»EPA
United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park NC 27711
EMB Report81-CUS-16
June 1982
Air
NSPS Revision
Nonferrous Smelter
Reverberatory
Furnace
Emission
Test Report
Pheips Dodge
Morenci, Arizona
-------
CORPORATION
DCN 82-222-018-06-19
RCN 222-018-06
FINAL REPORT
PRIMARY COPPER SMELTER
NSPS REVISION
EMISSIONS TESTING AT
THE PHELPS DODGE SMELTER
MORENCI, ARIZONA
Prepared for:
Frank R. Clay
U.S. Environmental Protection Agency
Emissions Meaurement Branch
ESED OAQPS MD-13
Research Triangle Park, NC 27711
Contract No. 68-02-3542
Work Assignment No. 6
ESED 80/13
Prepared by:
R.V. Collins
M.J. Krall
L.O. Edwards
Radian Corporation
September 13, 1982
8501 Mo-Pac Blvd. / P.O. Box 9948 / Austin, Texas 78766 / (512)454-4797
-------
coopoaortON
TABLE OF CONTENTS
Section Page
1.0 INTRODUCTION 1
2.0 SUMMARY . . 2
3.0 PLANT DESCRIPTION 4
3.1 Process Description 4
3.2 Sampling Port Description 7
4.0 SAMPLING METHODOLOGY 8
4.1 Sampling Locations 8
4.1.1 Matte Tapping Emissions 8
4.1.2 Slag Skimming Emissions 10
4.2 Sampling 10
4.3 Visual Emissions Observatinos 14
4.3.1 Observation Sites - Morenci Smelter 15
4.3.2 Methodology 15
4.3.3 Method 9 15
4.3.4 Method 22 18
4.3.5 Limitations of the Methods 19
4.4 Process Observations . . 20
5.0 ANALYTICAL METHODOLOGY 22
6.0 QUALITY ASSURANCE 24
6.1 Source Sampling Audit Results 24
6.1.1 Systems Audit 25
6.1.2 Performance Audit 26
7.0 CHAIN-OF-CUSTODY 33
-------
RADIAN
TABLE OF CONTENTS (Continued)
Section Page
8.0 RESULTS 34
APPENDIX A: PROCESS OBSERVATIONS 43
APPENDIX B: SYSTEMS AUDIT CHECKLIST AND CALIBRATION DATA 53
APPENDIX C: SOURCE SAMPLING FIELD DATA AND RESULTS SUMMARY .... 71
APPENDIX D: VISIBLE EMISSION OBSERVATION FORMS 99
ii
-------
LIST OF FIGURES
Number Page
3-1 Typical Primary Copper Smelter Flowsheet 5
4-1 Sketches Showing the Locations of the Matte Tapping
Emission Sampling Ports at the Phelps-Dodge Morenci
Site 9
4-2 Sketches Showing the Locations of the Slag Skimming
Emission Sampling Port at the Phelps-Dodge Morenci
Site 11
4-3 A Schematic of the Combined EPA Method 5 and 6
Sampling Apparatus 13
4-4 Visible Emissions Observer Locations for the No. 5
Reverberatory Furnace at Morenci 16
4-5 Visible Emissions Observer Locations for the No. 3
Reverberatory Furnace at Morenci 17
111
-------
CORPORATION
LIST OF TABLES
NUMBER
2-1
6-1
6-2
6-3
6-4
8-1
8-2
8-3
8-4
8-5
8-6
8-7
Summary of Emission Rates Calculated from Particulate and
Sulfur Dioxide Testing at Morenci
Attainable Accuracy and Precision of Test Results Based
on the EPA Collaborative Tests of Stationary Source
Summary of Data Reduction Check
Summary of EPA Audit Sample Analyses
Summary of EPA Method 5/6 Impinger Collection
Slag Skimming Data Summary Morenci No. 3 Reverb Furnace . .
Sampling Sequencing Phelps-Dodge Morenci, Arizona Copper
Smelter Matte Tapping - No. 5 Reverb
Sampling Sequencing Phelps-Dodge Morenci, Arizona Copper
Reverberatory Furnace No. 5 - Method 9 Observations ....
Reverberatory Furnace No. 5 - Method 22 Observations . . .
Reverberatory Furnace No. 3 - Method 9 Observations ....
Page
3
27
30
32
32
35
36
38
39
40
40
41
-------
1.0 INTRODUCTION
This report presents the results of a field testing effort
conducted at the Phelps Dodge smelter at Morenci, Arizona. This effort,
conducted under U. S. EPA Contract 68-02-3542 Work Assignment No. 6, was
part of a series of efforts designed to provide background data for a
portion of the revision of the New Source Performance Standards for the
primary copper industry. During the study, the emissions from the fugitive
gas collection hooding serving the reverberatory furnace matte tapping and
slag skimming operations were measured.
The particulate and sulfur dioxide emission rates were determined
using the combined EPA Reference Methods 5 and 6. Visible emissions (used
as a surrogate for particulate and sulfur dioxide) were monitored using the
techniques of EPA Reference Methods 9 and 22 (proposed) to determine the
capture efficiency of the fugitive gas collection systems.
The emission rates from these tests are used in conjunction with
observations of matte and slag production to calculate emission factors for
the mass of particulate and sulfur dioxide emitted per mass of matte or slag
produced.
Radian personnel performed all particulate, sulfur dioxide, and
visible emissions testing. Research Triangle Institute (RTI) personnel made
process observations during testing.
The remaining sections of this report present a summary of the
results, a process desription, sampling and analytical methodologies,
quality assurance documentation, and results.
-------
2.0 SUMMARY OF RESULTS
The particulate and sulfur dioxide testing performed at the Phelps
Dodge Morenci smelter yielded average results of 17 pounds of particulate
and 300 pounds of sulfur dioxide per hour of matte tapping and 1.9 pounds of
particulate and 17 pounds of sulfur dioxide per hour of slag skimming.
These tests were performed intermittently due to the nature of the
two processes evaluated. Only reverbratory fugitive emissions were
measured. Fugitive emissions from matte tapping were obtained at the No. 5
furnace and fugitive emissions from slag skimming were taken at the No. 3
furnace. Thus, the results are more meaningful if expressed in terms of
mass of pollutant per unit of production.
The average pounds per hour pollutant emission rates were con-
verted to the following average pound per unit of production emission
factors: 0.076 pounds of particulate and 1.3 pounds of sulfur dioxide per
ton of matte production and 0.024 pounds of particulate and 0.21 pounds of
sulfur dioxide per ton of slag production. These emission rates and factors
are presented in Table 2-1. The matte and slag production rates used for
these emission factors are estimated from observation of the matte tapping
and slag skimming operation made by RTI during the emissions testing. This
information "Primary Copper. Smelter NSPS Revision Emission Testing, Process
Observations, December 17, 1981" prepared by RTI is included as Appendix A.
Additional information, including copies of the operator logs, have been
requested from Phelps Dodge by RTI. Therefore, it should be realized that
the process observations, in particular the production rates, may be subject
to small changes; emission factors in this report are based upon the data
given in Table 2-1.
-------
I
TABLE 2-1. SUMMARY OF EMISSION RATES CALCULATED FROM PARTICULATE TESTING AT MORENCI
Estimated
Production
Test
Matte Tapping
(Reverb No. 5)
EMB-004 MMT
EMB-006 MMT
EMB-008 MMT
Average
Slag Skimming
EMB-003 MSS
EMB-005 MSS
EMB-007 MSS
Average
Tons
185
250
275
80
90
60
Taps
8
10
11
2
3
2
Particulate
Lb/Hr*
19
18
15
17
2.0
2.5
1.2
1.9
Lb/Ton**
0.1
0.072
0.054
0.076
0.025
0.028
0.020
0.024
Lb/Hr*
290
290
310
300
15
30
7.6
17
Sulfur
Dioxide
Lb/Ton**
1.6
1.2
1.1
1.3
0.19
0.33
0.13
0.21
ppm
683
491
516
557
188
367
100
215
DSCFM
42600
59200
60200
54000
8000
8200
7600
7933
*Lb. of pollutant/hr. of sampling.
**Lb. of pollutant/ton of matte or slag produced during sampling.
-------
3.0 PLANT DESCRIPTION
3.1 Process Description
A block flow diagram of a typical primary copper smelter is shown
in Figure 3-1. The Phelps-Dodge smelter at Morenci has two reverberatory
furnaces in operation, No. 3 and No. 5. Both furnaces are charged with
copper ore concentrate and are fired with natural gas and/or fuel oil.
Fluxing agents are also added to the furnace with the concentrate. In the
furnaces, the levels of sulfur and iron in the concentrate are reduced to
the point where iron sulfide (FeS) and copper sulfide (Cu^S) are present
in approximately equal amounts. This mixture is called matte. Excess FeS
is converted via an exothermic reaction to iron oxide (FeO) and sulfur
dioxide (S02). The FeO then combines with the silica-based fluxing agents
and forms slag. This slag floats on top of the matte which facilitates
their separation. The discharges from the furnace are the furnace off
gases, the slag, and the matte.
The off gases containing S02, particulate, metal fumes, and
combustion products leave the furnace and enter waste heat boilers and then
a dust collection system. The recovered dust is combined with the reverb-
eratory furnace feed and recycled to the furnace.
Removal of the slag layer from the furnace for disposal at a slag
dump is by a process referred to as "slag skimming". Molten slag exits the
furnace through a skimming port and is conveyed to a ladle by way of a
trough or "slag launder". The port, slag launder, and transfer point from
the slag launder to the ladle are all hooded. These hoods are designed to
collect fugitive gases from the slag skimming operation and convey them
through a duct to a blower and then to a discharge stack.
The matte is removed from the furnace by a process called "matte
tapping". The molten matte exits the furnace through a tapping port and
-------
ENTERING THE SYSTEM
LEAVING THE SYSTEM
Raw Charging
Calcines
Flux
Fuel
Air
Smelting
Furnace
Reverberatory
Matte
Siliceous Flux
Air
Miscellaneous Material
High in Copper
Reducing Gas/Fuel
Air
Slag
Converter
Blister
Copper
ippe
Fire
Refining
Furnace
Anode
Copper
ipper
Casting Wheel
Anodes
Gases and Oust to Waste
*• Heat Boilers, Control
Equipment and Stack
*• Slag to Dump
> Gases to Control Equipment
and Stack
-*» Gases to Stack
-*• Slag to Converter
-»»Exit Plant
Figure 3-1. Typical Primary Copper Smelter Flowsheet
-------
CORPORATMM
conveyed to a ladle by way of a trough or "matte launder". The port, matte
launder and point of transfer from the launder to the ladle are all hooded.
As in the slag skimming process, these hoods are designed to collect fugi-
tive gases from the matte tapping operation and convey them through a duct
to a blower and then to a discharge stack. The ladle of molten matte is
transferred by way of overhead cranes to converters for further refining.
In the converters, FeS is oxidized to FeO and SC^. Quartz is
added to the converters and binds the FeO to form a slag containing approxi-
mately four percent copper. This slag is periodically recycled to the
reverberatory furnace via ladles and an overhead crane. "White metal"
consisting of Cu2S remains after the removal of the iron. At this stage,
the converter blowing cycle stars. The Cu2$ is converted to a 98 percent
pure copper called "blister copper." Cold copper scrap is periodically
added during he blowing step to absorb the heat produced by the exothermic
reaction.
The blister copper is transferred to an anode furnace in batches,
again by ladles and cranes. Air blowing into the anode furnace removes the
remaining sulfide and results in copper containing Cu£0 as an impurity.
The Cu20 is then reduced with reformed natural gas and produces a 99+
percent pure copper which is ready for the anode pouring step and casting.
Casting 99+ percent pure copper into the anodes is the final production
operation that takes place at most smelters. The copper anodes, which
weight about 750 Ibs each, are then shipped by rail car to another facility
where they are refined one last time by means of electrolysis. This step
removes trace amounts of noble and precious metals, leaving pure copper as
the final product.
-------
3.2 Sampling Port Description
Emission rates for particultate and sulfur dioxide from the fugi-
tive gas collection hooding for matte tapping and slag skimming operations
were determined during this effort. During the pretest site survey for the
Phelps Dodge smelter at Morenci, and as a result of subsequent discussions
with EPA/EMB and RTI, two sampling locations were selected. One was for the
emissions from the slag skimming operations, while the other was for the
emissions from the matte tapping operations.
The first location is in a duct which carries emissions captured
by hooding at two of the three slag skimming locations on the No. 3 reverb-
eratory furnace. The second location is in the duct which carries emissions
from all tapping and skimming locations on the No. 5 reverberatory furnace.
Only emissions from the matte tapping operations at the No. 5 reverberatory
were sampled, thus requiring coordination to assure sampling was performed
when matte tapping only was in progress. The configuration of the duct work
and these ports are further described in Section 4.0.
-------
4.0 SAMPLING METHODOLOGY
The emission rates for participate and sulfur dioxide from the
fugitive gas collection hooding serving the matte tapping operations for the
No. 5 reverberatory and the slag skimming operations for the No. 3 reverb
were determined during this effort. In addition, the visible emissions
escaping these hooding systems were monitored. This section describes the
sampling locations and the methodologies utilized to collect these samples
and data.
4.1 Sampling Locations
4.1.1 Matte Tapping Emissions
The No. 5 reverberatory has eight tap bays and two skim bays. The
gases from these bays were conveyed to one common duct, through two fans in
series, and on to the main converter stack. The sampling ports were
downstream of the fan and upstream of the junction with the other gases
which are also conveyed to the main duct. The ports were was located in a
long horizontal run of the main fugitive gas duct as it crossed the top of
the reverberatory/converter building. These ports were two duct diameters
upstream of the 90° elbow, which turns the duct down the side of the
building. The duct diameter at the ports was 53".
The ports were installed specifically for these tests and were
accessible from an existing catwalk. The catwalk was enlarged for these
tests. One port axis was horizontal and the other was vertical. An "A"
frame was also constructed above the duct to support the sampling train.
The ports were 3" pipe nipples approximately 6" in length. Electrical power
was available at the sampling ports. Figure 4-1 is a sketch showing the
matte tapping emission sampling port locations. This test plane meets the
eight and two diameter criteria as described in EPA Method 1.
-------
FLOW
ROOF
3" PORTS
CATWALK
ROOF
Side view of No. 5 main fugitive duct.
Note: The "A" frame required above the
vertical nort is not shown.
2D
FLOW
7
,3" PORTS
Top view of No. 5 main fugitive duct.
Figure 4-1. Sketches Showing the Locations of the
Matte Tapping Emission Sampling Ports
at the Phelps Dodge Morenci Site
-------
Due to the availability of a more favorable site, approximately 225 feet
downstream of the first, matte tapping emission sampling operations were
moved to the second site after the first run.
These ports had been installed after the pretest site survey by
Phelps Dodge for their own use. The duct diameter and conditions were the
same for both locations. However, the gas velocity measured at the second
site was greater than for the first site. Adjustments to some of the hood
dampers overnight could account for this increased gas flowrate.
4.1.2 Slag Skimming Emissions
Unlike the No. 5 reverb, the slag skimming bay ventilation system
was isolated sufficiently to permit sampling without interference from
simultaneous matte tapping. The two 3" sampling ports were in an 18" dia-
meter duct which served two of the three skim bays for the No. 3 reverb.
This test plane met the eight and two diameter criteria with more than ten
duct diameters downstream from any disturbance, and approximately five
diameters upstream of any disturbance. The duct was horizontal and just
below the reverberatory gas offtake. Figure 4-2 is a sketch showing the
slag skimming emissions sampling port locations.
4.2 Sampling
Radian determined the emission rates by EPA Reference Methods 1,
2, 3, 5, and 6 (Code of Federal Regulations, 40, Protection of the Environ-
ment, Parts 53 to 80, Revised as of July 1, 1980). Radian performed EPA
Reference Methods 1 and 2 the first day at the plant at both sampling sites.
When the decision was made to move the matte tap testing to a more
accessible sampling location, Methods 1 and 2 were again performed at that
site.
10
-------
c\V
->10C
FLOW-
HORIZONTAL
T^
\ °;
D i
3" PORTS x
V X
1 " U
View toward No. 3 reverb from between No. 3
reverb and No. 3 waste heat boiler beneath
reverb offtake.
U
" PORTS
O
VERTICAL
View from floor between No. 3 reverb and
No. 3 waste heat boiler beneath reverb
offtake.
Figure 4-2. Sketches Showing the Locations of the
Slag Skimming Emission Sampling Port
at the Phelps Dodge Morenci Site
11
-------
During each test, Radian also performed EPA Reference Method 3
using a calibrated Fyrite®.
In the determination of particulate and sulfur dioxide rates, EPA
Reference Methods 5 and 6 were combined within a single sampling test. A
very prominent portion of this testing was the sequencing and communication
involved to coordinate the sampling with the plant operations. Walkie talk-
ies, provided by the U.S. EPA, were the main source of communication. RTI
officials observed when the slag was being skimmed and the matte being
tapped. This information was relayed to the sampling team and the sampling
commenced. If the plant operations ceased, then word was sent to terminate
the sampling. This aspect of the testing went smoothly. Ths U.S. EPA
requested that Radian sample each traverse for ?. 1/2 minutes per point and
sample each of the six points twice. The reasoning was that since each tap
had a duration of only a few minutes, the sampling would be biased if only
one point was sampled per tap (5-minute sampling point). With 2 1/2 minute
sampling points, at least two points across the duct would be sampled per
tap.
A schematic drawing of the sampling train and related equipment
is shown in Figure 4-3. Gas volume was measured with a dry gas meter
calibrated against a standard dry gas meter. Stack temperature was measured
by Type K, chrome-alumel, thermocouple calibrated against a mercury
thermometer. Velocity pressure and pressure drop across the orifice were
measured using Magnehelic® gauges having ranges from 0-1 and 0-3 inches
H20, respectively.
A leak check of the entire train was conducted before and after
each run and the leak rate noted. The leak check was performed at either
15" Hg vacuum or the highest vacuum obtained during the run. Pitot lines
were leak checked at 3" H^O before and after each run to insure proper
velocity pressure measurements. The barometric pressure was read daily from
an aneroid barometer. The measured static pressure of the stack gas was
added to this atmospheric pressure to obtain the actual stack pressure.
12
-------
TEMPERATURE
SENSOR
ISOKINETIC
NOZZLE
STYPE
PITOT TUBE
DRY
IMPINGER
FILTER
HOLDER
ICE BATH
SILICA GEL
DESICCANT
VACUUM
LINE
ORIFICE
ORIFICE
GAUGE
TEMPERATURE
SENSORS
\]
BYPASS
VALVE
VACUUM
GAUGE
MAIN
VALVE
702091 I
PUMP
Figure 4-3. A Schematic of the Combined EPA Method 5 and 6 Sampling Apparatus
-------
Dry gas meter volume was noted at the beginning of sampling for
each point. The velocity pressure (AP) was read and the sampling rate (AH)
adjusted by a predetermined factor to obtain an isokinetic sample. Stack
temperature, hot box temperature, gas temperature at the exit of the final
impinger, dry gas meter inlet and outlet temperature, and vacuum on the
train were all recorded at each sample point. Data forms containing the
above information for each run are included in Appendix C.
Both the probe liner and nozzle were brushed and rinsed with
acetone to collect particulate while at the sampling location. The hot
box/cold box assembly was taken to the mobile lab for the remaining portion
of the sample recovery. All glassware upstream of the filter and the front
half of the glass filter holder were rinsed with acetone into the probe wash
sample container. The filter was removed and placed in its original petri
dish. Care was taken to recover all fragments of the filter. Impingers were
weighed to obtain final weights to allow calculation of the moisture
fraction, and the h^Og impingers were quantitatively transferred to
polyethylene containers, marked, and sealed for shipment.
4.3 Visual Emissions Observations
Visual emission observations were made to visibly evaluate various
hood capture systems and pollution control equipment at the Phelps Dodge
Company copper smelting facilities in Morenci, Arizona.
All visible testing was performed by certified observer Mr. Craig
Beskid, Engineer, of Radian Corporation (Texas Air Control Board certifi-
cation date 9-24-81). The testing was performed using visible emissions
observations of particulate matter escaping capture of the hooding system or
pollution control equipment as a surrogate for sulfur dioxide (S02)
emissions that were not captured. Results from this visible emissions
testing, along with simultaneous SC^ and particulate measurements, will
provide input for revision of the New Source Performance Standards (NSPS)
for the primary copper industry.
14
-------
4.3.1 Observation Sites - Morenci Smelter
Two streams from the Morenci smelter were visually evaluated:
• The No. 3 reverberatory furnace slag skimming fugitive
emissions escaping capture by the hooding system, and
• The No. 5 reverberatory furnace matte tapping fugitive
emissions escaping capture by the hooding system.
Figures 4-4 and 4-5 show the position of the observer for each launder and
reverberatory furnaces No. 3 and No. 5, respectively.'
4.3.2 Methodology
Two visible emissions methods were employed in the testing. The
technique of EPA Method 9, "Visual Determination of the Opacity of Emissions
From Stationary Sources" was used to determine the opacity of fugitive hood
emissions and scrubber stack emissons. The technique of proposed EPA Method
22, "Visual Determination of Fugitive Emissions from Materal Processing
Sources" was used to determine the accumulated time that fugitive emissions
were observed escaping each hooding system evaluated.
All in-plant observations were made from a position approximately
15 feet directly in front of each launder.
4.3.3 Method 9
A certified observer is generally used by control agencies to
evaluate the opacity of an emission source. The observers are instructed at
opacity training schools. In order to become certified, observers must
evaluate plume opacity with _+ 7.5 percent accuracies relative to transmis-
someter measurement of plume opacity. Upon passing the course, they are
certified by the school for six months as capable of evaluating plume
opacity by visual inspection.
15
-------
CONVERTER AISLE
SLAG
RETURN
LAUNDERS
LAUNDERS
FLOOR HOLE
TO LADLES
REVERBERATORY
FURNACE •
NO. 5
10
Identifies observer
location points
Figure 4-4. Visible Emissions Observer Locations for the No. 5
Reverberatory Furnace at Morenci Matte Tapping
Operations.
16
-------
CONVERTER AISLE
SLAG
RETURN
' LAUNDERS
Launders
REVERBERATORY
FURNACE
NO. 3
FLOOR HOLE
TO LADLES
Observer
Locations
Figure 4-5. Visible Emissions Observer Locations for the No. 3 Reverberatory
Furnace at Morenci for Slag Skimming at Taps 1, 2, and 3.
17
-------
COOPORJmOM
When observing a plume, Method 9 requires that the observer stand:
• at a distance from the plume sufficient to provide a clear
view of the emissions,
• with his line of vision approximately perpendicular to the
plume direction, and
• with the sun oriented in the quadrant to his back.
The method also requires that readings be made at 15-second inter-
vals over a minimum six-minute period at the point of greatest opacity in
the plume. The average of these minimum 24 readings is reported as the
average visual opacity. Data forms with these recorded readings are
contained in Appendix D.
For the purpose of this study, modifications to Method 9 were
necessary. First, Method 9 was performed indoors at the Morenci smelter
reverberatory furnaces. This was an improper position relative to the
emissions, the light source, and the background. The emissions were read
most often with light from above the emissions as the emissions escaped the
hooding system. All opacity observations were made consistently using the
furnace facing approximately two feet above the launders as background.
Most observations were performed during a simultaneous stack test. Also,
all observations were halted during excessive visual interferences caused by
fugitives from other sources or moving equipment and personnel.
4.3.4 Method 22
This method is used to determine the amount of time that any fugi-
tive visible emissions occur during the observation period. Fugitive emis-
sions include emissions that:
18
-------
CORMHUmOt
• escape capture by process equipment exhaust hoods,
t are emitted during material transfer, building housing
material processing or handling equipment, and
f are emitted directly from the process equipment.
Since this procedure does not require certification according to
Method 9, a trained opacity observer is not required. However, it is neces-
sary that the observer be educated in the general procedures for determining
the opacity of emissions. Mr. Craig Beskid of Radian Corporation performed
all the Method 22 testing. The modifications to Method 22 are identical to
the modifications previously listed for the Method 9 observations.
4.3.5 Limitations of the Methods
Both Methods 9 and ?.?. were modified for conditions at the observa-
tion sites. The modifications to each method were made to minimize method
limitations and to include process dependent effects on emissions, light
deficiencies, and observer positions. The extent of these modifications
were determined during the observations at each smelter process point. The
modifications and limitations of the methods, including their effect on the
observations, are discussed below.
4.3.5.1 Method 9 Limitations
Visible emissions best evaluated by this method are from stack
plumes, while observations at the Morenci smelter were performed on fugitive
emissions. The procedures for Method 9 specify outdoor observations. The
majority of observations at the Phelps Dodge smelter were performed indoors.
Indoor observations decrease observed opacities due to a decrease in the
amount of light and background contrast.
19
-------
Method 9 procedures implicitly include provisions for a cyclic
process such as a varying process load. However, characterization of fugi-
tive emission opacity for an intermittent process such as slag skimming or
matte tapping is not addressed.
Other limitations to obtaining an accurate opacity assessment of
the fugitive emissions is listed below.
• Inability to obtain proper observer position relative to the
light source.
• Inconsistent amount of light.
• Visible emissions interferences due to hood leaks and nearby
processes (increases observed opacity).
4.3.5.2 Method 22 Limitations
The most significant modifications and limitations to the Method
22 observations are listed below.
t Visible emissions interference due to hood leaks and nearby
processes (increased recorded emission time).
• Less than adequate amount of light (<100 lux, decreases visi-
bility of emissions).
• Inability to attain proper observer position relative to
light source (may increase or decrease visibility of emis-
sions, highly dependent on position of light source).
4.4 Process Observations
The matte tapping from the No. 5 reverberatory and slag skimming
from the No. 3 reverberatory were observed from the furnace floor during
sampling. The observers also coordinated the sampling and communicated with
the two-way radio.
20
-------
At Morenci, the No. 5 reverberatory tap and skim bays were served
by a single hooding system. Samples were collected from a location common
to other types of bays. Thus, it was necessary to sample only during the
tapping operation and stop sampling if a skim or lancing in preparation of a
skim began. The observer noted the times for the start of any lancing
operations, the initation of metal flow (at which time sampling could
begin), and the completion of the top or the start of lancing for another
tap or skim (at which time sampling was stopped). The observer also noted
the extent that each ladle was filled and any other information which was
thought to be pertinent. Simultaneous taps occurred several times during
the sampling. For these, sampling was stopped only during lancing, but was
resumed as soon as metal flow at the second bay began. Sampling continued
if just one of the taps were completed.
The observer's task at the No. 3 reverberatory skim bay was
essentially the same as above. However, the duct work in which the sampling
ports were installed served only two of the three skim bays for the furnace
and none of the tap bays. The observer, as for the No. 5 reverberatory,
recorded the times of lancing and metal flow and skim completion, as well as
the time of any interfering operations. The observer also noted the extent
to which each ladle was filled and any other information thought to be
pertinent.
The process observation for Morenci was performed primarily by RTI
with support from Radian. Coordination of the sampling activities was
performed by Radian. The process observations document provided by RTI is
Appendix A in this report. Operating logs from the Morenci have been
requested. This information will be used to corroborate the data used, and,
where possible, confirm some assumptions which were made for matte and slag
production notes and schedules.
21
-------
5.0 ANALYTICAL METHODOLOGY
This section deals with the handling, preparation and analysis of
all the recovered samples. EPA Reference Methods 3, 5, and 6 were per-
formed using the same reference stated in Section 4.2.
In determining the dry molecular weight of the flue gas, the
EPA's Reference Method 3 was used. Radian used a Fyrite during each test.
During mid-run, the flue gas was hand pumped through a pitot line into a
sealed container holding either C02 or 03 absorbing solution. The gas
was thoroughly mixed with the solution and the C0j> or 02 percentile was
read off the scale on the container. This procedure was done at least
twice. Once both C02 and Q£ percentages were established, the remainder
of the gas was assumed to be composed of nitrogen.
Upon return to Radian's Austin laboratories, final weights were
obtained for all filters and the acetone/deionized water wash residues in
compliance with EPA Reference Method 5. Filters were desiccated for 24
hours and weighed to a constant weight. Initial and final weighing was
performed on a Mettler semi-micro balance. The acetone/deionized water wash
was quantitatively transferred to equilibrated tared glass beakers and taken
to dryness. These samples were then desiccated 24 hours and weighed to a
constant weight. Acetone and deionized water blanks, 100 ml, were treated
in this same manner, and the samples corrected for these blanks.
Using EPA Reference Method 6, the h^O;? samples containing the
S02 were diluted up to 500, 1000, or 2000 mL in volumetric flasks depend-
ing on the SO? concentration. A 5 mL aliquot was pipetted into a 250 mL
Erlenmeyer flask and 20 mL of 100% isopropanol added. Two to four drops of
thorin indicator were added and the sample titrated to a pinkish-orange end-
point with .01 N 83(0104)2. The equation in Method 6 for calculation of
SO? concentration was followed. The 63(0104)2 was standardized daily
22
-------
against 0.01 N Hj>S04 prepared from a purchased analytical concentrate.
Blanks and QA audit samples were titrated as was a duplicate of every tenth
sample.
23
-------
CORPORATION
6.0 QUALITY ASSURANCE
The work performed at the Phelps Dodge Morenci smelter
incorporated a comprehensive quality assurance/quality control (QA/QC)
program as an integral part of the overall sampling and analytical effort.
The major objective of the QA/QC program was to provide data of known
quality with respect to:
• completeness,
• accuracy,
• precision,
• representativeness, and
• comparability.
The quality assurance function was organized to provide inde-
pendent review and assessment of project activities and their ability to
achieve the stated data quality objectives. The QA coordinator for the
project had the responsibility of evaluating the adequacy and effectiveness
of the QC system and providing assurance that it was, in fact, responsive to
the specific needs of the program.
While the system of QA activities was necessarily independent of
the technical effort per se, the QC system was an integral part of the daily
technical effort. It was designed to provide an overall system for
generating data of a specified quality. This section provides an assessment
of the QC program and a summary of resulting data quality as determined by
the QA audit.
6.1 Source Sampling Audit Results
As part of the quality assurance program for this project a
performance and systems audit was performed during the period 13 November to
14 November 1981. Audit activities, results and conclusions are presented
below.
24
-------
CORfHMUmOM
6.1.1 Systems Audit
A systems audit is an on-site qualitative review of various
aspects of a total sampling and/or analytical system to assess its overall
effectiveness. The systems audit results represent a subjective evaluation
of a set of interactive systems with respect to strengths, weaknesses and
potential problem areas. The audit was designed to evaluate the following:
t Adherence to accepted procedures in performing
reference method source sampling,
• Adequacy of internal quality control procedures
• Equipment and facilities,
• Qualification and training of personnel,
• Calibration procedures and documentation,
• Sample handling, custody and storage, and
• Data recording, review and handling.
The systems audit checklists, which are presented in Appendix B,
delineate the specific aspects of the sampling/analytical system which are
deemed to be especially important in obtaining quality data. The activities
which were observed during the audit included determinations of:
• Velocity and volumetric gas flowrate (EPA Method 2)
• Gas phase molecular weight (EPA Method 3),
t Gas phase moisture (EPA Method 4),
• Particulate concentration (EPA Method 5),
• Gas phase (Sf^) concentration (EPA Method 6), and
• Visible fugitive emissions (EPA Method 9 and proposed
Method 22).
25
-------
CORPORATION
As indicated on the audit checklists, careful compliance with
accepted sampling procedures was observed for all sampling activities. The
sampling crew exhibited an obvious familiarity with the equipment and
methods used. Internal QC checks such as pre- and post-test leak checks of
sampling train, intermediate calculation of isokinetics, replicate Fyrite
analyses, etc., were carefully followed. The facilities and procedures used
in sample handling and storage were judged to be quite adequate. All data
records were well organized and utilized preformatted data sheets in most
instances. All equipment calibration data was complete and similarly well
organized. Overall, the systems audit indicated an efficient, well
orchestrated sampling effort which was judged to be adequate for achieving
the data quality for each of the EPA Methods as shown in
Table 6-1.
6.1.2 Performance Audit
A performance audit is a quantitative assessment of the data
quality of a sampling and/or analytical system. Both field and laboratory
(analytical) operations were addressed in the performance audit for this
program. Audit activities included:
• field checks of dry gas meter/control console
calibration,
t field check of the laboratory balance
(Mettler PC4400),
• field checks of the Fyrite analyses,
• checks of field calculations,
t check of the data reduction program used for the
sampling data reduction, and
• analysis of S0£ audit samples.
26
-------
RADIAN
CORPORATION
TABLE 6-1. ATTAINABLE ACCURACY AND PRECISION OF TEST RESULTS
RASED ON THE EPA COLLABORATIVE TESTS OF STATIONARY
SOURCE METHODS**
Parameter
Volumetric Gas Flow Rate
Molecular Weight
Moisture
Particulate Mass
S02
Emissions Opacity
Emissions Visibility*
Method
2
3
4
5
6
9
22
Accuracy
(%)
I11
±25
+10
+20
^20
+7.5
+10
Precision
(Standard
Deviation)
(%)
20
10
11
10
10
5
10
*Radian estimate.
**Midgett, M. Rodney, Environmental Science and Technology,
JJ., No. 7, pp. 655-659.
27
-------
Results of the performance audit supported the results of the systems audit,
indicating that the test data quality should be as shown in Table 6-1.
Performance audit results presented below are expressed in terms
of relative accuracy. The relative accuracy for each parameter is
calculated as:
where,
% A = relative accuracy, percent
M = measured value
T = true value of reference standard
100 = factor for conversion to percentage basis
Dry Gas Meter/Control Console
Field checks of the dry gas meters/control consoles were performed
using a Kurz Model 543 flow calibrator (Serial No. 769). The Kurz instru-
ment had recently been calibrated by the manufacturer using an NRS traceable
mass flow meter (NBS test numbers 213-21/190522). The calibration factor,
Y-j, was determined by averaging triplicate measurements at each of three
meter rates (nominally 0.25, 0.50 and 0.75 ACFM).
A second calibration indicated that the dry gas meter correction
factors (Y-j) had changed by less than five percent of their original
calibration value. Thus, the use of the original calibration factor was
appropriate for further data reduction.
28
-------
RADIAN
CORPORATIOM
Laboratory Balance
The accuracy of the laboratory balance (Mettler PC4400) was
checked using a set of NBS traceable Class S weights. Replicate weighings
were made on weights ranging from 0.01 g to 100 g. The greatest difference
between balance reading and actual weight was observed with the 0.5 g
weight. This difference was 0.01 g or 2.0 percent.
Fyrite® Analyses
Ambient air was analyzed during the performance audit using the
Fyrite gas analyzer for carbon dioxide and oxygen. The precision and
accuracy for these determinations were within the readability of the
analyzers (+0.5 percent). Based on this data and the systems audit results
the gas composition determinations should be within the precision and
accuracy ranges estimated for this method (Table 6-1).
Field Calculations and Computerized Data Reduction
A check of field measurements and calculations used for
determining the location of sampling traverse points (EPA Method 1)
indicated that the points were correctly located. Radian's computerized
Source Sampling Data Reduction Program was used to reduce all velocity,
flowrate, molecular weight, particulate mass and SO? data. Example data
sets for Methods 2, 3, 4, 5 and 6 were submitted for reduction using this
program and the results compared to those obtained by hand calculation using
the equations and procedure specified in the Reference Methods. The
comparison of results indicated excellent agreement between the two
procedures with the magnitude of the difference attributable to rounding
differences. The results of this comparison are summarized in Table 6-2.
29
-------
TABLE 6-2. SUMMARY OF DATA REDUCTION CHECK
Parameter
Velocity (ft/sec)
Volumetric Flow Rate (dscf/min)
Molecular Weight
(Ibs/lb-mole)
Particulate Concentration (gr/dscf)
Data
Reduction
Program
Results
15.3505
24184
28.85
0.637
Hand
Calculated
Results
15.3657
24203
28.85
0.639
Accuracy
(%)
-0.04
-0.08
0.0
-0.3
30
-------
COOPORATIOM
jkilfur Dioxide neterminations
The sulfate content of the impinger solutions resulting from the
absorption of sulfur dioxide were analyzed per EPA Reference Method 6 using
the barium perchlorate-thorin titration method. The data quality for these
determinations were assessed by submitting blind EPA Stationary Source
Quality Assurance SOg Reference Standards. These audit standards were
analyzed with the impinger samples. The results for these determinations
are summarized in Table 6-3. The percent accuracy for these anlayses is
within the +7.0 percent control limit.
Two of the six Method 5 and 6 tests incorporated a third hydrogen
peroxide impinger to document the S02 collection efficiency of the
impinger train. As shown in Table 6-4, the results from the analyses of
these additional impinger solutions indicate the collection efficiency using
two impingers was greater than 98 percent.
31
-------
TABLE 6-3. SUMMARY OF EPA AUDIT SAMPLE ANALYSES
Sample
Number
8136
6065
Measured
ng S0?/dscf
805
1317
Actual
ng SOp/dscf
762.6
1334.6
Accuracy
(%)
+5.6
-1.3
Control
Limit
^7.0
+7.0
TABLE 6-4. SUMMARY OF EPA METHOD 5/6 IMPINGER COLLECTION EFFICIENCY RESULTS
mM S02 in mM S02 in Collection
Sample Impingers Impinger Efficiency
Number 1 and 2 3 (%)
EMB 007 MSS 98.1 1.9 >98
EMB 008 MMT 99.3 0.7 >99
32
-------
7.0 CHAIN-OF-CUSTODY
While in the field, all samples were under the care of Michael J.
Krall during the period of November 12 through November 24. Throughout this
period, all samples were processed for transport back to Radian's Austin
laboratories within 24 hours after collection. This involved placing
filters in petri dishes and sealing them with tape. Probe washes and S0£
samples were kept in polyethylene bottles, their lids taped shut, and
initial volume noted. All samples were stored in the Radian Mobile Lab on
site.
On November 24, the samples were transported by automobile from
Deming, New Mexico, to El Paso, Texas. The samples were placed in the cus-
tody of Continental Airlines aboard Flight No. 75. Upon arrival to San
Antonio, Michael J. Krall transported the samples by automobile to Radian's
Austin laboratories where they remained overnight. Final processing of the
samples was completed on December 7, and the samples were placed in storage.
At no time during the process of collection, storage on site, transport, or
final analysis was there any evidence that these samples were tampered with.
33
-------
CORPORATION
8.0 RESULTS
A summary of participate and sulfur dioxide sampling results of
the matte tapping and slag skimming are presetned in Tables 8-1 and 8-2.
The sampling time for each run was 60 minutes while the total elapsed time
from beginning to end was much longer - exceeding 24 hours in some
instances.
The particulate and sulfur dioxide results from the computerized
data reduction are in units of mass of pollutants per hour (of sampling).
These results were converted to mass of pollutants per ton of matte or slag
produced (during sampling) using the following expression:
Mass of Pollutant _ PMR 0
Ton of Production n
(60 min/hr) Z M.T.L
1=1 1 n
where: PMR, in pounds per hour, is the pollutant mass rate of particulate
or sulfur dioxide as calculated through the computerized data
reduction.
0, in minutes, is the sampling time.
n is the number of trapping or skimming operations which were
sampled. (Note: more than one operation may be in progress during
sampling.)
Ti is the fraction of the total duration of the ith tap or
skim which was sampled.
MJ is the fraction of the normal ladle which was removed from
the furnace during the observed itn tap or skim.
L, in tons, is the weight of a normal ladle of matte or slag as
specified by the plant personnel.
34
-------
CO
01
Run
Date
Time
Lbs Particulate/Hr
Lbs Particul ate/Ton
Lbs S02/Hr
Lbs S02/Ton
Matte Production, Tons
Episodes
Gas Flow, DSCFM
TABLE 8-1. MATTE
EMB 004 MMT
11/12-13/81
1247-0844
19
0.10
290
1.6
185
8
42,600
TAPPING DATA SUMMARY
EMB 006 MMT
11/13-14/81
1413-1248
18
0.072
290
1.2
250
8
59,200
EMB 008 MMT
11/14-15/81
1328-1145
1.9
0.054
310
1.1
275
11
60,200
1
5
Average
-
-
17
0.076
300
1.3
237
-
_
-------
00
TABLE 8-2. SLAG SKIMMING DATA SUMMARY MORENCI NO. 3 REVERB FURNACE
Run
Date
Time
Lbs Particulate/Hr
Lbs Part icul ate/Ton
Lbs S02/Hr
Lbs S02/Ton
Slag Production, Tons
Episodes
Gas Flow, DSCFM
EMB 003 MSS
11/12/81
1556-1747
2.0
0.025
15
0.19
80
2
8,000
EMB 005 MSS
11/13/81
1532-1733
2.5
0.028
30
0.33
90
3
8,200
EMB 007 MSS
11/14/81
0854-1057
1.2
0.020
7.6
0.13
60
2
7,600
Average
-
-
1.9
0.024
17
0.21
77
-
_
s
-------
The T-j, M-j, and L terms are presented in Tables 8-3 and 8-4.
These were derived from correlating the start and stop times on the data
sheets with the process observations. The test results are expressed in
units of mass of pollutant per unit sampling time (not clock time). Also,
the mass of matte and slag production on those which occurred during
sampling not that which occurred during the elapsed time of the test.
Visible Emissions
Two types of visible emission testing were performed: Method 9
which yields results that are a time weighted opacity and Method 22 which
indicates the fraction of time visible emissions were present. Neither
method produces data on the percent of fugitive emissions captured, but
lower opacity reported for Method 9 generally indicates a greater capture
efficiency for the hooding system evaluated if all other factors are
constant.
Method 9 is generally used for sources where natural lighting
conditions are present. The use of Method 9 in indoor locations, where
lighting conditions may not be optimal and where observer locations may or
may not be a problem, will not produce data as reliable as data gathered
under the conditions for which the method was intended.
Reverberatory Furnace No. 5 Matte Tapping (Lancing Emissions
Not Included)
Method 9
Matte tapping launder No. 3 hood system was observed most
frequently. Based on approximately.one hour of observations, the average
opacity was 15 percent. Launder Nos. 2 and 4 had average opacities of
10 percent. Based on 14 minutes of observations the launder No. 5 hood.
system had an average opacity of 35 percent. Table 8-5 presents these
results.
37
-------
coRponart
TABLE 8-3. SAMPLING SEQUENCING PHELPS-DODGE MORENCI, ARIZONA
COPPER SMELTER MATTE TAPPING - NO. 5 REVERB
EMB-004MMT EMB-006MMT EMB-008MMT
Date
11/12/81
11/13/81
11/14/81
11/15/81
Episode Ti
1257-1304 0.8
1334-
1445-1453 1
1656-1703 1
1705-1709 1
1733-1740 1
1745-1752 1
0829-0835 1
0836-0844 1
1213-1223
1502-1509
0756-0806
0944-0950
1058-1106
1120-1127
1131-1135
1244-1248
1328-1337
1402-1409
1434-1441
1449_1454
1509-1513
1602-1606
1635-1639
0929-0935
0945-0951
1118-1123
1140-1145
Mi L Ti Mi L Ti
1 25
1
1 25
1 25
1 25
1 25
1 25
1 25
1 25
1 1 25
1 1 25
1 1 25
1 1 25
1 1 25
1 1 25
1 1 25
1 1 25
1 1 25
1 1 25
1
1
1
1
1
6
1
1
1
1
1
Mi
1
1
1
1
1
1
1
1
1
1
1
1
L
25
25
25
25
25
25
25
25
25
25
25
25
T-j is the fraction of the total duration of the -jth tap or skim which
was sampled.
M-j is the fraction of the normal ladle which was removed from the furnace
during the observed ^th tap or skim.
L, in tons, is the weight of a normal ladle of matte or slag as specified by
the plant personnel.
38
-------
TABLE 8-4. SAMPLING SEQUENCING
PHELPS-DODGE MORENCI, ARIZONA COPPER SMELTER
SLAG SKIMMING - NO. 3 REVERB
Date
EMB-003MSS
EMB-005MSS
Episode
Ti
Mi
Ti
Mi
EMB-007MSS
Ti
Mi
11/12/81
11/13/81
11/14/81
1556-1626
1717-1747
1532-1602
1659-1706
1711-1733
0854-0924
1025-1057
40
40
1 1
1 0.3
1 1
40
40
40
1
0.5
40
40
39
-------
TABLE 8-5. REVERBERATORY FURNACE NO. 5 - METHOD 9 OBSERVATIONS
Process
Observed
Matte Tapping
Matte Tapping
Matte Tapping
Matte Tapping
Launder
Number
2
3
4
5
TABLE 8-6. REVERBERATORY
Process
Observed
Matte Tapping
Matte Tapping
Matte Tapping
Launder
Number
3
4
5
Number
of
Events
Observed
7
7
7
2
FURNACE NO. 5
Number
of
Events
Observed
1
1
1
Avergae
Opacity
10
15
10
35
- METHOD 22
Time
Emissions
Observed
(%)
100
82
100
Total
Observation
Time
(Min/Sec)
55:00
66:00
42:00
14:00
OBSERVATIONS
Total
Observation
Time
(Min/Sec)
6:38
5:00
5:33
40
-------
RADIAN
comxMumoft
Method 22
Fugitive emissions were observed escaping capture by launders
No. 3 and No. 5 hood systems continuously. The hood system for launder
No. 4 was observed for approximately five minutes. Emissions escaped
capture 82 percent of the time. Table 8-6 presents these results.
Reverberatory Furnace No. 3 Slag Skimming (Lancing Emissions
Not Included)
Method 9
Based on over two hours of observations, the average opacity of
emissions escaping capture by the launder No. 1 hood system was less than
five percent. The average opacity of the emissions from launder No. 2 was
also less than five percent. Table 8-7 presents these results.
TABLE 8-7. REVERBERATORY FURNACE NO. 3 - METHOD 9 OBSERVATIONS
Process
Observed
Launder
Number
Number
of
Events
Observed
Average
Opacity
Total
Observation
Time
(min/sec)
Slag skim
Slag skim
1
2
4
2
<5
<5
122:00
33:00
41
-------
Method 22
Launder No. 2 was evaluated using Method 22 for approximately 30
minutes. Fugitive emissions were observed escaping capture by the hood
system three percent of the time.
The visual emissions testing indicate each of the hoods for the
matte handling systems on the No. 5 reverb perform equally as well. The
system for tap bay No. 5 may be an exception to this with an opacity
approximately three times greater than the others. Each of the three tests
incorporated approximately equal fractions of the No. 5 bay and the reduced
collection efficiency of this hood cannot be verified.
The performance of the skim bay hooding was much better than that
of the tap bays as determined from both the visible and sampling data.
Examination of the data presented in Tables 8-1 and 8-2 does not
show a strong dependence of the emission rate, mass of pollutant emitted per
unit time, on production rate. Thus, the emission factors, mass of pol-
lutant emitted per unit of production, are not constant and in some cases
actually decrease with increasing production. A possible explanation for
this situation is the strong dependence of the emission rate on the exposed
surface area of the molten material and the time it is exposed. This type
of dependence would result in reduced emission factors with increased
flowrates of molten material. With this approach the exposed surface of
molten material in the ladles would be a significant contributor with its
contribution being inversely proportional to the matte or slag flowrate in
the launder. Sampling was terminated when the flow of molten material
stopped, thus emissions from a ladle of material waiting for transfer, were
not addressed by this testing protocol.
42
-------
APPENDICES
At the request of the Project Officer, a number of copies were
published without the appendices which consisted of original field data
sheets and raw, reduced data. This is one of those copies. Please contact
the Project Officer if details of those data are required.
43
------- |