EPA-650/2-74-111
OCTOBER  1974
Environmental Protection  Technology Series







                                             £  £*  \
                                             I  *****  5
                                             V
                     SE2

                                     :::::::;::::::;:;:;:;:gj[fi^:::»g::g:^^
                                        ;liili;l;i;^^M^^.l^H^3i

-------
                             EPA-650/2-74-111
MEASUREMENT  OF  SULFUR  DIOXIDE,
PARTICIPATE, AND TRACE ELEMENTS
  IN COPPER  SMELTER  CONVERTER
  AND ROASTER/REYERBERATORY
            GAS STREAMS
                   by

               Robert M. Statnick

            Control Systems Laboratory
         National Environmental Research Center
       Research Triangle Park, North Carolina 27711
              ROAP No. 21ADM-012
            Program Element No. 1AB012
       NATIONAL ENVIRONMENTAL RESEARCH CENTER
        OFFICE OF RESEARCH AND DEVELOPMENT
       U.S. ENVIRONMENTAL PROTECTION AGENCY
        RESEARCH TRIANGLE PARK, N. C. 27711

                October 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

-------
                              CONTENTS
                                                              Page
  List of Figures                                               v
  List of Tables                                               vii
  Acknowledgments                                              vii i
  Sections
  I    Introduction                                              1
      Background                                                1
      Conclusions                                               4
      Recommendations                                           5
      Conversion  Factors                                        6
      References                                                7
 II    Plant Operating  Data                                     8
      Converter Operations                                     8
      Roaster and Reverberatory Furnace Operation             10
III    Sulfur Oxide  Tests                                       15
      Experimental                                             15
      Results                                                 17
      Discussion                                               20
      References                                               24
 IV    Particulate Tests                                       25
      Experimental—Particulate Mass                           25
           Particulate Control Equipment                      25
           Sampling Port  Locations                            25
           Sampling Methods                                   32
           Analytical. Procedures                              37
                                        111

-------
                          Contents (Cont.)

                                                            Page
         Sampling Approach                                  38
    Results—Particulate Mass                               38
         Converter ESP Inlet                                38
         Converter ESP Outlet                               42
    Experimental—Particulate Size                          43
    Results--Particulate Size                               48
    Discussion                                              48
         Converter ESP Tests                                48
         R&R ESP Tests                                      60
    References                                              64
V   Elemental Composition                                   65
    Experimental                                            65
    Results                                                 66
    Discussion                                              69
    References                                              71
Appendix A— Sampling and Analysis of Mercury                A-l
            Vapor in Industrial Streams Containing
            Sulfur Dioxide
    Introduction                                            A-2
    Experimental Methods                                    A-3
    Field Sampling                                          A-10
    Results                                                 A-l3
    Conclusions and Summary                                 A-l3
                                  iv

-------
                          FIGURES
No.                                                         Page
 1    Schematic of Tacoma Smelter                              9
 2   Sampling Log for Converter Runs 1A (Inlet) and          11
       IB (Outlet)
 3   Sampling Log for Converter Runs 2A (Inlet) and          12
       2B (Outlet)
 4   Sampling Log for R&R  Runs 3 and 4                       13
 5   S03/S02 Sampling Apparatus                              16
 6   Sampling Location  at  Water Spray Chamber Inlet          18
 7   Particulate Sampling  Sites                              26
 8   Sampling Point Diagram--R&R Furnace ESP Outlet          28
 9   Sampling Point Diagram—Converter ESP Outlet            30
10   Sampling Point Diagram—Converter ESP Inlet             33
11    Brink Sampling Train                                    44
12   Andersen Sampling  Train                                 46
13   Cumulative Percent of Particles Less than               49
       Given Size—Converter Inlet
14   Cumulative Percent of Particles Less than               50
       Given Size—Converter Outlet
15   Cumulative Percent of Particles Less than               51
       Given Size—R&R  Inlet
16   Cumulative Percent of Particles Less than               52
       Given Size—R&R  Outlet
17   Differential Size  Distribution—Converter               53
       ESP Inlet
18   Differential Size  Distribution—Converter               53
       ESP Outlet

-------
                           Figures  (Cont.)

No.                                                        Page
19   Differential  Size Distribution—Converter               55
       ESP Inlet and Outlet
20   Converter ESP Efficiency  as  a  Function of               55
       Particle Geometric Diameter
21   Differential  Size Distribution—R&R ESP Inlet           59
22   Differential  Size Distribution—R&R ESP Outlet          59
23   Fractional Collection Efficiency of R&R ESP's           60

-------
                            TABLES

No.                                                         Page
 1   Log of Samples Acquired                                  3
 2   Average Converter ESP Operating Variables                14
 3   Average R&R ESP Operating Variables                      14
 4   Measured Concentrations of SOx at Standard               19
       Conditions
 5   S02 Emission Levels                                     22
 6   Effect of Acid Plant on SOX Emission                     22
       Concentrations
 7   Summary of ASARCO Particulate Mass Measurements          31
 8   ASARCO Parti cul ate Mass Data                            39
 9   Brink Impactor Operating Conditions—ESP  Inlet           47
10   Andersen Mark III Operating Conditions—ESP  Outlet       47
11   Converter Particulate Mass  Collection  Efficiency        57
12   Representative Precipitation Rates                      57
13   Chemical Composition of Solids and Residues  from        64^
       the Particulate Mass Sampling Train,  % by  Weight
14   ASARCO Concentration of Vapor-Phase Mercury  in          65
       Flue Gas
15   Elemental  Collection Efficiencies of Converter ESP      67
16   Mass Emission Rate of Selected Elements                 67
                             VII

-------
                       ACKNOWLEDGMENTS
     The author wishes to express his appreciation to
Messrs. R. Steiber, W. Fowler, and R. Ogan for assisting
in the performance of the measurements;  and  to Monsanto
 Research  Corporation  and Research  Triangle  Institute  for
 performing several  of the sampling and analysis  tasks.
     The author is especially indebted to Mr. J.  A. Dorsey
who encouraged and supported this effort; and a special thank
you to Miss C. Atkinson for tolerating many rewritings.
                           viii

-------
                         SECTION I
                        INTRODUCTION

BACKGROUND
     The primary non-ferrous smelting industry is a significant
source of sulfur oxides, particulate, and chemical species which
could be classified as hazardous (for example, Pb, Zn, or As
emissions).  Smelters in the continental United States emit
1,920,000^ short tons* of sulfur to the atmosphere annually.
Over 97 percent of the sulfur oxide emissions are due to smelting
operations west of the Mississippi River.  The primary non-ferrous
smelting industry emits an estimated 243,000 tons^ of particulate/
year.  (It is the ninth major source of particulate pollution.)
Due to a paucity of data, no reliable estimates of the yearly fine
particulate emissions can be made.^  An EPA sponsored program to
determine the elemental distribution of ores (or concentrates)
smelted in the United States reveals^ a wide distribution of abun-
dances of the analyzed elements.  As in the case of fine particulates,
very sketchy data exists on the elemental abundances in the emitted
particulate.
     To acquire data useful for developing a viable control program,
the Control Systems Laboratory (CSL) sought access to American smel-
ters.  In June 1973, EPA Region X's OR&D** office sought the
 *A1though EPA's policy is to use the metric system in all  its docu-
  mentation, certain non-metric units are used in this report both
  for convenience and to reflect actual  test conditions.   Readers more
  familiar with metric units may use the conversion factors provided
  on page 6.
**0ffice of Research and Development.
                             1

-------
 assistance of  CSL's  Process Measurements Section to obtain data
 on  emissions and  control efficiencies from the American
 Smelting  and Refining Company's  (ASARCO) Tacoma, Washington,
 smelter.
      A  program to evaluate the efficacy of the existing control
 equipment—both particulate and  sulfur oxides—was developed.
 The test  program  was conducted September 25-27, 1973, by personnel
 from the  Process  Measurements Section and Monsanto Research
 Corporation.
      A  log of  the samples acquired, date of acquisiton, fre-
 quency  of acquisition, and sampling location is given in Table 1.
      The  particulate data presented in this report is based on
 the dry filterable solids.  Previous studies5 have shown that
 when S02  is present  in the gas stream sampled, the Greenburg-
 Smith impinger catch is principally (> 90 percent) sulfuric acid.
 However,  the analysis for the elemental abundances in the emitted
 stream  included both the dry filterable solids and the impinger
 catch.  As will be shown later in the report, the observed control
 efficiencies for  the dry filterable solids and the observed con-
 trol  efficiency of the specific  elements are nearly identical, which
•supports  the concept of using only the cyclone-filter catch in
 assessing the  particulate mass emission rate.
      The  data  presented in this  report is of significant value
 for assessing  the control efficiencies and the emission levels of

-------
           Table 1.   LOG  OF SAMPLES ACQUIRED
Date Frequency
9/25/73 3
3
3
3
Type of
sample
S02
Hg (vapor)
Brink impactor
Andersen Impactor
Location
Converter ESP inlet
Converter ESP outlet
Converter ESP inlet
Converter ESP outlet
9/26/73
9/27/73
2   Particle mass  loading   Converter ESP inlet
2   Particle mass loading   Converter ESP outlet
3       Brink impactor
3     Andersen impactor
3            S02
3         Hg (vapor)
3            S02
3            S02
R&R inlet
R&R outlet
R&R ESP outlet
R&R ESP outlet
Acid plant inlet
Acid plant outlet
            2  Particle mass loading  R&R outlet

-------
the processes sampled.  The data Is technically valid and may
be used for this purpose.  A* times, strict adherence to the
varied technicalities of compliance testing was not possible;
at other times,  it was neglected because of the limited resources
and the goals of the  project.  The jdata should be considered
carefully and utilized within the above mentioned limitations.
CONCLUSIONS
     Based on the data collected, the mass emission rates of
sulfur dioxide,  particulate, and selected elements were deter-
mined.  The size distribution data was used to determine the
fractional size  collection efficiency of the particulate con-
trol equipment.  The analysis of these results led to the
       i
following conclusions:
     1.  The sulfur dioxide emission rates from the R&R and
converter effluent streams are 518 Ib/min and 587 Ib/min,
respectively.  The R&R flue represents 47 percent of the total
sulfur dioxide mass emission rate.
     2.  The efficiency of the acid plant for the control of
sulfur dioxide was 96.8 percent which is comparable to the
efficiency of other acid plants tested in the same application.
     3.  The mass collection efficiency of the converter electro-
static precipitator (ESP) for dry filterable solids was 95 percent.
     4.  The ESP's appear to have a minimum collection efficiency
for dry filterable solids between 0.8 and 1.2 pm.  Negative collec-
tion efficiencies below 0.8 ym were noted on both streams and

-------
could arise from condensation,  non-representative sampling
locations, or ESP operations.   Insufficient data  is  available
to differentiate between these  possibilities.
     5.  Analyses of the emitted particulate for  selected ele-
ments gave mass emission rates  (Ib/hr) of:   arsenic  (58.05);
lead (24.65); cadmium (1.32);  zinc (15.7);  chromium  (0.065);
and copper (4.825).  The control  efficiency for the  elements
analyzed ranged between 90 and  98 percent which is comparable
to the overall mass collection  efficiency.
RECOMMENDATIONS
     Two basic recommendations  resulted from the  studies con-
ducted during this program.
     1.  It is recommended that quantitative information be
developed for fugitive emissions from the smelting process.
These low-level, low-velocity emissions could contribute signi-
ficantly to As and Pb concentrations in local  ambient air.
Several attempts have been made, but each was unsuccessful.
     2.  A program of abatement is recommended based on sulfur
oxides and particulate control  for the roaster/reverberatory
flue.  Such a program would achieve a large reduction in emis-
sions from these smelting process operations, because the R&R
flue represents approximately 50 percent of the stilfur dioxide
emissions and about 75 percent of the analyzed elemental emissions.

-------
CONVERSION FACTORS
     Although it is EPA's policy to use metric units for all
quantitative descriptions, certain non-metric units are used
in this report for convenience.  Readers more familiar with
the metric system, may use the following factors to convert
to that system.
Multiply
non- metric
cu ft
Op
ft
gr
in.
Ib
sq ft
sq in.
ton (short)

By.
28.32
5/9 (°F-32)
30.48
0.06
2.54
0.45
0.09
6.45
0.91
907.18
To convert
to metric
liter
°c
cm
g
cm
kg
sq m
sq cm
ton
kg

-------
REFERENCES
1.  Systems Study for Control of Emissions, Primary Nonferrous
    Smelting Industry, Vol. I, Arthur G. McKee & Company, NTIS
    No. PB184-884.  June 1969.  p.  ,11-1.
2.  Particulate Pollutant System Study, Vol. I, Mass Emission,
    Midwest Research Institute, NTIS No. PB 203-128.  May 1971.
    pp. 5, 133.
3.  Particulate Pollutant System Study, Vol. II, Fine Particle
    Emissions,  Midwest Research Institute, NTIS No, PB 203-521.
    August 1971.   p. 113.
4.  vonLehmden, D.  Private communication.  September 1972.
5.  GASP-Tacoma Area.   Newsletter.   February 1973.   Dr.  S.
    Milhan, State of Washington Health Services Department.

-------
                        SECTION II
                   PLANT OPERATING DATA

     A solids and effluent flow diagram of the Tacoma  Smelter
is given in Figure 1.  The major gas.effluent streams  are
depicted by the solid lines and the flow of the copper-containing
mineral is shown by the dashed lines.   The emission streams
indicated in the diagram are those which vent into either  of
two effluent ducts which feed into the 535-foot-tall brick
stack.  Two oil burners are constantly firing into the base  of
the stack to raise the effluent gas temperature.
     The converters and roaster/reverberatory furnaces at  the
ASARCO plant were in operation only during the evening and
night hours.  Careful attention was paid to the meteorological
conditions:  smelting would begin only when the emissions  from
the stack would rise and not settle over populated areas.
CONVERTER OPERATIONS
     The emissions from the four converters at the plant are
controlled by an electrostatic precipitator (ESP).  The ESP
had 159,450 square feet of surface area and processed  approxi-
mately 130,000 scfm of flue gas.  The ESP parameters are given
in Table 2.
     A series of four lights on the roof of the building housing
the inlet to the ESP and another group of lights on the building
over the breeching indicated which units were in operation.
                              8

-------




FACK
t















1

t









1
\ \ PLATE
P TYPE -* 	
ESP
1
>»
v
^. ^'
vy
i
i
PLATE / /
D iTPt Q[ i
t tSP t 1 1

' 'i

ARSENIC
PLANT
-«« 	
1 |
TUBE \ REVERBERA- ROASTER
TYPE O -~* 	 TORY — - * 	 (NICHOLS- 	 *"n, r
ESP |~" FURNACE HERRESCHOFF) PILE
1 1
i
J 1

<^
f ^ 	 CONVERTER - ^- 	 iSf?

L „ — ^W- Fl IIP RAS FinW
i 1 	 -^- 	 SOLIDS FLOW
1 f n«i 	 SAMPLING LOCATIONS
T
D < . A RAS STRFANI CnNniTinNIN(i
U ACID PLANT *U EQUIPMENT NOT SHOWN
lWri 	

Figure 1. Schematic of Tacoma smelter.

-------
During each of the sampling periods, the information as  to
whether finish or slagging operations were being conducted
was supplied by plant personnel.   In general,  only two of the
four converters operated at any given time.
     Once the sampling run began, a notation was made of how
long each lamp was lit.  Sampling would continue as long as
at least one lamp remained illuminated.  Sampling was stopped
when all lamps were extinguished.  Figure 2 is the sampling
log of runs 1A and IB, showing the sampling times and which
converters were in operation during the sampling run.  Figure
3 is the same type of sampling log for runs 2A and 2B.  As  can
be noted on the sampling logs, sampling was stopped about 1
minute after the lamps were turned off.  This delay was caused
by the distance from the sampling sites to the lamp observation
location.
ROASTER AND REVERBERATORY FURNACE OPERATION
     The R&R flue gas was treated by dry pipe type ESP and then
by parallel type ESP's.  The ESP's  treat a nominal 475,000 scfm
flue gas.  The pipe ESP's had  57,500 sq ft of surface area and
the plant  ESP's had 174,240 sq ft.  The ESP parameters are
shown  in Table 3.
     Under normal conditions at  least  six  of  the  plant's eight
roasters would be in operation at any  given time.   Sampling  was
therefore  started as soon as word was  received  from  the  plant
personnel  that at least  six of the  units were on  stream.   Figure  4
is a  log of R&R operations during sampling runs  3 and 4.
                              10

-------
 4- SLAG
  3-
 2- SLAG
1-FINISH
OUTLET
  INLET
1 1 1
A A
0132 0220




0220
¥
A
0225
0159 0221
1 T
A A A
0117 0208 0230
0220
v
A A
0120 0226
1 1 1
1 1 I 1


0415
f
A A
0319 0430
0245 0333
r v
A A
0252 0337
0246 0300
» Y
A A A A A
0253 0305 0347 0408 0450
0245
T
A A
0252 0415
till
      0100
0130       0200
0400
                              0300
                       TIME, 24-hour clock
Figure 2.  Sampling  log for runs 1A (inlet) and 1B (outlet).
0500
                                         11

-------
4-FINISH
3-
 2- SLAG
 1- SLAG
OUTLET
  INLET
2152 1
12156

A
2155



2152
12159
s
[2205
2157
• 2153
f


2155
L-
k
2206

1
1 1 1
2247 2310
1 f
M A
22421 2316
2249



S 2226 2320
f r
A
2306
2312
2244 1 2322 2338
f * I I
A A A A
2235 2306 | 2334
2316
2220 2300 2312 2324
T 1 » »
i M
2224 2310 f
2318
1 1 1
I







2355 0017
L. !_
A A
2353 0004

2355 0018
T f
A A
2353 0006

2356 0018
L L_
~i A
2355 0005

1
1 1 1 1




0105 0150
V V
* \ —
i i
0144 0203
0047 0103
f f
A A
0045 0102
0041
10047 0108
tr T
A i A 4
0036 T 0103 0132
0045
0048 0104
f T
A A A
0045 0103 0203

1 I 1 1
      2130        2200                  2300                  2400                  0100                 0200




                                                  TIME, 24-hour clock






                            Figure 3.  Sampling log for runs 2A (inlet) and 2B (outlet).
                                                  12

-------
                                                        2300
                                         Run No. 4, 9/27/73
                                                                       0330
cc
LU
a.
o
1900
Run No. 3, 9/27/73
                          2300
3
cc

Ll_
O

cr
LJ
CQ
                1900
      2100
                                       2300
                                                                           0100
                                                                 0300
                                                TIME, 24-hour clock


                            Figure 4.  Sampling log for R & R runs 3 and 4.
                                                13

-------
           Table 2.  AVERAGE CONVERTER ESP OPERATING VARIABLES
Variable, cfm/sq ft                                 0.81
Variable, ya/sq ft                                  7.8
Variable, spark rate/minute                       110
Temperature, ° F
  Inlet                                           253
  Outlet                                          245.5
           Table 3.  AVERAGE R&R ESP OPERATING VARIABLES


Variable,
Variable,
Variable,

cfm/sq ft
ua/sq ft
spark rate
Pipe
8.25
12.5
0
Plate
2.72
8.2
14.3
Temperature, ° F
  Inlet                             NA
  Outlet                            —             178.5
                              14

-------
                       SECTION III
                   SULFUR OXIDE TESTS

EXPERIMENTAL
     The primary concern of EPA Region  X  centered  upon  the  emission
of sulfur dioxide.   It was decided to utilize a slightly modi-
fied Shell procedure because the sampling team had extensive
experience with this method.  Since the method requires that
the first absorber be filled with 80 percent isopropanol (IPA),
it was decided to analyze the IPA solution for sulfate  rather
than discard it although this analysis  would provide an esti-
mate of the acid mist emission since isokinetic sampling was
not employed.   Identical sulfur oxide sampling procedures were
employed at all sampling locations.
     The sampling apparatus, shown in Figure 5, differs from
that used in the Shell procedure primarily in that a fine frit
was not used between the first and second absorbers.  Data  sub-
mitted in the Wai den Research Corp. report'  point  out the
                                                         P
equivalency, if not superiority, of Lamp  Sulfur Absorbers  to
midget impingers as gas contactors over a wide range of sampling
rates.
     The collection .media used are identical to those in the
Shell method and in the Federal Register3 (80 percent IPA for
SOs and 3 percent ^2 for S02).  Further details  of the sampling
apparatus are  obtainable from various other  sources.^
                            15

-------
HEATED
QUARTZ
 PROBE
FIBER GLASS PLUG
                                     A 30 ml 80% IPA
                                     B & C 30 ml EACH 3%
                                                                    SILICA
                                                                     GEL
                                       LAMP SULFUR ABSORBERS
                                                              Hg
                                                          MANOMETER
                              Figure 5.  S03/SO2 sampling apparatus.

-------
      The sampling location  selected on  the converter flue
 duct was preceded by  a  200-300  foot run of brick/concrete
 duct.  A single-point sample was acquired from a depth of
 6 feet from the  top of  the  duct (11.5 feet high x 24 feet
 wide),  at the  inlet to  the  electrostatic precipitator.
      The sulfur  dioxide sample  from the roaster/reverberatory
 (R&R) flue was acquired after the electrostatic precipitator
 and the sample was again acquired at a depth of 6 feet from
 the top of the duct (16 feet high x 20 feet wide).  The sam-
 pling location for the  R&R  flue was preceded by a straight
 run of about 100 feet of duct work.
      Sulfur oxide samples were  also acquired from the inlet
 and outlet of a  150-ton/day Monsanto-designed H2S04 acid plant.
 The inlet sample was  acquired just prior to the acid plant
 water-conditioning sprays.  The spray chamber was about 12 feet
 wide  x  16 feet high, with the sampling point approximately 2-1/2
 feet  from the top of  the duct.  (See Figure 6.)  The acid
 plant outlet sample was acquired from the midpoint of a 36-inch
 diameter  fiberglass-reinforced plastic (FRP)  duct.
      The  samples collected were transferred to Teflon bottles
 for shipment to NERC-RTP.   The samples were titrated using barium
 perchlorate ion to the Thorin endpoint per Federal  Register
Method 6.3
 RESULTS
     The  results  of the S02 tests are  given in Table 4.   The
wide range of observed converter S02 values can be  ascribed to
                              17

-------
                                                     FIRST H20
                                                     SPRAYS
Figure 6.  Sampling location at water spray chamber inlet.
       18

-------
   Table 4.   MEASURED CONCENTRATIONS OF SOX AT STANDARD CONDITIONS
Location

Converter
stream



Date Time
(P.
9/25/73 8:

8:
9:
9/27/73 6:
m. j
25

50
15
50
S02
ppm
13

37
25
35
,856

,871
,766
,909
S03a
Comments
ppm
66.

36.
14.
83.
3

7
4
2
Taken

Taken
Taken
Smell
prior

prior
prior
of SO;
to

to
to
ESP

ESP
ESP
> when
Average

Reverberatory
stream
Average

Acid plant
outlet
Average
          7:08  4Q.295   67.5



          7:25  27.720  108.7



                30,236   62.8

9/26/73   8:40   5,170.9 11.9

          9:10   5,609.0 20.6

          9:25   5,740.0  4.1

                 5,506.6 12.2

9/27/73   7:45   1,849.7 12.8

          8:00     677.9  8.4

          8:10     794.6  6.5

                 1,107.4  9.2
                                                 bottle opened;
                                                 taken at acid plant
                                                 inlet

                                                 Smell  of S02 when
                                                 bottle opened;
                                                 taken at acid plant

                                                 Smell  of S02 when
                                                 bottle opened;
                                                 taken at acid plant
                                                 inlet
Taken after ESP

Taken after ESP

Taken after ESP
Estimated gaseous $03.
                                 19

-------
the cyclic nature of the operations of the converter pot line.
The cyclic nature of the blowing operation leads to highly
variable S02 levels which are evident in the short sampling
duration manual test results.  These data are consistent with
                                     *••
reported typical converter acid plant inlet 862 values which
range between 1.2 and 4.5 percent SC^ v/v.^
     The results of the analysis of the 80 percent IPA solution
are also given in Table 4.  This information is given as an
estimate of the magnitude of the gaseous S03 concentration.
     Particulate control on the converter and R&R flues is
achieved by the use of electrostatic precipitators.  The high
resistivity of the metal oxide fume requires the conditioning
of the particulates with H2S04-  This acid addition contributes
to the SOs level as well as direct reaction with particulate to
form sulfates.5  (PbS04 was identified in the dust collected by
the converter ESP.)  The $03 values determined during testing
may be higher than in other locations.
DISCUSSION
     SOx associated with the R&R streams are emitted directly
to the atmosphere.  At the present time, no SOx control technology
is being utilized to abate the "weak" SOX streams.  (Interrupted
plant operations are used to achieve ambient air standards.)
The mass emission rate at typical  flow rates (approximately
480,000 scfm for the R&R stream) is given in Table 5.
     The Tacoma smelter converter duct is equipped with an acid
plant which has a throughput capacity equivalent to 18 percent  of
                             20

-------
the total converter flue gas rate (approximately 23,000 scfm).
The efficiency of the Monsanto acid plant for S02 is given in
Table 6.  The effluent values are similar to those observed at
National Zinc (Bartlesville, Oklahoma^  and Kennecott Copper
(Garfield, Utah).6
     The impact of operation of the acid plant on the SOx mass
emission rate is given in Table 5.  The overall emissions asso-
ciated with the smelting operation are almost evenly split
between the R&R stream (518 Ib/min) and the converter stream
(587 Ib/min).  It is to be noted that ASARCO is currently
installing the dimethylaniline process to handle the remainder
of the converter effluent.  The concentrated S02 stream will
be sent to a liquid SO? plant.  SOg removal efficiencies of
95-98.75 percent have been reported.?
     During the author's  field visit, the acid plant experienced
an "upset condition" (acid plant taken off-line).  During the
upset, gas samples were acquired at the converter ESP inlet.
The converter emissions,  at a nominal flow rate of 130,000 scfm
and with the acid plant off-line, are given in Table 5.  Since
no information could be obtained on the frequency of upsets,
the real impact of the acid plant's being off-line on an annual
basis cannot be evaluated.
     The requirement for adequate S02 control  technology appli-
cable to a reverberatory  or an R&R stream is seen if overall
removal  of SOx associated with copper smelter operations is to
                             21

-------
                Table 5.   S02 EMISSION LEVELS
Source                                           S02
                                               Ib/min.
Roaster/reverberatory flue                     518.01

Converter flue, acid plant on-line             587a

Converter flue, acid plant off-line            715.97




Estimated                   .  .




   Table  6.   EFFECT  OF ACID PLANT ON SOX EMISSION CONCENTRATIONS

Inlet
so.x, %
3.59
4.03
2.77
Avg. 3.46
Outlet
SOX> %
0.18
0.07
0.08
0.11
Average9
% removal
96.7
 aThe inlet and the outlet samples were not taken simultaneously,
  Therefore, only an average  removal efficiency is reported.
                                 22

-------
achieve a level >80 percent.   A comparison of the distribution
of the SOx contained in the converter stream and in the R&R
stream shows that at normal flow rates the R&R stream contains
47 percent of all SOX released by the two processes.  In prac-
tical terms then, SOX emissions associated with the smelting
process cannot be reduced by more than approximately 60 percent
unless control is utilized on both effluent streams.
     An additional requirement associated with this control
technology is the need for separation of entrained dust; for
example, the Tacoma smelter processes the collected particulate
to produce As20a as a recoverable by-product.  Other smelters
process collected particulate matter to yield such valuable
metals as cadmium, lead, and indium.  The cost of employing
such processes as wet limestone scrubbing would have to include
additional costs for processing large amounts of unreactive solids
or penalties for not processing collected particulate for by-
product values.
                              23

-------
REFERENCES
1.  Improved Chemical  Methods for Sampling and Analysis  of
    Gaseous Pollutants from the Combustion of Fossil  Fuels,
    Vol. I, Wai den Research Corp., NTIS No. PB 209-267,
    June 1971.
2.  Purchased from Ace Glass, Inc., 1430 N.W. Boulevard,
    Vineland, N. J.  08360.
3.  Federal Register,  36, 247, Method 6 (p. 24893-95).
    December 23, 1971.
4.  Godsey, E., ASARCO.  Personal communication.
5.  Wyatt, S. Miss, ESED, U.S. Environmental Protection Agency,
    Research Triangle Park, N. C.  Personal communication.
6.  Ajax, R. J., ESED/ETB, U.S. Environmental Protection
    Agency, Research Triangle Park, N. C.  Personal communi-
    cation.  October 3, 1972.
7.  Henderson, J. M., ASARCO.  Response to Questionnaire
    Regarding DMA Scrubbing System.  March 15, 1973.
                           24

-------
                          SECTION  IV
                        PARTICIPATE TESTS
 EXPERIMENTAL— PARTICIPATE MASS
 Particulate  Control  Equipment
      The ducted emissions from  the copper smelter are controlled
 by two  ESP systems:  one, a parallel  plate ESP, controls the
 emissions from the  converter pot line; the other, consisting
 of dry  tube  and parallel plate  ESP's  with effluent gas flowing
 through them in series, controls the  emissions from the R&R
 furnaces. The  ESP's are connected by individual breechings to
 a single brick  stack,  535 feet  tall.  A diagram of the ducts
 sampled for  particulate matter  is  shown in Figure 7.
 Sampling Port Locations
      All  sampling at this plant employed existing sampling ports.
 In all  cases  the ports were not located to meet Method 1 criteria.
 However,  no  ideal alternate locations for sampling ports were
 available consistent with program  goals and resources.  The duct
 and port locations at each site are described below.
      Roaster and Reverberatory  ESP Outlet.  This outlet was sam-
.pled from a  series of  ten sampling ports installed in the top
 of the  duct.  The ports were located  about 100 feet from the exit
 end of  the ESP  and 3-4 feet before a  berid in the breeching to
 the stack.   The duct at this point was 20 feet wide and 16 feet
 high.   The sampling port location, while not meeting Method 1
 criteria, was  the only site available with existing ports.  The
                             25

-------
                                          ROASTER AND

1
I SAMPLE POINTS
_ 2

PLATE
ESP
TUBE
ESP

ESP
REVERBERATOR*
) ? S
1
SAMPLE POINT
) ° (!
CONVERTER
              (PLAN VIEW)
                 ESP
          (ELEVATION)
Figure 7.  Particulate sampling sites.
               26

-------
sample ports were 3-3/4 x 2-1/2 inches  and consisted of pipe
sections that extended 17 inches above  the roof of the breeching.
     Considerable dust had collected on the bottom of the  duct,
reducing the area of air flow.   Soundings of the height of the
dust build-up were made by ASARCO personnel.  Figure 8, a  dia-
gram of the cross-section of the duct at the sampling points,
gives the measurements of free  area from the dust pile at  each
port to the roof of the breeching.
     As the sampling ports are  2 feet apart, the duct was  divided
into 2-foot square areas, with  the traverse point chosen to be
at the center of each area.  Thus, the  first point was 1 foot
inside the duct; the second point, 2 feet deeper, etc.  No
point was selected that would be very close to the layer of
particulate in the duct.  The location  of each of the 56 sampling
points is shown by a circle in Figure 8.
     The area of the duct was calculated by multiplying the free
air depth of each port in the duct by 2 feet, and then adding  the
segments together.  The resulting area  was 289.2 square feet,
instead of the 320 square feet indicated by the entire duct dimen-
sions.
     Roaster and Reverberatory ESP Inlet.  This location was
provided with 3/4-inch ID ports and the headroom from the  top
of the port to the underside of the building roof was about 6
feet.  The.small port diameter prevented the use of available
sampling equipment; therefore,  particulate mass was not sampled
at this location.

                            27

-------
A B C D E

SAMPLE PORTS
























<*~2~>"
i
t 1


C
7-4 9-
y •
.<* '• K
••'•''.('••
• • '•'»'"•




: i



i i



-9 12
1
J: J

<
' '. • *.v

••"•'••'
' ." •:"'"•• '

(


<

NOTE: ALL DIMEN- -'-; /'%:-. I






<
•10 13


<





SIONS ARE IN FEET- ' ' "' '••"••''"•• 1
NCHES. .••;-:>.'.V:Vy^._ '
F G H I i
IT
1
i
j
!
i
—
11










DOTS SHOW . -'• •'. :••:•• •>;-•:.•
SAMPLING POINT ' • .• V
LOCATION . '•-,'.•
•5
r
t


r
<
14


<


c


<

::-.-...






-3 U
i


1


^


<


' '•' ''.''.'•''.:':•• . '












i





? t
.-1 15-5 15-3 U
I '
J
i /
<









' ' \" '.'. '•''". 'V":"^-:';'
• <
1








I
> i
(



i




i
LL
L_
»


1
-n











i
i
i


2
3

4
5



6


7




* ~ : 	 —20 	 	 ^
Figure 8. Sampling point diagram-R&R furnace ESP outlet.
                      28

-------
     Converter ESP Outlet.  This outlet duct was provided with
a series of ten sampling ports 2 feet apart, beginning 1 foot
from the side wall of the duct.  These ports were approximately
2 inches ID and extended 30 inches above the roof of the duct.
The ports were located about 100 feet from the ESP and 3-4
feet from a bend in the breeching.  This site was used for
sampling, although Method 1 criteria could not be met at this
location.
     An initial traverse taken at the location indicated that
little or no flow, or in some cases slightly negative flow, could
be detected at the seven traverse points on ports B through G,
Figure 9.  As a result, the ports labeled A, H, I, and J were
employed for sampling, each with seven traverse points, for a
total of 28 points tested.  The duct dimensions at this location
were 20 feet wide by 15 feet high for an overall duct area of
300 square feet.   The area of gas flow was calculated from the
seven 2x2 foot traverse areas in the duct for those ports
indicating positive flow.  The resulting area was 112 square
feet instead of the 300 square feet indicated by the entire duct
dimensions.   The effective area utilized in the calculations given
in Tables 7 and 8 is  equivalent to a 14 x 8 foot duct.  Since
the duct had large areas of negative flow, the error introduced
by not including the remaining 1  foot of duct height (4.5 percent
of the total  area) is probably small  when compared to the uncer-
tainty in establishing the point of demarcation of the positive
and negative flows.
                             29

-------
     1
    •*»•
15
                   o 1
                   o 2
                   o 3
                   o 4
                   o 5
                   o 6
                   o 7
 NOTE: ALL DIMEN-
SIONS ARE IN FEET-
INCHES.
DOTS SHOW SAMPLING
 POINT LOCATION
                       2-6
                       Li
                                                                 SAMPLING PORTS
                   2
                  JL,
                              -20-
     Figure 9.  Sampling point diagram-converter ESP outlet.
                                30

-------
Table 7.  SUMMARY OF ASARCO PARTICULATE MASS MEASUREMENTS

Date
1973
9/26
9/26
9/26
9/26
9/27
9/27
Test
1-A
1-B
2-A
2-B
3
4
Average
velocity
Test site ft/sec
Inlet to converter ESP
Outlet to converter ESP
Inlet to converter ESP
Outlet to converter ESP
Outlet to reverberatory
ESP
Outlet to reverberatory
7.95
30.3
11.1
30.0
35.2
36.0
Avg. Parti cul ate
temp. emission rate
°F %Q2 %C02 %CQ %H2 lb/hr %H2P
256 20.2 0.4 0.0 79.4
214
250 19.5 0.7 0.0 79.7
277
176 20.0 0.4 0.0 79.6
181 19.4 0.8 0.0 79.8
1040
107
810
55.9
18
268
4.82
3.44
6.85
3.87
5.92
5.92
     ESP

-------
     Converter ESP Inlet.   As  shown  in  Figure  10,  the sampling
location for the inlet to  the  converter ESP was  in an 11-1/2
foot high x 24 foot wide horizontal  duct.  About 3-4 feet
beyond the sampling ports, this duct was connected to a  long
inclined duct that led to  the  electrostatic precipitator.
This location, while not meeting Method 1  criteria, was  the
best available site for sampling.
     The three sampling ports, 6 inches ID and 2 feet  long,
provided at this site were 6,  12, and 18 feet from the  side
of the duct.  These ports  were located in a  covered area with
6-foot 1-inch headroom.  As a result, it would be necessary
to have a sectional probe for complete traversing. As  only
one sectional probe was available, the duct was sampled with
a 6-foot probe.  Further,  since a preliminary velocity traverse
indicated uniform velocity and temperature across the measurable
portions of the duct, it was decided to sample at only one
point because of manpower and time limitations.  The  point
chosen was from the center port and at the maximum depth possi-
ble with the available probes.  As the overall length of a
6-foot probe when  removed from the sample box is  7 feet, the
sampling nozzle was located 5 feet below the top of the port.
Sampling Methods
      Particulate mass sampling at the ASARCO plant generally
followed the procedures outlined in  the Federal Register.1
                             32

-------
    -12-
                       SAMPLE PORT
T
1
                               T
                                5
                                I
                     SAMPLING POINT
                       LOCATION
                                      T
                                      11-6
                                             NOTE:  ALL DIMEN-
                                             SIONS ARE IN FEET-
                                             INCHES.
                        -24-
Figure. 10.  Sampling point diagranr-converter ESP inlet.
                         33

-------
Method 2 "Determination of Stack Gas Velocity and Volumetric
Flow Rate," was employed by EPA personnel  prior to the sampling
program to obtain preliminary velocity data.  During each parti-
culate sampling run, integrated gas samples were collected in
Tedlar bags following the procedures outlined in Method 3 "Gas
Analysis for Carbon Dioxide, Excess Air, and Dry Molecular
Weight."  The collected gas sampled was analyzed for 02, CO,
COg, and Ng at the site by the Orsat technique.  Approximate
determination of the moisture in the stack gases for the pur-
pose of defining the nomograph parameters (for Method 5) was
completed at the outlet of both ESP's following Method 4
"Determination of Moisture in Stack Gases."
     The particulate emissions were determined using a modi-
fication of Method 5 "Determination of  Particulate Emissions
from Stationary Sources."  The deviations from the method were
specifically in the probe, probe connectors,  traversing point
location, and number of runs; otherwise, the equipment and
techniques  used followed Method 5  procedures  exactly.
     At the outlets of the ESP's,  a Process  Measurements  Section
  (EPA/CSL) designed probe was used for  all sample runs.  This probe
  consisted of electrically heated 1/2-inch OD Teflon tubing2
wrapped with foam  insulation  and a polyvinyl  chloride outer
protective  coating.  The  tubing was heated by a  variable
transformer and  the  temperature was monitored by a  Type  K thermo-
couple.   As  the  tubing material  is quite  flexible,  it was encased
                            34

-------
 in sections of aluminum conduit 2 feet long  and fastened
 together with conduit coupling to provide  a  16-foot-long
 probe.   The original  intent was to add sections of  conduit
 to the  probe as the traverse proceeded;  however,  this was
 inconvenient and the  probe was left assembled  in the desired
 length.
      Two 1/4-inch OD  Teflon pi tot tube lines were taped to
 the Dekron tubing and placed in the aluminum conduit.  At
 the stack end of the  probe,  the Teflon pitot tube lines were
 connected to a 0.250-inch  probe tip.   Each section  of the con-
 duit was marked to allow alignment of  the probe tip into the
 air flow.   At the sample box end of the  probe,  the  pitot line
 connectors  were provided with quick connectors  to match the
 normal sample box lines.   A  special adapter  (consisting of a
 Swagelok tube unit and a stainless  steel female ball joint)
 connected  the sample  line  to  the  sample  box.  With  the flexi-
»
 ble probe  it was  possible  to  traverse  downward and  leave the
 sample box  in  a  normal position with respect to the impingers
 and ice  bath.
     At  the  inlet  to  the converter  ESP,  the standard heated
 Pj^rex-lined  stainless steel probe, stainless steel probe tips,
 and  stainless steel pitot tubes were used.   However, as the
sample traverse was in a vertically downward direction, the
probe could  not be connected to the sample box in the normal
                           35

-------
manner.  Monsanto Research Corporation (MRC) had previously
designed and constructed a flexible line to connect the
probe to the sample box.  This connector consisted of 1/2-
inch OD wire-braid-covered Teflon tubing, wrapped with heating
tape, and overlaid with pressure-sensitive Teflon tape.  At
each end, special adapters were made to join Swagelok tube
fittings to stainless steel ball joints:  a male on one end
(for the probe) and a female on the other (for the sample
box).  A Type K thermocouple was used to monitor the tempera-
ture of the connector, the heating of which was controlled by
a variable transformer.  Other than this flexible connector,
all train components were as specified by Method 5 procedures.'
A digital thermometer and Type K thermocouple were used to moni-
tor stack temperature.  Sampling trains designed and built by
MRC were also employed at the outlet sampling sites.  All
temperature readings on these trains are obtained from thermo-
couples and a digital thermometer.
     All thermocouple and thermistor read-out devices and the
dry test meters were calibrated prior to the beginning of the
program.  The orifice plate condition factor was determined and
noted on each instrument at calibration time.  The pi tot tube
used for the inlet sampling was calibrated  (coefficients of
0.85 were used in all calculations) and found to meet Federal
Register specificationsJ
                             36

-------
Analytical Procedures
     Samples from the Method 5 sampling trains were recovered
as outlined in the Federal  Register.^   After removal  of the
filter, all exposed surfaces of the sample were washed with
reagent grade acetone or distilled water as specified.  In
addition, impinger solutions were retained for analysis.
All sample bottles and the  petri dishes for sample filters
were previously acid-soaked with 1:1  HN03 for 1 day,  rinsed
with distilled water, and soaked with distilled water for 1
day.
     Subsequent handling of the Method 5 samples also followed
the Federal Register, with  one exception:   water from both
the impingers and washing of the glassware of the train was
extracted with chloroform and ether,  and then the extracted
portion was dried to constant weight.  In addition, the water
remaining after extraction  was evaporated to dryness at 212° F
to constant weight.
     Sample weights from the Method 5 samplers were reported
as "front half" (probe washings and filter collection weights)
and "total" (front half plus water, chloroform/ether extract,
and impinger acetone washing weights).  Only "front half" values
have been used in the reporting of results in the following
sections on particulate mass and efficiency.
     All particulate mass analyses were performed at Monsanto
Research Corporation, Dayton Laboratory.
                              37

-------
Sampling Approach
     Both the inlet and outlet of the converter ESP  were
sampled simultaneously to determine the efficiency of the
devices.  Due to time limitations, two runs  .(Runs  1  and 2)
were made at each site.  Runs 1A and 2A were at the  inlet;
and Runs IB and 2B, at the outlet of the ESP.  Sampling was
continued as long as at least one converter  was in operation.
A series of lights near the sampling location indicated how
many and which of the four converters were being used.  Outlet
sampling was completed by traversing all points showing a
positive flow, but only one traverse point was sampled at  the
inlet.
     The outlet of the R&R ESP was sampled by traversing all
points in the moving air stream.  Since this duct  contained  a
considerable amount of settled dust, sampling points were
chosen to avoid this material.  The inlet to this  ESP could
not be sampled due to inadequate overhead clearance  for the
sampling probe at the port location.
RESULTS—PARTICULATE MASS
     A summary of the test sites and emission data is given  in
Table 7.  Additional data is presented in Table 8, based on
the computer printout of the Method 5 calculations.
Converter ESP Inlet
     The inlet data from the converter ESP (Runs 1A  and 2A)
was obtained by sampling at one point in the 11.5  x  24 foot
                            38

-------
                                    Table 8.  ASARCO  PARTICULATE  MASS DATA
CO
10

Description
Duration of run, min.
Barometric pressure,
in. Hg
Avg. orifice press.
drop, in. H20
Vol . dry gas (meter
con.), dcf
Avg. gas meter temp.,
°F
Vol . dry gas (std.
cond.), dscf
Total H?0 collected,
ml
Vol. H?0 vapor (std.
con.), scf
Moisture by vol., %
Mole fraction dry gas
C02, %
02, %
N?, %
1-A
162.0
30.33
0.703
77.335
83.7
76.50
81.7
3.870
4.82
0.952
0.4
20.2
79.4
1-B
162.0
30.38
0.849
76.423
77.4
76.60
57.7
2.730
3.44




2-A
176.0
30.38
0.417
65.893
85.4
65.00
100.9
4.780
6.85




2-B
168.0
30.41
1.100
76.468
63.3
78.90
67.0
3.180
3.87




3
168.0
30.30
1.Q60
97.628
84.5
96.40
128.0
6.070
5.92
0.941
0.4
20.0
79.6
4
140.0
30.30
1.1QO
82.202
75.4
82.60,
109.8
5.200
5.92
0.941
0.8
19.4
79.8

-------
                             Table 8.   ASARCO PARTICULATE MASS DATA  (Cont.)

Description
Mol . wt. of dry
gas
Mol. wt. of stack
gas
Avg. stack velocity
head, in. HeO
Stack temperature,
°F
Stack pressure
(static), in. H20
Stack pressure (ABS),
1-A
28.9

28.4

0.015
256.0
-0.12
30.32
1-B
28.9

28.5

0.286
214.0
-0.10
30.37
2-A
28.9

28.1

0.029
250.0
-0.12
30.37
2-B
28.9

28.5

0.355
277.0
-0.10
30.40
3
28.9

28.3

0.324
176.0
-2.70
30.10
4
28.9

28.3

0.346
181.0
-2.70
30.10
  in. Hg
Avg. stack gas velocity,  477.0
fpm
Stack diameter, in.       225.00
Stack area, sq in.      39800.0
Stack flow rate (dry    94100.0
  std.), dscfm
Stack flow rate        132000.0
  (actual), acfm
  1820.0
664.0
1800.0
2110.0
   143.30      225.00
 16100.0     39800.0
157000.0    129000.0

204000.0    183000.0
             143.30      230.26
           16100.0     41600.0
          141000.0    481000.0

          201000.0    610000.0
  2160.0

   230.26
 41600.0
488000.0

623000.0

-------
Table 8.  ASARCO PARTICULATE MASS DATA (Cont.)
Description
1-A
Probe tip diameter,
in.
Percent isokinetic
Parti cul ate
mg
Parti cul ate
mg
Parti cul ate
gr/dscf
Parti cul ate
gr/dscf
Parti cul ate
gr/acf
Parti cul ate
gr/acf
Parti cul ate
Ib/hr
Parti cul ate
Ib/hr
(front),
(total),
(front),
(total),
(front),
(total),
(front),
(total),
0.
109.
6428.
7138.
1.
1.
0.
1.
1040.
1160.
482
5
7
9
2900
4400
9200
0300
000
000
1-B
0.
99.
395.
1281.
0.
0.
0.
0.
107.
347.
250
0
3
6
0795
2580
0612
1990
000
000
2-A
0.
98.
6913.
7539.
1.
1.
1.
1.
1810.
1980.
384
1
9
6
6400
7900
1600
2600
000
000
2-B
0
109
237
762
0
0
0
0
55
180
.250
.0
.2
.8
.0463
.1490
.0325
.1050
.900
.000
3
0.250
101.2
482.7
587.5
0.0771
0.0939
0.0608
0.0740
318.000
387.000
4
0.250
102.4
343.3
683.6
0.0640
0.1270
0.0501
0.0993
268.000
531.000

-------
 duct.  An initial velocity traverse at depths of 3 and 5 feet
 at each of the three ports indicated a pi tot reading of 0:02
 inches of water at each point and an average temperature
 of 129° F.  As shown in Table 8, the pi tot tube readings
 during Runs 1A and 2A were 0.015 and 0.029 inches of water,
 and the temperature averaged 256 and 250° F, respectively.;
 These results yield quite different values for the velocity,
 causing differences to appear in the gas flow and particulate
 emission rates.  The results can be considered approximate:
 only one location was sampled in a large duct, and the veloci-
 ties were quite low.
 Converter ESP Outlet
     Data from the outlet of the converter ESP (Runs IB and
 2B) was consistent, except for stack temperature:   214° F in
 Run IB and 277° F in Run 2B.  The average temperature from
 Run 2B was rather unexpected:  it was higher than the inlet
 average temperature during the same time period.   A possible
 explanation is that heat from the burner at the base of the
 stack which was used to create draft in the stack increased the
 temperature in the outlet breeching.   As quite a  large portion
of the duct showed essentially a zero flow, this  heat could be
 carried back into the sampling area.   Unfortunately, no data was
 recorded on the operation of this burner.
     The stack flow rate from Runs  IB and;2B,  in  terms of both
the dry standard conditions  and the actual  cubic  feet/minute,
                             42

-------
agree quite well;  however,  the particulate weights  from these
two runs are not very similar.  Review of the production data
does not indicate  any apparent reason for the difference in
weight of material collected.   However,  the ratio of the
front half mass loadings to front half plus back half mass
loadings is quite  close, indicating that the same type of
particulate was emitted during both runs.
EXPERIMENTAL--PARTICULATE SIZE
     The size distribution of the fine participates, <_ 5 urn,
was experimentally determined at the inlet and outlet of both
ESP systems, using aerodynamic sizing techniques.  Particle
size was measured with inertial classifiers manufactured by
Monsanto (Brink Model B) and by 2000 Inc. (Andersen Stack
Sampler, Mark III).
     Due to the high particulate concentration (> 1 grain/scf)
reported prior to  the control  device, the Brink impactor was
used to determine the particle size distribution of the inlet
streams.  Figure 11 shows the sampling arrangement used at the
ASARCO smelter.  All Brink impactor measurements were made with
the impactor in the duct and vertical.  Prior sampling at non-
ferrous smelters3 had established that the size distribution
of metal condensate fumes tended to be what is classified as
fine particles; therefore, pre-scalping cyclones were not used
at this smelter.
                             43

-------
 MAGNEHELIC
GAGE, inches Hg
   TUBE-
NOZZLE
 BRINK
VACUUM
 HOSE
DRIERITE
 TUBEx
                          DUCT WALL
                      PROBE EXTENSION
NEEDLE
 VALVE
                                      DRY TEST
                                       METER
                   Figure 11.  Brink sampling train.
                                44

-------
     After preparation  of the  impactor, an appropriate nozzle
and sampling probe were selected  to  provide  isokinetic rates
at the nozzle inlet.  The inlet duct to the  Converter and to
the R&R ESP were both 16 x 24  feet with the  air  flow wing
horizontal.  With the in-stack sampler, probes were used that
placed the inlet to the impactor  4 feet from the top of the
duct.  The impactor operating  conditions  are given in Table
9.
     The Andersen Mark  III impactor  was used at  the ESP out-
lets.  The anticipated  low grain  loading  (<  0.1  gr/scf) made
the selection of a high-sampling-rate inertial sizing device
mandatory.  The sampling arrangement used is shown in Figure
12.  As required by the impactor  inlet design, all Andersen
Mark III measurements were made with the  impactor horizontal.
Extreme care was used in removing the impactor from  the duct
so that collected particulates would not  be  jarred Ibose  from
the collection stage.  (The smelter  dust  has a tendency to  coke
and optical measurements of the  impactor  stages  following  the
sampling period revealed no evidence of jarring.)
     After preparation  of the  impactor, an appropriate  nozzle
and sample probe were selected to provide isokinetic rates  at
the nozzle inlet.  With in-stack  sampling, the probes  used
placed the inlet to the impactor 4 feet from the top of the
duct.  The Andersen impactor operating conditions are  given in
Table 10.
                              45

-------
NOZZLE
                        PROBE EXT.
                               VACUUM
                                TUBE
ANDERSEN
SAMPLER
                                                            DRY TEST
                                                             METER
                  Figure 12.  Andersen sampling train.
                               46

-------
    Table 9.  BRINK IMPACTOR OPERATING CONDITIONS—ESP  INLET


Temperature, ° F
Warm-up time, min.
Sampling time, min.
AP across impactor, in. Hg
Position
Converter Flue
220
30
15
ia
Verti cal
R&R Flue
170
30
15
la
Vertical
Corresponds to flow rate of approximately 0.1  acfm.
 Table 10.   ANDERSEN MARK III  OPERATING  CONDITIONS—ESP  OUTLET

Temperature,
Warm-up time,
Sampling time
Sampling rate
Position

0 F
min.
, min. (approx.)
, acfm

Converter Flue
220
30
50
0.75-0.93
Horizontal
R&R Flue
170
30
50
0.9
Horizontal
                                47

-------
     In all cases, the samples (Brink stages  or Andersen
filters) were weighed at NERC-RTP-.   The samples were dried
for 3.hours at 80° C, desiccated for about 2  hours,  and
weighed on a Mettler H20T balance.
RESULTS—PART ICULATE SIZE
     The raw weights combined with  impactor operating para-
meters were numerically manipulated using a computer program.4
The cumulative size distributions from the data reducing  pro-
gram are given in Figures 13, 14, 15, and 16:  Figures 13 and
14 represent the inlet and outlet cumulative size data for
the converter ESP; and Figures 15 and 16 are the inlet and
outlet cumulative size data of the R&R flue.   The observed
experimental scatter is not a typical of impactor data.5
DISCUSSION
Converter ESP Tests
     Based on the data obtained from the front half of the
particulate mass loading tests, the estimated particulate mass
removal efficiencies are given in Table 11.
     Differential size distributions for each run were obtained
by differentiation of the cumulative size distributions.  The
dM/(d log D) values which are the result of this step are in
gr/ft3/ym.  The shape of the differential distributions,  plotted
in Figures 17 and 18, is of significant interest.  The numeri-
cal values, used to determine size collection efficiency, are
established by the following procedure:
                            48

-------
     90
     40
g   10


5
C/J
c/>   t

UJ   ^
or>   .

3   L
o
                  8
     10.0   8.0   6.0      4.0           2.0           1.0   0.8   0.6      0.4           0.2




                                         PARTICLE SIZE, urn



     Figure 13.  Cumulative percent of particles less than given size-converter inlet.



                                          49
0.1

-------
   99.99
8   90

g.
a>
_>


73
M

un
     50
P   10
<£

Q.
    0.01
                                      2.0
1.0     0.8    0.6         0.4      0.2
10.0 8.0   6.0      4.0




                                    PARTICLE SIZE,  ym



  Figure 14.  Cumulative percent of particles less than given size—converter outlet.



                                      50
0.1

-------
99.99
 0.01
   10.0  8.0   6.0      4.0
2.0           1.0  0.8   0.6      0.4

      PARTICLE SIZE,  pm
0.2
    Figure 15.  Cumulative percent of particles less than given size— R&R  inlet.
                                            51

-------
    99.99
 «

 o>
 _»
 CO
M
oo
«•>
LLJ
_l
O
 1
10.0  8.0   6.0      4.0           2.0           1.0   0.8   0,6     0.4

                                PARTICLE SIZE,  pm

   Figure 16.  Cumulative percent of particles less than given size-R&R outlet.

                                     52
                                                                                               0.1

-------
    10
 ° 1.0
CO
oe
M


_l

-------
     1.  The mean value curves in Figures  17 and 18 are
         transcribed to a single plot (Fi-gure 19).
     2.  Values of dM/d log D are interpolated from these
         curves at preselected geometric mean diameters.
     3.  The collection efficiency is determined at each
         diameter by:
         P   (dM/d log D) in. - (dM/d log  D) out,   x 100%
         L "           (dM/d log D) in.
     The collection efficiencies calculated in step 3 are
presented in Figure 20 as a function of geometric mean diam-
eter.  Several apparent features of the resultant curve are:
     1.  The collection efficiency for particles larger than
         3 ym exceeds 90 percent.
     2.  A monotonic decrease in the collection efficiency is
         observed for particulate between  3 ym and about  1.2
         ym.
     3.  A plateau in the collection efficiency occurs in the
         size range between 0.8 and 1.2 ym.
     4.  A strong decrease and negative efficiencies (the
         effluent from the control device  contains a greater
         mass of particulate in a given size range than the
         inlet to the control device) occur through the
         remaining size cuts.
     The first two features of this curve  have been observed  in
field studies of very high efficiency ESP  performance at  coal-
fired power plants:  an ESP model has been developed for  coal
                             54

-------
    10
°  1.0
o
i
M
    0.1
   0.01
                               ESP INLET
                                   ESP OUTLET
                                                   90
                                                   80
                                                _  70
                                                8
                                               >: 60
                                               o
                                               C 50
                                                  40
                                                  30
                                                  20
                                                  10
     0.1                  1.0

       GEOMETRIC MEAN PARTICLE DIAMETER,  ym
                                              10
 Figure 19.  Differential size distribution—con-
 verter ESP inlet and outlet.

                                             55
 0.5     1.0    1.5     2.0    2.5

PARTICLE GEOMETRIC DIAMETER,  urn
3.0
                                                    Figure 20.  Converter ESP efficiency as a
                                                    function of particle geometric diameter.

-------
flyash which predicts a shallow minimum between 0.1 and 1  ym.
Therefore, the observed decrease in efficiency as a function
of size was expected, but the degree of particulate penetra-
tion of the control device at the minimum was unexpected.
Possible explanations for the depth of the minimum in the
fractional efficiency curve are the relatively low overall
efficiency of the ESP and the high resistivity of metal oxide
fumes which leads to low precipitation rates.  Table 12 lists
representative precipitation rates for various applications:
the smelter dust precipitation rate is about a factor of 7
lower than that for utility flyash.  The lowering of the pre-
cipitation rate could amplify the magnitude of the minimum in
the fractional efficiency curves.
     While the preceding explanation would explain a minimum
in the fractional efficiency curve, it would not satisfactorily
explain the negative efficiencies observed.  Probable causes
for this feature of the curve include:
     1.  Sampling locations were not representative of the dust
         loading distribution.
     2.  Agglomerated particulate broke up impaction.  This
         would have occurred only at the outlet to produce the
         results observed.
     3.  Condensation and/or reaction of a flue gas component
         with the collection substrate.
                             56

-------
Table 11.  CONVERTER PARTICULATE MASS COLLECTION  EFFICIENCY

Run
1
2
Inlet, gr/scf
1.29
1.64
Outlet, gr/scf
0.0795
0.0463
Efficiency, %
93.8
97.1
   Table 12.   REPRESENTATIVE  PRECIPITATION  RATES

Application
Utility flyash
Pulp and paper
Sulfuric acid
Cement (wet)
Smelter
Open hearth
Cupola
Blast furnace
Precipitation rate
avg ft/sec
0.43
0.25
0.24
0.35
0.06
0.16
0.10
0.36
                             57

-------
      4.   Re-entrainment and/or evaporation of collected
          particulates  due  to electric sparkover.
      5.   Back  corona churning of the collected dust.
      Causes  1  and  2 are the least probable for these observa-
tions, since they  would require that (in 12 tests conducted
over  a 2-day period) the same errors and/or conditions were
encountered  and  repeatable.  There is some credibility to
cause 3,  based on  the  other test results observed; however,
Southern  Research  Institute4 has reported this same type of
efficiency behavior at an  ESP installation when the particu-
late  is a dry fluffy material.  This observation leaves items
4 and 5 as the most probable causes.
R&R ESP Tests
     Since inlet mass  data could not be collected, particulate
mass removal efficiencies  could not be estimated.  Emission
values were reported earlier in the section, under Results--
Parti cul ate Mass.
     The fractional efficiency of the in-series R&R ESP's
was determined using the procedures described previously.
Figures 21 and 22  represent the results of numerically differen-
tiating the cumulative size distributions;  Figure 23 represents
the fractional  collection efficiency of the two ESP's as a
function of size.
                          58

-------
    10
 on

cf
 oa
   1.0
m
cc
oo
a
LU
M
to


I  0.1
UJ
cc
   0.01
                                 .8
   10
   1.0
CO
or
te
M
oo
   0.1
     0.1                  1.0    .               10

       GEOMETRIC MEAN PARTICLE DIAMETER,  ym
 0.01
              o DISCARDED
                                                                       DISCARDED o
                                                              DISCARDEDO.
    0.1                  1.0                  10

       GEOMETRIC MEAN PARTICLE DIAMETER,  y-m
Figure 21. Differential size distribution-R&R   Figure 22. Differential size distribution-R&R
ESP inlet.                                      ESP outlet.
                                             59

-------
     10
      50
>-
o
o

u.
o

f-
o
o
o
     -50
    -100
    -150
   -200
                              1                       2                       3



                              GEOMETRIC MEAN PARTICLE DIAMETER,  ym


                 Figure 23.  Fractional collection efficiency of R&R ESP's.
                                         60

-------
      The  shape  and  the  level of control achieved by the R&R
 ESP  system  in general exhibit the same features as the frac-
 tional  efficiency curve for the converter ESP:  a notable
 exception is an  upturn  in the curve at particle diameters
 less  than 0.7 urn.   It is felt that the same causes/effect
 relationships exist for the R&R ESP as for the converter ESP.
 This  feeling is  reinforced by the similarity between the two
 fractional efficiency curves.
 REFERENCES
 1.  Federal Register, 36, 247, December 23, 1971.
 2.  Tubing is Dekron, Samuel Moore Co., Aurora, Ohio.
 3.  Harris, D. B., R. M. Statnick, D. C. Drehmel, and
    D. K. Oestreich.  Measurement of Air Pollutants from
    Selected Non-Ferrous Smelting Processes, unpublished.
4.  Southern Research Institute computer program, modified
    by F. Briden (EPA/CSL/PMS),
5.  Field Measurements of Particle Size Distribution with
    Inertial Sizing Devices, Southern Research Institute,
    NTIS No. PB  226-292/AS.   October 1973.
6.  An Electrostatic Precipitator Systems  Study;  A Manual  of
    of Electrostatic Precipitator Technology:   Part I,  Fundamentals,
    Southern Research  Institute.   NTIS No.  PB  196-380,  August
    1970.
                             61

-------
                        SECTION V
                  ELEMENTAL COMPOSITION

EXPERIMENTAL
     The total participate train catch (front and back half
residues of the EPA participate mass sampling train)  was
submitted for chemical analysis of selected elements.  The
samples were brought into solution by adding 10 ml of aqua
regia to each of the following:  residue from acetone washings
of exposed surfaces of all samples; residue from acetone
washing of the Greenburg-Smith impingers; residue from eva-
poration of the water in the Greenburg-Smith impingers; and
residue from the chloroform/ether extract.  The fiber glass
filters were placed in separate beakers and also treated with
10 ml of aqua regia.  All samples were digested at room
temperature for 72 hours and the supernate liquors from each
test run were combined into a single sample.
     Samples IB, 2B (converter ESP outlets), 3, and 4 (R&R
ESP outlets) were free of undissolved solids after the initial
treatment with aqua regia.  The combined supernate liquors
from these samples were evaporated to dryness.  The residue
was redissolved in 5 ml of aqua regia, transferred to a 25 ml
volumetric flask, and diluted to the mark with deionized water.
     The samples from Runs 1A and 2A (converter ESP inlets)
contained undissolved solids after initial treatment with aqua
                              62

-------
 regia.  The liquors were heated to 90° C for 2 hours and cooled.
 Any remaining residue was filtered.   The filtrate was evaporated
 to dryness and the residue redissolved in aqua regia.  The sam-
 ples were transferred to 250 ml volumetric flasks and diluted
 to the mark with deionized water.
      All  analyses were carried out on a Perkin-Elmer Model  403
 Atomic Absorption Spectrophotometer,  calibrated using commer-
 cially available standards.   Sample  concentrations  were derived
 from calibration data,  fitted to a linear regression least
 squares program.   The following wavelengths  were used from  the
 appropriate hollow cathode lamp:   arsenic (193.7 nm), cadmium
 (228.8 nm), chromium (357.9  nm), copper (324.7 nm),  mercury
 (253.6 nm), lead  (283.3  nm),  and zinc (213.9 nm).   All  analyses,
 except for mercury,  were performed by direct aspiration into  an
 air/acetylene  flame.  The mercury  analyses were  performed by
 flame!ess  atomic  absorption  proceduresJ
      In addition  to  the  analysis of mercury  in  the  particulate
 samples, gas-phase mercury was  determined using  acid  perman-
 ganate absorption preceded by a sodium carbonate prescrubber
 for S02 removal.  This procedure,  fully described elsewhere,2
 is given in Appendix A.
 RESULTS
     The results of the elemental analyses were expressed as
weight percent of the total material  analyzed and are given
 in Table 13.  The vapor-phase mercury concentrations are given
 in Table 14.
                            63

-------
            Table  13.   CHEMICAL COMPOSITION OF SOLIDS AND RESIDUES FROM THE PARTICIPATE MASS
                                     SAMPLING TRAIN, % BY WEIGHT9
05



Element

As
Cd
Crb
Cu
Hgb
Pb
Zn



1A

17.22+0.29
1.27+0.08
81.7+10.0
0.88+0.009
23.4+1.28
8.39+0.21
7.65+0.13



2A

16.16+0.87
0.96+0.005
60.6+9.8
0.598+0
13.2+4.6
9.04+0.08
4.69+0.487



IB

3.83+0.27
0.17+0.003
71.5+2.5
0.144+0.004
18.6+1.6
4.73+0.07
0.883+0.05


Run No.
2B

4.31+0.16
0.21+0.004
65.2+3.5
0.15+0
5.49+0.35
6.69+0.1
0.585+0.03



3

7.09+1.03
0.062+0.004
228.7+8.4
1.22+0
--
1.77+0.09
4.28+0.25



4

12.76+0.96
0.092+0.004
73.3+11.8
0.79+0.007
—
2.67+0.05
2.04+0.45
       aBased on weight of elements  in the zero oxidation state, m°.
       bln ppm  (rather than % by weight).

-------
Table 14.   ASARCO CONCENTRATION  OF VAPOR-PHASE MERCURY
                      IN FLUE  GAS

Sample
No.
5
6
7
11
12
13
aFormulas
Total
mol
Lit
Date &
location
9/25/73
Converter inlet
9/25/73
Converter inlet
9/25/73
Converter inlet
9/26/73
R&R outlet
9/26/73
R&R outlet
9/26/73
R&R outlet
used for calculations
^9 v 99 A - li+ave n-f
pr<»
SLi 	 ~n 	 = nnm
Volume Total Hg
liters ug
184.7 2.5
103.7 5.0
221.2 0.3
269.9 4.5
285.9 17.5
319.4 17.4
of ppm:
sample
ppma
0.0002
0.0005
0.0000
0.0002
0.0007
0.0007

                              65

-------
DISCUSSION
     Combining the percent composition data with the participate
loading, a collection efficiency for the converter ESP can be
calculated for each element.  The results of such a calcula-
tion are given in Table 15, where:
F-F-FI •:   w _ (gy/scf) in. x % comp. in. - (gr/scf) out, x % comp.  out.
trnciency                      (gr/scf) in7 x % comp. in.
     The collection efficiency of each individual element can
be compared with the particulate mass collection efficiency,
when this elemental collection efficiency is observed.  The
implication that the elemental composition of the emitted parti-
culate is homogeneous over all size ranges is supported by the
fractional size collection efficiency curves.
     The quantity of each element emitted to the atmosphere
from the two flues is given in Table 16.  All values are expressed
as the element in the zero oxidation state.
     The total emissions of all elements analyzed, except cad-
mium and lead, can be reduced significantly by concentrating
the control effort on the R&R flue.  Reducing the R&R effluent
grain loading from 0.055 to 0.0275 gr/scf would reduce arsenic
emissions from 58.05 to 34.2 Ib/hr (about 40 percent); whereas,
the same 50 percent reduction of the converter effluent grain
loading would reduce arsenic emissions from 58.05 to only 52.8
Ib/hr (about 10 percent).  For lead, the same reduction in R&R
grain loading would reduce lead emissions from 24.6 to 19.3
Ib/hr, or about 25 percent.  Complete control of R&R emissions
will result in a maximum reduction of 40-50 percent.
                              66

-------
Table 15.  ELEMENTAL COLLECTION  EFFICIENCIES OF
                  CONVERTER ESP

Element
As
Pb
Cd
Zn
Cr
Cu
Collection efficiency, %
(average of two runs)
96
90
97
98
95
97
Table 16. .MASS EMISSION RATE  OF SELECTED  ELEMENTS

Element
As
Pb
Cd
Zn
Cr
Cu
Total
Total
Ib/hr
58.05
24.65
1.34
15.7
0.065
4.825
104.63
Converter
Ib/hr Percent
10.5
14.2
0.97
2.05
0.005
0.375
28.1
18
58
72
13
8
8
27
R&R
Ib/hr
47.55
10.45
0.375
13.65
0.06
4.45
76.535
Percent
82
42
28
87
92
92
73
                         67

-------
REFERENCES
1.  Hatch, W. R. and W. L. Ott.   Anal.  Chem.,  40_,  2085
    (1968).
2.  Statnick, R. M., D. K. Oestreich, and R. Steiber.
    Sampling and Analysis of Mercury Vapor in  Industrial
    Streams Containing Sulfur Dioxide.   Presented  at ACS
    annual meeting, Chicago.  August 26-31, 1973.
                             68

-------
                           Appendix A

                SAMPLING AND ANALYSIS  OF MERCURY

           VAPOR IN INDUSTRIAL STREAMS CONTAINING SULFUR

                            DIOXIDE
                              by
                        R.  M.  Statnick
                        D.  K.  Oestreich
                        R.  Steiber
                        Research Branch
                   Control  Systems  Laboratory
For Presentation at 1973 ACS Annual  Meeting,  Chicago,  Illinois,
                      August 26-31,  1973
               U.  S.  Environmental  Protection Agency
                Office  of Research  and  Development
              National  Environmental  Research Center
           Research Triangle  Park,  North  Carolina   27711
                              A-l

-------
 Introduction
     The measurement of  the  level of mercury contained in off-gases
 resulting from various high  temperature industrial processes often
 requires the collection  of samples from streams high in sulfur
 dioxide concentration.   For  example, waste gas streams from non-
 ferrous smelters and fossil  fuel combustion contain sulfur dioxide
 in concentration ranges  several orders of magnitude greater than the
 concentration of mercury.  Mercury most often occurs in these gas
 streams as elemental mercury vapor because most inorganic mercury
 compounds are thermodynamically unstable at high temperatures with
 respect to elemental mercury.  At 25°C the relative nobility of
 mercury metal is indicated by the standard oxidation potentials:
     2Hg° -»• HG2 "** + 2e~                E° = -0.79 volts1
     Hg° + Hg * + 2e~                   E° = -0.85 volts1
     It can be seen from the potentials that elemental mercury is
 thermodynamically stable with regard to oxidation at lower temperatures.
     Even after considerable adiabatic cooling has taken place in the
waste gas streams the mercury formed from high temperature reactions
will  be present as the vapor since the gas phase concentrations based
on equilibrium vapor pressure considerations far exceed concentrations
actually found in typical source waste gas streams.  The variation of
 the equilibrium vapor pressure of mercury with temperature is given in
'  'W. M. Latimer and J. H. Hildebrand, "Reference Book of Inorganic
   Chemistry," 3rd ed, Macmillan Company, New York, N. Y., 1964, p 529.
                                  A-2

-------
Figure I.2
     The chemical inertness of mercury vapor, combined with its
high vapor pressure, implies that it might best be sampled by col-
lection in a wet scrubber containing an oxidizing medium.  Numerous
                                                3 4
oxidizing substances or mixtures have been used: '   the most common
oxidant is acidic potassium permanganate (Couple E° = -1.695) ,
closely followed by iodine-monochloride solution (Couple E° = -1.19).
     The high chemical reactivity of sulfur dioxide, combined with
its presence in industrial gas streams in concentrations which are at
least 3 orders of magnitude greater than the mercury concentration,
interferes seriously with any oxidizing system for the collection
of mercury.  In the case of acid permanganate, sulfur dioxide reduces
the permanganate ion and precipitates manganese dioxide; in the case
of iodine monochloride, elemental iodine is precipitated by sulfur
dioxide.
     This paper describes the development of a modification to existing
mercury sampling procedures which makes it possible to utilize these
procedures on sulfur-dioxide-containing waste gas streams.
Experimental  Methods
                  Mercury Collection Efficiency Tests
     Of the collection media  which have been  used for mercury,  the nitric
(2)
   Q.  R.  Stahl, "Preliminary Air Pollution  Survey  of  Mercury  and  Its
   Compounds," NAPCA,  NTIS  No.  PB 188-074,  p  74  (1969).
^Federal  Register, 36,  23245,  1971.
(4)
  'C.  E.  Billings  and  W.  R.  Matson, Science,  176 ,  1232  (1972).
                             A-3

-------
    105
5   5
o

o
o

DC.
=3
O
    103
    102
        -   I       I       I        I       I       I       I       I
       -   /
             -30     -20     -10
              0      10      20      30     40     50      60

                     TEMPERATURE,°C

Figure 1.  Saturation concentration  of mercury in air.

                         A-4
70       80

-------
acid/potassium permanganate solution was selected because of ease
of preparation in the field and the general  availability of reagents.
To determine mercury collection efficiency,  a series of laboratory
tests was conducted with 2.0 yg total mercury present.   A standard
mercury solution of 1000 yg Hg/ml was purchased from Harleco Division
                                 5
of American Hospital Supply Corp.  and was used as the primary
standard for this work.  A standard solution of 1 yg/ml mercury was
prepared by appropriate dilution of the primary standard.  Two ml
aliquots of the 1 yg/ml stock solution were  used to prepare the
charges for the mercury reduction-evolution  apparatus.   The procedure
of Hatch and Ott  was used to generate elemental mercury vapor.  The
experimental apparatus is described in Figure 2.
     A 2 ml aliquot of the appropriate stock solution was transferred
to the evolution flask and appropriate amounts of nitric acid,
stannous chloride, and hydroxyl amine hydrochloride were added.  The
system was then quickly closed and the bellows pump operated for 10
minutes.  Previous tests with this system, operated in an open loop
manner, indicated that all the mercury was evolved in 2 to 3 minutes.
     The resultant mercury-laden collection  media were placed in
100 ml volumetric flasks and diluted with distilled, de-ionized
water to the mark.  One to four milliliter aliquots of this sample
were then analyzed utilizing the atomic absorption procedure of
^1740 Ridge Avenue, Evanston, 111. 60201
(6)
   W. R. Hatch and W. L. Ott, Analytical Chemistry. 40, 2085-2087
   (1968).
                             A-5

-------
>
OS
                MERCURY EVOLUTION
                      FLASK
                (500 ml ERLENMEYER)
                                                                               LAMP SULFUR ABSORBERS
                                                                                  (30 ml K2Wn04/HN03
                                                                                   SOLUTION EACH)
                                  BELLOWS PUMP
                                           Figure 2.  Material recovery test apparatus.

-------
 Hatch and Ott.   The  results are reported  in Table 1.  Similar
 recoveries were  observed  by the Occupational Safety and Health
 Administration.

                      Table 1.  MERCURY RECOVERY
                 mercury found in Lamp Sulfur Absorbers)
Test Run
No.
1
2
, 3
Total ug Hg
Present
2.0
2.0
2.0
Absorber
No. 1 h
1.6
1.7
1.6
Absorber
No. 2
ND
NO
ND
Absorber
No. 3
ND
ND
ND
rj
% Recovery1
80
85
80
ND = not detected

   Adaptation of Procedure to Sulfur Dioxide Containing Streams
     In order to make the sampling procedure for mercury compatible
with streams containing 0.1 to 8% v/v of sulfur dioxide, a procedure
was sought for selectively and quantitatively removing all of the
sulfur dioxide present.  Experience has shown various alkaline
scrubbing media to be effective for sulfur dioxide removal.   A
            p
recent study  has shown sodium carbonate solutions to be an  exceedingly
effective sulfur dioxide removal  medium.  Based on this experience,
a prescrubber containing saturated sodium carbonate solution was
   R.  L. Larkin, OSHA,  Cincinnati,  Ohio,  personal  communication,  1973.
(8)
  ;D.  C. Draemel, EPA/OR&M/NERC-RTP/CSL,  personal  communication,  1973.
                              A-7

-------
prepared and tested for effectiveness in a synthetic flue gas stream.
The prescrubber was followed by acid/permangate scrubbers; after
sampling 0.18 m^, no visible deterioration of the acidic permanganate
was observed.  This was true even at sulfur dioxide levels approaching
90+% v/v.
     The question of the possible retention of mercury by the sodium
carbonate solution was then addressed.  A test stand, shown in Figure
3, was assembled to determine the fate of mercury in sodium carbonate
solution.
     For these tests, mercury vapor was generated by aspirating air
through a midget impinger containing several grams of elemental
mercury.  The mercury-containing stream was bypassed around the
sodium carbonate scrubber until a steady state value was observed
with the atomic absorption instrument (Perkin Elmer Model 403)
operated in the fTameless mode.  The stopcock was then turned to
direct the flow through the scrubber; when a steady state was again
                                                         9
achieved, the recording was noted.  Indirect measurements  indicated
that the mercury content of the scrubber was less than 20 ppb.  A
10% mercury loss into the prescrubber liquor would have resulted in
a mercury level of 60 ppb being observed in the scrubber liquor.
(9)
  'These results were also observed by Radian Corporation and TRW
   Systems Group personnel at a later date.
                               A-8

-------
                          SODIUM CARBONATE
                            BYPASS SYSTEM
>
CD
                           DRYING
                           AGENT
        MIDGET IMPINGER
      (ELEMENTAL Hg AT 0°C)
 BUBBLER
  (200ml
SATURATED
  Na2C03)
    FLAMELESS
ATOMIC ABSORPTION
   CELL (10 cm)
                                    PUMP
                              Figure 3.  Test stand used to determine fate of mercury.

-------
                            Sampling Procedure
     The apparatus used for sampling various gas streams for mercury
is shown in Figure 4.  Prior to the taking of a sample, a fresh
glass-wool plug was inserted in the borosilicate glass sample probe
to act as a particulate filter.  The probe was then preheated to
150°C to approximate the temperature of the gas stream sampled and
to avoid condensation in the probe.  The Lamp Sulfur Absorbers
were prepared by charging each with 30 ml of the potassium
permanganate/nitric acid solution.  The Greenberg-Smith impinger was
loaded with 200 ml of saturated sodium carbonate solution.
     After assembly of the complete sample train (as shown in Figure
3) and the initial heating of the probe, an initial dry gas meter
reading was taken.  The pump was then started and the pressure to
the dry gas meter noted.  After a suitable volume had been sampled,
the pump was shut down and the dry gas meter reading again noted.
The contents of the Lamp Sulfur Absorbers were then transferred into
prewashed (with acid) polyethylene bottles.  A sample of fresh acid/
permangate solution was also stored in a polyethylene bottle to serve
as a reagent blank.  After proper dilution and aliquoting, the samples
were analyzed, using the cold atomic absorption method of Hatch and Ott.^
      Sampling
     To ascertain the precision of the procedure described above under
field sampling conditions, a series of experiments was conducted with a
^  'Purchased from Ace Glass, Inc., 1430 N.W.  Boulevard, Vineland,
    N. J.  08360.
                              A-10

-------
GLASS PROBE
(WASHED WITH
CONC. HN03)
                BUBBLER
               '  (200ml
                 SODIUM
               CARBONATE)
   LAMP SULFUR ABSORBERS
(30 ml OF COLLECTING SOLUTION)
                                                                   PRESSURE
                                                                  TRANSDUCER
                                              Figure 4.  Mercury sampling train.

-------
coal-fired boiler.  Two sampling probes, held close to each other,

were inserted in the duct.  A gas sample was aspirated through both

sampling trains simultaneously to determine the sampling-analysis

precision.

     The results given in Table 2 indicate:  a 32% total coefficient

of variation; and a 13.7% average dispersion for duplicate analysis.


   Table 2.  MERCURY FIELD SAMPLING PROCEDURE TEST RESULTS


Test Run                    Train A                 Train B
  No.                        pg/m3                   yg/m3

   1                           31                      25
   2                           25                      20
   3                           15                      15


     Information presented in Table 3 indicates that the precision of

the mercury field sampling procedure compares favorably with those of

the typical field sampling procedures applied to other types of analyses,


   Table 3.  COEFFICIENT OF VARIATION FOR VARIOUS FIELD SAMPLING
                              PROCEDURES

                                              Coefficient of
     Procedure                                 Variation. %

Mercury                                             32
Sulfur Dioxide                                       8-20
Carbon Dioxide                                    0.97-9.311
Particulate Mass Loading                             6-60^2
   ^Walden Research Corp., Improved Chemical Methods for Sampling and

    Analysis of Gaseous Pollutant from the Combustion of Fossil Fuels.
(12)
v  'W. J. Mitchell, EPA Internal Report, A Method to Obtain Replicate
    Particulate Samples from Stationary Sources.


                               A-12

-------
      Much of the lack of precision  observed with various  procedures
 is the result of variability in  the streams being sampled as  a
 function of time, rather than of the procedures  being  used.
 Results
      Field experience using  the  new mercury sampling procedure was
 acquired by sampling  a number of different industrial  processes.
 Included in the  processes  sampled were:
      1.   A zinc  smelter roaster.
      2.   A lead  smelter blast furnace.
      3.   A coal-fired steam  generating station.
      4.   An experimental coal  gasifier.
      Chemical  analysis  of  the samples taken using the  fTameless
 atomic  absorption method gave results as indicated in  Table 4.
 Conclusions and  Summary
      A mercury sampling procedure has been developed which is
 operable  in  the  presence of  large concentrations of sulfur dioxide
 and which  is suitable for  field use.
      The  procedure employs a  sodium carbonate prescrubber for sulfur
 dioxide removal,followed by  two scrubbers containing nitric acid/
 potassium  permanganate  solution for mercury collection.  The prescrubber
may be operated  for 45 minutes at a flow rate of about 2 liters/minute
and sulfur dioxide concentration of 80,000 ppm before the solution is
exhausted.  These figures may be used to predict the useful lifetime
of thet-prescrubber under various conditions.
     The long sampling times possible provide  greater analytical
sensitivity as well  as a time-integrated sample which is less  subject
                               A-13

-------
       Table 4.  MERCURY ANALYSIS OF VARIOUS  INDUSTRIAL GAS STREAMS
    Source

Experimental
Coal Gasifier

Coal-Fired Boiler.
175-Megawatt
Zinc Smelter
Roaster
Lead Smelter
Blast Furnace
SOX Level,
ppm
30
-2500


80,000


JOG-250013

Vol ume
Sampled, 1
19.64
233.47
180.17
285.05
93.9
80.1
64.2
170
283
Hg, yg/m3
484
91.5
73.2
11.9
220
20
970
3380
2590
Total Stream Flow
Rate (Q), m3/min.
36
9000
9000
9000
1250
1250
1250
-9000
9000
(13)
    Estimated from data presented in  A.  G. McKee  report, Systems Study for

    Control of Emissions:   Primary Nonferrous  Smelting  Industry, NTIS No.

    PB 184-884, 184-885, and 184-886, June 1969.  (Vols. I-HI.)

                                A-14

-------
to short-term fluctuations in sample concentration  caused  by anomalies
in process operations than are real  time analyses done  with  instrumental
mercury monitors.  Instruments which monitor mercury directly are  also
subject to severe interference from hydrocarbons  in that they operate
in the ultraviolet spectrum.
     While the mercury collection efficiency shown  in Table  1 is not
100%, a consistent efficiency of collection was observed;  consequently,
the low bias may be compensated for with a good degree  of  confidence.
                             A-15

-------
                                  TECHNICAL REPORT DATA
                           (Please read Instructions on the reverse before completing)
 i. REPORT NO.
 EPA-650/2-74-11T
                             2.
                 3. RECIPIENT'S ACCESSION-NO.
 4. TITLE AND SUBTITLE Measurement of Sulfur Dioxide, Parti-
 culate, and Trace Elements in Copper Smelter Con-
 verter and Roaster/Re verberatory Gas Streams
                 5. REPORT DATE
                 October 1974
                 6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)

 Robert M. Statnick
                 8. PERFORMING ORGANIZATION REPORT NO
 9. PERFORMING ORSANIZATION NAME AND ADDRESS
 EPA,  Office of Research and Development
 NERC-RTP, Control Systems Laboratory
 Particulate and Chemical Processes Branch
 Research Triangle Park. NC 27711	
                 10. PROGRAM ELEMENT NO.

                 1AB012; ROAP 21ADM-012
                 11. CONTRACT/GRANT NO.
                 NA (In-House)
 12. SPONSORING AGENCY NAME AND ADDRESS
 NA
                                                        13. TYPE OF REPORT AND PERIOD COVERED
                                                        Final: 6/73-4/74    	
                                                        14. SPONSORING AGENCY CODE
 15. SUPPLEMENTARY NOTES
 is. ABSTRACT
               repOrf gives the results of the analysis of data on particulate, sulfur
 dioxide (SO2), and trace element emissions and control efficiencies for a copper
 smelter.  The SO2 emission rates from the roaster /reverberatory and converter
 effluent streams were 518 and 587 Ib/min, respectively. The acid plant's SO2 con-
 trol efficiency was 96. 8 percent.  The mass collection efficiency of the converter's
 electrostatic precipitator (ESP) for dry filterable solids was 95 percent.  The ESPs
 appear to have a minimum collection efficiency  for dry filterable solids  of 0. 8-
 1.2 jum. Analyzing emitted particulate gave the following mass emission  rates (in
 Ib/hr) for selected elements: arsenic  (58.05), lead (24.65), cadmium (1.32), zinc
 (15.7), chromium  (0.065), and copper (4.825).  Control efficiency for the analyzed
 elements was between 90 and 98 percent.
 7.
                              KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
    b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
 Air Pollution
 Copper Converters
 Smelters
 Measurement
 Sulfur Oxides
 Trace Elements
    Air Pollution Control
    Stationary Sources
    Particulates
 13B
 11F

 14B
 07B
 8. DISTRIBUTION STATEMENT
                                           19. SECURITY CLASS (ThisReport/
                                           Unclassified
                                                                    21. NO. OF PAGES
 Unlimited
    20. SECURITY CLASS (This page)
    Unclassified
                             22. PRICE
EPA F.orm 2220-1 (9-73)
A-16

-------