PB85-1 15 152
EPA-600/3-84-093
October 1984
EVALUATION OF TRAINED VISIBLE EMISSION
OBSERVERS FOR FUGITIVE EMISSION OPACITY MEASUREMENT
by
Thomas H. Rose
Eastern Technical Associates
2412 Atlantic Avenue
Raleigh, North Carolina 27658
Contract 68-02-3480
Project Officer
Willian D. Gonner
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triarvgle park., North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711

-------
PB85-115152
Evaluation of Trained Visible
Emission Observers for Fugitive
Emission Opacity Measurement
Eastern Technical Associates, Raleigh, NC
Prepared for
Environmental Sciences Research Lab.
Research Triangle Park, NC
Oct 84

-------
TECHNICAL REPORT DATA
(Piease read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-600/3-84-093
3. RECIPIENT'S ACCESSION NO.
PM S 115152
4. TITLE AND SUBTITLE
EVALUATION OF TRAINED VISIBLE EMISSION OBSERVERS FOR
FUGITIVE EMISSION OPACITY MEASUREMENT
5. REPORT DATE
October 1984
6. PERFORMING ORGANIZATION COOE
7. AUTHORISI
Thomas H. Rose
S. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Eastern Technical Associates
2412 Atlantic Avenue
Raleigh, North Carolina 27658
10. PROGRAM ELEMENT NO.
CDTA1D/ll - 0409 (FY-84)
11. CONTRACT/GRANT NO.
Contract 68-02-3480
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 10/1/80 to 10/1/81
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A smoke generator was modified to emit white or black smoke in a horizon-
tal direction near ground-level to simulate fugitive emissions. Observers were
placed at a slightly higher elevation than the smoke generator and were then
shown smoke plumes at four different opacity values against a dark terrestrial
background. They read the opacities of the smoke and recorded their observations.
The observers also evaluated smoke plumes from a conventional vertical emitting
smoke generator under the same clear sky lighting conditions to determine that
the observers had no bias when viewing normal plumes.
For the white smoke, the observer's opacity readings of the simulated fug-
tive emission plumes were not significantly different than the opacity readinqs
of the conventional plumes for opacities around 15-20%. Above 20% opacity, the
observer readings of the white fugitive plumes become increasingly lower than
the vertical plume readings. At 40% opacity, they were lower on average by
about 8% opacity. For the black smoke, the observer readings of the simulated
fugitive plumes were lower at all opacity levels. At 15% opacity, they were
on average lower by about 5% opacity, and at 40% opacity, they were lower
by about 11% opacity. For both white and black simulated fugitive emission
plumes, the observers sensitivity to changes in opacity .levels declined
relative to the conventional verticle stack plumes..
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI I'icld/Croup



18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
27
20. SECURITY CLASS (This page;
UNCLASSIFIED
22. PRICE
EPA Farm 2220-1 (R.v. 4-77) PACVIOUJ EDITION It OBSOLKTt

-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse-
ment or recommendation for use.
ii

-------
ABSTRACT
The stack of a visible emission training school smoke generator was
modified to emit white or black smoke in a horizontal direction near
ground-level to simulate fugitive emissions. Trained awoke observers
measured the opacity of these simulated fugitive emissions against a
dark terrestrial background and under clear sky lighting conditions.
They measured both white and black emissions generated at four different
opacities  15, 20, 30 and 40%. The observers also measured the same
opacity of sr>oke plumes emitted from an unmodified training school smoke
generator. Th^se vertical-flow gnokestack plunes were evaluated against
a sky background to provide reference data on the observers ability to
measure the opacity of conventional stack plums under the same clear sky
lighting conditions used for the fugitive plume tests.
The observers measured opacities of white fugitive emissions with
accuracies similar to the conventional stack emissions when opacities
were in the range of 15 to 20%. As opacity increased, however, the
observers increasingly underestimated opacity. At 40% opacity, observer
measurements were lower on average by about 8% opacity. The opacities
of black fugitive emissions were underestimated at all opacity levels.
Opacity was lower on average by 5% opacity at the 15% opacity level and by
11% opacity at the 40% opacity level. It was also observed that for joth
white and bl ck simulated fugitive emission plunes, the observers sensi-
tivity to changes in opacity levels declined relative to the conventional
vertical stack plinies.
This report was submitted in fulfillment of Contract No. 68-02-3480
by Eastern Technical ssociates under the sponsorship of the U.S. Environ-
mental Protection Agency. The report covers the period of September 30,
1980 to September 30, 1081 and work was completed November 16, 1981.
iii

-------
FIGURES
Number	Page
1	Modified smoke generator for emitting simulated fugitive
emission plunes of known opacities . . . . 		5
2	Conventional smoke generator for emitting verticle
stack plumes of known opacities 	 6
3	Observer opacity evaluations versus transmisscmeter
measured opacities of conventional vertical white smoke
plumes viewed against a sky background on a clear day	13
4	Observer opacity evaluations versus transmisscmeter
measured opacities of conventional vertical black smoke
planes viewed against a sky background on a clear day 	 14
5	Observer opacity evaluations versus transitussometer
measured opacities of simulated white fugitive plumes
viewec" against a terrestrial background on a clear day	16
6	Observer opacity evaluations versus transmisscmeter
measured opacities of simulated black fugitive plumes
viewed against a terrestrial background on a clear day 	 17
TABLES
1	Conventional Smoke Transnissometer Calibration Data 	 8
2	Simulated Fugitive Smoke Tranonissonneter Calibration
Data	9
3	Test Conditions			12
vi

-------
CONTENTS
Abstract				iii
Figures 		 		vi
Tables	.... .	vi
Acknowledgements 				vii
1	Introduction 	 		1
2	Conclusions and Reccnmendations 		 2
3	Experimental Design . 		 3
Test Location ............. 	 3
Snoke Generators 	 3
Observers				3
Observer Tests 		4
4	Data Acquisition and Quality Assurance 		7
Generator Calibration 	 		7
Generator Calibration Error . . . . . . 			7
Generator 2oro and Span Drift ...... 		7
Neutral-Density Filter Specifications 		7
Strip Chart Data Rscord 		10
Confidence Intervals-Generator Calibration and
Operation ............... 		10
Observers ................ 		10
Dacumentation . . . 			 . . 		10
Tests-Plume Configuration . 				11
Obscuring Mechanism 		11
5	Results . . 				. . 		12
References					18
Preceding page blank

-------
ACKNOWLEDGMENTS
Hie cooperation and assitance of William Conner of the U.S. Environ-
mental Protection Agency is gratefully acknowledged. We are particularly
indebted to Ken Averitte, Eugene Jackson, Jerry Clayton, David Daniel,
Donna Quidley, Ernie Fuller, Bryant Foust, and Gary Saunders who partici-
pated as gnoke readers in the field portion of the study.
The modification of the smoke generator was designed by Mr. Willie Lee
of Environmental industries. Analysis of the data was performed by
Ms. Judy Maples, Mr. David Savage and Mr. David Brande of Eastern Technical
Associates.
vii

-------
SECTION 1
INTRODUCTION
Many types oE airborne particles and aerosols are not emitted from
conventional vertical stacks. Instead, these emissions may come from
windows, roof monitors, vents, conveyor belts, hoppers, storage piles,
construction sites, roads, and a variety of other sources. Often these
unconfined emissions originate close to the ground and move with the wind
in a horizontal direction; however, unlike conventional smokestack plumes
that are viewed against the sky, these emissions are usually viewed
against a darker terrestrial background. Researchers such as Conner
(1,2), Himil (3), and Hood (4) have long recognized that the contrast
betwen the plume and the backgrouind have an effect on the opacity assigned
by observers to the emissions. The introduction to EPA Revised Method 9
states: "Studies of the theory of opacity and field studies have demon-
strated that a plume is most visible and presents the greatest apparent
opacity when vievred against a contrasting background (5)." Atmospheric
visibility studies have reached the same conclusions and even in nature
it is well known that the background matching ability of chameleons
reduce their visibility.
This study was designed to determine the biases, if any, introduced
when a certified Method 9 snoke reader assigns opacity values to ground-
level emissions viewed against non-sky backgrounds.
A conventional snoke generator, used in the training and certifi-
cation of trained visible emission (VE) observers, was modified to emit
anoke plumes horizontally below the eye level of the observers, who then
viewed the sroke plumes against a terrestrial background. To determine
that the observers selected for the panel were capable of making unbiased
opacity readings of vertical-flow stack plumes under good viewing con-
ditions, an unmodified generator was used to produce vetical plumes for
evaluation at the same time.
1

-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
The study indicated that the panel of smoke readers evaluated the
opacities of simulated white (horizontal-flow) fugitive emissions wi'-h
accuracies similar to the (vertical-flow) stack plume opacity measurements
when opacity levels were around 15 - 20%. As opacity levels increased,
howaver, the panel increasingly underestimated them; opacity was lower on
average by about 8% opacity at the 40% opacity level.
Black (horizontal-flow) fugitive amissions were underestimated at
all opacity level c.. At 15% opacity, the observers measured the emissions
5% opacity low; at 40% opacity, they measured them 11% opacity low.
Overall, the observer panel's sensitivity to opacity changes was less for
simulated (horizontal-flow) fugitive emissions than for conventional
(vertical-flow) stack plunes. The change in panel opacity measurement
per change in transraisscmeter opacity measurement declined frcm near 1.00
for stack plumes to 0.62 and 0.77 for white and black fugitive plumes,
respectively.
It is recommended that controlled tests of observer and transnis-
scmeter opacity measurements of simulated black and white fugitive
emissions be conducted for other environmental lighting conditions, e.g.,
on an overcast day. It is also reccnmended that a lower opacity level
around 5 or 10% be included in the tests.
2

-------
SECTION 3
EXPERIMEOTAL DESIGN
TEST LOCATION
The field tests ware conducted at a location in Raleigh, North
Carolina where the topography of the site was well suited for the study.
An embankment allowed the readers to be located above the fugitive smoke
generator where they viewed the plunes against a terrestrial (vegetation)
background. For the conventional vertical stack plume simulation the
other generator was used and the observers moved down the embankment
where their view was more representative of normal stack plumes with a
sky background.
SMOKE GENERATORS
Two smoke generators ware utilzied during each field test. The
generators v^re commercially available units that are camonly used at
visible emission training schools. Both generators meet the EPA specifi-
cations for Method 9 training and certification as specified in the
Federal Register (5).
One generator was modified to emit a horizontal plume near ground
level. The modification was achieved by leaving the stack in the horizon-
tal (stored) position and installing a 90 elbow between the lower
vertical section of the stack and the upper horizontal section of the
stack. The elbow was caulked and shimmed to prevent leaks. This method
of generating a horizontal piune was selected after stratification tests
shewed that the smoke remains uniformly dispersed in both vertical and
horizontal axes near the duct exit (where a transmisscmeter was located)
and in the stack exit plumes. The strati "ication tests consisted of
viewing a white plume against a black velv^r background at close range,
and viewing a black plume against a white background. The appearance of
both plumes was uniform along both axes. It) further simulate the fugi-
tive nature of a horizontal fugitive plume, a panel was; placed in front
of the stack end in order to block the observers' direct view of the
emission exit point.
The other smoke generator was set up with the stack in the normal
upright position to emit vertical-flow snokestack plumes in the conven-
tional manner.
OBSERVERS
The observers selected for the study were six inspectors frcm a
state agency and two air pollution personnel frcm local contractors.
Each participant possessed a current Method 9 certification and a past
3

-------
record of accurate snoke reading, as evidenced by his ability to certify
promptly at a smoke school. The Method 9 certification procedure is
described in the Federal Register (6). The participants ranged in age
from 26 to 38 years old with an average age of 31. Three possessed
college degrees, three had associate degrees and the other two had sane
education beyond high school. Their experience at reading visible
emissions ranges from one to eight years, with an average of 4.1 years.
Since the test session was relatively long, observer fatigue was of
concern. The general concensus of the observers, hoover, was that no
fatigue problem was apparent despite the fact that they were required to
read emissions for six hours, three in the morning and three in the
afternoon.
OBSERVER TESTS
Both black and white simulated (horizontal-flow) fugitive emissions
and (vertical-flow) stack plumes were tested under clear sky conditions.
The vertical-flow plumes v*ere tested to confirm that the observers were
able to assign unbiased opacity values to conventional stack plunes.
During each test, the observers were required to observe for 12 minutes.
Within that 12-rr>'nute period, there was a period of just over six minutes
of steady opacity groke. The average o 24 consecutive readings by an
observer during this six-minute period was the observers fciPA Method 9
evaluation of the snoke opacity. The opacity levels selected for the
study were 40%, 30%, 20%, and 15%. The observers were exposed to the
sane levels from each generator, but in randem order. Observers weve
given no review of emission opacities prior to testing and were given no
instructions except how to complete the field test form and to read on
signal. Opacity values were recorded on computet compatible data forms
in bound books. The observers were instructed to read by i prerecorded
vape at 15-second intervals to assure Method 9 compatibility and standard-
ization of the time reference. At the same time, the data logger made a
reference mark on the transmissometer strip chart record.
To minimize any potential bias effects that may result frcm observer
fatigue or change in the sun angle, the vertical generator was operated
for only a few runs, then the horizontal generator was operated. This
cycle was repeated throughout the six-hour test. Consequently, observers
were required to shift their observation positions between those illus-
trated in Figures 1 and 2.
4

-------
Emission Point
Obscuring Board
Fig-,ne 1. Modified smoke generator for emitting simulated fugitive
emission plumes of known opacities.
5

-------
'-'.t
.."V >
Fi.ur. 2. Conventional .ok g.n.r.to. to, fitting vrtUL .tk
plumes of known opacities.
6

-------
SECTION 4
DATA ACQUISITION AND QUALITY ASSURANCE
All transmisscmeter data were accumulated on a strip chart recorder
operated at one inch per minute. The recorder unit was connected to the
trananissometer of each of the generators during its phase of operation.
In addition to the continuous strip chart trace of the opacity levels,
the observation times vere also marked on the (.harts.
Each VE observer was provided a bound book of data forms adequate to
last over the test period. The forms vrere designed for ease in data
reduction and to provide information on the test type, test number,
position of the observer in relation to the sun, and the viewing condi-
tions.
GENERATOR CALIBRATION
Both generators were span calibrated after allowing a minimum of 30
minutes warmup by alternately producing simulated opacities of 0% and
100%. The stable 0% and 100% opacity values of each generator were
verified by alternately switching the power to the licst source on and
off while the smoke generator was not producing smoke. This procedure
was performed prior to conducting the laboratory calibration error tests
described below, and the actual field session of the fugitive emissions
study.
GENERATOR CALIBRATION ERROR
The error between the ritual response and the theoretical linear
response of the snoke meter was measured by inserting a series of three
NBS traceable neutral-density filters of nominal opacity of 20, 50, and
75% in the anoke meter pathlength. The calibration error for the
trananisscmeter in each generator was determined by the filter insertion
procedure described in the Federal Register (5). The calibration results
are shown in Tables 1 and 2.
GENERATOR TRANSMISSOMCTER ZERO AND SPAN DRIFT
Zero and span checks of the transmissometers were performed periodi-
cally to measure the amount of opacity drift from the generators. This
procedure was conducted at the end of every two tests.
NEUTRAL-DENSITY FLTER SPECIFICATIONS
National Bureau of Standards (NBS) traceable neutral-density filters
of ncminal opacity of 20, 50, and 75% were used to calibrate the anoke
generators. These filters were calibrated within + 2%.
i

-------
adi
mbc
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
CONVENTIONAL SMOKE TRANSMISSOMETER FILTER CALIBRATION DATA
Transmissometer
Reading	
72
22
46
72
22
46
72
22.5
46
72
21.5
46
72
21.5
46
23.5
46.5
72.5
Filter
Value
73
23.5
47
73
23.5
47
73
23.5
47
73
23.5
47
73
23.5
47
23.5
47
73
8

-------
'LjLj
ad
mb
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
ia
SIMULATED FUGITIVE SMOKE TRANSMISSOMETER FILTER CALIBRATION DATA
Transmissoroecer	Filter
Reading
Value
80.5
80
52
52
18.5
.19
80.5
80
52
52
18.5
19
80.5
80
52.5
52
20
19
80.5
80
52
52
18.5
19
80.5
80
52
52
19
19
21.5
19
53
52
80.5
80
9

-------
STRIP CHART C&TA RECORD
Each test ws~ recorded on strip chart paper by a continuous trace of
the opacity and listing the following information: * - start point, + -
stop point, test number, test type, each reading number and point at
which a reading was made. Calibration error checks included filter
value - percent opacity, measured value - percent opacity, calibrated by,
date of calibration, location, generator, and filter set number.
CONFIDENCE INTERVALS - GENERATOR CALIBRATION AND OPERATION
Statistical analysis was conducted to determine the 95% confidence
interval. Information rvacessary for this evaluation were (1) filter
calibration error (from manufacturer), (2) filter calibration error check
(actual filter value vs meas.ured filte" value), (3) plume configuration
tests - one test for each of the four luvels \
-------
Generator. Operator Log
A generator data form, comments sheet, and plume configuration curves
for all tests were in bound book format to assist the generator operator
in providing accurate opacity levels for the tests. The generator data
form includes a test number, test type, flat trace opacity percent, time
period for flat trace opacity, generator operator, generator type - ETA
for horizontal or EPA for vertical, time of test, recorder chart speet,
and catments.
TEST-PLUME CONFIGURATIONS
Emissions were generated at four different opacity levels. The
levels were selected at random by die tossing. Each test was conducted
over a 12-minute period (48 readings at 15-second intervals). The timing
of the test and reading intervals were verified by stopwatches. During
this time period, a six-minute period for the selected flat trace opacity
was generated for the required 24 readings at 15-second intervals. Flat
trace opacity percent levels for both black and white smoke were (1) 15%,
(2) 20%, (3) 30%, and (4) 40%.
OBSCURING MECHANISM
An obscuring mechanism (plywood board) was used to prevent the
readers from directly viewing the emission point frcm the hori7>ntal
stack of the generator. This was done to make the plume more analogous
to a fugitive plune.
11

-------
SECflON 5
RESULTS
To observe the effect each smoke color and mode of emission had upon
the observers when they evaluated fugitive emissions, observer results
were plotted against actual transmisscmeter values. All tests were
conducted on a clear day between 9 AM and 3 FM.
Table 3 presents a matrix of the test conditions and a list of the
corresponding figures where the data are shown.
TABLE 3. TPST CONDITIONS
Plume Configuration
Smoke
Color
Background
Fiqure
Conventional stack plune
vfoite
Clear Sky
->
Conventional stack plume
Black
Clear Sky
4
Simulated fugitive plume
Vjhite
Dark terrestrial
5
Simulated fugitive plume
Black
Dark terrestrial
6
TVie test results of VE observer measuranents of (vertical-flow) stack
plumes under blue sky conditions (Figures 3 and 4 for white and black
plumes, respectively) show that panel (averaged observer) opacity measure-
ments and their regression lines ware close to the 1 to 1 line of agree-
ment with the stoke generator tranamissometer opacity measurements over
the range of opacities used in the study. Slopes of 1.17 and 1.01 were
observed. The between-observer variation of the data as evidenced by the
confidence intervals of the means was more than expected and previously
observed for the method (2,3). The reason for this increased variation
was not investigated but may have been due to the procedure of alternating
the observers betwen the smoke generators during the test. The between-
observer variation is not the focus of this study. It is the panel
(averaged observer) opacity readings when viewing the conventional and
simulated fugitive plumes that are compared to determine any bias in
Method 9 trained VE observers when evaluating fugitive plumes.
12

-------
60
71
B
/
0
MEAN ~ 95%
CONFIDENCE INTERVAL
REGRESSION LINE
r
30
4 0
i
50
TRANSMISSOMET E R
% opaci ty
6 0
Figure 3. Observer opacity evaluations versus transraissometer measured
opacities of conventional vertical white smoke plumes viewed
against a sky background on a clear day.
13

-------
7
V
MEAN i 95%
CONFIDENCE INTERVAL
REGRESSION LINE
2 0
~t
3 0
4 0
50
TRANSMISSOMETER
% opaci t y
6 0
Figure ^ ODserver opacity evaluations versus transraissoraeter measured
opacities of conventional vertical black smoke plumes viewed
against a sky background on a clear day.
14

-------
The simulated fugitive emissions test results for white plunes on a
clear day (Figure 5) indicate no systematic difference (bias) between
observers and transnisjometer values for opacities around 15 - 20%. At
higher opacities Figure 5 indicates an increasing negative observer bias.
At 40% opacity, this bias resulted in observer panel readings around 8%
ooacity lower than the transsmissometer values. The black simulated
fugitive auissionij test results (Figure 6J indicate lower observer panel
values at all opacity levels. At 15% opacity they wera 5% opacity low,
and at 40% opacity they were 11% opacity lower than the tcansmisscmeter
values. Ttie data also indicate that the observer sensitivity to opacity
changes, as evidenced by the slopes of the curves, declined for both the
white and black fugitive emission plumes. The change in observer measured
opacities per change in transnissoroeter measured opacities declined fvcu
near 1.0 for the conventional verticle stack plume measurements to 0,62
and 0.77 for the white and black simulated fugitive plume measurements.
15

-------
6 0-
4 0-
MEAN ~ 95%
CONFIDENCE INTERVAL
REGRESSION LINE
50
3 0
TRANSMISSOMETER
% opacity
Figure 5. Observer opecJ.ty evaluations versus, transmissometer measured
opacities of simulated white fugitive plumes viewed against
a terrestrial background on a clear day.
16

-------
6 0
5 0-
2 0-
MEAN ~ 95%
CONFIDENCE INTERVAL
REGRESSION LINE
3 0
6 0
T R ANSMISS 0 MET E R
% opacity
Figure 6. Observer opacity evaluations versus transmissometer measured
opacities of simulated black fugitive plumes viewed against
a terrestrial background on a clear day.
17

-------
REFERENCES
1.	Conner, W.D. and J.R. Hodkinson. Optical Properties and Visual
Effects of aroke-Stack riunes, U.S. Public Health Service Report No.
999-AP-30. NTIS PB 174-705, Springfield, Virginia, 1967. S9 pp.
2.	Conner, W.D. A Comparison between In-Stack arvd Plume Opacity Measure-
ments at Oil-Fired Fower Plants, In: Proceedings of the Fourth
National Conference on Energy and the Environment, Dayton Sect. Am.
Inst, of cfrem. Eng., Dayton, Ohio, 1976. pp. 478-83.
3.	Hamil, H.F., R.E. Thomas, and N.F. Swynnerton. Evaluation and Collab-
orative study of Method for visual Determination of Opacity of
Emissions from Stationary Sources, NTIS PB 257 948/OBA, Springie\d,
Virginia, 1975. 70 pp.
4.	Hood, K.T., and A.L. Caron. The Relationship between Particulate
Mass Emission Pate and Observed Plune Appearance from Kraft Recovery
Furnaces, Paper 74-AP-08, presented at FNNIS-APCA Meeting, Boise,
Idaho, November 197h.
5.	Federal 'Register Vol. 39, No. 219:39372-39875, November 12, 1974.
6.	Federal Register Vol. 39, No. 247:24877, December 23, 1971.
18

-------