AIR POLLUTION CONTROL
IN THE
PRIMARY ALUMINUM INDUSTRY
VOLUME I OF II
SECTIONS 1 THROUGH 10
23 July 1973
SINGMASTER&BREYER
235 East 42nd Street
New York, N.Y. 10017
-------
0369
AIR POLLUTION CONTROL
IN THE
PRIMARY ALUMINUM INDUSTRY
VOLUME I OF II
SECTIONS 1 THROUGH 10
23 July 1973
SINGMASTER & BREYER
235 East 42nd Street
New York, N.Y. 10017
-------
0370
Acknowledgement
With the exception of source sampling tests, the work
reported herein was performed by Singmaster & Breyer,
New York, New York, under Contract No. CPA 70-21 with
the Environmental Protection Agency, Office of Air
Programs, Triangle Park, North Carolina. We gratefully
acknowledge the assistance and cooperation of EPA
Project Officers Reid E. Iverseri and Robert V. Hendriks,
and representatives of all United States primary alumi-
num producers, working through the Primary Aluminum
Industry Liaison Committee.
SINGMASTER & BREYER
John C. Russell
Dumont Rush
-------
0371
TABLE OF CONTENTS
Report in Brief
1.0 introduction
2.0 The Primary Aluminum industry
3.0 Technology of Aluminum Production
4.0 Sources and Characteristics of Effluent Releases
5.0 Technology of Emission Control
*6.0 Emission Sampling and Analytical Techniques
*7.0 Reported Industry Effectiveness and Costs
8.0 Systems Analysis of Pollution Abatement
9.0 Analysis of Control and improvement
10.0 Potential Research and Development Fields
Appendices
1A Data Acquisition Questionnaire
4A Particle Size Weight Distribution
5A Fractional Removal Efficiency Curves
6A Sampling and Analytical Technique
*6B Method 13 - Determination of Total Fluoride Emissions
7A EPA Source Sampling
*8A Emission Flow Diagrams
8B Removal Equipment Purchase Costs
9A Sample Calculation of Industry Control
Improvement Costs
* EPA Sampling information contained in these sections.
-------
7
2
REPORT IN BRIEF
A study was made of the technical and economic
aspects of the emissions and control of air pollutants
in the primary aluminum industry for the Environmental
Protection Agency under Contract No. CPA 70-21, and
was completed in late 1971. The cost and performance
data base for the study included detailed information
supplied to the contractor in confidence by the domes-
tic producers, typical performance data obtained from
equipment suppliers, and published information from
the technical literature. Engineering analysis of the
data by the contractor resulted in systems evaluation
of current industry control in terms of present costs
and performance. Systems analysis was applied to
growth projections at various control levels to esti-
mate future costs and emissions. Recommendations were
made covering the direction of research and develop-
ment efforts towards improvement in pollution abate-
ment by the industry.
The Industry
The domestic primary aluminum industry is
based on the electrolytic reduction of alumina, most
of which is either directly imported or produced from
imported bauxite. At the end of 1971 there were 30 re-
duction plants, operated by 13 producers, distributed
among the Pacific Northwest, the Gulf Coast, and the
East Coast sections of the United States. The indus-
try has grown since 1946 at an average rate of 10.2
percent, and in 1970 had reached a production rate of
nearly 4 million short tons per year. United States
capacity was some 46.6 percent of the world total.
Industry growth is expected to continue to re-
spond to the long term steady increase in demand, and
domestic primary production is expected to double or
triple by the end of the century.
-1-
-------
0373
Technology of Aluminum Production
Primary aluminum metal is produced by the
Hall-Heroult process involving the electrolytic reduc-
tion of the aluminum oxide (alumina) dissolved in a
molten electrolyte of sodium aluminum fluoride (cryo-
lite) . Carbon anodes are immersed in the bath contain-
ed in a carbon lined cell which acts as the cathode.
Approximately 7-8 kwh of direct current is
consumed per pound of aluminum metal produced. From
90 to 180 reduction cells are connected electrically
in series to form a potline, the basic production unit
of the reduction plant.
The molten aluminum metal produced has a
slightly greater specific gravity than the cell bath,
and collects as a layer on the bottom of the cell,
from which it is periodically syphoned and transfer-
red to a cast house.
As electrolysis progresses, the alumina con-
tent of the bath is decreased, and is intermittently
replenished by feed additions to maintain the content
at about 2 to 5 percent in solution in the cryolite.
When this content falls to about 1.5 to 2.0 percent,
the phenomenon of "anode effect" may occur, in which
the bath fails to wet the carbon anode and a gas film
collects under the anode. This film causes a high
electrical resistance and the normal cell voltage in-
creases 10-15 fold. Correction is obtained by addi-
tion of alumina to the bath.
Bath composition is adjusted by addition of
aluminum fluoride and cryolite as required to obtain
maximum current efficiency.
Reduction cells are of two basic types, the
prebake cells using multiple prebaked carbon blocks as
anodes, and the Soderberg cells using large single
anodes which are charged with carbon paste and contin-
uously baked in place. Prebaked anodes are replaced
periodically by new assemblies and the butts recycled
for reclamation and re-use in the anode preparation
plant.
-2-
-------
037^
Soderberg anodes are completely consumed; the
in-place baking of the anode paste results in the re-
lease of hydrocarbon fumes and volatiles derived from
the pitch binder of the paste mixture. Ihese compo-
nents of the Soderberg effluents, which are essential-
ly absent from the prebake cell gases, require modifi-
cation of the effluent treatment techniques applied to
emission control from the reduction cells.
The preparation of anode materials is usually
an ancillary operation at the reduction plant site. A
carbon plant, or "green mill", crushes and sizes coal
and coke, mixes them with pitch, and produces Soderberg
anode paste or green anodes. The latter are fired and
baked, and assembled with connectors for use in the
cells. Return prebake butts are cleaned and then re-
crushed for incorporation in the green anode mix.
Molten cell metal is transferred to a cast
house where it is fluxed to remove trace impurities
and cast into a variety of ingot forms.
Fluoride losses from the potline operations
occur as cryolite absorption into cell linings and as
evolution and dusting of fluorides from the cell sur-
faces. The latter, about three quarters of the total,
presents the major portion of the potential pollutant
emission of the primary aluminum reduction operations.
Approximately one-third of the fluorine content of the
cell effluent is in the form of particulate fluorides,
some of which are susceptible to direct mechanical
separation and recovery as dry recycled return feed.
Finer sizes of fluoride particulates present greater
removal difficulty and are a potential source of pol-
lutant emissions. Gaseous fluorides, evolved directly
from the cell or formed by hydrolysis from other flu-
orine compounds, are amenable to nearly complete emis-
sion control by chemisorption on alumina or by solution
in aqueous media.
The economics of fluorine replacement of loss-
es from the reduction cell favor maximum recycle of
fluorine in a form which can be directly added as cell
feed without the need to be reconverted to cryolite
for this purpose.
-3-
-------
Effluents
The airborne effluents from primary aluminum
reduction plant operations include dusts of carbon and
alumina from materials handling and preparation, and
particulates and gases evolved from potlines, anode
bake furnaces, and cast house. Of these, the dusts
present an in-plant problem of industrial hygiene and
housekeeping; the bake plant effluents are primarily
a smoke abatement problem and the cast house effluents,
mainly chlorides from intermittent metal fluxing oper-
ations, may not be of significance. The potline efflu-
ents are the greatest quantity and potentially the
most damaging.
Composition and quantity of potline effluent
vary within wide limits among the plants in the alumi-
num industry, being affected by type of cell installa-
tion, and potline operating conditions. The most
significant components from a pollutant viewpoint are
fluoride particulates, gaseous fluorides and nonflu-
oride particulates. Sulfur dioxide (derived from
sulfur content of the anode content) may or may not
present a pollutant problem, depending on plant loca-
tion. Effluent generation rate, per 1000 pounds of
aluminum* made in prebake potlines is of the order of:
Lb/1000 Ib Al
SO2 30
"F" as gaseous fluorides 14
F as solid fluorides 9
Total F 23
Total Solids 46
Effluents from Soderberg cells include, in addition,
volatilized hydrocarbons at the cell operating temper-
atures.
* Throughout this report, most effluent and emission
rates are expressed as pounds per 1000 pounds of
aluminum production, equivalent to kilograms per
tonne.
-4-
-------
0376
Bake plant effluents may include products of
firing combustion, burned and unhurried hydrocarbons
derived from the heating and carbonizing of the paste
binder pitch, sulfur oxides derived from the carbon
paste materials, and fluoride. The order of magnitude
of these effluents is:
Lb/1000 Ib Al
Total Solids 1-5
Hydrocarbons 0.25-0.75
Total F 0.15-0.75
Sulfur (in oxides) 0.35-1.0
Cast house effluents are largely fumes of alu-
minum chloride which, in the presence of atmospheric
moisture may hydrolyze to HCl and A12O3-
Emission Sampling
The problems of obtaining representative sam-
ples of effluent and emission streams in the primary
aluminum plant to determine the effectiveness of pol-
lution abatement are discussed. They are concerned
with the difficulty of accurately sampling very large
volumes of low velocity air flow in secondary collec-
tion systems, complicated by low particulate and gas
loadings, by non-uniform content of the oas systems,
and by water saturation of the gas stream after wet
scrubbers. More dependable sampling can be made in
the primary collection systems. Sampling techniques
to obtain isokinetic samples are discussed, as are
the problems of differentiation of solid and gaseous
fluorides due to the high reactivity of the gaseous
fluorine compounds.
Note is taken of the need to correlate partic-
ulate emission size distribution with settling veloc-
ities and conditions which affect dust-fall and ambient
air quality in peripheral areas. Sampling equipment,
including particulate separation and gas trains, are
discussed. References are given for analytical meth-
ods.
-5-
-------
03 II
Reported Industry Effectiveness
and Cost of Emission Control
Based on confidential quantitative information
furnished by the primary aluminum producers in response
to a detailed questionnaire, reported industry emission
control practice was analyzed and summarized.
The weighted average effluent rate from carbon
anode bake plants was calculated to be of the order of
214 standard cubic feet per pound of prebake plant alu-
minum capacity, with gas loadings of 0.015 grain total
gaseous fluoride and 0.085 grain total particulates per
standard cubic foot. Fluoride in particulates was re-
ported to be negligible. No emission data on anode
bake plants was reported by industry, although some 40%
of the bake plants capacity is under some sort of con-
trol, much of it experimental.
The weighted average emission control data re-
ported by the industry for all types of potlines which
exercise some control are shown in the following tabu-
lation. While the responses represented some 93% of
the industry tonnage, not all provided breakdowns 'of
gaseous and particulate fluoride data, and the weighted
averages are regarded as representing good orders of
magnitude rather than exact data.
-6-
-------
• 0378
Lb/1000 Ib Aluminum Produced
Total Solid Total
Solids "F" HF_ "F"
Cell Effluent 47.7 8.8 13.1 22.5
Primary Collection
System 40.3 7.5 11.7 19.3
Secondary Collection
System 6.9 1.1 1.2 2.3
Primary System
Emission 5.9 1.6 0.9 2.3
Secondary System
Emission 6.4 1.1 1.2 2.3
Total Emission 12.3 3.0 2.1 5.1
Overall Emission
Control Efficiency* 73% 66% 84% 77%
* Based on reported data from most of the controlled
segment of the United States aluminum industry.
The emission from the domestic primary aluminum
industry on a 1970 production of 4.0 million tons of
metal are estimated to have been as shown below:
Tons Emissions (1970)
Potrooms Bake Plants Total
Total Fluorine 23,200 650 23,800
Gaseous Fluorides 10,200 600 10,800
Fluorine in Particulates 13,000 50 13,000
Total Solids 53,000 4,200 57,200
Ninety-seven percent of the potline tonnage
reported some degree of emission control on their
effluents.
-7-
-------
03
7Q
The reported costs of emission control, after
adjustment to 1970 dollars and extrapolation to include
the 12 percent of controlled tonnage not reported, pro-
vided data for an estimate that the total capital in-
vestment for installed pollution control for the 1970
annual production capacity of about 4.1 million tons
is about $240,000,000, of which $236,000,000 is in pot-
room pollution control.
It is estimated from the reported data that the
total industry annual operating costs for pollution
abatement, adjusted to include interest, taxes, insur-
ance and depreciation amount to about $65,000,000 per
year.
From the reported data the United States pri-
mary aluminum industry, as a whole, has an investment
of some $58 per annual ton of capacity in pollution
control installation and spends a net of about $16 per
ton of aluminum produced to control its emissions to
the level of 6.0 pounds of total fluorine (2.7 pounds
in gaseous form, 3.3 pounds in particulates) and 14.4
pounds total solids per 1000 pounds of aluminum pro-
duced.
Systems Analysis of Pollution
Abatement Control
To provide a tool by which the effects of vari-
ations in industry-wide levels of emission control
could be evaluated in terms of resultant needs for mod-
ification of control schemes and consequent costs, plant
effluent control models were constructed representative
of the combinations of reduction cell types, collection
schemes, and emission control devices of the potline
units now operating in the industry. These plant mod-
els could then be combined by proportional tonnage to
obtain an overall model representative of industry per-
formance and costs.
Typical effluent and collection parameters were
established from reported industry data as a basis for
the determination of plant model performances. Cumula-
tive removal efficiencies of individual control devices
used in combination determined the effectiveness of
-8-
-------
0380
control schemes, and these, applied to the effluent and
collection parameters, established the overall control
efficiencies and thus the emissions.
Capital and operating costs of the model con-
trol schemes were calculated by summation of the costs
of the elements, including collection systems, pollu-
tant removal equipment stages for primary and secondary
effluent treatment, and cost of scrubber water treat-
ment for recycle.
Equipment purchase costs were estimated from
supplier information. Operating costs included cred-
its, where appropriate, for recovered alumina and flu-
orine values.
Cost effectiveness comparisons were made among
model control schemes applicable to collection scheme
alternatives used for each type of potline, and rela-
tionships were established between overall control ef-
ficiency and control cost- per ton of aluminum. These
comparisons illustrate the strong influence of collec-
tion or hooding efficiency on overall control efficien-
cy as well as the orders of magnitude differences of .
costs associated with the various control efficiency
levels for the different potline types.
Industry Control Improvement
Analysis by Models
By operating on the individual plant control
models which are combined in proportion to the respec-
tive tonnages representative of industry practice,
systems analysis of the overall industry may be carried
out. The degree of industry control can be analyzed
and costed, and the effects of improvements in individ-
ual process segments of the industry can be evaluated.
Current industry control practice was analyzed
in detail through use of the models by cell type, efflu-
ent collection systems, and finally by emission control
schemes. Comparisons were developed to show the rela-
tionships between percent total aluminum capacity of
each segment of the industry and the contributions to
total fluoride emission inventory, relative control
efficiency, and fluoride emission rate per ton aluminum
produced.
-9-
-------
0381
The base case was used to project the effec-
tiveness of improving the present level of industry
overall control efficiency by selectively upgrading
individual model schemes to others of higher perfor-
mance. Unit cost increments involved were applied to
the capacity tonnages represented by the model modi-
fication, and the resulting cost and emission model
of the total industry was restructured.
For these projections, improvements from the
present estimated 74.3 percent overall control effi-
ciency were analyzed to estimate performances and
costs to achieve the following four levels of improved
performance:
All plants to achieve at least 80 per-
cent overall control efficiency
Apply best demonstrated primary control
technology in all plants
Raise all plants to at least 90 percent
overall control efficiency
Apply the best demonstrated technology,
both primary and secondary, to all
plants
The result of the systems analysis, summarized
below, illustrates the sharply rising costs involved
as higher levels of emission control are achieved by
the industry.
A base case, adjusted to capacity constructed
at the beginning of 1971, was developed to correlate
the results of the systems analysis technique with the
adjusted estimates derived from industry reporting
through 1969. Good correlation was obtained within
the accuracy of the original data, as noted below:
-10-
-------
Estimated From
Reported Data
1969 Capy.
Systems
Analysis
1971 Capy.
Aluminum, MM Tons
4.0
4.6
Emissions, Total Fluorides
Annual Tons F 23,200
Rate, Ib F/1000 Ib Al 5.8
27,600
5.9
Overall Control
Efficiency
74.6
74.3
Capital Investment,
$MM Total
$236l/
$236
Annual Control Cost,
$MM
$6527
$61.43/
_!/ Reported direct investment $182 MM.
_2/ Reported direct cost $11 MM without capital charges
or credits.
3/ Includes assigned credits.
-11-
#382
-------
Industry Overall Control
Efficiency
Emission, Total Fluoride
Rate, Ib F/1000 Ib Al
Annual Tons
Total Control Investment
$MM (1970)
Annual Operating Cost
$MM/Yr (1970)
Cost Effectiveness of
at 1971
1971
Model
mtrol
74.3
oride
> Al 5.9
27,500
tment
$236
>st
$ 61
Industry
Capacity
Control Improvement
Level
Control Level
Min. 80% Best Prim. Min. 90%
Efficiency Control Efficiency
84.0
3.7
17,100
$473
$119
85.6 90.9
3.3 2.1
15,300 9,800
$494 $713
$116 $201
O
CO
CO
CO
Best
Demonstrated
Technology
92.5
1.7
8,000
$813
$200
-------
038^
Future costs of pollution control have been
estimated, premised on forecasts of industry capacity
growth and on two assumptions for required control
level: a) that new capacity will be required to con-
trol emissions to the level of best demonstrated tech-
nology, and b) that new capacity will be required to
apply best demonstrated technology to the primary cell
collection stream.
The median of growth forecasts predict total
capacity increase to some 22.5 million short tons by
the year 2000, with a range in variation of some 30
percent, plus and minus. Cumulative investment and
operating costs, will, as capacity replacement occurs,
approach those established for the limiting cases of
best demonstrated technology, whatever control stand-
ards may be adopted by existing plants.
It is estimated that, under these assumptions,
the projected industry performance and costs at the
year 2000 will be as indicated in the following table:
Best
Demonstrated Best
Primary Demonstrated
Control Technology
Aluminum Capacity, MM Tons 22.5 22.5
F Effluent, 1000 Tons 518 518
Overall Control Efficiency,% 94.6 96.5
F Emission, 1000 Tons 28.0 18.1
Total Capital Investment,
MM 1970$ 1350 2320
Unit Capital, $/Ton 60 103
Total. Annual Cost, MM 1970$ 202 553
Unit Annualized Cost, $/Ton 9 25
-13-
-------
0385
Potential Research and Development
Fields for Pollution Abatement
The costs of reaching the presently achievable
limit of emission control by best demonstrated technol-
ogy are high, and the thrust of needed research and
development work is towards the reduction of these
costs, rather than in closing a technological gap in
emission control.
This effort could well be directed towards im-
provements in cell operations leading to the reduction
in the amount of effluents produced by the Hall-Heroult
electrolysis process, improvement in the performance of
collection and removal equipment which now falls below
that of the best systems, and investigation of basic
electrochemical inter-relationships affecting fluoride
evolution in the Hall-Heroult process to minimize cell
effluents.
-14-
-------
03 as
Table of Contents
Section 1
1.0 Introduction
1.1 Objective of the Study
1.2 Procedures for the Study
1.3 Data Base
1.3.1 Technical Literature Sources
1.3.2 industry Questionnaire
1.3.3 Industry Contacts
1.3.4 Equipment Suppliers
-------
0387
1.0 Introduction
The Environmental Protection Agency of the United
States Government entered into contract with Singmaster
& Breyer, New York, (Contract No. CPA 70-21) to engage
in and report on a study embracing technical and eco-
nomical aspects of the emission and control of air pol-
lutants in the primary aluminum smelting industry. This
study, both a survey of the state of the art and a sys-
tems analysis of measures to improve pollution abatement,
enhances a general understanding of emissions control in
the industry and provides guidance in the establishment
of control standards.
1.1 Objective of the Study
This study was designed to:
a) Establish a clear understanding of the
technical and economic aspects of air
pollution control in the primary aluminum
smelting industry of the United States,
b) Determine the potential for improving air
pollution abatement using existing tech-
nology,
c) Estimate present and future costs of
control, and,
d) Define areas of investigation which can
be benefitted by accelerated research and
development.
The present report of this investigation is written
with the cooperation of the United States Government
Environmental Protection Agency, Office of Air Programs,
the United States and Canadian aluminum producers, and
manufacturers of equipment useful in the control of alu-
minum smelter emissions.
1-1
-------
'.0388
1.2 Procedures for the Study
The first phase of the project was concerned es-
sentially with obtaining, compiling, and presenting
statistical and technical information about the industry
and about the pollution emission control practices in it.
Information was gathered on the locations, types and ca-
pacities of all aluminum reduction plants in the United
States. Cooperation among the producers was obtained in
supplying available control information in these plants
to assist in evaluation of industry practice by erection
of models representative of industry segments.
Projections of historical and statistical data were
made in an attempt to present a picture of future growth,
and technological modifications.
A second phase of the study involved the analysis
and reduction of this information to forms which were
used in a systems analysis of emissions control technol-
ogy in the industry. By appropriate modeling based on
confirmed flow diagrams, the cost and effectiveness of
emission control were evaluated for individual sources
and for the entire reduction process. New foreign and
domestic developments were considered in addition to
current United States practices.
Based upon all of the preceding work, recommenda-
tions were made as to useful research and development
efforts. The recommendations fall into the following
general areas of investigation:
a) Improvement in effluent hooding or collection
systems,
b) Reduction of effluents from the Hall-Heroult
alumina reduction process,
c) Improvement in the performance of existing
control equipment,
d) Fundamental research in pollution abatement
technology,
1-2
-------
^•389
A final part of the work was an economic study of
emission control in the primary aluminum industry. A
model was developed which utilizes the information gath-
ered in previous phases to make possible estimates of the
capital costs and annual operating and maintenance costs
of pollution control equipment now in operation in the
industry. A projection model was developed to provide
the capability of projecting expenditures for achieving
desired higher levels of emission control for a reason-
able time into the future.
1. 3 Data Base
Data in five related areas of knowledge were col-
lected through three principal activities. The five data
categories are:
a) Production Statistics - Determination of past,
present, and probable future aluminum produc-
tion at all present and projected smelter sites
in the United States.
b) Process Technology - Description of the tech-
nology of aluminum smelting and of the asso-
ciated effluents production and control.
c) Control Equipment - Description and evaluation
of equipment and techniques available for emis-
sion control in the aluminum industry.
d) Air Pollution - Determination of quantitative
and qualitative aspects of emissions attrib-
utable to aluminum smelters.
e) Projections - Construction of models for future
aluminum production and the needs for emission
control.
A data base in these five areas was developed
through three coordinated activities:
a) Literature Search - Extraction from the techni-
cal literature of data pertaining to the han-
dling of fluorides and to all emission control
in the primary aluminum industry.
1-3
-------
0390
b) Smelter Industry Questionnaire - Collection
of data from primary aluminum producers
through a comprehensive formal questionnaire.
c) Contact and Communicatiori - Personal and cor-
respondence contacts with aluminum producers
and pollution control equipment manufacturers,
to enlarge the data base and to improve under-
standing of the problems and technologies of
emission control in the primary aluminum in-
dustry.
, These data were correlated and analyzed in prep-
aration for Phase II, the evaluation of current and
future cost effectiveness and the identification and
analysis of research and development projects in the
field of emission control technology.
1.3.1 Technical Literature Sources
Searches through the technical literature dis-
closed several hundred references which relate to alumi-
num smelting and the control of associated pollutant
emissions. Of particular value in these searches were
several index and abstracting services, especially:
Engineering Index
Aluminum Abstracts
Air Pollution Control Association Abstracts
Chemical Abstracts
Air Pollution Technical Information Center
of the Environmental Protection Agency
In addition, an earlier study for Environmental
Protection Agency, "Control Techniques for Fluoride Air
Pollutants", provided more than one hundred useful
reference articles.
1.3.2 Industry Questionnaire
Through the assistance and cooperation of the
Primary Aluminum Industry Liaison Committee representing
all aluminum producing companies in the United States, a
comprehensive questionnaire was developed which sought to
bring together all available plant data on aluminum smelt-
ing effluents and their control. Appendix 1-A is a
1-4
-------
0331
condensed example of this questionnaire. Responses to
this questionnaire were prepared by most United States
companies representing the current status of air pollu-
tion control at 30 primary smelters, located in 16
states and ranging in individual plant production capac-
ity from approximately 35,000 short tons to 275,000
short tons of aluminum per year. No company could an-
swer all the questions but the composite represents a
fair picture of the effluents and diverse methods of
control practiced in the United States.
1.3.3 Industry Contacts
To better interpret the data and to better under-
stand the problems of controlling aluminum smelter ef-
fluents, teams of engineers visited nearly all of the
American smelters to see control measures in operation
and to discuss the technology of pollution control with
the plant operating personnel directly involved with
the problems.
1.3.4 Equipment Suppliers
Based on an analysis of the different types of
control equipment applicable to the various components
of reduction plant effluents, technical and cost infor-
mation was obtained concerning such equipment by direct
contact with representative suppliers. These data were
correlated, where possible, with information contained
in industry questionnaire response. Comparative costs,
performance characteristics, and claimed efficiencies
were obtained among the equipment types currently avail-
able to the industry.
Contacts were also made tp gain an insight into
the current trends of manufacturers development activ-
ities leading to future improvements and changes.
1-5
-------
0332
Table of Contents
Section 2
2.0 The Primary Aluminum Industry
2.1 Background
2.1.1 Power
2.1.2 Alumina Shipments
2.1.3 Marketing Areas
2.1.4 International Integration
2.2 Production Statistics
2.2.1 World Production
2.2.2 Domestic Production
2.3 Primary Aluminum Capacity
2.3.1 World Capacity
2.3.2 Domestic Capacity
2.4 Domestic Plant Location
2.5 Peripheral Process Operations
2.6 Projection of Industry Demand Growth
2.6.1 Review of Forecasts
2.7 United States Capacity Growth
2.7.1 Secondary Supply
2.7.2 Stockpile
2.7.3 Foreign Trade
2.7.4 Duties and Tariffs
2.7.5 Internationalization
2.7.6 Projection of Growth Possibility
-------
en 3
\j \s \j
2.0 The Primary Aluminum Industry
2 .1 Background
Aluminum is the most abundant metallic element in
the earth's crust. Its supply potential is limited only
by the technology and economics of extraction, which cur-
rently restrict the commercial raw material to bauxite
ore, although recent announcements have been made by the
USSR of a major project utilizing nephelite, an ore with
a lower aluminum content than bauxite. According to the
Russians, this economic disadvantage can be overcome by
a marketable by-product yield of cement, soda, and potash.
Poland and Mexico are reported to be seriously consider-
ing processing non-bauxite aluminum ores. Poland is re-
ported to be constructing plants for obtaining aluminum
from clays. Developmental work on this approach has been
carried out in the United States and technical feasibil-
ity demonstrated, but economics are unfavorable as com-
pared to the use of bauxite. Mexican researchers have
announced the successful development of a process for
treating alunite ores containing relatively low (10%) con-
tent of alumina. Like nephelite treatment, the economics
are concerned with the marketability of co-products, in
this case potassium and ammonium sulfate fertilizers.
Bauxite ores are plentiful, although the principal
deposits are located in tropical areas away from the main
aluminum metal producing and consuming centers of North
America, Europe and Japan. Known reserves are constantly
being expanded by geological exploration and discoveries;
the largest are located in Australia, Guinea, Jamaica,
and Surinam. The reserves of commercial bauxite in the
United States are relatively small, and account for less
than 1% of the world total.
Bauxite is converted to the intermediate refined
product, alumina, by extraction with caustic soda, precip-
itation of purified aluminum hydrate, and calcination.
This Bayer process treatment is used almost universally
throughout the world, although in Norway a small tonnage
of alumina is extracted commercially from high-iron baux-
ite by the Pedersen smelting process in which bauxite,
limestone, and coke are smelted in an electric furnace to
produce pig iron and calcium aluminate slag containing
30-50% alumina. The slag is leached with sodium carbonate
2-1
-------
0.39 »*
solution and the aluminum trihydrate is precipitated with
carbon dioxide.
Alumina is the principal feed material for the pro-
duction of aluminum metal by electrolytic reduction. In
its commercial form it is a fine white powder, 40 mesh to
submicron in particle size, with about 0.5% impurities,
principally soda (Na20), with minor contamination of cal-
cium oxide, silica, and iron oxide. About one ton of alu-
mina is produced from two tons of bauxite.
The economics of shipping costs, combined with polit-
ical considerations, have resulted in the location of many
alumina plants at sites closer to the mining operations
than to the reduction plants.
2.1.1 Power
Electric power (7.2 kwh per pound of metal) is,
after cost of alumina, the largest cost item in the produc-
tion of aluminum metal, and its availability and cost has
historically been a determinant factor in smelter location.
The early plants in the United States and Canada
were based on cheap hydroelectric power from Niagara Falls
and the upper St. Lawrence River, and later in North Car-
olina and Tennessee. Development of hydroelectric power
in the Northwest by Bonneville Power Authority attracted
reduction plants to Oregon and Washington.
In the early 1950's with most of the available low
cost hydropower in the Northeast, Southeast, and Northwest
developed, there was a move to other sources of power.
The Gulf Coast offered natural gas, and attracted major
plants in Texas. One Texas plant was based on lignite
coal.
Development of large fossil fuel generating stations
in the Ohio River Valley, some of them located at mine
mouth, resulted in power costs which, although higher than
those from hydro, were.low enough to draw new reduction
plants. In the Tennessee Valley Authority area, as well
as in the Pacific Northwest, fossil fuel stations have
supplemented hydro in the power distribution systems. Nu-
clear power stations are being built within these systems
for the same purpose.
2-2
-------
0335
2.1.2 Alumina Shipments
Aside from the considerations of low cost power,
another important factor in plant location has been acces-
sibility to cheap water-borne transportation for raw ma-
terials, especially alumina from domestic and overseas
plants. With the exception of three, domestic reduction
plants built since 1950 have been on waterways navigable
by ship or barge. The exceptions have been able, by spe-
cial circumstances, to obtain sufficiently attractive
trans-shipment and rail facilities to compensate.
2.1.3 Marketing Areas
The relative location of reduction plants to mar-
keting areas for primary metal has not, in the past, been
a major determinant in plant location, being overbalanced
by the pressure to obtain low cost power. As the latter
has become increasingly scarce, transportation cost of
metal has assumed a greater importance, and has acted to
shift reduction facilities closer to markets, both those
for semi-fabricated forms of metal and those for,finished
end product.
2.1.4 International Integration of the Aluminum Industry
About 80% of the free world productive capacity for
bauxite, alumina, and aluminum is operated by some six
corporate groups, or their subsidiaries or affiliates. All
are fully integrated from the mining of bauxite through to
the production of semi-finished and finished aluminum prod-
ucts. They are involved, jointly or with other firms, in
virtually all major aluminum projects of international sig-
nificance in the free world.
Somewhat more than 10% of the current free world
primary metal capacity is accounted for by producers in
which a majority interest is held by the respective na-
tional governments as partner with major aluminum produc-
ers. Somewhat less than 10% of free world capacity is in
the hands of smaller organizations whose integration ex-
tends only towards semi-finished products.
This internationalization and vertical integration
of the industry has developed by reason of the geograph-
ical distribution of ore reserves with respect to low cost
2-3
-------
0396
power and to markets, and because of the very large cap-
ital investments required by the economic scale of opera-
tions, particularly in mining and alumina refining.
Increasing costs, both capital and production, in
the past twenty-five years have led to a shift in the eco-
nomic center of gravity away from markets towards raw
materials. As a result, the trend is toward location of
primary metal reduction plants nearer the sources of baux-
ite, or at best, to areas where cost of product shipment
and marketing is more than offset by lower cost power,
and/or savings in material transport.
2.2 Production Statistics
Primary aluminum is defined as the commercially
pure metal containing about 99.5% aluminum produced usual-
ly by the electrolytic reduction of alumina and cast in
the form of ingot, slab or billet for subsequent working
into semi-fabricated shapes or finished products.
2.2.1 World Production
The estimated world production of primary metal in
1969 was 10,019 thousand short tons _!/. Primary aluminum
was produced in 33 countries, with the United States account-
ing for 37.9% of the world output. Seven other countries,
(the USSR, Canada, Japan, Norway, France, West Germany, and
Italy) produced 43.5% of the world output. Table 2.1 pre-
sents world production statistics over the period 1960-1969
which illustrate the industry growth increase over this
period.
2.2.2 Domestic Production
Production of primary aluminum metal in the United
States was begun in 1858, and data on production starts
in 1893, covering the industry from its very early period.
The semi-log graph, Figure 2.1, illustrates the rapid
growth of the domestic primary industry from 1893, and
Table 2.2 gives the actual production figures. The aver-
age rate of growth since 1893 has been 14.8% and since
1946, 10.2%.
_!/ Aluminum Association
2-4
-------
Table 2.1 WORLD PRODUCTION OF PRIMARY ALUMINUM, 1960 - 1969
Thousands of short tons
COUNTRY
WORLD TOTAL*1)
NORTH AMERICA - Total
Canada
United States
Mexico
SOUTH AMERICA - Total
Brazil
Surinam
Venezuela
ASIA - Total*1)
China*1)
I India
1-71 Japan
South Korea
Taiwan
AFRICA - Total
Cameroon
Ghana
1969p
10,019
4,927
1,098
3,793
36
121
48
59
15
935
132
145
627
7
24
176
52
125
1968r
8,876r
4,265
985
3,255
25
105
46
48
11
786
99
132
532
-
22
170
50
120
1967r
8,343r
4,268
975
3,269
24
79r
42r
34
3
63 2r
88r
106
421
-
17
97
53
44
1966
7,583
3,881
890
2,968
22
58
30
28
-
593
110
92
372
-
19
54
54
-
1965
6,951
3,606
831
2,754
21
35
34
1
-
529
110
74
324
-
21
56
56
-
1964
6,553
3,415
843
2,553
19
29
29
-
-
487
110
62
293
-
21
57
57
-
1963
5,862
3,038
719
2,313
6
19
19
-
-
431
110
61
247
-
13
58
58
-
1962
5,580
2,808
690
2,118
-
22
22
-
-
350
110
39
189
-
12
58
58
-
1961
5,185
2,567
663
1,904
-
22
22
_
-
310
110
20
169
-
10
52
52
-
1960
4,950
2,777
762
2,014
-
20
20
.
-
264
88
20
147
-
9
48
48
-
OCEANIA - Australia 139 ' 107 102 101 97 88 46 18 15 13
r - Revised
p - Preliminary
(1) Estimated by the Bureau of Mines, Department of Interior.
Detail may not add to totals due to rounding.
O'
SOURCE: U.S. Department of the Interior, Bureau of Mines, Co
as reported by the Aluminum Association, 1969. CO
--J
-------
NJ
I
Table 2.1 WORLD PRODUCTION OF PRIMARY ALUMINUM, 1960 - 1969 (Cont.)
Thousands of short tons
COUNTRY 1969p 1968r 1967r 1966 1965 1964
1963
1962
1961
r - Revised
p - Preliminary
(1) Estimated by the Bureau of Mines, Department of Interior.
Detail may not add to totals due to rounding.
SOURCE: U.S. Department of the Interior, Bureau of Mines,
as reported by the Aluminum Association, 1969.
1960
EUROPE - TOTAL^1) 3,
Common Market
Countries - Total
France
Germany, West
Italy
Netherlands
European Free Trade
Association - Total
Austria
Norway
Sweden
Switzerland
United Kingdom
Other - Total ^ 1,
Czechoslovakia^ '
Germany, East(l)
Greece
Hungary
Iceland
Poland
(includes secondary)
Rumania
(includes secondary)
Spain
U.S.S.R/1) 1,
Yugoslavia
720
937
409
290
159
79
865
106
564
74
85
37
918
72
88
88
71
14
107
99
114
213
53
3,444r
892
403
278
157
54
799r
95
516r
62
85
42
l,753r
72
88
84
69
-
103
84
98r
l,102r
53
3,164r
853
398
279
141
35
645
87
398
38
80
43
l,666r
72
88
79
68
-
102
58
86r
1,064
49
2,897
833
401
269
141
22
592
87
357
32
76
41
1,472
68
88
40
67
-
61
52
70
980
46
2,629
770
375
258
137
-
540
87
304
35
74
40
1,319
68
77
-
64
-
52
25
57
930
46
2,477
718
348
242
127
-
513
86
288
34
71
36
1,245
65
72
-
63
-
53
-
55
900
38
2,269
660
329
230
101
-
452
84
248
19
66
34
1,157
65
50
-
61
_
51
-
50
840
40
2,325
613
325
196
91
-
419
82
227
18
55
38
1,293
65
50
•_
58
_
53
_
46
990
31
2,220
590
308
190
92
-
364
75
189
17
47
36
1,266
55
50
_
56
_
52
_
42
980
30
1,827
541
263
186
92
-
350
75
182
18
44
32
936
44
44
_
55
„
29
_
32
705
28
CD
CO
CD
CO
-------
^0339
MILLION
POUNDS
3,000
6,000
4,000
Figure 2.1
Primary Aluminum Production in the
United States 1893 - 1969
1893 - 1969
2,000
1,000
800
600
400
200
too
80
60
40
20
1.0
.8
.6
1893 1900 '05 '10 '15 '20 '25 '30 '35 '40 '45 '50 '55 '60 '65 70
SOURCE. The Aluminum Association
2-7
-------
D^OO
Table 2.2
PRODUCTION OF PRIMARY ALUMINUM IN THE UNITED STATES
Year
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1907
1908
1909
1910
Millions
of
Pounds
0.2
0.5
0.5
1.0
2.4
3.0
3.3
5.1
5.8
5.8
6.6
8.1
10.8
14.1
16.3
10.7
29.1
35.4
Year
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
Millions
of
Pounds
38.4
41.8
47.3
58.0
90.5
115.1
129.9
124.7
128.5
138.0
54.5
73.6
128.5
150.6
140.1
147.4
163.6
210.5
228.0
229.0
Year
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
Millions
of
Pounds
177.5
104.9
85.1
74.2
119.3
224.9
292.7
286.9
327.1
412.6
618.1
1,042.2
1,840.4
1,552.9
990.1
819.3
1,143.5
1,246.9
1,206.9
1,437.2
Year
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
Millions
of
Pounds
1,673.8
1,874.7
2,504.0
2,921.1
3,131.4
3,357.9
3,295.4
3,131.1
3,908.2
4,029.0
3,807.4
4,235.9
4,625.1
5,105.5
5,509.0
5,936.7
6,538.5
6,510.1
7,586.1
7,942.0
(1) Data prior to 1907 represent years ending August 31. Production during last
four months of 1906 totaled 5.4 million pounds.
SOURCES: U.S. Department of the Interior, Bureau of Mines and The Aluminum
Association.
Reference: Aluminum Statistics - 1969, The Aluminum Association.
2-8
-------
In the United States, the Aluminum Company of •*•
America (Alcoa) was the only domestic producer from 1886
to 1940. Under the impetus of World War II the domestic
aluminum production was sharply increased, and by 1946
Reynolds Metals Company and Kaiser Aluminum & Chemical
Corporation had become operators of reduction plants.
Anaconda Aluminum Company began producing in 1954, and
in 1958 Harvey Aluminum, Inc., and Ormet Corporation
(a subsidiary of Olin Mathieson Chemical Corporation and
Revere Copper & Brass, Inc.) entered the field. Con-
solidated Aluminum Corporation (now jointly owned by
Phelps Dodge and Swiss Aluminum Ltd.) began production
in 1963. Intalco Aluminum Corporation, owned 50% by
American Metal Climax, Inc., and 25% each by a U.S. sub-
sidiary of Pechiney (France) and by Howmet Corporation,
started production of primary metal in 1966. National
Southwire Aluminum, a partnership between Southwire (a
fabricator) and National Steel Company, began operations
in 1970, as did Revere Aluminum in their own plant and
Eastalco, a partnership of Howmet and Pechiney.
In addition to these producers, two more, Noranda
and Gulf Coast Aluminum, had reduction plants under con-
struction at the end of 1970.
2.3 Primary Aluminum Capacity
The capacity ratings given to primary aluminum
reduction plants are design capacities, and are nominal
rather than exact. They are rates which are below the
level at which a plant can actually produce by making
(temporary) operational changes, perhaps at the expense
of optimum electrometallurgical efficiency. While occa-
sions for production exceeding design capacity have
existed, the industry prefers to operate normally at,
or somewhat below, capacity.
The consequent uncertainty in determination of
an actual production capacity for an individual plant
is compounded when the attempt is made to assess indus-
try production capacities. Estimates are understood to
represent nominal industry capability, rather than max-
imum potential.
2.3.1 World Capacity
Piant-by-plant compilations of primary aluminum
capacities in the free world have been made and presented
2-9
-------
OM)2
by the technical press which represent perhaps the best
detailed analysis of the worldwide industry. One of the
most recent was published by Metals Week, in "Aluminum -
A Profile of an Industry", and is shown in Table 2.3 to
illustrate the international character and interrelation-
ships of the industry elements, as well as to give an
approximate estimate of world primary aluminum capacity.
Based on the information from this source the cur-
rent (1970) world distribution of primary aluminum capac-
ity is:
Area 1000 S.T.
Free World Annual Capacity Percent
North America 5,352 46.6
Europe 2,281 19.9
Asia 1,004 8.7
Latin America 250 2.2
Africa 231 2.0
Oceania 209 1.8
Subtotal, Free World 9,327 81.2
Sino-Soviet Block 2,161 18.8
World Total 11,488 100.0
A survey of announced plans for capacity growth
within the free world was published by Metals Week
(March 23, 1970) and illustrates the spreading interna-
tional growth pattern of the industry. Tables 2.4a and
2.4b analyze the anticipated growth in terms of geograph-
ical expansion and numbers of producers in the free world.
Of the 30 new producers expected to enter the in-
dustry between 1969 and 1974, all except four are located
outside of the continental United States. Of the 227 ex-
pansions of existing or proposed plants indicated in this
forecast, only 32 are anticipated by U.S. producers.
In terms of distribution of capacity growth, the
United States is expected to account for about 24%, divided
almost equally between new plant capacity and expansion of
2-10
-------
Table 2.3 FREE WORLD PRIMARY ALUMINUM CAPACITY AND OWNERSHIP
(Estimated Capacities in 1000 Short Tons at End of Year)
Expansions
COUNTRY
NORTH AMERICA
Canada :
Alcan Aluminium Ltd. (Alcan)
Canadian British Aluminium Co. Ltd.
(CBA)
United States:
(See Table 2.5 for breakdown)
LATIN AMERICA
Brazil :
Aluminio Mina Gerais, S.A.
Companhia Brasileira de Aluminio
(CBA)
Companhia Mineira de Aluminio S.A.
(Alcominas)
Mexico :
Aluminio S.A. de C.V.
Sur inam :
Surinam Aluminum Co. (Suraico)
Venezuela :
Aluminio del Caroni ,S .A. (Alcasa)
OWNERSHIP
Reynolds Metals 83.5%
Alcan 100%
Government Owned
Alcoa 50%
Hanna Mining 24%
Brazil 26%
Alcoa 46%
Mexicans 54%
Alcoa 100%
Reynolds 50%
Government 50%
1966
900
100
1,000
3,183
/, 1 QO
f , 1OJ
18
22
-
. —
40
22
50
-
1967 1968
950 970
100 115
1,050 1,085
3,339 3,602
4OQQ /. f.Q-7
,joy 4-, DO/
22 25
34 34
-
— _
56 59
22 33
50 77
11 11
1 ^Q 1 RO
Ljy J.OU
1969
970
115
1,085
3,787
4879
3 O / /
25
34
-
_
59
33
77
11
1970
1,000
145
1,145
4,207
^ Q R9
J , J J/
33
55
27
— —
115
33
77
25
9 ^n
ZJU
1971
1,000
175
1,175
4,754
^ Q9Q
:>,yzy
33
55
27
_
115
44
77
25
1972
1,000
175
1,175
5,134
6*5 C\ Q
j juy
33
55
27
___
115
44
77
25 v
'
^i>
CD
Co
-------
Table 2.3 FREE WORLD PRIMARY ALUMINUM CAPACITY AND OWNERSHIP (Cont.)
(Estimated Capacities in 1000 Short Tons at End of Year)
Expansions
COUNTRY
EUROPE
Austria:
Vereinigte Metal Iwerke '
Ranshofen-Berndorf AG (VMRB)
Salsburger Aluminium GmbH (SAG)
France :
Compagnie Pechiney (Pechiney)
Ugine-Kuhlmann (Ugine)
NJ
I West Germany:
to Aluminium-Huette Rheinfelden GmbH
Vereinigte Aluminium-Werke AG (VAW)
Leichmetall-Gemeinschaft
Gebruder Giulini, GmbH
Greece :
Aluminium de Grece S.A.
Iceland:
OWNERSHIP
Government Owned
Alusuisse 1007=
Alusuisse 100%
Government Owned
Alusuisse 50%
Metallgesellschaft 50%
Pechiney 72%
Ugine 18%
Government 10%
1966
80
12
92
314
89
403
61
218
-
-
279
69
1967
80
12
92
316
90
406
66
218
-
-
284
80
1968
80
12
92
316
^90
406
69
218
-
-
287
83
1969
90
12
102
316
90
406
69
218
-
22
309
86
1970
90
12
102
316
90
406
69
218
-
22
309
86
1971
90
12
102
316
90
406
69
218
76
22
385
86
1972
90
12
102
316
90
406
69
218
88
22
397
86
Icelandic Aluminum Co. Ltd,
(Isal)
Alusuisse 100%
36
36
36
-------
Table 2.3 FREE WORLD PRIMARY ALUMINUM CAPACITY AND OWNERSHIP (Cont.)
(Estimated Capacities in 1000 Short Tons at End of Year)
Expansions
COUNTRY
EUROPE (Cont.)
Italy:
Alcan Alluminio Italiano S.p.A.
Montecatini -Edison
Societe Alluminio Veneto per Azioni
S.p.A. (SAVA)
Alluminio Sarda (Alsar)
to Netherlands:
H Aluminium Delfzijl N.V. (AldeL)
U)
Norway ;
A/S Aardal og Sunndal Verk (Aardal)
Det Norske Nitrid A/S (DNN)
Mosjoen Aluminiumverk A/S (Mosal)
Alnor A/S (Alnor)
Elektrokemisk Aluminium A/S
Soer-Norge Aluminium A/S (Soral)
OWNERSHIP 1966
Alcan 100% 6
88
Alusuisse 100% 55
EFIM (Govt) 52%
Montecatini-Edison 24%
Societe Generale
de Belgique 24%
149
Hoogovens 50% 33
Alusuisse 33%
Billiton 17%
Government 50% 216
Alcan 50%
British Aluminium 50% 32
Alcan 50%
Alcoa 50% 68
Elektrokemisk 50%
Norsk Hydro 51%
Harvey Aluminum 49%
Alcoa 50%
Elektrokemisk 50%
Alusuisse 100% 63
379
1967
6
88
63
-
157
36
220
32
68
85
_
63
468
1968
6
93
63
-
162
78
285
40
95
93
_
63
576
1969
6
100
64
-
170
78
285
40
95
95
_
64
579
1970
6
100
64
55
225
99
285
40
95
100
_
64
584
1971
6
102
64
110
282
99
285
45
95
100
36
64
625
1972
6
102
64
110
282
99
285
45
95
120
36
64
645
-------
Table 2.3 FREE WORLD PRIMARY ALUMINUM CAPACITY AND OWNERSHIP (Cont.)
(Estimated Capacities in 1000 Short Tons at End of Year)
Expansion
COUNTRY
EUROPE (Cont.)
Spain :
Aluminio Espanol S.A. (Alumespa)
Empresa Nacional de Aluminio S.A.
(Endasa)
Aluminio de Galicia (Alugasa)
Sweden :
A/B Svenska Aluminiumkompaniet
(Sako)
Switzerland:
Swiss Aluminium Ltd. (Alusuisse)
Usine d1 Aluminium Martigny SA
United Kingdom:
British Aluminium Co. Ltd. (Baco)
Rio Tinto-Zinc/British Insulated
Callendars Cables Ltd.
Alcan Aluminium (UK) Ltd.
Yugoslavia :
State Industry
Trti-al T?ni-nr>o _
OWNERSHIP
Pechiney major interest
Government 75%
SECEN 197=,
Banco de Bilbao 37,
Pechiney 507o
Endasa 107o
Alcan 217o
Svenska Metallwerken 797=
Giulini 1007=.
Tube Investments 49% 7=
Reynolds Metals 487=
Rio Tinto Zinc) fio% tentative
BICC )
Kaiser 407=
Alcan 1007,
Government Owned
i
1966
10
46
28
84
33
70
10
80
39
-
-
39
54
&Q/,
1967
10
46
36
92
55
71
10
81
39
-
-
39
54
1 R/,/,
1968
14
53
36
103
72
72
10
82
42
-
-
42
54
9 m7
1969
14
71
36
121
72
72
10
82
42
-
-
42
54
9 i m
1970
14
83
36
133
77
72
10
82
42
-
-
42
100
9 9«1
1971
14
83
36
133
77
72
10
82
154
112
' 67
333
100
9 7/, fi
1972
14
83
36
133
77
72
10
82
154
112
67
333
220
9 «Q«
-------
Table 2.3 FREE WORLD PRIMARY ALUMINUM CAPACITY AND OWNERSHIP (Cont.)
(Estimated Capacities in 1000 Short Tons at End of Year)
Expansion
COUNTRY
AFRICA
Angola
Aluminio Portugues
Cameroon :
Alucam
Ghana :
Volta Aluminum Corp. (Valco)
South Africa :
Aluminium South Africa (Alusaf)
>L ASIA
01 Bahrein;
Aluminium Bahrein (Alba)
INDIA :
Aluminium Corp. of India Ltd. (Alucoin)
Indian Aluminium Co. Ltd. (Indal)
Hindustan Aluminium Corp. Ltd.
(Hindalco)
Madras Aluminium Co. Ltd. (Malco)
Bharat Aluminium
OWNERSHIP 1966 1967 1968
Pechiney 8.6%
Pechiney-Ugine 60% 57 57 57
Cobeal 10%
Kaiser Aluminum 90% - 115 115
Reynolds Aluminum 10%
Alusuisse 22% - - -
57 179 172
Bahrein Government 27%% -
British Metal Corp. 25%
Akiebolaget Elektro-
koppar 25%
Western Metal Corp. 12%%
Guiness Mahon 10%
8 8 10
Alcan 65% 40 40 41
Kaiser 27% 53 66 66
Birla & Public 73%
Montecatini 27% 11 11 13
Government Owned -
1969 1970
27
57 57
115 115
32
172 2^1
.L / £. &*J -L.
-
10 11
55 73
66 66
13 22
-
1971
27
57
160
55
299
£• x ^
88
13
73
77
28
-
1972
27
57
160
55
OQQ
£.77
88
13
94
110
28
28
112 125 130 144 172 191 273
-------
Table 2.3 FREE WORLD PRIMARY ALUMINUM CAPACITY AND OWNERSHIP (Cont.)
(Estimated Capacities in 1000 Short Tons at End of Year
COUNTRY OWNERSHIP 1966 1967 1968 1969
IRAN:
Iran Aluminium Co. Iran Government 65% -
Reynolds 25%
Pakistan Government 10%
JAPAN:
Nippon Light Metal Co. Ltd. (NLM) Alcan 50% 140 162 184 184
Showa Denko K. K. (Showa) 85 100 155 175
Sumitomo Chemical Co. Ltd.
(Sumitomo) 88 100 144 171
Mitsubishi Chemical Industries 66 74 123 123
(Mitsubishi)
Mitsui Aluminium Co. (Mitsui) -
379 436 606 653
TAIWAN :
Taiwan Aluminium Corp. (Taialco) Government Owned 22 22 22 27
TURKEY:
Government Government Owned - - -
Tr,f-a1 Ao-ia S13 SRI 7SR R9A
Expansion
1970 1971
55
245 311
175 317
200 200
173 173
41
793 1,042
39 42
-
i nnA i AIR
-c-
0
GO
1972
55
311
317
232
173
83
1,116
42
66
i fiin
-------
Table 2.3 FREE WORLD PRIMARY ALUMINUM CAPACITY AND OWNERSHIP (Cont.)
(Estimated Capacities in 1000 Short Tons at End of Year)
Expansion
COUNTRY
OWNERSHIP
1966 1967 1968 1969 1970 1971 1972
OCEANIA
Australia:
Alcoa of Australia Ltd.
I
t—>
^J
Comalco Industries Pty. Ltd.
(Comalco)
Alcan Australia Ltd.
New Zealand;
Comalco/Sumitomo/Showa Denko
Alcoa 51%
Western Mining Corp. 20%
Broken Hill South 17%
North Broken Hill 12%
Kaiser 50%
Conzinc Riotinto 50%
Alcan 100%
Comalco 50%
Sumitomo 25%
Showa Denko 25%
44
62
Total Oceania 106
44
81
125
44
88
88
81
81
40
81
40
125
209
209
88
81
40
79
288
99
81
40
106 125 125 209 209 209 220
118
338
GRAND TOTALS
6,665 7,252 7,959 8,358 9,327 10,941 11,745
Source: Farin and Reibsamen, "Aluminum, A Profile of an Industry," 1969
•T-
O
CD
-------
Table 2.4 a SUMMARY OF ANTICIPATED CAPACITY GROWTH, 1969-1974
(Free World)
a) Capacity Units
1969-1974
Producers
Expansions By Producing Countries
Mid -1968
-1969
-1970
-1971
-1972
-1973
-1974
Existing
51
51
56
59
73
77
80
New
5
3
14
4
3
1
Total Existing Producers New
56
59
73
77
80
81
39
40
46
50
29
23
2
-
4
2
1
1
Total
23
25
25
29
31
32
33
30
329
10
b) Distribution of Anticipated Growth. U. S. & Others (1969-1974)
Estimated Metric Tons
U. S. Others Total
By Plant Expansion 737 1,289 2,026
In New Plants 634 3,135 3,769
Percentage of
Capacity Increase
U. S. Others Total
12.7 22.2 34.9
11.0 54.1 65.1
Total Capacity
Increase
(1969-1974)
Mid-1968 Estimated
Capacity
Estimated
1974 Capacity
1,371 4,424 5,795 (82.770) 23.7 76.3 100.0
6,988 (100.0%)
12,783 (182.7%)
(Estimates published - Metals Week, March 23, 1970)
2-18
-------
Table 2.4b SUMMARY OF ANTICIPATED CAPACITY GROWTH, 1969-1974
(Free World)
c) Patterns of World Capacity Expansion (1969-1974)
Est. Quantity, 1000 M.T.
Distribution
Increase, 7o of 1969
Base
1969
7548
Capacity Expansion
1974
By Plant
Expansions
2026
357,
27%
In New
Plants
3769
657»
507o
Total Capacity
Increase
5795
1007o
777,
Pattern by Geographical Area
(% of Free World)
EEC-/
EEC Associates-'
EFTA-/
Other Europe
Total Europe
North America
Latin America
Africa
Asia
Oceania
Total Free World
11%
3
11
_2
277o
58
2
2
19
2
1007o
107o
4
16
_5
35%
37
4
3
17
4
1007o
227o
1
12
__3
387,
24
4
2
28
4
10070
1770
2
13
_5_
377o
28
4
2
25
4
1007o
I/ European Economic Community (France, West Germany, Italy, Netherlands,
Belgium-Luxembourg).
2f Cameroon, Greece, Surinam, Turkey.
3_/ United Kingdom, Austria, Norway, Sweden, Switzerland, Iceland.
SOURCE: Metals Week Roundup - March 23, 1970.
2-19
-------
OM12
existing plants. The growth in the rest of the free world
is anticipated to be more than three times that of the
domestic industry, with 70% of that growth as new plant
capacity.
2.3.2 Domestic Cap a c it y
As of December 31, 1948 there were three domestic
producers of primary aluminum in the United States with a
combined annual capacity of 641,500 short tons. Capacity
increases were frequent through 1959, when total annual
capacity stood at 2,402,750 short tons at year end, and
three new producers had entered the field.
During the next five years through 1964, total an-
nual capacity increased only 8.2% and only one new producer
appeared. During the four years, 1965 through 1968, capac-
ity increased 49.6%, an average annual rate of 8.4%.
During 1969 capacity increased 203,300 short tons
to 3,888,300 short tons, a gain of 5.5% over 1968. This
is compared to an expansion of 11.0% in the previous year.
A ninth primary producer came on-stream in 1969.
Table 2.5 shows the reported annual capacities of
the producers during the period 1948-1969.
Six of the existing producers have announced plans
for additional capacity or modernization in the period
1970-1972. Four new producers have new plants under con-
struction at the start of 1970, and a fifth new producer
appears to be a probability in the next three years.
Table 2.6 lists the individual production plants
in operation, or under construction in 1970, with the data
indicating their capacity increases during the 1966-1972
period. It is noted that the capacities given in this
table do not exactly coincide with totals presented for
companies in Table 2.5. The estimates are from different,
but equally authoritative sources and the differences are
considered to be not significant.
The amount of excess domestic capacity in relation
to production which has existed in the past two decades
is indicated in Figure 2.2 from data given in Table 2.7.
Capacity increase approximately matched that of production
2-20
-------
Table 2.5 REPORTED CAPACITIES OF PRIMARY ALUMINUM PRODUCERS, 1948 - 1969
Yeard-
1948
1949
1950
1951
1952
1953
1954
1955
1956
N> 1957
L 1958
H 1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
) Total
641,500
654,000
751,250
800,750
1,155,700
1,335,700
1,413,200
1,634,700
1,775,500
1,839,000
2,194,250
2,402,750
2,468,750
2,483,750
2,488,750
2,510,750
2,599,100
2,758,284
3,165,284
3,321,000
3,686,000
3,888,300
Alcoa
325,000
294,000
369,750
371,250
484,250
548,000
570,500
706,500
792,500
792,500
798,250
798,250
853,250
853,250
853,250
853,250
858,100
950,000
1,050,000
1,150,000
1,200,000
1,325,000
Reynolds
188,000
227,000
238,500
259,500
353,250
359,500
414,500
440,000
488,500
488,500
601,000
701,000
701,000
701,000
701,000
701,000
725,000
725,000
815,000
815,000
895,000
935,000
Kaiser
128,500
133,000
143,000
170,000
318,200
428,200
428,200
428,200
434,500
498,000
537,000
609,500
609,500
609,500
609,500
609,500
650,000
650,000
670,000
670,000
690,000
710,000
1948 - 1969
Short Tons
Intalco Ormet
_
-
-
-
- .
-
-
-
-
-
144,000
180,000
180,000
180,000
180,000
180,000
180,000
184,284
152,000 184,284
152,000 240,000
255,000 240,000
265,000 240,000
Anaconda
_
-
-
-
-
-
-
60,000
60,000
60,000
60,000
60,000
65,000
65,000
65,000
67,000
67,000
100,000
100,000
100,000
175,000
175,000
Consolidated Harvey
_ _
-
-
-
-
-
-
-
_ _
-
54,000
54,000
60,000
75,000
80,000
20,000 80,000
32,000 87,000
62,000 87,000
106,000 88,000
106,000 88,000
140,000 91,000
140,000 91,000
National-
Southwire
_
-
-
_
-
-
-
-
_
-
•
-
-
_
-
-
-
-
_
-
-
7,300
(1) Capacities shown are as reported for December 31.
SOURCE: The Aluminum Association.
-------
Table 2.6
INDIVIDUAL ALUMINUM SMELTER CAPACITIES
(Thousands of Short Tons at Year End)
EXISTING COMPANIES
Alcoa;
Alcoa, Tenn.l/
Badin, N.C.
Evansville, Ind.
Massena, N.Y.
1966
1967
1968
1969 1970 1971 1972
21
Point Comfort, Tex.—'
Rockdale, Tex.
Vancouver, Wash.
Wenatchee, Wash.
Alcoa Total
Anaconda:
21
Columbia Falls, Mont.-'
Sebree, Ky.
Anaconda Total
Conalco:
New Johnsonville, Tenn.
Harvey:
21
The Dalles, Ore.^'
Near John Day Dam in
southern Washington—'
Harvey Total
Intalco;
Fernda1e, Wash.
Kaiser:
Chalmette, La.—'
Mead, Wash.
Ravenswood, W.Va.
Ta coma, Wa sh. -L'
Kaiser Total
Ormet:
Hannibal, Ohio
125
50
175
125
175
175
100
125
1,050
105
-
105
106
88
,
88
165
260
206
163
41
670
125
100
175
125
175
175
100
175
1,150
105
-
105
106
88
«__
88
165
260
206
163
41
670
125
100
175
125
175
175
100
175
1,150
175
-
175
140
88
,..
88
264
260
206
163
61
690
200
100
175
125
175
225
100
175
1,275
175
-
175
140
88
88
264
260
206
163
81
710
200
100
225
125
175
275
100
175
1,375
175
-
175
140
88
_ r
88
264
260
206
163
81
710
T
1,425
175
-
175
140
88
100
188
264
T
740e
T
l,475e
175
50e
225e
140
88
100
188
264
j
1
770e
184
240
240
240
240
240
240
2-22
-------
Table 2.6 (Cont.)
CM* 15
INDIVIDUAL ALUMINUM SMELTER CAPACITIES
(Thousands of Short Tons at Year End)
EXISTING COMPANIES
Reynolds:
Arkadelphia, Ark.—'
4/
3/
Corpus Christi, Tex.i'
Jones Mills, Ark.
Sheffield, Ala.i'
Longview, Wash.2.'
Massena, N.Yo—
Troutdale, Ore.
Reynolds Total
Total Existing
Companies
1966
1967
1968
1969
1970
1971
1972
63
111
122
221
70
128
100
815
63
111
122
221
70
128
100
815
63
111
122
221
110
128
100
855
63
111
122
221
150
128
100
895
63
111
122
221
190 z
128
140
' 1
975 l,005e 1,065
3,183 3,339 3,602 3,787 3,967 4,177e 4,367e
NEW COMPANIES
Eastalco:
Frederick, Md.
Gulf Coast Aluminum:
Lake Charles, La.
National-Southwire
Aluminum:
Hawesville, Ky.
Noranda;
New Madrid, Mo.
Revere:
Scottsboro, Ala.
Total New Companies
Grand Total
85
85 170
35
90 180
75
35
180
75
72 112
175 447 572
3,183 3,339 3,602 3,787 4,142 4,624e 4,939e
NA not available
e METALS WEEK estimates
SOURCE;
Note: All plants prebaked anode cell except:
I/ Both prebake and VSS potlines
2J VSS potlines only
3_/ HSS potlines only
4_/ Both prebake and HSS potlines
Farin and Reibsamen, "Aluminum,
A Profile of an Industry," 1969
2-23
-------
(M6
FIGURE 2.2
EXCESS CAPACITY
U.S. PRIMARY ALUMINUM INDUSTRY
z
o
o
1948 1950 1952 1954 1956 1958 1960 1962 1964 1966 1968 1970
YEAR
DATA SOURCE: ALUMINUM STATISTICS, 1969 THE ALUMINUM ASSOCIATION
2-24
-------
Table 2.7
U. S. CAPACITY vs. PRODUCTION (See Fig. 2.3)
1000 Tons
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
Rated
Cap.l/
641.5
654
751
800.7
1155
1336
1413
1635
1776
1839
2194
2408
2469
2484
2488
2511
2599
2758
3165
3321
3666
3888
Production!.'
623.5
603.5
718.6
836.9
937.4
1252.0
1460.5
1565.7
1679.0
1647.7
1565.5
1954.1
2014.5
1903.7
2118.0
2312.0
2553.0
2754.5
2968.4
3269.0
3255.0
3798.0
Prod. /Cap.
7o
97
92.5
95.5
104.5
81.2
94.0
103.0
95.6
94.6
89.7
71.2
81.3
83.5
76.5
85.0
92.0
98.0
100.0
93.8
98.5
89.0
97.6
Cap. /Prod
°/
103
108
104.5
96
123
106
97
103.5
105.5
111
140
123
120
130
126
109
102
100
107
102
112
102
I/ Source: Aluminum Statistics - 1969, The Aluminum Association.
2-25
-------
during 1948-1956 and a narrow margin (approximately
of excess maintained. A falloff in production in 1956-
1958, coupled with the completion of scheduled capacity
increases during 1956-1959, resulted in a period of
some five years during which the excess capacity margin
rose as much as 40% and averaged about 25%. The recovery
of demand and production, coupled with a sharply reduced
rate of capacity increase between 1960-1964, rapidly
closed the gap to 5% (average) and restored what might
be considered a normal relationship. Resumption of capac-
ity increase after 1966 maintained this balance.
With the completion of the announced new plants
and expansions in the period 1971-72 it appears likely
that, in the short range, domestic capacity will again
exceed domestic production by a significant amount.
During the last half of 1970 and the first half of
1971 the economic conditions in the aluminum industry have
caused many producers, both domestic and foreign, to cut
back production substantially from existing facilities, to
stretch out construction schedules on new facilities, and
to postpone announced plans for expansions. United States
primary production in 1970 increased less than 5% over 1969
output, and 1971 may well show a decrease over 1970. The
excess capacity gap in 1971-72 may be of the order of 20%,
which has been exceeded in several previous years, and is
a situation which gives the industry concern, for the
short term, but not alarm in considering the longer view.
2.4 Domestic Plant Location
Figure 2.3 shows the geographical "distribution of
United States smelters and graphically indicates their
relative current normal capacities by the size of the
symbol at each location.
In addition to those listed, it appears quite pos-
sible that two more plants will be constructed by a new
entrant into the United States primary aluminum industry.
One of these would be located at Warrenton, Oregon on the
site of the former Northwest Aluminum project, and the
other in Puerto Rico. Firm announcement of these plants
had not been made in 1970, and it is assumed that they
would not come into production before 1973.
2-26
-------
r iiLw^G^or --
^3«?' YT^^r
runt 2.1
ALIIIIIIM IMITIIS
>N
IIITfl fTIIES
MfW flANTt AT PARTIAL CAFACITV IN t*70
EASTAICO- FREDERICK. MD
NATION*! iOUTMWIRE- HOWESVIUI, KT
REVERE ALUMINUM* SCOTTSBOBO. ALA
NEW PLANTS UNDER CONSTRUCTION IN 1*70
ANACONDA- SEBRfE KT
OULF COAST-LAKE CMAfiilS.lA
CD
CD
NORONDA-NEW MADRID. MO
-------
0^20
Most of the aluminum reduction plants in the
United States are located in predominantly rural areas
with a sparse population density, as estimated by the
total population of towns or cities within a ten mile
radius of each plant given in 1960 or later census fig-
ures.
The distribution of plants (1971) with respect to
population is:
Number
of Plants
13
9
2
7
Percent
Capacity
41.1
28.7
5.7
24.5
Surrounding 300 Square Mile Area
Population Population/Sq.Mi.
Less than 10,000
10-25,000
25-50,000
More than 50,000
Less than 32
32-80
80-160
More than 160
One plant is surrounded by residential sections in
an urban community. The other six plants in the high den-
sity areas are located on the outskirts of medium sized
communities where the surrounding land is utilized for
dairy farming or truck farming.
The distribution of plant capacity with respect to
the type of surrounding land use is shown in Table 2.8.
2.5
Peripheral Process Operations
Peripheral process operations are defined as those
in-plant units, controlled by the reduction plant manage-
ment and located on the reduction plant site, which sup-
ply power, steam, or materials to the reduction plant, or
recover by-products from the plant.
Alumina production and/or coke calcining are usual-
ly carried out at locations which are remote from the alu-
minum production plant and they are not included in the
scope of this study.
Power supply to reduction plants is generally pur-
chased from utilities, some of which operate central sta-
tions largely dedicated to the reduction plant load. These
stations, being independently operated by the utilities,
are excluded from the scope of the study.
2-28
-------
Table 2.8 ESTIMATED DISTRIBUTION OF PLANTS BY ENVIRONMENT
Environmental Category
Urban
Orchard Growing
Dairy Farming
Truck Farming
Cattle Raising
Lumbering
General Agriculture
Dairy plus Truck
Dairy plus Cattle
Dairy plus Agriculture
Dairy plus Lumber
Truck plus Cattle
Truck plus Lumber
Number
of
Plants
1
4
3
4
1
1
2
2
1
1
1
1
4
Percent of Total U. S.
Aluminum Capacity in
Environmental Category
5.5
15.4
9.4
10.9
1.9
5.6
2.4
6.3
4.4
3.0
1.7
3.0
11.3
Truck plus General
Agriculture
Lumber plus General
Agriculture
Truck Farming plus
Cattle plus General
Agriculture
Total
31
3.5
4.7
11.0
100.0
Source: Industry Questionnaires, EPA Contract CPA 70-21 and
Encyclopedia Britannica.
2-29
-------
0^22
In areas where cheap natural gas is available,
five aluminum reduction plants have installed in-plant
generation units incorporating gas engines, gas turbines,
or gas-fired boiler steam turbine units. In two plants
these installations are the normal power source, in an-
other they are supplemented by purchased power, and in
the remainder they supplement the power purchased from
utilities.
Low Pressure Steam for process and heating require-
ments is generated at all plants, in almost all cases by
gas-fired boilers.
Cryolite Recovery
Synthetic cryolite is produced from treatment of
reclaimed pot materials in six plants which are part of
primary reduction plant operations, as well as in another
separate recovery plant operated by a major primary alu-
minum producer. Reclaimed pot materials from other reduc-
tion plants are generally shipped or sold to these cryolite
plants on a toll basis.
Fluoride containing products of the water treatment
from gas cleaning installations are generally impounded,
although they may be recovered and treated with reclaimed
pot materials where cryolite facilities are available.
A few plants recover fluorine from this source alone.
The processing operations of cryolite recovery do
not involve generation of airborne pollutants other than
those incident to the crushing of feed and the final han-
dling of calcined product. For this reason the reported
questionnaire information covering these plants was scanty
and in such form that it could not be meaningfully sum-
marized with respect to industry operations or emissions
from this source.
Table 2.9 indicates the peripheral process plant
installations associated with domestic reduction plants,
and Table 2.10 summarizes reported industry data concern-
.ing fired low pressure steam generation units.
2-30
-------
Table 2. 9 PERIPHERAL PLANT PROCESSES ASSOCIATED WITH U. S.
PRIMARY ALUMINUM INDUSTRY, BY PLANTS, 1970
Power
Cryolite
Recovery
Plant
Alcoa:
Alcoa Tenn.
Badin, N. C.
Evansville, Ind.
Massena, N. Y.
Pt. Comfort, Tex.
Rockdale, Tex.
Vancouver, Wa sh.
Wenatchee, Wash.
Anaconda:
Columbia Falls, Mont.
Sebree, Ky.
Conalco:
New Johnsonville, Tenn.
Harvey:
The Dalles, Ore.
John Day, Wash.
Intalco:
Ferndale, Wash.
Kaiser:
Chalmette, La.
Mead, Wa sh.
Ravenswood, W.Va.
Tacoma, Wash.
Ormet:
Hannibal, Ohio
Reynolds:
Arkadelphia, Ark.
Corpus Christi, Tex.
Jones Mills, Ark.
Sheffield, Ala.
Longview, Wash.
Massena, N. Y.
Troutdale, Ore.
U
U
U
U
P - U
U
U
U
U
U
U
U
U
U
U - P
U
U
U
U
U
P
U - P
U
U
U
U
SH
SH
SH
SH
SH
SH
SH
SH
SL
SL
SL
SL
SL
R
SL, R
SL
SL
R
R
R
SH
R
R
R
R
2-31
-------
Table 2.9 PERIPHERAL PLANT PROCESSES ASSOCIATED WITH U. S.
PRIMARY ALUMINUM INDUSTRY, BY PLANTS, 1970 (Gont.)
Eastalco;
Frederick, Md.
Gulf Coast Aluminum:
Lake Charles, La.
National-Southwire
Aluminum:
Hawesville, Ky.
Noranda;
New Madrid, Mo.
Revere;
Scottsboro, Ala.
Power
U
U
U
U
Cryolite
Recovery
Plant
R
SL
SL
SL
SL
Notes:
Power: U - Utility furnished.
P - In-plant generation by gas-fired units.
Cryolite: R - Recovery plant.
SH - Materials treated for recovery in parent
company plant.
SL - Materials sold to other companies.
SOURCE: Industry Questionnaires and Plant Visits,
EPA Contract CPA 70-21.
2-32
-------
Table 2.10
PRIMARY ALUMINUM INDUSTRY; FIRED LOW PRESSURE STEAM
GENERATION INSTALLATIONS, SUMMARY DATA
Total Units Reported
Normal Fuel - Natural Gas
Powdered Coal-
Emergency Back-up Fuel
Residual Oil
Distillate
None
Normal Operating Time - 100%
50-1007,
Less than 50%
Approx. Heating Rate - Maximum
Minimum
Average
Type of Boiler - Water Tube
Fire Tube
Direct or Indirect Heaters
Date of Installation (gas-fired)
Stack Heights
31
30 units
1 unit
5 units
7 units
19 units
17 units
6 units
8 units
117 x 106 BTU/Hr
4 x 106 BTU/Hr
30 x 106 BTU/Hr
19 units
6 units
6 units
1941-1970
30-175 ft.
Region Units
Pacific Northwest 10
Gulf Coast and Texas 3
TVA and Mississippi 8
Eastern States 10
Heat Rate Range
4-56 x 106 BTU/Hr
5-30 x 106 BTU/Hr
7-84 x 106 BTU/Hr
5-117 xlO6 BTU/Hr
Installations
1941-1968
1952
1942-1970
1953-1970
I/ Peak winter load only.
Source: Responses to Industry Questionnaires, EPA Contract CPA 70-21.
(15 of 30 plants reported relevant data)
2-33
-------
OU26
2.6 Projection of Industry Demand Growth
Forecasts of future growth in the aluminum in-
dustry have been made periodically by informed minerals
economists both inside and outside of government.
Referenced to projections of compounded annual growth
factors of increasing gross national product, expanding
markets in end-use consumer markets in virtually all
sections of the economy, availability of raw materials,
aggressive marketing efforts, and the metal's versatil-
ity, these forecasts are consistent and there is every
reason to believe that aluminum will continue to be a
major growth metal for many years.
It should be noted, however, that the forecasts
of economic growth in the primary aluminum industry used
in this report, were not obtained from the primary pro-
ducers or from any other sources within the primary alu-
minum industry. Consequently, these economic forecasts
are not intended to reflect any judgment by the primary
producers as to long-term future growth of the primary
aluminum industry.
There is a considerable range of optimism among
these forecasts - according to the period in which they
were made and the interpretations of the growth trends
on which they were based. In the light of past experience
they are sound, but not precise, and as they are extend-
ed beyond a relatively short period (10-15 years) the
spread between high and low estimates becomes very wide.
Translation of these forecasts of world consump-
tion into terms of demands for domestically produced
primary aluminum requires consideration of the second-
ary metal contributing to total supply, the proportion
of world demand represented by U.S. consumption, and
the international trends which may affect the competi-
tive position in the U.S. primary markets.
Five of these forecasts are compared in Tables
2.11 with extrapolation based on their findings. There
is a reasonable area of agreement with respect to United
States consumption in the median projections of the
latest three extended out to the year 2000, and the
growth rate assumed by the most recent is accepted as
being credible for the analysis.
2-34
-------
'"G4£7
Table 2.11a COMPARISON OF FORECASTS OF FUTURE ALUMINUM CONSUMPTION
(a) World Consumption
Metal si/
Landberg— Petrickfi' Brubaker—' Week
Stamper—'
Year Made
World Consumption
Base Year
Base Tonnage
x 106 short tons
Growth Factor
Median 7»
Estimates
10 short tons/yr
1970
1980 Low
Median
High
1984 Low
Median
High
2000 Low
Median
High
1963
1960
1967 1965 1969
(non-communist)
1965 1965 1967
7.42
9.5
28.73
5.8
6.8
14
19
22
8.65
8.1
32.5
1970
1968
11.3
6.4
53.9
83.2
112.4
2-35
-------
CMS
Table 2.lib COMPARISON OF FORECASTS OF FUTURE ALUMINUM CONSUMPTION
(b) U. S. Consumption
Year Made
Base Year
Base Tonnage
x 10 short tons
Landberg—
1963
1960
2.13
Petrick;?./ Brubaker^/
1967
1965
2.75
1965
(U.S. plus
Can.)
1965
Metals4./
Week
1969
1969
4.15
Stamper ^
1970
1968
4.31
Growth Factor
Median 5.5
Estimates
10 short tons/yr
1970 (3.63)
1980
5.8
6.2
7.0
6.45
Prim. Ingot
Sec . Supply
Total - Low
Median
High
2000
Prim. Ingot
Secondary
Total - Low
Median
High
5.66
0.61
3.82
6.27
11.18
13.28
2.94
7.31
16.22
34.15
(8.0)
(1.4)
6.41 9.4
(21.1)
(3.7)
11.1 (24.8)
- 11*
2*
13*
(28.1)
(5.1)
(32.2)
-
-
™
(25.8)
(5.8)
21.2
31.6
42.0
* For 1984
( ) Extrapolated or ratioed
I/ Landberg, Fischman and Fisher, Resources in America's Future,
Johns-Hopkins Press, 1963
2/ A. Petrick, Proc. Council Economics, AIME, 1967
3/ S. Brubaker, Trends in World Aluminum Industry, Resources for the Future,
Johns-Hopkins Press, 1967
4/ Farin and Riebsamen, Aluminum, a Profile of an Industry, McGraw Hill, 1969
5_/ J. Stamper, Aluminum, Minerals Facts & Problems, U.S. Bureau of Mines, 1970
2-36
-------
2.6.1 Review of Forecasts
The industry trade publication, Metals Week, in
their study, "Aluminum, A Profile of an Industry", com-
ments on three previously made studies on the future of
aluminum as follows:
"Alfred Petrick, Jr., economist for the Bureau of
Mines, estimated in 1967 that world aluminum con-
sumption in 1980 would reach 29-million tons, or
about four times the 1965 level of 7.4-million.
In a study entitled "World Demand for Mineral Prod-
ucts and the Shifting Supply of Mineral Raw Mate-
rials", Petrick emphasized that there is still
plenty of room for world expansion despite major
inroads into copper's electrical markets. Petrick
used a 9.5% growth rate for the 1965-1980 period.
"Resources in America's Future, a 1963 book of un-
precedented scope backed by the Ford Foundation,
expected US aluminum consumption to be three and
one-half times as high in the year 1980 as in 1960,
and about two and one-half times as high in the
year 2000 as in 1980. Since US consumption in 1960
was about 1.6-million tons, this would place 1980
use at 8.7-million tons, rising to 14.7-million
tons by 2000.
"Yet, the book projects three consumption levels.
It says that, 'if all factors favor the demand for
the metal, consumption by 1980 may rise to six
times' what it was in the early 1960's, and by 2000
'to three times that of 1980'. However, it adds
that even under the least favorable circumstances,
an increase of about 120% by 1980 seems to be in
the cards, although the subsequent 20 years might
see only a 90% gain. The third projection is a
median of the high and low.
"Sterling Brubaker, in his 1967 book, Trends in
the World Aluminum Industry, concludes that through
1980 a 5-8% annual growth rate for the Free World
is a possibility, with 7% as the most likely fig-
ure. He derived these figures from per capita
growth rates worked out by the United Nations in
1962.
2-37
-------
QUO
"By extrapolating from 1964 information, Brubaker
then arrived at a Free World consumption for 1980
of about 19-million tons. Accordingly, he estab-
lished a range of about 14-million to 22-million
tons as being possible for that year.
"Brubaker's analysis is easily the most exhaustive
and perceptive of the three largely, presumably,
because unlike the others, he deals only with alu-
minum. Appropriately, he notes that forecasters -
including the industry itself - have 'had a very
spotty record of success in making estimates of
future consumption by a variety of statistical
techniques'. He adds, however, that no definitive
forecast of future demand is required, since,for-
tunately, long lead times are not required for sup-
ply to react to demand."
Metals Week, in preparing its own forecasts of the
future of aluminum, enlisted the cooperation of the manage-
ments of seven of the major world aluminum producers in
analyzing the factors which would affect demand and supply,
during the period from 1969 to 1984. Considered in this
study were anticipation of application developments and
changes, interrelationships of the industry throughout the
world, future of power costs and sources, forward and back-
ward integration, influence of emergent nations, compe-
tition with other materials, possible technological changes
and analyses of market and utilization changes.
The conclusion of this study is that the past pat-
tern of world aluminum growth trend is not expected to
change much in the coming 15 years, and that between now
and 1984 total aluminum shipments are expected to triple.
That magnitude of increase calls for a continued demand
growth of 7-8% reaching perhaps 11-14 million tons in
1984.
In the 1970 volume of the authoritative "Mineral
Facts and Problems" published by the U.S. Bureau of Mines,
Mr. John Stamper has made an in-depth analysis of future
aluminum demand, projecting to the year 2000. His fore-
cast, based on analysis of the probabilities of techno-
logical, social and economic changes and their possible
effects during the period, places the range of United
States demand for aluminum metal in the year 2000 as be-
2-38
-------
0'.
•
tween 21.2 million and 42 million tons, corresponding to
an average annual rate of growth between 5.1% and 7.4%.
To establish a forecast base, domestic aluminum
demand in each of 14 end-use categories was projected by
Stamper on the basis of a number of considerations which
included:
Qualitative and quantitative correlations with
projections of growth in gross national product
(GNP). Federal Reserve Board index, and other
general economic indicators;
Short term (5-10 years) and long term (10-40 years)
projections published by Resources for the
Future, Inc., and other organizations;
Projections for specific items, such as number and
type of motor vehicles and aircraft, growth in
expenditures for irrigation, pollution control,
and other equipment, and growth in industrial
chemicals and steel.
Using econometric techniques and judgment, a fore-
cast base in each of the end-use categories was estab-
lished.
A number of contingent changes in technology and
in the economic mix were then considered that could have
a positive or negative influence on the forecast base for
each of the end-uses, in order to arrive at a high and
low forecast range for each end-use. Interrelations be-
tween end-uses were evaluated to determine compensating
inverse demand effects and the resulting probability that
total aluminum demand would fall within the forecast
range.
The results of these forecasts are shown in Table
2.12.
Primary aluminum demand is defined by Mr. Stamper
as total demand for aluminum less that quantity recovered
from secondary sources, and the forecast for primary de-
mand is based on the assumption that the present propor-
tion of secondary recovery to total demand will not under-
go a marked change. Continuation of the current share of
2-39
-------
QV32
Table 2.12
CONTINGENCY FORECASTS OF DEMAND FOR ALUMINUM
BY END USE, YEAR 2000
(Million Short Tons)
End Use
Metal :
Building and
construction
Motor vehicles
Aircraft and parts
Shipbuilding and repair
Railroad
Electrical
Fabricated metal parts
(consumer durables)
Machinery and equipment
(except electrical)
Metal cans and containers,
and packaging
Highway and street
construction
Other manufacturing and
fabrication^-
Total
2
Bauxite and Alumina :
Abrasives, aluminous
Chemical and allied
products
Nonclay refractories
Total
Grand Total
Demand
1968
1.00
.67
.17
.02
.02
.60
.47
.31
.46
.06
.53
4.31
.08
.16
.16
.40
4.71
U.S.
forecast
base
2000
4.5
2.7
.8
.4
4.5
4.0
4.2
2.3
.4
2.1
-
.3
.7
.6
_
Demand in Year 2000
United
States
Low
4.0
2.7
.4
.4
4.0
3.5
2.5
1.4
.3
2.0
21.2
.2
.5
.5
1.2
22.4
(Median
High
6.0
6.0
1.0
1.0
8.0
5.0
6.0
5.0
1.0
3.0
42.0
.4
1.0
1.0
2.4
44.4
33.4)
Rest
of
the World
Low
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
31.5
(Median
High
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
68.0
49.8)
NA—Not available.
1—Includes aluminum content of some alumina and bauxite (see fig. 1).
2—Aluminum content of bauxite and alumina.
SOURCE: Stamper - Mineral Facts & Problems, U.S. Bureau of Mines, 1970.
2-40
-------
the market by the independent secondary aluminum industry
and the proportion of aluminum recovered from old alumi-
num scrap in relation to that recovered from new scrap is
also implicit in this assumption. The forecast presents
the future breakdown between primary and secondary metal
as follows:
United States Aluminum Demand
(Millions of Short Tons)
1968 2000
Low Median
Primary 3.5 17.3
Secondary 0.8 3.9
Total 4.3 21.2 31.6 42.0
2.7 United States Capacity Growth
If the order of magnitude of future U.S. demands
for new primary aluminum metal is accepted as being be-
tween 9 and 14 million short tons in 1984, rising to be-
tween 17 and 34 million tons in 2000, the problems affect-
ing future domestic capacity growth require analysis. The
degree to which domestic capacity will expand is not, as
it has generally been in the past, in a simple relation-
ship to demand growth.
Historically, domestic consumption of aluminum has
been supplied in the greater part by domestic production
from primary and secondary sources? lesser but occasion-
ally important supply has been furnished by withdrawals
from accumulated Government stockpile, and by a small net
balance of imports over exports.
2.7.1 Secondary Supply
As has been noted, before, the bases for forecasts
of future new primary metal demand have included the prem-
ise that the future contribution to total supply from
secondary metal will continue as a constant proportion of
that supply.
2-41
-------
2.7.2 Stockpile
The U.S. Government aluminum stockpile, after re-
vision in its objectives, has been used in recent years
as a mechanism to permit orderly expansions of the indus-
try with private capital by accepting "puts" to the
stockpile during periods of slack demand and "takes" by
participating companies to ease pressures when demand
pushed capacity. Large amounts of aluminum had been added
to this stockpile in the mid-1950's, a systematic disposal
plan of the surplus was instituted in 1965 which provided
for the participating companies to purchase about 1.45
million tons over a period of 16 years.
The relationship of this stockpile movement to
domestic primary production is illustrated in Figure 2.4.
The use of Government stockpiles to adjust pres-
sures on primary production has been an influence on the
industry growth in the past. It is expected that stock-
piling policies with respect to defense goals, acquisi-
tions, and releases will continue to have impact on domes-
tic supply and demand, but to a lesser relative degree
than in the past as the production and capacity of the
industry increase.
2.7.3 Foreign Trade
Imports of primary aluminum in the form of ingot
and mill products have been a minor, but increasing,
factor in the domestic industry supply since the mid-
1940 's. A major component of this import trade has been
the movement of metal from the largest Canadian producer
to its U.S. fabricating subsidiaries, representing an
intra-company activity. Such shipments have accounted
for 70 to 80% of primary aluminum metal imports in recent
years (1967-1968).
Exports of primary aluminum ingot have also oc-
curred, growing from a few thousand tons in the early
1950's to several hundred thousands in the 1960's, but a
net import balance has existed in every year except 1960.
The trend of this net import balance has shown a steadily
increasing growth over the past two decades, with cyclic
interruptions, as shown in Figure 2.5.
2-42
-------
FIGURE 2.4
0^35
PROPORTION OF PRIMARY PRODUCTION
REPRESENTED BY STOCKPILE MOVEMENT
1948 1930 1952 1954 1956 195t 1960 19*7 19*4
1966 19M
1970
YIAR
bATA SOURCI US IUREAU OF MINES MINERAL YEARBOOKS
2-43
-------
FIGURE 2.5
ALUMINUM INGOT
U.S. IMPORTS & EXPORTS
700
1941 1*50 1*52 954 195* WSi
YEAR
V/ i »a i9
19«t
LEGEND
• IMPORTS
—~<~ • MET IMPORTS
.»_ a EXPORTS
DATA SOURCE : ALUM. STATISTICS lv«9
ALUMINUM ASSOCIATION .
U.S.B.M. MINERALS YEARBOOKS
2-44
-------
The interrelationship among the interests of the
major producers, both foreign and domestic, increasingly
facilitates exchange of materials and metal, and tends
to result in end-use markets being supplied from the most
economical sources. Aluminum price stability has resulted
from a high degree of integration of the larger producers,
and this integration has extended through participatory
interests to all parts of the world.
It seems probable that the importance of imports
in contributing to domestic primary aluminum supply will
continue to grow and exert considerable influence on the
future expansion of domestic capacity. When the net im-
port position is compared to domestic primary aluminum
production over the past twenty years it is apparent that
foreign primary metal, once a modest part of the supply
picture, is finding an important market in the United
States.
2.7.4 Duties and Tariffs
U.S. trade policy has steadily reduced tariffs on
aluminum products, while eliminating the dutues on baux-
ite and alumina by renewals of temporary suspensions.
Other important aluminum-consuming countries, with few
exceptions, have limited reductions of their own duties
from high levels to levels still above the U.S. duties.
Some countries have imposed special taxes on imported
aluminum, either not applicable to domestic aluminum or
else higher than similar taxes imposed on domestic metal.
Protectionism for domestic aluminum industries has been
especially strong in Australia, India, Japan, France,
Italy and Brazil.
2.7.5 Internationalization
The international character of the industry in terms
of producers' interests extending beyond national bound-
aries has become increasingly marked during the past de-
cade, with joint ventures participating in new plants and
sharing in both material supplies and markets for outputs.
As an instance, on the primary production level alone, the
three major United States producers collectively have in-
terests in Brazil, India, Mexico, Surinam, Venezuela, the
United Kingdom, Norway, Ghana, Australia, and Canada.
Conversely, European producers hold interests in several
2-45
-------
0^38
United States operations. The world-wide character of
foreign investment in existing and planned primary plants
by some of the world's largest producers is indicated
below:
Foreign Production
Producer Investment-Countries
Alcoa 5
Reynolds 6
Kaiser 4
Alcan 8
Pechiney 5
Swiss Aluminum 8
This internationalization has many obvious ramifi-
cations affecting aluminum supply, fabrication, marketing,
and technical exchange, and influences production patterns,
economics, and inter-country movement of primary metal.
An instance of this is cited previously where it is noted
that the greater portion of U.S. imports of primary metal
to-date have represented shipments of Canadian produced
ingot to U.S. subsidiaries for fabrication into semi-
finished or finished products.
It is noted that the internationalization of the
aluminum industry which has gained momentum in the past
decade will undoubtedly have an effect on the extrapola-
tion of growth in the domestic primary aluminum production.
Requirements fo.r 'new capacity and expansion in the United
States will be affected by moves on the part of major pro-
ducers to add primary capacity outside of the United
States - near bauxite, for instance - in order to reduce
costs, and then import the resultant metal. If world
tariffs continue to decline, as they are expected to do,
this line of action would appear to be more and more
attractive.
There is certainly a further factor which may in-
fluence the growth of domestic capacity, that of increas-
ing local governmental pressure in areas of bauxite supply
to carry the conversion of ore through refined alumina to
primary metal within the national boundaries, in order to
obtain maximum value utilization of indigenous raw mate-
rials resources.
2-46
-------
0^39
To put this latter problem into perspective, the
most recent (1968) figures reported by the U.S. Bureau of
Mines indicate that only 6% of the U.S. metal output was
derived from domestically mined bauxite. The balance of
the primary aluminum production was derived from imports
of alumina (17%), or from alumina domestically produced
from imported bauxite (65%). It is doubtful that the
domestic production of bauxite, 95% of which is mined in
Arkansas, could be expected to supply 10% of the future
U.S. needs for alumina. Economic utilization of alternate
domestic sources of aluminum, such as non-bauxite clays,
is still in the future.
The nationalistic pressures noted above are partic-
ularly evident in the newly independent and undeveloped
countries in which an important portion of the world's
bauxite production comes; Jamaica, Guyana, Surinam, and
Guinea may be cited. All are forcing conversion of baux-
ite to alumina before shipment as a matter of government
policy, and some are pressing actively for further inte-
gration including metal production.
2.7.6 Projection of Growth Possibility
Considering the impact of the factors discussed
above on the supply-demand relationship of the domestic
industry in the future, it appears possible that the
average increase in future domestic production may be
within the limits indicated in Figure 2.6, derived from
Stamper's projections of demand and an assumption, be-
lieved to be conservative, that imports will supply as
much as 20% of domestic requirements of new metal by the
year 2000.
The gap between 1971 capacity and future production
requirements will be closed by capacity expansion on the
part of existing producers, as well as by the entry of a
certain number of new domestic producers attracted by the
expanding domestic market. Whether imports will expand
to the degree assumed, both proportionately and in total,
seems very possible.
The apparent ranges of increased domestic produc-
tion are to levels of 8.5-15 million short tons in 1984
and to 14-28 million short tons by the year 2000.
2-47
-------
36
34
32
30
FIGURE 2.6
UNITED STATES PROJECTIONS FOR
DEMAND, CAPACITY AND PRODUCTION
OF PRIMARY ALUMINUM
26
Z 24
i
* 72
at
< 20
18
o
:
w
O
I
14
12
10
I '
I I
T-T TT-r-r^r
r ' ' I ' ' ' ' I
HIGH AND 1OW ESTIMATES
PRIMARY METAL DEMAND-
CAPACITY AT 110% PRODUCTION—^^ \^
PRODUCTION* DEMAND-IMPORTS ~~V..'
i
/
I I I I I I I I I I I I I I I I I I I I I I I I 1
J I
1970
1975
1980
1985
YEAR
1990
1995
2000
ASSUMED TO GROW FROM 10% TO 20% OF DEMAND FROM 1*70 TO 2000.
2-48
-------
Table of Contents
Section 3
3.0 Technology of Aluminum Production
3.1 Reduction of Alumina by Hall-Heroult Process
3.1.1 Operations
3.1.2 Prebake Cells
3.1.3 Soderberg Cells
3.2 Carbon Plants
3.2.1 Paste Production Operations
3.2.2 Baking Operations
Ring Furnaces
Tunnel Kilns
3.3 Prebaked Anode Rodding
3.4 Cast House Operations
3.5 Fluoride Recovery from Cell Effluents
3.5.1 Dry Collection of Solids
3.5.2 Adsorption and Dry Collection
3.5.3 Wet Collection
3.6 Cryolite Production
3.6.1 Low Grade Cryolite Recovery
3.6.2 Standard Grade Cryolite Recovery
3.6.3 Cryolite Recovery Costs
3.7 Heat Generation
-------
3.0 Technology of Aluminum Production
An understanding of the technology governing the
production of aluminum, and of the mechanisms by which
pollutants are released from the processes, provides a
background for the evaluation of the problems reducing
effluents through adjustments of operating techniques
and the application of control systems best suited to
the various kinds of aluminum smelters.
3.1 Reduction of Alumina by Hall-Heroult Process
All of the production of primary aluminum metal in
the United States has been, and is, by the electrolytic
dissociation of alumina dissolved in a molten bath of
cryolite (Hall-Heroult process), and the following dis-
cussion is limited to that technology.
Several theories have been proposed to account for
physical changes which occur during the electrolysis of
aluminum oxide. However, the high reactivity of the com-
plex electrolyte combined with the high operating tem-
perature make it difficult to determine experimentally
which ions are present. Little is known about the exact
reaction mechanism beyond that it is complex and is var-
iable with both temperature and with the concentrations
of the several bath constituents.
Alumina (A1203) is dissolved in molten cryolite
(Na3AlF5) and is reduced to aluminum metal by direct cur-
rent electrolysis. The released oxygen rises through
the electrolyte and reacts with the sacrificial carbon
of the anode, while the molten aluminum settles to the
bottom of the reduction cell.
Cryolite bath is gradually lost from the reduction
cell through absorption in lining materials, electrol-
ysis, and vaporization. Although the quantity varies
among aluminum reduction plants, about 20 to 50 pounds
of cryolite must be added to the bath per 1000 pounds of
aluminum produced in order to make up for these losses.
The stoichiometric weight ratio of sodium fluoride
to aluminum fluoride in molten cryolite is 1.50 but ex-
perience has shown that maximum current efficiency for
the reduction of alumina occurs when this bath ratio is
3-1
-------
CM 3l
adjusted to fall between 1.30 and 1.45. Sodium impuri-
ties carried by the feed alumina accumulate in the cell
bath; they react with cryolite to form additional sodium
fluoride and thus increase the bath ratio. 3.1/ Two
additional factors act to deplete the bath of aluminum
fluoride or enrich it in sodium. The vapor pressure of
NaAlF4 over molten cryolite is approximately 30 times
that of NaF with the result that vaporized bath is de-
pleted in aluminum by vaporization. Also, cryolite,
Na3AlF6, will react with water to form NaF and Al2C>3,
which tend to remain in the bath, and HF which leaves
as a gas. Optimum bath ratio is maintained by the peri-
odic addition of aluminum fluoride to the bath at a rate
of approximately 0.2 to 0.7 pound of A1F3 per pound of
cryolite addition, or 10 to 30 pounds of aluminum flu-
oride per 1000 pounds of aluminum produced.
Table 3.1 presents approximate quantities of feed
materials required to produce 1000 pounds of aluminum by
the Hall-Heroult process as reported in response to the
industry questionnaire. These numbers vary from plant
to plant and should not be used in determining process
weight at specific plants.
Table 3.1
Feed Materials per 1000 Pounds of Aluminum
Alumina 1950 Ib
F (cryolite, aluminum
fluoride, fluorspar) 44 Ib
Calcined Coke 457 Ib
Pitch 138 Ib
Electric Power 7-8 MWH
3-2
-------
Fluoride Balance
Solids and gaseous fluoride effluents from an alu-
minum reduction cell constitute the most serious aspects
of the air pollution abatement problem. Their rates of
release vary over a wide range as a result of changes in
operating conditions and, more importantly, due to the
occasional "working" of the pot when the crust is broken
to admit alumina and cryolite to the molten bath or to
tap aluminum metal. Furthermore, over the useful life
of a pot lining,perhaps three years, a large fraction of
the fluoride bath materials added to the pot is absorbed
into the lining. These factors, together with inherent
imprecision in measuring effluents, have made it imprac-
tical in plant operation to determine accurate fluorine
balances around the reduction process.
Balances representing the order of magnitude of
long term distribution may be obtained by using data on
fluoride consumption and recovery, measured over a year
or more. One such balance, illustrative of an existing
prebake operating potline and its pollution control sys-
tem, is given in Figure 3.1. This balance indicates a
cell effluent of 11.8 pounds of solid "F" and 20.8 pounds
of gaseous HF released per 1000 pounds of aluminum pro-
duced. These values are higher than those of the indus-
try weighted averages shown in Tables 7.1, particularly
with regard to gaseous fluoride, and the differences may
illustrate the inherent imprecision in measurements of
this sort as well as the variations among operating
plants.
Out of the approximately 43.4 pounds of "F" shown
as being added to the potline as various solid salts,
approximately 32.6 pounds is released from the cell in
airborne effluents. Of this, 27.2 pounds is captured
by the cell hood collection system for a total "F" efflu-
ent collection efficiency of 83.5 percent. The remain-
ing 5.4 pounds, 16.5 percent, escapes to the potroom and
leaves by way of roof monitor ventilators. These and
other efficiency characteristics of this particular
plant air pollution control system example are shown in
Table 3.2
In constructing this material balance, the cell in-
put of raw materials and of cyclone collect were deter-
mined from annual measured usages. Fluoride in the
3-3
-------
FIGURE 1.1
ILLUSTRATIVE
POTROOM FLUORINE BALANCE
PREBAKED ANODE POTLINE
BASIS-POUNDS FLUORINE PER 1000 POUNDS ALUMINUM
CRYOLITE
Na3 Al
22 Ib Solid
1.4 Ib Solid
MULTICLONE
COLLECT
AljO3 and Bath
12 Ib Solid
8 Ib Solid
INPUT 43.4 Ib F as Soli
POTROOM
Loss 10.8 Ib Solid
010 Solid
1J0.8 Solid
(Y) Measured Quantities
(2) Derived Quantities
(V) Control equipment removal efficiencies
are assumed values.
(4) Hooding Efficiency
for Gas 86%
Solid 79%
EFFLUENT
11.8 Ib Solid
20.8 Ib Gas
(V)2.5 Solid
02.9 Gas
Released
to Potroom
(uncollected
but sampled)
(2)9.3 Solid
(5)17.9 Gas
(4) Cell Collection
Removed
(D 1.0 Solid
1.1 Gas
Released
to Atmosphere
3-4
-------
DM*
c
Table 3.2
Potroom Fluorine Balance Control Efficiencies
(Prebaked Anodes with Multiple Cyclones and
Scrubbers on a Primary Effluent Collection System
Reference Figure 3.1)
Indicated Efficiencies, %
Primary Cyclones Scrubber Primary Overall
Collection Removal Removal Removal Control
Gaseous HF 86 0 94 94 81
Total Solids 79 85 25 89 70
Total Effluent 84 29 89 92 77
3-5
-------
recovered carbon of spent butts and potlinings was deter-
mined by averaging chemical analyses of these materials.
Quantities of solid and gaseous F escaping the potroom
through roof monitors were measured by means of a regular
sampling program and other flow rates were derived from
assumed removal efficiencies of control equipment and by
differences to establish a balance. This illustrative
fluoride balance points out the fact that only a part of
the consumed fluoride is lost as an effluent; a substan-
tial fraction is absorbed in the cell lining and insula-
tion.
Potline Configuration
The reduction cell, or pot, is a strongly rein-
forced steel box, lined with heat insulation and with pre-
baked carbon blocks or a rammed monolithic carbon liner
inside the insulation. The carbon liner forms the cath-
ode of the electrolytic cell and provides high electrical
conductivity and good corrosion resistance to the highly
reactive molten electrolyte. The carbon lining contains
steel electric current collector bars that extend through
the sides of the steel shell and are connected to a ring
collector bus, which is, in turn, connected to the main
bus which is usually made of aluminum bars, serving as
the electrical connection to a line of cells connected in
series.
The anode, also made of carbon, is suspended over
the steel pot shell and is immersed in the molten elec-
trolyte. It is connected to the main bus system through
flexible conductors.
From 90 to 180 reduction cells are linked together
electrically in series to form a potline, the basic pro-
duction unit of the reduction plant. Figure 3.2 shows a
schematic electrical diagram. A typical late design pot-
line may consist of 180 cells connected in series and
operating at 100,000 amperes and 830 volts, about 4.5
volts drop per cell. Such a potline operating at 83,000
KW would produce approximately 275,000 pounds of aluminum
per day with an energy consumption of approximately 7.2
kilowatt hours per pound of aluminum produced.
3-6
-------
SCHEMATIC ELECTRICAL DIAGRAM
ALUMINUM REDUCTION CELLS
140 CELLS CONNECTED IN SERIES
tHb Jr
f
jT—Tl
S H
J:
3
J
SCHEMATIC ELECTRICAL DIAGRAM
ALUMINUM REDUCTION CELL
FLEXIBLE CONNECTOR
STEEL
ELECTRODES
t
DIRECT CURRENT
CARBON ANODE
CRYOLITE WITH
ALUMINA ELECTROLITE
ALUMINUM PAD
ALUMINA
CRUST
<•— CARBON CATHODE
INSULATED FLOOR
3-7
-------
Potline configuration, cell types, and cell dimen-
sions vary according to the design and capacity of the
individual aluminum reduction plants, and cell modifica-
tions exist within single plants reflecting development
of design as capacity expansion has been constructed.
The rectangular bath cavities of modern reduction
cells are usually about 10 feet wide and 36 feet long,
varying a few feet either way depending on cell type,
method of crust breaking and production capacity. Inher-
ent economies associated with the construction of very
large cells are offset by structural problems resulting
from the swelling of pot linings as they absorb bath ma-
terial and by operating problems associated with the
strong magnetic fields which accompany large electric
currents. Recently designed reduction cells range from
about 100,000 to about 180,000 ampere capacity, producing
1540 to 2800 pounds of aluminum per day.
Reduction cells are of two basic types, the pre-
bake cell using prebaked carbon multiple anodes, and the
Soderberg cell using one large self-baking anode.
3.1.1 Operations
The cell cavity contains molten bath consisting
of approximately 85 percent cryolite (Na3AlF6), 8 to 10
percent fluorspar (CaF2) and 2 to 5 percent alumina
(Al2C>3) . The molten bath is covered by a crust of frozen
electrolyte and alumina. This crust both diminishes heat
loss from the top of the cell and protects the anode from
oxidation. Periodically part of the crust is broken and
stirred into the bath, and fresh alumina is added to
cover a newly formed crust.
The electric current decomposes the alumina in
solution in the bath. Aluminum is deposited as molten
metal on the bottom of the cell and the oxygen is lib-
erated at the surface of the anode where it reacts to
form carbon dioxide which is released in the cell gases.
The aluminum at the temperature of operation of the cell
(about 970°C) is slightly more dense (2.3 gms/cc) than
the molten bath (2.1 gms/cc) and thus forms a metal pad
on the bottom of the cell. A small portion of the molten
aluminum mixes with the bath and is carried to the anode
by the circulation of the bath. Here it is oxidized,
reducing some of the carbon dioxide to carbon monoxide.
3-8
-------
0^50
According to Faraday's law, 1000 amperes should produce.
17.76 pounds of aluminum per day per cell. In .practice,
only about 15.4 pounds of metal is produced, so that the
current efficiency is about 87 percent and the anode
gases contain from 25 to 30 percent carbon monoxide.
The theoretical decomposition voltage of alumina
to yield aluminum and carbon dioxide is 1.68 volts. Owing
to the resistance of the electrolyte, the carbon lining
and the electrical connections, the cell operates at
about 4.5 volts resulting in an effective energy effi-
ciency about 35 percent. The remaining 65 percent of the
electrical energy input is converted to heat, maintain-
ing the cell at its operating temperature of about 970°C.
Normally the cell operates with about 2 to 5 per-
cent of alumina in solution, in the bath, but as the
electrolysis proceeds the alumina content is decreased,
being intermittently replenished by feed additions. When
this content falls to about 1.5 to 2.0 percent the phenom-
enon of an "anode effect" occurs. It is believed that
at this alumina concentration the bath fails to wet the
carbon anode and a gas film collects under the anode.
This film causes a high electrical resistance and the
normal cell voltage increases 10 to 15-fold. To correct
the condition the cell crust must be broken and more alu-
mina added to bring the concentration back to its normal
content. The gas film under the anode is dispersed and
the cell returns to normal voltage.
Whereas the cell bath solidifies at the top and
around the sides, forming an insulating crust, the molten
aluminum on the bottom of the cell extends under the
crust to the cell lining and, therefore, provides a
relatively low resistance heat leak through the pot to
the surroundings. The depth of molten metal is usually
regulated in the range of 4 to 10 inches in order to
regulate the operating temperature of the cell by means
of this adjustable heat leak. Operating temperature
can be held constant in the face of increased cell cur-
rent by allowing the depth of molten metal to increase
and thus increase heat leak to the building.
Metal accumulates in a 100 kiloampere cell at the
rate of about 1540 pounds per day. At suitable inter-
vals, usually daily, a large thermally insulated steel
3-9
-------
crucible is brought to the cells. This crucible has a
thermally insulated airtight steel cover with a cast iron
siphon attached. The siphon is placed in the cell so
that it reaches the bottom of the cell cavity, immersed
in the molten metal pad lying on the bottom of the cell.
A vacuum is applied to the steel crucible and molten
metal is drawn up into the crucible. The amount of met-
al taken out of a cell is usually about the amount the
cell produced in the last twenty-four hour period. The
crucible usually taps three or more cells before the
steel lid is removed and molten metal in it is "skimmed"
to remove lighter bath that might have been tapped with
the metal. The crucible is then transported to a cast
house where the metal is either poured into a reverbera-
tory holding furnace for casting into various shapes or
is poured directly from the crucible into cast iron
molds to form "pigs" or "sows". A pig is usually a 50
pound piece of aluminum while a sow varies from 700 to
2000 pounds.
3.1.2 Prebake Cells
Modern prebake cells use a number of anodes sus-
pended in the electrolyte. The anodes are press-formed
from a carbon paste and are baked in a ring furnace or
tunnel kiln.
A mixture of coke and pitch is pressed or vibra-
tion molded into blocks which are baked at approximately
1200°C to drive off volatiles and to coke the coal tar
pitch, cementing the anode into a strong block. The
baked anode blocks are moved to a rodding plant where
steel stub electrodes are bonded into preformed holes in
the blocks. Completed anode assemblies are delivered to
the potlines, ready for the replacement of consumed an-
odes.
Figure 3.3 shows a sectional view of a typical
prebake reduction cell with a hood for cell effluent col-
lection.
The newer design prebake cells use up to twenty-
six anode assemblies per cell, attached to the anode bus
on the cell superstructure by means of clamps. The an-
ode bus is attached to the steel superstructure by anode
jacks which may be driven by an air motor or other means,
3-10
-------
FIGURE 3.3
PREBAKE REDUCTION CELL
SCHEMATIC ARRANGEMENT
ALUMINA (ORE) BIN
CRUST BREAKER
RISER BUS TO
NEXT CELL
ANODE BUS
ANODE ROD
CLAMP
IP 41
s %
CARBON
ANODE
#• • - • + ,
CARBON LINING
SIDE HOOD FOR
VENT CONTROL
ALUMINA
CRUST
CRYOLITE BATH
MOLTEN
ALUMINUM
CATHODE
RING BUS
FLOOR
-ALUMINA INSULATION
STEEL CRADLE
STEEL CATHODE
COLLECTOR BAR
3-11
-------
/~ ' •
•140
giving a travel of from 10 to 14 inches and permitting
the raising or lowering of all twenty-four assemblies in
the cell simultaneously. Each of the twenty-six assem-
blies may also be raised or lowered individually by means
of an overhead crane after the anode clamp is loosened.
The anode assemblies are usually installed in two
rows extending the length of the cell. In some arrange-
ments the two rows are closely spaced in the center of
the cell, providing a working area on each side of the
cell between the cell side lining and the anodes. In
other cases the rows are separated and placed closer to
the cell side lining, providing: the working area in the
center of the cell between the rows of anodes.
The sacrificial carbon anodes are replaced period-
ically by new anode assemblies, the total operating time
being dependent on the size of the anode blocks and the
amperage of the potline.
The general trend in prebaked anode design has
been toward larger anode blocks, obtaining greater effec-
tive anode/cathode surface ratios and lower current den-
sities at the anodes for equivalent power inputs.
3.1.3 Soderberg Cells
There are two types of Soderberg cells, each hav-
ing a single large carbon anode, but differing in the
method of anode bus connection to the anode mass. They
are termed the horizontal spike suspension (HSS) Soder-
berg and the vertical spike suspension (VSS) Soderberg.
In both, the anode material is a green anode paste which
is fed periodically into the open top of a rectangular
steel compartment and baked by the heat of the cell to a
solid coherent mass as it moves down the casing. This
casing is mounted on the steel superstructure of the cell
and is raised or lowered by means of powered jacks. Green
paste is added to the upper section to replenish the an-
ode as it is consumed.
In the HSS pot, rows of steel studs in channel
assemblies project laterally into the paste and move with
the anode. They transmit current into the baked anode,
and are extracted when, as a result of the progressive
consumption of the anode, they have been moved down to a
point where they are close to contacting the molten bath.
3-12
-------
In the VSS Soderberg, steel current-carrying studs
project vertically into the anode through the unbaked
paste portion and into the baked portion of the anode.
These steel studs are also extracted before they are ex-
posed at the bottom of the anode.
In both cases the steel studs are connected to an
anode bus that transmits electric current from the main
bus.
Figures 3.4 and 3.5 show schematic diagrams of the
two Soderberg cell designs.
In both types of Soderberg cells the in-place bak-
ing of the anode paste results in the release of hydrocar-
bon fumes and volatiles derived from the pitch binder of
the paste mixture. These products are a component of the
Soderberg cell effluents and are essentially absent from
those of the prebaked cells. Their tarry nature requires
modification of the control treatment techniques applied
to the effluents, as it interferes with pollutant remov-
al devices. With VSS Soderberg cells this modification
involves the combustion of the collected hydrocarbon
fumes at the cell by means of a burner, converting the
tars to gaseous fractions which do not interfere with the
operation of subsequent control devices.
The construction of the VSS Soderberg cell with
respect to electrode positioning is such that gas collec-
tion skirts may be installed between the anode casing
and the bath surface, and effluents may be ducted to a
burner and the following control devices without inter-
ference with cell operations. However, the requirement
for side-working the cell results in surface areas with
uncontrolled exposure during the crust breaking and charg-
ing operations, with accompanying losses of cell efflu-
ents from the unhooded surfaces during these operations.
The construction of the HSS Soderberg cell pre-
vents the installation of an integral effluent collection
device such as a gas skirt, since the anode casing is
formed by removable sections supporting the horizontal
spike electrodes, and these sections are periodically
changed as the anode moves downward and is consumed. Hood-
ing is restricted to canopy suspension with removable
side panel closures, resulting in so much air dilution
3-13
-------
0^55
FIGURE 3.4
HSS SODERBERG CELL
SCHEMATIC ARRANGEMENT
ALUMINA HOPPERS
PASTE COMPARTMENT
COVER
REMOVABLE
CHANNELS
ALUMINA
CRUST
STEEL SHELL
INSULATION
CARiON LINING
PASTE COMPARTMENT
CASING
MOLTEN
ALUMINUM
POT ENCLOSURE
DOOR
ANODE
STUDS
GAS AND FUME
EVOLVING
CATHODE
COLLECTOR BAR
3-14
-------
FIGURE 3.5
VSS SODERBERG CELL
SCHEMATIC ARRANGEMENT
0^56
ANODE BUS
ANODE ROD
STEEL ANODE
STUD
ANODE CASING
GAS COLLECTING
SKIRT
MOLTEN
ELECTROLYTE
TO EFFLUENT
COLLECTION
SYSTEM
PARTIALLY BAKED
ANODE PASTE
CARBON
BLOCK
LINING
CRUST
ALUMINA '
MOLTEN ALUMINUM
CATHODE
STEEL CATHODE
COLLECTION BAR
STEEL
CRADLE
CATHODE BUS
3-15
-------
:QU57
of collected effluent that self-supporting combustion in
burners is not possible.
As in the VSS Soderberg the side-working area in-
herent in the cell design complicates the problem of effi-
cient effluent collection.
The relative advantages and disadvantages of the
different types of cells are not universally agreed upon
within the primary aluminum industry. In the United
States, the larger number of installations employ pre-
baked anode cells. No new VSS Soderberg plants and only
one new HSS Soderberg plant was built between 1959-1970,
although expansions were carried out in existing Soder-
berg plants using the existing types of cells. One VSS
Soderberg plant was under construction in 1970.
Table 3.3 following summarizes this trend in the
United States.
3.2 Carbon Plants
Reduction plant anodes and cathodes are made from
anthracite and/or calcined petroleum coke, bonded by
pitch and baked to form solid carbon masses.
The preparation of this electrode material in one
form or another is usually carried out as an ancillary
operation at the reduction plant site. In the United
States there is only one reduction plant in which, be-
cause it is supplied with fabricated prebaked anodes made
at another company-owned facility, the carbon plant is
not a part of the on-site operation to supply anode mate-
rial. Cathode material in the form of prebaked block may
be supplied from off-site sources, but usually is pre-
pared on-site in the form of paste used for monolithic
rammed lining.
3.2.1 Paste Production Operations
Carbon paste preparation consists of crushing,
grinding, screening and classifying, combining of care-
fully sized fractions with a pitch binder, and mixing.
The preparation plant is termed the "green mill" by the
industry and may produce anode paste for Soderberg cells,
cathode paste, or green pressed anodes for prebake treat-
ment.
3-16
-------
Table 3.3 U. S0 TRENDS IN ADOPTION OF CELL TYPES
U)
Soderberg Plants
Prebake Plants
Installations Expansion
New Cumulative
Pre 1946
1946
1947-1950
1951-1958
1959-1965
1966-1969
1970
19713-/
197 2^
-
(-3)i/
ll/
4
1
1
3
2
1
10
7
8
12 2
13
14 1
17
19
20
HSS
Installations Expansion
New Cumulative
2
2
2
3 5 2
16-
6 2
6
6
6
VSS
Installations
New Cumulative
-
-
-
3 3
3
3
3
3
1 4
Expansion
-
-
-
-
2
-
-
_
I/ - 3 Government-owned (DPC) plants deactivated.
2j - 1 DPC plant reactivated under private ownership.
3/ - Under construction.
-------
0^59
Figure 3.6 shows a typical flowsheet for a Soder-
berg paste plant.
Figure 3.7 shows a typical flowsheet for the paste
preparation and green anode pressing of prebake anodes.
Forming of the green anodes is accomplished either by hy-
draulic molding or vibratory jolting of the stiff anode
paste into dimensionally stable blocks ready for baking
and rodding.
Solid raw materials (calcined petroleum coke, an-
thracite coal, solid pitch, and green petroleum coke, as
required for various kinds of paste mixes) are received
in bulk and conveyed to carbon plant storage. Wetting
agent sprays are used in some green mills to reduce dust-
ing conditions inherent in materials handling.
Material is reclaimed from storage, usually by
front-end loaders with enclosed cabs, and fed to combi-
nations of crushing equipment in closed circuit with
vibrating screens followed by grinding units. Sized
fractions of crushed and ground material are separated
and stored in mix bins for make-up of paste composition.
Cleaned reclaimed spent anodes and anode scrap
from prebake plant operations are similarly crushed and
sized for recycle to prebake anode preparation.
Dry solids are drawn from the mix bins in weighed
proportions to provide batches of carefully controlled
size distribution and composition, which are then trans-
ferred to steam jacketed hot mixers. For baked anode
pastes the mixer feed contains either solid crushed coal
tar pitch which is softened and blended in the mixers or
hot liquid pitch to provide the paste binder. For Soder-
berg paste, a liquid pitch is used, metered to the mixers.
The hot Soderberg paste is discharged directly
from the batch mixers to transfer cars which convey it
to the cell rooms for anode replenishment, or may be
cooled and briquetted.
The prebake paste, less fluid than the Soderberg
material, is transferred from the mixers to anode molds,
in which the self-supporting green anode is formed by
compaction.
3-18
-------
FIGURE 3.6
SODERBERG ANODE PASTE PLANT FLOWSHEET
CALCINED
PETROLEUM COKE
STORAGE
U>
I
STEAM
LIQUID PITCH
STORAGE
PUMP
• ALL MILL
METER
TRANSFER CART
CD
-Sr
O")
CD
-------
FIGURE 3.7
PREBAKE ANODE PREPARATION FLOWSHEET
10
o
CD
-r
CD
COO1INO
CONVITOI
PIT •AKINO
FURNACI
MOltINO
PRISS
-------
0^62
3.2.2 Baking Operations -_ Ring Furnaces
Green anodes are delivered to the baking plant,
in most of which the furnaces are ring type, sunken,
baking pits with surrounding interconnecting flues.
Anodes are packed into the pits, with a blanket
of coke or anthracite filling the space between the an-
ode blocks and the walls of the pits. A 10 to 12 inch
blanket of calcined petroleum coke fills the top of each
pit above the top layer of anodes.
The pits are heated with natural gas or oil fired
manifolded burners for a period of about 40 hours. The
flue system of the furnace is arranged so that hot gas
from the pits being fired is drawn through the next sec-
tion of pits to gradually preheat the next batch of an-
odes before they are fired, in turn, when the manifold
is progressively moved. The anodes are fired to approx-
imately 1200°C, and the cycle of placing green anodes,
preheating, firing, cooling, and removal is approximate-
ly 28 days.
The ring type furnaces use outside flues under
draft, and since the flue walls are of dry type construc-
tion, most volatile materials released from the anodes
during the baking cycle are drawn, with the combustion
products of the firing, into the flue gases.
Flue gases may be passed through scrubbers and
perhaps electrostatic precipitators to reduce temperature
and scrub or precipitate out a portion of the hydrocar-
bons before exhausting to a stack.
The furnace buildings spanning the lines of baking
pits are usually open at the side and ventilated through
gravity roof monitors without emission controls.
The baked anodes are stripped from the furnace
pits by means of an overhead crane on which may also be
mounted pneumatic systems for loading and removing the
coke pit packing.
3-21
-------
0^63
Baking Operations - Tunnel Kilns
A second type of furnace, the tunnel kiln, has
been developed for application in the baking of anodes.
The kiln is an indirect fired chamber in which a con-
trolled atmosphere is maintained to prevent oxidation
of the carbon anodes. Green anode blocks are loaded on
transporter units which enter the kiln through an air
lock, pass successively through a preheating zone, a fir-
ing zone, and a cooling zone, and leave the kiln through
a second air lock. The refractory beds of the cars are
sealed mechanically to the kiln walls to form the muffle
chamber, and yet permit movement of the units through the
kiln.
The muffle chamber is externally heated by combus-
tion gases, and the products of combustion are discharged
through an independent stack system.
Effluent gases from the baking anodes may be in-
troduced into the fire box so as to recover the fuel
value of hydrocarbons and reduce the quantity of unburned
hydrocarbon to approximately one percent of that coming
from a ring furnace. Further reduction of solid and gas-
eous effluent may be achieved by the use of heat exchang-
ers, scrubbers and electrostatic precipitators.
While the tunnel kiln presents mechanical problems
in design and operation, it is reported to have several
appreciable advantages over the ring type of furnace.
Baking cycle from green to finished anode is much short-
er. Anode baking is more uniform. Space requirements
for equal capacity furnaces is less. Smaller gas volumes
are handled through the furnace emission control system.
The successful development of the tunnel kiln in
this application is recent, and to date only one installa-
tion is in normal operation.
Baked anodes are delivered to air blast cleaning
machines utilizing fine coke as blasting grit. Fins,
scarfs, and adherent packing is removed by this treatment,
and the baked anodes are then transferred to the rodding
room.
3-22
-------
3.3 Prebaked Anode Rodding
Anode assemblies returned from the cell room, after
initial separation of the spent butts, are delivered to
the rodding room where the thimbles forming the connec-
tion between the anode blocks and the current carrying
rod supports are cracked off and the rod stubs cleaned
by grit blasting.
In the green anode forming operation depressions
had been molded into the top surfaces of the blocks to
receive the rod stubs.
Cleaned baked anode blocks are transferred from the
bake plant storage area on roller conveyors to the rod-
ding area for make up into rodded anode assemblies.
Rod yoke assemblies, supported by overhead convey-
ing mechanisms, are indexed in position over anode blocks
and connected to the blocks by pouring a cementing mate-
rial, usually cast iron as a thimble in the holes around
the rod stubs. After rodding, the anodes may, in some
plants, be sprayed with a metallic aluminum coating.
The completed rodded anode assemblies are then
stored for later transfer to the potrooms.
3.4 Cast House Operations
Molten aluminum metal is syphoned from the reduc-
tion cells into transfer crucibles, sampled, and conveyed
to the cast house, where it may be placed directly in
gas-fired holding furnaces, or may be cast into pigs or
large sows for later remelt.
Primary metal, after fluxing for removal of minor
impurities such as oxides, bath electrolyte, and gas
inclusions, may be cast into a variety of ingot forms,
including 50 pound unalloyed or alloy ingot, 30 pound
casting alloy ingot, or cast extrusion billet, sheet in-
got or .forging alloy ingot. Cast ingot may be produced
continuously from horizontal direct chilled casting ma-
chines or intermittently from vertical direct chilled
casting machines; other forms are produced on casting
wheels or in-line casting machines.
3-23
-------
Metal may also be shipped in the molten state
directly from the reduction plant to a customer's plant
in insulated ladles.
3.5 Fluoride Recovery from Cell Effluents
Every thousand pounds of aluminum produced from
alumina requires the replacement to the cell bath of ap-
proximately 44 pounds of fluoride, primarily as cryolite
(Na3AlF5) and aluminum fluoride (A1F3) . These salts
serve only as the bath in which alumina is dissolved, and
this 44 pounds is lost from the reduction cell mainly
through evaporation and dusting from the cell surface.
Cryolite and aluminum fluoride presently cost approximate-
ly $260 and $360 a ton respectively, approximately $0.25
per pound of F, and their consumption or loss contribute
significantly to the cost of producing aluminum. Several
methods are used to recover fluoride values which escape
from the cell bath as effluents.
3.5.1 Dry Collection of Solids
Some smelters employ multiple cyclone dust collec-
tors on the effluent streams from prebake or VSS Soder-
berg reduction cells to capture a major portion of the
solids which are carried away from the' cells in primary r
effluent collection systems. These particulates consist
mainly of alumina but they may contain about nine pounds
of F per thousand pounds of aluminum produced, in the
form of cryolite, aluminum fluoride and calcium fluoride.
Collected in a multiple cyclone separator, these solids
may be returned directly to the reduction cells to make
up a substantial fraction of the 44 pounds of F required.
Figure 3.1 of this section shows an F material balance
around a reduction cell in which eight pounds of F is
returned to the cell as a multiple cyclone collect.
3.5.2 Adsorption and Dry Collection
Normally, a reduction cell releases nearly twice
as much gaseous F as solid F, in the order of 11 to 18
pounds per thousand pounds of aluminum produced, so that
its recovery can be even more economically beneficial
than the recovery of solid F. Several process systems
are in use in which particulate fluorides are captured
dry and gaseous fluoride, HF, is adsorbed on alumina and
returned to the cell in this form.
3-24
-------
3.5.3 Wet Collection
Wet scrubbers in various forms are used to remove
gaseous and particulate fluorine from potline effluents.
Liquors from these scrubbers, when used with fluoride-
bearing spent pot linings or insulation materials, can
serve as raw material to plants which produce synthetic
cryolite or aluminum fluoride.
3. 6 Cryolite Production
As discussed in the preceding section, one route
for the recovery of fluorides lost from the reduction
cell bath is to capture them in scrubber liquor, combine
this with spent pot lining and insulation materials and
make synthetic cryolite from these raw materials. Any
discussion of fluoride recovery through this route must
of necessity include consideration of the various types
of cryolite that may be produced and the effects of their
uses in aluminum smelting.
There are basically three types of synthetic cryo-
lite; a high purity type containing approximately 95 per-
cent cryolite, a standard type, approximately 90 percent
cryolite, and a low grade type containing approximately
50 percent cryolite. Most of the balance is alumina.
High Purity Cryolite
This cryolite contains only a few percent free
alumina and very low iron and silica impurities (less
than 0.20 percent each).
High purity cryolite may be used on any reduction
cell, even those producing 99.90 percent purity aluminum,
and it will have little if any degrading effect on the
metal purity. This grade of cryolite is the most expen-
sive and its production from cell effluents is not
possible without the use of purchased hydrofluoric acid
because of the impurities contained in the effluents.
Standard Grade Cryolite
This cryolite contains approximately six percent
free alumina and total iron and silicon impurities in
the range of 0.70 percent maximum.
3-25
-------
CkST
Standard grade cryolite may be used throughout
most reduction plants with little if any adverse effects
on metal grade or cell operation. Only requirements for
highest purity aluminum would limit the use of standard
grade cryolite. This grade may be produced from cell
effluents using equipment which is more expensive and
more complicated than that necessary to produce low grade
cryolite.
Low Grade Cryolite
This cryolite contains approximately 50 percent
free alumina and significant amounts of iron and silica
impurities. For this reason, low grade cryolite has
limited utility in a reduction plant. Upon addition to
a reduction cell, the cryolite is purified. That is, the
metallic impurities pass into the molten aluminum in the
bottom of the cell leaving the molten electrolyte low in
metallic impurities. This phenomenon makes practical the
use of the reduction cell^ primarily for the purpose of
purifying low grade cryolite.
Several cells in each cell room may be selected as
bath cells. They will be fed the low grade cryolite on
a routine basis, and they will be "bath tapped". Bath
tapping is done in a method similar to tapping molten alu-
minum from the reduction cell with the exception that the
cast iron siphon is held above the molten aluminum pad.
The bath is tapped and is transferred to cast iron tubs
where it is allowed to solidify and cool. It is then
dumped from the tubs and broken into pieces. It is term-
ed "purified" bath and may be used on any reduction cell
in the plant.
The rate at which low grade cryolite may be puri-
fied depends upon the number of bath cells that are used
and upon the free alumina content of the cryolite being
used.
The number of bath cells is usually limited by the
amount of high grade metal the plant is trying to produce
and by the poorer operating efficiency of bath cells. Of
necessity bath cells must be operated at higher voltages
than normal cells and they produce a much lower grade alu-
minum. They have a tendency to run at higher cell tem-
peratures and because of the high free alumina content of
3-26
-------
the low grade cryolite, they have a tendency to collect
muck on the bottom of the cells. When this happens,
they must be removed as bath pots until the cell is
cleared of the muck. The muck consists of undissolved
alumina and causes a higher electrical resistance in the
cell than that found in normal cells. Mucky, or "sick"
cells as they are sometimes called, operate at much low-
er ampere efficiency than normal cells and are therefore
uneconomical aluminum producers.
It is usually not economical to operate a reduc-
tion plant solely on low grade cryolite. Depending upon
the grade of aluminum that the plant is required to pro-
duce, some combination of low grade and standard or high
grade cryolite will be used.
An evaluation of the economics involved in recover-
ing the fluorine value in the cell effluents as either
standard or low grade cryolite requires a careful study
of their costs relative to their utilities in the reduc-
tion process.
Another factor to be considered is whether the cell
effluents contain hydrocarbons or if they do, whether or
not they can be easily removed. The following paragraphs
describe processes to recover fluorine from scrubber ef-
fluents in the absence of hydrocarbons. If the scrubber
liquor contains hydrocarbons, a different process flow
design must be used to produce standard grade cryolite.
3.6.1 Low Grade Cryolite Recovery
Figure 3.8 shows a block flow diagram of a low
grade cryolite recovery system. Potline effluent scrub-
ber liquor containing dissolved fluorides is fed to a
reaction tank where sodium aluminate is added. This
precipitates cryolite from the liquor and the slurry is
fed to a clarifier. The overflow from the clarifier is
returned to the potroom scrubbers and the underflow,
containing the cryolite, is fed to a classifier, a filter,
and a roaster for drying.
Spent pot lining and spent alumina insulation also
may be crushed, ground fine and fed to the roaster.
3-27
-------
G!
FIGURE 3.8
LOW GRADE CRYOLITE
PROCESS BLOCK FLOW DIAGRAM
SODIUM
ALUMINATE
POTLINE
SCRUBBERS
//
OVERFLOW
ALUMINA INSULATION
SPENT POTLININGS
No Al O2
BATCHING
REACTION
TANK
UNDERFLOW
CRUSHING
GRINDING
1
IFIER
I
/
(
\
ROASTING
CRYOLITE
STORAGE
3-28
-------
The product of the roaster is a low grade cryolite
containing approximately 50 percent cryolite and 50 per-
cent alumina. No information is available on pollution
controls associated with the roaster operations.
3.6.2 Standard Grade Cryolite Recovery
Figure 3.9 shows a block flow diagram for one
process producing standard grade cryolite from scrubber
liquor and spent pot linings.
Spent pot lining material containing cryolite is
crushed and ground to about minus 1/2 inch mesh and is
reacted in the digester with caustic to produce soluble
sodium aluminate (Na2Al0.) with sodium fluoride (NaF) .
Slurry from this digestion process is washed in
a mud wash thickener to recover entrained fluoride.
Liquor from the thickener containing dissolved
aluminum, sodium, and fluoride values, is adjusted to
the stoichiometric ratio of cryolite by the addition of
sodium fluoride from scrubber water treatment and is
acidified by reacting with carbon dioxide, precipitating
cryolite from a sodium carbonate solution.
A filter separates cryolite product from sodium
carbonate spent liquor and the cryolite, now about 90
percent pure, is dried for storage.
Part of the spent liquor is used to dissolve
fluorides in the scrubber water and the remainder is
reacted with lime to produce caustic for recycle to
digestion.
3.6.3 Cryolite Recovery Costs
Figure 3.10 shows the relationship between esti-
mated capital cost to erect a standard grade cryolite
recovery plant and its production capacity. The curve
was derived by using in-house figures for the capital
cost of a cryolite recovery facility in 1968. The plant
capacity is 8,000 tons of standard grade cryolite per
year. The costs were escalated to 1970. The cost of
3-29
-------
"7 1
S 1
FIGURE 3.9
PROCESS BLOCK FLOW DIAGRAM FOR
STANDARD GRADE CRYOLITE RECOVERY
POTLINE
SCRUBBERS
SPENT
POTLINING
/I \ V
CAUSTIC
SODA
SCRUBBER
LIQUOR^
LIME
REACTION
TANK
CLARIFIER
LIME
STORAGE
SLAKER
CRUSHING
GRINDING
SCRlENING
CAUSTIC
STORAGE
STORAGE
DIGESTER
STEAM
J
CO,
CAUSTICIZER
THICKENER
CARBONATOR
FILTER
DRYER
-•»•• TO CaCO3 POND
3-30
-------
FIGURE 3.10
CAPITAL COST CRYOLITE PLANT
VS
PRODUCTION CAPACITY
7 2
ANNUAL CRYOLITE PRODUCTION CAPACITY, 1000'S SHORT TONS
SOURCE-SINGMASTER & BREYER ESTIMATE
3-31
-------
the plants with less than 8,000 tons per year capacity
were calculated by using the following equation:
Capital Cost = Constant + X(
0.6
8000'
where: Constant = Engineering cost and fixed indirect
construction costs.
X = Labor and material costs to con-
struct a plant of 8,000 tons per
year capacity
Y = Capacity of other plants, tons per
year
Figure 3.11 shows the unit production cost of cryo-
lite for plants of varying capacity. The double scale on
the X-axis shows the relationship between the capacity of
the cryolite recovery facility and the production of a
reduction plant that would be necessary to supply the
spent pot lining and scrubber liquor to sustain the op-
eration of the cryolite facility. Production costs in-
clude labor: maintenance and services; raw materials; tax-
es and insurance; and depreciation and interest.
Labor, maintenance, and service costs were esti-
mated from the actual- experience of an 8,000 ton per year
plant.
Raw material costs were calculated from the ton-
nages involved and the market price of the materials.
Spent pot lining was priced at $180 per ton of contained
fluorine.
Taxes and insurance were estimated at two percent
and depreciation and interest at sixteen percent of cap-
ital costs.
3.7
Heat Generation
All aluminum smelters include a heat generating
plant; a few also generate electric power. All but one
of the United States plants normally burn natural gas in
their boilers, a fuel which emits negligible pollutants.
The one exception is a small powdered coal unit, 33 x 10
Btu per hour, which is used only for peak winter heating.
3-32
-------
450
H 400
g
Of
U
350 -
Z
o
O
-------
01 "'
UT I
Among the 31 heating units reported in the Indus-
try Questionnaire, the unit sizes range from 4 x 10^ to
117 x 10^ Btu per hour with a weighted average about
30 x 10^ Btu per hour. None uses control equipment for
emissions other than stacks ranging from 35 to 175 feet
tall.
Although most smelters purchase their electric
power from utility companies, one generates power with
gas-fired steam boiler-turbines and gas turbines and a
few produce power from gas-fired piston engines.
3-34
-------
0^76
Reference - Section 3
Garcia, A. F., and Lewis, R. H., with Reese, K. M.
editor, "Aluminum—Light Metals King", Industrial
and Engineering Chemistry 4^7 (10) : 2066-2072
(October 1955).
3-35
-------
GW7
Table of Contents
Section 4
4.0 Sources and Characteristics of Effluent Releases
4.1 Potline Effluents
4.1.1 Composition
Particulate Composition and Particle Size
Gaseous Composition
4.1.2 Factors Influencing Effluent Generation
4.2 Carbon Plant Effluents
4.3 Bake Plant Effluents
4.4 Cast House Effluents
4.5 Other Effluents
-------
ows
4.0 Sources and Characteristics of Effluent Releases
The airborne effluents from primary aluminum
reduction plant operations include dusts of carbon
and alumina from materials handling and preparation,
and particulates and gases evolved from potlines, an-
ode bake furnaces and cast houses. The greater quan-
tity (and potentially most damaging) are evolved in
the actual electrolytic reduction process at the pot-
line. An understanding of the operations causing the
release of the various kinds of pollutants, and the
identification of sources, provides a basis for con-
sideration of abatement by control over process vari-
ables and for the selection of effective pollutant
removal equipment. The following paragraphs discuss
factors which influence the production of effluents.
4.1 Potline Effluents
The quantities and composition of potline efflu-
ents vary within wide limits among modern aluminum
smelters, being strongly influenced by operation con-
ditions such as temperature, bath ratio, frequency of
anode effects, and method of crust breaking. Moreover,
the effluent may vary with time for any given plant,
due to gradual changes which may occur in potline op-
erations.
Normal cell operation is interrupted by occa-
sional anode effects, cell working to introduce alu-
mina feed, periodic tapping of molten aluminum and
in the case of prebake cells, the periodic changing
of anodes.
According to one investigator, the normal flu-
oride evolution from a crusted-over cell is approxi-
mately 15 pounds F per thousand pounds of aluminum
produced but during an anode effect the fluoride evo-
lution increases to approximately 378 pounds per
thousand pounds of aluminum, 4.I/. Normally individ-
ual cells in this country may experience from less
than one-half to two anode effects per cell day. The
duration of an anode effect is dependent upon how
quickly the cell operator corrects it and may range
from five to fifteen minutes.
4-1
-------
Breaking the crust of the cell for a cell work-
ing causes the fluorine evolution to rise to approxi-
mately 53 pounds per thousand pounds of aluminum. 4.I/
The duration of a cell working varies according to the
size and type of the cell and whether or not the cell
is equipped with automatic crust breakers. With the
automatic crust breaker on a prebake cell, working is
accomplished very quickly, taking only one or two min-
utes. For a normal size prebake cell of approximately
90,000 amperes, a manual working may be accomplished
in five to ten minutes depending upon the hardness of
the crust. Soderberg cells and side-working prebake
cells are normally worked by means of a pneumatic crust
breaker similar to a paving breaker. A working may be
accomplished in approximately five minutes on a 90,000
ampere side-worked cell.
Tapping and changing anodes cause the least
increase in fluorine evolution depending upon how much
of the molten electrolyte is exposed.
Sulfur oxide effluents originate from the coke
and pitch from which the sacrificial anodes are made.
4.1.1 Composition of Potline Effluents
A typical prebake cell effluent, derived from
published information in the literature, and the indus-
try Questionnaire, is shown in Table 4.1. These data
indicate that:
a) Roughly 25-63 pounds of particulates are
released per 1000 pounds of aluminum
produced, of which 10-25 percent of the
weight is fluorine content.
b) Gaseous fluorine content is approximately
half again the weight of fluorine contained
in the particulates.
4-2
-------
. 0^80
Table 4.1
Reduction Cell Effluents
Component
C02
CO
S02
"F" (Gas)
"F" (Solid)
Total "F"
Total Solids
Quantity
European i/
1500
250
6.5
10.3
6.3
16.6
25 to 63
Ib/M Ib
' U.S.
-
-
30
13.1
8.8
22.5
45.6
Al
I/
V
3/
I/
3/
3/
Ref. 4.2/, 4.3/, 4.4/, 4.5/, 4.6/t 4.7/, 4.8/, 4.9X
Estimate based on 3 percent sulfur in anode coke.
_3/ Industry Questionnaire, Weighted Average Response.
4-3
-------
J. L. Henry, contributing to a 1962 international
symposium on the extractive metallurgy of aluminum 4.2/,
reviewed the nature of reduction plant fumes. The follow-
ing summary of effluent compositions is based largely on
this review.
Particulate Composition and Particle Size
The particulate phase of the cell effluents con-
tains material derived from dusting of the alumina and
other raw materials during the feeding operation of the
cells, solid matter which originates from the volatiliza-
tion of the fused salt bath, and material mechanically
entrained by the air sweep over the cell surface into the
collection system. The greatest portion of the airborne
particulates consists of alumina from dusting; some of
the alumina is formed by thermal hydrolysis of the vol-
atized bath materials. Carbon particles result from the
mechanical and electrochemical dusting of the anodes.
Other components have been identified in the par-
ticulate matter including cryolite (Na3AlFg), aluminum
fluoride (A1F3), calcium fluoride (CaF2), chiolite
(Na5Al3Fl4) and iron oxide (Fe2C>3) .
Particle size distribution is a principal deter-
mining factor in the particulate removal efficiency for
most types of air pollution control equipment, and know-
ledge of the size distribution for a given pollutant may
aid in estimating the removal efficiencies of alternative
selections among types of removal equipment. If the
fractional removal efficiency characteristic of a piece
of control equipment is known and if the particle size
distribution of a pollutant can be determined by the same
or comparable measuring equipment used for determining
the fractional efficiency, then these data may be com-
bined to calculate an overall removal efficiency for the
equipment operating on the pollutant dust in question.
Published or reported cell effluent particle size
distribution data are sparse and techniques for measure-
ment are subject to variations, even among different
investigators using similar equipment, so caution should
be exercised in drawing conclusions from these data or in
comparing data from one source with those from another.
4-4
-------
0482
However, a limited amount of available information pro-
vides a basis for estimating the performance of some
types of control equipment applied to potline effluents.
Reported determinations of particle size distri-
butions of the dust and fume collected in primary efflu-
ents are plotted in Figure 4.1. Two plots are shown for
prebake potlines, one reported as the average of four
samples of pot emissions, the other as the average of
five samples of electrostatic precipitator intake. A
single plot of average samples is shown for HSS Soderberg.
No comparable data have been obtained for VSS Soderberg
effluents.
These plots are illustrative of the comparative
size characteristics of the primary dusts from two types
of cells. The slopes of these data give an indication
of the range of particle sizes in the samples and the
placement of the curves on the plot indicates that a sub-
stantial fraction of the prebake and HSS particulate
weight is submicron, or in the range where particulate
removal efficiencies of most equipment are low.
f
Additional particle size data are reported in
Appendix 4A.
Gaseous Composition
The gaseous phase of the cell effluent, before com-
ing into contact with air, consists principally of carbon
dioxide and carbon monoxide formed by the oxidation of
carbon anodes by oxygen released from Al2C>3 on electrol-
ysis. Although the mechanism of formation is subject to
argument, the variation in volume ratio of the two gases
during the normal cycle of cell operation is generally
known. The ratio of CO2 to CO decreases when the cell
temperature is abnormally high and also during the pe-
riods when the anodes are polarized (anode effect). The
carbon dioxide content of the unburned gases varies be-
tween 60 and 85 percent; the balance is largely carbon
monoxide. Contact of the hot gases with air results in
a substantial decrease in the carbon monoxide through
combustion.
Other gases may be found in small amounts during
cell operation. These include sulfur dioxide (S02)» hy-
drogen sulfide (H2S), carbonyl sulfide (COS), carbon
4-5
-------
0^83
FIGURE 4.1
PARTICLE SIZE WEIGHT DISTRIBUTION
POTLINE PRIMARY EFFLUENT
KM
90
WEIGHT PERCENT LARGER PARTICLES
*• M 70 M SO 40 M M 10
0.1
JO *0 40 M M 70 SO 90 *S
WEIGHT PERCENT SMALLER PARTICLES
4_g INDUSTRY MPOKTID DATA
0.2
0.1
-------
disulfide (082)/ silicon tetrafluoride (SiF4); hydrogen
fluoride (HF), and water vapor.
During the period of an anode effect, fluorocar-
bons are known to be produced. These consist almost en-
tirely of carbon tetrafluoride (CF4) together with very
small amounts of hexafluoroethane (C2F5).
Fluorine
Henry reports that the ratio of gaseous to partic-
ulate fluoride in reduction cell fumes varies over a
range of about 0.5 to 1.3. These values are given for
fumes which have burned in contact with air. Weighted
average data from the Industry Questionnaire indicate
that this ratio is about 1.2 to 1.7 for prebake and HSS
Soderberg cells and about 3.0 for VSS Soderberg cells
with hydrocarbon combustors. Unburned fumes usually show
a lower ratio of about 0.3, according to Henry. Burning
of the hot gas-particulate mixture when it contacts air
results in thermal hydrolysis of some of the particulate
fluoride with the formation of additional hydrogen flu-
oride.
Thermal hydrolysis of volatized bath materials
appears to be responsible for a substantial part of the
hydrogen fluoride found in reduction cell fumes. This
reaction of solid or vaporized fluorides with water vapor
at elevated temperatures takes place primarily at the
point where the hot gases escape through vents in the
crust.
A source of hydrogen is necessary for the genera-
tion of hydrogen fluoride. Water vapor in the air is a
contributor of part of this hydrogen. Other sources in-
clude residual moisture in alumina and bath raw materials
and hydrocarbons in the carbon anodes.
*
Some gaseous hydrogen fluoride is removed from
the effluent stream by interaction with the contained par
ticulate matter. Chemical reaction is responsible for
some of this pickup, while some is due to chemisorption,
absorption, and adsorption.
While the determination of total fluoride content
of fumes may be quite reliable, estimates of the distri-
bution of fluoride between gaseous and particulates forms
4-7
-------
0^85
is subject to uncertainty due to such factors as the de-
gree of thermal hydrolysis during burning of the gas and
method of separation of gas and particulates during sam-
pling.
Sulfur Oxides
The sacrificial carbon anodes of aluminum reduc-
tion cells are made from calcined petroleum coke which
characteristically contains sulfur. Whereas coke with a
maximum sulfur content of 2.5 percent was once readily
available, it has become scarce and smelters now buy coke
with as much as 5 percent sulfur. The average petroleum
coke now in use may contain 3 percent.
Since this coke has already been calcined at tem-
peratures above the anode baking temperature, the sulfur
is fixed and will be released only as the anode is con-
sumed in the electrolysis process. Depending on sulfur
content of anode carbon, the S02 effluent may range from
15 to 50 pounds per 1000 pounds of aluminum produced.
Soderberg Effluents
The effluents from Soderberg cells have, in addi-
tion to the constituents mentioned above, the character-
istic of containing the hydrocarbons evolved by the
baking of the anode paste in the cells. The presence of
these hydrocarbons, which are in gaseous form at the cell
operating temperatures, complicates in some degree the
emission control problems associated with Soderberg cells.
If they cannot be converted by combustion to stable gas-
eous compounds the hydrocarbons will condense to tarry
compounds which are difficult to handle and remove from
the gas stream. As will be noted in a later section, the
collection and combustion of these materials is practical
with the VSS Soderberg cell, but not with the HSS Soder-
berg cell. In the latter case the problem of tar separa-
tion becomes somewhat similar to that encountered in pre-
baked anode furnace plants, where similar tars are present
in the effluents.
4-8
-------
0^86
4.1.2 Factors Influencing Potline
Fluorine Effluent Generation
J. L. Henry reported 4.2/ on experimental work
which established correlations between three cell opera-
ting parameters and effluent production for a 10,000
ampere laboratory experimental prebake type aluminum
reduction cell. It was shown that increasing bath ratio
(NaF/AlF^), increasing alumina content of the bath, and
decreasing temperature all tend to result in a decrease
in the fluoride content of cell effluent. Table 4.2 sum-
marizes the findings of these tests.
Table 4.2
Experimental Effect of Three Operating
Variables on Fluoride Effluent
Range of Variables
Alumina Fluoride
Bath Ratio Content Temperature Effluent
(1.44 to 1.54) 4% 975°C 31% Decrease^/
1.50 (3% to 5%) 975°C 20% Decrease^/
1.50 4% (982 to 972°C) 24% Decrease^/
_!/ Within range of variable denoted by ( ) .
Henry calls attention in his paper to the fact
that "determination of the effect of operating variables
on the fluoride emission from electrolytic reduction
cells is difficult to accomplish with a high degree of
certainty. This is true even with small-scale experi-
mental cells operated by research personnel. It appears
from the work reported here, however, that cell tempera-
ture, bath ratio, and alumina concentration are the most
important variables affecting total fluoride emission".
The absolute relationships reported by Henry may
not hold for full-scale cell operation.
4-9
-------
4.2 Carbon Plant Effluents
Aluminum reduction cell anodes and cathodes are
made from anthracite and/or petroleum coke, bonded by
pitch and baked to form solid carbon masses. The prep-
aration of the carbon materials, consisting of crushing,
grinding, classifying, blending of carefully sized frac-
tions, and mixing with pitch binder, is carried out in
the green mill. In all but a few cases, this carbon
plant operation is carried out on the reduction plant
site, the exceptions being plants utilizing prebaked
anodes and cathode blocks which are shipped to the plant
already fabricated.
The effluents from these operations consist of coke
and coal dusts and fines generated by comminution, screen-
ing, and materials handling. Control is practiced to
maintain plant housekeeping and industrial hygiene stand-
ards. Effluents are generally coarse particulates, eas-
ily controlled by collection to bag houses, and do not
constitute a significant air pollution problem beyond the
boundaries of an aluminum reduction facility.
Volatile hydrocarbon fumes are generated to a limit-
ed extent by the paste mixing operation in which the hot
pitch binder is added to the dry materials. This efflu-
ent is usually vented directly to atmosphere. In some
operations, however, these fumes are partially removed
from their gas streams by using a wet scrubber.
4.3 Bake Plant - Effluents
For plants using prebaked anodes the carbon paste
is pressed to green forms and baked for extended periods,
during which time effluents are generated and released.
Bake plant effluents may include products of firing
combustion, burned and unburned hydrocarbons derived from
the heating and carbonizing of the paste binder pitch,
SC>2 and 503 derived from the carbon paste materials, and
fluorine. The source of the latter is recycled anode
butt scraps which carry absorbed or adherent bath mate-
rials back into the anode cycle.
4-10
-------
Little information has been obtained or published
concerning the quantitative amounts of bake plant efflu-
ents. A very limited amount of testing has been carried
out on baking emissions. The order of magnitude of the
problem is indicated in Table 4.3, supplied as an average
by a multi-plant aluminum producer. It is reported that
total F effluents can be maintained at less than 0.4
pound per ton of aluminum produced by exercising partic-
ular attention to cleaning the spent anode butts of
adherent bath before they are crushed for recycle.
Table 4.3
Anode Baking Ring Furnace Emissions -=/
Flow Rate, cfm 75,000 - 184,000
Stream Loading, gr/cf
Total Solids 0.021 - 0.10
HF 0.003 - 0.03
Pitch Condensate 0.01 - 0.30
Quantities, lb/1000 Ib Al
Total Solids 1.0 - 5.0
Hydrocarbon 0.25 - 0.75
Total F 0.15 - 0.75
Sulfur . 0.35-1.0
I/ While the direct fired ring furnace has been the
normally used type for prebaked anodes, attention
has been given to the development of continuous
tunnel kilns for this purpose. Combustion condi-
tions are significantly different and zonal tem-
perature control closer, with one result being
that the above emission levels may be reduced by
factors of 0.01 in total solids and 0.02 in hydro-
carbons, fluorine, and sulfur.
4-11
-------
4.4 Cast House Effluents
Cast house operations include the receipt of
molten metal in tapping crucibles from the potlines,
transfer to holding furnaces or casting in large sows
for later remelt, fluxing for the removal of impuri-
ties, alloying, and casting into ingots or billets.
Effluent releases from these operations consist large-
ly of fluxing fumes, periodically evolved.
Furnaces are usually gas-fired, and in themselves
present no potential pollution problem.
Blended metal from the holding furnaces can be
gas-fluxed or salt-fluxed for the removal of certain
impurities such as oxides, bath electrolyte, or gas
inclusions and the skimmings, or dross, subsequently
treated for metal recovery.
Gas fluxing involves the bubbling of chlorine,
nitrogen, argon, helium, or mixtures of chlorine with
any of the inert gases, through the molten metal in the
holding furnace. Salt fluxing, in which the salt is
added to the bath surface, may utilize hexachlorethane,
aluminum chloride, and magnesium chloride. In all
cases where chlorine or chlorine compounds are used for
fluxing, copious quantities of fumes are evolved, which
must be collected and removed from the working area and
may or may not be treated. The fume is primarily alu-
minum chloride which, in the presence of atmospheric
moisture, may hydrolyze to HCl and A12O3- If a melt
is overfluxed, free chlorine may evolve. Traces of
fluorine may be present in the fumes, originating with
electrolyte impurities fluxed from the metal.
4.5 Other Effluents
The handling of dry bulk materials, alumina,
cryolite, and fluorspar is accompanied by dusting at
transfer points. These particulate effluents are nor-
mally an industrial hygiene problem but may create an
air pollution problem. Normally they are collected at
the points of evolution and returned to the handling
systems.
4-12
-------
n' - Q n
U 4 0 U
References - Section 4
Miller, S.V., et al. "Emission of Fluorine Com-
pounds From Electrolysis Cells in the Production
of Aluminum" Sverdlovsk Nolich, Issled Inst.
Henry, J.L., "A Study of Factors Affecting Flu-
oride Emission from 10 KA Experimental Aluminum
Reduction Cells", Extractive Metallurgy of Alu-
minum, Volume 2, pp 67-81, Interscience Publish-
ers, New York (1963).
Oehler, R.E., "Emission of Air Contaminants in
Aluminum Electrolysis", TMS of AIME Paper No.
A70-11, Presented at the TMS-AIME Meeting, Feb-
ruary 16-19, 1970 at Denver, Colorado.
"Restricting Dust and Gas Emission in Bauxite and
Aluminum Processing Plants", Verein Deutscher
Inginieure Paper No. 2286 (November 1963).
Barrand, M.P., et al. "L1Aluminium", Volume II,
Published by Editions Eyrolles, Paris (1964).
Calvez, C., and Pailhiez, A., "Compared Technol-
ogies for the Collection of Gases and Fumes and
the Ventilation of Aluminum Potlines", Presented
at the Extractive Metallurgy Division of AIME
Symposium, Chicago, Illinois, December 11-13,1967.
Callaioli, G., Lecis, U., and Morea, R., "Systems
of Gas Collection and Cleaning in Electrolytic
Furnaces of Montecatini Edison Aluminum Plants",
TMS OF AIME Paper No. A70-23, Annual Meeting at
Denver, Colorado, February 16-19, 1970.
Pailhiez, A., "Collection and Washing of Gases
from Aluminum Reduction Cells", TMS of AIME Paper
No. A70-57, Annual Meeting at Denver, Colorado,
February 16-19, 1970.
4-13
-------
Moser, E., "The Treatment of Fumes from Primary
Aluminum Reduction Plants", Paper No. 12 Pre-
sented at International Conference on Air Pollu-
tion and Water Conservation in the Copper and
Aluminum Industries, Basle, Switzerland,
October 21-23, 1969.
Hanna, T.R., and Pilat, M.J., "Size Distribution
of Particulates Emitted from a Horizontal Spike
Soderberg Aluminum Reduction Cell", Presented at
the 1970 Annual Meeting of the Pacific Northwest
International Section of the Air Pollution Con-
trol Association, November 1970.
4-14
-------
Table of Contents
Section 5
5.0 Technology of Emission Control
5.1 Effect of Potroom Operating Conditions
5.1.1 Suppression of Anode Effects
5.1.2 Reduced Bath Temperature
5.1.3 Hood Maintenance and Operation
5.1.4 Mechanization and Computer Control
5.1.5 Effect of Ancillary Operations
5.2 Effluent Collection Systems
5.2.1 Roof Monitor Collection Systems
5.2.2 VSS Soderberg Primary Collection System
5.2.3 Prebake and HSS Soderberg Cell Hooded
Collection Systems
5.3 Emission Control Techniques
5.3.1 Pollutant Removal Mechanisms
Particulates
Gaseous Fluoride
5.3.2 Removal Equipment
a) Burners
b) Cyclone collector
c) Baghouse Filter
d) Dry Scrubbing Systems
e) Electrostatic Precipitator
f) Spray Tower
g) Spray Screen
h) Venturi Scrubber
i) Chamber Scrubber
j) Cross Flow Packed Bed Scrubber
k) Floating Bed Scrubber
1) High Pressure Spray Screen
m) Other Wet Scrubbers
n) Incinerators
-------
5.0 Technology of Emission Control
The effective control of effluents from an alumi-
num smelter potline involves attention to:
a) Operating conditions in the cells,
b) Collection of effluents from the cells,
c) Removal of pollutants from the collected
effluent streams.
Anything which can be done to reduce the quantity of nox-
ious effluent from the process will improve the ultimate
emission picture from a pollution control system. Also,
since overall control efficiency is the product of collec-
tion and removal efficiencies, a highly effective scrubber
system, for example, loses its value if a substantial
fraction of the effluent escapes the collection system.
All three factors: operating conditions, collection effi-
ciency, and removal efficiency should be treated carefully
to achieve optimum control over plant emissions.
5.1 Effect of Potroom Operating Conditions
Section 4.1 describes several conditions in reduc-
tion cell operations which give rise to higher than aver-
age effluent production and careful control over cell
operation can minimize potline effluents. Specifically,
the following measures may be used to achieve these ends.
5.1.1 Suppression of Anode Effects
During an anode effect the cell voltage differen-
tial rises from its normal 4.5-4.8 volts to 40-60 volts
and the line current is reduced by from three to five
thousand amperes. The net effect is that the power in-
put to the cell increases more than tenfold. The entire
power increase is converted into heat, which in turn
raises the temperature of the cell electrolyte. At the
higher cell temperature, the fluorine evolution is in-
creased. For the anode effect effluent rate compared
with quiet cell operation, Less and Waddington 5.I/
found a 27-fold increase in solid F and a 2.7-fold in-
crease in HF. Depending on the promptness with which
the cell operator reacts, this anode effect may last from
three to fifteen minutes. Occasionally cell operators
5-1
-------
will deliberately allow anode effects to continue in or-
der to soften an unusually hard crust. Automatic crust-
breakers help to minimize the need for this practice. In
normal cell operation, with manual crust breaking, the
frequency of anode effects is from less than one-half to
as many as six anode effects per cell day.
Placing cells on an anode effect suppression sys-
tem, that is, scheduled workings of the cell in order that
the alumina content of the electrolyte is replenished be-
fore it falls below the concentration causing the anode
effect, can reduce the frequency of anode effects to the
range of one-half to one anode effect per cell day. The
newer computer controlled potlines may operate almost free
from anode effects.
5.1.2 Reduced Bath Temperature
The higher the bath temperature, the more will the
bath salts vaporize and be carried into the cell efflu-
ents. Normal operating temperatures for cells with a bath
ratio of approximately 1.40 are between 970 and 980 de-
grees centigrade. This relatively low operating tempera-
ture is near the freezing point of the electrolyte and
close attention by cell operators is necessary to prevent
the cell from becoming too cold, or "mucking up". A cold
cell is corrected by increasing the cell voltage and allow-
ing the electrolyte to increase in temperature. Abnormal
or "sick" cells operate at temperatures in excess of
1000°C and sometimes they do not crust over. Under these
conditions, the high temperature molten electrolyte is
exposed and there is a large increase in volatilization
of bath salts with a corresponding increase of fluorine
in the cell effluents. Operation of cells at the lowest
possible temperature to minimize fluorine effluents re-
quires trained, conscientious cell operators, or computer
control.
While the temperature of the cell may be lowered
by the additions of lithium salts to the electrolyte, low-
ering its freezing point, the net benefit of these addi-
tions is the subject of controversy. One overseas inves-
tigator 5.2/ reports, among other advantages, a substan-
tial decrease of fluorine losses in waste gases which
resulted in a reduction of fluorine emissions. In this
country experiments undertaken by a major producer, were
5-2
-------
01*95
reported to have demonstrated quite the contrary of the
referenced conclusions.
The electrolyte system is complex, and electro-
lyte conditions which reduce fluorine emissions from the
molten bath but which simultaneously destroy the ability
of the bath to crust over and carry a cover of alumina
may result in a net increase in cell emission; the alu-
mina cover intercepts a substantial quantity of fluoride
and returns it directly to the molten bath.
5.1.3 Hood Maintenance and Operation
Most aluminum smelters which have pollution con-
trol equipment designed to achieve high overall control
efficiencies depend on a primary hooding system to col-
lect most of the cell effluent and conduct it to removal
equipment. Any effluents which escape the primary col-
lection systems are exhausted through roof monitors with
or without treatment. Overall control efficiency is
limited by the collection or hooding efficiency with which
cell effluents are drawn into the primary control system.
Present cell designs, regardless of how well enclosed or
shielded they may be, do not achieve 100 percent hooding
efficiency because the shields need to be opened for cell
working, at least for anode replacement in the case of
prebake lines, and for metal tapping in all lines. How-
ever, careful attention to the design and construction of
hoods, and strict insistence that potroom operators keep
shields in good repair and that they open or remove them
no more than necessary, will contribute significantly to
improved pollution control.
Some potlines are provided with means for increas-
ing the air flow into the primary collection system at
individual cells when hoods need to be opened. This con-
tributes strongly to high collection efficiency and per-
mits a realization of economy of low air flow when the
cell is properly enclosed.
In recognition of the particular importance of
hooding or collection efficiency, one operator of prebake
potlines makes a special effort to aim for 100 percent,
and estimates a realization of 95 to 97 percent based on
total F effluent. The design of this system incorporates
provision to double the flow rate on any cell which is
5-3
-------
opened for working, tapping, or anode changing. Older
design cells and those without dual air flow provisions
usually do not achieve these high collection efficien-
cies; in fact they may average ten or more percentage
points lower.
5.1.4 Mechanization and Computer Control
Mechanization of crust breaking and cell feeding
allows the cell operators time to maintain close watch
over the operating cells and to control them within nar-
row temperature ranges. The overall effect is lower
average operating cell temperature, fewer and briefer
anode effects, and a reduction in the fluorine content
of cell effluent gases compared with normal manual cell
operation.
Full mechanization of reduction cells makes it
possible to apply computer control which incorporates
the frequent scanning of operating variables on each
cell and the triggering of automatic corrective action
for any variation that is outside set operating limits.
Such control makes it possible for all cells in a pot-
line to be operated at the lowest practical temperature
and with nearly complete freedom from upsets caused by
anode effects. Cell feeding and crust breaking opera-
tions can be cycled in response to the needs of individ-
ual cells, and the number of abnormal or "sick" cells
usually associated with manual potline operation can be
reduced. Variations in the cell operation occasioned
by having different shift personnel tending the cells
over the 24-hour period are largely avoided.
Many plants are developing various degrees of com-
puter control in combination with mechanization. Although
full automation has not yet been satisfactorily accom-
plished, several potlines are approaching this goal on
an experimental basis.
5.1.5 Effect of Ancillary Operations
The bake plant at prebake anode aluminum smelters
releases effluents which are partially subject to control
through provisions in operating practice. The bake
plant effluents present primarily a smoke abatement prob-
lem resulting from tars and volatile hydrocarbons re-
5-4
-------
leased from the pitch binder in the baking operations.
Other effluents are SC>2 and HF. Most of the effluent SC>2
derives from the sulfur content of the pitch and coke
used in the manufacture of coherent anode blocks. By us-
ing low sulfur pitch and coke, effluent S02 may be held
to a minimum. It should be noted that low sulfur pitch
and coke are becoming increasingly difficult to obtain.
HF gas in bake plant effluents comes from bath ma-
terial adhering to anode butts which are recycled in the
manufacture of new anodes. Special care in cleaning the
butts before crushing will reduce F effluent to about one
quarter the quantity experienced when adherent cryolite
is simply knocked off the butts at the potroom prior to
sending them to the anode plant.
5.2 Effluent Collection Systems
Effective control of aluminum potline effluents
requires first that they be collected for treatment in
removal equipment. Three types of collection systems are
in use, each having certain advantages over the others
for particular applications.
5.2.1 Roof Monitor Collection Systems
Many European reduction plant potlines and a few
prebake potlines in the United States are designed to
pass all airborne effluents through pollutant removal
equipment located in roof monitors or ducted from collec-
tion points at the tops of the potline buildings. All
potline effluent and all normal room ventilation air is
intended to pass through the collection and removal equip-
ment.
Although collection efficiency might be assumed
to be 100 percent in this scheme, deficiency in the de-
sign of the provisions for air intake to the buildings
may bring about a reduction in the collection efficiency.
Some potline buildings have openings in the side walls
at working floor level through which ventilation air
enters as shown in Figure 5.1. This air is supposed to
sweep past the cells and up through the roof monitor col-
lection system, but adverse winds may blow through the
buildings in such a way as to carry potline effluents out
through some building wall openings, thus short circuit-
ing the collection system and reducing its efficiency.
5-5
-------
0^38
FIGURE 5.1
ROOM COLLECTION SYSTEM SIDEWALL ENTRY
INDUCED DRAFT FAN
ROdF MONITOR SPRAYS
FIGURE 5.2
ROOM COLLECTION SYSTEM BASEMENT ENTRY
FLOOR GRATING
5-6
-------
Figure 5.2 shows a building arrangement which helps to
avoid this short circuiting of the collection system.
Fresh air is drawn into the building below the working
floor level and is allowed to pass up through gratings
past the cells to the monitor collection system. Al-
though sub-floor intakes may make the collection virtu-
ally 100 percent efficient, all roof monitor collection
systems suffer the disadvantages that very large quanti-
ties of ventilation air must be handled, 30,000 to 60,000
cfm per cell.
x
5.2.2 VSS Soderberg Primary Collection System
Soderberg cells utilize an anode consisting of a
rectangular container, open at the top and bottom, sus-
pended above the cell, into which carbon paste is fed at
intervals, becoming baked by the heat of the cell as it
gradually descends in the container. This baking of the
carbon paste releases substantial quantities of hydro-
carbon gases and fumes which may interfere with proper
operation of pollution control equipment.
The VSS or vertical spike suspension Soderberg
design utilizes steel pins driven into the anode verti-
cally from the top to conduct the electric current from
the anode bus into the anode. These pins are removed
from the anode by withdrawing them from the top before
they are carried by the anode into the bath zone. Since
no side channels are needed as in the HSS Soderberg, it
is possible to install a permanent gas collecting skirt
around the bottom of the anode which is sealed to the
electrolyte crust by a blanket of alumina. (See Figure
3.5). Most gases and particulate effluents pass into
the skirt enclosure and to a gas collection system.
This system has the advantage of very low air
dilution and sufficiently high hydrocarbon concentration
that the effluent gases may be burned; most of the hydro-
carbons, both gaseous and particulate, and much of the
CO, is oxidized to C02 and water vapor. The effluent
gas leaving the burner is low in hydrocarbons and may
then be treated in a manner similar to prebake cell efflu-
ents .
One typical VSS Soderberg collector passes about
20 scfm of cell gas and 200 scfm of secondary combustion
air into each of two burners on the collecting skirt.
5-7
-------
ObUU
The discharge from the burners, about 400 scfm per cell,
passes into ducts for transport to pollutant removal
equipment.*
Skirt and burner type systems on VSS Soderberg
cells collect most of the cell effluents during approxi-
mately 95 percent of the time when the cell is operating
without disturbance of the crust and the alumina seal.
When the crust is broken, the alumina seal falls into
the molten electrolyte and some gaseous and particulate
effluents escape the collection system and pass^into the
room atmosphere for subsequent discharge through the
roof monitor. Any hydrocarbon effluents escaping from
the top of the anode pass directly into the room atmos-
phere and also are discharged through the roof monitor.
Average fluoride collection efficiencies for VSS
Soderberg potline systems have been reported to range
from 70 to 95 percent.*
5.2.3 Prebake and HSS Soderberg Cell
Hooded Collection Systems
The physical arrangements of prebake and HSS
Soderberg reduction cells lend themselves to hooding or
enclosures connected with duct systems to collect most
of the cell effluents and deliver them to primary pollu-
tant removal equipment. (See Figures 3.3 and 3.4).
These systems characteristically draw 2000 to 8000 cfm of
air into the hoods at each cell. The concentrations of
cell effluents in the entraining air are intermediate be-
tween those found in the VSS Soderberg skirt and burner
system and the European designs which permit all efflu-
ents to mix with the building ventilation air.
The effectiveness of the cell enclosure primary
collection system depends essentially on three factors.
a) Hood Designs - Designs which leave a minimum
of leaks to the potroom and which limit the
exposure during crust breaking and anode chang-
ing1 operations will outperform poorer designs.
Center working pots in which crust breaking
* Industry Questionnaire Response
5-8
-------
0501
and alumina charging operations take place
in the middle between rows of anodes, espe-
cially pots with cell mounted automated
crustbreakers, permit tight hood designs
which need be opened only occasionally to
change anodes or to remove molten aluminum
from the cells.
b) Operator Care - Careful supervision of pot-
room operating personnel to ensure that they
maintain shields in good repair and close
hoods promptly after working cells will re-
duce emission to the cell room and improve
overall collection efficiency.
c) Air Flow Rate - Increasing the air flow rate
into primary collection systems reduces the
tendency for effluents to escape through
openings in cell hoods and improves the col-
lection of dust and gases during periods when
the hoods are open for cell working, alumina
feeding, or anode replacement. A dual volume
exhaust system permits an increased exhaust
rate to provide a greater collection efficiency
when hoods are removed.
Aluminum smelters in the United States have report-
ed primary collection efficiencies for hooded prebake and
HSS Soderberg potlines to range from 71 to 98 percent*,
averaging in the 90's for prebake cells and somewhat low-
er for HSS Soderberg cells.
5.3 Emission Control Techniques
The emission control problems of the primary alu-
minum reduction plant are a result of the effluent con-
ditions described in the previous section, and are con-
cerned with the removal of particulates and of HF gas
from the stream in which they occur.
* Industry Questionnaire Response
5-9
-------
0502
5.3.1 Removal Mechanisms
Particulates
Particulates include solid particles of carbon,
alumina and bath materials, tar fogs (hydrocarbons) and
inorganic fumes. Their removal from gas streams may be
accomplished by inertial segregation from the gas phase,
by collection in a liquid phase, by interception on a
porous medium, or by segregation through electrostatic
agglomeration and attraction.
These mechanisms are affected in greater or less-
er degree by the size particulate to be removed. Par-
ticulates larger than about two microns may be separated
from gas streams by gravitational or centrifugally in-
duced forces.
Collection of particles into a liquid medium, as
by wet scrubbing, is effected by inertial impaction,
interception, and diffusion, depending upon the particle
size and its inertial behavior. The inertial impaction
mechanism assumes that particles have sufficient mass or
inertia to leave the flow streamlines and strike a liquid
surface around which the streamlines are bending. The
interception mechanism assumes that the particles have
size, but no mass, and follow the gas streamlines so that
a particle is collected only when the gas streamline is
closer to an obstruction than half the diameter of the
particle.
Collection of submicron particles which lack suf-
ficient mass to exhibit significant inertia and are too
small to be collected by interception may be collected
by a diffusion mechanism that visualizes the particles
as moving about in the gas in a manner characteristic of
the thermodynamic behavior of gas molecules.
Interception by a porous medium involves the me-
chanical entrapment of the particles in the interstices
of built up layers of the medium. These interstices may
become submicron in size.
Electrostatic attraction among charged particles
and between charged particles and a collecting surface
may effect the agglomeration and capture of fine par-
ticulates. Consideration of their electrostatic proper-
5-10
-------
0503
ties with respect to specific dusts may be a factor in
the selection of filter fibers. Electrostatic attrac-
tion between effluent feed particles and the bed mass is
a factor in fluid bed agglomeration and capture.
In electrostatic precipitation, forces acting on
electrically charged particles in the presence of an im-
pressed electric field are utilized to remove solid or
liquid particles down to and including submicron sizes.
The particulates are attracted to and retained on the
collection electrode and are removed either by intermit-
tent rapping or continuous irrigation.
Tar fogs generated by volatilization from Soder-
berg anodes present added problems in particulate emis-
sion control. The wet scrubbing mechanisms apply to
their capture; however, particle reentrainment may occur
because tars resist wetting and thus reduce the effective
removal. Where these fogs are in relatively concentrated
form, as from VSS Soderberg potlines, they may be burned
by self-supporting combustion. If the effluent gas
stream is dilute, as from HSS Soderberg potlines, pitch
mixers, or anode bake plants, self-supporting combustion
is not possible. In these cases agglomeration of the
tar fog may occur, together with condensation of hydro-
carbons forming sticky deposits which interfere with col-
lection devices and removal mechanisms.
Gaseous Fluoride
The principal gaseous pollutant in potline and
bake plant effluent streams is fluorine as HF, and the
control mechanism for its removal is absorption in a
selected liquid solvent or adsorption on the surface of
a selected solid. The HF gas molecules to be removed
must diffuse from the bulk gas stream to the surface of
the liquid or solid, pass the phase boundary, and in the
case of the liquid solvent, diffuse into the liquid phase.
In liquid absorption there is a tendency for dis-
solved molecules to migrate back through the interface
and escape to the gaseous phase (a measure of this tend-
ency to escape is vapor pressure). This back pressure
effect is reduced by use of an alkaline solution which
reacts with the dissolved HF molecules to form an ionic
solution. Effective vapor pressure and the contact area
control the removal efficiencies.
5-11
-------
In dry scrubbing, the mechanism of adsorption is
controlled by the reactivity of the adsorbent and the
surface area available for gas-solid contact, the in-
timacy of contact and, to a lesser extent, the contact
time.
5.3.2 Removal Equipment
Table 5.1 summarizes the removal equipment con-
sidered for emission control of the different types of
effluent streams in the aluminum industry. The table
includes the devices which are used singly or in combi-
nation by the domestic producers, and by some of the
foreign plants. It also includes equipment which has
been, or is known to be under application development,
as well as some which experience indicates to have poten-
tial application.
Reportable performance and corresponding operat-
ing parameters of control equipment items in current use
and considered applicable are summarized in Tables 5.2a
through 5.2d, presented by type of effluent control duty.
These data provide the basis for the model systems used
in the analyses of Sections 8 and 9.
Details concerning the individual types of equip-
ment and their appropriate applications are discussed
in the following paragraphs.
a) Burners
The burners on the VSS Soderberg cells, pre-
viously noted and illustrated in Figure 3.5, assist in
the control of cell effluents by converting hydrocarbons
to CC>2 and water vapor. When the cell is operating nor-
mally, burners maintain continuous flame. However, irreg-
ularities in operation can result in a flame-out. With-
out the use of igniters or a satisfactory manual ignition
program, from 5 to 10 percent of the burners of a potline
may be out at any time, giving rise to unburned hydrocar-
bons in the potline effluent gas stream entering the
removal equipment.
The important combustion variables include ef-
fluent tar composition and concentration, and air supply
to the burner. The latter is determined by the size,
5-12
-------
Table 5.1
CONTROL EQUIPMENT CONSIDERED FOR THE PRIMARY ALUMINUM INDUSTRY
PREBAKE POTLINES
Prim. Sec. Sec.
no with no
Sec. Prim. Prim.
VSS POTLINES HSS POTLINES
Prim. Sec. Prim.
no with no
Sec. Prim. Sec.
Sec.
with
ANCILLARY PROCESSES
Bake Dry Paste Cast
Prim. Plant Mtls. Mix House
Ul
I
Burner
Incinerator
Multiple Cyclone A
Baghouse Filter
Fluid Bed Dry Scrubber A
Coated Filter Dry Scrubber A
Injected Alumina Dry Scrubber A*
Dry Electrostatic Precipitator A
Wet Electrostatic Precipitator
Spray Tower A
Spray Screen
High Pressure Spray Screen B
Wet Centrifugal Scrubber
Venturi
Chamber Scrubber A
Wet Impingement Scrubber
Cross Flow Packed Bed B
Floating Bed (Bouncing Ball) A
Sieve Plate Tower
Self-Induced Spray (Bubbler) B
Vertical Flow Packed Bed A
A
A
B
A
A
A*
A*
A*
A
A
B
D
A
B
B
A*
A*
A
A
C
A
B
D
D
B
C
A
A
C
A
B
B
B
A
A
A
B
A
B
B
B
B
B
B
Prim.
Sec.
A
A*
Primary collection stream.
Secondary or potroom system.
In current use in the United States.
In current use outside the United States.
Used in one foreign plant. Not considered
economically feasible in the United States.
B Considered feasible but not known to be
in use.
C In development stage.
D Superseded by other equipment.
Considered not feasible, economically
and/or technically.
Ol
O
01
-------
050b
Table 5.2a - EQUIPMENT REMOVAL EFFICIENCIES
PREBAKE POTLINES
In Current Use
Primary Control
Multiple Cyclone
Fluid Bed Dry Scrubber System
Coated Filter Dry Scrubber I/
Injected Alumina Dry Scrubber*
Dry Electrostatic Precipitator
Spray Tower
Floating Bed Scrubber
Chamber Scrubber
Vertical Flow Packed Bed
Secondary (No Primary)
Spray Screen
Cross Flow Packed Bed (3 ft.).5-/
Applicable
Primary Control
Cross Flow Packed Bed (5 ft.)
Self-Induced Spray
Ventur i
High Pressure Spray Screen — '
Secondary (No Primary)
Efficiencies Derived
from Reported Data—'
Particulate
78
98
98
98
89-98
80 3/ 4/
80
85
85
45 $/
84 y
HF
99
76-92
98
89-98
98
88
66
93 I/
99
Estimated Efficiencies
Particulates
87
65
96
93
HF
98
96
99
98
Operating
Conditions
HP/Mcfm£/ Gal/Mcf
0.8-1.6
4.4
1.5
2.2
0.3-0.7
0.4-0.9 1.7-10
1.9 18
1.0 5
0.7 5
0.2 1.3-10
0.5 10
Operating
Conditions
HP/Me fn£/ Gal/Mcf
1.5-1.8 10+
3.3 6-10
9-10 5-10
8-12 28-42
Floating Bed Wet Scrubber
Secondary (With Primary)
Spray Screen
75
25
87-95
80
0.3-1.0
0.2
3-10
10
j./ Footnotes follow Table 5.2d
* Denotes foreign application
5-14
-------
G507
Table 5.2b - EQUIPMENT REMOVAL EFFICIENCIES -
VSS SODERBERG POTLINE EFFLUENTS
In Current Use
Primary
Burners
Multiple cyclones
Fluid Bed Dry Scrubber-'*
o /
Injected Alumina Dry Scrubber—'*
Dry Electrostatic Precipitator*
Wet Electrostatic Precipitator
Spray Tower_/
Self-Induced Spray
Floating Bed*
Sieve Plate Tower*
Venturi Scrubber
Efficiencies Derived Operating
from Reported Data!/* Conditions
Particulate
"
98 —
98 —
90-99 .
75 y
NR §.'
78
96-97
96 y
HF
-
99
98
-
-
99
99 I'
97
99
99
HP/Mcfm2-/
1.6
4.4
2.2
NA
.66-1.36
1.0-1.3
1.4
1.7
4.5-6.2
9-10
Gal/Mcf
-
-
-
-
5-10
30
33
11
7
5-10
Secondary
(With Primary Collection)
Spray Screen
Applicable
Primary
42 y
88
Estimated Efficiencies —
6/
Particulate
High Pressure Spray Screen
(3 Stage)
Chamber Scrubber
Crossflow Packed Bed (5 ft. Bed)
93 y
87
HF
98
94
98
NA NA
Operating
Conditions
HP/Mcfm^ Gal/Mcf
6.1
1.24
1.5-1.8
26
4.5-5.0
10+
_!/ Footnotes follow Table 5.2d
* Denotes foreign application
5-15
-------
.05U8
Table 5.2c - EQUIPMENT REMOVAL EFFICIENCIES -
HSS SODERBERG POTLINE EFFLUENTS
In Current Use
Primary
Spray Tower
Floating Bed*
Efficiencies Derived
from Reported Data
Particulate HF
'
63-80 -/
78 I/I/
91-93
98
Operating
Conditions
21
HP/Mcfm-/ Gal/Mcf
0.3-0.5
1.9-2.8
1.6-5.0
20-40
Applicable
Primary
Estimated Efficiencies—
.
Particulate
Wet Electrostatic Precipitator 98
Crossflow Packed Bed (5 ft. Bed) 81 *
Self-Induced Spray 62
Spray Screen 75
Chamber Scrubber 94
HF
98
96
93
94
Operating
Conditions
HP/Mcfm^ Gal/Mcf
1.4
1.5-1.8
3-6
0.3-0.5
1.2
5-10
10+
6-10
1.6-5.0
4.5-5.0
Secondary
(With Primary Collection)
Spray Screen
25
80
0.2
10
I/ Footnotes follow Table 5.2d
* Denotes foreign application
5-16
-------
" 0509
Table 5.2d - EQUIPMENT REMOVAL EFFICIENCIES -
BAKE FURNACE
In Current Use Reported and Vendor Operating
Estimated Efficiencies Conditions
9 /
Particulate HF HP/Mcfm^ Gal/Mcf
Dry Electrostatic Precipitator 90 ^ - NR NR
Spray Tower NA \'. 96 NA NA
opj. ciy J.UWCJ. JLXTL -p . j\j ixn.
Self-Induced Spray JLO/ 98-' 96 I/ 3.6
6-10
Operating
Applicable Estimated Efficiencie&S' Conditions
Particulate HF HP/Mcfn£/ Gal/Mfc
Incinerators 90
Wet Electrostatic Precipitator 99—' - 3.8 0.3-0.4
\l Refer to notes on following page.
5-17
-------
Obiu
Notes - Tables 5.2a through 5.26.
I/ Calculated from industry supplied data.
2_/ Horsepower, assuming 50 percent fan and pump efficiency, exclud-
ing collection system losses. Precipitator corona power converted
to horsepower.
3/ Efficiencies derived from reported information.
^/ Fluoride particul'ate.
_5/ Efficiencies reported by equipment manufacturer.
^/ The estimated particulate efficiencies are obtained from manufac-
turers, from reports in the literature or from calculations
based on dust size distribution and fractional efficiency. The
fluoride gas efficiencies and operating conditions are based on
equipment manufacturers' information or literature citations.
l_l Estimated efficiencies for equipment not reported but in use.
J!/ Not Reportable (NR) to preserve confidentiality.
9/ Not Available (NA).
10/ Maximum efficiency for hydrocarbon combustion, manufacturers*
estimate.
ll/ JDesign basis for precipitators being tested applied to the ef-
fluent gas stream after 20 to 30 percent of the particulate has
been removed from the cooled effluent gas by a wet device, as
per private communication with equipment manufacturer.
5-18
-------
4f *< !K f I
U -J i J.
shape and location of the air inlet openings and by the
draft through the burner. The neck between the skirt and
burner must be designed wide enough to prevent blockages
which would otherwise occur when the bath splashes. 5.4/
The composition of VSS gas collected and fed
to the burners is reported to include about 3 volume per-
cent of hydrocarbons, primarily methane, ethane, propane,
butane and their concomitant isomers. After combustion,
less than 0.1 volume percent hydrocarbons remain, indi-
cating a combustion efficiency of 96.7 percent (assuming
little change in the molar gas throughput). 5.5/
Burners are not applicable to HSS Soderberg
cells because the cell construction does not permit skirt
collection of relatively undiluted cell effluents, and
the resulting hydrocarbon concentration in the effluent
gases is too low to support combustion.
b) Cyclone Collector
The most widely used type of dust collection
equipment is the dry cyclone, in which the dust-laden gas
enters a cylindrical vessel tangentially, creating inner
and outer vortices. The entering gas initially traces
the outer vortex down carrying the particulate matter
along, then the gas stream reverses direction and traces
the inner vortex, heading up towards a centrally located
outlet. The dust, because of its inertial characteris-
tics, tends to follow the outside wall down without re-
versing direction, exiting through a gravity discharge
hopper at the bottom. Generally, cyclones are effective
for the removal of solids larger than 5 microns diameter
as indicated by a typical fractional efficiency curve
for a multiple small diameter tube cyclone, Figure 5.3.
The trend in modern installations is to employ
multiple tube cyclones for a preliminary particulate re-
moval stage, taking advantage of the economic return to
the potline of alumina and particulate fluoride.
Performance data for centrifugal collectors,
reported in the literature and by industry questionnaire
responses, are summarized in Table 5.3 for prebaked anode
and for VSS Soderberg potline applications. From these
data average removal efficiencies of 80 and 50 percent
5-19
-------
FIGURE 5.3
FRACTIONAL REMOVAL EFFICIENCY
MULTIPLE TUBE CYCLONE
9 9.99
99.9
99.8
99.5
99
98
U
u 95
&
O 90
lu
O 80
z
Ul
60
so
O
40
3 0
20
10
0.01
I I I I I 1 I
III I I I TT
J i
1 I I I I I
I I I I I I I
0.05 0.1 0.5 1 S
PARTICLE SIZE, MICRONS
i o
50 100
Source: Ref. 5.6/
5-20
-------
Table 5.3 - PARTICULATE REMOVAL EFFICIENCIES OF CYCLONE SEPARATORS
Effluent Source
Prebake
VSS Soderberg
i
NJ
Equipment
Designator
MC
MC
MC
MC
MC
C
C
C
Efficiency
85
81.4
80.5
77.9
40-50
,50
40
25
Pressure
Drop
in.Water
2.5-3
5.1
3.6
3.5
5
N.A.
N.A.
N.A.
Source
IQ
IQ
IQ
IQ
Equipment Designator Code
MC - Multiple Tube Cyclone
C - Cyclone
Source Code
IQ - Industry Questionnaire
O
01
-------
have been used in model analyses for prebake and VSS
Soderberg potline applications. The multiple tube cy-
clone is limited to prebake and VSS Soderberg installa-
tions where tar fog in the gas stream treated is either
absent (prebake) or at a high enough temperature that it
does not condense and agglomerate in the equipment, with
resultant plugging.
c) Baghouse Filter
One of the most effective methods for removal
of dry particulate matter from effluent gas streams is
by fabric filtration, in which interstices of the built-
up layers become submicron in size. Collection efficien-
cies of nearly 100 percent can be expected on particulates
with a median size of 0.5 micron over the size distribu-
tion range of the pollutant dust. 5.9/
Fabrics are woven or felted, with finer fibers
being more effective because they provide more surface
area per unit gas volume, resulting in better capture by
inertial impaction and interception. The initially clean
surfaces pass the gas and some of the particles finer
than the space between the fibers, but rapidly accumulate
a layer of the entrapped larger particle with resultant
reduction in the dimension of the gas flow passages, en-
trapping finer and finer particulates. In this manner
particulate removal efficiencies approaching 100 percent
are rapidly achieved, unequalled at comparable power re-
quirements by other means of separation. 5.9/ Typical
fractional efficiency curves are shown in Figure 5.4.
The filter cake is removed intermittently by
reversal of the gas flow and/or shaking the filter cloth.
Bag filters are suitable for dry materials handling op-
erations and all of the carbon plant operations, (except
pitch fume abatement and anode baking), since these efflu-
ent gases are free from condensible constituents which
might blind the filters. Bag filters are not used on
potline effluents except as part of a dry scrubbing sys-
tem.
5-22
-------
FIGURE 5.4
FRACTIONAL REMOVAL EFFICIENCY
BAGHOUSE FILTERS
" -': ~n r~ -1
UoL
z
ui
U
X
o
U
z
o
3 0
20
1 0
0.01
0.05 0.1
0.51 S
PARTICLE SIZE, MICRONS
30 100
Source: Ref. 5.10y
5-23
-------
Ubib
d) Dry Scrubbing Systems
By combining the mechanisms of HF adsorption
on alumina and mechanical separation of particulates,
removal of both particulates and gaseous fluorides can
be accomplished in a single system. A major advantage
is the return of the recovered fluoride materials to the
potlines without need for further processing to make
them available to the reduction process.
The dry scrubbing of HF by alumina is made
possible because the adsorbed HF appears to form a chem-
ical bond with the alumina, as evidenced by the fact
that the HF is not driven off when the reacted alumina
is heated on feeding to cell bath. The adsorbed or
"chemisorbed" HF reacts with alumina to form aluminum
fluoride and water.
Three processes involving the dry scrubbing
principle have been developed and put into practice.
Table 5.4 summarizes removal efficiencies for these sys-
tems as developed from user information.
Variations among the systems lie in the char-
acter of the alumina used as HF adsorbent, the methods of
obtaining gas contact with-the adsorbent, and the designs
of the mechanical separation devices to remove the par-
ticulates, including the absorbent, from the gas stream.
Coated Filter Dry Scrubber Systems, the Alcoa
173 (reference 5.ll/) and the Wheelabrator (reference
5.12/), are similar in that finely ground reactive alu-
mina is injected into the effluent gas stream to form a
layer on fabric filter bags. Effluent particulates are
captured on this layer and HF is adsorbed on the alumina
when passing through it. The amount of injected alumina
used is noted in the Alcoa patent as being between 3 and
20 percent by weight of the total alumina feed to the re-
duction cell, an amount far in excess of that theoreti-
cally required to convert the gaseous fluorides to alumi-
num fluoride. Catch from the baghouse is blended with
additional cell grade alumina and fed to the pots.
Figure 5.5 shows a process flow diagram.
This type of system has been used on prebake
potlines but is being replaced by other dry scrubbing
systems in several installations.
5-24
-------
Table 5.4 - REMOVAL EFFICIENCIES OF DRY SCRUBBING SYSTEMS
Equipment Fluoride Removal Efficiencies, %
System/Application Designator HF Particulate Total F
Coated Filter Dry Scrubber/PB CFDS 90 98 93.4
Fluid Bed Dry Scrubber/PB FBDS 99 98 98.6
Fluid Bed Dry Scrubber/VSS* FBDS 99 98 98.9
Injected Alumina Dry Scrubber/PB* IADS 98 98 98
Injected Alumina Dry Scrubber/VSS* IADS 98 98 98
* Denotes Foreign Application
-------
FIGURE 5.5
COATED FILTER DRY SCRUBBER
PROCESS FLOW DIAGRAM
C3
(^
l_, '
cc
CLEAN AIR TO
ATMOSPHERE
ACTIVATED
ALUMINA
I
KJ
cn
O-O\-/ \—/O<5
METAL GRADE
ALUMINA
-/
REDUCTION
CELL
DUST
COLLECTOR
ALUMINA
CONTAINING
FLUORIDI
STORAGE
-------
The Fluid Bed Dry Scrubber (Alcoa System 398)
employs a fluidized bed of sandy alumina to contact and
adsorb HF and to trap particulates. Ore grade alumina
is continuously fed to the reactor bed in amounts up to
100 percent of the potline feed requirements and the re-
acted bed material overflows and is used as cell feed.
Virtually all of the effluent particulate is trapped in
the fluid bed, (perhaps by electrostatic agglomeration).
Fugitive particulate, primarily alumina, is stopped by a
bag filter mounted over the reactor. The bags are clean-
ed intermittently, dropping the catch back into the flu-
id reactor bed. Figure 5.6 illustrates this process.
5.137 5.147 5.15/ 5.16/
Alcoa reports that, with proper operating and
maintenance procedures, this system is capable of 98 per-
cent particulate and 99 percent HF removal efficiencies
on prebake potline effluents.
The fluid bed dry scrubber has been applied
in foreign plants to VSS Soderberg effluents with pilot
lights or other devices being used to ensure all burn-
ers being lit. The system has not been applied to HSS
Soderberg effluents and has not been used in domestic
VSS service.
The Injected Alumina Dry Scrubber was devel-
oped by Aluminum Company of Canada for the removal of
particulates and gaseous fluoride from the effluent
streams of prebake and VSS Soderberg anode potlines.
The process depends on the chemisorption of HF on ore
grade alumina followed by fabric filtration to capture
all particulates including the adsorbed HF.
The claimed removal efficiencies compare with
those of the fluid bed dry scrubber but the method of
achieving contact between the effluent gas stream and
the alumina adsorbent differs.
For prebake potline applications, ore grade
alumina is injected into the effluent stream. After
sufficient time to achieve gas-solid contact, this dust
cloud is intercepted by bag filters operating at an air-
to-cloth ratio of about 6 cubic feet of air per minute
per square foot of filter area and a pressure drop of
approximately 6 inches water gage. Collected alumina,
5-27
-------
FIGURE 5.6
FLUID BED DRY SCRUBBER
PROCESS FLOW DIAGRAM
CD
01
ro
c
en
I
to
oo
MITAl GRADE
ALUMINA
\2
ALUMINA
STORAGE
CLEAN AIR TO
ATMOSPHERE
ft
MJ1M1
FLUID IID REACTOR
ft BAGHOUSE
ALUMINA
CONTAINING
FLORIDE
REDUCTION
CELL
-------
•0521
its adsorbed HF, and potline solid effluents are fed to
the cells without further treatment. .
Effluent streams from VSS Soderberg potlines
have higher concentrations of HF than prebake effluents
and they may contain unburned tar fumes. Here again,
ore grade alumina is injected into the effluent stream,
but from this point on, the Alcan process is modified
slightly. An arrangement is made to separate the bulk
of the alumina containing adsorbed HF from the portion
containing unburned hydrocarbons. The latter minor
quantity of alumina is calcined to remove the tar prior
to being returned to the cells along with the main por-
tion of the collected alumina. This system does not re-
quire that all burners be lit all the time.
Removal efficiencies of 98 to 99 percent on
dust and HF have been reported for a full prebake pot-
line application. The vendor expects to achieve 97 to
98 percent on dust and HF on a full line VSS Soderberg
installation.
e) Electrostatic Precipitator
The electrostatic precipitator is contained
in a relatively large chamber through which effluent gas
streams pass at low velocity, usually 3 to 5 feet per
second. In its usual form, high negative voltage corona
discharge wires are suspended across the air stream and
grounded collector plates form parallel passageways for
the air. The ionizing field surrounding the discharge
wires ionizes part of the gas stream and imparts elec-
tric charge to most particles, some positive but most
negative. Positively charged particles migrate toward
the discharge wires and negatively charged particles mi-
grate to the grounded collection plates. When collected
particles lose their charges, they tend to agglomerate
and collect on the surfaces. Figure 5.7 illustrates the
particle charging collecting mechanism. Figure 5.8
shows the arrangement of a dry electrostatic precipita-
tor with mechanical rappers to dislodge particles from
the discharge wires and collector plates.
The removal efficiency of electrostatic pre-
cipitators for many kinds of particulate is improved if
the entering gas is conditioned by raising its moisture
content. Two tests on similarly designed dry electro-
5-29
-------
FIGURE S.7
ILICTRONIC CHARGING OP DUST PARTICLES
ELECTROSTATIC
FIELD LINES
(HIGH VOLTAGE)
DISCHARGE
ELECTRODE
PARTICLE PATH
DUST PARTICLE
GROUNDED)
COLLECTING
ELECTRODE
WIRE WEIGHT
5-30
COURTESY UNIVERSAL OIL PRODUCTS
-------
FIGURE 5.i
DRY ELECTROSTATIC PRECIPITATOR ARRANGEMENT
0523
DISC'
COURTISY INOU$T«IAl CAS CLEANING INSTITUTi
5-31
-------
•0-524
static precipitators operating on prebake potline efflu-
ents showed 98 percent removal efficiency with 2.4 mol
percent moisture and 91.5 percent removal with 0.7 mol
percent moisture. 5.17/ Sprays or other devices up-
stream of an electrostatic precipitator may be required
to realize its full removal potential. When applied to
VSS or HSS Soderberg potlines, precipitators are usually
preceded by a wet scrubbing device which both conditions
the gas and removes most HF.
Electrostatic precipitators fall into two cat-
egories, dry or wet, depending on whether the collected
particulates are knocked off the plates and wires by me-
chanical rapping to be gathered dry in a hopper, or
whether the plates and wires are washed with falling
water or electrostatically collected mist and the par-
ticulates removed as a slurry.
Dry electrostatic precipitators find useful
application in prebake and VSS Soderberg primary potline
effluent cleaning, and in the collecting of dry particu-
lates from materials handling, grinding, and grit blast-
ing operations such as occur in carbon plants. Wet
electrostatics are applied effectively to VSS and are
being tested on a prototype basis on HSS Soderberg pri-
mary potline effluents.
Electrostatic precipitators, both wet and dry,
operating on potline primary effluent streams report de-
sign and operating removal efficiencies ranging from 60
percent to 99 percent on particulate. Unlike many types
of control equipment, electrostatic precipitators may be
designed for almost any selected efficiency. By using
conservative design dimensions, by controlling humidity
of the incoming gas, and by operating at high voltage,
both wet and dry precipitators can achieve 98 to 99 per-
cent removal of potline effluent particulates.
Table 5.5 summarizes available data on per-
formance of systems incorporating electrostatic precipi-
tators in potline applications and shows the values
selected for use in the model systems analysis in Sec-
tion 8 of this report.
5-32
-------
Table 5.5 - ELECTROSTATIC PRECIPITATOR PERFORMANCE DATA AND MODEL SUMMARY
IJI
1
co
co
System
Number
1
2
3
4
5
6
7
8
Model
Model
Model
Model
Anode
Type
PB
PB
PB
PB
VSS
VSS
VSS
HSS
PB
VSS
HSS
VSS
Equipment
Designator
DESP
DESP
DESP & WS
MC & WESP &
MC & DESP
WS & WESP
WS & WESP
WS & WESP
DESP
DESP
WESP
WESP
Removal Efficiencies, % Fluoride
Gaseous Particulate Total
WS
99.9
98
91.5
90-98
98.9
98 ,
98
98
98
99
94-96
87-93
99.6
98.89
Data
Source
5.5 5.8 5.18 5.197
System No. 1
System No. 5
System No. 8
System No. 6
Equipment Designator Code
DESP - Dry Electrostatic Precipitator
WS - Wet Scrubber
MC - Multiple Tube Cyclone
WESP - Wet Electrostatic Precipitator
Source Code
I.Q. - Industry Questionnaire
P.C. - Private Communication From Equipment Manufacturer
O
cn
ro
cn
-------
0526
f) Spray Tower
The spray tower is the most common type of
removal equipment used in aluminum potline effluent
service. Properly operated and maintained these devices
can achieve removal efficiencies for potline HP ranging
from the low to high nineties. Particulate removal
efficiencies for spray towers alone are much lower, but
in combination with multiple cyclones in prebake and VSS
Soderberg service overall particulate efficiencies in
the eighties are achieved.
The term spray tower is applied to gas scrub-
bing devices in which the gas passes through an enclo-
sure at relatively low velocity and is contacted by wa-
ter, alkaline liquor or limed water liquor, sprayed from
headers usually in counterflow with the gas. In prebake
or HSS potline service the units may range frorii 38,000
to 630,000 ACFM capacity and may spray from 1.7 to 10
gallons of liquor per thousand cubic feet of gas. A typ-
ical spray tower in prebake service uses water or limed
water and consists of an open top redwood tower, 12 to
15 feet diameter and 40 to 70 feet high, with cyclonic
inlet breeching and a mist eliminator at the top. Liquor
may be sprayed down from the top or at several eleva-
tions in the tower.
\
Compared with other types of wet scrubbing
equipment, spray towers show relatively low removal ef-
ficiency for fine particulates. Figure 5.9 shows remov-
al efficiency versus particle size for a gravity spray
tower spraying at a rate of 18 gallons of water per thou-
sand cubic feet of air. The spray tower for which frac-
tional efficiency curves are shown in Figure 5.10 uses
only 5 gallons per thousand cubic feet in high pressure
sprays and achieves much higher particulate removal ef-
ficiency.
HSS Soderberg effluent gases contain unburned
hydrocarbons to the extent that dry collectors upstream
of a scrubber cannot be used because they would become
fouled by condensed tars. Spray towers in HSS Soderberg
service appear to perform less efficiently than similar
scrubbers in prebake or VSS Soderberg service. This has
been suggested to be the result of an interference by
the hydrocarbons in the wetting of the particulates and
diffusion of HF to the spray droplets.
5-34
-------
FIGURE 5.9
FRACTIONAL REMOVAL EFFICIENCY
GRAVITY SPRAY TOWER
0527
99.99
u
at
X
o
z
HI
O
2
3 0
70
1 0
001
0.05 0.1
051 5
PARTICLE SIZE, MICRONS
so too
Source: Ref. 5.21y
5-35
-------
.99
u
te
X
o
Ul
5
y
U
z
O
lu
tt
99.9
99-8
99.5
99
98
95
90
80
70
6 0
5 0
4 0
3 0
20
1 0
o.oi
FIGURE 5.10
FRACTIONAL REMOVAL EFFICIENCY
HIGH PRESSURE SPRAY TOWJRS
I 1 1 | 1 I 1 1
1 1 i II 1 !
SPRAY PRESSURE
400 PSI
' '
i i I ;
i r 1 I I 1
\ \ 1 1 1 1 1 1
0.05 O.I
0.5 1
1 0
PARTICLE SIZE, MICRONS
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
so 100
Source: Ref. 5.22/
5-36
-------
0529
Table 5.6 presents operating characteristics
and removal efficiencies for several aluminum potline
effluent control systems which include spray towers.
Removal efficiency data for spray scrubbers alone are
sparse. The interpretation of industry response is com-
plicated by the fact that various efficiencies were re-
ported for systems using various types of scrubbing
liquors, various inlet scrubbing liquor concentrations
and various bases (overall fluoride, gaseous fluoride
and total particulate, and gaseous fluoride and partic-
ulate fluoride).
The scrubber liquor chosen for the model is a
limed water with an inlet concentration of 200 parts per
million fluoride. Model efficiencies correspond to this
type liquor at this concentration; any higher inlet liq-
uor concentrations with the same inlet gas concentration
can be expected to yield lower gas efficiencies. This
liquor circuit includes a treatment facility which pre-
cipitates the fluoride as calcium fluoride and returns
the overflow liquor at 200 ppm fluoride to the scrubber.
HF Removal Efficiencies
The removal of HF in the systems reported in
Table 5.6 results from the scrubbing action of the spray
towers, as the other components are dry removal equip-
ment preceding the scrubber in the system of a wet elec-
trostatic precipitator following.
From the internal evidence contained in the
industry reports, the HF removal efficiencies which can
be achieved for prebake and VSS Soderberg service spray
tower scrubbing with lime treated water recycle circuits
assumed for the model are 95 and 99 percent respectively.
The best HF removal efficiency reported for a composite
of several spray towers in HSS Soderberg service was 93
percent. These removal efficiencies were adopted for
use in the model analyses of Section 8.
Particulate Removal Efficiencies
Spray tower removal efficiencies for partic-
ulate fluorides are directly derivable from only very
limited reported data and usually have little utility
because most control systems include some device in ad-
dition to a spray tower for particulate removal. However
5-37
-------
Table 5.6 - REMOVAL EFFICIENCIES OF SPRAY TOWER SYSTEMS
en
I
OJ
00
No.
1
2
3
4
5
6
7
8
9
10
11
12
Anode
Type
PB
PB
PB
PB
PB
PB
PB
PB
VSS
VSS
HSS
HSS
Equipment Air Pres. Drop
Designator
MC & ST
MC & ST
MC & ST
ST
DESP & ST
DESP & ST
DMS & ST
MC & DESP & ST
B & ST
B & ST & WESP
ST
ST
In. Water
0.5
1.8
3.0
3.0
3.0
8.0 (Total)
4.0 (Total)
2.5
8.0 (Total)
-
-
Liquor
Gal/Mcf
6.7
10. O1
3.5
1.7
5.3-8.02
5.3-8.03
6.7-9.S2
3.53
9.43
30
1.6-5.04
3.9
Removal Efficiency, %
HF
88.9
-
—
98.4
95*
98.4
89.8
-
_
934
90.7
Particulate
94.8
87.5
—
89-90
95
_
_
-
_
63. 54
79.6
Total F
88.5
-
79.2
73
94-96
94-96
78-86
87-93
90-95
98-99
81. 44
87.0
Notes
1. Scrubber liquor contained 2000 to 4000 ppm dissolved fluoride for cryolite recovery.
When operated with limed liquor, reported efficiency was 93 to 95 percent.
2. Once through water circuit.
3. Recycled limed liquor circuit.
4. Composite of multiplant results.
* Derived from reported data.
Legend
MC - Multiple Cyclone
ST - Spray Tower
DESP - Dry Electrostatic Precipitator
DMS
B
- Dynamic Mechanical Separator
- Burner
WESP - Wet Electrostatic Precipitator
-------
0531
for the sake of model analysis the following solid F
removal efficiencies have been derived for spray towers
alone. Where a spray tower follows a solids collector,
the particulate removal of the spray tower has been
ignored.
Prebake
Comparison of the fractional efficiency curves
for spray towers and multiple tube cyclones indicate that
both types of equipment should show approximately equal
overall removal efficiencies for prebake particulate ef-
fluent. Therefore, the value 80 percent was taken, equal
to multiple tube cyclone particulate fluoride removal
efficiency for prebake.
VSS Soderberg
Assuming that VSS Soderberg particle size dis-
tribution is somewhat finer than prebake, the assumed
removal efficiency for spray towers on VSS solid fluoride
particulate is taken as 75 percent.
HSS Soderberg
A composite report of multiplant experience
on HSS Soderberg solid fluoride particulate removal in
spray towers shows 64 percent removal efficiency. This
is used in the model analysis of Section 8.
Table 5.7 summarizes the model efficiencies
used in Section 8 for spray towers.
g) Spray Screen
The term spray screen scrubber is applied to
wet scrubbing equipment in which the liquor is sprayed
into an effluent gas stream and on to screens or open
mesh filters enclosed in a plenum chamber. The assembly
also usually includes a mist eliminator. Effluent gas
flow may be powered by exhaust fans, or may be moved by
unpowered convection.
The removal mechanisms, as in other wet scrub-
bers, are inertial impaction on, interception by, and
adsorption into, the liquid droplets or filters.
5-39
-------
Table 5.7 - SUMMARY OF MODEL REMOVAL EFFICIENCIES FOR SPRAY TOWERS
ESTIMATED FROM INDUSTRY QUESTIONNAIRE DATA
r
C
C
r
i
*>
o
Application, Cell Type
Prebake
VSS Soderberg
HSS Soderberg
Model Efficiencies, % Fluoride
Gaseous Particulate Total
95
99
93
80
75
64
88.7
93.0
82.0
-------
0533
The low gas pressure drop across spray screen
scrubbers and the relatively low power cost recommends
them for secondary, or potroom scrubbing service. Figures
5.11 through 5.14 illustrate several designs of spray
screen scrubber installations which have been used in the
primary aluminum industry.
Performance data on spray screen installations
are scant and are difficult to establish with accuracy
because of the large volumes and low concentrations han-
dled in the systems to which they are applied. One
foreign prebake plant reports 90 percent total F removal,
5.26/, and another reports greater than 90 percent for
HF and approximately 50 percent for particulate fluoride
in secondary scrubbing without primary control, 5.27/.
One VSS Soderberg plant using a spray screen secondary
in addition to a primary system reports for the secondary
system 88 percent HF, 77 percent total particulate, 42
percent solid F, and 72 percent total F removal efficien-
cies, 5.20/.
Removal efficiencies used in the model anal-
ysis of Section 8 were derived from these data and from
limited information in the industry questionnaire re-
sponses. Values chosen are:
Solid F HF
Secondary System
No Primary 45 93
Secondary System PB 25 80
With Primary VSS 42 88
HSS 25 80
While more sophisticated scrubbing devices
achieve higher removal efficiencies on both particulates
and HF than does the spray screen, the costs are 30-100
percent greater and the cost effectiveness much lower,
when applied to secondary treatment. It is the consensus
of the industry that, for secondary treatment in combina-
tion with primary control, the cost differential would be
more effectively invested in improved primary collection
and removal equipment. Among the alternative secondary
scrubbers only the spray screen is considered econom-
ically feasible.
5-41
-------
053H
FIGURE 5.11
UNPOWERED ROOF SPRAY SCREEN
Source: Ref. 5.8/
5-42
-------
FIGURE 5.12
POWERED CELLROOM SPRAY SCREEN SCRUBBER
CLEANED AIR
AIR FROM
POTROOM
SCREEN
V
V
-c
-<
-c
'AIR FROM
POTROOM
FOUR BANKS OF
FINE SPRAYS
COARSE
SPRAYS
MIST
ELIMINATORS
Source: Ref. 5.20/
CD
CJ1
CA.)
01
-------
FIGURE 5.13
POWERED SPRAY SCREEN SCRUBBER
O
cn
c>>
cn
un
I
FAN
SPRAY NOZZLES
SEPARATOR
SECTION
PUMP
Source: Raf. 5."6
-------
FIGURE 5.14
POWERED MONITOR SPRAY SCREEN SCRUBBER
EXHAUST FAN
Ul
I
t_n
SPRAY NOZZLES
PLASTIC WIRE MESH
WATER COLLECTING TROUGH
Source: Ref. 5.2V/
CD
01
GO
-------
h) Venturi Scrubber
The venturi type scrubber contains a reduced
flow area or throat section in the main air duct in
which the velocity is increased many-fold, followed by
a diffuser section. Scrubbing liquor introduced at the
throat is sheared by the high velocity gas and is dis-
persed and mixed with the gas as an extremely fine spray.
Droplets interact with HF and particulates in cell efflu-
ent streams and are removed from the gas stream in some
form of entrainment separator. Full advantage of the
potential of a venturi as a scrubber is realized if the
gas entering the throat is saturated with water vapor.
Additional injected water evaporating in the throat will
condense out when the static pressure rises in the dif-
fuser section, thus providing additional sites or nuclei
for absorbing gas and particulates.
Normally operated at 20 to 40 inches of water
pressure drop, the venturi shows removal efficiencies
in the high nineties for both HF and most particulates.
Incomplete tests with the effluent from HSS Soderberg
potlines indicate that the removal efficiencies for hy-
drocarbon fumes may be lower.
Venturi scrubbers have been applied to the
control of VSS Soderberg potline effluents and are cur-
rently undergoing tests in bake plant service. Oper-
ating costs for high particulate removal are considered
by the industry to be unacceptable for prebake service.
Industry questionnaire response for VSS Soder-
berg services reports for a venturi system (with a pack-
ed bed separator section), an overall fluoride removal
efficiency of 98 to 99 percent with a gas pressure drop
of 30 inches water column and a liquor rate of 5 to 10
gallons per thousand cubic feet. This overall efficien-
cy, based on reported fluoride distribution between gas
and particulate and an assumed gas removal efficiency
of 99 percent (because of packed separator section),
yields a fluoride particulate removal efficiency of 96
percent. These values were adopted for the model anal-
ysis in Section 8.
5-46
-------
0539
i) Chamber Scrubber
The chamber scrubber consists of spray nozzles
and venturi throat pieces, assembled in a staggered ar-
rangement in a horizontal cylindrical steel vessel which
may be lined for corrosion protection. The effluent
gases are scrubbed by liquid spray action which provides
enough suction to partially entrain particulate and gas
and carry them into the throat section where thorough
mixing occurs. Gases cannot bypass the spray and the
accelerated gas velocity provides opportunity for high
impact of spray particles. The numbers of passes, ven-
turi throats and nozzles are a function of the volume
and characteristics of the gas and the level of removal
efficiency desired. The final design is customized to
meet service conditions.
This type of scrubber has been recently in-
stalled in a prebake potline primary control system but
no operating data are available. Costs for application
in secondary treatment are considered by the industry to
be unacceptable.
j) Cross Flow Packed Bed Scrubber
Packed bed wet scrubbers remove entrained par-
ticulates by inertial impingement against wet surfaces
of packing material and remove HF gas by diffusion. The
interception mechanism, by which particulates in the one-
micron range are most readily collected, plays a minor
role in packed bed scrubbers. It has been suggested that
similar electrostatic charges on the particulates and
packing tend to repel the lighter particulates. 5.23/
For this reason, packed bed scrubbers may show relative-
ly poor efficiency in controlling the approximately half-
micron particulates which contribute most to opaque
emissions.
The crossflow configuration shown in Figure
5.15 may have advantages over the older vertical packed
tower arrangement where the gas flows up countercurrent
to scrubbing liquor which drips down through the bed.
The crossflow bed has an inherently lower pressure drop
than a vertical bed for equal capacity because the liq-
uid and gas streams interact at right angles rather than
in direct opposition. Further, this flow geometry is
responsible for the non-flooding characteristics of the
5-47
-------
FIGURE 5.15
CROSSFLOW PACKED BED SCRUBBER
O
cn
JT
O
LIQUID INLET
ui
I
*>
00
CONTAMINATED
GAS
CLEANED
GAS
COURTESY CEIICOTE COMPANY
-------
crossflow bed. The first foot or so of packing may be
washed with considerably more liquor than the back part
of the bed effecting turbulent washing action to remove
most of the particulate and to prevent blinding. In
addition, if the gas can be cooled below its dewpoint
early in the bed, condensate nuclei formed downstream
will provide additional sites for absorbing gas and par-
ticulates.
The crossflow packed bed is used in one sec-
ondary application on prebake potlines. Based on avail-
able information the following values have been selected
for model analysis:
Removal Efficiency, Percent
Application Solid F HF
Primary Systems
(5-foot bed)
Prebake 87 !/ 98
VSS Soderberg 87 -i/ 98
HSS Soderberg 81 I/ 98
Secondary Systems (No Primary)
(3-foot bed)
All Potlines 84 2/ 99
_!/ Equipment manufacturers estimates
_2/ Industry reported data for prebake application
k) Floating Bed Scrubber
The floating bed scrubber is a special case
of the packed bed scrubber concept in which the packing
consists of hollow plastic spheres, approximately 1.5
inches diameter, buoyed and agitated by the rising efflu-
ent gas stream. Scrubbing liquor flows down through the
bed carrying away the collected particulate and dissolved
gas. Although the typical unit has only one stage, two
or more may be used. One example of a single-stage float-
ing bed scrubber uses perforated grid plates 18 inches
apart caging a 12-inch bed of balls when they are at rest.
At normal gas velocities, 400 to 600 feet per minute, and
10 to 20 gallons per minute liquor flow per square foot
of scrubber cross section, the gas pressure drop ranges
from about 2 to 6 inches water gage.
5-49
-------
The self-cleaning action of the balls rubbing
against each other makes the floating bed scrubber suit-
able for applications in which the particulate pollutant
may have sticky or flocculent characteristics which could
plug a fixed bed scrubber. HSS Soderberg potlines re-
lease tarry mists which have been controlled successfully
in a foreign installation by floating bed scrubbers ob-
taining 90 percent total fluoride and 98 percent HF re-
moval. 5.257
The VSS Soderberg Sako plant at Sundsvall,
Sweden, uses floating bed scrubbers on both cell gas or
primary collection streams (Figure 5.16) and on potroom
or secondary streams. The operators report 78 percent
removal efficiency of solids and 97.5 percent of total F
with two-stage scrubbers used for primary system control.
See Figure 5.16. In the secondary system with single-
stage scrubbers, Figure 5.17, they report 70 percent re-
moval of solids and 87 percent of total F. Operating con-
ditions are given in Table 5.8. 5.4/
One prebake plant in the United States uses
floating bed scrubbers. The operators report HF removal
efficiency in the high nineties. At another plant,
floating bed scrubbers are installed on HSS Soderberg
lines.
1) High Pressure Spray Screen
The high pressure spray screen wet scrubber
is characterized as having high pressure liquor sprays
which impinge on a grid or screen to bounce off and pro-
vide violently agitated mixing between the liquor and
gases passing through the spray and screen. Figure 5.18
illustrates one such scrubber using 200 psig cocurrent
sprays striking a membrane.
Results of pilot scale tests of high pressure
spray screens on HSS Soderberg primary effluents using
river water as scrubbing liquor are shown in Table 5.9.
Removal efficiencies for particulates, especially the
total solids which are primarily non-fluorine, are lower
than might be expected. This may be attributable to the
suspected tendency of HSS hydrocarbons to interfere with
wetability of particulates. Undocumented tests on pre-
bake effluents are reported to show the same or slightly
5-50
-------
•'•8543
FIGURE 5.16
FLOATING BED SCRUBBER IN PRIMARY VSS SODERBERG SERVICE
i
ENTRAPMENT SEPARATION GRID
PLASTIC NET
FLOATING BED
SPRAY NOZZLE
SLURRY
DRAIN
Source: Ref. 5.47
5-51
-------
FIGURE 5.17
FLOATING BED SCRUBBER IN SECONDARY VSS SODERBERG SERVICE
STACK
ENTRAPMENT
SEPARATION GRID
CELL ROOM
AIR
Source; Ref. 5.4/
5-52
-------
05^5
Table 5.8 - GAS CLEANING OPERATING PARAMETERS
FOR SAKO, SUNDSVALL SWEDEN PLANT
USING FLOATING BED SCRUBBER FOR VSS SODERBERG SERVICE 5._4/
Primary
Gas Flow, scf/lb Al 212
Water Flow, gal/Mscf 10.9
Pressure Drop
Scrubber, in. water 5.1
Total, in. water 27.6
Sprays, psi 7.1
Contacting Power
Gas, HP/Mef 0.8
Liquid, HP/Mef 0.045
Total, HP/Mef 0.845
Removal Efficiency, %
Solids 78
Fluorides 97.5
Secondary
31,750
3.1
0.59
0.99
17.1
0.0925
0.0370
0.1295
70
87
5-53
-------
FIGURE 5.is
HIGH PRESSURE SPRAY SCREEN
DIRTY GAS
QUENCH SPRAYS
REBOUND ZONE
ACCESS HATCH
ROD SCREEN MEMBRANE
SPRAY
WATER-^
MANIFOLD
SPRAY-
NOZZLES
LIQUOR EFFLUENT
SEAL POT
BASE FRAME —
MOISTURE ELIMINATORS
CLEAN AIR
5-54
COURTESY KREBS ENGINEERS
-------
Table 5.9 - PILOT SCALE TEST RESULTS - HIGH PRESSURE SPRAY SCREEN
ON HSS SODERBERG PRIMARY SYSTEM 5.28/
Scrubbing
Stages
Single
Two
Fresh
Total gpm
26
38
Water
Gal/Mac fm
5.7 to 17
8.3 to 25
Removal Efficiency, %
HF
95.3
96.6
Solid F
82.7
81.2
Total Solids
71.6
74.4
Three 50 10.9 to 33 98.4 86.5 77.3
Ul
I
Ul
Ul
Operating Parameters
Spray Pressure - 200 psig
Water Quantities - 14 gpm prequench, 12 gpm per stage
Superficial Gas Velocity - 250 to 767 feet per minute
o
en
-------
higher removal efficiencies for total fluoride. These
test results are combined and interpreted to yield the
HP and solid F removal efficiencies used in the primary
systems model analysis of Section 8.
Table 5.10 shows results of test work with
single and three-stage high pressure spray screens treat-
ing HSS Soderberg primary effluents after they had been
treated by a spray tower.
m) Other Wet Scrubbers
There are a number of other wet scrubbing de-
vices which have been used to remove particulates and sol-
uble gases from an effluent stream. Some of these have
been applied to domestic aluminum reduction plant efflu-
ents and have been replaced by other equipment. Others
have been reported to be in use in overseas plants, but
the information concerning them is open to question. The
equipment has not been widely adopted in the industry.
i. The Wet Centrifugal or Multivane Gas
Scrubber consists of a tower with two or more vane assem-
blies which impart a swirling action to dust-laden gas
passing upward through them from a bottom inlet. Scrub-
bing liquor is distributed through low pressure sprays
over the lower vane assemblies, forming a turbulent layer
of liquid in which fine particulate are absorbed as the
gas passes through. Liquor flows down through the vane
assembly, wetting all surfaces, and passes through a gas
inlet region at the bottom of the tower where the heav-
iest particles in the gases are removed. Particulates
leave the system as a slurry from the conical bottom of
the tower. A vane-type mist eliminator at the top tends
to prevent droplets leaving with the exit gas.
The multivane wet centrifugal scrubber was
applied to a domestic VSS Soderberg potline but has been
replaced with higher efficiency scrubbers. No perform-
ance data are reportable.
ii. Dynamic Wet Scrubbers for particulate
control have been developed in several configurations.
Means are provided to wet the blades and case of a high
speed gas fan or blower which mixes gas, dust and scrub-
ber liquor in extreme turbulency to force dust particles
5-56
-------
Table 5.10 - TEST RESULTS - HIGH PRESSURE SPRAY SCREEN AS SECOND STAGE SCRUBBER,
HSS SODERBERG PRIMARY SYSTEM 5.28/
I
U1
Single Stage
Test
No.
1A
2A
Capacity
acfm
3000
3000
Fresh Scrubbing Water
Temp.
°F
58
58
Flow
30
30
Pressure
psig
200
200
Removal Efficiency
As Total F
86.71
86.67
3 Stages
IB 3000
2B
3000
58
58
60
60
200
200
89.00
89.00
O
Cn
-f-~
-------
0550
into the liquid. One embodiment of the dynamic scrubber
is as a second stage integral with a multivane wet scrub-
ber. Gas, dust and liquid droplets from over the vane
assembly are drawn into a centrifugal fan where more liq-
uor is introduced at the hub. The fan discharges into
a cyclonic mist eliminator from which gas exits at the
top and liquor drops down to the vane assembly. The
scrubber is not known to have been applied in the alumi-
num industry.
iii. Self-Induced Spray, or bubble type scrub-
ber is one in which dust-laden gas is forced down into a
pool of scrubber liquor forming small bubbles with vi-
olent agitation. This type of scrubber is particularly
applicable to dusts which are sticky, since the equipment
has no close clearances, grids, or surfaces which could
built up deposits and plug. It has been applied success-
fully to VSS Soderberg and bake plant effluents but no
performance data are reportable.
iv. The Sieve Plate Tower accomplishes its
scrubbing action by bubbling the effluent gas stream up
through flooded perforated plates which retain layers of
scrubber liquor a few inches deep. Liquor is introduced
by sprays into the upward moving gas stream and over-
flows from an upper to a lower stage plate at a controlled
rate, carrying the captured particulates finally to a
sump at the bottom of the tower.
Three and four-stage sieve plate towers have
been applied to VSS Soderberg effluents at a smelter in
Norway. Although data are incomplete, these installa-
tions are reported to achieve removal efficiencies of 70
percent on total particulate, 80 percent on solid flu-
oride and greater than 97 percent on HF. 5.29/ Later
tests were reported to achieve as high as 97 percent on
water soluble fluoride and 99.3 on HF. 5.77
v. The Wet Impingement Scrubber Tower con-
tains flooded perforated plates somewhat similar to those
of the sieve plate tower, but close to and facing the
discharge of each orifice is a wetted surface against
which the gas strikes before turning and bubbling up
through a layer of scrubber liquor on top of the plate.
Sprays under the plates introduce water to wash particu-
lates from the impingement surfaces.
5-58
-------
0551
The wet impingement scrubber probably could
be used in potline effluent service but no report of such
installation or test has been found. One test on bake
plant effluent was unsuccessful; the scrubber plugged.
n) Incinerators
Unburned pollutants in effluent streams from
anode bake plant ring furnaces may include particulate
carbon, particulate hydrocarbons, tar mists and combus-
tible gases, but the concentrations are too low to sup-
port combustion. A gas-fired incinerator or burner has
been proposed to control these pollutants but no appli-
cation in the aluminum industry has been made to date.
5-59
-------
0552
References - Section 5
5.I/ Less, L.N. and Waddington, J. "The Characteriza-
tion of Aluminum Reduction Cell Fume" Light Metals.
Proceedings of Symposia, 100th AIME Annual Meet-
ing, New York - March 1-4, 1971.
5.2/ Wendt, G. "Operating Experiences With Electro-
lytes Containing Lithium Fluoride" TMS of AIME
Paper No. A70-39 - February 16, 1970.
5.3/ Henry, J.L. "A Study of Factors Affecting Flu-
oride Emission from 10 KA Experimental Aluminum
Reduction Cells" Extractive Metallurgy of Aluminum
Vol. 2:67-81 Interscience Publishers, New York,
(1963).
5.4/ Brenner, E.M. "Gas Collection, Cleaning and Con-
trol at Sako, Sundsvall Works" TMS of AIME Paper
No. A70-14 - February 16, 1970.
5.5/ Schmitt, H. "The Fluorine Problem in Aluminum
Plants" (Trans, from German) Aluminum 2:97-102
(1963).
5.6/ Strauss, W. "Industrial Gas Cleaning" Pergammon
Press, New York (1966).
5.7/ Erga, 0. et al. "The Gas Purification Plant at
Mosj0en Aluminum Works" (Trans, from Norwegian)
Tekn, Ukebl, 114 (12): 232-236.
5.8/ Callaioli, G., et al. "Systems for Gas Cleaning
in Electrolytic Cells of Montecatini Edison Alu-
minum Plant" TMS of AIME Paper No. A70-57 -
February 16, 1970.
5.9/ "Fundamentals of Fabric Collectors and Glossary
of Terms", F-2, industrial Gas Cleaning Institute,
Inc., A Publication prepared by Fabric Collectors
Division, Rye, New York (1969).
5.10/ Private Communication, Wheelabrator Corporation,
Wheelabrator Division, Mishawaka, Ind.
(September 1970).
5-60
-------
0553
Doerschuk, V.C. "Electrolytic Production of Alu-
minum" Canadian Patent No. 613,352 (Jan. 24, 1961).
Pring, R.T. "Method for Removal of Halides from
Gases" U.S. Patent No. 2,919,174 (Dec. 29, 1959).
Cook, C.C. and Knapp, L.L. "Treatment of Gases
Evolved in the Production of Aluminum" U.S. Pat-
ent No. 3,503,184 (March 31, 1970).
Cook, C.C. and Swany, G.R. "Evolution of Fluoride
Recovery Processes Alcoa Smelters" Light Metals
1971, A Publication of Proceedings of Symposia
100th AIME Meeting, NlY. (March 1-4, 1971).
Cook, C.C., Swany, G.R. and Colpitts, J.N.
"Operating Experience With the Alcoa 398 Process
for Fluoride Recovery", Journal of the Air Pollu-
tion Control Association, Vol. 21, No. 8 (Aug.1971)
Cochran, C.N., Sleppy, W.C. and Frank, W.B.
"Chemistry of Evolution and Recovery of Fumes in
Aluminum Smelting" TMS'of AIME Paper No. A70-22
(Feb. 16, 1970).
Oglesby, S. and Nichols, G.B. "A Manual of Elec-
trostatic Precipitator Technology" Part I and II,
Southern Research Institute, Birmingham, Ala.,
A Publication of National Air Pollution Control
Admini stration.
"Restricting Dust and Gas Emission in Bauxite and
Aluminum Processing Plants" Verein Deutscher
Inginieure Paper No. 2286 (Nov. 1963)„
Schmitt, H. "Further Developments in the Solution
of the Fluorine Problem in the Aluminum Industry"
(in German, translated by the Franklin Institute
Research Laboratories) Erzmetall, 18(3): 111-115
(1965). r
Byrne, J.L. "Fume Control at Harvey Aluminum"
Annual Meeting Pacific Northwest Section, Air
Pollution Control Association, Spokane, Wash.
(Nov. 16-18, 1970).
5-61
-------
Stairmand, C.J. "The Design of Modern Gas Clean-
ing Equipment" Journal of the Institute of Fuel,
London, (Feb. 1956).
Private Communication, S. H. Holt, Buffalo Forge
Company, Buffalo, N. Y. (Dec. 1970).
Eckert, J.S. "Use of Packed Beds for Separation
of Entrained Particles and Fumes from an Air
Stream", Journal Air Pollution Control Association,
Vol. 16, No. 2 (Feb. 1966).
Private Communication, E. Hanf, Ceilcote Company,
Berea, Ohio (Jan. 1971).
Kielback, A.W. "Progress by the Aluminum Company
of Canada, Ltd. in Air Pollution Control" A Paper,
818-11, prepared for the National Conference on
Pollution and Our Environment, Montreal (Oct. 31-
Nov. 4, 1966).
Calvez, C. et al. "Compared Technologies for the
Collection of Gases and Fumes and the Ventilation
of Aluminum Potlines" A Paper presented at the
AIME Meeting, Chicago (Dec. 11-16, 1967).
Moser, E. "The Treatment of Fumes from Primary
Aluminum Reduction Plants" Paper No. 12, Presented
at International Conference on Air Pollution and
Water Conservation in the Copper and Aluminum In-
dustries (Oct. 21-23, 1969).
Private Communication, J. Melin, Krebs Engineers,
Menlo Park, California (Nov. 1970).
Erga, 0., et al. "Selective Absorption of Flu-
orine from the Gases from Aluminum Reduction Cells
With Vertical Spike Soderberg Anodes" Aluminum:
83-103 Tidsker, Kjemi Bergv. Met. 23(5), Chem.
Abstracts, 1965 No. 14,400a, (1963).
5-62
-------
Table of Contents
Section 6
6.0 Source Sampling, Ambient Air Sampling and
Analytical Techniques
6.1 Emission Sampling and Analytical Techniques
6.2 Source Sampling - Primary System Emissions
6.2.1 Sample Extractions
6.2.2 Sample Recovery
6.2.3 Sample Treatment
6.3 Source Sampling - Secondary Systems
6,13.1 Methods of Secondary Sampling
6.3.2 Sampling Trains
6.4 Costs and Manpower Requirements, Source Sampling
6.5 Analytical Determination of Fluorides
6.6 Ambient Air Sampling
6.7 EPA Sampling and Analytical Techniques
-------
n
L
6.0 Source Sampling, Ambient Air Sampling
and Analytical Techniques
The effective management of air pollution abate-
ment in connection with the production of aluminum
requires reliable information on the effluents and
emissions of the plant and the effects of plant emis-
sions at locations outside the plant property. Source
sampling and analysis provides the data with which to
evaluate the performance of pollution generating proc-
esses and the equipment used to collect and remove
pollutants from exit streams. Ambient air sampling
provides quantitative data on the concentration of
specific pollutants at locations outside the immediate
plant site.
This section of the report presents a brief
introduction to some of the problems and procedures
presently associated with sampling technology.
Table 6.1 indicates the orders of magnitude of
total fluoride concentrations representative of the
various samples which may be obtained in source and
ambient air testing.
Table 6.1
Typical Total Fluoride Concentration
(Micrograms per Cubic Meter)
Primary Effluent 120,OOQl/
Controlled Primary Emission 12,000i/
Secondary Effluent 600i/
Controlled Secondary Emission 240—'
Ambient Air 0
_!/ Derived from Industry Questionnaire data.
_2/ Approximate limit of analysis 6.5/.
6-1
-------
0557
6.1 Emission Sampling
and Analytical Techniques
The procurement and analysis of data with which
to determine the effectiveness of pollution abatement
systems in the primary aluminum plant presents a number
of problems of sampling which are difficult to solve.
These problems are concerned primarily with obtaining
representative samples of pollutants from large volume,
low velocity air flows in secondary emissions, partic-
ularly in unpowered roof monitors. They are also com-
plicated, in some streams, by water saturation and mist
carryover from wet scrubbing operations, and also by
cyclical or erratic condtions in the potrooms which
result in non-uniform content of the gas system.
In plants where primary collection is employed,
the larger portion of the total emissions occurs in
relatively low flow, high velocity exhausts, and a
good degree of confidence can be placed in the results
of the sampling, reducing the probable errors in total
emissions determination introduced by less dependable
sampling of the secondary volumes.
Emission analysis needs to discriminate between
gaseous and particulate compositions of the emissions,
and this is complicated by the high reactivity of the
gaseous fluorine compounds. The usual chemical anal-
ysis techniques cannot distinguish between fluoride
originating from HF or SiF4 and that originating from
dissolved solids, cryolite or aluminum fluoride, so
where discrimination is required, special techniques
are employed to collect gaseous fluorides and solids
separately.
Another problem is to provide information from
emission sampling which can be used to correlate par-
ticulate quantity in emissions with ambient dust fall
at points removed from the source. Data on particle
size and settling velocity obtained by aerodynamic
sizing of particulates in the emission sample will
assist in this correlation.
6-2
-------
0558
T*he current practices for handling these prob-
lems are discussed in general terms. Appendix 6A con-
tains information concerning specific sampling and
analysis procedures employed at several aluminum smel-
ters. Although all companies obtain samples in tech-
nically acceptable ways, their methods, types and
materials of construction of equipment, and analytical
methods differ enough that results from one company may
not be directly comparable with those from another.
Standardization of sampling and analysis procedures
would be desirable.
6.2 Source Sampling - Primary Systems
Where primary collection systems are used on
potlines, the effluents and emissions may usually be
sampled in the ductwork of the control equipment, and
as such present no particularly difficult problems of
sampling other than those which may be purely mechan-
ical. The compositions of the gas flows are reasonably
uniform over long time periods and velocities are with-
in ranges which may be determined with accuracy for
flow calculation.
6.2.1 Sample Extraction
Representative samples are extracted from the
gas stream at an isokinetic rate. Simultaneous total
flow rate determinations permit relating the measured
quantities of collected pollutants with the quantity
of air flow and, therefore, with aluminum production.
Flow Rates
Quantification of sampled pollutant to determine
total emission and relate it to aluminum production re-
quires measurement of the volume flow rate in the sam-
pled duct. Although high velocity ducted gas streams
may be metered accurately with calibrated orifice plates
or venturi restrictions and manometer pressure gages,
the gas streams of most primary systems in aluminum
smelters flow at such velocities that determinations by
Pitot-static tube and differential pressure manometer
are usually preferred. The Pitot-static tube measures
velocity pressure which is converted to the local gas
velocity through use of the gas density by the rela-
tionship,
6-3
-------
where:
C = experimentally determined coefficient,
near unity
= velocity pressure
P = gas density
g = gravitational constant
Velocity measurements may be made by traversing
the duct, (dividing the duct into a number of equal
cross sectional areas or annuli and determining the
total flow as a sum) or by making a single velocity
determination at the center of the duct and multiplying
by an arbitrary factor to correct this reading to an
average velocity across what is usually assumed to be
a fully developed turbulent flow profile. Hie former
method, while more laborious, is the more accurate.
Because of fluctuations in flow conditions caused
by cell operations over a period of time, multiple flow
measurements are required to obtain acceptable averages.
Sampling Rates
Isokinetic flow conditions in the sampling tube
are necessary if the ratio of solids to gas in the
sample is to equal that ratio in the gas stream. Iso-
kinetic conditions are realized when the sampling flow
rate and the inlet diameter of the sampling tube are
adjusted so that the gas velocity entering the tube
equals the stream velocity at the location of the sam-
pling tube inlet. When gas streamlines bend, entrained
particulates tend to continue straight, and it can be
visualized from Figure 6.1 that too slow a sampling
rate will result in a sample which is relatively richer
in particulates than the main gas stream; and sampling
too fast will draw in a gas stream relatively lean in
particulates.
6-4
-------
FIGURE 6.1
Gas Streamlines in Isokinetic and Improper Sampling
I
ui
-•»•
-»•
Sampling Too Slow
Isokinetic
Sampling Too Fast
O
O"?
cb'
CD
-------
0561
Sampling time must be long enough to ensure that
the collected sample is representative of long-time
average conditions and is not atypical because of dis-
turbances caused by periodic pot working or anode re-
placement. Three hours to three days of continuous or
frequent intermittent sampling in one location have
been reported as industry practice,
Dew point control in a sampling probe is often
required to prevent the condensation of water vapor
within the tube or on filters. Electric resistance or
radiant heaters on the sample tube, or the placement
of the particulate filter assembly within the main duct
being sampled, can overcome difficulties arising from
humid gas streams.
6.2.2 Sample Recovery
The equipment used for extraction of the sample
from the stream flow at the positions, velocity, and
time required to obtain representative samples of the
total flow, includes in sequence the sample probe, mist
eliminator, particulate separator, gas train, gas flow
meter, and gas pump. The latter is controlled to aspi-
rate the rate of gas sample corresponding to isokinetic
flow, as determined by conditions of gas flow and tem-
perature, and by size of sample probe.
There are two types of fluoride, particulate
and gaseous, in the emissions from aluminum reduction
plants which have quite different properties as to air
pollution, and which therefore require separation in
the emission sample.
Particulate separation may be accomplished by
some type of filtration through porous media, by cy-
clonic separation of relatively coarse particles, or
by settling out from the sample stream in specially
designed equipment. In the gas concentrations typical
of primary systems source sampling, the effect of HF
adsorption on collected particulates may be unimpor-
tant, 6.I/, and may be neglected. This may not be true
for low concentrations of pollutants typical of ambi-
ent air or secondary system sampling. Systems which
attempt the removal of particulates before the gas ad-
sorber have been found to yield misleading information
6-6
-------
0562
because some of the HF is collected on trapped partic-
ulate and reports out of the analysis as solid F.
Total parti.culates separated by these devices
are weighed, usually indirectly by tare differential of
the collectors, to determine total dust loadings.
Gas sampling trains collect gaseous effluents
usually by impingement in liquid filled impingers, but
sometimes, especially in ambient air sampling, by chem-
ical adsorption on treated filters or by adsorption on
alkali coated tubes. Samples may be collected in con-
tainers which are used to transport unaltered gas sam-
ples to the analytical laboratory.
Appendix 6A, Sections 6A.1 through 6A.5 describe
sampling trains and analytical techniques used at sev-
eral aluminum smelters for source testing.
6.2.3 Sample Treatment
Particulate size analysis and aerodynamic size
properties may be important in establishing a correla-
tion between emission rate and effective dust fall as
measured in ambient air quality evaluation.
Among the devices used to make this particle
size distribution analysis, the Andersen sampler is
simple and effective for measuring equivalent aerody-
namic diameter. In its usual form a series of perfor-
ated plates will separate the particulate into eight
aerodynamic size ranges, from 7.7 microns and above
on the first stage down to 0.47 micron on the eighth
stage. Particulates finer than 0.47 micron are col-
lected on a backup filter. John Nan-Hai Hu 6.2/ des-
cribes development work in which an Andersen sampler
was modified to extend the lower range to 0.17 micron
by changing the flow rate and adding filter paper baf-
fles to inhibit bouncing of particulates off the col-
lection surfaces.
6.3 Source Sampling - Secondary Systems
Collection and analysis of samples of emission
from controlled secondary systems present problems of
a different order of magnitude from those of primary
system emissions.
6-7
-------
0563
Pollutant loadings of the air streams are often
much less than 5 percent of those of primary emission
streams, and total air flows are some ten times greater.
Unpowered flows pass through very large cross sectional
areas at very low velocities, making it difficult to
find suitable sampling locations as well as reducing
the accuracy with which measured flow determinations
can be extrapolated to account for total air flows.
Low loadings result in long sampling periods to
collect significant amounts of emissions, increasing
the difficulty of maintaining constant sample extrac-
tion rates and accuracy. Cyclical operating disturb-
ances in the potroom can contribute to variations in
the secondary loading which require prolonged sampling
periods to collect representative samples.
6.3.1 Methods of secondary sampling may follow three
different routes, each with advantages and shortcomings.
The most reliable but most costly method re-
quires a number of sampling stations arranged so that
their aggregate results will fairly represent the total
emission from a whole secondary system. A single cell-
room building may be 1200 feet long with 12,000 total
square feet of secondary controlled exhaust openings.
Total flow rate may be two million cubic feet per min-
ute, corresponding to an exit velocity in the order of
three feet per second. With a velocity pressure of
only a few thousandths of an inch of water, sensitive
anemometers rather than Pitot tubes are used to meas-
ure flow rate and overall accuracy is poor. However,
four or more sampling stations on one building taking
simultaneous samples over periods of six hours or more
can yield useful data if sufficient care is exercised
in selecting sampling locations.
A second method of sampling roof emissions uses
multiple fixed sampling points connected by ducting to
draw a simultaneous composite sample to a convenient
location. This type of sampling is inexpensive when
compared to the effort involved in collecting samples
by alternate means. The cost of a manifold system is
in the vicinity of $10,000. Also, some operators be-
lieve it is better to adjust flow to an average of
6-8
-------
6
-------
.056
At the other end of the scale, the major multi-
plant aluminum producers maintain very large central
research and development groups which include fluoride
control work as a part of their activities supporting
the efforts of the plant operating staff. No meaning-
ful figures were obtained for the total annual industry
expenditure for fluoride control, but it is certainly
very substantial.
6.5 Analytical Determination
of Fluorides
Quantitative fluorine analysis of the particulate
and gaseous components of the emission samples is car-
ried out by several methods and techniques which are
authoritatively summarized in complete detail by the
recent (1968) publication "Intersociety Committee Man-
ual of Methods for Ambient Air Sampling and Analysis"
6.3/. The section of this procedure concerned with
fluorine analysis describes fluoride ion isolation by
the techniques of Willard and Winter distillation, ion
exchange, and diffusion. It includes the quantitative
determination of fluoride titrimetric and spectrographic
methods.
A later development has been the availability of
the fluoride ion activity electrode, the use of which
is analogous to hydrogen ion determination by pH meter.
6.6 Ambient Air Sampling
The foregoing discussion has been concerned with
source sampling and analysis, testing done at a smelter
control system to determine effluent and emission quan-
tities and to evaluate the performance of pollution
abatement equipment. Under these conditions, gas
streams are usually moving at finite and measurable
velocity so that isokinetic sampling is important and
the pollutant concentrations are usually relatively
high. Ambient air testing, on the other hand, is used
by several aluminum smelters to determine very dilute
concentrations of particulate and gaseous fluorides in
the air surrounding smelter sites. Only the finest
particulates remain suspended and isokinetic sampling
is not applicable in ambient air sampling.
6-10
-------
05G6
Whereas source sampling is usually conducted on
a campaign basis with close attention given to con-
trolling sampling rate and measuring flow conditions
over a period of a few hours, ambient sampling may be
automated to the extent that the apparatus can run un-
attended for long periods of time.
The Boyce Thompson Institute for Plant Research,
Inc., among others, has done a considerable amount of
research and development work on ambient sampling and
analysis for fluorides, and they have attempted to cor-
relate ambient and dust fallout fluoride levels with
damage to various kinds of plant life. References 6.4/
and 6.5/ report some of their work.
Three different kinds of sampling train have
gained acceptance for ambient sampling, namely:
1) The filter and wet impinger type, also used
in source sampling
2) A dual tape filter sampler in which an acid-
treated tape captures particulates and an
alkali-treated tape captures HF
3) A sodium bicarbonate coated tube for HF
adsorption followed by a filter.
The Halogen Subcommittee of the Intersociety
Committee, 6.3/ and a Task Group on Fluorides organ-
ized by the ASTM has studied various methods of sam-
pling and analyzing ambient air for fluorides and has
prepared descriptions of technique following several
alternative methods. Reports are expected to cover
the above three sampling methods and reviews of lab-
oratory and semi-automatic methods of fluoride analysis
of collected samples.
6-11
-------
0567
6.7 EPA Sampling and Analytical Techniques
Source tests were conducted in 1972 on selected
plants by a contractor, EPA, or a combination of both.
Analysis for fluorides and particulate material for
the first two plant tests was done by the contractor;
the rest of the analytical work was conducted in EPA
laboratories.
Where possible, sampling and analytical proce-
dures were used that conformed to EPA Methods 5, 6,
and 7 (determination of particulate, sulfur dioxide,
and nitrogen dioxide emissions, respectively, from
stationary sources) as described in the Appendix to
the December 23, 1971, Federal Register (Volume 36,
Number 247). Measurements of oxygen (02)/ carbon
dioxide (C02)» and carbon monoxide (CO) were conduct-
ed with an Orsat analyzer.
Organic particulates were determined from the
impinger solution by extracting, first with ethyl
either and then with chloroform, and drying the ex-
tract to a constant weight.
Emission samples from inlets and outlets of pri-
mary control systems were collected isokinetically for
fluoride analysis with the sampling train described in
EPA Method 5, and the stack was traversed in accord-
ance with EPA Method 1, "Sample and Velocity Traverses
for Stationary Sources." However, some source tests
were conducted using non-EPA methods that were similar
to techniques used by the company.
When possible, the primary and secondary systems
were sampled simultaneously. Secondary air flow meas-
urements were provided by the company and used by EPA.
When emission samples upstream and downstream of a con-
trol device could not be taken simultaneously, sequen-
tial sampling was accomplished as quickly as possible.
Length of sampling time varied from 2 to 24 hours on
primary systems and from 8 to 24 hours for secondary
emissions.
Special sampling techniques were required to
measure emissions of exit gases where we could nei-
ther traverse nor sample isokinetically. These spec-
ial sampling techniques were used to sample secondary
6-12
-------
0568
emissions from some roof monitors. Some plants used a
control system, others did not. Traverse sampling was
not practical because of the cross-sectional area to be
covered and location of the sampling stations. Iso-
kinetic sampling conditions were not ideal because of
the low air velocity, which lowers the air sampling rate
below the point of efficient collection of the pollutant
(fluoride) in the impinger section of the EPA train.
Therefore it was decided to sample at a constant rate
of 1 cubic foot per minute at a single point, close to
the center of the air stream, to reduce interferences.
This single-point method is a reasonable sampling tech-
nique since about 70 percent of the particulates releas-
ed at the cell could be less than 3 microns in size. A
very important criterion was to ensure that the sample
rate permitted maximum efficiency of the impinger sec-
tion of the sampling train. EPA felt these sampling
techniques for testing roof monitors were reasonable
under the circumstances, but because they were subject
to error, EPA feels that the results are satisfactory
only as an estimate of emissions.
A second and more elaborate method of sampling
the monitor was provided by one company that has in-
stalled a multipoint sampling system in the roof mon-
itor (Figure 6.2). Continuous air samples were drawn
through the common manifold to a discharge stack.
Standard sampling techniques could be performed at the
stack and the results prorated over the total air flow
out the monitor. Air flow through the roof monitor
was simultaneously measured at many points with ane-
mometers. The anemometers were monitored by a computer
that provided air velocity readings every few minutes
for an accurate measurement of air flow during the
sampling period. This system was attractive because
it permitted increased confidence in the sampling tech-
nique. Results of this type of test on a secondary
system should be more reproducible than those from the
single-point nonisokinetic sample. The company reports
excellent correlation between this sampling method and
elaborate manual techniques for sampling in the mon-
itors.
Samples were analyzed for both water-soluble and
water insoluble fluorides. The water-soluble fluorides
were determined by the SPADNS-Zirconium Lake Method,
after the sample was first distilled with sulfuric acid.
6-13
-------
T0 CO
I
£
TO
^ATMOSPHERE
SAMPLING
STATION
CD
(J1
CD
CD
to
m
n
O
Z
O
>
73
(ft
CO
Z
O
Z
o
m
Z
c
71
1"
MONITOR PICK-UP SYSTEM
-------
n r
UO
70
Water-insoluble fluorides were also determined by
the SPADNS method after the sample had been fused
with NaOH. These are both standard fluoride analyt-
ical techniques used for many years by industrial
plants and enforcement agencies.
The proposed EPA sampling and analytical tech-
nique for fluorides (Method 13) is given in Appendix
6B. This may be subject to some minor changes before
Method 13 becomes part of the Federal Register in its
final form. The proposed Method 13 for fluorides is
included in this report in its present form so that
industry may become familiar with the procedures.
The results of EPA source sampling program are
reported in Section 7.5 of this report.
6-15
-------
0571
References - Section 6
Less, L. N. and Waddington, J., "The Character-
istics of Aluminum Reduction Cell Fume", Light
Metals 1971, Proceedings of. Symposia, AIME
Annual Meeting, New York, N. Y. March 1-4, 1971.
Hu, J. N-H, "An Improved Impactor for Aerosol
Studies - Modified Andersen Sampler" Environ-
mental Series: Technology 5 (3): S 251-213,
March 1971.
Intersociety Committee Manual of Methods for
Ambient Air Sampling and Analysis: Health
Laboratory Science, 6_ (2) April 1969, Published
by American Public Health Association, Inc.
Mandl, R. H., Weinstein, L. H., Weiskopf, G. J.,
and Major, J. L., "The Separation and Collection
of Gaseous and Particulate Fluorides", Proceed-
ings of The Second International Clean Air
Congress, Academic Press, New York (1971).
"Biologic Effects of Atmospheric Pollutants -
Fluorides", pp 51 to 56, National Academy of
Sciences, Washington, D. C. (1971).
6-16
-------
0572
Table of Contents
Section 7
7.0 Reported Industry Effectiveness and Costs
of Air Pollution Abatement
7.1 Industry Emission Control Practice
7.2 United States Emission Inventory (1970)
7.3 Emission Control Costs
7.4 industry Cost Effectiveness
7.5 EPA Source Sampling
7.5.1 Description of Facilities Tested
7.5.2 Discussion of EPA Source Test Results
7.5.3 Summary
-------
7 -3
7.0 Reported Industry Effectiveness and
Costs of Air Pollution Abatement
Quantitative data on aluminum smelter effluents
and emissions have been collected from responses to a
comprehensive questionnaire distributed to all primary
aluminum producers in the United States, and from ref-
erences in the technical literature. The responses cov-
ered the operations of all but one plant in the United
States, and varied in completeness of detail furnished.
Technical literature reference and data on emission
control were concerned largely with operations in seven
.European plants.
The questionnaire information was furnished under
an agreement to preserve its confidentiality with re-
spect to individual plants and to report only composite
averages representative of industry practice.
7.1 Industry Emission Control Practice
Almost no data have been reported for the quanti-
ties of effluents and emissions from aluminum smelters
for process areas other than the potlines.
Materials handling operations in the potrooms and
various process operations in carbon plants evolve dusts
of alumina, cryolite, carbon and other materials in
minor quantities. These effluents are normally collect-
ed in baghouses, cyclones or dry electrostatic precipi-
tators as a means for conserving process materials and
maintaining clean working conditions within the plants.
Bake Plant Emissions
Anode baking furnaces evolve objectionable quanti-
ties of smoke, some S02/ and small amounts of fluorides.
These effluents are sometimes treated in scrubbers or
other control equipment, mainly to remove visible compo-
nents of the smoke. Only limited quantitative informa-
tion has been found to be available.
7-1
-------
057k
From the segment of the industry operating anode
baking furnaces, information on bake plant effluents
was received which represented 62 percent of the total
aluminum production by prebake anode reduction plants.
The weighted averages of the reported data indicated
that the effluent flow rate from a typical ring furnace
is of the order of 214 standard cubic feet (at 70°F)
per pound of prebake aluminum plant capacity, with gas
loadings of 0.015 grain total gaseous fluoride and 0.085
grain total particulate per standard cubic foot. Flu-
oride in bake plant effluents is reported to be negli-
gible amounting to only about one percent of the total
,smelter plant fluoride effluent. Emission data on an-
ode bake plants was not reported by the industry, al-
though some 43 percent of the bake plant capacity has
some sort of emission control, much of it experimental.
Based on observations and impressions gained
through industry and suppliers contacts, an assumption
has been made that a removal efficiency of 96 percent
on gaseous fluoride and 75 percent on particulates can
be achieved on bake plant effluents. It is estimated
that some 40 percent of total bake plant effluents are
currently treated with this order of control.
While tunnel kilns are reported to produce lower
loadings of effluents than ring furnaces, the propor-
tion of total prebaked anodes now baked in tunnel kilns
is small enough (perhaps 7 percent) that to disregard
this difference does not affect the limited accuracy of
the calculations made.
The presence of fluoride compounds in bake plant
effluents can be largely controlled by more complete
cleaning of bath material from spent anode butts which
are recycled through the anode baking circuit, since
this material is the largest source of fluorine in the
process.
7-2
-------
0575
Potroom Emissions
Tables 7.1 show a breakdown of potline effluents
and emissions as reported for a composite of United
States smelter experience. All quantities are expressed
as pounds of pollutant per 1000 pounds of aluminum pro-
duced (equivalent to kg/tonne). Values have been com-
puted by summing the total annual quantities of pollu-
tants calculated for each of the companies responding to
the industry questionnaire, and dividing by the annual
aluminum production of those companies.
The fact that the values shown in this table for
"total F" does not equal the sums of their respective
"solid F" and "HF" values results from the fact that not
all questionnaire respondents provided breakdowns of
solid and gaseous fluorine. As much of the reported
data as possible was used.
Tables 7.2 show effluents and emissions as report-
ed in the literature for seven European aluminum smelters;
these data may be compared with corresponding weighted
average data for all the United States plants responding
to the industry questionnaire.
7-3
-------
Table 7.la - U.S. INDUSTRY REPORTED EFFLUENT CONTROL
Total Solids
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Solid F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
HF
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Total F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
LEBAKE POTLINES
Percent
Tons
Reporting
60
83
65
76
65
65
69
76
65
58
65
47
65
76
65
52
65
47
65
80
65
80
65
65
Lbs/1000
Ibs Aluminum „ ..
High Wtd Avg
88.6
84.5
7.8
12.5
7.8
16.3
14.8
14.0
1.6
3.7
1.6
4.4
17.4
17.0
1.2
9.8
1.2
3.2
33.0
31.0
1.7
12.8
1.7
5.1
47.2
43.8
4.0
4.6
4.0
8.1
81%
10.2
9.2
0.7
1.4
0.7
2.5
75%
12.4
10.6
0.5
0.9
0.5
1.5
88%
23.2
19.8
1.3
1.8
1.3
4.3
81%
Low
22.5
21.6
2.0
1.1
2.0
3.5
4.7
4.0
0.6
0.7
0.6
1.8
8.1
6.0
0.2
0.1
0.2
0.5
12.8
15.4
1.0
1.3
1.0
1.4
7-4
-------
Table 7.1b - U.S. INDUSTRY REPORTED EFFLUENT CONTROL
VSS SODERBERG POTLINES
U 5 [ (
Percent
Tons
Reporting
Lbs/1000 Ibs Aluminum
Total Solids
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Solid F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
HF
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Total F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
89
64
100
64
100
64
64
64
100
64
100
64
64
100
100
100
100
100
100
6.5
30.6
29.0
6.8
2.4
6.8
8.0
Wtd Avg
39.2
22.0
11.2
4.4
6.7
11.7
70%
23.2
19.5
3.9
1.5
3.4
4.9
79%
Low
2.2
5.6
5.5
4.4
1.2
4.4
5.6
5.0
3.1
3.0
0.8
2.9
3.8
24%
2.8
1.0
1.3
0.3
0.8
1.6
17.5
23.5
2.4
1.9
2.4
2.5
15.2
16.4
2.2
0.7
1.6
1.6
89%
10.0
7.5
1.1
0.0
0.3
0.3
15.4
11.6
2.8
0.3
1.0
2.0
7-5
-------
•057-8
Table 7.1c - U.S. INDUSTRY REPORTED EFFLUENT CONTROL
Total Solids
Effluent
Primary Collection
Secondary Collection
.Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Solid F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
HF
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Total F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
1 SODERBERG
Percent
Tons
Reporting
93
93
93
93
93
93
93
93
93
93
93
93
100
100
100
100
100
100
93
93
100
100
100
100
POTLINES
Lbs/1000
Ibs Aluminum
High Wtd Avg
52.0
42.0
10.4
10.1
10.4
20.5
8.1
7.4
1.8
3.2
1.8
4.2
14.4
12.0
3.6
2.8
3.6
4.6
21.6
18.8
5.4
6.0
5.4
7.6
49.2
39.1
10.1
8.9
10.1
19.0
61%
7.8
6.9
1.0
2.3
1.0
3.3
58%
13.3
11.3
2.0
1.0
2.0
3.0
77%
21.2
18.1
3.0
3.3
3.0
6.3
70%
Low
41.8
31.4
10.0
8.5
10.0
18.5
7.2
5.4
0.7
1.1
0.7
2.9
12.6
10.8
0.6
0.8
0.6
2.3
21.0
16.2
1.6
2.1
3.0
5.8
7-6
-------
Table 7.Id - U.S. INDUSTRY REPORTED EFFLUENT CONTROL
ALL TYPES OF POTLINES
Total Solids
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Solid F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
HF
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Total F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Percent
Tons
Reporting
63
82
71
82
71
71
Lbs/1000 Ibs Aluminum
72
64
74
75
74
65
74
86
75
75
74
64
77
86
79
88
79
79
Wtd Avg
Low
88.6
84.5
7.8
24.4
7.8
23.5
47.7
40.3
6.9
5.9
6.4
12.3
22.2
16.4
2.0
1.1
2.0
3.5
33.0
31.0
6.8
12.8
6.8
8.0
73%
14.8
14.0
4.4
3.7
4.4
5.6
8.8
7.5
1.1
1.6
1.1
3.0
667=
2.8
1.0
0.6
0.3
0.6
1.6
17.5
23.5
2.6
9.8
2.6
3.2
13.1
11.7
1.2
0.9
1.2
2.1
84%
8.1
4.1
0.2
0.1
0.2
0.3
22.5
19.3
2.3
2.3
2.3
5.1
77%
12.8
4.2
1.0
0.3
1.0
1.4
7-7
-------
0580
Table 7.1
Notes;
1. "Percent Tons Reporting" signifies the fraction of total U.S. annual
production in each potline type which contributed data used to compute
the weighted average effluent and emission factors. For example,
plants producing 60 percent of the total U.S. prebake potline aluminum
reported data which were used to compute a weighted average of 47.2
pounds of total solids effluent per 1000 pounds of aluminum produced.
2. High and low values for individual plant factors and "Percent Tons
' Reporting" are given only in cases where the weighted average values
are computed from three or more data.
3. Weighted average factors represent the total annual reported quantity
of a given effluent or emission divided by the total annual production
of aluminum corresponding to that quantity.
4. Values for "Overall Control Efficiency" for each component are calcu-
lated from weighted average factors according to the formula,
(Effluent Factor - Total Emission Factor) x
Effluent Factor
5. Collection and emission factors for those plants which use no primary
collection system per se but treat all effluents passing through roof
monitors, are included in the calculations for primary collection and
emission factors.
6. U.S. aluminum companies reporting effluent and emission data are sum-
marized in the following table:
Estimated Reporting
Plant U.S. Production Data Percent
Type 109 Ib/yr 109 Ib/yr Reporting
PB 4.433 3.906 88
VSS 1.012 1.012 100
HSS 2.042 2.042 100
All 7.487 6.960 93
7. Data for Table 7.1 were derived from responses to the Industry
Questionnaire.
7-8
-------
Table 7.2a - REPORTED EUROPEAN EFFLUENT CONTROL PRACTICE
0581
Prebaked Potlines
(Lbs/1000 Ibs Aluminum)
Total Solids
Plant*
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Solid F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
HF
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Total F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
A
NR
8.0
8.0
3.8
3.8
52%
17.6
17.6
4.3
4.3
761
B
NR
8.8
8.8
4.5
4.5
49%
9.6
9.6
0.51
0.51
95%
11.0
11.0
0.56
0.56
95%
19.8
19.8
5.1
5.1
74%
* See notes, Table 7.2
7-9
-------
0582
Table 7.2b - REPORTED EUROPEAN EFFLUENT CONTROL PRACTICE
(Lbs/1000 Ibs Aluminum)
Soderberg Potlines
Plant*
Total Solids
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Solid F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
HF
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
Total F
Effluent
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Overall Control Efficiency
c
17.8
12.2
5.6
0.1
2.8
2.9
80%
4.2
3.0
1.2
0.02
0.60
0.62
86%
11.9
7.8
4.1
0.01
0.98
0.99
92%
16.0
10.8
5.2
0.04
1.60
1.54
90%
VSS
D
36.8
21.0
15.8
1.5
5.2
6.7
82%
NR
^
8.8
NR
V
16.3
11.6
4.7
0.17
0.96
1.13
93%
NR
I
^ r
E . F
63.2 NR
40.0 NR
23.2 NR
1.3 NR
23.2 NR
24.5 NR
61% 72%
NR NR
t
^ v
22.3 NR
20.0
2.3
0.46
2.32
2.78
87% v
NR 17.0
12.8
4.2
NR
NR
1.0
94%
HSS
G
27.7
12.7
15.0
3.8
15.0
18.8
32%
NR
v
7.1
5.1
2.0
0.95
2.0
2.95
58%
NR
1 1
*See notes Table 7.2
7-10
-------
0583
Notes - Table 7.2
Plant Lit.
Designation Company Reference
A Alusuisse 7.. I/
B Alusuisse (Rheinfelden) 7.2/
C Montecatini 7.3/
D Pechiney (Nougeres) 7.4/. 7.5/
E German Practice 7.6/
F Svenska Aluminum . 7.7/, 7. 8/
G German Practice 7.6/
NR Not Reported
7-11
-------
058**
7.2 United States Emissions Inventory (1970)
The emissions from the domestic primary aluminum
industry in 1970 are estimated from reported industry
data to have been as shown below and in Table 7.3.
Tons Emissions (1970)
Potrooms Bake Plants Total
Total Fluorine 23,200 650 23,800
.Gaseous Fluorides 10,200 600 10,800
Fluorine in Particulates 13,000 30 13,000
Total Solids 53,000 4,200 57,200
These estimates are based on a 1970 production of
4 million short tons of aluminum, of which about 2.5 mil-
lion tons was produced in prebake anode plants. Reported
potroom emission data were available on 84 percent of the
industry tonnage. No data were reported on bake plant
emissions, and the estimates given above were derived
from reported data on furnace gases, the control equip-
ment identified in individual bake plants, and estimated
control efficiencies ascribed to the control systems.
A more detailed analysis of this inventory is shown
in Table 7.3 which puts into focus the relative impor-
tance of potroom and bake plant control.
7.3 Emission Control Costs
The industry questionnaire requested detailed cost
information for individual items of collection and re-
moval equipment, including capital costs and breakdowns
of the various elements of operating costs. From these
data costs of total plant and industry emission control
could be built up.
The responses to this cost request were as complete
as the aluminum companies were able to provide from their
records, and included nearly 88 percent of the pollution
controlled aluminum production (85 percent of the total
production). Dates of installation were given so that
capital investments could be adjusted to terms of 1970
dollars.
7-12
-------
Table 7.3 - ESTIMATED TOTAL INDUSTRY EFFLUENTS AND EMISSIONS (1970)
Tons
Potlines
Bake Plant
Tons per
1000 tons Al
Produced** Controlled Uncontrolled Total
Controlled Uncontrolled
Total
Industry
Total
-j
i
h-1
U)
Total Aluminum
Production
Effluents
Total Fluoride
F as Gas
F as Solids
Total Solids
Emissions
Total Fluorine
F as Gas
F as Solids
Total Solids
Overall Control
Efficiency
Total Fluorine
F as Gas
F as Solids
Total Solids
23.0
13.9
9.1
45.6
3,854,000 123,000 3,976,000* 1,090,000 1,435,000 2,525,000
88,600
53,600
35,100
175,700
77%
84
66
73
2,800
1,700
1,100
5,600
91,400
55,300
36,200
181,300
5.3
2.2
3.1
12.3
20,400
8,500
11,900
47,400
2,800
1,700
1,100
5,600
23,200
10,200
13,000
53,000
74.6%
81.5
64.0
70
470
440
25
2,650
615
590
30
3,500
1,085
1,030
55
6,150
92,500
56,300
36,200
187,500
25
20
5
650
615
590
25
3,500
640
610
30
4,160
23,800
10,800
13,000
57,200
92.5%
95
80
75
41%
41
45
32
73.2%
80.8
64.0
69.6
* USBM est. 1970 production
**Unit effluent and emission data derived from reported controlled
potline effluent data with adjustments to reconcile balances.
f,n
oo
en
-------
The details of the responses were not uniform, and
interpretations were necessary to bring them to a common
base and reconcile obvious discrepancies and omissions.
Escalation of reported capital costs to 1970 equiv-
alent investment, and extrapolation to include the 12
percent of controlled tonnage not reported, resulted in
the summary figures shown in Table 7.4, and in the esti-
mate that the direct capital investment in potroom pol-
lution control for the 1970 annual production capacity
of about 4.1 million tons is about $182,000,000.
Cost data reported for investment in bake furnace
emission control were incomplete and fragmentary, and did
not justify derivation of an estimate of industry invest-
ment. It is certainly of a different order of magnitude
and perhaps might be composed of $3.5 million for stack
and collection systems required on all plants plus $1.5-
$2 million for the control equipment now installed in
the industry.
The annual direct operating cost ascribable to pot-
line pollution control is estimated from reported data
to be of the order of $11 million per year, with the re-
ported range between $1.20-$6.63 per ton of aluminum
illustrating the different degrees of control employed
among the reporting plants. It is noted that the report-
ed direct operating cost includes power, labor and main-
tenance materials only, and excludes capital charges.
The foregoing report averages for capital invest-
ment and operating costs represent direct costs. Sec-
tions 8.3 and 9.2 deal with estimated costs for model
control systems using costs which include indirect instal-
lation costs at 30 percent of direct capital (purchase
and direct installation). Model operating costs include
capital sensitive factors amounting to 23 percent of
total capital. Adjusted to include indirect charges as
used in the model systems analysis, the total industry
reported investment in potroom pollution control facili-
ties would be $236,000,000 or about $57.60 per capacity
ton. Similarly, if the reported $11 million annual
direct operating cost is adjusted to include interest,
taxes, insurance, and depreciation, it becomes about
$65,000,000 or about $15.84 per ton of aluminum.
7-14
-------
0587
Table 7.4
INDUSTRY-WIDE COSTS FOR AIR POLLUTION CONTROL
IN ALUMINUM POTLINE OPERATIONS
Reported Costs in $/Ton Aluminum
Component High Weighted Avg Low
Reported Capital Costs \_l
Collection Equipment 54.90 17.00 7.70
Removal Equipment 62.90 27.80 3.90
Total Control Equipment 75.20 44.30 18.10
Adjustment for Indirect Costs 13.30
Estimated Total Capital Cost 57.60
Reported Operating Costs _2/
Electric Power 2.00 .89 .19
Materials 1.55 .66 .21
Labor 3/ 4.66 1.19 .24
Total Operating Cost 6.63 2.61 1.20
Annualized Capital at 23% 13.23
Estimated Total Operating Cost 15.84
JL/ Expressed as (1970) dollars per annual ton of aluminum production
capacity.
2_l Expressed as dollars per ton of aluminum produced.
3_/ Includes both operating and maintenance labor.
7-15
-------
0588
7.4 Industry Cost Effectiveness
From the reported data the United States primary
aluminum industry, as a whole, is accomplishing an over-
all potline emission control of about 75 percent of
total fluorine, 81 percent of gaseous fluorides, 64 per-
cent of fluorine in particulates, and 70 percent of
total solids, with an adjusted total capital investment
of some $236 million (in 1970 dollars) and an adjusted
total operating cost of about $65 million per year. The
industry costs represent investment of some $58 per
annual ton of capacity and total net pollution control
pperating costs of about $16.00 per ton of production
at this level of pollution abatement.
The control now exercised results in an emission
of an average of some 12 pounds total fluorine, 5.4
pounds gaseous fluorides, 6.5 pounds fluorine in par-
ticulates, and 28.6 pounds of total solids per ton of
aluminum produced.
7-16
-------
0589
7.5 EPA Source Sampling
After the analysis and systems study of the in-
dustry-reported data had been completed, EPA carried
out a program of source sampling at plants selected to
represent applications of best control technology, to
verify the data supplied by industry and to develop
further information.
The program included source testing of potline
installations at two VSS plants, three prebake plants,
and one HSS plant, and source testing at the anode bake
plants associated with two of the prebake plants.
The digest of the results of these tests as pre-
sented by EPA is given in this section. Data are re-
ported in terms of pounds per ton of aluminum produced,
in distinction to the units of pounds per thousand
pounds aluminum produced (numerically equal to kilo-
grams 'per tonne) used elsewhere in this report.
7.5.1 Description of Facilities Tested
Plant A - uses vertical-stud Soderberg cells with
both primary and secondary emission control systems.
When this plant was first tested, the primary control
system consisted of a "bubbler scrubber" followed in
series by a redwood-tower scrubber. The secondary con-
trol system consisted of a spray-screen scrubber. AI
and A2 are the same plant as A, but these tests were
done after the company had modified the primary control
system to a "bubbler scrubber" followed by a wet elec-
trostatic precipitator. The secondary system remained
a spray screen. The duration of sampling and the dates
are the only differences between tests AI and A2-
Plant B - uses prebake cells with a fluid-bed
dry scrubber as the primary emission control system
and has no secondary control. Test EI is a repeat
source test of this plant to get more complete data.
Other features in this plant that reduce emissions are
the use of computerized pot lines for crust breaking,
alumina additions, and minimum anode effects. The
company had good pot-hooding control and a maintenance
program to keep the hoods in good condition. This
plant had a secondary monitor sampling and air veloc-
ity system as described in Section 6.7 on secondary
emission sampling.
7-17
-------
0530
Plant C - uses prebake cells with both primary
and secondary emission control systems. The primary
system consisted of a cyclonic wet scrubber and the
secondary control system was a spray screen scrubber.
Plant D - is very similar to Plant B. It uses
prebake cells with a fluid bed dry scrubber as the pri-
mary emission control system and has no secondary con-
trol. Test DI is the average of the results of two
non-EPA test methods conducted at this plant. Other
features in this plant that reduce emissions are the
use of computerized pot lines for crust breaking, alu-
mina additions, and minimum anode effects. The company
had good pot-hooding control and a maintenance program
to keep the hoods in good condition. This plant had a
secondary monitor sampling and air velocity system as
described in Section 6.7 on secondary emission sampling.
Plant E - uses horizontal-stud Soderberg cells
with a wet scrubber followed by a wet electrostatic
precipitator as the primary emission control system.
This plant had no secondary controls.
Plant F - uses vertical-stud Soderberg cells with
a wet scrubber followed by a wet electrostatic precipi-
tator as the primary emission control system. This
plant had no secondary controls.
Plant G - is an anode bake plant using a wet
preconditioner ahead of an electrostatic precipitator
as the emission control system.
Plant H - is also an anode bake plant with a wet
preconditioner ahead of the electrostatic precipitator
as the emission control system. However, during EPA
source tests the wet preconditioner was not operated.
7-18
-------
0531
7.5.2 Discussion of EPA Source Test Results
EPA conducted 20 primary inlet, nine secondary
inlet, 12 controlled secondary outlet, and 10 uncon-
trolled secondary outlet emissions tests for total
fluoride.
In addition, 24 primary and three secondary in-
let and 27 primary and three secondary controlled out-
let particulate tests were conducted. No uncontrolled
secondary outlet particulate tests were taken.
Table 7.5 presents the results of particulate
and fluoride tests conducted by EPA on potline efflu-
ents and two anode bake plants. Potline effluent
emission units are expressed as pounds total fluoride
or pounds particulate material per ton of aluminum
produced (Ib TF/TAP and Ib P/TAP, respectively).
Anode bake plant effluents are expressed as pounds
total fluoride per ton of anode produced (Ib TF/TAnP).
Source tests were conducted on the inlet to the con-
trol systems and are shown on the table as primary or
secondary collection. Outlet source tests are shown
on the table as primary or secondary emission. The
individual data points shown in the table are an aver-
age of the tests taken where more than one test was
conducted. From the inlet and outlet data, efficien-
cies of the various emission control systems were
calculated. Sampling and analytical techniques em-
ployed are described in Section 6.7. The source test
data sheets in Appendix 7A give some other details on
each test not shown in Table 7.5.
In Section 6.7, the potroom facilities and type
of control systems for each were described. Therefore,
the following discussion will just present source test
results.
7-19
-------
Table 7.5 - RESULTS OF EPA SOURCE TESTS FLUORIDE AND PARTICULATE
PRIMARY ALUMINUM INDUSTRY
Ibs/Ton Aluminum*
Plant
Cell Type
Control (Primary)
(Secondary)
PARTICULATES
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Primary Efficiency (7.)
Secondary Efficiency (7=)
Overall Control Efficiency (7.)
TOTAL FLUORIDE (Participates and Gaseous)
Primary Collection
Secondary Collection
Primary Emission
Secondary Emission
Total Emission
Primary Efficiency (7.)
Secondary Efficiency (7.)
Overall Control Efficiency (%)
ANODE BAKE PLANTS^
Particulate Emissions
Total Fluoride Emissions
Plant
Control
A
VSS
BS-ST
y) SS
NS
-
-
:
53.50
3.65
1.65
1.651
3.30
96.91
54.79
94.22
G
ESP
1.56
0.88
Al-
VSS
BS-WESP
SS
91.26 (67. 60)2
27.12 (26.68)
0.12 (0.11)
9.53 (5.83)
9.65 (5.94)
99.86 (99.84)
64.85 (78.15)
91.85 (93.70)
37.69
3.02
0.01
0.81
0.82
99.97
73.17
97.98
H
ESP
3.96
1.25
A2
VSS
BS-WESP
SS
NS
-
-
-
-
NS
NS
0.02
2.05
2.07
99.94
32.12
94.92
B
PB
FBDS
None
110.10
NS
13.79
NS
87.46
NC
NS
Bl
PB
FBDS
None
100.30
NS
1.80
NS
98.19
NC
37.80 (48.05)
1.20
0.14
1.20
1.34
99.62 (99.54)
NC
96.89 (97.46)
C
PB
ST
SS
NS
-
"
-
-
28.10
9.06
69. 643
7.30
76.94
19.42
D
PB
FBDS
None
72.26
NS
4.00
NS
94.43
NC
61.40
1.17
0.87 (0.49)*
1.17
2.04 (1.66)
98.64 (99.23)
NC
96.89 (97.46)
Dl
PB
FBDS
NS
-
-
-
-
NS
NS
0.35
1.17
1.52
99.31
NC
97.54
E
HSS
ST-WESP
81.80
NS
5.95
NS
NS
92.73
NC
46.53
2.06
0.41
2.06
2.47
99.11
NC
94.95
F
VSS
ST-WESP
None
38.20
NS
1.34
NS
NS
96.58
NC
NS
* Refer to legend following for notes.
-------
CSS 3
LEGEND FOR TABLE 7.5
VSS - Vertical stud Soderberg cell
PB - Prebake cell
HSS - Horizontal stud Soderberg cell
BS - "Bubbler" scrubber
ST - Spray tower
WESP - Wet electrostatic precipitator
SS - Spray screen
FBDS - Fluid bed dry scrubber
NS - No sampling
NC - No control
(1) - Average of two tests; one test suspected to be
contaminated and deleted.
(2) - Average of two tests; one test deleted due to
stud blow during test.
(3) - Samples suspected to be contaminated during
sampling; these plant data are suspect.
(4) - Data with two tests deleted due to suspected
control system malfunction.
(5) - Anode plant emission units are Ib/ton anode
produced.
7-21
-------
01—' ,*-% 1
,5^
Plant A data show the average of three each pri-
mary and secondary inlet tests and three each primary
and secondary outlet tests for fluorides. Primary col-
lection data ranged from 42.20 to 62.30 Ib TF/TAP with
an average of 53.50. Primary emission data ranged from
1.28 to 1.97 Ib TF/TAP with an average of 1.65. Second-
ary collection data ranged from 1.38 to 5.74 Ib TF/TAP
with an average of 3.65. Secondary emission data ranged
from 1.07 to 37.20 Ib TF/TAP. The 37.20 Ib TF/TAP fig-
ure is not consistent with company or EPA data; it was
thus deleted, and the average for the two tests is 1.65
Ib TF/TAP. These high points were probably due to pick-
ing up a highly contaminated particulate or water drop-
let during the sampling period. The primary emission
control system, from which these tests were conducted,
was shut down and a new primary system, installed. A^
and A2 represent data from the "new" plant or system.
AI data show three primary inlet tests for partic-
ulate ranged from 44.90 to 138.63 Ib P/TAP with an av-
erage of 91.26. Three primary outlet tests ranged from
0.10 to 0.13 Ib P/TAP with an average of 0.12. Three
secondary inlet tests ranged from 22.38 to 30.98 Ib
P/TAP with an average of 27.12. Three secondary out-
let tests ranged from 7.44 to 16.91 Ib P/TAP with an
average of 9.53. The high figures of 138.63 and 16.91
Ib P/TAP came during a source test in which a stud blow
occurred. Though the company states that stud blows
are a rare occurrence, these figures were used in cal-
culating the averages shown in Table 7.5.
The data shown in parenthesis in Table 7.5 for
Plant A, are with the stud blow test deleted.
Test AI data for fluorides show three primary in-
let tests ranged from 32.57 to 42.80 Ib TF/TAP with an
average of 37.69. Three primary outlet tests ranged
from 0.010 to'0.016 Ib TF/TAP with an average of 0.01.
Three secondary inlet tests ranged from 2.67 to 3.41 Ib
TF/TAP with an average of 3.02. Three secondary outlet
tests ranged from 0.64 to 1.07 Ib TF/TAP with an aver-
age of 0.81.
Test A2 data for fluorides show three primary out-
let tests ranged from 0.006 to 0.027 Ib TF/TAP with an
average of 0.016. Three secondary outlet tests ranged
7-22
-------
f At r
from 1.42 to 2.93 Ib TF/TAP with an average of 2.05.
No inlet fluoride tests were made during test A2«
Test A2 was conducted for a 24-hour sampling period
wherever possible. Test number 3 had to be termi-
nated after 16 hours due to a malfunction discovered
in the control device.
A2 efficiencies were calculated using the inlet
data from test A]_. Although AI and A'2 primary data
correlate closely, the secondary outlet data show a
wide range between the two tests and efficiencies. The
high primary efficiency on this plant compares closely
with a similar system on another plant (E).
Some experimental tests were conducted at
Plant B using different filters in the sampling train.
Three primary inlet tests for particulates using glass
filters ranged from 105.3 to 115.3 Ib P/TAP with an
average of 111.0. Three tests using paper filters
ranged from 86.6 to 131.7 Ib P/TAP with an average of
109.4. The overall average of 110.2 Ib P/TAP shown
on Table 7.5 is for the six tests.
•
Three primary outlet tests each were conducted
for particulates using a glass, paper, and membrane
filter. The three tests for the glass filter ranged
from 10.60 to 17.30 Ib P/TAP with an average of 15.81.
Three tests using the paper filter ranged from 10.86
to 21.20 Ib P/TAP with an average of 16.20. Three tests
using the membrane filter ranged from 6.72 to 12.50 Ib
P/TAP with an average of 9.91. The average for the
nine tests was 13.79 Ib P/TAP. No organic extractions
were run on these samples. Fluoride tests conducted
at this plant were not reported due to questionable
analytical procedures.
Tests BI show three primary inlet tests for par-
ticulates ranged from 97.4 to 101.0 Ib P/TAP with an
average of 100.3. Inlet particulate tests for test B
and BI compare favorably. Three primary outlet tests
ranged from 1.32 to 2.54 Ib P/TAP with an average of
1.80. A comparison of the primary outlet tests between
B and BI show a wide variance. However, no explanation
can be determined for the higher results of B. No sec-
ondary particulate tests were taken at this plant.
7-23
-------
0536
p'
Test BI data for fluorides show three primary
inlet tests ranged from 17.4 to 51.7 Ib TF/TAP with an
average of 37.80. The figure 17.4 Ib TF/TAP is sus-
pect as it does not compare to EPA or company data.
However, the number was used in calculating the aver-
age and efficiency but the data in parenthesis in
Table 7.5 show the average and efficiency calculated
with the 17.4 figure deleted. Four primary outlet
efficiency calculated with the 17.4 figure deleted.
Four primary outlet tests ranged from 0.06 to 0.27 Ib
TF/TAP with an average of 0.14. Four tests of second-
ary or roof emissions ranged from 1.10 to 1.37 Ib
TF/TAP with an average of 1.20. The secondary emis-
sions are not controlled but are the emissions that
escape the pot hooding in the potroom building.
Plant C data show three primary inlet tests for
fluoride ranged from 26.90 to 29.5 Ib TF/TAP with an
average of 28.10. Three primary outlet tests ranged
from 8.42 to 157.30 Ib TF/TAP with an average of 69.64.
Considerable dirty water droplets were being discharged
from the outlet stacks of this plant's primary control
system, it was reasonable to expect that some of these
droplets were picked up in the sample train during the
course of sampling and most, if not all, data from this
plant is suspect. Three secondary inlet tests ranged
from 8.55 to 9.53 Ib TF/TAP with an average of 9.06.
Three secondary outlet tests ranged from 7.02 to 7.45
Ib TF/TAP with an average of 30. No particulate tests
were taken at this plant.
Plant D data for three primary inlet tests for
particulates ranged from 63.83 to 83.23 Ib P/TAP with
an average of 72.26. Three primary outlet tests using
a glass filter ranged from 2.16 to 2.30 Ib P/TAP with
an average of 2.22. No secondary particulate tests
were taken at this plant.
Plant D data for three primary inlet tests for
fluorides ranged from 56.20 to 64.80 Ib TF/TAP with an
average of 61.40. Six primary outlet tests ranged
from 0.32 to 1.66 Ib TF/TAP with an average of 0.87.
However, two tests conducted in one day were found to
have much higher results (1.60 and 1.66) compared to
the other four tests (0.32 to 0.61X It was determined
that the control device was malfunctioning during the
two runs and the data are suspect. Therefore, the
7-24
-------
O
averages were calculated from the primary emission
control system using all test data; for comparison,
the data deleting the two high tests are shown in
parenthesis. Two secondary or roof emission tests
ranged from 0.86 to 1.48 Ib TF/TAP with an average
of 1.17. The roof emissions were not controlled but
were emissions escaping the pot hooding in the pot-
room building.
Three experimental primary outlet tests for
fluorides (designated DI) ranged from 0.38 to 0.50
Ib TF/TAP with an average of 0.44. These tests were
conducted using the EPA impinger train and heated
filter but using a coated stainless steel probe and
flexible tubing between the stack and impinger train.
The difference between tests D and DI was only the
sampling train modification. Three other experimen-
tal primary outlet tests for fluoride ranged from
0.20 to 0.36 Ib TF/TAP with an average of 0.27. These
tests were conducted using the same sampling train
as above except that sampling was conducted at a
single point at the point of average gas velocity
through the stack. The average for the six tests is
shown in Table 7.5.
Plant E data for three primary inlet tests for
fluorides ranged from 35.00 to 54.20 Ib TF/TAP with
an average of 46.53. Three primary outlet tests
ranged from 0.38 to 0.42 Ib TF/TAP with an average
of 0.41. Four secondary or roof emission tests
ranged from 1.10 to 2.85 Ib TF/TAP with an average
of 2.06. The wide variation in the roof emissions
was due the duration of the sampling period, potroom
activity during the sampling period, and sampling at
a single point in the monitor. For example, test
number one was conducted over a 16 hour period and
the potroom activity was at a minimum. Test number
four was conducted for eight hours and potroom activ-
ity under the single sampling point was high.
Plant E data for three primary inlet tests for
particulates ranged from 74.5 to 90.8 Ib P/TAP with
an average of 81.80. Three primary outlet tests
ranged from 3.21 to 8.91 Ib P/TAP with an average of
5.95. No secondary particulate tests were taken.
7-25
-------
O1^
w w
r r:
Plant F data for three primary inlet tests for
particulates ranged from 32.35 to 39.35 Ib P/TAP with
an average of 38.20. Three primary outlet tests
ranged from 1.08 to 1.35 Ib P/TAP with an average of
1.34. The fluoride data from this plant were taken
by a contractor who analyzed the samples for "gaseous
fluorides" and used midget impingers for the sampling
train. Therefore, the fluoride data were suspect be-
cause they did not agree with any company or EPA data.
Plant G and H are anode bake plant furnace
emission data. Plant G data for three outlet tests
for particulates ranged from 1.40 to 1.74 Ib P/TAnP
with an average of 1.56. Three outlet tests for flu-
orides ranged from 0.70 to 0.98 Ib TF/TAnP with an
average of 0.88. No inlet emission data were taken.
Plant G data for three outlet tests for flu-
orides ranged from 1.20 to 1.28 Ib TF/TAnP with an
average of 1.25. One outlet test for particulates
was 3.96 Ib P/TAnP. No inlet emission data were
taken and time (due to poor weather conditions) pre-
vented any more outlet particulate tests to be run.
Sulfur oxide data ranged from 5 PPM for a pre-
bake plant to 80 PPM for a vertical stud Soderberg
plant. No sulfur oxide data were obtained on a
horizontal stud Soderberg plant.
Nitrogen oxide emissions averaged less than
5 PPM for all plants.
Efficiencies were calculated for the various
plants based on the inlet and outlet loadings on the
control system. Only one set of inlet data was taken
in most cases. If those data were within the range
of values determined by the company, the original in-
let data were used to determine efficiencies based on
outlet data not taken on the same tests. For example,
the fluoride inlet data determined for test A^ were
used to calculate fluoride efficiencies for test A2-
This is also true for tests D and D]_.
7-26
-------
Q599
7.5.3 Summary
Primary efficiencies for particulate average
95 percent with a range of 85.76 to 99.86 percent.
Lack of secondary particulate data precludes any over-
all efficiency control for particulates in the primary
aluminum industry.
Primary efficiencies for total fluoride average
99.42 percent with the range 96.91 to 99.98 percent.
However, the 96.91 percent efficient system is no
longer in operation. Secondary emission control
ranged from 0 percent to 78 percent, but more meaning-
ful efficiency data is shown by the overall control of
fluorides. The overall control efficiency for flu-
oride averaged 96.85 percent with the range from 94.95
to 98.56 percent after deleting Plant A data.
7-27
-------
.-> f~ r\ f*l
•UbOO
References - Section 7
Oehler, R.E., Emission of Air Contaminants in
Aluminum Electrolysis; TMS Paper No. A70-11,
Metallurgical Society of AIME, February 1970.
Moser, Dr. E., Treatment of Fume from Primary
Aluminum Plants; International Conference on air
pollution and water conservation in the Copper
and Aluminum Industries, Basle, Switzerland,
October 1969; British Non-ferrous Metals Research
Association.
Callaioli, G., Lecis U., Morea, R., Systems of
Gas Collection and Cleaning in Electrolytic Fur-
naces of Montecatini Edison Aluminum Plants; TMS
Paper No. A70-23. Metallurgical Society of AIME,
February 1970.
Barrond, M.P. et al. (Group Pechiney), "L1alumi-
nium" , Vol. II, published by Editions Eyrolles,
Paris, 1964.
Calvez, C., et al, "Compared Technologies for the
Collection of Gases and Fumes and the Ventilation
of Aluminum Potlines": Proceedings of Extractive
Metallurgy Division Symposium, December 1967,
EMD of AIME.
"Restricting Dust and Gas Emissions in Bauxite
and Aluminum Processing Plants", VDI 2286,
November 1963; Leichmetall-Fachauschmss (VDI Kom-
mission Reinhaltung der Luft).
Lindberg, G., "Air Pollution Control in the Swed-
ish Aluminum Industry", 2nd International Clean
Air Congress of International Union of Air Pollu-
tion Prevention Association, December 1970,
Washington, D. C.
Brenner, E.M., "Gas Collection, Cleaning and Con-
trol at Sako, Sundsvall Works"; a TMS of AIME
Paper Presented at the TMS-AIME Annual Meeting,
February 1970, Denver, Colorado.
7-28
-------
OG01
Table of Contents
Section 8
8.0 Systems Analysis of Pollution Abatement by Models
8.1 Models of Potline Effluent Controls
8.1.1 Model Structure
8.1.2 Model Parameters
8.1.3 Overall Emission Control
8.1.4 Control Schemes Applied to Models
8.2 Costs
8.2.1 Capital Costs - Model Control Schemes
8.2.2 Operating Costs
8.2.3 Control System Credits
and Net Annualized Costs
8.2.4 Summary of Cost Elements
8.3 Cost Effectiveness of Control Models
-------
0602
8.0 Systems Analysis of Pollution Abatement by Models
The total emissions from an aluminum smelter in-
stallation includes controlled or uncontrolled effluents
from three groups of process operations, electrode prep-
aration, potline reduction, and cast house operations.
The character and quantity of emissions from these three
sources differ considerably. Emissions from electrode
preparation are largely carbon dusts, with some hydrocar-
bon volatiles generated in carbon paste materials proc-
essing, as well as combustion gases, smoke and volatiles
resulting from anode baking. Potline emissions include
particulates and gases arising directly from the produc-
tion of the molten metal. Cast house emissions are
combustion gases and fumes, primarily aluminum chloride
which may hydrolyze in the presence of atmospheric mois-
ture to HC1 and A12O3.
Insufficient quantitative and qualitative data are
available from reporting or published sources concerning
carbon plant or cast house emissions to provide the
basis for systems analysis of their generation and con-
trol. Their quantities are orders of magnitude less than
those from potline operations; emission control is exer-
cised on most effluent streams, but not measured.
Some scanty data are available on bake plant emis-
sions, but are not adequate for purposes of a meaningful
systems analysis on an industry basis.
A systems analysis approach applied to models rep-
resenting current aluminum industry reduction practices
provides insight into the potential effectiveness of
pollution control equipment used in various combinations.
For this analysis typical effluent parameters have been
established from industry data as the basis of models
defining types of operation in the industry. The per-
formance and costs of alternative pollution control
equipment combinations have been evaluated as they apply
to the effluent models.
These evaluations indicate the degrees of overall
control efficiency which should be attainable within
limits of capital investment and annual operating costs.
8-1
-------
0603
Descriptions of removal equipment and derivations
of their assumed removal efficiencies for solid and
gaseous fluoride are given in other parts of this report.
In this section applications of control schemes applied
to industry effluent models are described. The compo-
nents of capital and operating costs for each control
scheme are combined and compared with performance param-
eters to give a measure of cost effectiveness for the
control model and to provide the basis for estimating
costs involved in upgrading the overall pollution con-
trol of the primary aluminum industry.
8.1 Models of Potline Effluent Controls
Most of the attention to air pollutant control in
the primary aluminum industry has been directed toward
fluoride effluents from the alumina reduction potlines.
Although some objectionable pollutants derive from anode
baking operations and other process activities at a
smelter, it is the large quantities of gaseous and solid
fluorides emanating from the cryolite bath of the pot-
lines which present the most serious hazard to plant and
animal life and which have received the major portion of
technical effort to control.
The control of emission opacity, as judged by
Ringlemann standards, is complicated by the presence of
submicron hydrocarbon aerosols in effluents from anode
bake plants and Soderberg anode potlines. Plume opacity
problems may be more severe in plants which combine and
discharge emissions from a central system through a sin-
gle stack. The same emissions from a number of separate
points might individually meet the Ringleman standard
adopted. Some of the control schemes considered herein
among the models, while effective for the abatement of
particulate and gaseous fluoride air pollution, might
not satisfy the additional requirement of controlling
submicron hydrocarbon aerosols which contribute most to
plume opacity in central stack emissions.
8-2
-------
8.1.1 Model Structure
To assess the performances of various emission
control schemes applied to similar effluent situations
in the primary aluminum industry, a number of plant ef-
fluent models have been established. These models rep-
resent various combinations of effluents and effluent
collection systems which are applicable to the three
major kinds of reduction cell installations, viz., pre-
bake anodes, HSS Soderberg anodes and VSS Soderberg
anodes. To these streams can be applied various combi-
nations of emission control devices, or control schemes.
Such a control model structure is shown diagrammatically
in Figure 8.1.
Primary stream collection is defined as the di-
rect removal of cell effluent through hoods on individ-
ual pots, ducted to a common emission control system
serving a group of pots.
Secondary stream collection is defined as the
gathering of potroom effluents, including those not col-
lected by primary hooding, together with potroom venti-
lation airflow, by using the potroom structure as a
containment envelope and exhausting through one or more
roof plenums.
Although a part of the particulate material pass-
ing through the roof monitor secondary system originates
from mechanical operations within the potroom and not
from cell operations per se, collection or hooding effi-
ciencies and model effluent quantities are considered on
the basis that all particulates as well as gaseous flu-
orides are assumed to originate at the cell surface.
Estimates of pollutant removal performance and equipment
costs are not influenced by this simplification in the
model.
Collection efficiencies are expressed as the per-
centage of the total effluent collected into a primary
stream, in terms of total fluorides, calculated from
components of solid fluorides and gaseous fluorides.
Cognizance is taken in the control model struc-
ture of the potential of returning collected dry fluorine
bearing products to the reduction cell without reproc-
essing. This potential is assumed to exist only with
8-3
-------
FIGURE 8.1
SCHEMATIC FLOW DIAGRAM
ALUMINUM SMELTER AIR POLLUTION CONTROL
SYSTEM MODEL: „
OVERALL CONTROL EFFICIENCY:.
CD-
CD
O
01
REMOVAL
EFFICIENCY
I.
1.
If
1c= TOTAL F COLLECTION EFFICIENCY
-------
control equipment applied to the primary stream.
The following effluent models have been selected
to represent the various types of primary aluminum re-
duction plants.
Collection Arrangement
Type of Cell Primary Secondary
IA Prebake Anode Controlled Uncontrolled
IB Prebake Anode Controlled Controlled
1C Prebake Anode None Controlled
ID Prebake Anode None Uncontrolled
IIA VSS Soderberg Controlled Uncontrolled
IIB VSS Soderberg Controlled Controlled
IIIA HSS Soderberg Controlled Uncontrolled
IIIB HSS Soderberg Controlled Controlled
Primary collection is always used with Soderberg
cells and thus no model corresponding to 1C is appli-
cable to Soderberg cell plants.
The control schemes applied to the models have
been selected to suit the physical characteristics of
the model effluent streams. Removal efficiencies of
control schemes are defined as the percentage removal
of the component (solid fluoride or gaseous fluoride)
from the feed stream to the control device. Removal
efficiencies are considered to be independent of equip-
ment size or capacity.
Most removal efficiencies for solid F are derived
from test data on actual potline effluents. (Ref. Table
5.2) Where no data exist, efficiencies were estimated
from fractional efficiency curves and the assumption
that solid F particle size distributions are similar to
those for total solids evolving from potline operation.
The validity of this assumption is questioned by some
investigators.
8-5
-------
0607
8.1.2 Model Parameters
Effluents
The quantities and characteristics of effluents
entering the several control systems described by the
above models are determined partly by the type of reduc-
tion cell and partly by the efficiency with which pri-
mary collections systems separate the effluents into
primary and secondary component streams.
All three types of reduction cell, so far as can
be determined, generate the same total quantities of
total fluorides in their effluents amounting to an av-
erage of 23 pounds per 1000 pounds of aluminum produced.
In the prebake and HSS Soderberg cell effluents
there are approximately equal amounts of solid and gas-
eous fluorides in which a large part of the particulate
fluoride is submicron in diameter.
In the VSS Soderberg, after the burner, the col-
lected effluent contains approximately 90 percent of the
total fluorine as HF, and has a low particulate fluoride
content, resulting probably from a greater increased
opportunity for hydrolyzation by contact of the fume
with moist air from combustion of hydrocarbons after
passing through the burner. 8.I/
Primary Collection Efficiency
The efficiency of primary collection at the re-
duction cell depends on the mechanical design of the
hoods, which is governed by the type and configuration
of the cell, the extent to which the hooding devices are
kept intact and in good repair, and the amount of inter-
ference by cell operations with the hooding effective-
ness.
A high collection efficiency for both gases and
solids can be obtained with modern prebake cells in
which the hood design is integral with cell construction,
and in which feeding and crust breaking operations can
be carried out at the center between rows of anodes with
minimum disruption of the hooding protection. Side-
worked prebake cells, on which hooding has been added
8-6
-------
0608
after original design, characteristically show lower
collection efficiency. The terms "new" and "old" are
used in model analyses to designate this difference.
Primary collection efficiency for side-worked
Soderberg cells is lower than for the newer center-
worked prebake cells. In the VSS Soderberg, there is
a substantial area of cell surface outside the skirt
through which cell working is accomplished with conse-
quent evolution of uncollected pollutant. In the HSS
Soderberg, the hooding enclosure must be breached par-
tially to work the cell or entirely to pull the anode
channels for stud relocation.
The collection efficiency with respect to partic-
ulates in an effluent stream appears to be somewhat less
than that of gaseous collection, and the order of mag-
nitude of this differential has been estimated from data
reported in the industry responses to the questionnaires.
Part of this apparent difference in collection effi-
ciency may be attributable to the fact that pot working
operations account for a large part of solids effluent
and it is characteristic of many pot working conditions
that part of the hooding may be removed at the time.
Also, since fluoride dust released to the potroom from
sources other than the cell itself is counted as cell
effluent escaping the hooding, this unmeasured quantity
reduces the apparent collection efficiency.
Controlled Effluent Streams
The quantitative parameters adopted for the efflu-
ents referenced to the models are summarized in Table 8.1.
To be valid, the model parameters must approxi-
mate those of the corresponding plant installations in
the industry. They have, therefore, been chosen after
evaluation of industry reported data, which are not en-
tirely consistent within themselves. However, it has
been possible, by a technique of weighted averages, to
arrive at parameters which can be taken to be illustra-
tive of industry conditions and practice within reason-
able limits.
8-7
-------
Table 8.1
MODEL EFFLUENTS ENTERING CONTROL SYSTEMS
Basis: 1000 Ib Aluminum
I New* PB I Old* PB II VSS
Component Tot .
Solid F, Ib 10
HF, Ib 13
Total F, Ib 23
Alumina. Ib 20
00 .
l
00 Total Solids, Ib 48
Air, 106 ACF 27.5
Collection, Eff.
Solid F
HF
Total F
Prim.
9.5
12.6
22.1
19.0
46.1
2.5
95
97
96.0
Sec. Tot. Prim.
0.5 10 8.0
0.4 13 11.7
0.9 23 19.7
1.0 20 16.0
1.9 48 41.3
25 27.5 2.5
80
90
85.6
Sec. Tot. Prim. Sec.
2.0 3 1.5 1.5
1.3 20 17.0 3.0
3.3 23 18.5 4.5
4.0 3 1.5 1.5
6.7 39 25.9 13.1
25 35.5 0.5 35
50
85
80.4
CD
CD
CD
CD
III HSS
Tot. Prim. Sec.
10 8.0 2.0
13 11.7 1.3
23 19.7 3.3
20 16.0 4.0
49 38.2 10.8
38.5 3.5 35
80
90
85.6
*Refers to average collection efficiencies between those of "new" potlines
and those to which hooding may have been added after original design.
-------
0610
Collection Systems
Primary effluent collection systems comprise the
hooding devices installed at the reduction cells (in-
cluding the skirts and burners on VSS Soderberg cells),
the individual cell ducting to common headers serving
groups of cells, and the main ducting leading to control
devices. Practice varies among aluminum smelters as to
the number of cells connected with a single control sys-
tem, in centralized systems, an entire potline of 150
or more cells may be ducted to a single control system,
whereas the decentralized systems, where smaller control
units are usually located in courtyards between potlines,
may connect 20 or fewer cells to each control system.
Figures 8.2 illustrate schematically several possible
collection system configurations for PB, VSS and HSS pot-
lines. The illustrated courtyard schemes are patterned
after existing installations and were selected as the
bases for control system models.
Costs for a central scheme usually are greater
than for courtyard schemes as will be shown in Section
8.2.1. These cost differences may be offset by differ-
ences in the total costs of very large units of removal
equipment for central systems compared with smaller
units for courtyard systems.
Considerations other than costs, such as flexibil-
ity by provision of duct interconnections for continued
pollution control when part of a system may be out of
service, and the ease of cleaning deposits from the in-
side of ducting, may influence the design of collection
systems. Maximum collection efficiencies are realized
when the system designs provide for continuous exhaust-
ing of all operating cells through removal equipment even
when parts of a potline are being serviced, and when
provisions are made in the collection system to increase
air flow rate from a cell which may have part of its
hooding removed for cell working or anode replacement.
8.1.3 Overall Emission Control
The overall emission control, measured by either
the overall control efficiency (OCE) or by the weight of
the emitted total fluorides per 1000 pounds of aluminum
produced, is a function of three independent variables,
8-9
-------
0611
FIGURE 8.2a
PRIMARY COLLECTION SYSTEMS
Typical ducting layouts for a Single Prebake Potline-160 cells, two rooms.
-Manifold Duct
Manifold Duct-
COURTYARD SCHEME
[20 cells per manifold duct]
[80 cells per manifold ductj
Main Duct
? ?
R
20 CELLS JJ \} IJ LJ/
-71 ! !
^JLJ i
D
/ 9 9
/
i i
i !
ti fi A i
K
6 i
Main Duct-
9 \ ? ? 9
• i
1 I
i i
1 1
A A A 4
80 CELLS
CENTRAL SCHEME
J- R
Symbol
Removal Equipment
8-10
-------
0612
FIGURE 8.2b
PRIMARY COLLECTION SYSTEMS
Typical ducting layout for a Single VSS Potline - 160 cells,two rooms.
Manifold Duct
Main Duct-
9
R
»-
ll
on
[
R
-
in
R
r
cS\e/j
R
•—
\
t ~\.
I ^
i
i
LI pi? 99 g
•* R
ii!] ill 6 ^loi d fi
CELLS
*• R
•B
i
-0
ft od rtfl fi
COURTYARD SCHEME
[lO cells per manifold-main ducting]
Symbol
Removal Equipment
8-11
-------
u
U-,'1 X
- U „)„ w
FIGURE 8.2e
PRIMARY COLLECTION SYSTEMS
Typical ducting layout for a Single HSS Pot line-160 cells, two rooms.
Manifold Duct
Main Duct
re
99 90
R •-
^ R
OD am
10
CELLS
od ofi as a
COURTYARD SCHEME
[lO eel Is per manifold ductj
Manifold Duct-
Main Duct-
9 \ 9 9 9
i i
fl fl 0 d
80 CELLS
CENTRAL SCHEME
•- R
JL
[80 cells per manifold duct]
Symbol:
Removal Equipment
8-12
-------
•OGH
primary collection efficiency (r^c) , removal efficiency
of devices used to treat the primary collected stream
(njp) , and the removal efficiency of secondary stream con-
trol devices (ns) . The relationship among these may be
expressed as,
OCE = rjc
The determinant factors in obtaining high overall
control efficiencies are, in order of impact,
1. Primary removal efficiency
2. Primary collection efficiency
3. Secondary removal efficiency
For example, the best demonstrated combination of
available technology for collection and primary removal
alone in a prebaked potline (96 and 98.6 percent respec-
tively) results in an overall control efficiency of 94.6
percent, corresponding to about 1.2 pounds total flu-
oride emission per 1000 pounds aluminum produced.
Addition of secondary treatment to the primary
control system results in small, but significant, incre-
ments to overall control efficiency. A 63.6 percent
efficient secondary system added to the best available
primary system raises overall control efficiency to 97.1
percent (0.67 pound total fluoride emission per 1000
pounds aluminum) .
It is apparent that to reach a control level with
prebake cells of say 96 percent (about 0.92 pound flu-
oride emission per 1000 pounds of aluminum produced) a
secondary system of at least 33 percent removal effi-
ciency would be needed in combination with the best
available primary collection and removal techniques, and
that if the secondary efficiency were greater, the com-
bination of optimum primary parameters could be somewhat
relaxed.
The above line of reasoning can be applied to the
situations of Soderberg potlines. However, the nature
of the design of both VSS and HSS Soderberg cells make
it virtually impossible to achieve collection efficien-
cies as high as for modern prebake potlines.
8-13
-------
Pel
V> w J« >
For situations where no primary collection is
used, and all emission control is exercised on the
secondary stream, overall emission control is limited
to the secondary removal efficiency. Low pollutant
loadings in the air streams with correspondingly low
removal driving force and inordinately high costs for
high performance removal equipment, as referenced in
Tables 8.4, make high levels of overall control effi-
ciency impractical.
8.1.4 Control Schemes Applied to Models
From among the various control devices and com-
binations applicable to the removal of reduction plant
pollutants, a selection has been made for control
schemes to be applied to the potline effluent models,
including those control schemes which are in current
plant use as well as others which represent the higher
ranges of technically achievable performance with ef-
fluents characteristic of the models.
In practice, the actual values of the effluent
conditions in individual plants will not correspond
exactly to the average parameters assumed for the models
as shown in Table 8.1. The overall emission control
plant performances obtained with similar removal equip-
ment will, therefore, vary to some extent from those
derived from the model control schemes which are repre-
sentative of the industry groupings.
Tables 8.2 list, by models and model control
schemes, the control scheme removal efficiencies and
the overall control efficiencies corresponding to the
collection efficiencies and flow rates postulated for
the various models.
Appendix 8A presents the simplified flow diagrams
and calculated emission data for some of the model con-
trol schemes listed in Tables 8.2.
8-14
-------
06:
Table 8.2a
MODEL CONTROL SCHEMES AND EFFICIENCIES
PREBAKE POTLINES - MODEL I
Model
Number
Equipment*
Designator
IA - Prebake Primary
-1
-2
-3
-4
-5
-6
-10
-11
-12
-13
-14
-15
-16
-17
FBDS
MC+VS
MC+HPSS-3
DESP+ST
MC4CFPB-5
MC+ST
CFDS
MC
ST
MC+DESP+ST
MC+VPB-3
MC+FBWS
IB - Secondary
(with Primary)
-9 J3S
1C - Secondary Only
(no Primary)
-7 CFPB-3
-8 FBWS
-9 SS
Equipment Removal Eff.
Solid F
98
96
93
98
87
80
98
78
80
90
85
80
85
98
HF
99
99
98
95
98
95
90
-
95
95
66
98
88
98
Total F
98.6
97.7
95.9
96.2
93.4
88.8
93.4
33.6
88.7
93.0
74.8
90.8
86.4
98
25
84
75
45
80
99
93
93
92.5
85.2
72.1
Overall Control
Eff. Total F
Collection Eff.
96.0
94.6
93.8
92.1
92.4
89.6
85.2
89.6
32.2
85.
89.
71.8
87.2
83.0
94.1
,2
.2
85.6
84.4
83.6
82.1
82.3
80.0
76.0
80.0
26.7
75.9
79.6
64.0
77.2
74.0
84.0
1.9** 6.7**
All Effluent to Sec.
92.5
85.2
72.1
* Underlined equipment designators signify equipment in current
use on prebake potlines. Legend follows Table 8.2c.
** Overall control efficiency values for Model IB are to be added
to values for Model IA to evaluate performances of combined
control schemes.
NR = Not reportable.
8-15
-------
0617
Table 8.2b
MODEL CONTROL SCHEMES AND EFFICIENCIES
VSS SODERBERG POTLINES - MODEL II
Model
Number
IIA - VSS
-1
-2
-3
-4
-5
-6
-8
-9
Equipment*
Designator
Primary
ST+WESP
FBDS
DESP+ST
MC+VS
MC+HPSS-3
MC+CFPB-5
ST
IADS
IIB - VSS Secondary
(with Primary)
-7 SS
Equipment Removal Eff.
Solid F HF Total F
99
98
98
96
93
87
75
98
42
99
99
99
99
98
98
99
98
99.0
98.9
98.9
98.8
97.6
97.1
93.0
98.0
Overall Control
Eff. Total F
Collection Eff.
^c = 80.4
79.
79.
79.
88 72.7
79.4
78.4
78.0
74.6
78.7
14.2**
* Underlined equipment designators signify equipment in current
use on VSS Soderberg potlines. Legend follows Table 8.2c.
** Overall control efficiency values for Model IIB are to be added
to values for Model IIA to evaluate performances of combined
control schemes.
8-16
-------
Table 8.2c
MODEL CONTROL SCHEMES AND EFFICIENCIES
HSS SODERBERG POTLINES - MODEL III
Model Equipment*
Number Designator
IIIA - HSS Primary
-1 CFPB-5+WESP
-2 ST+WESP
-3 CFPB-5
-4 ST
-5 FBWS
IIIB - HSS Secondary
(with Primary)
-6 SS
Equipment Removal Eff.
Solid F HF Total F
98
98
81
64
78
25
98
93
98
93
98
80
98
95
91.1
82
89.9
45
Overall Control
Eff. Total F
Collection Eff.
Re = 85.6
83.9
81.3
78.0
70.2
76.9
6.5**
* Underlined equipment designators signify equipment in current
use on HSS Soderberg potlines. Legend follows.
** Overall control efficiency values for Model IIIB are to be added
to values for Model IIIA to evaluate performances of combined
schemes.
8-17
-------
0619
Q
JLegend for Tables 8.2a. b and c
CFPB-5 Cross Flow Packed Bed Scrubber (5 ft. Deep)
CFPB-3 Cross Flow Packed Bed Scrubber (3.25 ft. Deep)
CS Chamber Scrubber
DESP Dry Electrostatic Precipitator
FBDS Fluidized Bed Dry Scrubber
FBWS Floating Bed (Bouncing Ball) Wet Scrubber
HPSS-3 High Pressure Spray Screen (3-Stage)
IADS Injected Alumina Dry Scrubber
MC Multiple Cyclone
SS Spray Screen
ST Spray Tower
VPB-3 Vertical Flow Packed Bed Scrubber (3ft. Deep)
VS Venturi Scrubber
WESP Wet Electrostatic Precipitator
8-18
-------
8.2 Costs (Based on 1970 Dollars)
Cost effectiveness analyses for control schemes
applied to model potlines are based on both capital in-
vestment and annual costs. The following paragraphs
outline the elements of these costs assuming new plant
installations. Modifications to existing plants involve
additional costs which may vary widely according to
specific circumstances.
8.2.1 Capital Costs - Model Control Schemes
The capital costs of the model control schemes
are the sums of the capital costs of the elements, in-
cluding collection systems, pollutant removal equipment
stages for primary and secondary effluent treatment,
and costs of scrubber water treatment for removal.
The capital costs of each element are expressed
in dollars per annual capacity ton of aluminum and are
made up (a) equipment purchase costs plus (b) direct
installation costs plus (c) indirect installation costs.
Purchase costs were estimated from cost curves
developed from manufacturers information and, in some
cases, from actual costs reported by respondent plants.
The cost curves are presented in Appendix 8B as func-
tions of air flow-rate capacity.
Primary removal equipment capacities were based
on the use of collection systems corresponding to court-
yard arrangements (40 cell groups for prebake, 20 cell
groups for Soderberg). Equipment unit capacities
assumed were 100,000 cfm dry air for prebake collection
systems, 10,000 cfm dry air for VSS applications and
60,000 cfm moist air for HSS Systems. An arbitrary
10 million cfm of secondary systems air per grouping of
control units was chosen as the basis for costing the
equipment in secondary removal systems.
8-19
-------
.0621
Direct installation costs, which include costs
for transportation of equipment, site preparation, foun-
dations, erection, utility connections and auxiliaries
such as fans, inlet ductwork, motors and control instru-
mentation, were calculated by applying the factors shown
in the following table to the control equipment purchase
cost. 8.27
Installation Cost,
Equipment Type Percent
Dry Centrifugal 70
Wet Collectors
Venturi 200
All Others 100
Electrostatic Precipitators
Dry (plate type) 70
Wet (plate type) 75
Fabric Filters 75
Indirect installation costs were estimated as a
percentage addition to direct installed costs, and were
made up of:
Engineering 7%
Construction Overhead 1%
Start-up Expense 3%
Initial Spares 2%
Sales Taxes 2%
Contingencies
Total 30%
_!/ This figure is representative; sales tax rates
vary from state to state and in addition many
states exempt air pollution control facilities
from sales and use taxation.
8-20
-------
t CS22
Scrubber water treatment costs were added as an
element of total removal scheme costs where the removal
scheme includes the solution of HF in water.
Although some existing smelters discharge scrub-
ber water to moving bodies of water without fluoride re-
moval treatment, increasing awareness of the need to
protect the total environment makes it likely that any
new installation would be required to remove most flu-
oride from scrubber water prior to release from the
plant. Normally this is accomplished by precipitating
dissolved fluoride as calcium fluoride by reaction with
lime, with subsequent separation of the precipitate and
accompanying removed particulates in a clarifier for im-
pounding. The capital cost for this process equipment
is estimated to be $130 per gpm of water flow required
by the scrubber equipment used.
There are special situations where scrubber liquor
may be economically treated with sodium aluminate and
caustic soda to recover dissolved fluoride as cryolite.
A smelter located adjacent to a Bayer process alumina-
from-bauxite plant may have a convenient and low cost
source of sodium aluminate solution, the Bayer process
green liquor, which may be reacted with dissolved flu-
oride in scrubber liquor to produce low grade cryolite
(approximately 45 percent cryolite, 55 percent alumina).
Alternatively, high grade cryolite plants, which use
spent pot lining as primary source of fluoride, recover
additional fluoride from scrubber liquor. Except as
these special circumstances obtain, fluoride recovery
from scrubber liquor appears uneconomical.
Collection systems costs were estimated for both
courtyard and central systems, using the configurations
shown in Figures 8.2, model effluent flow-rates given
in Table 8.1, and the assumption that the cells produce
one pound of aluminum per minute. A summary of these
estimates is presented in Table 8.3.
Most United States plants have employed the court-
yard collection system and this pattern has been followed
in the structuring of the control schemes.
8-21
-------
Table 8.3
ESTIMATED CAPITAL COSTS OF PRIMARY COLLECTION SYSTEMS
($ Per Ton Aluminum at Full Plant Capacity)
(New Construction)
Collection System
Prebake Cells
VSS Soderberg
Air Flow
per Cell acfm
2,500
500
HSS Soderberg
00
1
to
to
Configuration Courtyard Central
Cell Hood (or Skirt
and Burner) and
Branch Duct $ 6.18 $ 6.18
Manifold Duct 10.89 24.96
Main Duct 5.62 2.54
Total $22.69 $33.68
Basis:
Annual Production
Per Cell, Tons 262.8
Cell Current, Amps 90,000
Courtyard Central Courtyard Central
$ 5.71 Not $ 6.18 $ 5.71
Used 17.92 34.40
13.40
9.92 2.62
$19.11 $34.02 $42.73
262.8 262.8
90,000 90,000
3,500
Source - Estimates by Singmaster & Breyer
-------
Secondary control systems, which treat all build-
ing ventilation air, have been assumed to have no collec-
tion system costs which would not be incurred for build-
ing ventilation without control. Therefore the model
analyses of Section 9 have no added costs for secondary
collection.
8.2.2 Annualized Costs
Annualized costs for the control of potline ef-
fluents on the basis of dollars per ton of aluminum pro-
duced at plant capacity are taken as the sum of direct
operating costs I/• various indirect costs 2/, and an
annualized cost of invested capital _3/. Pollution control
schemes which recover fluoride and alumina values earn
corresponding credits which are applied to the total an-
nualized cost to obtain net annualized costs. The ele-
ments of estimated costs are tabulated for each of the
several model control scheme components in Tables 8.4,
pages 8-27, 8-28, 8-29, and are detailed following.
Operating labor and materials charges against col-
lection and removal equipment were analyzed on the basis
of limited data on manpower to perform various routine
tasks and on the basis of responses to the industry
questionnaire.
A weighted average of industry reported operating
labor and material costs amounted to $0.40 per ton of
aluminum produced. Using an average labor rate of $5.35
per manhour including 23 percent fringe benefits, $0.40
per ton is approximately equivalent to one man full time
per 2% potlines. Expressed another way, this is approx-
imately equivalent to 0.08 manhour per ton of aluminum.
For the purpose of allocating labor charges to each com-
ponent of control equipment, the following breakdown was
assumed.
_!/ Operating labor and materials, electric power, water,
chemicals, maintenance labor and materials.
2/ Royalty, other operating costs including taxes, in-
surance and depreciation.
_3/ Equivalent to interest at 8 percent on a continuing
debt equal to the capital cost.
8-23
-------
Manhour/Ton
Straightening and changing
collection hood shields 0.015
Unloading dry removal
equipment 0.006
Sump cleaning, nozzle and
pipe changing for wet
scrubber systems 0.034
Monitoring total systems 0.025
Total 0.080
Electric power requirement was calculated from
manufacturers information on the equipment, and was
priced at 6 mills per kilowatt hour.
Water was assumed to be consumed at a rate of
3 percent of the circulation rate required for the
equipment in question and was assumed to cost $0.013
per 1000 gallons.
Chemicals for the neutralization of acid scrub-
bing water from primary scrubber units were calculated
to cost $0.54 +_ 0.02 per ton of aluminum produced, de-
pending on the removal efficiency for HF and based on a
price of $18.00 per ton of lime (90 percent CaO).
Maintenance labor and material is estimated to
cost 5 percent of the total capital cost per year.
Royalty costs are associated with the dry scrub-
bing of effluent gas by ore grade alumina with the re-
sultant adsorption of the fluorine on the alumina.
A cost evaluation of the presently reported
licensing terms has been made by assuming an annual
interest payment for an indefinite period for a debt
that covers the fees and royalty payments, the debt
interest rate being 8 percent.
8-24
-------
Other operating costs include the following com-
ponents :
Depreciation 8% of total capital costs
Administration 5% " " "
Property taxes and
insurance 2% "
Total 15% "
The figure used for property taxes and insurance
is representative for locations in which present alumi-
num reduction plants are located, considering that the
classes of property taxes vary from state to state, and
that some states exempt air pollution control facilities
from property taxation.
Interest at a rate of 8 percent of total capital
costs on a debt of indefinite duration is assumed in
lieu of capital recovery (interest and amortization)
over a finite period of time.
8.2.3 Control Systems Credits and
Net Annualized Unit Costs
Cost credits for returned alumina and F values may
be earned by some control schemes, such as dry particu-
late collection and dry gas scrubbers which remove flu-
orine from the effluent stream in a form which may be
recycled to the cells without reprocessing. Alumina is
valued at $0.032 per contained pound and fluoride at
$0.25 per pound of contained F in dry recycle. Twenty-
five cents per pound of contained F corresponds to 90
percent cryolite at $245 per ton or to 90 percent alumi-
num fluoride at $305 per ton.
Where applicable, control scheme credits are sub-
tracted from total annualized costs for individual equip-
ment components to arrive at a net annualized cost.
An analysis of the material balance concerning
the return to the cell of adsorbed HF discloses that each
pound of F requires 0.9 pound of alumina to reform the
bath constituent, aluminum fluoride. In terms of the ef-
fluents assumed to be removed by the dry scrubbing proc-
esses as applied to the model prebake potlines in Section
8, nearly two-thirds of the removed alumina is required
8-25
-------
to react with the removed HF before that fluoride be-
comes again available to the bath from which it was
vaporized. In the cases of removal equipment in which
dry scrubbing of HF occurs, the credit for returned
alumina has been adjusted downward to account for that
which is chemically associated with the returned HF in
reforming aluminum fluoride.
A difference of opinion exists among users of
dry scrubbing systems within the aluminum industry as
to the validity of this analysis. However, the denial
of credit to 0.9 pound of alumina per pound of F results
in a conservative estimate of financial credit earned
by dry scrubbing systems.
8.2.4 Summary of Cost Elements
Tables 8.4 summarize the above cost elements for
each major component of the control schemes, including
the collection systems, for each of the three types of
potline, based on unit capacities of equipment which
conform to the requirements of the effluent flows in
the collection systems modeled.
Judgment should be exercised in the application
of these model element costs to systems defined by other
parameters. Variations from individual plant installa-
tions in airflow quantity per ton of aluminum produced
and differences between actual equipment investment
costs and the values developed for the model can have
a significant influence on both capital and net annual
costs. The model element costs presented here are use-
ful for making comparisons among control schemes, one
scheme to another.
8-26
-------
TABLE 8.4a
COST ELEMENTS - PREBAKE MODELS
EQUIPMENT DESIGNATOR
\^
^ Equipment
^\ Name
Cost \.
Component ^^\^
^\
CAPITAL COST, $/Ann Ton Al
Purchase Cost I/
Direct Installation U
Indirect Installation!/
TOTAL CAPITAL COST
ANNUALIZED UNIT, $/Ton Al
Operating Labor & Mtls
Electric Power 8/
Water i/
Chemicals 12/
Maint Labor & Mtls 12/
Royalty
Other Operating CostsiS'
Subtotal
Interest 15/
TOTAL ANNUALIZED COST
CREDITS
Returned Alumina !£/
Returned F Values 17/
TOTAL CREDITS
NET ANNUALIZED COST
PRIMARY CONTROL SCHEMES *
§
JOI
3
£S
22.69
I
22. 69i/
o.osZ/
0.35
-
-
1.13
-
3.40
4.96
1.82
6.78
-
-
-
6.78
FBDS
•o
SB
"O £1
•O W
3 >>
Efi
28.54
8.56
37.1012/
0.34S/
1.67
_
1.86
0.34^
5.57
9.78
2.97
12.75
(0.40)
(10.14)
(10.54)
2.21
VS
31 S
e M
> CO
5.71
11.42
5.14
22.27
0.31Z/
3.60
0.02
-
1.11
3.34
8.38
1.78
10.16
-
-
-
10.16
HPSS-3
3 £
a 0)
U CO 00
ffi CO O
9.51
9.51
5.71
24.73
0.311/
1.90
0.04
-
1.24
-
3.71
7.20
1.98
9.18
.
-
-
9.18
DESP I
U 0
S a
>, 0 -H
h " a
PUT*
U U
W £
7.80
5.46
3.98
17.24
0.03Z/
0.13
.
-
0.86
-
2.59
3.61
1.38
4.99
(1.07)
14.171
(5.24)
(0.25)
CFPB-5
^_^
3 *a (x
O 0) U
CO 0) 4J
0) .M >W
O U
5.52
5.52
3.31
14.35
0.31Z/
0.25
0.02
-
0.72
-
2.15
3.45
1.15
4.60
-
.
-
4.60
ST
^
»
E-i
X
CD
P.
CO
2.47
2.47
1.49
6.43
0.312/
0.25
0.02
-
0.32
-
0.96
1.86
0.51
2.37
.
-
-
2.37
MC
r-t V
a. a
1.23
0.87
0.63
2.73
0.03Z/
0.35
-
-
0.14
-
0.41
0.93
0.22
1.15
(0.87)
(3.40)
(4.27)
(3.12)
CFDS
a> y,
$ >.
13.30
10.00
7.00
30.30
0.34i/
1.67
.
.
1.52
.
4.55
8.08
2.42
10.50
(0.40)
(9.65)
(10.05)
0.45
FBWS
•8
™ 5
•3 0
SB
5.33
5.33
3.19
13.85
0.31Z/
0.41
0.02
.
.69
-
2.08
3.51
1.11
4.62
-
-
-
4.62
CS
h a
01 J3
e s
45 U
U CO
4.28
4.28
2.57
11.13
0.31Z/
0.25
0.02
.
.56
-
1.67
2.81
.89
3.70
.
-
-
3.70
VPD-3
•8
•u ji
>•&
3.33
3.33
2.00
8.66
0.31Z/
0.32
0.02
.
.43
-
1.30
2.38
.69
3.07
.
_
-
3.07
IADS
S
.a
u B u
S > S
at 0}
a B
CO CO
14.27
14.27
8.56
37.10
0.4Q6/
1.94
0.25
1.86
5.57
10.02
2.97
12.99
.
_
-
12.99
S
§,-,
8 « B
Z a 5
s£~
5.76
I
5.761/
0.201/
0.48
0.05U/
0.29
-
0.86
1.88
0.46
2.34
-
-
-
2.34
* Control device unit capacities are assumed at 100,000 acfm for dry units and 82,000 acfm for wet units.
** Based on 10 million acfm of equipment capacity.
Footnotes follow Table 8.4c.
cn
ro
CD
-------
TABLE 8.4b
COST ELEMENTS - VSS SODERBERG MODELS
CD
t 4J
to O
PU O
19. nV
r
19.11^
o.osZ/
0.06
.
_
0.96
-
2.87
3.97
1.53
5.50
-
.
-
5.50
ST
to
H
^
&
CO
1.52
1.52
0.76
3.80
0.31Z/
0.15
0.02
_
0.19
-
0.57
1.24
0.30
1.54
-
-
-
1.54
WESP
i-< to
u 0
co *J
4J CO
0) JJ
U O -H
cp
•~* to
a &
10.71
8.03
2.41
21.15
0 . 03i/
0.06
-
.
1.06
-
3.17
4.32
1.69
6.01
.
-
-
6.01
FBDS
d) to
W ol
JZ
•o ja
,
Sfi
8.0512/
A2
10.47H/
0.55ii/
0.2419/
.
.
0.52
0.3413/
1.57
3.22
0.84
4.06
(0.04)
(9.15)
(9.19)
(5.13)
DESP
u 0
CO U
jj cd
en JJ
>i O -H
O JJ -H
U U
0> CU
M fi
5.71
4.00
2.91
12.62
0.03Z/
0.04
.
_
0.63
-
1.89
2.59
1.01
3.60
(0.10)
(0.74)
(0.84)
2.76
MC
0)
.-1 01
o- e
•H 0
>-4 O
15-
0.48
0.34
0.25
1.07
0 . 03l/
0.10
.
.
0.05
-
0.16
0.34
0.09
0.43
(0.05)
(0.38)
(0.43)
0.00
VS
to
•H V
to .a
3 43
u 3
a to
a) u
> CO
3.81
7.62
3.43
14.86
0.31Z/
0.61
0.02
.
0.74
-
2.23
3.91
1.19
5.10
-
-
-
5.10
HPSS-3
a>
to C
3 S
01 a)
n to ^
cu u d>
>| CO 00
£ co
>* u
J3 CO W
60 i: I
SD.CO
CO ^
3.01
3.01
1.81
7.83
0.311/
0.27
0.02
.
0.39
-
1.17
2.16
0.63
2.79
-
-
-
2.79
CFPB-5
o.
3 13 0
O V It
1-4 to Q
fc
T3 -
CO ^ |Jc<
0 0
M CO 10
<5 &* ^^
3.01
3.01
1.81
7.83
0.311/
0.10
0.02
.
0.39
-
1.17
1.99
0.63
2.62
-
'
-
2.62
IADS |
3
f>
J3
•a 3
V co to
"> S
-i 3 >*
S1&
10.95li/
oTss
11.83ii/
0.40i/
0.06
.
_
0.59
- 18/
1.77
2.82
0.95
3.77
(0.04)
(9.15)
(9.19)
(5.42)
c
-^
M *-l 1)
at
>,
it
CO
19.98
19.98
11.98
51.94
0.40i/
2.72
0.35
_
2.60
-
7.80
13.87
4.16
18.03
-
-
-
18.03
/-s
U V -H
sse
8.005/
i
s.ool/
0.201/
0.62
.
0.12
0.40
-
1.20
2.54
0.64
3.18
-
-
-
3.18
00
I
* Control device unit capacities are assumed at 10,000 acfm for dry units and 7,000 acfm for wet units.
** Based on 10 million acfm of equipment capacity.
Footnotes follow Table 8.4c
-------
TABLE 8.4c
COST ELEMENTS - HSS SODERBERG MODELS
EQUIPMENT DESIGNATOR
Equipment
\ ^ Name
Cost ^^
Component ~^-^^
~~\_
CAPITAL COST, $/Annual Ton Al
Purchase Cost *•'
Direct Installation 2/
Indirect Installation 2/
TOTAL CAPITAL COST
ANNUALIZED UNIT, $/Ton Al
Operating Labor & Materials
Electric Power 8/
Water i/
Chemicals W.I
Maint. Labor & Materials JLr/
Other Operating Costs i*/
Subtotal
Interest il/
TOTAL ANNUALIZED COST
CREDITS
pel-tirneri AlimrfnA !&/
Returned F Values iZ/
NET ANNUALIZED COST
PRIMARY CONTROL SCHEMES *
Jo
01
tH
ffio
34.02^
1
34.02*/
o.osZ/
0.43
_
-
1.70
5.10
7.31
2.72
10.03
.
_
10.03
WESP
I*
O
n)
4J -H
a) i o.
3 -H
Q)
w«
57.00
42.75
29.93
129.68
0.03S/
0.51
.
-
6.48
19.45
26.47
10.37
36.84
_
_
36.84
CFPB-5
§T3 Cb
a) o
rn pa a)
T3
ffl U 4J
00
8.18
8.18
4.91
21.27
0.312/
0.51
0.02
.
1.06
3.19
5.09
1.70
6.79
..
_
6.79
ST
1
Sag
4J 01 -H
g S^
6 085-/
i
6.08£/
0.20#
0.44
_
0.54
0.30
0.91
2.39
0.49
2.88
_
_
2.88
SECONDARY CONTROL **
SS
c
u
CO
CO
19.98
19.98
11.98
51.94
0406/
2.72
0.35
2.60
7.80
13.87
4.16
18.03
_
_
18.03
4J
a e
u « -3
S S~
8.005/
1
8.005-/
0.206/
0.62
_
0.07il/
0.40
1.20
2.59
0.64
3.23
.
_
3.23
* Control device unit capacities are assumed at 60,000 acfm each, corresponding to 70,000 acfm dry hot gas entering.
** Based on 10 million acfm of equipment capacity.
Footnotes follow.
cn
Oi
o
-------
0631
Footnotes - Tables 8.4
!_/ Purchase cost from capacity - cost curves and tables in
Appendix 8B.
2/ Direct installation cost factored from purchase cost.
3/ Indirect installation cost equal to 30 percent of the sum of
purchase and installation costs.
4_/ Based upon Singmaster & Breyer estimates for courtyard systems
summarized in Table 8.3.
5_/ Based on 1970 costs of facilities to neutralize acid liquor -
$130 per gallon per minute of treated liquor. Water treatment
capacity required for a primary scrubber system is less than
for a secondary scrubber for equivalent aluminum output.
6f Estimated from industry questionnaire data.
7/ Based on estimated direct labor manhours at $5.35 distributed
according to the following table:
Operating Labor
Description $/Ton of Aluminum
Dry Device
Unloading collected solids .03
Wet Device
Sump cleaning, nozzle changing,
pipe cleaning .18
Monitoring wet-dry device
combination .13
.31
Primary Collection
Straightening and changing,
cell hood shields .08
Water Treatment
Monitoring water treatment plant .20
8_/ Based on 6 mills per kwh.
£/ Based on $0.013 per 1000 gallons.
10/ Based on using lime at $18.00 per ton (90% CaO) delivered.
This cost for primary systems varies + $0.02 depending on
HF removal efficiency of the scrubber and assumes prior dry
collection of solid fluoride. (In the Pacific Northwest
lime is reported (1971) to cost $25 per ton.)
8-30
-------
0632
ll/ This figure and corresponding totals apply where a smelter
includes both primary and secondary collection and removal
systems. Where a smelter includes only a secondary control
system, these costs are increased by $0.52 per ton. Small
variations introduced by differing efficiencies among con-
trol schemes are averaged.
12/ Based on 570 of total capital cost.
13/ This figure represents the sum of the licensee's schedule
of declining payments reduced to a present value, assuming
an interest rate of 870 and converted to a schedule of uni-
form payments over an indefinite period; it applies to a
smelter of 264,000 tons annual capacity. For a 35,000 ton
per year smelter, the corresponding payment is $0.46 per
ton aluminum.
14/ Based on 1570 of total capital cost with taxes and insurance
at 2%, administration at 57<> and depreciation at 87=.
15/ Based on a continuing debt equal to total capital cost with
interest at 87o.
16/ Based on a value of $0.032 per pound of aluminum returned
dry to the cells.
17/ Based on a value of $0.25 per pound of contained F returned
dry to the cells.
18/ A technical know-how fee is included in the capital cost.
19/ Cost estimates for the FBDS and IADS systems are derived
from data given by exclusive suppliers of these systems on
a turnkey basis.
8-31
-------
,*S f"* /•> if**
1 n T, ^
W W V*-' V/
8.3 Cost Effectiveness of Control Models
To compare the cost effectiveness of the various
emission control schemes represented by the models, the
component costs from Tables 8.4 have been summed and
tabulated against the overall control efficiency pre-
dicted for each model control scheme. These data are
shown in Tables 8.5.
Control schemes known to be in current plant use
for the indicated potline applications are underlined in
Tables 8.5 and their control efficiencies are derived
from reported data. Efficiencies for the other poten-
tially applicable schemes were developed from test data
or from fractional removal efficiency curves for the
equipment.
The tabulation of the performance of the model
control schemes, representing the application of the
most effective available kinds of effluent removal
equipment, illustrates the strong influence of collec-
tion or hooding efficiency on overall control efficien-
cy, and the cost effectiveness relationships illustrate
the orders of magnitude of costs associated with var-
ious control efficiency levels for the different potline
types. Control schemes using primary control alone are
limited in overall control efficiency to the value of
the collection efficiency. If secondary scrubbers are
added to primary systems, overall control efficiencies
ranging in the high 90's can be achieved even with
relatively low collection efficiency but at increased
cost.
8-32
-------
Table 8.5a
COSTS AND OVERALL CONTROL EFFICIENCIES
PREBAKE POTLINES - MODEL I
Overall Control
Eff. Total F
Model
Scheme
No.
IA-1
-2
-3
-4
-5
-6
-10
-11
-12
-13
-14
-15
-16
-17
IB-1,9
-2,9
-3,9
-4,9
-5,9
-6,9
Control Scheme*
Primary
Stage 1
FBDS
MC
MC
DESP
MC
MC
CFDS
MC
ST
MC+DESP
MC
MC
CS
IADS
FBDS
MC
MC
DESP
MC
MC
Stage 2
.
VS
HPSS-3
ST
CFPB-5
ST
-
-
-
ST
VPB-3
FBWS
-
-
_
VS
HPSS-3
ST
CFPB-5
ST
Sec-
ondary
.
-
-
-
-
-
-
-
-
-
-
-
-
SS
SS
SS
SS
SS
SS
Model Costs
Cap-
ital
60
51
53
49
43
35
53
25
32
52
37
42
37
55
106
94
96
92
86
78
, $/ton
Al Collection
Annualized
Unit
19.80
19.70
18.70
15.80
14.20
11.90
17.30
7.90
10.80
16.90
12.60
14.20
12.10
17.00
35.90
35.00
34.00
31.10
29.50
27.20
Credit
10.
4.
4.
5.
4.
4.
10.
4.
-
5.
4.
4.
-
10.
10.
4.
4.
5.
4.
4.
50
30
30
20
30
30
00
30
20
30
30
50
50
30
30
20
30
30
Net
9.30
15.40
14.40
10.60
9.90
7.60
7.30
3.60
10.80
11.70
8.30
9.90
12.10
6.50
25.40
30.70
29.70
25.90
25.20
22.90
^ c
96.0
94.6
93.8
92.1
92.4
89.6
85.2
89.6
32.2
85.2
89.2
71.8
87.2
83.0
94.1
96.5
95.7
94.0
94.3
91.5
87.1
9
85.
84.
83.
82.
82.
80.
76.
80.
26.
75.
79.
64.
77.
74.
84.
91.
90.
88.
89.
86.
82.
Eff.
6
4
6
1
3
0
0
0
7
9
6
0
2
0
0
1
3
8
0
7
7
All Effluent
IC-7
-8
-9
_
-
-
-
CFPB-3
FBWS
SS
68
99
43
22.50
32.40
15.30
22.50
32.40
15.30
to
92
85
72
Sec.
.5
.2
.1
Underlined equipment designators signify equipment in current use
on prebake potlines.
* Legend follows Table 8.5c.
8-33
-------
OE
Table 8.5b
CONTROL SCHEME COSTS AND OVERALL CONTROL EFFICIENCY
VSS SODERBERG POTLINES - MODEL II
Model
Scheme
No.
Control Scheme*
Primary
Stage
1
Stage
2
Sec-
ondary
Model Costs, $/ton Al
Cap- Annualized
ital Unit Credit Net *{c = 80.4
Overall
Control
Eff. Total F
IIA-1
-2
-3
-4
-5
-6
-8
-9
ST
FBDS
DESP
MC
MC
MC
ST
IADS
WESP
-
ST
ys
HPSS-3
CFPB-5
-
-
IIB-1,7 _ST
-2,7 FBDS
-3,7 DESP
-4,7 MC
-5,7 MC
-6,7 MC
49
30
41
40
33
33
28
31
15.60
9.70
13.20
13.60
11.30
11.10
9.60
9.30
-
9.20
0.80
0.40
0.40
0.40
-
9.20
15.60
0.50
12.40
13.20
10.90
10.70
9.60
0.10
79.6
,5
,5
.4
79.
79.
79.
78.4
78.0
74.6
78.7
WESP
-
ST
VS
HPSS-3
CFPB-5
SS
SS
SS
SS
SS
SS
109
90
101
100
93
93
36.80
31.00
34.40
34.80
32.50
32.30
-
9.20
0.80
0.40
0.40
0.40
36.80
21.80
33.60
34.40
32.10
31.90
93.8
93.7
93.7
93.6
92.6
92.2
Underlined equipment designators signify equipment in current use on
VSS Soderberg potlines.
*Legend follows Table 8.5c.
8-34
-------
Table 8.5c
CONTROL SCHEME COSTS AND OVERALL CONTROL EFFICIENCY
HSS SODERBERG POTLINES - MODEL III
Model Control Scheme*
Scheme Primary Sec-
No. Stage 1 Stage 2 ondary
IIIA-1 CFPB-5
IIIB-1,6 CFPB-5
-2,6 ST
-3,6 CFPB-5
WESP
WESP
WESP
WESP
SS
SS
SS
Model Costs. $/ton Al
Cap- Net
ital Unit Credit Unit
Overall
Control
Eff. Total F
c = 85.6
191
181
61
51
61
251
241
121
56.50
53.50
19.70
16.70
19.70
77.80
74.80
41.00
56.50
53.50
19.70
16.70
19.70
77.80
74.80
41.00
83.9
81.3
78.0
70.2
76.9
90.4
87.8
84.5
Underlined equipment designators signify equipment in current use on
HSS Soderberg potlines.
* Legend follows.
8-35
-------
0637
Legend for tables 8.5a, b and c
CFPB-5
CFPB-3
CS
DESP
FBDS
FBWS
HPSS-3
IADS
MC
ss
ST
VPB-3
VS
WESP
Cross Flow Packed Bed Scrubber (5 ft. Deep)
Cross Flow Packed Bed Scrubber (3.25 ft. Deep)
Chamber Scrubber
Dry Electrostatic Precipitator
Fluidized Bed Dry Scrubber
Floating Bed (Bouncing Ball) Wet Scrubber
High Pressure Spray Screen (3-Stage)
Injected Alumina Dry Scrubber
Multiple Cyclone
Spray Screen
Spray Tower
Vertical Flow Packed Bed Scrubber (3 ft. Deep)
Venturi Scrubber
Wet Electrostatic Precipitator
8-36
-------
References - Section 8
Less, L.N. and Waddington, J., "The Character-
istics of Aluminum Reduction Cell Fume", Light
Metals 1971, Proceedings of Symposia, 100th
AIME Annual Meeting, New York, March 1-4, 1971.
U.S. Department of Health, Education and Welfare,
"Control Techniques for Particulate Air Pollu-
tants" AP-51, National Air Pollution Control
Administration, January 1969.
8-37
-------
Table of Contents
Section 9
9.0 Analysis of Control and Improvement
By Industry Models
9.1 Potential Industry Control Practice
In Terms of 1971 Capacity by Models
9.1.1 Control By Cell Type
9.1.2 By Collection Type
9.1.3 By Emission Control Scheme
9.2 Industry Control Costs
In Terms of 1971 Capacity by Models
9.3 Improvement in Industry Emission Control
9.3.1 Improvement in Overall Control Efficiency
to a Plant Minimum of 80 Percent
9.3.2 Improvement in Overall Control Efficiency
by the Application of Best Primary System
Control
9.3.3 Improvement in Overall Control Efficiency
to a Plant Minimum of 90 Percent
9.3.4 Improvement in Overall Control Efficiency
to the Best Demonstrated Technology
9.4 Future Control Costs 1975-2000
-------
9.0 Analysis of Control and Improvement
By Industry Models
In Section 7.0 of this report, effluent and emis-
sion data obtained concerning the various plants in the
primary aluminum industry were presented as industry
totals and weighted averages to conform with the restric-
tions on the confidentiality of individual plant data
under which responses were made.
To permit systems analysis of the control and im-
provement of the process variations within the industry,
individual process models (representing capacity and
production portions of the total) have been structured
specific to the various and differing control problems.
(See Section 8). These model types and schemes, com-
bined in the proportions of capacity tonnages represen-
tative of industry practice, are totaled to provide an
overall industry model. Operations on individual seg-
ments of the model may then be made to effect systems
analyses of the industry as a whole. The degree of in-
dustry control can be analyzed and costed, and effect
of improvement in individual process segments of the
total industry can be evaluated.
By considering the factors involved in altering
one model to another of higher performance characteris-
tics, estimates may be made with respect to the cost
and practicability of the pollution abatement control
of the total industry, or of its major segments. Such
estimates are developed in Section 9.3.
Direct comparison between an individual plant and
the model system representing a process and/or control
grouping is not intended. The models are hypothetical
and are representative of the achievable performance
levels of the various components that are, or may be,
assembled to treat typical sets of plant conditions.
Operating and equipment parameters of actual installa-
tions will, in each plant case, deviate from these con-
ditions. Calculated model performances may be used
only as a guide to those to be expected in actual prac-
tice.
9-1
-------
U6<*1
9.1 Potential Industry Control Practice
in Terms of 1971 Capacity by Models
Table 9.1 shows the comparative relationships
among levels of control effectiveness of the portions of
the total United States aluminum industry represented by
the individual cell type models and methods of collect-
ing effluents, as the industry was structured at the end
of 1971.
9.1.1 Control by Cell Type
The analysis indicates that prebake anode plants
(Model I) accounted for about 65 percent of the total
aluminum capacity in 1971, and that their total emission
control was approximately 74 percent. These plants
accounted for about 66 percent of the industry potential
emissions of total fluorides, including the emissions
from one plant which exercised no control over potroom
effluents.
VSS Soderberg plants (Model II) accounted for
about 13 percent of the total aluminum capacity and about
9 percent of the total fluoride emissions. The general
overall emission control exercised in these plants was
approximately 83 percent.
HSS Soderberg plants as a group comprise 22 per-
cent of the total aluminum capacity and were responsible
for 26 percent of the potential total fluoride emissions.
Overall emission control efficiency was lower (70 per-
cent) than in the other types of plants.
The industry model as a whole showed a 74 percent
overall emission control with respect to total fluorides.
9.1.2 By Collection Type
Considering the breakdown into the methods of ef-
fluent collection utilized by the industry in the model
types, Table 9.1 shows that less than 3 percent of the
aluminum capacity of the industry was in prebaked pot-
lines without emission control, accounting for about 10
percent of the total fluoride emissions from the in-
dustry potlines.
9-2
-------
Table 9.1
DISTRIBUTION OF EMISSION CONTROL, 1971,
BY ANODE CONFIGURATION AND COLLECTION TYPES
VD
I
Model
I-A
I-B
I-C
I-D
II-A
II-B
II
(Prebake Primary Only)
(Primary and Secondary)
(Secondary Only)
(Uncontrolled)
All Prebake Potlines
(VSS Primary Only)
(Primary and Secondary)
All VSS Soderberg
III-A (HSS Primary Only)
Industry
Aluminum
Capacity
1000 Tons
2,160
100
638
122
3,020
410
191
601
1,033
4,654
Tons Total
Effluent
49,680
2,300
14,674
2,806
69,460
9,430
4,393
13,823
23,759
107,042
Fluoride
Emissions
11,341
398
3,567
2,806
18,112
2,134
272
2,406
7,056
27,574
7» Overall
Control
Efficiency
77.2
82.7
75.7
0
73.9
77.4
93.8
82.6
70.3
74.2
Emissions
Lb/1000 Ib Al
5.3
4.0
5.6
23.0
6.0
5.2
1.4
4.0
6.8
5.9
C3
CO
-------
Controlled emission prebake potlines (62 percent
of total aluminum capacity) released 56 percent of total
industry fluoride emissions. The greater portion of
these prebake potlines employ primary collection and
control only, with overall control efficiency amounting
to 77 percent of cell effluent. A somewhat better con-
trol efficiency (83 percent) is obtained on the smaller
tonnages of prebake lines employing both primary and
secondary control in separate collection systems. Lines
from which total collection to a secondary system is
effected without the use of primary hooding treat about
21 percent of the prebaked capacity tonnage, (14 percent
of total capacity), and achieve about 76 percent control
efficiency.
On the VSS Soderberg production (13 percent of
total capacity) overall control efficiency is 77 percent
for those lines using primary collection only which ac-
count for 68 percent of the VSS Soderberg tonnage. Over-
all collection efficiency is 94 percent for lines using
both primary and secondary control on 32 percent of the
VSS Soderberg tonnage.
All HSS Soderberg capacity (22 percent of total)
is in potlines with primary collection and control only,
and overall emission control efficiency is relatively
low, at 70 percent. HSS Soderberg capacity accounts for
26 percent of the industry fluoride emissions.
9.1.3 By Emission Control Scheme
The analysis of model performance of the sections
of the industry which correspond to the emission control
schemes in use on the three types of potlines is shown
in Tables 9.2 and 9.3.
Prebake Potlines (Table 9.2) have been separately
divided into "old" and "new" cell categories* with dif-
ferent levels of primary collection efficiency as noted
in the previous section. The "new" group accounts for
the smaller (21 percent) portion of prebake cell capacity,
and the effluent is treated in primary collection/control
systems with an overall control efficiency of 85 percent.
* See pages 8-6 and 8-7 for definitions.
9-4
-------
Table 9.2
DISTRIBUTION OF EFFLUENT CONTROL, 1971,
BY CONTROL SCHEMES
PREBAKE POTLINE CAPACITY
Model
No.
IA (New)*
-1
-15
-10
-11
IA(New)
IA (Old)*
-6
-1
-4
-11
-13
-10
-16
-12
-14
IA (Old)
IA Total
IB-6,9 (Old)
IC-7
-9
1C Total
ID
Equipment
Designator*
Prim. Sec.
FBDS
MC+FBWS
CFDS
MC
Subtotal
MC+ST
FBDS
DESP+ST
MC
MC+DESP+ST
CFDS
CS
ST
MC+VPB-3
Subtotal
MC+ST SS
CFPB-3
SS
None
Percent
Total
Capacity
6.5
3.9
1.9
1.5
13.8
12.9
7.3
3.5
3.5
2.6
1.5
0.8
0.3
0.1
32.5
46.3
2.1
2.4
11.3
13.7
2.6
Percent
Total
F Emission
1.4
1.9
0.8
3.9
8.0
12.0
4.4
2.4
9.9
2.0
1.2
0.9
0.3
0.2
33.3
41.3
1.4
0.7
12.2
12.9
10.1
Overall
Control
Efficiency
94.6
87.2
80.0
32.2
85.0
76.0
84.4
82.3
26.7
79.6
80.0
74.0
75.9
64.0
73.6
77.0
82.7
92.5
72.1
75.7
0
Emission
Rate, Ib F
/1000 Ib Al
1.2
2.9
2.4
15.6
3.4
5.5
3.6
4.1
16.9
4.7
4.6
6.0
5.5
8.3
6.1
5.3
4.0
1.7
6.4
5.6
23.0
Total Model I (PB)
64.9
65.8
73.9
6.0
* Definitions follow Table 9.3, page 9-7.
9-5
-------
Table 9.3
DISTRIBUTION OF EFFLUENT CONTROL, 1971,
BY CONTROL SCHEMES
VSS AND HSS SODERBERG POTLINE CAPACITY
AND INDUSTRY TOTAL
Model
No.
Equipment
Designator*
Prim. Sec.
Percent
Total
Capacity
Percent
Total
F Emission
Overall
Control
Efficiency
Emission
Rate, Ib F
/1000 Ib Al
VSS Potlines
IIA-4 MC+VS
-8 ST
-1 ST+WESP
HA Subtotal
IIB-1,7
ST+WESP
SS
Total Model II (VSS)
3.8
3.8
1.3
8.8
4.1
.2.9
3.0
3.7
1.0
7.7
0.1
8.7
79.4
74.6
79.5
77.4
93.8
82.6
4.7
5.8
4.7
5.2
1.4
4.0
HSS Potlines
IIIA-9 ST
22.2
25.5
70.3
6.8
Total Industry
100.0
100.0
74.3
5.9
* Definitions follow.
9-6
-------
Legend for Tables 9.2 and 9.3
CFPB-3 Cross Flow Packed Bed Scrubber (3.25 ft. Deep)
CS Chamber Scrubber
DESP Dry Electrostatic Precipitator
FBDS Fluidized Bed Dry Scrubber
FBWS Floating Bed (Bouncing Ball) Wet Scrubber
MC Multiple Cyclone
SS Spray Screen
ST Spray Tower
VPB-3 Vertical Flow Packed Bed Scrubber (3 ft. Deep)
VS Venturi Scrubber
WESP Wet Electrostatic Precipitator
For definitions of "New" and "Old", see page 8-6
9-7
-------
u £' t7
The "old" prebake group, comprising about 32 per-
cent of the total industry capacity, is largely under
primary collection/control only, with overall control
efficiency of 73 percent. It is noted that in this
group the production tonnage represented by Model IA-11
contributes 10 percent of the total industry emissions
from less than 4 percent of the industry tonnage: over-
all control efficiency is very low compared to the rest
of the group, and no attempt is made in this control
scheme to remove gaseous fluorides from the potline
effluents.
Better emission control (83 percent) is obtained
on the small portion of "old" cell capacity in which a
secondary control system is used to supplement the pri-
mary (IB-6,9). Production in which all prebake efflu-
ent is treated by secondary control only (Models 1C),
represents 14 percent of the total tonnage and emits 13
percent of the total fluorides of the industry, achiev-
ing 76 percent emission control.
The contribution to the total fluoride emission
inventory by the production elements using various con-
trol schemes on prebake potline effluents varies quite
widely. Emission rates for the better controlled con-
ditions of the newer lines range between 1.2 and 2.9
pounds total fluoride per 1000 pounds of aluminum pro-
duced. The corresponding range for older type instal-
lation with primary collection only is 4 to 6 pounds,
excluding the one model group (IA-11) which is greatly
in excess of the rest.
VSS Potlines emission control schemes are com-
pared in Table 9.3. Control of primary effluent alone
results in overall emission control efficiency of 75-80
percent as applied to 9 percent of the industry capac-
ity, and contributes 8 percent of the total industry
emission inventory. A marked improvement in control
is shown for the plants which collect and treat VSS
potline effluent in both primary and secondary systems.
The resulting emission rate of this group is about 1.4
pounds of total fluorides per 1000 pounds of aluminum
produced.
9-8
-------
ESS Soderberg Potlines utilize only one control
scheme, of which the overall control efficiency is about
70 percent. HSS capacity is 22 percent of the industry
total, and accounts for 26 percent of the industry emis-
sion inventory of total fluorides.
9.2 Industry Control Costs in Terms
of 1971 Capacity by Models
The unit cost and capital cost of present indus-
try control can be estimated from the models by summing
the cost elements of the various collection, removal,
and treatment schemes of the models which are detailed
in Section 8.0, and applying these model costs to the
corresponding industry tonnages after adjusting for dif-
ferences in flow rate between actual plants and the flow
rates assumed for the models.
Table 9.4 presents this estimate for the model
breakdown of capacity tonnages of the 1971 industry. It
is noted that this estimate is constructed from derived,
not reported data as explained in Section 8. Direct cost
comparison with individual plants is not justified because
plants may incorporate several models in their operations,
and individual operating conditions will vary from the
models.
The derived industry costs given in Table 9.4 in-
clude cost elements which were not accounted for in the
industry responses to the questionnaire (Section 7.3).
Model capital investment totals include indirect instal-
lation costs amounting to 30 percent of the sum of pur-
chase and direct installation costs for equipment. When
this adjustment is made to the reported industry invest-
ment, the agreement with the industry model is good, and
the model capital cost may be regarded as being conserv-
ative.
Model net annual costs include allowances for cap-
ital recovery costs amounting to 23 percent of the total
capital cost for depreciation, interest, administration,
taxes and insurance. Applying these allowances to the
reported industry direct operating costs reconciles them
with the estimate constructed from the industry models.
9-9
-------
Table 9.4
1971 ALUMINUM INDUSTRY COSTS FOR CONTROL BY MODELS
1A-1
IA-1
IA-4
IA-6
IA-10
IA-10
IA-11
IA-11
IA-12
IA-13
IA-14
IA-15
IA-16
IB-6,
IC-7
IC-9
ID
Model
(N)
(0)
(0)
(0)
(N)
(0)
(N)
(0)
(0)
(0)
(0)
(N)
(0)
9 (0)
Equipment
Designator
Prim.
FBDS
FBDS
DESP+ST
MC+ST
CFDS
CFDS
MC
MC
ST
MC+DESP+ST
MC+VPB-3
MC+FBWS
CS
MC+ST
I (Prebake)
IIA-1
IIA-4
IIA-8
IIB-1,7
Model II (VSS)
IIIA-9 (HSS)
ST+WESP
MC+VS
ST
ST+WESP
ST
Sec.
SS
CFPB-3
SS
SS
Capacity
1000 Tons
302
338
165
601
90
70
70
163
15
120
6
180
40
100
112
526
122
3,020
60
175
175
191
601
Flow
Adjustment
Factor
1.00
1.00
1.30
1.12
1.00
1.00
1.12
1.23
1.82
1.00
0.97
1.17
1.32
1.20
1.00
1.00
-
-
1.00
1.00
1.15
1.16
-
Model Capital Cost
Unit
$/Cap'y Ton
60
60
64
39
53
53
28
31
58
52
36
49
49
100
72
48
0
50.11
49
40
32
109
60.50
Total
MMS
18.1
20.3
10.5
23.6
4.8
3.7
2.0
5.1
0.9
6.2
0.2
8.8
2.0
10.0
8.1
25.2
0
149.3
2.9
7.0
5.6
20.8
36.4
Model Annual ized Cost
Unit
$/Ton
9.30
9.30
13.80
8.50
7.30
7.30
4.10
4.60
19.70
11.70
8.00
11.60
16.00
29.40
23.80
16.90
0
11.59
15.60
13.20
11.00
42.70
21.96
Total
MM $/yr
2.8
3.1
2.3
5.2
0.7
0.5
0.3
0.9
0.4
1.4
0.1
2.1
0.7
2.9
2.7
8.9
0
35.0
0.9
2.3
2.0
8.0
13.2
1,033
1.00
49
50.6
16.70
17.2
Total Industry
4,654
50.77
236.3
14.05
65.4
-------
9.3 Improvement in Industry Emission Control
The effectiveness and costs of improving the pres-
ent level of overall emission control of the industry
can be evaluated by selectively upgrading individual
model schemes to others of higher performance, applying
the unit cost increments involved to the capacity ton-
nages represented by the model modification, and then
reconstructing the cost and emission model of the total
industry.
Emission data have similarly been applied to ca-
pacity tonnages, resulting in the analysis representing
improvement in emission control under conditions of full
production. Operation at less than full capacity would
reduce the cost effectiveness (pounds of emission reduc-
tion per dollar of capital cost) of the upgrading.
For the purpose of this evaluation, added cost and
performance improvement estimates have been made to im-
prove air pollution abatement in the existing aluminum
industry from its present overall control efficiency of
approximately 74 percent to four higher levels of per-
formance, to wit:
Raise all plants to a minimum 80 percent overall
control efficiency.
Apply best demonstrated primary control technol-
ogy in all plants.
Raise all plants to a minimum 90 percent overall
control efficiency.
Apply the best demonstrated technology, both pri-
mary and secondary, to all plants.
In performing these upgrading analyses, no judgment
has been made as to their economic feasibilities.
In raising overall control efficiency to the inter-
mediate levels, model schemes have been modified with
consideration given to the least capital investment re-
quired to effect the level of improvement; in the optimum
performance case, some models require investments which
may be considered out of proportion to the improvements
realized.
9-11
-------
06^
The analysis is made by taking each of the model
schemes represented by an industry segment, determining
the new model scheme represented by modification re-
quired to reach the new control level, calculating the
unit costs of the model scheme conversion, and obtaining
the incremental costs for that segment. Emissions are
recalculated from the overall control efficiency of the
new model, and industry costs and emissions combined to
determine cost effectiveness.
For determining the costs of upgrading the models
it is assumed that, in most cases, modifications will
be made and new ducting added to the primary collection
systems amounting to an arbitrary 160 percent of the
estimated cost for main ducting in courtyard systems.
It is further assumed that, when elements of existing
control systems are changed from one type to another,
the original element will be either bypassed without
cost, or will be removed to provide physical space for
the new element. In the latter case the net cost of
demolition, including salvage credit, is estimated to
be 75 percent of the direct installation cost of equip-
ment removed.
Capital and annualized costs for upgraded segments
of the industry are developed from the model costs shown
in Tables 8.4 with adjustments to allow for the fact
that the installation of a piece of control equipment in
an existing plant costs more than the installation of
the same equipment in a new plant. It has been assumed
that upgraded capital costs are 15 percent higher than
new construction. This is reflected as a 12 percent
penalty on annualized cost.
It is recognized that these cost assumptions can
only be applied to the general case of model modifica-
tion, and that they may not be applicable to individual
plant modification. They are considered to be valid in
structuring the overall industry model costs.
In the upgrading of model schemes it was not con-
sidered practical to overcome the inherent limitations
imposed by the existing types of prebake collection sys-
tems; no attempt was made to upgrade the collection ef-
ficiency of the older generation of prebake potlines to
that of the new, as the cost of such structural modifi-
cation is both major and indeterminant for the models.
9-12
-------
UbbZ
However, a plant, faced with the need to improve its
overall control efficiency, may determine that provi-
sions for improved collection efficiency give greater
benefit for the capital expenditure than the installa-
tion of some types of new removal equipment.
A sample calculation, illustrating the develop-
ment of costs to upgrade the overall control efficiency
in one segment of the existing aluminum industry, is
given in Appendix 9A.
The summary of the results of the various up-
grading analyses is shown in Table 9.5. Working from
a base of an industry model of the capacity available
for production in 1971 and its control costs and effec-
tiveness, improvement of industry control to a minimum
of 80 percent can be realized by modification of ten
of the twenty-two models at an investment of $237 mill-
ion. (Ref. Table 9.6) Industry emission rate is re-
duced from 5.9 to 3.7 pounds total fluoride per 1000
pounds aluminum produced at capacity and at an average
added cost of $21 per ton aluminum, applied to the
modified segment of the industry.
Conversion of the United States plants to incor-
porate the best demonstrated control on primary gas
streams without the addition of secondary control would
raise the industry average efficiency from 74 to 86
percent. Fourteen of the twenty-two models would be
affected at an added capital cost of $314 million. The
average cost of producing aluminum at these fourteen
plants would increase by about $18 per ton. (Ref.
Table 9.7) An examination of the cost effectiveness
parameters for improved models, shown in Table 9.5,
suggests that this project may be more attractive than
others. Based on net annualized cost per ton of alu-
minum produced at plant capacity, this alternative
shows the highest capture of effluent F per dollar
spent for pollution control, 1.6 pounds per dollar.
Raising industry control to a minimum 90 percent
level would involve modification of eighteen of the
twenty-two models at a cost of $477 million, and result
in reduction of the industry emission rate from 5.9 to
2.2 pounds total fluoride per 1000 pounds of aluminum.
The added operating cost of this improved control would
be $35 per ton, applied to the modified plant. (Ref.
Table 9.8)
9-13
-------
TABLE 9.5
SUMMARY OF PERFORMANCE AND COSTS
UPGRADING EMISSION CONTROL
o
cr
on
Annual Capacity, 1000 Tons Al
Annual Effluent, 1000 Tons Al
Annual Emisston, 1000 Tons Al
Emission Rate, Ib F/1000 Ib Al
Overall Control Efficiency
Capture Rate, Ib F/1000 Ib Al
Capital Investment
Total Before Upgrade, MM$
Added Total, MM$
Total After Upgrade, MM$
Unit Before Upgrade, $/Ton
Added Unit
Unit After Upgrade, $/Ton
Annual ized Cost
Total Before Upgrade, MM$/Year
Added Total, MM$/Year
Total After Upgrade, MM$/Year
Unit Before Upgrade, $/Ton
Added Unit, $/Ton
Unit After Upgrade, $/Ton
Cost Effectiveness,
Ib F Captured/$ Spent
Capital Investment
Annualized Cost
Base
1971
Industry
4,654
107
27.5
5.9
74.3
17.1
236
51
65
14
0.7
2.6
Min. 807. OCE
Unchanged
1,903
44
5.7
3.0
87.0
20.0
121
-
121
64
-
64
29
-
29
15
-
15
0.6
2.9
Modified
Segment
2,751
63
11.4
4.1
82.0
18.9
115
237
352
42
86
128
36
58
94
13
21
34
0.3
1.1
Total
Industry
4,654
107
17.1
3.7
84.0
19.3
236
237
473
51
50
101
65
58
123
14
12
26
0.4
1.5
Best Primary
Unchanged
1,443
33
4.5
3.1
86.4
19.9
97
-
97
67
• -
67
25
-
25
17
-
17
0.6
2.3
Modified
Segment
3,211
74
10.8
3.4
85.3
19.6
139
257
396
43
80
123
40
55
95
13
17
30
0.3
1.3
Total
Industry
4,654
107
15.3
3.3
85.6
19.7
236
257
493
51
55
106
65
55
120
14
11
26
0.4
1.5
Min. 907. OCE
Unchanged
695
16
1.1
1.6
93.4
21.4
51
-
51
73
-
73
14
-
14
20
-
20
0.6
2.2
Modified
Segment
3,959
91
8.7
2.2
90.3
20.8
185
477
662
47
120
167
51
133
184
13
34
47
0.2
0.9
Total
Industry
4,654
107
9.8
2.2
90.5
20.8
236
477
713
51
102
153
65
133
199
14
29
43
0.3
1.0
Best Technology
Unchanged
191
4
0.3
1.4
93.8
21.6
21
-
21
110
-
110
8
-
8
42
-
42
0.4
1.0
Modified
Segment
4,463
103
7.7
1.8
92.1
21.2
215
576
791
48
129
177
57
139
196
13
31
44
0.2
1.0
Total
Industry
4,654
107
8.0
1.8
92.2
21.2
236
576
812
51
124
175
65
139
204
14
30
44
0.2
1.0
-------
065^
Maximum control within limits of best demon-
strated technology would require a capital investment
in modification of all but one of the models amounting
to $584 million and an added operating cost of $32 per
ton of aluminum. Under these conditions, total indus-
try emissions could be reduced to a rate of 1.8 pounds
total fluoride per ton aluminum. At this level, total
investment by the industry in emission control would be
more than 800 million dollars, and net operating cost
of air pollution control would be some $43 per ton of
aluminum.
This analysis illustrates the sharply rising
costs involved as higher levels of emission control are
achieved by the industry.
9.3.1 Improvement in Overall Control Efficiency
to a Plant Minimum of 80 Percent
Table 9.6 indicates the results of performance
and cost analyses to upgrade ten of the industry models,
representing 59 percent of the total aluminum capacity,
so that all plants achieve at least 80 percent overall
control efficiency. The average control efficiency of
these ten models increases from about 66 percent to 82
percent with a resulting total industry efficiency of
84 percent.
9.3.2 Improvement in Overall Control Efficiency by
the Application of Best Primary System Control
If all models of the United States aluminum in-
dustry were to adopt the best demonstrated control on
primary collection streams, fourteen models represent-
ing 69 percent of the capacity, would be modified as
indicated in Table 9.7. Their average control effi-
ciency would increase from about 69 percent to more than
85 percent, boosting the total industry efficiency to
nearly 86 percent.
9-15
-------
TABLE 9.6
UPGRADING INDUSTRY MODELS
MINIMUM OVERALL CONTROL EFFICIENCY 80%
CD
rn
en
Oi
Old
Model
No.
IA-6(0)
IA-11(N)
IA-11(0)
IA-12(0)
IA-14(0)
IA-16(0)
IC-9
ID
IIA-8
IIIA-9
Upgraded
Segment
Total
Industry
Cap'y
1000
Tons
601
70
163
15
6
40
526
122
175
1,033
2,751
4,654
Old
OCE, %
76.0
32.2
26.7
75.9
64.0
74.0
72.1
0
74.6
70.3
65.6
74.2
New
Model
No.*
IA-5
IA-6
IA-5
IA-5
IA-5
IA-5
IA-6, 9
IA-5
IIB-8,7
IIIA-2
New
OCE, %
80.0
85.2
80.0
80.0
80.0
80.0
84.6
80.0
88.8
81.3
82.0
84.0
Annual Emission
Old
3,318
1,092
2,748
83
50
239
3,375
2,806
1,022
7,056
21,789
27,648
New
2,765
238
750
69
28
184
1,863
561
451
4,443
11,352
17,137
Tons F
Decrease
553
854
1,998
14
22
55
1,512
2,245
571
2,613
10,437
10,437
Capital Investment
Unit, $/Ton
Old
39
28
31
58
36
49
48
32
49
42
51
Add
31
12
20
46
18
26
40
49
69
167
86
50
- New
70
40
51
104
54
75
88
49
101
216
128
101
Total, MM$
Old
23.6
2.0
5.0
0.9
0.2
2.0
25.2
5.6
50.6
115.1
236.4
Add
18.6
0.8
3.3
0.7
0.1
1.0
21.1
6.0
12.1
172.9
236.6
236.6
New
42.2
2.8
8.3
1.6
0.3
3.0
46.3
6.0
17.7
223.5
351.7
473.0
Annualized Costs
Unit, $/Ton
Old
8.50
4.10
4.60
19.70
8.00
16.00
16.90
11.00
16.70
12.90
14.05
Add
3.30
4.48
5.15
(0.19)
2.50
5.80
8.58
11.08
23.75
42.68
20.90
12.36
New
11.80
8.58
9.75
19.51
10.50
21.80
25.48
11.08
34.75
59.38
33.80
26.41
Total, MM$/Year
Old
5.2
0.3
0.9
0.4
0.1
0.7
8.9
2.0
17.2
35.7
65.4
Add
2.0
0.3
0.8
0.0
0.0
0.2
4.5
1.4
4.2
44.1
57.5
57.5
New
7.2
0.6
1.7
0.4
0.1
0.9
13.4
1.4
6.Z
61.3
93.2
122.9
* Definitions follow Table 9.9, page 9-21
-------
TABLE 9.7
UPGRADING INDUSTRY MODELS
BEST PRIMARY CONTROL
Old
todel
No.
IA-6(0)
IA-10(N)
IA-10(0)
IA-11(N)
IA-11(0)
IA-12(0)
IA-13(0)
IA-14(0)
IA-15(N)
IA-16(0)
IC-9
ID
IIA-8
IIIA-9
Upgraded
Segment
Total
Industry
Cap'y
1000
Tons
601
90
70
70
163
15
120
6
180
40
526
122
175
1,033
3,211
4,654
Old
OCE, 7,
76.0
89.6
80.0
32.2
26.7
75.9
82.3
64.0
87.2
74.0
72.1
0
74.6
70.3
69.3
74.2
New
Model
No.*
IA-1
IA-1
IA-1
IA-1
IA-1
IA-1
IA-1
IA-1
IA-1
IA-1
IB-1,9
IA-1
IIA-3
IIIA-2
New
-OCE, 7.
84.4
94.6
84.4
94.6
84.4
84.4
84.4
84.4
94.6
84.4
91.1
84.4
79.5
81.3
85.3
85.6
Annual Emission
Old
3,318
215
322
1,092
2,748
83
563
50
530
239
3,375
2,806
1,022
7,056
27,775
27,648
New
2,156
112
251
87
585
54
431
22
224
144
1,077
438
825
4,443
10,849
15,407
Tons F
Decrease
518
103
71
1,005
2,163
29
132
28
306
95
2,298
2,368
197
2,613
11,926
12,241
Capital Investment
Unit, $/Ton
Old
39
53
53
28
31
58
52
36
49
49
48
-
32
49
43
51
Add
62
62
62
60
65
93
61
55
67
58
69
69
27
167
80
55
New
101
115
115
88
96
151
113
91
116
107
117
69
59
216
123
106
Total, MM$
Old
23.6
4.8
3.7
2.0
5.0
0.9
6.2
0.2
.8.8
2.0
25.2
-
5.6
50.6
138.6
236.4
Add
37.3
5.6
4.3
4.2
10.6
1.4
7.3
0.3
12.1
2.3
36.3,
8.4
4.7
122.9
257.7
257.7
New
60.9
10.4
8.0
6.2
15.6
2.3
13.5
0.5
20.9
4.3
61.5
8.4
10.3
173.5
396.3
494.1
Annualized Costs
Unit, $/Ton . Total, MMS/Y
Old
8.50
7.30
7.30
4.10
4.60
19.70
11.70
8.00
11.60
16.00
16.90
-
11.00
16.70
12.58
14.05
Add
2.64
3.12
3.12
7.12
7.12
(.86)
(1.83)
2.64
0.12
(1.33)
10.41
10.41
3.90
42.68
17.13
11.86
New
11.14
10.42
10.42
11.22
11.72
18.84
9.87
10.64
11.72
14.67
27.31
10.41
14.90
59.38
29.71
25.91
Old
5.2
0.7
0.5
0.3
0.9
0.4
1.4
0.1
2.1
0.7
8.9
-
2.0
17.2
40.4
65.4
Add
1.6
0.3
0.2
0.5
1.2
0.0
(0.2)
0.0
0.0
0.0
5.5
1.3
0.7
44.1
55.2
55.2
ar
New
6.8
1.0
0.7
0.8
2.1
0.4
1.2
0.1
2.1
0.7
14.4
1.3
2.7
61.3
95.6
120.6
* Definitions follow Table 9.9, page 9-21
CD.
m
en
-------
I ! I- *••> /
U U <_» i
9.3.3 Improvement in Overall Control Efficiency
to a Plant Minimum of 90 Percent
Eighteen models representing 85 percent of the
total industry capacity would need to be upgraded to
achieve at least 90 percent control efficiency in all
plants. Table 9.8 details the most economical changes
as developed in the model analysis. Only three of the
twenty-two existing control models now achieve 90 per-
cent or better.
9.3.4 Improvement in Overall Control Efficiency
to the Best Demonstrated Technology
Of the twenty-two model control schemes, twenty-
one could be improved, usually by both changing the
primary control equipment and adding secondary control.
One model, IIB-1,7, a VSS Soderberg potline with spray
tower plus wet electrostatic precipitator on the primary
stream and a spray screen secondary control, is now
achieving about as good overall control at 93.7 percent
as present technology can demonstrate. Table 9.9 shows
the projected performances and costs for upgrading 96
percent of the industry to best demonstrated technology.
9.4 Future Control Costs 1975-2000
To estimate the future costs of pollution control
in the primary aluminum industry several premises are
used as developed in earlier parts of this report.
The production capacity used as a base is the me-
dian of the capacity projections made in Section 2,
Figure 2.6. This median capacity projection is illus-
trated in Figure 9.1 which also shows the range of the
capacity estimate.
Control costs are a function of the levels of con-
trol applied to the capacity, and the industry model
costs developed in Section 9.3 have been applied to the
1971 capacity as reference points for the projection of
industry capital investment.
9-18
-------
TABLE 9.8
UPGRADING INDUSTRY MODELS
MINIMUM OVERALL CONTROL EFFICIENCY 907.
Old
Model
No.
IA-1(0)
IA-4(0)
IA-6(0)
IA-10(0)
IA-11(N)
IA-11(0)
IA-12(0)
IA-13(0)
IA-14(0)
IA-15(N)
IA-16(0)
IB-6,9(0)
IC-9
ID(0)
IIA-1
IIA-4
IIA-8
IIIA-9
Upgraded
Segment
Total
Industry
Cap'y
1000
Tons
338
165
601
70
70
163
15
120
6
180
40
100
526
122
60
175
175
1,033
3,959
4,654
Old
OCE, 7.
84.4
82.3
76.0
80.0
32.2
26.7
75.9
82.3
64.0
87.2
74.0
84.6
72.1
0
79.5
79.4
74.6
70.3
70.9
74.2
New
Model
No.*
IB-1,9
IB-4,9
IB-4,9
IB-4,9
IB-5,9
IB-4,9
IB-4,9
IB-13,9
IB-4,9
IB-15,9
IB-4,9
IB-4,9
IB-4,9
IB-4,9
IB-1,7
IB-4,7
IB-1,7
IIIB-1,6
New
OCE, 7.
91.1
89.0
89.0
89.0
91.5
89.0
89.0
86.3
89.0
89.1
89.0
89.0
89.0
89.0
93.8
93.6
93.8
90.4
90.3
90.5
Annual Emission
Old
1,208
672
3,318
322
1,092
2,478
83
563
50
530
239
398
3,375
2,806
283
829
1,022
7,056
26,324
27,648
New
689
417
1,521
177
137
412
38
378
15
451
101
253
1,331
309
86
258
250
2,281
8,852
10,176
Tons F
Decrease
519
^ 255
1,797
145
955
2,336
45
185
35
79
138
145
2,044
2,497
179
571
772
4,775
17,742
17,742
Capital
Unit, $/Ton
Old
60
64
39
53
28
31
58
52
36
49
49
100
48
—
49
40
32
49
47
51
Add
60
60
83
99
82
98
96
49
80
49
96
35
57
106
69
93
69
261
120
102
New
120
124
122
152
110
129
154
101
116
98
95
135
105
106
118
133
101
310
167
153
Investment
Total, MM$
Old
20.3
10.5
23.6
3.7
2.0
5.0
0.9
6.2
0.2
8.8
2.0
10.0
25.2
"
2.9
7.0
5.6
50.6
184.6
236.4
Add
20.2
9.8
49.6
6.9
5.7
16.0
1.4
5.9
0.5
8.9
3.8
3.5
30.0
12.9
4.1
16.3
12.1
296.4
477.0
477.0
New
40.5
20.3
73.2
10.6
7.7
21.0
2.3
12.1
0.7
17.7
5.8
13.5
55.2
12.9
7.0
23.3
17.7
347.0
661.6
713.4
Annuali
Unit, $/Ton
Old
9.30
13.80
8.50
7.30
4.10
4.60
19.70
11.70
8.00
11.60
16.00
29.40
16.90
"
15.60
13.20
11.00
16.70
12.83
14.05
Add
17.98
17.98
21.19
21.67
24.95
25.67
17.70
17.39
21.19
17.39
20.35
4.02
11.79
28.96
23.75
30.49
23.75
74.04
33.64
28.62
New
27.28
31.78
29.69
28.97
29.05
30.27
37.40
29.09
29.19
28.99
36.35
33.42
28.69
28.96
39.35
43.69
34.75
90.74
46. S3
42.67
zed Costs
Total, MM$/Year
Old
3.1
2.3
5.2
0.5
0.3
0.7-
0.4
1.4
0.1
2.1
0.7
2.9
8.9
"
0.9
2.3
1.9
17.2
51.0
65.4
-
Add
6.1
3.0
12.7
1.5
1.8
4.2
0.3
2.1
0.1
3.1
0.8
0.4
6.2
3.5
1.4
5.3
4.2
76,5
1)3.2
133.2
New
9.2
5.3
17.9
2.0
2.1
4.9
0.7
3.5
0.2
5.2
1.3
3.3
15.1
3.5
2.3
7.6
6.2
93,7
184.2
198.6
* Definitions follow Table 9.9, page 9-21
-------
TABLE 9.9
UPGRADING INDUSTRY MODELS
BEST DEMONSTRATED TECHNOLOGY
CD
en
en
CD
Old
ttodel
No.
lA-l(N)
IA-1(0)
IA-4(0)
1A-6(0)
IA-10(N)
IA-10(0)
LA-ll(N)
IA-11(0)
IA-12(0)
IA-13(0)
1A-14(0)
IA-15(N)
IA-16(0)
IB-6,9(0
IC-7
IC-9
ID
IIA-1
IIA-4
IIA-8
IIIA-9
Upgraded
Segment
Total
Industry
Cap'y
1000
Tons
302
338
165
601
90
70
70
163
15
120
6
180
40
100
112
526
122
60
175
175
1,033
4,463
4,654
Old
OCE, %
94.6
84.4
82.3
76.0
89.6
80.0
32.2
26.7
75.9
79.6
64.0
87.2
74.0
84.6
92.5
72.1
0
79.5
79.4
74.6
70.3
73.5
74.2
New
todel
No.*
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1,9
IB-1.9
IB-1,9
IB-1,7
IB-1,9
IB-1,9
IIB-1,7
IIB-4,7
IIB-1,7
IIIB-1 , 6
New
OCE, 7.
96.5
91.1
91.1
91.1
96.5
91.1
96.5
91.1
91.1
91.1
91.1
96.5
71.1
91.1
96.0
91.1
91.1
93.8
93.6
93.8
90.4
92.1
92.2
Annual Emission
Old
375
1,208
672
3,318
215
322
1,092
2,748
83
563
50
530
239
398
192
3,375
2,806
283
829
1,022
7,056
27,376
27,648
New
243
689
338
1,230
72
143
56
334
31
246
12
145
82
205
103
1,077
250
86
258
250
2,281
8,131
Tons F
Decrease
132
519
334
2,088
143
179
1,036
2,414
52
317
38
385
157
193
89
2,298
2,556
197
571
772
4,775
19,245
8,043 19,245
Capital Investment
Unit, $/Ton
Old
60
60
64
39
53
53
28
31
58
52
36
49
49
100
72
48
49
40
32
49
48
51
Add
60
60
119
111
111
111
109
114
143
110
104
116
107
66
69
69
118
69
93
69
261
129
124
New
120
120
183
150
164
164
137
145
201
162
140
165
156
166
141
117
118
118
133
101
310
177
175
Total. MM$
Old
18.1
20.3
10.5
23.6
4.8
3.7
2.0
5.0
0.9
6.2
0.2
8.8
2.0
10.0
8.1
25.2
2.9
7.0
5.6
50.6
215.5
236.4
Add
18.0
20.2
19.6
66.7
10.0
7.8
7.6
18.6
2.1
13.2
0.6
20.9
4.2
6.6
7.7
36.3
14.4
4.1
16.3
12.1
269.4
576.4
576.4
New
36.1
40.5
30.1
90.3
14.8
11.5
9.6
23.6
3.0
19.4
0.8
29.7
6.2
16.6
15.8
61.5
14.4
7.0
23.3
17.7
320.0
791.9
812.8
Annualized Costs
Unit. $/Ton
Old
9.30
9.30
13.80
8.50
7.30
7.30
4.10
4.60
19.20
11.70
8.00
11.60
16.00
29.40
23.80
16.90
15.60
13.20
11-00
16.70
12.86
14.05
Add
17.98
17.98
16.29
19.51
19.99
19.99
23.99
23.99
16.01
15.04
19.51
16.99
15.54
2.34
10.11
10.11
27.28
23.75
30.49
23.75
74.04
31.07
29.8
New
27.28
27.28
30.09
28.01
27.29
27.29
28.09
28.59
35.71
26.74
27.51
28.59
31.54
31.74
33.91
27.01
27.28
39.35
43.69
34.75
90.74
43.83
43.9
Total, MMS/Year
Old
2.8
3.1
2.3
5.2
0.7
0.5
0.3
0.9
0.4
1.4
0.1
2.1
0.7
2.9
2.7
8.9
"
0.9
2.3
2.0
17.2
57.4
65.4
Add
5.4
6.1
2.7
11.9
1.8
1.4
1.7
3.6
0.2
1.8
0.1
3.1
0.6
0.3
1.2
5.5
3.4
1.4
5.3
4.2
76.5
138.7
138.7
New
8.2
9.2
5.0
17.1
2.5
1.9
2.0
4.5
0.6
3.2
0.2
5.2
1.3
3.2
3.9
14.4
3.4
2.3
7.6
6.2
93.7
195.6
204.1
* Definitions follow.
-------
Components of Upgraded Models, Tables 9.6 to 9.9
Model
IA-1
-5
-6
Prim.
FBDS
MC+CFPB-5
MC+ST
Sec.
IB-1,7 FBDS
-1,9 FBDS
-4,9 DESP+SP
-15,9 MC+FBWS
CFPB-3
SS
SS
-5,9 MC+CFPB-5 SS
SS
Description
Fluid Bed Dry Scrubber, Primary only
Multiple Cyclone and Cross Flow
Packed Bed Primary
Multiple Cyclone and Spray Tower on
Primary
Fluid Bed Dry Scrubber Primary, Cross
Flow Packed Bed Secondary
Fluid Bed Dry Scrubber Primary, Spray
Screen Secondary
Dry Electrostatic Precipitator and
Spray Tower Primary, Spray Screen
Secondary
Multiple Cyclone and Cross Flow
Packed Bed Primary, Spray Screen
Secondary
Multiple Cyclone and Floating Bed Wet
Scrubber Primary, Spray Screen
Secondary
IIA-3
DESP+ST
IIB-1,7 ST-fWESP SS
-4,7 MC+VS SS
-8,7 ST SS
Dry Electrostatic Precipitator and
Spray Tower Primary
Spray Tower and Wet Electrostatic
Precipitator Primary, Spray Screen
Secondary
Multiple Cyclone and Venturi Scrubber
Primary, Spray Screen Secondary
Spray Tower Primary, Spray Screen
Secondary
IIIA-2
ST+WESP
IIIB-1,6 CFPB-5
+WESP
SS
Spray Tower and Wet Electrostatic
Precipitator Primary
Cross Flow Packed Bed and Wet Elec-
trostatic Precipitator Primary,
Spray Screen Secondary
9-21
-------
OGG1
FIGURE 9.1
PROJECTED PRODUCTION CAPACITY
U.S. PRIMARY ALUMINUM
1970 t« 2000
3
jo
„
10
Q
_4
_i
I
o I L
i i I
LOW
I I I I I I I I i I I i I I I I
i»ro
!»7S
19M
YEAR
9-22
-------
Two cases are considered for the projection of
future costs.
In Case I the assumption is made that all new
construction, whether it represents replacement of exist-
ing capacity or additional capacity, will be required to
control emissions to the level of best demonstrated tech-
nology, or 96.5 percent overall control efficiency for
these installations.
Two alternative situations are considered under
this assumption: a) that existing plants are not up-
graded before replacement, and b) that existing plants
make the investment required to bring overall emission
control of present capacity to the level of best demon-
strated technology (92.5 percent) industry average. As
capacity replacement occurs, the cumulative investment
and operating costs of the total industry under the
first alternative will approach those estimated for the
limiting case of the second alternative.
In Case II the basic assumption is made that new
construction will apply best demonstrated technology to
primary control only, with no control on secondary efflu-
ent, realizing a 94.6 percent overall control efficiency
on new installations.
Again, two alternative situations applying to
existing plants are considered: c) that no investment
is made in upgrading present control, and d) that in-
vestment is made to raise present control to the level
of best demonstrated technology on primary collection
streams only, with no control of the secondary stream.
This later control level is 85.6 percent for the aggre-
gate of existing plants. Similarly to Case I, as ca-
pacity replacement occurs, the capital and investment
costs of the total industry under the first alternative
will approach those estimated for the limiting case of
the second alternative as the total industry control in-
creases to 94.6 percent.
The projections also assume that most of the new
capacity to be built in the next thirty years will be
designed as prebake potlines with high efficiency col-
lection systems. While it is probable that some new VSS
Soderberg capacity expansion or new plant construction
9-23
-------
6 £6 3
will take place, the proportion of the total is expected
to be small enough that its costs will not have a sig-
nificant effect on this assumption within the accuracy
of the estimation.
Table 9.10 compares the projected costs and con-
trol performances, 1971 and 2000, for Cases I and II and
the two alternative improvement plans. Future unit cap-
ital and unit annualized costs are lower than their cor-
responding values for an upgraded 1971 industry because,
as existing plants are retired they are replaced with
new ones having improved control incorporated in the
initial construction rather than added to or replacing
less efficient systems.
Figure 9.2 shows the two cases of estimated pat-
tern of projected capital investment in emission control
to the year 2000, expressed in 1970 dollars. The fam-
ilies of curves represent the different situations ex-
pected when starting from the two levels of current
investment required to reach the initial industry com-
pliance with the emission control standards noted.
Figure 9.3 shows the corresponding families of
annual operating cost curves, projected to the year 2000,
expressed in 1970 dollars.
Because all projections are based on the median
capacity growth curve, the range of variation of both
actual investment and actual cost can be as much as 30
percent, plus or minus, in the latter years of the period
considered.
The presentation of these cost projections is not
intended as a recommendation for the adoption of the
best demonstrated technology regardless of cost but sim-
ply as an indication of the estimated costs in 1970 dol-
lars, should such a strategy be adopted.
9-24
-------
Table 9.10
PRESENT AND PROJECTED POLLUTION CONTROL
PERFORMANCE AND COSTS (1970 DOLLARS)
Aluminum Capacity, MM Tons
F Effluent, M Tons
Overall Control Efficiency, °L
F Emission, M Tons
Emission Rate, Ton F/M Ton Al
Unit Capital Investment, $/Ton
Total Capital Investment, MM$
Unit Annualized Cost, $/Ton
Total Annual Cost, MM$
Present
4.7
107
74.3
27.5
5.9
51
238
13
62
1971
Best
Prim.
4.7
107
85.6
15.3
3.3
106
494
25
116
2000
Best
Tech.
4.7
107
92.5
8.0
1.7
175
813
43
200
Best
Prim.
22.5
518
94.6
28.0
1.2
60
1350
9
202
Best
Tech.
22.5
518
96.5
18.1
0.8
103
2320
25
553
9-25
-------
0665
FIGURE 9.2
PROJECTED INDUSTRY CAPITAL INVESTMENT
IN EMISSION CONTROL CI970 DOLLARS?
TO THE YEAR 2000
o
o
z
o
CASE I BEST DEMONSTRATED TECHNOLOGY
CASE H BEST PRIMARY CONTROL
3000
2500
2000
1 500
1 000
CASE I
CASE JI
500
PRESENT
INVESTMENT
l r i T T i I 1 i I 1 [ ||1TITIIII1III
3000
2500
2000
15.00
CASE TJ
1 000
CO
500
0
1»70
I I I I I I I I I I I I I I I i I I I
r»/s
1*15
1»»5 2000
YEAR
9-26
-------
FIGURE 9.3
PROJECTED ANNUAL OPERATING COSTS
FOR EMISSION CONTROL (1970 DOLLARS)
TO THE YEAR 2000
0666
<
ui
>-
O
o
O
K
CASE I BEST DEMONSTRATED TECHNOLOGY
CASE H BEST PRIMARY CONTROL
700
600
500 _
400 _
300
Z CASEI200L^
CASEJJ
T970
COST
100
700
600
500
400
300
200
100
1970
2000
9-27
-------
CG67
Table of Contents
Section 10
10.0 Potential Fields for Research and
Development in Pollution Abatement
10.1 Survey of Problems
1. Carbon Plant Effluents
2. Anode Bake Plants
3. Anode Rodding
4. Potline Materials Handling
5. Potroom Effluents
6. Hot Metal Operations
7. Power Generation
10.2 Priorities
10.3 Priority Problems and Possible Solutions
1. Cell Effluents
2. Anode Baking Effluents
3. Cast House Fluxing
4. Paste Mixing
10.4 Research and Development Subject Areas
1. Measurement and Sampling
2. Improved Characterization of Effluents
3. Reduction in Effluent Generation
4. Elimination of Fluorides from Aluminum
Production
5. Improved Effluent Collection
6. Improved Pollutant Removal
10.5 Suggested Research and Development Projects
10.6 Suggested Research and Development Program
-------
10.0 Potential Fields for Research and
Development in Pollution Abatement
An objective of this study is to determine where
research and development may be undertaken to achieve
the desired control of emissions to the air from the
operations of the primary aluminum industry.
To accomplish that objective, this section sum-
marizes the effluent control problems which exist in the
reduction plant processing areas, and evaluates them in
terms of relative factors of level of uncontrolled emis-
sions, knowledge of emission problems, extent of avail-
able control, and cost impact of control. Priorities
and programs are suggested for research and development
effort.
10.1 Survey of Problems
The problems involved with the separate effluent
sources are briefly summarized below:
1. Carbon Plant Effluents
Carbon plant effluents are, with one excep-
tion, large quantities of carbon dust particulates
generated by handling, crushing, grinding, sizing and
mixing fractions of coke used for the preparation of
reduction cell electrodes. Universal practice is to
collect these dusts at point of origin and recover them
in dry separation equipment. The problem has been re-
duced to one of in-plant housekeeping. Little quanti-
tative information is available on the dust generation,
and control is imperative to obtain a minimum of emis-
sion from equipment in normal operating condition.
The use of softened or melted pitch as a
paste binder for the carbon particles of the electrodes
results in the evolution of an unknown, but minor, quan-
tity of volatile hydrocarbon during processing. The
generation is localized, and is considered to be small
enough so that normal practice is to exhaust these fumes
to atmosphere. The fumes can be largely removed from
these exhausts by application of available wet scrubbing
technology, although treatment is not usually given to
them.
10-1
-------
2. Anode Bake Plants
Anode bake plants are a component of nearly
all prebake smelters and are associated with approx-
imately 50 percent of the United States production capac-
ity. Minor sources of carbon dust to the atmosphere lie
with the mechanical handling of the anode packing mate-
rials in the bake pits, and in the blast cleaning of
finished anodes before rodding. Both are controlled to
empirical limits with local collection and separation
systems, the losses constituting a housekeeping, rather
than a pollutant problem.
Some volatile hydrocarbons may be released
during anode pressing operations, but the amounts are
minor and they are exhausted to atmosphere. As with the
paste fumes noted above, they can largely be removed by
available technology.
The gases from baking furnaces contain, in
addition to combustion products, hydrocarbons and limited
amounts of SC>2 released from the anode paste by baking,
small amounts of fluorides originating in recycled scrap
anodes, and smoke. The volume of effluent is large. The
amount of condensible hydrocarbons is great enough to
constitute a major removal problem. Visible smoke con-
stitutes the major abatement problem and a considerable
effort is being directed at controlling it.
3. Anode Rodding
Effluents generated in the anode rodding op-
erations are minor in quantity and are locally controlled
through collection and removal by available technology.
They consist of dusts arising from spent butt cleaning
and crushing operations, dust and metallics caused by
blast cleaning of rods, and fumes evolved in melting and
pouring thimbles for rod resetting. Control does not
present a major problem.
4. Potline Materials Handling
Bulk handling of alumina and of bath materials
generates particulate dusts at transfer points which, if
uncontrolled, represent both material losses and indus-
trial hygiene problems. Control is exercised by enclos-
ure, collection, and dry recovery, usually in baghouse
10-2
-------
0-8-70
systems. No quantitative information is available on
emission losses, but good control can be maintained with
existing technology.
5. Potroom Effluents
Problem
Reduction cell effluents include fluoride and
nonfluoride particulates, gaseous fluorides, CO and CC>2»
SC>21 and varying amounts of volatilized hydrocarbons.
If uncontrolled, these effluents may result in damaging
levels of fluorides and significant levels of other pol-
lutants. Much attention has been given to the fluoride
aspects of the control problem, and only incidental at-
tention to the emission of SC>2 and CO, on which there is
little information.
Level of Control
For prebake potlines, a high level of control
of total particulates and gaseous fluorides in collected
effluent streams is achievable with currently available
technology. The larger portion of the total effluent of
prebake cells may be collected by local hooding at the
cells; emissions may be controlled by dry adsorption of
gaseous fluorides on alumina and subsequent essentially
complete removal of particulates from the gas stream.
With optimum primary collection efficiency, treatment of
the remaining cell effluent in a secondary collection
and control system in much larger volume contributes rel-
atively little (2 percent) to overall control efficien-
cies.
A control level of 95 percent is obtainable with
primary control only on effluents from the prebake cells
representing some 15 percent of total industry capacity
whose design permits installation of nearly completely
effective cell hooding. On other prebake cells (some 50
percent of industry capacity) hooding efficiency is low-
er, and the attainment of overall emission control in
the 90-95 percent range would require either improved
collection efficiency or the additional use of secondary
collection and effluent treatment.
10-3
-------
n r* 1
uc «
VSS cells, which represent some 14 percent of
industry capacity, present air pollution abatement prob-
lems similar to those of prebake cells, since burner
combustion of tar fogs emitted from the VSS cells prac-
tically eliminates these constituents. The primary
collection efficiency is lower than with the fully hood-
ed prebake cells, and secondary effluent treatment is
required to reach an overall control efficiency range
comparable to the best control performance achievable
with prebakes.
HSS cells differ in their effluent control
problems from the other types of cells. Hydrocarbon
fumes, not removable by direct combustion as in VSS
cells, condense to tars which interfere with high effi-
ciency dry control systems. Unless hooding and opera-
tional techniques can be modified to achieve acceptable
levels of primary collection, secondary removal systems
would be required for high levels of overall emission
control. Wet devices can be used which are capable of
obtaining good efficiency for the 20 percent of indus-
try capacity represented by HSS installation.
Cost of Control
Present potline emission control is estimated
to account for about 6 percent of total plant investment
of $800-$900 per annual capacity ton with a range of
4-14 percent for this figure in individual installations.
The net annualized cost of current emission control is
3-5 percent of a nominal production cost of $390 per ton,
with a range of 1-9 percent in individual installations.
The emission control achieved with present installations
is of the order of 73-80 percent.
The effect on capital and production costs of
upgrading control in existing plants to best demonstrated
control technology in new plants, is shown in Table 9.5.
It is evident that the impact of potline emis-
sion control on capital and operating costs is substan-
tial, and is significantly greater on HSS installations
than on prebake or VSS plants. Capital cost impact is
greater in upgrading existing plants, where conversions
are involved, than on new plants constructed with highly
efficient emission control systems. Costs for individual
10-4
-------
OB72
plants vary widely, and can be a factor in their compet-
itive position with respect to other domestic producers,
as well as to foreign producers which may have lighter
control cost burdens.
6. Hot Metal Operations
Removal of metal from the cells is carried out
in closed systems with little evolution of pollutant
fumes in tapping, ladle skimming, hot metal transfer to
the cast house, retention in holding furnaces, alloying
to adjust metal composition, or casting.
Fluxing of hot metal to remove impurities car-
ried over with the cell metal produces copious, but in-
termittent, quantities of gas and fumes which vary in
composition with fluxing practice and are potential air
pollutants. Gases include HCl, occasional chlorine,
nitrogen and inert carriers. Fumes are primarily alumi-
num chloride, which may hydrolyze to alumina and hydro-
chloric acid. Little or no specific information is
available on amounts and composition of the effluents,
which are collected in hooding systems and exhausted
through empirically designed fume removal equipment.
7. Power Generation
Although aluminum smelters consume very large
blocks of base load electric power, few generate it in
company-owned power plants. One company uses gas tur-
bines and gas-fired boiler steam turbines to generate
power and a few use gas-fueled piston engines. Gas is
a clean burning fuel, essentially free from sulfur, but
nitrogen oxides and CO may present pollution problems.
Little has been done to control these.
10.2 Priorities
An assessment of air pollution abatement problems
facing the primary aluminum industry suggests that the
sources of emissions derive from five process areas which
can be ordered in rank of priority as regards the desir-
ability of developing better control. Factors influencing
the selection of priorities include the levels of cur-
rently uncontrolled emission, the extent to which control
techniques exist, costs for achieving control, and the
10-5
-------
n,c "7 o
u o T. o
state of knowledge of the problem. Considering these
factors, three pollutant sources present higher prior-
ity problems than the other two:
1. Potline effluents from side-worked prebake,
VSS Soderberg and HSS Soderberg cells,
2. Anode bake furnace effluents,
3. Potline effluents from center-worked prebake
cells.
Other emission sources are not considered to pre-
sent the same order of magnitude of potential pollution.
However, investigation of two of them with respect to
their character and composition could be included in a
second order priority. These are:
4. Cast House fluxing,
5. Paste mixing.
10.3 Priority Problems and Possible Solutions
Priority 1
Cell Effluents
Arranged in the approximate order in which posi-
tive results appear to be most likely to be realized,
the investigative areas concerning improved pollution
emission control over cell effluents are:
a) Improvement in Hooding and Collection Systems
As explained in Section 8.1.3, the efficiency
with which cell effluents are collected at the cells
and ducted to removal equipment has a strong impact on
the overall control efficiency of most potline control
schemes. Reported primary collection efficiencies range
from about 70 percent to higher than 96 percent, and
perhaps the most dollar-effective way to improve pollu-
tion abatement within the industry would be to develop
improved collection technology at the poorer installa-
tions. Hood designs which minimize openings, ventila-
tion systems which permit greatly increased gas flow
10-6
-------
0671*
from cells which have hoods open, and potroom supervi-
sion which assures that hoods are kept in place and
in good repair, all contribute to high collection ef-
ficiency.
At present,hoods are not used on VSS Soderberg
cells; only effluents entering the skirt are treated in
primary control systems. cell enclosure in addition to
the skirt would improve collection efficiency.
Soderberg anode fumes normally pass directly
into the potroom. Although the hydrocarbons might com-
plicate removal equipment, the collection and treatment
of these fumes would improve overall control and would
alleviate the industrial hygiene problems of tar fog in
the potroom.
b) Reduction of Fluoride Effluents
from Hall-Heroult Cells
Short of eliminating fluorides from the smelt-
ing process, measures to reduce the quantity of efflu-
ents from the cryolite bath of Hall-Heroult cells may
prove effective and economical ways to abate the air
pollution problem.
"These measures include experimental programs
to better define and quantify the electrochemical in-
terreaction in the cell bath during electrolysis, and
development work on the mechanical aspects of cell opera-
tions which result directly or indirectly in the genera-
tion of cell effluents.
They also include investigations which would
provide a sound basis for the extrapolation of the re-
sults of small-scale cell experiments on the effect of
changes in bath ratio, alumina concentration, and bath
temperature on fluoride effluent rates.
c) Development of Improved Operating Technology
Improvement in the physical operations of cell
feeding, working, and tapping which, because they result
in the breaking of the cell crust, result in an oppor-
tunity for effluents to escape from the bath, is an area
for investigative development.
10-7
-------
0675
d) Composition of Cell Effluents
While the fluoride component of cell effluents
is generally well defined, less has been established
concerning the other gaseous components which represent
potential pollutants, such as S02» CO, and volatilized
hydrocarbons. The size distribution, composition, and
quantity of solid particulates is not well established,
and have an effect on the design and selection of appro-
priate removal equipment.
Information on the total composition of cell
effluents is incomplete, particularly with respect to
potential pollutants other than fluorides. Data on char-
acter, composition, and quantity of hydrocarbons is
sparse, as is information on SC>2 and C02- The amount of
hydrolyzation occurring in practice and the resultant
state transfer of fluorides affecting removal mechanisms
is not satisfactorily known. Investigation into these
areas would improve the knowledge of the conditions un-
der which pollutant removal is to be effected.
e) Particle Size Analysis
The prediction of removal efficiency for many
kinds of particulate control equipment depends on know-
ledge of the particle size distribution of the pollutant
entering the equipment. Measurement of effective par-
ticle diameter and the weight distributions of micron
and submicron dusts is uncertain and subject to difficul-
ties in obtaining reproducibility of results. A well-
supported project aimed at developing accurate methods
of determining particle size distributions in aluminum
potline and bake furnace effluents, particularly in the
submicron ranges, and at correlating these distributions
with particulate removal equipment performance would pro-
vide valuable information by which equipment design might
be refined and improved.
f) Particulate Agglomeration
There is strong evidence that some particulates
in potline effluent streams tend to agglomerate during
their passage from cell to removal equipment. Few quan-
titative data exist and little is known about the factors
favoring agglomeration. A substantial experimental re-
10-8
-------
G676
search project could be performed at several kinds of
potlines to develop information which might assist the
design of more efficient particulate removal equipment.
g) Particulate Solubilities
The capacity of wet scrubbers to remove gas-
eous fluorides may become limited by the concentration
of the fluorine ion, and this in turn may become con-
trolled by the solubilities of fluoride particulates.
A limited effort at determining particulate solubilities
at various temperatures and in various liquors might
provide useful data.
h) Hydrocarbon Volatilization,
Condensation and Oxidation
Basic research into the behavior and charac-
teristics of various hydrocarbons used in the making of
anodes and cathodes, particularly Soderberg type anodes,
could yield insight into better ways to control the
gases and fumes evolved during baking of anodes, both in
bake plants and in Soderberg cells. Better information
on volatilization rates, condensation temperatures and
characteristics, and on oxidation rates could help in
formulating anode paste with improved properties as re-
gards pollution control.
i) HF and Scrubber Liquor Solutions
Maximum removal efficiency of wet scrubbers for
gaseous fluorides is sometimes limited by the vapor pres-
sure of HF in the scrubber solution. This occurs when
the scrubber liquor is at a neutral or acid pH. Data are
sparse on vapor pressures and solubilities of HF in water,
lime water, and caustic solutions at relatively low con-
centrations .
j) Capture of Submicron Particulates
Although filters and electrostatic precipita-
tors may be designed to capture submicron particulates
efficiently, wet scrubbers, generally low in submicron
removal efficiency, may be better suited to some kinds
of effluents, such as those from HSS potlines and bake
furnaces. Improvement in cost-effectiveness in the sub-
micron ranges would be very desirable.
10-9
-------
0677
k) Removal Efficiency on Very Dilute Streams
Pollutant concentrations, both particulate
and gaseous, in secondary gas streams, especially when
primary cell collection efficiency is high, may be only
a hundredth as great as concentration in the primary
streams - equivalent to the discharge from a 99 percent
efficient removal device on the primary. Present tech-
nology does not offer equipment at reasonable cost which
is capable of high removal efficiency on these dilute
streams.
Priority 2
Anode Baking Effluents
Development work is needed in the control of the
anode baking effluents.
The composition of the baking plant gases is not
well defined, and even the major potential pollutants
are only incompletely identified at most plants. Because
of the considerable quantities of condensible hydrocar-
bon fumes in the effluents, satisfactory control is dif-
ficult and may present a problem. Large volumes of gas-
es complicate sampling and control.
In general the problem areas are similar to those
of the HSS cell effluents; incomplete information on
character and composition of both gaseous and particu-
late components, control of effluent generation and the
physical state of the particulates contained in the
effluents.
Priority 3
Cast House Fluxing
A lower priority is assigned to the investigation
of effluent control applied to gases and fumes from
fluxing operations in cast houses. While emission con-
trol might be improved by a more quantitative understand-
ing of the effluent composition, specific removal of
chlorine, and greater efficiency in elemination of sub-
micron particulates from the emissions, the problem is
less important than that of either cell room or anode
10-10
-------
6678
baking effluents. The potential pollution and the cost
impact of pollution control are relatively minor.
Priority 4
Paste Mixing
Volatile hydrocarbon fumes released during hot
processing of pitch binders represent a minor, but poten-
tial air pollutant of undetermined effect. Little, if
any, emission control is practiced. The problem needs
to be better defined with respect both to the quantita-
tive and qualitative aspect of the effluent and to the
applicability of existing control devices. Cost impact
of problem solution is relatively small.
10.4 Research and Development Subject Areas
The solutions of the major air pollution problems
in the industry identified earlier in this section re-
quire research and development work in several general
subject areas to generate more and better information.
These areas are:
Measurement and sampling
Improved Characterization of Effluents
Reduction in Effluent Generation
Elimination of Fluoride Pollutants
Improved Effluent Collection
Improved Pollutant Removal
The following section presents a general discus-
sion of these areas in relation to the major pollutant
problems and suggests specific research and development
projects concerned with them.
1. Measurement and Sampling
One of the problems encountered in this study
has been the difficulty in obtaining consistent or com-
plete data concerning the volumes and composition of
the effluent streams. The data are highly variable; the
sampling and analytical methods are not uniform, and
there are limited data on a number of effluent parameters,
10-11
-------
Relatively low volume gas streams such as those
in primary cell collection systems are moved at moderate
velocities and are sufficiently uniform that measurement
and sampling can be accomplished with reasonable confi-
dence.
The effluent streams of secondary collection
systems and anode baking gases may be orders of magni-
tude greater in volume, much lower in velocity, and may
in both cases be subject to wide variation in pollutant
loading over considerable time spans and intervals.
Unpowered flows of secondary potroom effluents
pass through large cross-sectional areas at very low
velocities, which are measurable only with sensitive
anemometers; air flow distribution can be erratic, com-
pounding the difficulty of utilizing a sufficient number
of sampling points and sampling determinations on which
to confidently base extrapolations to total air flows.
Powered exhausts of air flow, as in controlled
secondary potroom systems, reduce the flow measurement
problem and more accurate data are obtainable, but still
require use of a multiplicity of sampling points to
gauge the results of non-uniform flows in the system.
Representative sampling of the large volume
flows is difficult. Very large samples are required to
obtain accurate estimation of the pollutant loadings,
per unit of volume, which are both low and variable.
The confidence which can be placed in data ob-
tained from this type of gas volume sampling is not
great, and efforts are recommended to improve the tech-
niques and practice. Project 1, Development of Improved
Techniques for Gas Volume Sampling, is suggested to im-
plement this recommendation.
Sampling of pollutants which are carried in
the effluent streams is accomplished by passing the sam-
ple through a train which separates the constituents for
analysis. Difficulties leading to analytical inaccura-
cies arise in the simultaneous collection of tar fumes
with other particulates, and in adsorption of gaseous
fluorides on particulates during the sample collection,
among others. Project 2, Improved Techniques in Pollu-
tant Sampling of Effluent Streams, is suggested for
10-12
-------
0680
research and development effort to improve the reliabil-
ity of pollutant sampling.
2. Improved Characterization of Effluents
The attention in the aluminum industry with
respect to air pollution has been largely focused on
fluorides and particulates in cell effluents; only minor
effort has been made to evaluate the occurrence of other
potential pollutants in the various plant effluents.
Reliable data on the amount, size, and composition of
particulates, and amount and composition of gaseous com-
ponents is less than complete for primary cell effluents,
and very sparse for other effluent flows.
Identification of potential pollutants such
as S02/ CO, volatile hydrocarbons is necessary and de-
termination of these pollutant amounts is needed to
adequately evaluate their contribution to the problem.
Reliable information concerning particulate size dis-
tributions and better data on relative proportions of
particulate composition is basic to improvement in the
emission controls applied.
This information acquisition is desirable on
all effluent streams identified, but particularly so on
the large volume effluents from primary and secondary
cell collection and anode baking, and on the effluents
from paste processing and cast house fluxing. Such
activity will require the cooperative participation by
various aluminum companies representing the full range
of processing operations. Given access to the process-
ing equipment itself, investigators would be able to
obtain the necessary analyses.
Project 3, Characterization of Emissions from
Potrooms, and Project 4, Characterization of Effluents
from Anode Baking, are examples of high priority under-
takings of this type.
3. Reduction in Effluent Generation
Short of elimination of potential pollutants
from the effluent stream, development of measures to re-
duce their generation in the processing segment can re-
sult in reducing the magnitude of the air pollution
problem.
10-13
-------
0681
As noted earlier in this report, some experi-
mental research results have been published on the
various factors influencing the quantities and charac-
teristics of aluminum reduction cell effluents. The
major aluminum producers have also investigated various
aspects of the process affecting effluents as a part of
basic research programs designed to improve the econom-
ics and efficiency of metal production, and they have
developed understandings of the cell operating condi-
tions which yield high current efficiency and ease of
operation. Conditions which minimize pollutant effluent
may not be compatible with stable and efficient cell op-
eration. Although it has been shown that changes in
cell operating conditions can have marked influence on
effluent generation, details of cell operating param-
eters are among the most jealously guarded secrets in
the highly competitive aluminum smelting industry and
there is little likelihood that the foreseeable future
will bring much disclosure in this field. The reduction
of effluent generation by adjustments in cell operating
conditions is an aim of most producers but it probably
will not prove acceptable as a subject for publicly dis-
closed research and development because the work would
be too closely associated with technology which affects
directly the profitability of aluminum production.
Aside from cell operating parameters, three
areas of investigation with the objective of reducing
pollutants carried in potroom effluents include:
a) Provision to keep VSS Soderberg burners ig-
nited;
b) Provision to install new prebake anodes with
a minimum of pollutant escape;
c) Improved handling of cell feed materials in
the potroom.
Project 5, Improvements in Potroom Technology to Reduce
Pollutants, is suggested as an approach to research and
development activity in this area.
Anode baking is another important effluent
source to which attention should be directed to reduce
the amounts of potential pollutants.
10-14
-------
CS82
SC>2 in this effluent originates with the sul-
fur content of the paste which is released during bak-
ing, and of the fuel used for furnace firing. Fluorides
originate with bath materials associated with scrap
butt anodes returned for recycling. Tar fumes originate
with the pitch used for the anode binder. Each of these
components is subject to reduction, which may be ef-
fected by optimization of materials choice and prepara-
tion.
Project 6, Reduction in Anode Baking Pollu-
tants, is suggested as a possible investigation program
in this area.
4. Elimination of Fluorides
from Aluminum Production
Virtually all commercial production of alumi-
num is by the Hall-Heroult electrolytic reduction of
alumina, characterized by the fact that the electrolyte
is a mixture of molten fluorine bearing salts,primarily
cryolite, operating at nearly 1000°C. Airborne flu-
orides, the principal pollutants from the aluminum in-
dustry, are derived from this electrolyte and elimina-
tion of fluorides from aluminum production would relieve
the major part of the air pollution of the industry.
Although no significant departure from the
original Hall-Heroult process for the reduction of alu-
mina to aluminum has been achieved on an industrial
scale, much development work has been performed in the
past thirty years on alternative processes.
Among the avenues which have been explored by
major aluminum producers have been the direct reduction
of aluminum oxide by carbon in electric furnaces, the
reduction of alumina with manganese chloride to form
aluminum chloride, followed by conversion to aluminum
metal with manganese, electrolysis of fused aluminum
chloride, and electrolysis of aluminum sulfides. The
direct reduction processes produce aluminum-silicon al-
loys requiring additional processing to recover aluminum
metal.
The low cost of producing pure alumina by the
Bayer process, the development of improved efficiencies
10-15
-------
068
for the Hall-Heroult process, and the existence of many
older, fully depreciated plants using the Hall-Heroult
process make it difficult to develop an economical com-
petitive process. However, it may be that, taking the
pollution aspects of the Hall-Heroult process into con-
sideration, another process which does not evolve flu-
orine effluents may become attractive. An acceptable
new process would have to be one that produces little
toxic effluents or one where the toxic effluents could
be easily treated and removed.
Although replacement of the Hall-Heroult proc-
ess by another might ease air pollution problems, this
is not recommended as a proposed research and develop-
ment project because no potentially successful process
has been identified.
5. Improved Effluent Collection
Control of pollutant emissions is contingent
on the collection, as well as the treatment, of the
effluents concerned, and in most process areas effluent
collection is good, or can be made so by application of
existing suitable design techniques. The process area
in which this statement is not entirely applicable is
that of the potlines, where the results of improvement
can contribute significantly to improvement in control.
There is current need for the development of
better hoods and collection systems, particularly for
Soderberg cells, which will operate at higher efficiency
without seriously hampering the primary business of pot-
room operations, that of producing aluminum.
It is expected that development programs de-
signed to improve cell collection would require close
cooperation with producers to achieve practically ap-
plicable results. The interrelation of hooding designs
with potline operability imposes restrictions on hooding
modifications.
Project 7, Improvements in Cell Effluent Col-
lection, deals with this problem, which is placed high
on a priority list because of its potential cost effec-
tiveness as applied to a large proportion of the indus-
try.
10-16
-------
6. Improved Pollutant Removal
The problem of removal of specific pollutants
from effluents are common to several of the process seg-
ments of an aluminum plant.
Improvement of HF removal over the high level
(98-99 percent) obtainable with presently available
mechanisms and equipment for wet scrubbing or dry absorp-
tion appears to be in the area of fundamental research,
rather than development. The priority which might be
placed on such effort is low when the potential gain to
control is considered. It is therefore disregarded.
Essentially complete removal of particulates
from gas streams also is accomplished with combinations
of existing equipment when the particulates are dry and
non-sticky. Separation of coarse (plus 5 micron) par-
ticulates can be made in centrifugal separators such as
cyclones, and the finer particles escaping cyclone sepa-
ration can be removed in baghouses or electrostatic
precipitators.
When a portion of the particulate loading is
viscid, as may be the case with tar fogs or hydrocarbon
condensates, interference with the operation of dry re-
moval equipment results from fouling of separation sur-
faces with tar agglomeration, and resort must be made
to some form of wet treatment. The efficiency of wet
scrubbing in particulate removal, a liquid phase absorp-
tion mechanism, falls off rapidly with decreasing par-
ticle size at constant energy input, and this effect is
compounded by the hydrophobic character of the tars.
Project 8, Improvement in Removal of Hydro-
carbon Fumes and Particulates from Effluent Streams, is
suggested as a specific development program in this prob-
lem area.
Removal of SC>2 from aluminum plant effluents
presents a difficult problem because of the low concen-
trations in the gas streams, an order of magnitude low-
er than occurs in other industrial effluents such as
fossil fuel combustion products.
10-17
-------
0685
Removal of HCl and gaseous chlorides from cast
house fluxing effluents has not been definitely in-
vestigated, so far as is known, but needs a relatively
low priority of attention.
10.5 Suggested Research and Development Projects
A number of specific research and development
projects have been identified which should be consid-
ered in a program to accumulate information leading to
solution of the priority pollution abatement problems
in the primary aluminum industry.
Project 1 - Development of Improved Techniques
of Gas Volume Sampling
Scope - Large volume, low velocity effluent streams.
Purpose - To improve accuracy of quantitative determi-
nation for large effluent volumes.
Impact - More precise definition of pollutant levels
for abatement purposes.
Requirements - In-plant development of specific stand-
ard procedures for reproducibility and accuracy of gas
sampling in high volume systems; improvement in low
velocity gas flow determinations.
10-18
-------
0686
Project 2 - Improved Techniques in Pollutant
Sampling of Effluent Streams
Scope - Primary and secondary collected cell effluents,
anode baking effluents, anode paste effluents, cast
house fluxing effluents.
Purpose - To improve discrimination among samples taken
for total solids, soluble and nonsoluble fluoride par-
ticulates, hydrocarbon particulates, gaseous fluorides,
chlorides, S02 and CO.
Impact - More precise definition of pollutant levels
for abatement purposes.
Requirements - Development of specific standard proce-
dures for reproducible and accurate simultaneous or
sequential separation of potential pollutant components
carried in effluent gas streams.
Project 3 - Characterization of Pollutants in
Cell Emissions
Scope - Total solids, soluble and nonsoluble particu-
late fluorides, hydrocarbons, gaseous fluoride, SO2/
CO, N02 and CO2 in primary and secondary collected cell
effluents.
Purpose - Quantitative determination of composition
and character for potential pollutants in cell efflu-
ents .
Impact - More precise definition of potential pollutant
levels for abatement control purposes.
Requirements - Continuous monitoring of cell effluents
from operating aluminum potlines and analysis, includ-
ing size distribution of particulates, of potential
pollutants, using techniques developed in Projects 1
and 2.
10-19
-------
0687
Project 4 - Characterization of Pollutants
in Anode Baking
Scope - Total solids, particulate and non-particulate
hydrocarbons and fluorides, S02» CO, N02/ and C02 in
anode baking effluents.
Purpose - Quantitative determination of composition
and character for potential pollutants in anode baking
effluents.
Impact - More precise definition of potential pollu-
tant levels for abatement control purposes.
Requirements - Continuous monitoring of anode baking
effluents from operating bake plants and analysis, in-
cluding size distribution of particulates, of poten-
tial pollutants using techniques developed in Projects
1 and 2.
Project 5 - Improvements in Potroom Technology
to Reduce Pollutants
Scope - Fluorine evolution from Hall-Heroult elec-
trolysis.
Purpose - Improved correlation between the effect of
cell operating variables and gaseous and particulate
fluoride generation in cell effluents.
Impact - Reduction of pollutants requiring abatement
control.
Requirements - Experimental and plant scale develop-
ment of devices and operating techniques which will
reduce the quantities of pollutants entering control
systems.
10-20
-------
U688
Pro-ject 6 - Reduction in Anode Baking
Plant Pollutants
Scope - Fluoride, SC>2» and hydrocarbon evolution from
anode baking operations.
Purpose - Reduction in pollutant generation.
Impact - Improvement in pollutant control problem,
reduction of fouling problem in tar removal equipment.
Requirements - Development of improved paste formula-
tions, control of sulfur in anode materials, use of
binders other than pitch.
Project 7 - Improvement in Effluent
Collection at Cells
Scope - Cell collection system designs.
Purpose - Increase of primary cell collection effi-
ciency.
Impact - Improved overall control efficiency.
Requirements - Development of improved cell hooding
designs and primary collection of cell effluents at
point of origin without interference with cell
operations.
10-21
-------
0689
Project 8 - Improvement in Removal of
Hydrocarbon Fumes and
Particulates from Effluent
Streams
Scope - Treatment of effluents from Soderberg cells,
Soderberg anodes, anode baking, and anode paste op-
erations .
Purpose - Reduction in the interference with the re-
moval of other particulates from gas streams.
Impact - Reduction in fouling problem in ducting and
separation equipment; lower maintenance expense, im-
provement in fine particulate removal.
Requirements - Development of better methods for
agglomeration and removal of viscid particulates from
gas streams.
10-22
-------
G63Q
10. 6 Suggested Research and Development Program
T*he projects noted above could be organized
into an integrated research and development program
to be undertaken by or under the sponsorship of the
appropriate air pollution abatement office of the
government in cooperation with the producing industry.
In planning such a program and obtaining in-
dustry cooperation, account must be taken of the keen-
ly competitive relationship among the relatively small
numbers of corporate producers. Historically, there
has been a minimum of "know-how" exchange between pro-
ducers, designed to protect operating techniques which
are regarded as economically advantageous trade secrets,
VJhile these tight information policies are relaxing
somewhat with the advent of new producers, the research
and development climate is not conducive to participa-
tion by outside groups. The projects suggested for
the program require access to operations at the least,
and for some the direct participation of the operating
staffs in the project is highly desirable.
In this situation, careful planning is required
to set up acceptable and effective conditions of data
access and discrimination which will guard the confi-
dentiality and use of plant information.
10-23
------- |