitude
(% above
allowable)
1
2
3
5
4
6
Zall
ranking
6
7
8
12
14
16
TABLE 48. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
INTEGRATION OF RANKINGS
CAUSAL CODE 4 (UNFORESEEN).
Rank
1
2
2
4
5
6
Number
Category
(103 acfm) ,
samples
<10 4
500 - 1000 8
50 - 100 3
10-50 8
100 - 500 8
>1000 1
Total = 25
Individual ranking
Frequency
3
2
6
5
4
1
Duration
(hr)
1
4
2
5
3
6
Magnitude
(% above
allowable)
3
4
2
1
5
6
Eall
ranking
7
10
10
11
12
13
95
-------
The most significant of the integrations, the one involving all causal
codes is found in Table 49. A tie existed for the overall number one ranking
between the 100,000 to 500,000 acfm category and the 500,000 to 1,000,000 acfm
size group. The former size cut ranked first in both frequency of incident
and average magnitude and would have been clearly the top rank except for its
last place duration rank. Nineteen of the twenty-three sources in this group
reported problems. The other large-size category ranked second in both fre-
quency duration indicators even though three of this group's six sources
reported no problems.
TABLE 49. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
INTEGRATION OF RANKINGS
ALL CAUSAL CODES.
Category
,, _ a .. \
(103 acfm)
Number
of
Individual rankings
, „ Duration /. . *^ ranking
samples Frequency ,, * (% above
1 ' allowable)
1
1
3
3
5
6
500 - 1000
100 - 500
10 - 50
<10
>1000
50 - 100
6
23
49
31
1
9
2
1
5
6
3
4
2
6
3
1
4
5
4
1
2
3
6
5
8
8
10
10
13
14
Total =119
Normalized Credits/Normalized Excess—
Finally, we review the relationship between normalized credits and nor-
malized excess. This data is presented in Table 50. A general trend is
apparent in the normalized credit column. As the size of control device in-
creases, so does the amount of normalized credit. Since our procedure for
normalizing both credits and excess has been used to eliminate the bias of
large sources, this trend is informative. It seems to indicate that the
larger sources are controlled to an increasingly greater degree than required.
It can also be inferred that larger control devices are purposely over-designed
so as to leave the source with some latitude in meeting state emission regula-
tions, should the design parameters of the equipment be altered for some reason.
A similar general trend, though not a direct correlation, is seen in the
normalized excess data. Again, the smaller the source, the greater the excess
emissions problem, with the smallest size category having the greatest nor-
malized excess while the largest control devices have the least.
96
-------
TABLE 50. PARAMETER D - CONTROL DEVICE SIZE (ACFM)
RANKING OF NORMALIZED
CREDITS AND EXCESSES.
Rank
order
1
2
3
4
5
6
Category Normalized Category
(103 acfm) credits (103 acfm)
50 - 100
100 - 500
<10
10 - 50
500 - 1000
>1000
0.41
0.48
0.52
0.53
0.60
0.73
<10
100 - 500
10 - 50
500 - 1000
50 - 100
>1000
Normalized Category
excess (103 acfm)
0.40
0.38
0.14
0.08
0.02
0.01
100 - 500
<10
50 - 100
10 - 50
500 - 1000
>1000
Credits -
excess
0.10
0.12
0.39
0.43
0.52
0.72
The third column in this table, the difference between the normalized
values shows a less defined control size/excess emissions correlation. All
categories have a net surplus of annual emissions as excesses are more than
completely offset by credits. A wide range of surplus exists, however, ranging
from 10 percent of allowables for the 100,000 to 500,000 acfm class to 72 per-
cent of allowables for the greater than 1,000,000 acfm category. While this
latter large category has the lowest rank in all three normalized areas, this
data is based on only one source survey, and care should be taken in inter-
preting the results. The small emissions surplus of the 100,000 to 500,000
acfm categorv reinforces the previously discussed number one rank of this
category wit) respect to all causal codes and suggests that sources with
control devices in this size range experience the greatest excess emissions
problems.
97
-------
SECTION 3
CONCLUSIONS AND RECOMMENDATIONS
DATA REVIEW
All excess emissions data analyzed for this study was extracted from 119
separate source surveys. In developing our various averages and the resultant
category ranks and integrations we have used all sources, including those which
reported no excess emissions incidents. A summary of the relevant source and
emissions data pertaining to these sources is presented in Table 51. We feel,
however, that the true extent of the excess emissions problem is larger than
that which is shown by these numbers. Unlike water quality programs which can
utilize discharge permit data to quantify pollutant emissions, air programs
do not establish explicit ambient emissions levels from each process operation.
Shortages of detailed verifiable emission data forced the individual contractors
to fill in data gaps with recollections of plant personnel concerning control
equipment performance. A conservative tendency was therefore necessarily
adopted in making judgements concerning the amount and extent of excess emission
episodes. In addition, plant personnel exhibited some lingering hesitation
to divulge process and control equipment performance data even though they
were instructed that no enforcement action would be taken on the basis of their
statements. We feel these factors lead us to underestimate the true extent of
the excess emissions problem. The extent of this underestimation cannot be
quantified.
A second factor which contributes to an underestimation of the actual
excess emissions problem concerns the interpretation of available process and
control equipment data by the individual contractors. No widely acceptable
methodology exists to determine the affect on stack emissions of a process
or control equipment perturbation. Each contractor used what information was
available and his best engineering judgment to estimate increases in stack
emissions due to a change in process and control equipment parameters. The
affect of boiler soot blowing or loss of one bag in a baghouse has not been
widely investigated. What is known in these areas was utilized. However,
we feel that the emissions estimates which were developed were conservative
and tended to underestimate the actual magnitude of source emissions. Since
we did not have access to individual contractor methodologies, we cannot be
sure of the extent nor the degree of this underestimation. Data from different
98
-------
TABLE 51. UNMODIFIED DATA SUMMARY
VO
Itea
uncontrolled emissions (tons)
Credits (tons)
Excess (tons)
Allowable emissions Ctoos)
Frequenc^
Duration (hours)
Magnitude (X)
Magnitude (absolute tons)
Capital cost (103$)
O&M cost (103$)
Control size (103acfm)
Control age
Normalized excess
Normalized .credits
Causal code 1
(design)
8,048.9
178. 5
118.3
310.0
9.5
158.2
1,003.
12.4
2,274
264
273
3.2
105.5
44.1
Causal code 2
(process)
14,566.1
85.3
18.4
234.2
65.9
6.6
421
0.3
1.218
177.
113.2
5.8
4.6
42.0
Causal code 3
(control O&M)
10,406.2
119.1
10.6
225.8
8.6
45.9
399
1.2
2,067.
267.
145.1
5.0
12.4
42.6
Causal code 4
(unforeseen)
16.314.8
202.9
9.2
337.8
1.9
83.9
31.7
4.9
1,747.
192.
192.3
4.8
25.8
50.1
£all causal codes
14,603.1
110.9
22.5
207.8
21.7
19.3
438
1.0
1,567
170
110.5
4.9
21.6
50.0
-------
contractors on similar sources did vary and we cannot be sure if this variation
was due to the nature of the source visited or the way in which available data
was converted to emissions estimates. This variation in individual contractor
estimates is perhaps best pointed out by the number of sources which reported
no emissions incidents. Of the 119 sources surveyed, 41 (34.5 percent) reported
no incidents during the study period Q.976 to 1977). The percent of these
sources in individual contractor summaries varied from 26.8 percent to 64.7
percent of the sources surveyed. We feel this variation is due to different
interpretations of the available source data and the identification of a source
as having no problems when no substantive data was available on which to make
an excess emissions estimate.
DATA MODIFICATION
Reviewing the raw excess emissions data presented in Table 51, we see
that many of the averages were greatly influenced by one or two typical cases.
The effect that the 882 incidents at one boiler facility and the 700 incidents
at one iron and steel plant had on average frequency tabulations, or the high
incident magnitude averages that resulted from operation of an uncontrolled
cement plant were repeatedly cited in Section 2. Since the purpose of this
study was to present the typical excess emissions problem that currently
exists, we felt that inclusion of these extreme data points would bias the
entire sample. Many avenues of data modification were investigated. While
the standard statistical procedure In this case is to compute geometric mean
values for all pertinent data, we were unable to do this for the frequency,
duration, and magnitude data due to the existence of data points with zero
values and our inability to use the log of these points. To account for the
existing statistical bias, we have chosen to eliminate extreme excess emissions
incidents based on the application of best engineering judgement. This method-
ology involved completely eliminating from the data base those sources which
had frequencies in excess of one per day or (364 per year), and those sources
with reported incident durations in excess of 4380 hours (% year of continual
plant operation). For our data base, this technique served to eliminate two
sources with high frequencies, and three sources with long durations. An addi-
tional three sources had individual incidents with long durations, and these
incidents, but not the entire source, were deleted. The remaining incidents for
these three sources were then used to calculate new source averages.
Finally we addressed the representativeness of the 41 sources which reported
no excess emissions problem. Each of the four contractors was contacted and
asked to guage the accuracy of the data supplied by these problem-less sources.
Specifically, we asked each contractor if, in their opinion, these sources has
actually had problem free case histories or could they have had a problem which
was unreported. The reluctance of many sources to divulge excess emissions
related information, in spite of assurances that this data would not be used
in enforcement actions, prompted this question. Collectively, the contractors
felt that approximately 42 percent of these sources may have had a problem which
went unreported. This figure represented 17 of their sources reporting no
emissions incidents. In order to eliminate this bias, 17 sources were randomly
selected from this group of 41, and their source records were deleted from
the data base. Our overall data base was therefore reduced by 22 sources. New
averages were then computed for each item and this modified data summary is
100
-------
presented in Table 52. The effect of this modification procedure on each
Causal Code summary can be seen by a comparison of Tables 51 and 52 and will
not be discussed. The way in which the average frequency, duration and
magnitude of all incidents has been changed is discussed in the following
paragraphs.
Twenty two of the 119 sources were eliminated in computing new indicator
averages. The modified averages indicate that:
• Frequency is reduced from 23.7 to 12.3 occurrences
per year
• Duration is increased from 19.3 to 23.7 hours
• Magnitude increases from 438 to 828 percent
The two highest frequencies of occurrence alone accounted for 57.3 per-
cent of all reported incidents. Elimination of additional frequencies which
were associated with long durations accounted for an additional 0.1 percent
for an overall reduction of 57.A percent. The average incidents per source
were therefore greatly reduced.
Sources with incident durations in excess of 4380 hours accounted for
46.7 of all hours in the excess mode. Sources with high incident frequencies
accounted for an additional 1.1 percent of all hours, for an overall reduction
of 47.8 percent. This indicates that long durations were associated with
relatively few incidents and the high frequencies are associated with relative-
ly few total hours. These factors tended to cancel each other out when both
high frequencies and long durations are eliminated and the average incident
duration only changed sightly, rising to 23.7 hours.
While sources with high incident magnitudes were not altered in our
data base, the average magnitude of incidents in our modified data base almost
doubled. This statistic reflects the. fact that excess emissions incidents
which occur frequently, or with a prolonged duration, do not have large
magnitudes. When these extreme frequency and duration episodes are eliminated
the average incident magnitude rises sharply.
101
-------
TABLE 52. MODIFIED DATA SUMMARY
S
Item
Uncontrolled emissions (tons)
Credits (tons)
Excess (tons)
Allowable emissions (tons)
Frequency
Duration (hours)
Magnitude (%)
Magnitude (absolute tons)
Capital cost ($ x 103)
O&M cost ($ x 103)
3
Control size (10 acfm)
Control age
N Normalized excess
Normalized credits
Causal code 1
(design)
8,801.9
196.3
123.8
333.2
10.4
102.0
1,012.0
11.9
2,484.3
264.3
299.7
3.3
107.9
48.5
Causal code 2
(process)
11,677.3
85.5
5.4
191. Q
22.6
5.3
107.5
0.2
1,295.7
204.2
109.7
5.4
3.2
44.2
Causal code 3
(control O&M)
11,117.9
125.3
11.4
233.8
8.7
27.1
426.0
1.3
2,150.9
274.5
148.9
4.9
12.9
43.7
Causal code A^/V
(unforeseen)
16,314.8
202.9
9.2
337.8
1.9
83.9
317.0
4.9
1,746.7
191.7
192.3
4.8
25.8
50.1
11 causal codes
13,341.5
100.1
22.0
179.2
12.3
23.7
828.0
1.8
1,623.3
180.9
105.4
5.1
24.7
50.0
-------
PROBLEM AREA PROFILES
In order to give the reader an overview of those areas which have specific
excess emissions problems, we have summarized the results of our various param-
eter integrations. Tables 53 through 57 present the integrated rankings of the
four parameters for each causal code. A summary table is also provided for the
integration of all causal code data. By combining the integrations of all
source types, source sizes, control device types and control device sizes,
which have excess emissions problems due to design, process, control or unfore-
seen causes. The summation number by which each parameter's data was ranked is
also provided to serve as a relative guide in comparing categories within a
parameter. We call these various combinations "problem area a profiles" inasmuch
as they readily identify those source and control equipment areas which we have
associated with various excess emissions problems.
We must re-emphasize two points which have to be kept in mind when reviewing
these tables. The first is that they were based on our total, unmodified, sample
population of 119 sources. As such they may reflect an unusually large value
contributed by one source. Questions concerning the factors which contribute to
any ranking will be answered by referring to the appropriate discussion in
Section 2.
The second point concerns the actual interpretation of each table. These
profiles are not to be read by taking the top-ranked category for each param-
eter and assuming that this is the worst combination for that causal code.
For example, referring to Table 53, we should not assume that grain-handling
operations with source sizes of 500 to 1000 tons and using "other" types of
controls which are sized at less than 10,000 acfm are those with the most
design problems. While each of these categories ranks first, it may not
follow that they are typically found combined in this way. Grain-handling
operations typically use baghouses for controls and are usually large sources
with large control devices. In fact, none of the five grain-handling facil-
ities which we surveyed were in the 500-1000 tons/yr category, none used
"other" control devices and none utilized control devices sized in the less
than 10,000 acfm range.
The profiles presented may be used by regulatory personnel to pinpoint
areas in need of closer attention. Table 53 presents the design problem
profile. Elimination or minimization of design problems requires a greater
in-depth review procedure by control agencies for new sources being installed.
If a new design review regulation were to be adopted, state and local agencies
might use a table of this nature to focus their attention on specific indus-
trial categories or control device types. Using the rankings of Table 53,
emphasis would be placed on samll grain handling and steam-generating plants
for design review. Similary, new novel control devices would get first
priority when control equipment design reviews are conducted.
103
-------
TABLE 53. PROBLEM AREA PROFILE OF CAUSAL CODE 1 (DESIGN)
Rank
1
2
2
4
5
6
Parameter A
Category
Grain handling
operation!
Stean generating planta
Petr»l«um product i and handling
Petrc:hemical
Iron and ateel
Stone, clay
/Pulp and paper
plant!
plant*
mill!
Paraneter B Parameter C Parameter D
Eall Category V^ all - k c.,..0,v V* all Category V* •"
ranking (10J ton/yr) pranking "•"" *•«•»"? pranking (101 acfm) pranking
? 0.5-1 7 1 Other 6 <10 7
9 0.1-0.5 7 2 Scrubber 7 100 - 500 8
) 1-10 7 2 Baghouie 7 10-50 9
' . 10-100 10 3 ESP 10 >1000 10
>. 0.1 14 500 - 1000 11
15 100 - SOO 18 50 - 100 18
-
1 Incineratora
1 Cement planta
yUptwlt planta
7 I
Aluminum plante
Braaa and bronze planta
Lumber and wood planta
Surface coating operations
Food and drug
Other
planta
-
-
-------
Table 54 presents the process problem profile. Attention in this area
might concern an in-depth study of the higher ranked categories to determine
how their plant operations may be altered to reduce the potential of process-
related excess emissions incidents. Based on our data, cement plants would
be prime candidates for such a study.
Control related problems are profiled in Table 55. Minimization of these
problems may entail the adoption of specific operating and maintenance pro-
cedures for the high ranking industrial categories and control devices. Our
data seems to indicate that hydrocarbon emitting sources may be the best place
to start when investigating a recommended method of equipment operation and
maintenance. Our data also points out that novel methods of control may be
the area that requires the most attention when conducting source inspections.
These devices appear to suffer the most from lack of attention.
The profile for unforeseen incidents is presented on Table 56. Here the
rankings identify those categories which appear to be more susceptible to
uncontrollable emissions incidents. A policy which requires these facilities
to have some method of being alerted to process upsets may be in order here.
Also, these categories might be required to have an approved backup procedure
should their equipment malfunction. While these policies may not eliminate
unforeseen incidents, they may reduce their impact once an incident occurs.
The data indicates, with respect to unforeseen events, that hydrocarbon sources
would be the types of facilities in need of followup studies.
The profile for all causal codes is given on Table 57. As this profile
is compiled from all problem areas, these categories may provide a listing of
areas in need of heightened daily surveillance by state and local enforcement
personnel. The frequency of on-site inspections could possibly be increased
for these sources to spot problem areas and to alert plant personnel to the
susceptibility of their sources to excess emissions problems. A requirement
that these selected categories report all process and control upsets to the
local agency might be adopted where it does not currently exist. This would
pinpoint the recurring problem sources for the agency and force the source to
deal with the problem.
INDICATOR PROFILES
An overview of the problem categories with respect to the excess emission
incident indicators, frequency, duration and magnitude, is also informative.
Table 58 through 60 present these indicator profiles. The causal code inci-
dent data for each of the four parameters is presented in respective tables.
Since the cumulative excess emissions problem is influenced by the frequency,
duration and magnitude of the individual excess emissions incidents, control
strategies that can be developed to minimize the extent of any of these indi-
cators will serve to reduce the overall problem.
Table 58 presents the frequency indicator profile. As can be seen in
this table, the distribution of frequencies of incident occurrence are con-
centrated in relatively few categories. The key to minimizing the total number
of frequencies appears to be limiting the repetitious events that constantly
cause an excess emission.All of the top-ranked categories are associated
105
-------
TABLE 54. PROBLEM AREA PROFILE OF CAUSAL CODE 2 (PROCESS)
o
a\
Hank
1
2
3
i>
5
5
7
8
8
10
11
12
Parameter A
Category
Ctocnt plant
Surface coating operation!
Pulp and paper ail Is
Stua generating plants
Petroleum products and handling
Iron and steel plants
Other
Petrochemical plants
Brass and bronse plant!
Incinerator a
Asphalt planta
Aluninua planta
(store, clay and glaaa plants
Nrain handling operations
I Lunber and wood planta
i^Food and drug planta
Paraaeter B ' Parameter C
Eall Category V* all r-r..a™ V* •"
ranking (103 ton/yr) Cranking Category ^rinking
; 100 - 500 7 ESP 4
13 <0.l 8 other 6
10 - 100 10 Scrubber 8
I': 1-10 11 -Bsghouse 12
19 O.S - 1 11
19 0.1 - 0.5 16
20
21
21
23
27
•)'•
-
-
-
Parameter D
Category T"* all
(103 acra) pranking
10 - SO 6
100 - 500 6
500 - 1000 10
<10 10
>1UOO 13
18
-------
TABLE 55. PROBLEM AREA PROFILE OF CAUSAL CODE 3 (CONTROL O&M)
Rank
1
i
}
'
i
6
7
*,
9
9
'1
• >
> )
: *
s
:',
Parameter A
Category
Surface coating operationa
''tone, clay and glaia plant*
Petroleua products and handling
'.'•jment piano
Aiphalt plantt
Lumber and wood plant*
Iron and ateel planta
luc ineratora
Grain handling operationa
Other
Muninum plants
''ftrochcmical plant*
Bra** and bronco pluntr
lood and drug planta
Steam generating plant*
!'ulp and paper Bill*
Parameter B Parameter C Parameter D
Eall Category V"» all V^ all Category y all
ranking (10- ton/yr) ZJranking ^"Sory ^^rlnkin> (103 Mcfm) pranking
16 . - 13 7 Other 5 10-50 6
17 0.1-0.5 9 ESP 7 100 - 500 7
It !••.) - '00 10 Baghouie 9 <10 «
1' -0.1 11 Scrubber 9 500 - 1000 U
!•' u.3 - 1 12 50 - 100 1-.
21 10 - 100 14 >1000 16
22
23
25
25
2*
2
;.i
24
-
-------
TABLE 56. PROBLEM AREA PROFILE OF CAUSAL CODE 4 (UNFORESEEN)
D
00
Rank
I
Paraaeter A
Category
Surface coating
operations
2 Petroleua products and handling
1
Cer»f!nt plants
< Iron and steel plants
;
6
6
B
8
10
Olher
Grain handling operations
Incinerator*
Stone, clay and
Steaa generating
glass plants
plants
Paraaeter B Paraaeter C Paraaeter D
Eall Category y» all c....orv V* all Category V^ all
ranking (101 ton/yr) ^Lfranklng «-""i°*y pranking (103 ac(n) pranking
9 0.1-0.5 6 Scrubber 6 <10
11 1-10 10 Other 7 500 - 1000
14 <0.1 11 ESP 8 50 - 100 ;•
14 100 - 500 11 Baghous* 9 10-50 .1
16 i.5 - 1 i: 100 - 500
17 10 - 100 13 >1000 1J
17
18
18
Pulp and paper aills
Asphalt planta
Aluminua planta
Btjs» and bronie
plants
-
-
-
Petrochenvcal plant*
Lunber and wood
planta
-
Food and drug plants
-------
TABLE 57. PROBLEM AREA PROFILE OF ALL CAUSAL CODES
o
\o
Rank
1
2
3
<•
5
6
7
8
9
10
11
12
U
14
15
16
Parameter A
Category
Cement plants
Surface coating
Cra.n hindling operations
Petroleum products and handl >•
Incinerators
Stone, clay and glass plants
Steam generating planta
Pulp and paper mills
Petrochemical plants
Lumber and wood plants
Asphalt plants
Iron and steel plants
Other
Brass and bronze plants
Food snd drug plants
Aluminum plants
Parameter B Parameter C Parameter D
Eall Category V"* all V"1 all Category V"* all
ranking UO' ton/yr) pranking Late8ory pranking (103 acfm) ^Lrfran-ing
1- 100-500 7 ESP 5 500 - 1000 8
1- 0.1 - 0.5 10 Scribber 1 100 - 500 f
!•« Id - 100 10 Other 8 10 - >0 1C
<0.1 11 Baghouae 9 <10 10
21 1-10 U' MOOO 13
23 0.5 - 1 13 50 - 100 14
23
24
28
28
29
2Y
JO
33
3i
3'
-------
TABLE 58. FREQUENCY INDICATOR PROFILE
R.,k
1
2
1
-
5
6
7
8
9
10
11
12
1 J
14
15
16
Parameter A
Category
Steoa generating plants
Iron and ateel plants
Pulp and paper mils
CeeenC plants
Luaber and wood products
Other
Creln handling operations
Petroleum products and handling
Asphalt plants
Stone, clay and glass plants
Incinerators
Surface coating operations
AluBinun plants
Brass and bronie plants
Petrochemical pleats
Food and drug plants
Frequency Rank
89.3 1
84.2 2
jR.5 3
29.:
28.6 ->
18.9 6
16.5
9.4
6..S
3.9
3.4
2.9
l.1000 1S.C
<0.1 5.5 4 Scrubber 0.5 4 50-100 li.;
lOo - 300 5.4 5 10-^0 o.^
.1 - .5 1.3 6 <10 4.1
-------
with this phenomena. The general approach that may be used in this regard is
to ensure that all process and control equipment is maintained and checked on
a regular basis, as faulty equipment will constantly malfunction or cause the
bypassing of emissions. A requirement that all control equipment have an
agency approved operation and maintenance plan is one recommendation that may
have application in this situation.A followup requirement that this plan be
openly posted near the control deviceT and that the dates of the last mainte-
nance overhaul be added to this plan may also help. This posted plan could
then be inspected as part of a source's annual control agency inspection,
with any lack of proper maintenance noted. Sources must be held accountable
for the upkeep of their control equipment. As previously discussed, those
process operations which cause constant repetitious incident frequencies
should also be investigated and new operating procedures adopted, if possible.
The duration indicator profile is presented in Table 59. Limiting the
duration of an excess emissions incident requires the quick identification
and timely repair of a problem. The best way to ensure a quick response fra
an excess emission, regardless* of"pollutant, is to have an automatic, in-
stack, pollutant monitor that signals the equipment operator when something
has gone wrong. While in-stack devices have been widely used for particulates,
similar monitors can be required for the other pollutants as well. This device
would be especially relevant to hydrocarbon sources, as two of the top four
sources in Parameter A are typically identified with a hydrocarbon emission.
The top ranking of "other" control devices in Parameter C confirms this prob-
lem area as several of these devices are also associated with hydrocarbon
release (vapor recovery systems, afterburners). The colorless nature of this
pollutant makes visible detection difficult, if not impossible, and an in-
stack device keyed to a certain pollutant concentration may prove useful.
Once, identified, the problem must be quickly repaired.
Many long-duration incidents would have been minimized had there been
sufficient spare parts on-hand. A regulatory requirement mandating that
selected key spare parts be stocked on-site, or a committment from a local
supplier that such parts can be made available on short notice may help in
this area. While it is obviously impractical for a source, especially a
small source, to keep spares for all items, requiring them to know where to
locate these spares on short notice can minimize incident durations. A third
suggestion would be to require the source to shut down his process when a
control malfunction occurs.This type of regulation forces the facility to
keep similar parts inventories for control as well as process equipment and
again leads to shorter incident durations.
The magnitude indicator profile is found in Table 60. Limiting the mag-
nitude of an excess emissions incident is difficult. High magnitude emissions
are associated with two primary factors: sources with high uncontrolled
emission rates and complete, as opposed to partial, failure of the control
device. Obviously nothing can be changed or fixed with sources having high
emission rates. One regulatory tool that will help in this area is increased
surveillance. Air pollution inspectors can be trained to identify those"
potential high-magnitude emitting sources and ensure that they are kept under
constant surveillance.Increased inspection frequencies will also help ensure
that these sources regularly maintain their control equipment.Complete con-
111
-------
TABLE 59. DURATION INDICATOR PROFILE
1-0
Rank
1
2
3
4
5
6
7
8
9
10
U
12
13
14
15
16
Parameter A
Category
Petrochanieal plants
Food ani drug plant*
Surface coating opirationi
Petroleua product! and handling
Incinerator*
Stone, clay and glati plant*
Ceoent planta
Umber and wood product*
Grain handling operationa
AluBinun plant*
Other
Stea*> generating plant*
Pulp and paper Bill*
Asphalt plant*
Braa* and bronie plant*
Iron and iteel plant*
Paraaeter B Paraaater C Paraneter D
Duration - . Category Duration _ . /..,.__„_„ Duration - . Cateogrv Duration
(hr) Kank (10> ton/yr) (hr) Rank Cat«8«"y (hr) "-"k (103 acfm) (hr)
2S54.0 l 100 - 500 218.0 1 Other 3097.7 1 <10 ISA. 3
233.0 2 .1 - .J 181.6 2 Scrubber 276S.O 2 500 - 1000 28.5
154.) 1 <.l 85.0 3 ESP 1130.4 3 10-50 24. 4
122. A <• 10-100 13.2 4 Baghouae 15.8 4 >1000 20.3
45.9 5 1-10 11.5 5 50-100 12.0
42.2 6 .5-1 11.4 6 100 - 500 8.1
31.2
26.6
16.8
12.8
9.0
7.2
4.9
2.8
1.7
1.6
-------
TABLE 60. MAGNITUDE INDICATOR PROFILE
Rank
I
2
3
4
S
6
7
8
9
10
11
12
13
14
15
16
Parameter A
Category
Cenent plants
Surface coating operations
Grain handling operations
Brass and bronze plants
Incinerators
Asphalt plants
Stone, clay and glass plants
Pulp and paper mills
Petroleua products and handling
Steam generating plants
Iron and steel plants
Petrochemical
Other
Alunlnun plants
Luaber and wood plants
Food and drug plants
Magnitude
(2 above Rank
allowable)
2463 1
1566 2
1224 3
1150 4
614 5
458 6
253
180
167
163
158
147
123
59
34
25
Parameter B Parameter C Parameter D
Magnitude Magnitude _ Magnitude
..-3 °* ? (J above Rank Category (T above Rank /,03 »„) '* at>ov"
11 ton/yrj ,nouabie) allowable) acini ailowati..j
100 - 500 18,913 1 Scrubber 2817 1 100 - 500 468
.1 - .5 1,292 2 Baghouse 1941 2 10-50 406
10 - 100 571 3 Other 667 3 <10 214
<.l 225 4 ESP 170 4 500 - 1000 192
1 - 10 196 5 50-100 111
.5 - 1 161 6 >1000 100
-------
trol device failures can be minimized through equipment design. Control
devices which can be compartmentalized enable the source operator to isolate
a problem should it develop. This design recommendation may take the form of
requiring one extra compartment in baghouse or precipitator installations,
or the use of two smaller scrubber fans in series as opposed to one large fan.
Any similar control strategy will help in minimizing the magnitude of an emis-
sion once a problem develops.
RECOMMENDATIONS
While this study has pointed out many diverse and interesting aspects
of the excess emissions problem, it has not been without its problems. The
main shortcoming appears to be the size of the sample population investigated.
In spite of our ability to combine and analyze the results of four contractors
source surveys, we feel that too few sources were interviewed to draw meaning-
ful conclusions in certain areas. The inclusion of 16 categories in Param-
eter A, for example, made a review of this data highly dependent on the pecu-
liarities of the individual sources within each category. To remedy this
problem we feel that a few of the top-ranked industrial categories should be
selected for a followup study. This project could follow the same general
outline as this report but include at least 20 to 30 sources in each indus-
trial classification so as to add more significance to the results of the
data tabulation. This followup might also entail investigating a specific
process to determine the cause of excess emissions incidents, investigate how
widespread they are in the industry and make recommendations on how these
problems might best be minimized.
114
-------
APPENDIX A
RAW DATA SUMMARY
115
-------
9/io/79
PAGE
SOURCE
LUDE
B 1. 2.
fr 1. 2.
-» t. 3.
6 1. 3.
« ?. 1.
9 ?. P. 3.
& P. o.
n J. 2.
H u. t.
i» u. 2.
* u. 2.
8 u. 2.
6 o. 3.
a -. «.
B a. a.
B a. u.
B o. 5.
6 «. S.
B «. 5.
H S. 1,
B S. t.
B 5. 1.
H S . 1 .
6 *>. I.
H 5. 2.
B S. 2.
B 5. 2.
B 6. 1.
B 6. 1.
B 7. 2.
B 7. 2.
B 7. 2.
B 7. I.
B 7. 3.
B 7. 3.
B 7. 3.
B 7. 3.
BtO. 1.
Bit. 1.
811. 1.
811. 0.
812. 1.
812. 1.
612. I.
B13. 2.
BIS. 2.
814. 1.
BI«. 2.
BIO. 2.
814. 2.
Bl«. 2.
814. 3.
UNCONTDO
LEO EMIS
SION TRY
26000.0
28440.0
625.0
825.0
200000.0
34000.0
27000.0
16600.0
129&.0
1296.0
1296.0
1596.0
1296.0
828.0
2160.0
2160.0
2160.0
1980.0
1980.0
1980.0
5900.0
S9ao.o
5900.0
5900.0
792.0
792.0
792.0
792.0
6300.0
6300.0
12355.0
12355.0
12355.0
12355.0
21600.0
1 160 0.0
21600.0
21600.0
8600.0
505.0
505.0
372.0
2700.0
2700.0
2700.0
650.0
650.0
1000.0
2000.0
2000.0
2000.0
2000.0
210.0
CREDIT 1
TPY
89.0
89.0
9. a
9.0
fbo.o
138.0
12.5
5t3.o
267,1
287.)
287.3
287.3
287.)
320.0
bB.S
SB. 3
58.3
209.0
209.0
209.0
345.7
3U5.7
305.7
305.7
201.5
201.5
201.5
201.5
8.5
8.5
56.3
56.3
56.3
56.3
1262.0
1262.0
1262.0
1262.0
1512.0
20.0
20.0
9.2
350.0
3SO.O
350.0
6.0
8.0
60.0
80.0
SO.O
BO.O
80.0
tt.o
EXCESS
: MISSIONS
TPY
09.10
09.30
280.00
2BO.OO
0.0
0.0
0.0
0.0
52. 5f.
10.01
U. 00
7.00
7S.50
0.0
12.00
3.00
15.00
18.50
20.50
03.00
36.00
16.00
10.00
68.00
36.00
18.00
10.00
68.00
20.80
20.80
10.00
5.00
10. on
25.00
7.20
3.00
5.00
15.20
0.0
1.35
1.35
0.0
0.83
0.38
1.21
0.20
0.2*
0.0
3.50
3.50
27.00
30.00
s.oo
ALLOWABLE
EMISSIONS
TPY
102.0
102.0
225.0
225.0
600.0
n25.n
50.0
693.0
i79.0
SJ9.0
579.0
579.0
579.0
363.0
726.0
726.0
726.0
396.0
396.0
396.0
669.
669.
669.
669.
«75.
075.
075.
075.
008.0
008.0
202.8
202.8
202.8
202.8
1723.5
1723.5
1723.5
1723.5
2160.0
65.5
65.5
36.0
050.0
050.0
450.0
10.0
10.0
210.0
300.0
300.0
300.0
300.0
70.0
P9.EO DURA
UENCY TIVMI
\0/VR Hi
1 .0 2(90.00
1 .0 2190.00
0.5 P76S.OO
0.5 r>765.00
n . n n , ft
0.0 n.o
0.0 0.0
0.0 0.0
1.0 50u. 00
51.0 5.00
PO.O ).00
2.0 on.no
Su.O 15.10
0.0 0.0
31.0 M.JO
20.0 3.00
51.0 3.80
0.5 500.00
1.0 336.00
1.5 392.00
60.0 1.00
JO.O 0.00
0.0 08.00
90.0 5.90
60.0 0.00
30.0 0.00
0.0 08.00
90.0 5.90
00.0 0.00
00.0 o.OO
20.0 0.00
10.0 1.00
3.0 20.00
33.0 5.80
6.0 2o.OO
1.0 15.00
5.0 20.00
12.0 20.30
0.0 0.0
0.2 8000.00
0.2 8000.00
0.0 0.0
9.1 3.50
0.1 168.00
9.2 0.60
20.0 2.00
20.0 2.00
0.0 0.0
0.2 8760.00
0.2 8760.00
0.6 168.00
1.0 3605.00
0.3 069.00
MAGNT
TUOF
t
69
69
?«1 7
PAI7
0
n
0
n
300
71
too
100
A7
0
5%
75
63
300
300
300
200
200
200
200
?no
200
200
200
693
693
150
150
200
155
too
100
100
100
0
12
12
0
270
270
270
too
100
0
5
5
566
342
1067
ABS.MAG
TOM/
EPisone
U9. ion
119.300
560.000
S60.000
o.o
p. i
0.0
I). I)
5P.5.IO
0.323
o.?no
3.SOA
1 .360
o.o •
0.390
0.150
O.?90
37.000
2u.son
28.700
0.600
0.600
3.500
0.720
0.600
0.600
3.500
0.720
0.620
0.620
O.SOO
0.500
3.300
0.760
1.200
0.750
1.000
1 .000
0.0
7.500
7.500
0.0
0.090
6.500
0.130
0.010
0.010
0.0
17.500
17.500
•5.000
34.000
1S.OOO
CAJ
.SAL
cn.ie
i
5
1
5
S
5
S
S
1
S
1
II
S
S
p
3
«i
P
u
S
2
3
0
5
?
3
0
5
3
5
1
3
u
5
1
3
0
5
5
3
5
5
3
a
5
3
5
S
2
3
4
5
4
CON
THOL
rone
>
p
>
t
> .,
P ••*
U
J
U
1
3
3
t
1
3
3
3
3
3
a
?
?
?
3
3
3
3
1
CAPITAL
CdST
1000*
PI 01
PIOO
70
70
^330"
I-39V,
5tKr
360'
t 169
1 169
1169
t 160
1 16S
^570
06U
UflU
060
65?0
65?n
6520
56?7
5627
5627
5627
9(15
905
905
905
91
91
850
850
850
850
15000
15000
15000
15000
2500
11
1 1
750
750
750
603
859
859
859
859
106
OHM
COST
10005
1
1
1?
1?
17
fu
9*1
7|
7!
7l
71
71
So
70
70
70
l>00
64u
6(10
?0»
?U3
203
203
?60
?60
260
260
R
A
I860
1860
I860
I860
567
250
250
250
:CWTWOL
ST7E
ACF"
5SO
550
?5
25
SuO
Sr,,i
31
SOD
?
?
U
(1
U
U
?
2
7
J>
1
0
a
?
5
5
5
5
3
NQRMA
LtfED
tXCESSt
30.7
30.7
IPO.O
)2o. b
n.a-
• .0
n. )
o.o
9.1
1 .1
0.7
1 .2
l<».7
0.0
1 ./
o.a
2.1
«.7
6.2
10.9
5.0
?.7
2.1
10.2
7.6
3.8
2.9
la.)
6.1
6.1
0.9
2.5
0.9
12.)
0.0
0.2
0.3
0.9
0.0
2.1
2.1
0.0
0.2
0.1
0.3
2.4
2.4
0.0
1.2
1.2
9.0
11.3
7.1
NUNMA
LtZEO
caeonsx
62.7
62.7
U.2
u.2
9). 8
22.1
25.0
74.0
«9.6
"9.6
«9.6
49.6
M?.t>
89.)
8.0
8.0
S.3
b2.d
S2.8
S2.0
51.7
51.7
' 51.7
51.7
02.0
42.0
42.0
02.0
2.1
2.1
27.8
27.8
27.8
27.8
7). 2
73.2
73.2
73.2
70.0
37.3
37.)
25.6
77.8
77.8
77.6
80.0
80.0
28.6
26.7
26.7
26.7
26.7
19.7
-------
9/10/79
P4GE
SOURCE
CODE
B14. 3.
Bib. 1.
bib. 1.
Bib. 2.
Hlb. 3.
Hlb. 3.
Bib. 1.
016. 3.
Bib. u.
Bib. «.
Bib. «.
BIh. u.
416. 5.
416. b.
lib. 5.
Bib. 5.
G 1. 2.
G 1. 3.
G 1. 1.
G 1.11.
G 1.11.
G 1.12.
C 1.12.
G 1.13.
G 1.13.
6 1.14.
G 1.14.
G 2. 1.
G 2. 1.
G u. I.
G u. |.
G S. 1.
G 5. 1.
G 5. 2.
G 5. 2.
G 5. 2.
G 5. 3.1
G 5. 3.1
G 5. 3.2
G 5. 3.3
G S. u.
G 5. u.
G 5. «.
G b. 1.
b. 2.
b. 2.
7. 1.
7. 1.
7. 1.
7. 2.
7. *.
7«
• 9.
7. 4.
UNCONTRO
LED EM IS
SION TPt
210.0
1667.0
1667.0
100.0
'56. 0
756.0
756.0
7S6.0
seeo.o
5860.0
5880.0
5660.0
516.1
SlB. 1
516. 1
518. 1
9U2. '
U469.0
U089.C
286.7
286.7
173.0
171.0
191.9
191.9
202.1
202.1
15117.0
15117.0
218.5
218.5
343310.0
343330.0
156400.0
156400.0
156400.0
76606.0
76606.0
226000.0
136600.0
145526.0
145528.0
I4S528.0
247.5
9225.0
9225.0
7050.0
7050.0
70SO.O
1199.0
1199.0
4.4
112.2
EXCESS ALLOHA9LE
CREDIT EUISSIONS E«IS9IC\S
TPT TPT
11.0 5.00
26.7 0.25
26.7 0.25
27.0 0.0
200.0 1.00
200.0 0.10
200.0 1.20
200.0 2.30
4)0.6 8.00
430.6 2.00
ulO.b (i.OO
UlO.h 16.00
103.0 3.00
103.0 O.BC
103.0 u.OO
103.0 7.60
143.4 0.0
1069.8 u.70
1089.6 u.70
43. S 0.02
41.5 0.02
11.5 O.OS
11.5 O.OS
12.1 0.06
12.3 0.06
12.2 0.02
12.2 0.02
15.5 4.45
IS. 5 4.45
3.
3.
212.
212.
72.
72.
72.
Sb.
56.
174.
32.
0.
0.
0.
1 •
19.
19.
94.
94.
94.
I6i
.
8.
ft. 55
6.55
1.26
1.26
Ib.bS
0.90
17.55
11.00
11.00
0.0
0.0
66.00
110.50
176.50
0.0
0.01
0.01
9.90
l.bS
11. SS
1.90
1.90
0.0
7.0*
TPY
70.0
11.3
11.3
40.0
262.0
262.0
262.0
262.0
Sb4.3
Sbu.l
564.0
S64.0
216.0
216.0
216.0
216.0
172.0
161 1.0
1611.0
56.5
58.5
41.0
41.0
45.5
45.5
46.6
46.6
61.1
61.1
24.0
2u.O
U77.S
U77.5
136.
I3b.
lib.
198.
196.
525.
62.1
219.0
219.0
219.0
2.1
19.0
19.0
172. S
172.5
172. S
1B.1
16.1
»•
• B
17. b
FREO
UENC*
NO/V
0.4
0.2
0.2
0.0
10.0
1 .0
1.0
12.0
UO.O
10.0
1.0
SI .0
UO.O
10.0
0.5
50. S
0.0
O.S
0.5
1 .5
1.5
2.0
2.0
1.0
1.0
1.5
1.5
196.0
196.0
2.0
2.0
5.0
5.0
5.0
9.0
14.0
20.0
20.0
0.0
0.0
1.0
12.5
11.5
0.0
O.S
o.s
700.0
29.0
729.0
48.0
46.0
On
.0
b.O
OUR*
TION
MH
uo9.CO
1450.00
1950.00
0.0
2.00
2.00
16.00
0.80
u.OO
J.OO
166.00
7.20
U.OO
b.OO
16.80
0.56
0.0
life. 00
116.00
0.60
0.60
.50
.50
.30
.30
0.90
0.90
.60
.60
1 IU.OO
1 IU.OO
uS.bO
80
A. 000
0.150
0.0
Q , (1 0 0
Q . 4J Q ft
0.010
0.010
0.021
0.023
0.020
0.020
0.010
0.010
0.023
0.023
3.280
1.2BO
0.250
0.250
1.330
0.100
1 .253
0.550
0.550
0.0
0.0
66.000
a. 640
13.070
0.0
0.006
0.006
0.014
0.126
0.019
0.040
0.040
1.100
CAJ
SAL
coos
5
1
5
5
2
3
u
5
?
\
U
S
?
i
u
5
5
u
5
2
5
2
5
2
S
2
5
2
5
3
5
4
5
2
1
S
3
5
5
5
2
3
5
5
1
S
2
3
S
2
5
1
CON
TROL
ClIDt
1
3
1
2
U
It
a
a
t
?
f
2
?
p
?
p
?
1
1
?
2
?
2
2
2
2
2
U
U
4
U
1
1
1
1
1
3
3
1
I
1
1
1
3
3
1
1
1
1
2
2
i
C»PIT»L
COST
moos
106
1 10
i to
110
i to
US9
211
211
s>>o
560
560
ShO
560
ShO
SbO
560
1269
1269
397
397
looo
1000
1700
1 700
1700
22
123
123
979
979
«7
J 1
197
OJM
COST
10001
1
2"
?«
2a
2-*
l 2<*
120
120
120
120
120
120
120
ft
•
u«
U»
inn
too
t oft
32
\>
1 11
S7
75
75
75
3
7
7
50
SO
1 0
1 w
24
COMPOl CNTi.
SI7F 4f.«
AC*" T-«
t 1 <•
le r
\t-
1- *•
70 *
70 t
7« 1
7" «
1 ->" «
1 (JO s
luQ 1
1 >|9 1
>>5 ;
nS '
6S <
65
g i t
25^
2SS -
SO i
SO 2
SO 2
SO t
SO '
SO ?
50 2
50 2
23 *«
23 «
1
t }
ulO ;
• If
2SO
2SO r
250
too i ;
IOC 17
300 1 '
300 2
160 1.
160 |o
I (> r. | fl
u 2
50 A
5u *
245 7
245 7
245 7
27 12
27 12
9 4
6 1
.'.. A
1. 5
.) . :
0.9
0. 1
O.I
n.i
0. 1
0.0
0.0
7.3
7.1
-7.1
27.1
0. !
0.1
• 2 . ;
•• , "*
1 2. i
S.S
S.S
1. r>
0.0
'•7.0
"b.2
71. 8
0.0
9.0
0.0
S. '
?. 1
7.9
5.0
5.0
OA
• U
43.5
NORM*
LI ZED
CREDITS!
IS. 7
60.2
80.2
67.5
76.)
76.3
76.3
76.3
76.3
76.1
76.1
76.3
U7.7
7.7
7.7
7.7
3.4
7.6
7.6
7(1. U
74.4
31.4
31.4
27.0
27.0
26.2
26.2
25.3
25.3
lu.6
1U.6
ou.S
uU.S
S3. 4
53. u
53.4
26.2
26.2
31.2
14.6
0.0
0.0
0.0
61.9
46. 7
46.7
54.7
54.7
54.7
48. a
46.8
sola
-------
9/14/79
PA&E
ca
UNCONTRO
SOURCE LED EMIS
CODE SION TPT
G 7. o 312.2
(i 7. S
r, 7. 5
G 7. S
G H. |
G A. |
GIU. 1
Git. 1
Gil. 1
Gil. i
Gil. 3
Gil. 3
Gil. u
Gil. 4
Gil. 0
Gil. «
GI2. 1
GI2. 1
GI2. 2
GI4. 1
GI4. 1
GI4. 1
G14. 1
GI4. 2
Gib. 1
GI5. 2
CIS. 2
. 1
. 2
2
£
1 1
. 4
. 4
. 4
. 1
. 2
. 4
. 1
Pio! 2
PIO. 1
Pll. 1
Pll. 2
Pll. 1
Pll. 3
P12. 1
Pll. 3
Pll. 3
P14. 3
PI*. 3
t 1. 3
7 1. S.
168.4
168.0
168.0
87.3
87.3
37.8
8682.0
8682.0
9243.0
211.0
231.0
26.1
26.1
26.1
26.1
1960.0
1960.0
6603.7
61. -I
61. u
61.0
61.0
u|.8
120.0
9.2
9.2
9.3
0.3
4.3
360.0
1*0.0
1441.0
1441.0
•1441.0
406.0
506.0
2200.0
IBO.O
58.0
722.7
1971.0
302.0
0.5
63.0
61.0
6719.0
S.6
I.*
176.0
I7*.0
IM4.0
lMt.9
CREDIT
TPX
8. A
9.5
9.5
9.5
0.2
0.2
14.2
71.0
73.0
80.5
45.4
45.6
9.0
9.0
9.0
9.0
4.6
4.6
191.6
0.0
0.0
0.0
0.0
0.2
22.1
2.0
2.0
0.3
0.0
0.0
0.0
0.0
54.0
54.0
54.0
14.0
10.5
33.0
19.6
10.6
8.5
4.0
78.7
0.4
5.2
5.2
1*3.4
O.I
O.I
11.8
11.*
«».*
4*.*
EXCESS
EMISSIONS
TPY
7.65
1.60
0.11
ALLOWABLE ^?o
EMISSIONS 'F.' n.S
1 0.i. I.S
10.
..s
10. u O.S
10. u 1.5
.5 O.S
.5 O.S
21 .3 0.0
.
.
.
6.
18.0
5.0
O.S
23. S
O.S 0.0
25.
1 0.0
7.3 I.S
7.3 1.5
o.b o.o
1.
1.
1 1.0
1.0
36.0 I.S
16.0 1.5
144.0 1.0
144.0 3.0
144.0 a.O
19.4 0.0
19.
44.
19.
12.
9.
8.
79.
0.
7.
7.
183.
1.
1.
17.
17.
!«•«.
l»4.
o.o
0.0
0.0
0.0
0.0
0.0
0.0
0.0
12.0
12.
0.
1.
1.
1.
1.
0.
0.
DURA
no*.
HP
27.00
10. AO
3.70
A. 10
2.00
2.00
v.G
IS. 00
15.00
0.0
«*. 00
6 . 0 v
2bb.nn
2u.OO
IA.GC
102.00
72.00
72.00
0.0
1.00
3.00
1AO.OO
5.23
0.0
0.0
233.00
231.00
0.0
SII2.00
5112.00
2920.00
2920.00
462.00
2B.OO
136.50
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
16.00
16.00
0.0
2145.00
2145.00
1760.00
17*0.00
§111.00
•ill,**
•UGN1
TiiOE
S
1 301
558
558
5S8
67
h7
0
o8Au
qAi*u
0
6)
ftl
IOH
306
ISb
203
29S1S
295 IS
0
737
900
900
776
0
0
25
25
0
04
04
900
900
900
900
900
0
0
0
0
0
0
0
0
0
49
49
0
24
24
964
9**
1*07
M07
« -.S.MAG
TON/
EPISODE
1 .100
O.bon
0.220
O.OAO
0.020
0.020
0.0
-1.100
Ml. 100
0.0
3.050
o.oso
3.180
0.220
0.030
0.100
23.000
23.000
0.0
0.030
0.120
7.400
0.210
0.0
0.0
0.180
O.IBO
0.0
0.680
0.680
106.000
108.000
68.700
4.200
20.32S
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.014
0.014
0.0
0.480
0.480
79.700
79.700
7*9.400
7*9.4*0
fAu
SAL
CODE
5
3
a
5
1
5
5
3
5
5
u
5
1
3
u
5
1
S
5
2
3
u
5
5
S
3
5
5
3
S
1
5
1
0
S
5
5
5
S
S
5
5
5
5
3
5
5
3
5
•
S
2
S
en
TR
C3
f
?
y
t
i
l
t
i
3
1
a
u
u
0
3
1
?
2
2
0
a
u
u
o
0
0
2
3
1
3
3
u
4
2
2
1
1
LAPUAL Of CONTROL CNTL *0«MA NQRMA
COST COST SI7E AGE I TZCO LIZEO
10001 lOOOt >CFM rRS EXCESS* CREDITS!
107
20?
202
2U2
101b
lain
h
RP9
6fO
U7b
2010
•>OIM
SI
b2
09b
U96
096
a9(i
33
13
10
10
013
013
995
99]
993
10)
IS?
MO
66
U89
ISO
ISO
I JO
11
12
12
• 0*
•0*
91
9n
90
a?
a?
I
2S
h
h
3
I?
an
us
on
itf
5
ft
1
1
36
100
too
U7
(|7
U7
SO
2A
o2o
oo
an
IS
IS
s
It
11
1 l
52
5?
0
SO
50
ao
u
a
02
«2
u?
02
IB
IP
10
13
11
33
1}
26
b
u
u
1
25
35
II
37
sn
I
8
56
I
7
7
23
I
I
129
1
(i
A
b
II
U
7
20
20
1
I 1
13
1U
ll
a
0
a
a
l
i
i
i
a
u
1
%
1
3
la
a
1
A
II
0
a
a
a
a
03.5
6.3
1.)
7.5
I .0
l .a
0.0
37.1
37.1
0.0
0.0
0.6
0.9
I .1
0.2
2.1
180.0
i«o.o
0.0
10.2
9.6
ftO.7
80.7
0.0
0.0
3.7
3.7
0.0
ttSO.O
MSO.O
07.7
«.7
S6.S
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.
2.
2.
0.
• 0.
«0.
468.
«b8.
27.
27. •
so.o
37.
37.
37.
3.
3.
9u.i
U2.
02.
78.
93.
91.
8e.b
86.S
86.5
86.b
70.8
70.8
90.7
0.0
0.0
0.0
0.0
ilO.O
88.0
32.9
32.9
SO.O
0.0
0.0
0.0
0.0
37.S
37.S
37.S
35.
26.
7«.
99.
84.
92.
SO.
99.
too.
74.
7*.
89.
8.
8.
6*.
*9.
3.
3.
-------
9/10/79
PAGE
UNCONTRO
SOURCE LEO EMIS
CODE SION TPV
1 1.11.
r 1.11.
r 2. 2.
i 2. 2.
T 3. 1.
3. 1.
3. I.
J. 2.
3. 3.
3. 3.
3. 5.
i. b.
3. b.
3. b.
0. 1 .
o. 1.
o. 2.
u. 2.
4. I.
4. 3.1
4. 1.1
4. 1.2
4. 1.2
4. S.
4. S.
4. 5.
T 5. 1.
T 5. 1.
T 6. 1.
T 6. 1.
T b. 1.
T b. 2.
T b. 2.
T 6. 1.
T 6. 4.
7. i.
1. 2.
7. 1.
7. 1.
7. o.
7. o.
7. 4.
a. i.
a. i.
a. 2.
9. 1.
9. 2.
9. 1.
9. 4.
9. «.
4). «.
TIO. 1.
TtO. 1.
643.0
641.0
60000.0
60000.0
2265.0
2265.0
13.4
13. u
522.0
522.0
490.8
196.0
196.0
196.0
3b8.0
368.0
EXCESS ALLOWABLE
CREDIT EMISSIONS EMISSIONS
TPY
ioa.0
19B.O
221.0
221.0
84.2
84.2
1.4
l.o
70.0
70.3
0.5
0.0
0.0
0.0
10.9
10.9
162.5 10.?
182.5 10.2
182.5 10.2
1253.7 15.2
1251.7 15.2
2005. B 1570.2
2005.8 1570.2
619.2 16. B
619.2 16. B
619.2 16. B
65101.0 15. a
65101.0 15.8
1120.0 16.8
1120.0 16. A
1120.0 16.8
2IB7.0 24.9
2187.0 24.9
210.0 4.4
525.
27.
27.
140.
140.
12.
12.
12.
212.
212.
5.
160.
6.
90.
••00.
••00.
••00.
1.
t.
1.4
1.8
1.8
5.2
5.2
2.2
2.2
2.2
•1.6
•J.6
2.0
76.8
0.2
1.1
6.0
6.0
6.0
0.4
TPY
2.bO
2.60
0.75
0.75
O.I 1
0.11
0.01
O.P1
O./S
0.75
0.0
O.Ob
0.39
O.uS
0.10
4.10
1.80
S.bS
7. OS
190.10
190.10
5.94
5.94
0.78
0.40
1.18
0.01
0.01
0.06
0.12
0.18
O.OS
0.05
0.0
0.0
1.04
1.04
4.50
4.50
0.22
1.27
1.49
0.01
0.01
0.0
0.0
0.0
0.0
0.17
0.69
0.86
0.10
0.20
TPY
329.0
329.0
536. P
53s. 0
1 Op. ^
lOf.u
o.O
u.n
12-. 0
12".. P
P.b
u.O
it . o
x.O
55.0
55.0
16.3
1 ". 3
ie.3
107.5
107.5
1970.1
1970.1
67.5
67.5
67.5
87.6
87.
20.
20.
20.
32.
32.
lo. 0
1.8
0.0
4.0
21.1
21.1
0.9
0.9
0.9
87.6
87.6
2.6
77.0
0.4
2.2
90.0
90.0
90.0
O.S
O.S
FRED
•»G/Y =
982. )
8*2. •
15.0
35.0
2.S
? .s
12. C
12.:-
u.S
0 .^
0.0
1 . t*1
O.S
1 .5
O.S
O.S
1 .0
O.S
.s
.5
.5
.5
.5
2.0
0.5
2.5
27.0
27.0
4.0
4.0
8.0
16.0
16.0
0.0
0.0
I.S
1.5
1.5
1.5
0.5
1.5
4.0
1.5
1.5
0.0
0.0
0.0
0.0
1.0
2.0
1.0
1.0
1.0
Ou» A UACNI
T'N TuDE
MP*
0.17
0.17
24.00
24.00
0.19
0.10
O.I 7
0.17
20. on
20.««n
o.c
0.24
o.on
1.51
I2U.OO
120.00
96.00
120.00
Ob. 00
70J.OO
792.00
792.00
792.00
6.00
12.00
9.60
0.13
0.13
0.11
2.00
1.17
0.50
0.50
0.0
0.0
21.00
21.00
45.50
45.50
b.OO
5.19
5.29
O.Of
O.OB
0.0
0.0
0.0
0.0
1.00
2.00
1.67
720.00
2160.00
t
162
162
10
IP
2015
2035
211
211
SI
SI
0
2908
P99P
2098
5b7
Sb7
1038
uSO
«4?
1067
1067
2
2
aia
816
BIB
25
25
250
75
163
33
33
0
0
280
280
567
567
567
521
527
50
50
0
0
0
0
1150
USO
1150
100
SO
ARS."Ar,
TON/
EPISOJt
o.nos
0. jn 1
0.021
O.OPI
o.mo
ii.mo
0.001
o.oni
o. i *r>
o . i * r
0.0
O.OS7
0.780
O.P9H
o.inn
o. i nn
i .*nn
1 1 .100
0.470
126.700
126.700
0.000
0.000
0.390
0.790
0.470
0.001
0.001
0.015
0.030
0.0?3
0.003
0.003
0.0
0.0
0.690
0.690
3.011
3.011
0.040
0.363
0.370
0.001
0.001
0.0
0.0
0.0
0.0
0.170
0.345
O.?90
0.100
o.too
CiJ
SAL
CON CAPITAL
THOL CftST
I** Cf^TUOL
COST SUE
C'.TL
CJ1E COOt IC'OJ IIKOJ ^C^ > rs*
f
5
2
S
i>
S
P
s
5
s
s
1
0
s
i
s
1
0
s
1
5
3
5
2
0
5
?
5
2
3
5
?
5
S
S
3
S
3
S
1
0
S
?
5
5
5
5
0
u
?
?
?
y
f
f
1
1
?
2
P
?
0
0
0
0
0
f
y
f
f
1
1
1
f
t
?
2
?
?
P
1
1
0
511
531
595
SOS
icon
i OO'i
35
IPS
IPS
IPS
1?S
IPS
OS
95
95
PlSOP
67
67
67
54
SO
1?
1?
1?
6
ft
?s
10
30
30
71
71
9
9
0
2?
1 A
S?
\
IPS?
125?
125?
IPS
IPS
60
00
11
31
9
g
7sr
7sr
i
Ib
16
16
330
110
330
314
151
151
ISI
159
1S9
31
11
11
31
33
07
OS
IS
15
31
31
17
17
17
130
130
10
80
OS
6
050
050
450
11
11
1
1
9
0
S
s
1
1
.
b
u
1
1
1
P
)
>
y
t
1
1
t
1
f
?
P
in
10
i
i
i
17
17
7
S
1
^
S
5
1
1
1
1
1
1
S
4
5
IS
15
15
7
7
Ll2tO
t .CiSSX
0.6
n.8
n.l
0. 1
0. 1
n . 1
0.3
C.3
0.6
n.t>
r>. 1
i .5
V. 7
1 1 .2
7.5
7.S
9.8
10.9
" - . 7
176.6
176.8
0.3
0.3
1 .2
0.6
1.7
0.0
P.O
0.3
0.6
0.0
n.2
0.2
0.0
0.0
26.0
?6.0
21.3
21.1
o.S
25.9
30.0
0.0
0.0
0.0
0.0
0.0
n.o
0.2
0.6
1.0
20.0
40.0
NORMA
LliED
C4EOIT3S
60. 1
60. J
41.2
01.2
79.«
79.4
6S.O
85.0
S6.5
Sb.S
19.2
0.0
0.0
0.0
19. B
19.6
55.7
55.7
55.7
10. |
10.1
79.7
79.7
20.0
24.9
20.9
16.0
16.0
80. B
ao. a
60.8
75.7
75.7
31.4
80.5
45.0
45.0
24.6
24.6
44.9
44.4
04.9
49. a
49.8
76.9
99.7
50.0
50.0
6.7
6.7
6.7
80.0
10.0
-------
9/14/7*
PACE
N>
o
SOURCE
CODE
110. 1.
TIO. 2.
TIO. 2.
TIO. 2.
Til. 2.
Til. 2.
TI2. 1.
TI2. 1.
TI2. 1.
Til. 1.
Til. 1.
TIO. 1.
TIO. 1.
Tin. 2.
T14. 2.
T10. 3.
Tin. u.
TI4. «.
114. 4.
110. 5.
TI4. 5.
TI4. 6.
TI4. 7.
114. 7.
TI5. 1.
T16. 1.
UNCONTRO
LED EM|3
SION TPr
1.
519.
519.
519.
956.
956.
19710.
19710.
19710.
95.
95.
80.
80.
468.
468.
1779.
686.
686.
688.
383.
163.
50.
100.
100.
159.
285.
9
i
3
3
3
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
EXCESS ALLOWABLE
CREDIT EMISSIONS EMISSIONS
TPr TP» IP»
O.u
0.0
0.0
0.0
U.I
u.l
191.5
191.5
191.5
12.0
12.0
1.4
1.4
21.6
21.6
214.0
75.2
75.2
75.2
0.0
0.0
2.1
44.0
44.0
0.0
170.1
0.10
64.00
4.50
68.50
1.15
3.15
596.22
8.26
600.50
C.4S
0.05
0.03
0.01
0.01
0.01
0.0
4.15
0.15
4.50
1.05
3.05
0.0
0.01
0.01
0.0
0.0
0.5
78.0
76.0
76.0
9.6
9.6
261.0
263.0
261.0
59.0
59.0
5.0
S.O
u7.0
47.0
267.0
101.2
101.2
101.2
57.6
57.6
2.1
450.0
450.0
15.9
418.0
FRfcQ
SO/rR
2.0
1 .0
0.5
1 .5
17.0
17.0
7?.0
1 .0
71.0
61.5
61.5
6.0
6.0
3.0
1.0
0.0
0.5
0.5
1.0
0.5
0.5
0.0
1.0
1.0
0.0
0.0
DURA
T10N
IOUU.OO
6000.00
I2P.OO
0000.00
50.00
50.00
in. oo
18.00
is.no
1.70
1.70
0.50
0.50
0.11
0.11
0.0
06.00
o.OO
26.00
160.00
160.00
0.0
0.08
0.08
0.0
0.0
MACNI ABS.MAG
Tupf TON/
* EP1SOOF
75
81
Stoh
7.?oo
8.741
8.7R1
8.281
0.007
0.007
O.PPU
o . o n a
0.001
0.001
0.0
6.300
0.700
4.500
6.100
6.100
0.0
0.007
0.007
0.0
0.0
r»u
s
i
?
5
3
5
i
3
s
3
s
3
5
7
S
<3
3
4
5
2
5
5
1
5
5
S
CO*
'will
; >ot
?
u
u
u
^
3
\
3
1
u
u
1
1
u
0
0
u
0
0
0
4
?.
4
u
a
4
CAPITAL
COST
IOPOS
36
168
168
I6A
S"
SB
PIM
l\ A3
t\ 63
100
100
US
46
761
761
781
266
266
266
111
III
19
406
OOA
112
18
IliX CIl'.TROL
COST sm
1000* ACf
?
f
70
? 0
20
19
39
1
1
15
IS
70
126
128
126
202
20?
6
86
66
18
1 1
6
*>
h
1 fr
1 *
1 »»
1 5"
1 3R
3.
3d
16
16
53
51
07
56
56
56
81
81
4
29
?9
1 1
5
CML
TfcS
7
P
^
)
S
S
1
1
1
)
}
1
1
s
s
7
7
7
7
a
a
1
7
7
S
5
LUEO
t >asst
hO.O
*?. 1
S.8
K7.8
U.
31 .
PP6J
\.
779.
3.
n.
0.6
0.0
0.0
0.0
4.0
0.1
4.4
5.1
5.1
0.0
0.0
0.0
0.0
0.0
NORM*
LI7ED
CHEDUSl
80.0
0.0
0.0
0.0
uu.
U4.
72.
72.
72.
20.3
20.3
66.0
66.0
50.2
50.2
60.1
72.9
72.9
72.9
0.0
0.0
100.0
9.8
9.8
0.0
18.9
I
r»
I
S
8
218 RECORDS PRINTED
CONTROL CODES LEGEND
1 - EUctrottotic Pr«cipil«tor (ESP)
2 - Scrubber
3 - B«ghou»«
4 - "Other"
------- |