EPA-650/2-74-026
March 1974
Environmental Protection Technology Series
m
m
m
m
i*
ii
II
A me r i c a n Pet r oJeum;i:
1801 K SU««t>
llpilliiililiiilli;
-------
I/O
EPA-650/2-74-026
INVESTIGATION OF PARTICULATE
EMISSIONS FROM OIL-FIRED
RESIDENTIAL HEATING UNITS
by
R. E. Barrett, D. W. Locklin, and S. E. Miller
Battelle, Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
Contract No. 68-02-0230 (Task 9)
ROAP No. 21AFE-07
Program Element No. 1AB015
Project Officer: Robert E. Hall
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
AMERICAN PETROLEUM INSTITUTE
COMMITTEE ON AIR AND WATER CONSERVATION
1801 K STREET, NW
WASHINGTON, D. C. 20006
and
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D. C. 20460
March 1974
-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
ABSTRACT
Two residential oil-fired heating units (a warm-air
furnace and a boiler) were fired in the laboratory while
Bacharach smoke and filterable particulate emissions were
measured at several excess-air levels for both cyclic and
steady-state runs. In addition, particle-size distributions
were measured during runs on the boiler to determine if particle-
size variations might help explain the lack of correlation be-
tween smoke and particulate emissions, based on earlier field
measurements.
One unit was found to produce essentially no ex-
cessive smoke or particulate on startup; smoke one minute
after startup was equal to steady-state smoke, and particulate
emissions for cyclic and steady-state firing were equal. The
second unit produced high smoke on startup (relative to the
steady-state smoke) and produced higher particulate emissions
for cyclic runs than for steady-state runs, suggesting that
this unit had a significant "on puff" on startup.
It was determined that particulate emissions varied
with excess air in the same pattern as smoke number, being
higher at low excess air levels for both units. Correlations
between smoke and particuJate emissions appeared practical for
individual units firing at specific operating conditions. How-
ever, the data did not suggest that a general correlation between
smoke and filterable particulate emissions exists for cyclic
operation. For the two units examined here, particle-size dis-
tributions indicated that over 80 percent of the particles were
below 1.0 micron and that the particle-size distributions were
nearly identical for all runs.
ill
-------
ACKNOWLEDGMENT
The s-jchors v ish to acknowledge the assiscancs and helpful comments
of the EPA Project OHlcci', Kchert fi. Hal J, .ir.d the API SS-5 Task Force
during the course- of this puuj,i'dm. Membjrship on the SS-5 Task Force
was as follows:
E. Landau (Chairman). . . . Asiatic Petroleum Corporation
R. C. Amero Gulf Res&irch 6. Development Company
S. P. Cauley Mobil t.il Corporation
H. E. Leikkanen Texaco Inc.
B. L. Mickel American Oil Company
R. E. Pater son Chevron Research Company
C. W. Siegnund Esso Research ft Engineering Company
R. A. Seals National Oil Fu*l Institute, Inc.
J. R. GouJd American Petroleum Institute
IV
-------
TABLE OF CONTENTS
Page
OBJECTIVE 1
BACKGROUND 1
APPROACH 2
Heating Equipment Studied 2
Measurement Techniques and Equipment 3
Particle Sizing 3
Conditions at Which Measurements Were Made 4
COMMENTS ON FILTER MEDIA 5
RESULTS 6
Data Summary 6
Smoke Versus Time During Cycle 6
Smoke Versus Excess Air 9
Particulate Emissions Versus Excess Ait 10
Particulate Loading Versus Smoke 11
Particle Size 12
CONCLUSIONS 13
Additional Information Koeded 14
REFERENCES 16
-------
FIGURES
No.
1 Schematic Diagram Showing Principle of the
Cascade Impactor 17
2 Smoke Versus Time During Cycle -- Unit 36 18
3 Smoke Versus Time During Cycle -- Unit 37 19
4 Smoke Versus Excess Air -- Unit 36 20
5 Smoke Versus Excess Air -- Unit 37 21
6 Filterable Particulate Emissions Versus Excess
Air -- Unit 36 22
7 Filterable Particulate Emissions Versus Excess
Air -- Unit 37 23
8 Filter Catch Versus Excess Air -- Unit 36 24
9 Filter Catch Versus Excess Air -- Unit 37 25
10 Filterable Particulate Loading Versus Smoke --
Unit 36 26
11 Filterable Particulate Loading Versus Smoke --
Unit 37 27
12 Particulate Loading (Based on Filter Catch)
Versus Smoke -- Unit 36 28
13 Particulate Loading (Based on Filter Catch)
Versus Smoke -- Unit 37 29
14 Particulate Loading (Based on Filter Catch) Versus
Smoke at 1.0 Minute for Cyclic Runs 30
15 Filterable Particulate Loading Versus Smoke --
Cyclic Operation of Units 36 and 37 31
16 Particle-Size Distribution for Unit 37 at 35 Percent
Excess Air 32
17 Particle-Size Distribution for Unit 37 at 29 Percent
Excess Air 33
18 Particle-Size Distribution for Unit 37 at 27 Percent
Excess Air 34
19 Particle-Size Distribution for Unit 37 at 26 Percent
Excess Air 35
20 Particle-Size Distribution for Unit 37 at 23 Percent
Excess Air 36
vi
-------
Figures (Cont)
No.
21 Particle-Size Distribution for Unit 37 at 18 Percent
Excess Air 37
22 Part iculate on Fiberglass Filter 38
23 Clean Silver Filter (1000X) 38
24 Clean Silver Filter (5000X) 39
25 Parti culate on Silver Filter (20X) 39
26 Particulate on Silver Filter QOOX) 40
27 Particulate on Silver Filter (500X) 40
28 Particulate on Silver Filter (1000X) 41
29 Particulate on Silver Filter (5000X) 41
vii
-------
INVESTIGATION OF PARTICULATE EMISSIONS FROM
OIL-FIRED RESIDENTIAL HEATING UNITS
by
R. E. Barrett, D. W. Locklin, and
S. E. Miller
OBJECTIVE
The objective of this limited laboratory study was to investi-
gate the relationship between particulate emissions and excess air and
the relationship between particulate emissions and Barharach smoke number
from oil-fired residential heating units.
BACKGROUND
Battelle-Columbus has conducted a two-year field study of
emissions from 33 oil-fired residential heating units and 13 commercial
(1 2)*
boilers ' . During the course of that study, gaseous emissions (CO,
hydrocarbons, SO , and NO ) were measured at many different operating
conditions. However, due to the time and cost associated with particu-
late sampling, data were collected on particulate emissions at only a
few operating conditions. Consequently, several questions remained un-
answered after completion of the field study. These questions included:
(1) What is the relationship between particulate emissions
and excess air? (During the field study, particulate
emissions were measured at only one excess-air level
for each burner condition.)
(2) Is there a relation between particulate emissions and
Bacharach smoke number? (The field study data showed
no satisfactory correlation between particulate emis-
sions measured during cyclic operation and smoke
measured near the end of a 10-minute "on" time, as
normally measured by burner servicemen.)
* References are given on page 16.
-------
(3) Do particle-size variations explain the lack of
suitable correlation between participate emissions
and smoke reading?
(4) How much do transients at startup and shutdown con-
tribute to particulate emissions?
In an effort to obtain data that might answer some of the
above questions, the American Petroleum Institute and the U.S. Environ-
mental Protection Agency sponsored this laboratory study to measure par-
ticulate emissions, smoke, and particle size while firing two residential
oil-fired heating units at several operating conditions.
APPROACH
Battelle's approach to this study was to install two typical
oil-fired heating units within the laboratory and to fire these units
at a range of conditions while collecting data on filterable particulate
emissions, Bacharach smoke, and particle-size.
Heating Equipment Studied
The two heating units that were selected for this study were
intended to be representative of different types of oil-fired heating
units: a warm-air furnace and a hot-water boiler. The furnace was
equipped with a light-weight ceramic combustion chamber liner and a
conventional gun burner, while the boiler was equipped with a heavy re-
fractory (firebrick) liner and a flame-retention gun burner. The speci-
fic units were as follows:
Unit 36'
Up-flow oil furnace with ceramic-felt liner and 1.0 gph conven-
tional high-pressure gun burner. (This unit was identified as
Unit 35 in the Phase II studies^2^.
Unit 37*
Dry base, vertical fire-tube steel boiler with a dense cast-
refractory combustion chamber and a 1.0 gph high-pressure,
flame-retention-type oil burner with 3450 rpm motor.
* Unit numbers were continued sequentially from numbers used in the Phase I
and II reports. Unit 36 was given a new number as the air damper had been
modified since it was run as Unit 35.
-------
The units were equipped with solenoid shut-off valves (nondelay type)
to eliminate possible variations due to pump shut-off.
To increase the variety of equipment included in this study,
the burner was interchanged and one run was made witli the flame-retention
burner firing in the warm-air furnace. The combination of the flame-
retention burner firing in the warm-air furnace is identified as Unit 38.
Measurement Techniques and Equipment
Measurement techniques and equipment used in conducting this
study were identical to those described in Reference 2 with the follow-
ing exceptions:
(1) NO and NOX were not measured as most of the data from
the field study showed fairly consistent levels of
NOX emissions from oil-fired residential heating units.
(2) A back-up filter was used on the EPA particulate
sampling train to catch any material not collected
by the first filter. Particulate weights are based
on the sum of the mass collected on the two filters.
(3) The material collected in the impingers of the EPA
particulate sampling train was not dried and weighed,
as the interest was in correlation with filterable
particulate.
(4) For the runs on the first units examined (Units 36
and 38), fiberglas filters wore used in the EPA
sampling train [as specified by the EPA Method 5
procedure^)]. However, due Lo problems associated
with the hygroscopic nature of the fiberglas filter
and its fragility, silver LLiters were used for
runs on Unit 37. (Silver filters were used for the
Phase I and Phase II field studies.)
Particle Sizing
Particle-size measurements were made for six runs in Unit 37
using the Battelle Cascade Impactor. The design of the impactor is based
on the principle of particles in a moving aerosol impacting on a slide
placed in the air stream. The impactor classifies particles in the
-------
range of 0.25 - 16.0 microns into seven categories. If a particle is
sufficiently large it will impact on the first stage; the smaller
particles will continue to travel around the slide to subsequent
stages. The jet diameter of each succeeding stage decreases. Thus,
the particle will increase in velocity for various stages until it
(4)
obtains sufficient inertia to impact as illustrated in Figure 1
Procedure. Flue-gas was sampled from the center of the stack.
For the sampling done for this study, a 5/8-inch-diameter sample probe
about 20 inches in length (from probe tip to impactor inlet) was used,
It had a 90-degree, 6-inch radius bend such that the probe tip pointed
upstream. The impactor was operated horizontally and was heated to
stack gas temperature (about 400 F) prior to probe insertion in the
stack. Sampling at the required 12.3 liters per minute was initiated
immediately on probe insertion into the stack. Due to the small parti-
cle size measured in some preliminary runs, it was decided that the
particle size was so low as to not require isokinetic sampling and so
sampling was conducted at greater than isokinetic velocities. (Also,
the very low gas velocities in the stafk would have required an ex-
cessively large nozzle to obtain isokinetic sampling.) Sampling times
were 30 and 60 minutes. The impactor slides were covered with a disk
of 2-mil stainless steel shim stock so i:hat the entire glass slide did
not have to be weighed. Immediately after sampling, the impactor
slides were removed and returned to a constant temperature and humidity
room to equilibrate and for weighing.
Conditions at Which Measurements Were Made
Particulate emissions and smoke were measured at a range of
excess-air levels for two types of runs: cyclic runs with repeated cycles
of 10-min on and 20-min off cycles (the same cycle as used in the field
program) and steady-state runs at thermal equilibrium. In addition, CO.,
0«, CO, and HC were continuously monitored for all runs and particle-
size measurements were made during steady-state runs on Unit 37.
-------
Particulate emissions for the cyclic runs were collected during the
firing portion of six cycles, giving one hour of sampling during burner
operation. To assure obtaining samples during startup and shutdown
puffs, sampling was initiated about 30 seconds before burner startup
and was continued until about 30 seconds after burner shutdown. For the
steady-state runs, sampling was begun about 30 minutes after startup
and was continued for about two hours. Smoke readings were taken at the
1-, 2-, 3-, 5-, 7-, and 9.5-minute points during the 10-minute "on"
period for cyclic runs and at the 10-, 30-, 50-, 70-, and 100-minute
points for the steady-state runs.
The fuel fired during this investigation was identical to
(1 2)
the No. 2 reference fuel oil used during the field study ' .
COMMENTS ON FILTER MEDIA
The particulate emission measurements in this study were com-
plicated by the problems associated with weighing the very small quan-
tities of particulate accumulated on the filters during these experi-
ments with residential oil burners having relatively low particulate
emission levels. For example, for most of the runs, three hours of
sampling during cyclic operation (one hour of firing time) or two hours
sampling during steady-state operation resulted in collection of less
than 10 mg of particulate; for many runs, less than five mg of particulate
was collected. Moreover, no filter is completely satisfactory when attempt-
ing to measure these small quantities of materials as indicated by the
following:
Fiberglas filters (as required by EPA Method 5) are
highly hygroscopic, are somewhat fragile, and contain
significant impurities which make them less suitable
for detailed chemical analyses of particulate catch.
Quartz filters (compared to fiberglas) contain lesser
quantities of background elements but are even more
fragile and also are hygroscopic.
Silver filters are rugged and, essentially, not hygro-
scopic. However, the silver does react with sulfur
compounds to a greater extent than fiberglas.
-------
Fiberglas filters were chosen initially for use in conducting
the measurements for this study, primarily because of their being speci-
fied by the EPA Method 5 for sampling large sources. However, weighings
of fiberglas filters from runs on Units 36 and 38 sometimes produced
negative values for filter catch. These negative values were attributed
to one of two causes: (1) the hygroscopic nature of the filter, or (2)
the loss of small pieces of filter that stick to the glass filter holder
or break off when the filter is removed at the completion of the run.
At this point it was decided that the silver filters should be
used for runs on Unit 37, even though there was the possibility of re-
action with sulfur. Hence, silver filters were used as the filters for
the Unit 37 runs and as back-up filters for the particle-size runs on
the same unit.
RESULTS
Data Summary
Tables 1 and 2 give detailed data resulting from this investi-
gation. Table 1 lists operational conditions and emissions. Table 2
gives detailed smoke data. The data presented in these tables are plotted
in Figures 2 through 15, and their significance is discussed below.
Smoke Versus Time During Cycle
For the cyclic runs, smoke readings were taken at the 1-,
3-, 5-, 7-, and 9.5-minute points in the ]0-minute "on" period. These
data are plotted in Figures 2 and 3 for runs in Unit 36 and 37, respec-
tively. For Unit 36, the smoke readings during the first few minutes
of the cycle were usually high; that is, this unit had an appreciable
start-up transient. The CCL and CL data indicated a richer flame (com-
pared to steady state) immediately after startup of this unit, probably
related to nozzle characteristics. In contrast, the 1-minute smoke
-------
TABLE 1. SUMMARY OF EMISSIONS AND EMISSION FACTORS
Operational
Unit 36
Unit 37
Unit 38
Unit 36
Unit 37
Unit 38
Firing C02'
Cycle %
10/20(b) 9.7
11.0
12.1
12.3
12.3
12.9
13.3
13.6
10/20 11.3
11.7
11.8
11.8
12.0
12.3
12.9
10/20 12.5
steady-state 9.7
11.0
11.9
12.3
12.2
12.8
13.3
13.5
steady-state 11.3
11.8
11.9
11.8
12.0
12.3
12.9
steady-state 12.5
0.3
0.3
0.3
0.3
0.5
0.3
0.7
1.8
4.0
0.2
0.7
0.7
0.9
1.0
2.3
5.8
0.3
611
610
596
579
595
567
551
542
580
546
547
543
528
540
529
570
630
624
610
593
582
580
566
560
584
547
552
550
552
542
538
590
10.2
11.1
11.8
10.7
10.8
11. 1
17.9
37.4
25.9
24.5
23.0
25.3
28.8
26.9
65.7
17.5
8.0
8.0
10.0
12.0
12.0
10.0
11.0
11.0
18.0
27.0
20.0
20.0
22.0
27.0
52.0
11.0
HC
1.9
3.2
4.2
2.0
2.9
1.9
2.2
2.4
7.0
3.5
3.5
2.6
2.4
3.1
2.8
3.0
<1.0
0.85
0.86
0.95
1.12
1.70
1.13
3.27
4.09
0.46
0.66
0.58
1.39
0.97
1.89
2.48
0.90
0.97
0.64
0.47
0.84
0.25
0.35
0.83
1.27
0.47
1.06
0.5?
1.07
0.56
0.78
3.35
0.14
(a) Background levels were measured and found to Le less than 0.25 mg/snv*.
(b) Cycle of 10 minutes on and 20 minutes off.
(c) Data at 7 minutes.
(d) Negative value obtained, considered as zero when determining total filterable partlculate.
(e) Filter catch only.
(f) Filter catch plus probe wash (reported aa "filterable" partlculate In reports covering Phaaea I and TT»
-------
TABLE 2. SMOKE DATA
Unit 36
Unit 37
Unit 38
Unit 36
Unit 37
Unit 38
Operational
Firing Cycle C°2' *
10/20(a) 9.7
11.0
12.1
12.3
12.3
12.9
13.3
13.6
10/20 11.3
11 7
11.8
11.8
12.0
12.3
12.9
10/20 12.5
Steady state 9.7
11.0
11.9
12.3
12.2
12.8
13.3
13.5
Steady state 11.3
11.8
11.9
11.8
12.0
12.3
12.9
Steady state 12.5
Data
o2. %
7.5
5.8
4.2
3.9
3.8
3.3
2.6
2.3
5.8
5.0
4.6
4.5
4.4
4.0
3.2
3.8
7.6
5.9
4.3
4.0
3.9
3.4
2.7
2.4
5.8
5.0
4.6
4.5
4.4
3.9
3.3
3.7
Excess Air. TL
54
37
25
23
23
18
14
12
35
30
28
27
26
23
18
22
55
38
26
23
23
19
15
13
35
29
27
27
26
23
18
21
1 mm
0.3
0.7
2.9
1.9
4.3
7.9
8.1
8.3
0.2
0.8
0.6
1.0
1.0
1.9
5.2
0.3
3 mm
0.3
0.6
0.6
0.6
1.0
1.7
5.4
4.7
0.2
0.8
0.6
1.0
0.9
1.7
5.4
0.3
5 mm
0.3
0.3
0.3
0.5
0.7
1.0
3.9
4.8
0.2
0.8
0.5
0.9
0.8
1.6
5.4
0.3
7 mm
0.3
0.3
0.3
0.5
0.6
1.0
3.0
4.1
0.2
0.7
0.5
0.7
0.8
1.5
5.4
0.3
Smoke .
Bacharach
N umbe r
9.5 mm 10 nun 30 mm
0.3
0.3
0.3
0.5
0.6
0.3
3.0
4.0
0.2
0.8
0.5
0.7
0.7
1.5
--
0.3
0.3
0.3
0.3
0.5
0.3
0.7
1.8
4.0
0.2
0.7
0.7
0.9
1.0
2.3
5.8
0.3
0.3
0.3
0.3
0.5
0.3
0.6
2.1
3.9
0.2
0.8
0.7
0.8
1.0
2.2
5.8
0.3
50 mm
0.3
0.3
0.3
0.5
0.3
0.6
2.1
4.0
0.2
0.7
0.6
0.8
1.0
2.3
5.6
0.3
70 nun
0.3
0.5
0.3
--
0.3
0.7
?.l
4.1
0.2
0.7
0.7
0.9
1.0
2.3
5.4
0.3
100 mm
_.
0.3
0.3
--
0.3
--
?. 1
--
0.2
--
0.7
0.8
1.0
2.2
--
0.3
00
(a) Cycle of 10 minutes on and 20-mlnutes off.
-------
readings for Unit 37 were nearly as low as the steady-state values and,
hence, it appears that this unit stabilizes relatively quickly from a
cold start. The shorter start-up transient for Unit 37 was somewhat
unexpected in that it was the unit having the heavier refractory com-
bustion chamber and, thus, was expected to have slower temperature
response. The excellent start-up properties of Unit 37 are apparently
due to superior burner design and/or nozzle characteristics.
Smoke Versus Excess Air
Figures 4 and 5 show plots of Bacharach smoke number versus
air for runs with Units 36 and 37, respectively. Data plotted in these
figures are 9.5-minute smoke data for the cyclic runs and 100-minute
smoke data for the steady-state runs. By the 9.5-minute point in the
cyclic runs, the smoke levels were essentially equal to the steady-
state values.
Both units demonstrated low smoke at relatively low excess air
levels. Units 36 and 37 reached No. 1 smoke at about 18 and 26 percent
excess air (12.9 and 12.0 percent CO.), respectively. No field units
from the Phase II study exhibited smoke levels as low as No. 1 Bacharach
at such low excess-air levels. (Data at lower excess-air values were ob-
tained on Unit 37 during these runs than when this unit was investigated
during the Phase II study; this was accomplished by modifying the air
gate to reduce leakage in the closed position.)
For Unit 38, the flame-retention burner from Unit 37 firing
into the furnace, a 0.3 smoke number was obtained at 22 percent excess
air. This smoke reading agrees more with the observed smoke versus
excess air characteristics of the Unit 36 data than with that for Unit 37.
The flame-retention burner produced low start-up smoke at relatively low
excess air when fired in the furnace (0.3 smoke number at both 1.0 minute
and steady state), similar to its performance when fired into the boiler.
-------
10
Particulate Emissions Versus Excess Air
Figures 6 and 7 show plots of filterable participate emissions
(probe wash plus filter) versus excess air for runs on Units 36 and 37,
respectively. Figures 8 and 9 show similar plots but based on filter
catch alone, which is the portion of the total particulate catch that
should relate best to smoke data. Particulate emissions were collected
*
using the EPA Method 5, but only filterable particulate was dried and
weighed.
For Unit 36 (Figures 6 and 8), particulate emissions were
appreciably less for the steady-state runs than for the cyclic runs,
suggesting that cycling (probably startup) contributes significant
quantities of particulate to particulate measurements "integrated" over
the cycle. The ratio of cyclic particulate to steady-state particulate
was nearly four to one at low excess air; this ratio decreased as excess
air was increased so that, at 40 to 50 percent excess air, cyclic partic-
ulate emissions were less than twice the level of steady-state emissions.
The difference in particulate emissions between the cyclic and
steady-state runs is attributed to particulate generated during burner
startup and shutdown. Hence, for this unit, it appears that the startup
and shutdown "puffs" contribute between 30 and 75 percent of the partic-
ulate emissions measured while operating a burner on a 10-minute-on/
20-minute-off cycle. Because this unit also exhibited high smoke levels
during startup, it is concluded that high startup smoke may be an indica-
tion of high levels of particulate emissions during startup.
Both cyclic and steady-state particulate emissions increased
significantly as excess air was reduced below 20 percent.
For Unit 37 (Figures 7 and 9), the cyclic and steady-state runs
produced about the same particulate emission levels at given excess air
levels. These data, combined with the low smoke early in the cyclic runs,
* Silver filters were used for runs on Unit 37; whereas, EPA Method 5
specified fiberglas.
-------
11
suggest that start-up transients are minor for this unit. Again, both
cyclic and steady-state particulate emissions increased significantly as
excess air was reduced below about 20 to 30 percent.
The filter loadings and particulate emissions for runs on both
units are in the same range of values as the data from the Phase I and II
field studies(1'2).
_Part_iculate Loading. Versus Smoke
Figures 10 and 11 show the correlations between filterable partic-
ulate loading (probe wash plus tiltcr catch) and the 9.5-minute smoke
reading for cyclic runs or 10--.iinute smoko reading for steady-state runs
for Units 36 and 37, respectively. Figures 12 and 13 show similar data
for filter catch alone (without probe wash). For both units, there is
considerable scatter of data below No. 1 smoke, but a correlation between
particulate emissions and smoki- appears piactical above this smoke level
for a given unit operating on a given cycle.
Figures 10 and 11 show that, for both units, the rycli.c runs
gave a different correlation between filterable partirulatc emissions
and smoke than did tbc steady-stai.e runs. Figures 12 and 13 show that
the correlations between filter catch and smoke number tor the steady-
state run for Unit 36 and both runs for Unit 37 were similar. However,
the cyclic run for Unit 36 produced H qnitr: different correlation.
Figure 14 shows the correlation between particulate based on
filter catch only an.3 smoke- at the 1 ,0-irinute point for cyclic runs in
both units. The correlations between particulate emissions and smoke for
Units 36 and 37 arc inore nearly similar when the 1.0-minute smoke reading
is used as the data base. However, ther^ is considerable scatter in these
data for Unit 36.
Comparing the cyclic-run results shown in Figures 10 and 11, it
can be seen that Unit 37 emitted l(.v-:s parL jculate at a No. 5 smoke than
did Unit 36 at a No. 2 smoke. This illustrates the difficulty of con-
trolling air pollution emissions from residential heating by basing regu-
lations on smoke reading alone.
-------
12
Figure 15 shows participate emissions plotted against smoke
number (at 9.5 minutes) for cyclic runs and for steady-state runs for
all three units. The data are too few to draw certain conclusions about
a general relationship of particulate emissions and smoke. The scatter
in the data do not suggest that a strong correlation between particulate
emissions and smoke exists for the total of the cyclic data from these
two units. It appears that different start-up characteristics of the two
units are the primary factor preventing correlation of smoke number and
particulate emissions. This helps to explain why such correlations were
not possible with the field data from Phases I and II. However, within
limits of the available data, Figure 15 does show a fairly definite re-
lationship between steady-state particulate emissions and smoke.
Particle Size
Figures 16 through 21 show plots of paLticle-size distribution
for samples collected by a Battalle Cascade Impactor during six steady-
state runs on Unit 37 with excess air ranging from 18 to 35 percent.
Runs were made at smoke levels from 0.2 to 5.8 Bacbarach smoke numbers,
with duplicate samples collected during each run. All runs were for 60
minutes except Runs B in Figures 16 and 21 which were for 30 minutes. A
particle specific gravity of 2.0 was assumed for calculating the cutoff
particle size for each impactor sta^e.
These data indicate that 80- to 90-veight percent of the partic-
ulate was below 0.25 microns in size, even with the higher smoke levels.
Little difference in partLcle-size distribution Is evident for the various
runs.
An electron-microscope examination of selected particulate filters
was made to provide an alternative determination of particle size (an opti-
cal microscope cannot see below about 0.25 microns). Figures 22 through
29 show electron microscope photographs as follows:
-------
13
Figure 22 - 1000X view of a fiberglas filter
with collected participate (Unit 36
run at 14 percent excess air)
Figures 23-24 - 1000X and 5000X views of clean
(unused) silver filters
Figures 25-29 - ?OX, 100X, 50QX, 1000X, and 5000X
views of silver filters with
collected particulate (Unit 37
run at 23 percent excess air).
Although these electron-microscope photographs (particularly Figures 25
through 28) show the presence or some larye particles, Figures 22 and
29 suggest that a significant portion of i:he collected material is well
below one micron in particle: size. The possible presence of agglomerations
makes determination of the size of individual particler-. difficult. How-
ever, these photographs appear to confirm the facf. that most of the
particles emitted by the furnace and boiler were quite small, below one
micron, placing them withjn the respirable size range which is consid-
ered to be below about 3.5 ricrons ' ''.
CONCT I'S I0\'?
Although the conclusions of this invoscigacion must be con-
sidered in the context, of data limited to only two units, the following
conclusions can be made::
Filterable particalato emissions vary with excess
air in approximately the sam3 manner as smoke: readings;
that is, particular emi3Sions are telatively low at
high excens air anJ increase significanrl >r as excess
air is reduced. Thus, smoke readings arc indicative
of particulace enissiim trencs for a gi >;rn unit and
operating condition.
-------
14
o Correlations of particulate emissions and smoke appear
possible for given operating cycles for particular units
and possibly for different units at steady-state conditions,.
However, data for cyclic conditions of the two units ex-
amined do not suggest that a general relationship between
particulate emissions and smoke exists when considering
more than one unit, primarily due to differences in start-
up characteristics. This observation confirms the results
of the Phase I and II studies where no general relationship
between particulate emissions arid smoke was found.
o A greater difference between cyclic and steady-state partic-
ulate emissions was observed for the unit that had high
start-up smoke than for the unit that did not have high
start-up smoke. Hence, high start-up smoke appears to be
an indicator of nha startup contributing a disproportionately
large quantity of particulace emissions.
o Particle-size distribution did not change significantly for
a given unit aw excels air and smoke were varied.
o Particle-size measurements indicated that most particulate
emitted by these units was below one micron, and in the
respirable range. (The respirable range is not precisely
defined but is roughly below 3.5 microns.)
Additional Information Needed
Although the results do not show a firm correlation between
particulate emissions arid smoke when different cycles and equipment are
considered, the results show that there is a trend toward lower partic-
ulate emission with lower smoke numbera. Thus, if the desired partic-
ulate emission control level falls in the range of data observed for
low smoke number values, the smoke number might be used as a satis-
factory control. To do this, it: would be necessary to accumulate con-
v
siderably more data on the? relationship between smoke number and partic-
ulate emission for a large number of oil-fired units under different
operating conditions.
The following information that is presently not available
would contribute to a better understanding of this subject:
-------
15
Particulate data versus excess air and smoke for a
variety of burners and applications
Effect of nozzle firing-rate characteristic on starting
(including consideration of nozzle temperature)
Cycles other than 10-on and 20-off, especially shorfor
cycles
Effect of pump cut-off characteristics
Particular characterization versus particle size
Chemical composition: carbon, hydrogen, and
nitrogen contents aa.d polycyclic organic matter (POM).
Effect of filter material on particulate measurement
(roact-ion of silver filters with sulfur).
This latter work uould he- justified CMly if particulate emissions from
domestic oil-fired equipment are considered to be a significant contri-
bution in the overall particula-jt anatem^nt problem.
-------
16
REFERENCES
(1) Levy, A., Miller, S. E., Barrett, R. E., et al., "A Field Investiga-
tion of Emissions from Fuel Oil Combustion for Space Heating", API
Publication 4099, November 1, 1971, available from the API Publi-
cations Section, 1801 K Street, N.W., Washington, D. C. 20006.
(2) Barrett, R. E., Miller, S. E., and Locklin, D. W., "Field Investi-
gation of Emissions from Combustion Equipment for Space Heating".
This report is identified as API Publication 4180 (available from
API); PB-223148 (available from NTIS); and as EPA Publication
EPA-R2-73-084a, all June, 1973.
(3) "Standards of Performance for New Stationary Sources", Federal
Register, Vol. 36, No. 139, Part II, pp 24876-24895, December 23,
1971.
(4) Pilcher, J. M., Mitchell, R. I., and Thomas, R. E., "The Cascade
Impactor for Particle-Size Analysis of Aerosols", presented to the
Chemical Specialists Manufacturers Assoc., Inc., New York City,
December 6 and 7, 1955.
(5) Dunmore, J. H., Hamilton, R.J., and Smith, D.S.G., "Instrument for
the Sampling of Respirable Dust for Subsequent Gravimetric Assess-
ment", J. Scientific Instruments, 41. 669 (1964).
(6) Lippman, M., and Harris, W. B.."Size-Selective Sampling for Esti-
mating 'Respirable1 Dust Concentrations", Health Physics, 8, 155 (1962)
-------
17
Large particle-**
Impaction slide
"*-Small particle
First Stage:
Large jet
Low velocity
Large particles impact
Succeeding Stages:
Smaller jets
Higher velocities
Smaller particles impact
FIGURE 1. SCHEMATIC DIAGRAM SHOWING PRINCIPLE
OF THE CASCADE IMPACTOR
-------
18
CO
u
o
I
u
o
CD
25 % Excess air
23% Excess air
I
34567
Time From Start of Cycle, minutes
8
10
FIGURE 2. SMOKE VERSUS TIME DURING CYCLE -
UNIT 36
-------
en
u
o
I «
19
18% Excess air
26% Excess air
28 % Excess air
(cess airi
23% Excess air
1 a-
27 % Excess air
_L
, 35 % Excess oir -
T I T
30 % Excess air
±
JL
i
8
234567
Time From Start of Cycle, minutes
FIGURE 3. SMOKE VERSUS TIME DURING CYCLE - UNIT 37
10
-------
20
o
o
en
o
o
o
Operation
x - Cyclic (smoke at 9 5 mm)
o - Steady state (smoke at 100min)
Cyclic and steady state
Unit 38
Cyclic
Steady state
I
xo
I
1
10
20 30 40
Excess Air, percent
50
60
FIGURE 4. SMOKE VERSUS EXCESS AIR - UNIT 36
-------
21
Operation
x - Cyclic (smoke at 9 5 min)
o - Steady state (smoke at 100 min)
o
z
a>
o
o
o
o
u
-------
22
4.5
3.5
30
O
O
g
£ 25
2.0
6
UJ
a>
o
0.
I I*
0)
10
05
Steady state
I
I
Operation
x - Cyclic
o - Steady state
10
20 30 40
Excess Air, percent
50
60
FIGURE 6. FILTERABLE PARTICULATE EMISSIONS VERSUS EXCESS
AIR - UNIT 36
-------
23
3.5
30
25
O
O
g
£ 2.0
in
o
o>
B
o
I
il
15
10
0.5
10
Operation
x - Cyclic
o - Steady state
Cyclic and steady state
o o
x
°oX
1
1
20 30 40
Excess Air, percent
50
60
FIGURE 7. FILTERABLE PARTICULATE EMISSIONS VERSUS EXCESS
AIR - UNIT 37
-------
4.0
3.5
30
0)
o
on
025
O
O
O 2.0
jC
o
3
1.5
en
tf>
'E
UJ
1.0
0.5
Cyclic
Operation
x - Cyclic
o - Steady state
1
1
10
20 30 40
Excess Air, percent
50
60
FIGURE 8. FILTER CATCH VERSUS EXCESS AIR - UNIT 36
-------
25
4.0
3.5
Operation
x - Cyclic
o -Seady state
3.0
o
"55
o
~ 2-5
O
O
O
I 2-0
JC
o
o
O
0)
E
UJ
1.5
1.0
0.5
Cyclic \ \Steady state
o
I
1
I
I
10
50
20 30 40
Excess Air, percent
FIGURE 9. FILTER CATCH VERSUS EXCESS AIR - UNIT 37
60
-------
26
45
40
35
30
(A
>^
a*
I 25
o
9)
O
0)
r>
o
v
15
10
I
I
Operation
x - Cyclic
o - Steady state
I
234
Bacharach Smoke No (at 9 5 or 10 min)
FIGURE 10. FILTERABLE PARTICULATE LOADING VERSUS SMOKE
UNIT 36
-------
35
27
30
Operation
x - Cyclic
o - Steady state
' 20
o
o
5
3
O
o
a.
a
a3
15
10
12345
Bacharach Smoke No. (at 9.5 or 10 min)
FIGURE 11. FILTERABLE PARTICULATE LOADING VERSUS
SMOKE - UNIT 37
-------
28
40
35
30
25
20
15
10
Operation
x - Cyclic
o - Steady state
I
I
1
234
Bacharach Smoke No. (at 9 5 or 10 min)
FIGURE 12. PARTICULATE LOADING (BASED ON FILTER CATCH)
VERSUS SMOKE - UNIT 36
-------
29
30
25
10
-------
30
"E
en
40
35
o>
E 30
_
c
O
o
o
O
25
S 20
§ '5
O
10
Unit
x - Unit 36
o - Unit 37
I
I
I
I
34567
Bacharach Smoke No. (at 1.0 min)
6
FIGURE 14. PARTICULATE LOADING (BASED ON FILTER CATCH)
VERSUS SMOKE AT 1. 0 MINUTE FOR CYCLIC RUNS
-------
31
8
o>
o
jQ
O
45
35
30
25
15
10
Cyclic runs
n Unit 36
o Unit 37
A Unit 38
Steady-state runs
Unit 36
Unit 37
A Unit 38
I
I
234
Bacharach Smoke No (at 95 or 10 mm)
FIGURE 15. FILTERABLE PARTICULATE LOADING VERSUS SMOKE
CYCLIC OPERATION OF UNITS 36 AND 37
-------
32
99
OQ
yo
95
an
ftft
O\J
£ en
o» WJ
'a;
> KH
& DU
S* ^in
Q «»U
c -in
QJ OU
e
n o/^
Q- 20
V
_>
n iu
3 R
O 5
0.5
0.2
.1
f\ /NC
O.O5
0^\i
.01
X"
-x
).2 0.
x~
3 0
.
.
4
0
6
rfM
c
1.8
^
RunAx
Run BX
x*^
.
*
/
X
/
/
>
5 "
J
(
k
(
J
l(
D 2C
Equivalent Particle Diameter, microns
FIGURE 16. PAPTICLE-SIZE DISTRIBUTION FOR UNIT 37
AT 35 PERCENT EXCESS AIR
-------
33
99
98
95
90
80
70
I, 60
'o>
$ 50
o 40
30
Run B
Run A
20
10
<§ 5
0.5
0.2
O.I
0.05
0.01
0.2 0.3 0.4 0.6 0.8 I 2346
Equivalent Particle Diameter, microns
8 10
20
FIGURE 17. PARTICLE-SIZE DISTRIBUTION FOR UNIT 37
AT 29 PERCENT EXCESS AIR
-------
34
99
98
95
90
80
70
1, 60
o>
$ 50
.0 40
§30
£ 20
0)
>
| 10
1 5
2
1
0.5
02
O.I
005
O.OI(
v
fi^
X**"
^
^o-
-
* *
-.
«
^^t
<
t
>^
*
»i
^"
Run AX
^****'*
RunB^.
**'>
X
S
/
^
f
/
)2 0.3 0.4 06 0.8 2 3 4 6 8 10 2C
Equivalent Particle Diameter, microns
FIGURE 18. PARTICLE-SIZE DISTRIBUTION FOR UNIT 37
AT 27 PERCENT EXCESS AIR
-------
35
99
98
95
90
80
70
1. 60
'
>
5 10
I 5
2
1
05
02
O.I
005
n r»i
^
(
/
/
X^.
r»X
s
~~~
^
~* -
^-
^»J
>
^>
w
>^
X
^
X1
X
Run^x
x/
x^Run B
02 0.3 04
FIGURE 19.
0.6 08 I 2 34 6
Equivalent Particle Diameter, microns
8 10
20
PARTICLE-SIZE DISTRIBUTION FOR UNIT 37
AT 26 PERCENT EXCESS AIR
-------
36
99
98
95
90
80
70
I. 60
'o>
$ 50
£ 40
§30
20
0)
>
| 10
o
05
0.2
O.I
0.05
0.01
0.2 0.3 0.4
0.6 0.8 I 2346
Equivalent Particle Diameter, microns
8 10
20
FIGURE 20. PARTICLE-SIZE DISTRIBUTION FOR UNIT 37
AT 23 PERCENT EXCESS AIR
-------
37
99
98
95
90
80
70
I. 60
'55
$ 50
n 40
£ 30
u
0- 20
| 10
J 5
2
I
05
0.2
O.I
005
/
0.01
0.2 0.3 0.4
0.6 0.8 I 2346
Equivalent Particle Diameter, microns
8 10
20
FIGURE 21. PARTICLE-SIZE DISTRIBUTION FOR UNIT 37
AT 18 PERCENT EXCESS AIR
-------
1000X
Scale
'4-
1 Micron
FIGURE 22. PARTICULATE ON FIBERGLASS FILTER
1000X
Scale
-4-
1 Micron
FIGURE 23. CLEAN SILVER FILTER
-------
39
5000X
Scale: ~*j r« 1 Micron
FIGURE 24. CLEAN SILVER FILTER
2 OX
FIGURE 25. PARTICULATE ON SILVER FILTER
-------
100X
FIGURE 26. PARTICULATE ON SILVER FILTER
500X
Scale
1 Micron
FIGURE 27. PARTICULATE ON SILVER FILTER
-------
41
1000X
Scale:
1 Micron
FIGURE 28. PARTICULATE ON SILVER FILTER
5000X
* 1 Micron
FIGURE 29. PARTICULATE ON SILVER FILTER
-------
42
TECHNICAL REPORT DATA
(I'lcaw read liiwuc Horn
------- |