EPA-600/2-76-194
July 1976
Environmental Protection Technology Series
ELIMINATION OF WATER POLLUTION BY
RECYCLING CEMENT PLANT KILN DUST
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2, Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-194
July 1976
ELIMINATION OF WATER POLLUTION BY
RECYCLING CEMENT PLANT KILN DUST
by
N. R. Greening
F. M. Miller
C. H. Weise
H. Nagao
Portland Cement Association
Skokie, Illinois 60076
Grant No. 802196
Project Officer
Donald L. Wilson
Industrial Pollution Control Division
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Research
Laboratory, Cincinnati, U.S. Environmental Protection Agency, and ap-
proved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the U.S. Environmental
Protection Agency, nor does mention of trade names or commercial pro-
ducts constitute endorsement or recommendation for use.
11
-------
FOREWORD
When energy and material resources are extracted, processed,
converted, and used, the related pollutional impacts on our environment
and even on our health often require that new and increasingly more
efficient pollution control methods be used. The Industrial Environ-
mental Research Laboratory - Cincinnati (lERL-Ci) assists in developing
and demonstrating new and improved methodologies that will meet these
needs both efficiently and economically.
As part of these activities, efforts of this study have included
determination of the feasibility of separation of cement plant kiln dust
into fractions which are alkali-rich and alkali-poor with various pyro-
processing techniques. These have included fluidized bed and flame-
spray methods. The work was undertaken in an effort to avoid a major
disposal problem. Cement plant kiln dust must often be discarded,
because it contains unacceptably high levels of alkalies and sulfur,
which can adversely affect the cement product or disturb kiln opera-
tional continuity. If these deleterious constituents can be removed,
the beneficiated dust would often be recyclable. The much lower volume
of high-alkali dust could represent a potentially valuable fertilizer
supplement. The results of the work were encouraging in that such
separation was achieved: however, further work to optimize the system
and scale it up for plant tests is indicated. The study should be of
interest primarily to cement manufacturers, but also to possible
consumers of the high-alkali dust fraction. In an allied study, the
practicality of burning kiln dust directly to cement clinker and of
intergrinding the product with normal portland cement clinker has been
studied, and reported to EPA.
For further information on this subject, contact the Industrial
Pollution Control Division, Metals and Inorganic Chemicals Branch.
David G. Stephan
Director
Industrial Environmental Research Laboratory-Cincinnati
111
-------
ABSTRACT
The efforts of this study have included determination of the feasibility
of separation of cement plant kiln dust, into fractions which are
alkali-rich and alkali-poor, with various pyroprocessing techniques.
These have included fluidized bed and flame-spray methods.
The study included the investigation of the effect of varying a number
of process parameters on the achievement of four goals:
1) Effective feeding of the kiln dust raw material
2) Maintenance of flame stability and of adequate temperature to
achieve alkali volatilization
3) Achievement of separation of the two aforementioned fractions
until collection was complete
4) Efficient collection of the two kiln dust fractions.
The parameters varied were: The feeding system and fluidizing arrange-
ment, the portion of the system designed for alkali entrapment, the dust
collection mechanism, the temperature of the flame and collection
system, and the collecting medium itself.
Limited success was achieved in meeting these objectives.
Although the first two objectives were generally met, there seemed to be
a degree of mutual exclusivity in the third and fourth objectives in
some cases. However, optimization of operational parameters resulted in
simultaneous achievement of all four goals.
A theoretical study of the operative chemical parameters was made, and
suggestions for achievement of these goals in other ways have been
prepared. In light of the energy shortage which is a problem now and
for the foreseeable future, some of these suggestions may prove ulti-
mately more practical than separation by pyroprocessing.
IV
-------
CONTENTS
ABSTRACT iv
List of Figures vi
List of Tables vii
Introduction 1
Acquisition of Existing Information 2
Conclusions 3
Recommendations 5
Objectives of the Research Program 7
Theoretical Considerations 7
Experimental Investigations 8
Fluid Bed Tests 8
Muffle Furnace Experiments 10
Thermal Balance Experiments 12
Flame-Spray Experiments 12
Further Characterization of Kiln Dust 15
Analysis of Alkali Condensate & Recovered Dust 20
Chemical Analysis 22
Discussion of Required 31
Modifications
Operation & Product Measurements 33
Further Improvements 39
Test Run Results 42
Kiln Dust Properties 48
Installation of Metallic Mesh Filter 52
Experimental Results 53
References 60
-------
FIGURES
No. Page
1 Fluidized Bed Furnace 9
2 Schematic of SWIRL Preheater System 18
3 Flame Spray Burner with Cyclone and Feed Supply 19
4 Flame Spray System 21
5 Flame Spray Dust Collection Assembly 29
6 Exhaust Stack Arrangements 30
7 Temperature Profile of Duct Cross Section -
Thermocouple No. 4 40
8 Burner and Feed System 41
9 Sources of Composite Dust 43
10 Plot of Area Ratio vs. Loss of Ignition 47
VI
-------
TABLES
No. Page
1 Alkali Losses from Shale 11
2 Analysis of Specimen Kiln Dusts 13
3 Analysis of Condensate and Recovered Dust 14
4 Comparison of the Composition of Condensate
FS-4 with the Composition of Starting Material 14
5 Particle Size Analysis and Distribution of
Alkalies in a Specimen Kiln Dust 16
6 Analysis of Dusts and Condensates 23
7 Analysis of Dusts and Condensates 24
8 Comparison of the Oxide Ratios in the Treated
and Untreated Dusts 27
9 Analyses for Material Caught in Feed System 32
10 Analytical Data - Run of 5/8/74 34
11 X-Ray Diffraction Results - May 8, 1974 35
12 Material Balance - Run of 7/12/74 36
13 Potassium Balance - Run of 7/12/74 37
14 Temperature Profile of System 38
15 Temperature Profile, Runs of 10/28/74 and 10/30/74 44
16 Combustion Gas Composition, Run of 10/30/74 45
!7 Beneficiation of Kiln Dust Samples 46
18 Material Balance, Runs of 10/28/74 and 10/30/74 49
19 Kiln Precipitator Dust 50
20 Rates of Weight Loss at Various Temperatures 51
21 Process Parameters for Flame-Spray Runs 55
22 X-Ray Diffraction Results, Runs of 6/3, 6/11,
6/16, 6/20/75 56
23 Material Balance - Runs of 6/3, 6/11, 6/16,
6/20/75 57
24 Process Parameters and Material Balance, High
Alkali Kiln Dust Run of 7/23/75 58
vii
-------
INTRODUCTION
The cement manufacturing process requires that a lime-rich material
(normally limestone) be interground with an argillaceous material
(normally clays, shales, sands, or feldspar rocks) and that the mixture
be burned in a rotary kiln at temperatures of from 1350-1600 C (2450-
2910 F). The required chemical elements for present-day cement manu-
facture are calcium, silicon, aluminum, and iron. Generally, econo-
mically feasible sources of these elements contain other elements as
well, particularly magnesium, sulfur, and alkali metals (primarily
postassium and sodium). For example, such clay minerials as orthoclase
(KAlSi.Og) provide the necessary silica and alumina, but also contain
appreciable potassium (about 14% by weight). Some limestones can also
contain alkalies in significant quantities.
Excessive amounts of alkalies can have deleterious effects upon the
process of cement manufacture and upon the properties of the product.
Normally much of the alkali present in cement raw materials (especially
clays, shales, and feldspar rocks) is volatilized in the cement kiln and
condenses on the particles of kiln dust which are carried out of the
kiln by the combustion gases. Pollution control devices, such as
electrostatic precipitators and fabric filters, collect this dust. If
it is subsequently returned to the kiln, an equilibrium circulating load
of alkali is established. When this load is too high, serious kiln
"ring" formation can occur, which often leads to problems with opera-
tional continuity, as the alkalies alternately vaporize and condense.
These rings can be low-melting eutectic mixtures of alkali and calcium
sulfates, sometimes containing chlorides, which ultimately solidify as
raw materials are assimilated into the ring. As the kiln cross section
decreases, gas velocity profiles are altered, and feed surges can occur.
Eventually, of course, the alkalies will be incorporated into the
clinker product. High alkali levels in cement can cause serious concrete
distress, if the aggregates used to make the concrete are "reactive."
Normally, when reactive aggregates are present, the optional ASTM limit
of 0.6% alkalies (calculated as Na_0 equivalent) is specified. Whenever
aggregates are sound, it should be possible to tolerate higher alkali
levels. When excessive alkali is present as sulfate, it can limit set
control capability, however, and when alkali is present as silicates,
it can lead to unsatisfactory strength development. It is impossible to
set a single alkali goal which will be applicable to all plants.
In addition, if the alkalies in a cement are high and sulfate levels are
low, unfavorable reactions can occur, which can lead to poor strength
development and abnormal setting behavior in extreme cases. Hence, it
is advantageous to have alkali stoichiometrically balanced by sulfate.
For these reasons, it is often necessary to discard at least a portion
of the kiln dust, which represents a substantial investment of time,
effort, and processing energy. J This disposal also represents poor
land use, and can create a water pollution problem because of leachable
alkali salts.
-------
Acquisition of the Existing Information
The library of the Portland Cement Association was a source for deter-
mining the state of knowledge on alkali removal from kiln dust. Numerous
foreign authors (notable Sprung, Ritzmann, and Weber) have commented on
the effect of alkalies on the operation of rotary kilns, but few papers
address themselves to the removal of alkalies from dust. Several PGA
publications have emphasized the need for monitoring alkali levels in
clinker and dust, and many plants using multistage electrostatic precipi-
tators have been able to recycle all dust except that trapped in the
last stage or last two stages. Other than the Davis report however,
only German workers have devoted appreciable study to the problem.
Hence, the existing information acquired was chiefly concerned with the
properties of kiln dust, and the operation of existing fluid bed systems,
rather than,the actual separation of alkali from kiln dust. These
references, combined with the authors' experience with the material,
served as the basis for experimental design.
A number of publications by the Portland Cement Association have addressed
themselves to the problem of kiln dust usage and disposal. For example,
efforts have been made to use kiln dust for an agricultural lime sub-
stitute, as an ingredient for soil stabilization,| ' ^ and as a
material for neutralization of acid waste streams. When kiln dust
must be disposed of by stockpiling, study has been devoted to the con-
struction of facilities to minimize egress of water containing alkalies
from plant property. All solutions involving disposal heretofore pre-
sented have been costly either in terms of capital, labor, energy, or
combinations of these.
Other methods of treating kiln dust to remove alkalies,have been the
subject of studies previously sponsored by EPA. Davis ^ discussed the
use of leaching combined with electrodialysis to dissolve the alkalies,
then concentrate the leachate. Membranes used with the electrodialysis
technique must be protected so carefully that it is possible that main-
tenance could be a problem over a period of time, although Davis does
not discuss this aspect in detail.
-------
CONCLUSIONS
Modest levels of alkali removal from kiln dust are achievable with a
pyroprocessing technique hereafter referred to as "flame-spray."
Control of the system is difficult, but some encouraging results have
been obtained when the system was properly operated.
The degree of alkali removal generally obtained> however, was less than
expected. The primary reason for this discrepancy is in the nature of
the materials themselves. Although volatilization of alkali salts from
kiln dust can occur readily at 2,000-2,100 F (owing to the high surface-
to-mass ratio of the materials), keeping the components separate is a
much more difficult matter. If the collecting medium is a glass fiber
filter, the maximum allowable temperature is 500-550 F. At these
temperatures the alkali particles have already condensed and coalesced
to the point that they are easily plated out on the alkali-deficient
dust particles already on the filter. The result is that the dust
remains relatively alkali-rich.
Two approaches were investigated in an attempt to increase the alkali
removal:
1) Condensation surfaces were presented to the hot gas stream, in
an effort to induce condensation of the alkali vapors while
temperatures were high. The larger, bulkier dust particles,
which underwent no gas-liquid phase transition, would be
expected to avoid condensation and pass on to a collecting
medium. The alkali vapors, however, pass through a liquid-
phase transition and could be expected to be condensible.
2) A filter material was chosen which could withstand elevated
temperatures. Hence, the dust might be caught, and the
miniscule alkali particles would hopefully be able to pass
through the mesh of the filter and be trapped later, perhaps
in an electrostatic precipitator.
Neither approach was fully successful. The nature of the alloys avail-
able for the high-temperature study could not withstand high enough
temperatures to permit alkalies to pass through as vapor. Similarly,
although the condensation surfaces removed appreciable alkali from the
dust, separation was not quantitative, and the practical problem of
causing the release of the alkali concentrate from the condensation
surfaces was never satisfactorily solved. A literature search indicates
that brushes have been used in a German study and also the condensation
surfaces have been caused to rotate into a water filled trough. * '
-------
Substantial alkali volatilization occurs during flame treatment of kiln
dust samples. The problems yet unresolved involve maintenance of
separation of the alkali-rich and alkali-poor phases. The results of
one experiment, conducted just prior to the conclusion of this project,
suggest that some amelioration of these problems can be achieved, pro-
vided that gas velocities can be kept low and the collecting medium
maintained at very high temperatures ( 1000 F).
In a separate phase of the study, reported elsewhere, the kiln dust was
burned to clinker in a rotary kiln without pretreatment. The product
was interground with normal portland cement clinker at various sub-
stitution levels. Evidence was presented that at the 20% substitution
level, acceptable concrete performance could be achieved using this
interground cement.
-------
RECOMMENDATIONS
In the area of flame-spray dust treatment, time precluded investigations
in several areas which could prove fruitful:
1) It has come to our attention that certain other fibers are
available, fibers such as ceramic alumina, which could con-
ceivably tolerate higher temperature ranges. It is felt that
if materials could withstand 900-1,000 C (1,652°-1,832°F),
alkali vapors would likely pass through the filter, and dust
particles be retained. This could prove a promising avenue of
approach in light of the findings of the final experiment
outlined in this report.
2) German work^ ' has dealt with the use of condensation
surfaces in the temperature region of interest (900°C) to
remove alkalies prior to the use of electrostatic precip-
itators. A potential "moving chain" arrangement makes use of
chains caused to move alternately through the exit gases of a
rotary kiln, then through a water-filled trough or a system of
brushes to remove alkalies. This system has not, to our
knowledge, been adopted by any cement-producing plant, for
reasons that are not presently clear. Further developments in
this area might prove practical.
3) Owing to water pollution problems, the industry is usually not
in a position to discharge water used to leach kiln dust.
However, if this water is made recyclable, it may become
possible to leach out the water-soluble alkalies with hot
process water, cool the solution, and filter out solids
(chiefly lime, alkali-sulfate, and alkali chlorides). The
water could then be reheated to extract further alkali from
further dust. The lime has a slightly negative temperature
coefficient of solubility, and hence will not separate on
cooling. Thus, the filtered material will be high in alkalies
(probably chiefly K2S04' a Potentially saleable fertilizer
supplement.) The insoluble material could be used as a raw
feed supplement in wet-process systems, and waste heat from
many dry-process kilns would be sufficient to remove the water
prior to recycling the dust for these systems. Another
alternative is to investigate removal of the alkali from the
raw mix components having high alkali contents. This was
partially explored in muffle furnace burns (clays, shales)
prior to use. In this case, one would have 10-12% of the
total material for alkali removal instead of 100% (kiln dust).
Studies,have shown that kiln dust is effective in neutralizing acid mine
wastes. ' Although required at levels slightly in excess of lime for
this application, the kiln dust has been observed to react faster and to
generate a much lower volume of sludge than lime. In view of these
-------
findings, it would appear appropriate to study kiln dust as a replace-
ment for lime in certain S02 scrubber systems installed on industrial
boilers and the like. Four advantages would be expected to accrue:
1) The more rapid neutralization might be expected to remove
higher levels of S02 than lime with equivalent scrubber geometry.
2) If sludge volume for equivalent S02 removal is reduced, the
corresponding sludge disposal problem would be correspondingly
reduced. This would depend on the supply of kiln dust rela-
tive to the scrubber demand. However, at present, about
12,000,000 tons of dust are available per year.
3) The alkali values in kiln dust would be converted largely to
sulfates. Hence the sludge could consist of K~SO. and clay
minerals in addition to calcium sulfate which should heighten
its attraction as a soil conditioner, since it contans sig-
nificant quantities of a major plant nutrient, potassium.
However, recent research by Southern Research Institute has
revealed that some kiln dust samples contain minor but sig-
nificant quantities of toxic elements, such as zinc, lead, and
arsenic. Before utilization of kiln dust for agricultural
purposes, it is important to carry out a thorough chemical
analysis to assess this factor.
4) Whereas the lime used for SO,, neutralization is a purchased
material, kiln dust is presently regarded in some quarters as
a waste product. Economic advantages could accrue to the
purchaser and user.
Economic limitations include shipping costs and nonuniformity of dust
composition. Where the source of the dust was close to a market without
tight compositional specifications, these problems would be minor.
Any or all of the foregoing are useful approaches which are believed to
be appropriate extensions of the present work.
-------
Objectives of the Research Program
The research program was designed to assess the feasibility of separ-
ation of alkali salts from the balance of the kiln dust by fluidized
bed, and later by flame-spray, techniques. The plan was to take advan-
tage of the known volatility of alkali salts, and then to condense the
alkalies prior to dust collection, or to separate the coarser dust and
then collect the very fine alkali fume. In essence, there were four
steps required for successful operation:
1) The dust had to be introduced quantitatively into the pyro-
processing unit.
2) The temperature of the unit had to be sufficient to induce
alkali salt volatilization, but low enough to conserve fuel.
3) Between the point of volatilization and that of collection,
the alkali-rich and alkali-poor fractions had to be kept apart
insofar as was possible.
4) Collection of the two fractions separately had to be efficient.
Of the four necessary steps, the first two were accomplished with
relative ease. However, the third and fourth step seemed to be mutually
contradictory. Whenever the maintenance of separation of the two
fractions was efficient, the total collection efficiency decreased and
vice versa. It would appear that this remains the primary unsolved
problem in this work, except for the single successful experiment
carried out at the conclusion of the test series.
Theoretical Considerations
In theory, rather high temperatures are required to vaporize alkali
salts, such as potassium sulfate, from kiln dust. However, the temper-
atures required in practice are much lower, for several reasons:
1) The alkali compounds are condensed as very fine particles on
the surface of the dust. Because of the very high surface
area-to-mass ratio, the effective volatility is high.
2) There is a moving gas stream in the kiln, which promotes
volatility by carrying away alkali vapors as they are formed,
and shifts the equilibrium in favor of the vapor.
3) Carbon dioxide and water vapor are present in the burning
zone. These compounds are believed to enter into reaction with
alkali sulfates as:
K0SO. + CC- -> K0CO_ + S0_
24 2 23 3
KS0 + H0 ->- 2KOH + S0
-------
Although these equilibria are not favorable, any carbonate or hydroxide
formed can be expected to vaporize far more readily than the sulfate,
and hence be carried away from the reaction site. At lower temperatures
downstream from the burning zone, the reverse reaction can again occur,
effecting the facile volatilization. These phenomena have been investi-
gated by Vogel and Lehmann and Plassman.
Because of the preceding factors, kiln dusts can and do lose their
alkali at temperatures around 1,100 C. Only the alkali salts are
effectively vaporized at this temperature while alkalies which are in
the form of silicates or aluminosilicates are unaffected until and
unless they are metathesized to sulfate, halides, or the like. The
calcium oxide and calcium carbonate present in the kiln dust can help in
this metathesis, as shown in the following examples:
7CaC03 + 2KAlSi3Og -* 6CaSi03 + CaAl^ + K2
K CCL + CaSO. (in hot zone) -> CaO + CO + K2S°4
7CaO + 2KAlSi3Og -*• K20 + 6CaSi03
2KC1 + CaO
The presence of calcium silicates and aluminates in the treated kiln
dust from this study lends credence to this hypothesis. Also consistent
with it is the observation that the water-soluble alkali content becomes
a greater portion of the total after pyroprocessing at high oxygen
levels.
EXPERIMENTAL INVESTIGATIONS
I. Fluid Bed Tests
Initial investigations into the feasibility of alkali volatilization
from kiln dust were carried out using a fluidized bed furnace. The
design selected is illustrated in Fig. 1. A Hoskins wire-wound electric
bench furnace, 30.5 cm in length, has a tube diameter of 4 cm. The
furnace was mounted vertically and encompassed a 2.5 cm Vycor tube which
contained a fluidized bed. A stainless steel cap was machined to fit
the Vycor tube and provide an inlet for the fluidizing gases. The cap
is sealed to the tube with an asbestos paper gasket, and a fritted glass
or stainless steel mesh support was used to disperse the gas and prevent
clogging of the gas inlet. A glass "Y" tube connected to the inlet tube
permitted addition of two gases simultaneously. Initial experiments
incorporated the use of moist C02 and air. Air was supplied by the
house line and the pressure regulated with the ballast tank and pressure
-------
THERMOCOUPLES
STEEL CAP
\
STEAM
GENERATOR
GAS
INLET
VYCOR TUBE
FURNACE
FLUIDIZED BED
-SCREEN
HEATED
SAND BATH
C02
CYLINDER
FIG. I - FLUIDIZED BED FURNACE
-------
gauge of a paint sprayer. CO- was generated from sublimation of dry ice
or from a cylinder of the compressed gas, and steam insufflated into the
CO,, gas stream from a simple boiling flask.
Temperatures of the interior of the bed and of the furnace wall were
measured with chromel-alumel thermocouples inserted from the top of the
system either directly into the fluidized bed or between the furnace
wall and the Vycor tube. Readings were taken with a potentiometer
calibrated directly in degrees Fahrenheit.
The operation of the fluidized bed furnace required a very delicate
adjustment of the pressure of the fluidizing gas. In general it had
been the practice to initiate ebullience in the bed with the Vycor tube
lowered from the furnace for observation of the activity. When an
acceptable degree of agitation was observed the tube was raised into the
furnace. During an experiment the temperature of both the furnace wall
and the bed were recorded at frequent intervals. Samples were taken
directly from the bed for analysis by inserting a small porcelain
crucible from the top of the system.
Some experiments were performed in an attempt to use a torch flame to
fluidize and heat the bed directly. With this system these were not
successful. It was not possible to reduce the gas velocity sufficiently
to maintain a stable fluidized bed and keep the torch flame ignited.
Moreover in the relatively short length of this tube the particles of
material tended to fuse.
Muffle Furnace Experiments
Several tests were made using the electric muffle furnace on samples of
shale material. Shale was chosen as an example of argillaceous raw
material, because it contains many of the same general types of minerals
found in clays. Feldspathic rocks are known to release their alkalies
with more difficulty, and were therefore less promising candidates for
this treatment. The amount of work done in this area was limited, be-
cause further extensions were contrary to the term of the contract.
The results of these tests are shown in Table 1. Mixtures in varying
proportions of the shale material together with such additives as CaCl,,,
CaCO_, CaSO. and oxalic acid were made by dry trituration of the solids
or by preparation and drying of slurries. Samples of approximately 3
grams were heated in platinum crucibles at temperatures up to 1,100°C.
The products were examined by X-ray diffraction to detect changes in the
mineral composition and possible compound formation. Chemical analysis
by atomic absorbtion spectrometry was used to detect any losses of
alkali.
The results obtained indicate some promise that higher temperature or
other chemicals might bring about volatilization or solubilization of
the alkalies. Further work in this area may prove fruitful.
10
-------
Table 1- Alkali Losses from Shale
Sample
No. Additive
control
4 CaSO.
4
5 CaC03
6 CaF2
nu p n
n«vj — u j
224
12 CaCl2 2H20
13 CaCl2 2H20
% Add'nb T C°C)C
1100
2 1100
10 1100
1.7 1100
3.8 1100
4 1100
6 23-1100e
, , Percent
K-0 Na20 K-0 Lost
3.
3.
3.
3.
3.
2.
3.
87
85
84
86
12
63
53
1.
1.
1.
1.
1.
1.
1.
18
17
16
14
09
02
01
0
0
0
19
32
8
0
.56
.89
.32
.42
.00
.82
Percent
Na?0 Lost
0
2
3
7
13
14
0
.91
.08
.42
.96
.76
.28
a Shale Lot No. A-118.
b In weight percent.
c Samples heated 15 minutes at 1100°C except when noted.
d Alkali content of shale portion of sample; ignited basis.
e Thermal balance experiment, heating over one hour period.
11
-------
Thermal Balance Experiments
Thermal balance measurements were made on several potassium compounds to
determine the temperature at which volatilization began and to obtain
estimates of the rate of that volatilization. Samples of K?CO_, KOH,
and KC1 were mounted in the balance pan in a small furnace. The weight
changes of the sample during a programmed temperature rise from 70 to
1,050°C were recorded on a twin-pen strip chart recorder which gives a
simultaneous record of weight change and furnace temperature. Estimates
of the rate of weight loss could be taken from the slope of the weight
change versus time at a particular temperature. The results were as
follows:
Temp. Rate of Wt. Loss
KC1 775°-1026°C 100%/hr.
KQH(Step 1) lg()o_ 45Qoc 35.5%/hr.
KQH(Step 2) 8250_10500C 30.1%/hr.
K2C03 930°-1155°C Not Available
KS0 1030°C 0.375%/hr.
Flame-Spray Experiments
Because of the difficulties encountered with the fluidized bed approach
to alkali removal, it was decided to attack the problem in a manner
analogous to that used in certain cement kiln systems. In these systems,
the kiln dust is recycled through the burner pipe directly into the
firing zone. At the temperature used in the burning zone, the condensed
alkali is immediately "flashed" off. The concept of a fluid bed feeding
system was not abandoned, however, as it seemed a good method of delivering
a moderately constant dust feed rate to the burner.
Initial experiments incorporated a 500 ml Erlenmeyer flask, inverted to
form a conical chamber. A regulated supply of air was used to entrain
dust in the air supply to the burner using a spouted bed. Experiments
were carried out using glass-blowing torches and pilot burners for
rotary kiln use. The dust used for these initial experiments, and all
subsequent runs, was a high-alkali, high-sulfur sample with chemical
analysis as shown in Table 2. The flame-sprayed product gas stream was
directed into a steel tube of 20 cm diameter and 50 cm length. The
initial condensates observed (in very small amounts) had a partial
chemical analysis as shown in Table 3. In itself, this finding was
encouraging as it indicated that a high-alkali fraction could be sep-
arated from the dust by volatilization. Somewhat disconcerting, however,
was the low yield of this material. Glass wool was used as a filter
originally, but proved inefficient. The material (FS-3) caught in this
filter had a partial chemical analysis as shown in Table 3. Table 4
12
-------
Table 2 - Analysis of Specimen Kiln Dusts
Number
1
2
As Is
Ing. Basis
As Is
Ign. Basis
Description
Precipitator
Kiln Dust
High Chloride
Precipitator
Kiln Dust
Low Chloride
Si02
13.4
17.6
15.2
18.3
A1203 Fe2°3 Ca° M8°
2.66 1.25 37.0 1.30
3.30 1.64 48.7 1.71
4.18 1.90 38.7 1.16
5.03 2.29 46.6 1.40
S°3
2.01
2.64
12.34
14.86
Na2°
0.44
0.58
0.41
0.49
Number
1
2
As Is
Ign. Basis
As Is
Ign. Basis
Description
Precipitator
Kiln Dust
High Chloride
Precipitator
Kiln Dust
Low Chloride
K2°
10.31
13.57
8.9
10.71
Ign.
Loss P205 TiO F- Cl-
24.04 0.19 0.15 0.11 6.24
0.25 0.20 0.145 8,27
16.26 - 0.26
0.31
Water
Na20
0.30
0.395
0.22
0.26
Sol.
K20
7.80
10.26
6.99
8.42
-------
Table 5 - Analysis of Condensate and Recovered Dust
Lab
Designation
FS-1
FS-3
FS-4
(a)
(b)
(c)
CaO
MgO
SO-.
Na?0
11.0 3.20 3.62 26.0 0.80
1.60 38.5
0.38 7.36
23.47 0.80 22.8
(a)
Flame spray condensate (white), pilot burner:
from steel tube: Table 2, No. 1.
Cb)
Recovered dust corresponding to condensate FS-4,
(ignited basis) pilot burner: Table 2, No. 2.
(c)
Flame spray condensate (white), pilot burner:
from steel tube: S05 dust: Table 2, No. 2.
Table 4 - Comparison of the Composition of Condensate FS-4
with the Composition of Starting Material
Weight
Ratio
Original
Dust
Condensate
FS-4
CaO
Si02
2.55
2.36
Fe203 CaO
A12W3 MgO
0.455 33.6
1.13 32.5
A1203
~MgO
3.60
4.00
CaO
Fe203
20.37
7.18
CaO
A1203
9.26
8.12
MgO
13.10
13.75
14
-------
details the ratios of various oxides in the condensate sample. The
increase in content of Fe2°3 maX be attributable to corrosion of the
steel surface, while the moderate enrichment of silica and alumina
relative to lime is probably a result of the fineness of the clay frac-
tions of the dust relative to the limestone fractions.
A second kiln dust sample was studied, in these preliminary experiments
only. Its chemical analysis is shown in Table 2, sample 1; notable is
the high chloride level of the dust. When the flame-spray techniques
described were applied to this material, the alkali enrichment of the
condensate was even more substantial, a fact attributable to the chloride
content of the dust. Alkali chlorides are known to be considerably more
volatile than the corresponding sulfates. Alkali results for this run
appear in Table 3 (Sample FS-1).
Further Characterization of Kiln Dust
The high-sulfate kiln dust (Table 2, Sample 2) was examined for particle
size distribution, and the separate size fractions were analyzed for
total alkalies and water-soluble alkalies. The purpose of these tests
was to verify the following characteristics of the dust:
a) The actual particle size distribution
b) The distribution of total and water-soluble alkalies as a
function of particle size. The results, which appear in Table
5, show the great fineness of the dust, and also indicate that
the water-soluble alkalies are concentrated in the finest
fractions. This is to be expected if the hypothesis that
alkalies collect on dust particles is accepted, since the
surface area-to-mass ratio for such particles is higher than
for larger particles.
The encouraging preliminary results prompted construction of a more
permanent apparatus for flame-spray experiments. It consists of a flame
and condenser chamber 20 cm in diameter formed from stainless steel
sheet. The top of the tube was covered by a hood through which com-
pressed air is passed to cool the tube surface and promote condensation
of the alkalies. The preheated air was then passed into the dust
fluidizing chamber from the bottom through a perforated ceramic disc.
Chips of corundum were agitated by the air steam and helped to break up
the dust agglomerates. A second supply of preheated compressed air was
added to provide a tangential gas flow within the fluidizing chamber.
This reduced but did not eliminate the sticking of dust on the chamber
walls.
The dust that passed through the burner was drawn down the tube to the
glass cloth dust bag by means of a vacuum cleaner. The burner end of
the tube was partially closed by a metal plate to improve suction. A
vacuum was found to be essential since the dust bag impeded the gas flow
and caused blowback. The dust bag provided approximately 1 yard of
surface area and was suspended in the vertical stack by a wire form.
Directly below the bag a can, sealed into the system, served as a
collector.
15
-------
a
Table 5 - Particle Size Analysisv J and Distribution of Alkalies
in a Specimen Kiln Dust __
Particle Size
Range (Microns)
+68
-68+48
-48+34
-34+24
-24+17
-17+12
-12+6
-6
Weight
Percent
0
0.3
0.4
0.7
1.8
5.1
27.3
64.4
Total Alkalies
%
Na20
-
0.30
0.31
0.35
0.38
0.40
0.33
0.42
K20
-
3.62
3.46
4.51
5.08
5.15
5.35
10.72
Water Soluble Alkalies
%
Na20
-
(cO
(c)
0.094
0.117
0.134
0.134
0.242
K20
-
(c)
(c)
1.927
2,560
3,072
3.252
8.191
Water Insoluble
K20
%
~
-
-
2.58
2.52
2.08
2.10
2.53
(a)
Particle size analysis was carried out by the Allis-Chalmers
Corporation using an "Infrasizer" particle size analyzer.
(b)
Low chloride precipitator kiln dust cf. Table 1 No. 2
(c)
Insufficient samply for analysis
16
-------
The operation of this system was still very imperfect. Extinction of
the flame was no longer a problem with the larger burner although
"flashback" caused by pre-ignition within the burner sometimes occurred
if the gas fuel flow was improperly adjusted. The major difficulty
seemed to be that of breaking up the dust agglomerates so as to prevent
sticking and to ensure that only very small discrete particles pass
through the flame. The tangential secondary air supply and the corundum
pellets were helpful and this suggested that a swirl preheater chamber
operating on the principle of a cyclone dust separator could be neces-
sary. A schematic design of such a chamber is shown in Figure 2. The
heavy agglomerate particles and coarse chips of corundum were caused to
separate from the air stream by centripetal force. These were broken up
by collision with the walls and the corundum chips. The lightweight
fine particles were then carried off by the air stream to the burner.
Provision also needed to be made was to preheat the air more effectively.
The heat exchange between the large volume of rapidly flowing air and
the furnace or the condenser pre-heater was so poor that very little
preheating of the air and dust was actually found.
Substantial amounts of white condensate were formed on the condenser but
a sample when collected was dark gray indicating that corrosion of the
stainless steel had occurred. A relatively large amount of dust was
collected from the tube directly beneath the burner indicating that much
of the original 20-gram sample passed through the flame in the form of
heavy agglomerates. Some very fine dusts were gathered from farther down
the furnace tube and from the dust collector but only in very small
amounts.
The aforementioned difficulties prompted some changes in equipment
design. The next modification of the equipment, shown in Figure 3,
delineates the changes made.
Two identical 15 rpm electric motors were used to drive the feed screws
and stir the dust in the feed hopper. The feed screws were made from
ordinary 3/8" wood auger bits. A paddle arrangement in the feed hopper
was found to be necessary in order to prevent the feed from hanging up on
the walls. Mechanical supports for the stirring motors are not shown.
The hopper capacity was about 30 grams of test kiln dust. An attached
funnel (not shown) allowed for convenient addition of further material
by hand. The dust was fed directly into the air supply pipe that passes
through the aluminum base of the apparatus and enters the base of the
cyclone chamber tangentially. The cyclone chamber is a piece of lucite
pipe 76 cm high with a 10 cm inside diameter. The transparent pipe
permitted visual observation of the behavior of the dust in the cyclone.
The pressure of the air supply was regulated and adjusted between 1 and
60 psi by a paint sprayer ballast tank. Air pressure was usually
controlled between 40 and 60 psi. A small amount of sand and crushed
particles of lightweight aggregate was added to the cyclone chamber to
clean the dust from the walls and to assist in breaking up the dust
agglomerates. The cyclonic action in the chamber caused the agglomerate
lumps to circulate around the walls where they were readily broken up by
collision with the coarse sand and aggregate particles. Only the fine
17
-------
FUEL
BURNER
FLUIDIZED BED
PREHEATER
TO DUST COLLECTOR
OR KILN
SCRAPER
CONDENSATION SURFACE
__PREHEATED AIR
FROM CONDENSOR
FIG. 2-SCHEMATIC OF SWIRL PREHEATER SYSTEM
-------
Burner
Venfuri
Cyclone
chamber-^
Air —
Gas
Electric motor
Stirring p
Feed hopper
Electric motor
Feed screws
-Air
FIG.3-FLAME SPRAY BURNER WITH CYCLONE AND FEED SUPPLY
-------
dust particles were entrained and carried with the air flow into the
burner. A difficulty in the operation of the cyclone had required a
modification of the air supply system to the cyclone chamber. The air
that carries the dust into the chamber rose immediately to form the
cyclone vortex. The assending spiral of fast moving air left a "dead"
space at the base of the cyclone chamber and an appreciable portion of
the agglomerated dust feed fell out of the fast air flow and accumulated
in the "dead" space. A second tangential air supply diametrically
opposed to the first was added and minimized the problem.
With the original smaller burner there was still a strong tendency for
the dust leaving the cyclone chamber to re-agglomerate and deposit on
the interior walls of the burner. Using a larger exit pipe (2 inches
ID) reduced the amount of contact between the air stream and the walls.
A venturi section with reduced diameter (1 inch) in the burner itself in
front of the gas fuel supply not only improved mixing of the air and gas
but by accelerating the gas stream and the dust may also have inhibited
dust accumulation in the horizontal portion of the burner. The burner
tip was a disc perforated by 1/16 inch holes that served to stabilize
the flame.
Figure 4 shows a schematic of the entire flame spray system. A 20 cm
diameter tube formed from stainless steel sheet acted as a condenser. A
cowl covering the top third of the condenser and through which air was
passed to cool the condenser surface could also serve as a heat exchange
pre-heater for the air supply to the burner.
The flame from the burner flowed along the concave inner surface of the
tube and deposited the volatilized alkali vapors on the condenser
surface. Some dust also adhered to the alkali condensate. Most of the
alkali condensate collected on the surface directly above the flame and
the density of the deposit decreased rapidly along the path of the gas
flow. The dust-laden air was drawn down the tube by a large industrial-
type vacuum cleaner and through a filter bag made from glass cloth. It
was found that much of the dust passed through the filter bag and a new
glass cloth bag with a finer weave was made for future experiments. The
dust that did collect on the bag was shaken off and collected after an
experiment. It would be anticipated that a full-scale system would
require a spring-loaded dust bag that could be vibrated periodically to
dislodge the dust to the dust collector for continuous recovery.
Alternatively, an electrostatic precipitator could be used.
Analysis of the Alkali Condensate and Recovered Dust
The white condensate material from the condenser and the recovered dusts
were analyzed by X-ray diffraction before being submitted for chemical
analysis. Comparison of the diffraction patterns of the recovered dust
after the flame spray treatment with the untreated dust showed a marked
degree of calcination in the flame sprayed material. Most of the CaCO
had been decomposed and strong peaks for free CaO were found. In
addition, significant amounts of belite (dicalcium silicate, C S) were
found together with small amounts of alite (tricalcium silicate, C S).
•J
20
-------
Vacuum
Glass cloth
filter bag
Condenser surface-
stainless stee!
Air cooled chamber
and air preheater
Stainless steel tube
Dust collector
Cyclone
chamber
Air
Burner
-Gas
Feed
-Air
FIG. 4 - FLAME SPRAY SYSTEM
-------
Silica, present in the untreated dust, was only detectable as a trace
after the flame spray treatment. Spurrite (2C S CaCO ) was found as an
intermediate.
The diffraction pattern of the alkali condensate collected from the
surface of the condenser showed only a very strong pattern for K-SO
with some weak lines for free CaO. Microscopic examination of tne
material showed a trace of CaCCU and large amounts of non-birefringent*
glassy material in the form of very fine beads with diameters ranging
from 4 to 20 microns. These beads have the appearance of fly ash
particles and are not water soluble. They were probably formed from
fusible raw material constituents present in the kiln dust.
Chemicaj^ Analysis
The low-chloride precipitator dust was used in all of these experiments
(Table 2, No. 2). The untreated (as is) dust has an appreciable ignition
loss (16.26 percent) and accordingly reference to the composition of the
dust will be made to the material heated to 1000 C (ignited basis). On
this basis the K 0 content of the untreated dust was 10.71 percent; the
Na_0 content 0.49 percent.
Table 6 shows the analyses of the condensates and recovered dusts collected
in the system shown in Figure 1 but using the economite burner. The 30.8
percent K20 corresponds to 57 percent K2SO • 1.29 percent Na^O is equi-
valent to 2.95 percent Na_SO.. Sample FS-9 is the agglomerated dust
lumps that were collected under the burner. Very small amounts of fine
dusts FS-8 and FS-11 were recovered from the base of the end tube and the
dust collector bag respectively. So little dust was collected that the
analysis is probably somewhat uncertain. The magnitude of the K_0
content of the treated dusts, FS-8 and FS-11, corresponds very closely to
the water insoluble K_0 content in the original material. It is worth
noting that the premise of this technique was that the water soluble
alkalies were in the form of salts, especially sulfates, condensed on the
particle surfaces. The water insoluble alkalies are assumed to be
distributed through the bulk of the material and are present as alumino-
silicates and thus are not readily susceptible to volatilization.
Table 7 shows the analyses of the dusts and condensates collected using
the new system. The white alkali condensate that formed on the condenser
surface directly above the burner tip (FS-15) showed a tendency to
separate from the condenser in the form of coherent flakes. The conden-
sate further from the burner (FS-14) was collected as a fine powder. The
X-ray diffraction patterns of both condensates showed no conspicuous
difference. The only notable lines in both instances were for K_SO.
with traces of CaO. The chemical analyses of FS-14, 15 show significant
differences in Si02 and K20. FS-14 would require 25 percent SO for
complete combination with the alkali oxides; FS-15 required nearly 27
* non-birefringent - cubic
22
-------
*a
Table 6 - Analyses of Dusts and Condensates
Samples Na^O K00 Description
FS-8
FS-9
FS-10
FS-11
c —
0.23
0.49
1.29
0.32
E
2.8
7.02
30.8
2.2
Fire dust, base of end tube
Coarse dust, under burner
Condensate
Dust from collector
Using fluidized bed feed and "economite" burner
(burner with self-contained compressed air supply.
This burner was more efficient but had to be abandoned
because it was not conducive to feeding dust.)
23
-------
Table 7 - Analysesj__of_ Dusts and Condensates
Sample
Number
FS-12
FS-13
FS-14
FS-15
FS-4
Sample
Number
FS-12
FS-13
FS-14
FS-15
FS-4
as is
ign.
-
as is.
ign.
0.37
0.375
1.30
1.68
1.72
1.16
0.80
09 A19°3 Fe->°3 Ca°
24.1 6.5 2.8 50.5
24.4 6.58 2.85 51.15
_
17.5 3.3 1.2 20.0
17.97 3.39 1.23 20.5
10.5 2.6 1.0 20.1
11.0 3.20 3.62 26.0
Loss
on Soluble
ICO Ign. Na?0 TC^O
7.6 1.25 0.11 3.61
7.7 - 0.11 3.65
4.9 -
26.9 2.60
27.6 -
30.0 -
22.8 -
MgO S0g
1.5 4.4
1.52 4.46
-
0.9 21.9
0.92 22.5
0.2 23
0.8 23.47
Description
Dust from glass cloth
Dust bag
Dust from base of end
tube
Alkali condensate
fine powder (cf . text)
Alkali condensate
flakes (cf. text)
Alkali condensate
powder
-------
percent SO- to form the alkali sulfates. Again the relatively small
sample size and the single analytical determination make the SO analyses
uncertain by about 1 percent. The burning conditions are such "(excess
air) that oxidizing rather than reducing conditions would be expected.
It is possible that some simple decomposition of the alkali sulfate
vapors has occurred with condensation of oxide and loss of S0_. No
facile explanation can be given for the marked differences in the SiCL
content of the two condensate samples*. The analysis of a condensate
(FS-4) obtained in earlier experiments is shown for comparison.
The low SO content in the treated dust is especially interesting.
Assuming that in the original dust all of the alkalies are combined as
sulfates, 9.73 parts of the 14.86 percent total SO^ (ignited basis) then
is combined as K^SO,. In fact, one-half of the Na20 and approximately
20 percent of the K?0 in the original dust are water-insoluble and hence
unlikely to be present in the water soluble sulfate form. The differ-
ence of 5.13 percent SO. is a minimum that must be present in the dust
in a form other than alkali sulfate. No CaSO was detectable by X-ray
diffraction in the untreated dust. The possibility that some part of
the total SO, in the untreated dust was present as adsorbed gas was
considered. A sample of the untreated dust was heated to 300 C for 30
minutes to promote desorption. After the heat treatment the sample was
analyzed for total SO, by the Leco method. The 11.45 percent S03 in
this sample compares with the 12.34 percent in the original dust and the
result suggests that all or nearly all of the S03 in the dust is chemi-
cally combined rather than adsorbed. The observed loss of S03 in the
dust recovered after the flame spray treatment is evidence for a chemi-
cal decomposition of sulfate salts in the flame spray process.
K SO is detectable by X-ray diffraction in the treated dust. However
770 percent SO, is required for complete combination of Na^O and K20 as
the sulfate salts. Assuming that only the water soluble alkalies are
combined as sulfates requires only 3.23 percent S03 combined as alkali
sulfate.
In the treated dust only 47 percent of the total K20 is water soluble
whereas in the untreated dust approximately 73 percent of the K20 is
soluble. In absolute terms the water insoluble K20 is 4.05 percent by
ignited weight of the treated dust is only 2.39 percent by ignited
weight of the untreated dust. The increase in the water insoluble K20
is most likely due to decomposition of K2S04 with subsequent reaction of
K 0 to form water insoluble alkali silicates or alumino-silicates. Some
possible decomposition mechanisms are: 1) Simple decomposition of alkali
vapor, 2) reduction in the flame or 3) substitution reactions. These
possible reactons can be written:
*It is possible that some of the crushed lightweight aggregate particles
used in the cyclone feed chamber could have abraded and contributed high
silica fines as contaminants to the dust. In future work small plastic
beads should be used as abrasives in the cyclone chamber.
25
-------
R20(g) + S0
2) R2S04Cg) + 2e~ -* R20(
3) R2S04(s,g) + Si02 -> R2
The alkali oxide produced in these reactions could be absorbed in the
silicate reaction products, especially in the C2S (belite) phase.
Alkalies dissolved in a C«S phase is probably released slowly as the CJ$
hydrates, but during the ID-minute extraction period used in the ASTM
test, (C114) for water soluble alkali very little of the alkali in the
C2S would be released.
The increase in water insoluble K.,0 cannot be due simply to segregation
or retention of coarser particles in the filter bag. Reference to Table
5 shows that the proportion of water-insoluble K?0 in the separate
particle size fractions is nearly constant and ranges between 2.08 and
2.58 percent on the "as is" or 2.48 and 3.08 percent on the ignited
weight basis*.
Table 8 shows the comparison of the composition of the treated and
untreated dusts based upon the ratios of ignited weight percent com-
positions of the various oxides. The ratios suggest that the content of
Al~0_, Fe~0_ and SiCL is enhanced in the treated dust at the expense of
CaO and MgO. Variation of these ratios could be the result of contam-
ination OT segregation of particle sizes from a particle size distri-
bution with an inhomogeneous chemical composition. The constancy of the
CaO/MgO ratio implies that contamination from a source containing these
oxides is improbable and that the particle size distribution in the
treated dust is homogeneous with respect to CaO and MgO, i.e., the CaO
and MgO are together in the same particles, in the same proportion.
Similarly Fe20_ and A120 are found together, but their increased
proportion relative to CaO suggests that the Fe 0 and Al 0_ containing
materials constitute a coarser particle fraction which is better re-
tained in the dust bag. The relative increase in the proportion of SiO,,
may be due to greater SiO_ particle size but the increase may also be
caused by contamination of the dust by fine glass fibers from the filter
bag. Such very fine hair-like fibers could be observed in the recovered
sample of treated dust.
These results, while improved, did not solve completely the problems of
feeding, separation of alkalies, and collection. Therefore, further
changes were instituted to correct these problems. These consisted of
the following:
*There is a serious inconsistency in the analyses here. The water
insoluble K 0 in the bulk sample is 2.29 percent on the ignited weight
basis and tnis value should correspond to the weighted mean of the
corresponding values for the separated particle fractions. The sepa-
rated particle fractions, however, range in values between 2.48 and 3.08
percent for water insoluble ICO. This discrepancy needs to be recon-
ciled but it is unlikely to affect the argument given above.
26
-------
Table 8 - Comparison of the Oxide Ratios in the Treated and Untreated Dusts
CaO/SiQ2 CaO/MgO CaO/Fe203 CaO/Al203 SiQ2/Al203
Original
Dust 2.55 33.29 20.37 9.26 3.64 7.99 0.455
Treated
Dust 2.096 33.65 18.07 7.72 3.708 8.62 0.430
Percent
Change -17.80 +1.15 -11.29 -16.63 +1.87 +7.88 -0.549
-------
A. Cyclonic Dust Feeder (See Figure 5j
(1) The air inlet to the swirl feeder was reduced from 3/8" I.D.
to 5/32" I.D. This allowed a higher air velocity which
assisted in fluidizing and dispersing the dust.
(2) The feeder outlet pipe was shortened by 5 inches to reduce the
surface area of pipe upon which dust could collect.
(3) The truncated cone in the center of the feeder was extended
from 4-1/2" to 11". This alteration was effective in main-
taining the centrifugal flow pattern such that the upper walls
of the chamber were relatively free of accretions.
(4) A stainless steel "collar" was inserted between the feeder
outlet pipe and the lucite at the top of the feeder, to elimin-
ate dead air space and consequent dust accumulations.
(5) An interval timer was purchased to adjust dust flow rate so as
to avoid overfeeding.
(6) A vibrator was installed to assist in release of dust accumu-
lations from the lucite walls.
(7) Small wooden beads and styrofoam chips were substituted for
the sand and lightweight aggregate particles in the feeder.
These improved the efficiency of deagglomeration.
B. Condensate Separation (Figure 5)
(1) Four aluminum right angle bars were inserted into the stain-
less steel heat exchanger on the dust shell. These improved
heat transfer considerably, allowing for a greater temperature
differential, and hence better alkali collection.
(2) Thermocouples were installed at key positions in the ductwork,
to permit temperature monitoring in the gas stream, which is
vital to understanding the separation mechanism. In addition,
a portable pyrometer was purchased to allow monitoring of
additional selected locations, where appropriate.
C. Exhaust Stack Arrangements for Product Collection (Figure 6)
(1) A fine weave glass cloth bag was substituted for the coarser
material used previously. This increased collection effi-
ciency markedly.
28
-------
EXHAUST STACK
VARIBLE ARRANGEMENT
THERMOCOUPLE NO.
INSERTED
FROM SIDE
N>
34 5
/// /.HEAT EXCHANGER
2" J.D.
BURNER
PIPE
/// /
///
1st SECTION
CONDENSER
STAINLESS
29'
SHELL
STEEL
•ae
36
2nd SECTION
CONDENSER SHELL
GALVANIZED STEEL
'
DIA.
5
TOP
CAP
GLASS |
CLOTH |
BAG |
\ ____ J
8"
FIG. 5- FLAME SPRAY DUST COLLECTION ASSEMBLY
-------
04
o
8" DIA. #100 MESH SCREEN
'WITH No. I WHATMAN
FILTER PAPER
DIAL
THERMOMETERt
BAFFLED l^
CJ
~(O
/
J
TOP
CAP
GLASS j
1 CLOTH I
n
1 i
_i i
BOTTOM
CAP
8"
t
1
"b?1
i
s.
"to
8 DIA. DISK
WITH 5" DIA.
TO
EXHAUST C
MESH
SCREEN
#100 MESH
FILTER
PAPER
#200 MESH
SCREEN
GLASS
l CLOTH
1 BAG
L_
TT-
I
BAFFLE
-1 i PLATE
DIA. HOLE
_|RI
CO
B
FIG. 6 - EXHAUST STACK ARRANGEMENTS
-------
(2) The bag was inserted with the sealed end upstream, such that
the material adhered better to the bag. This permitted more
efficient use of the cloth surface.
(3) A #1 Whatman filter paper was supported on a 100 mesh sieve
and installed downstream of the glass filter. This was in-
tended to trap particles passing through the bag, and did
indeed collect a small amount of very high alkali dust.
(4) Baffle plates as shown on Figure 2 were installed to assist in
condensate collection and to introduce turbulence into the gas
stream for the effective filtration with the glass bag.
Discussion of Required Modification
A. Feeding System
In spite of the improvements in dust feeding alluded to above,
there was a tendency for some material to accumulate in the bottom
of the swirl feeder and in the exit pipe to the burner. This
material, unfortunately, is classified by the centrifugal action of
the swirl feeder, and analytical data presented in Table 9 in-
dicates that this fraction does not have the same composition as
the "as received" dust. Therefore, only by weighing and analyzing
this material after each run can the composition of the materials
entering the flame be known. Suggestions for dealing with this
problem are presented on page 39.
B. Separation and Collection
While these have benefited greatly from the changes instituted,
they still leave something to be desired. The condensate is still
partially escaping, as evidenced by the fact that the net potassium
content of the product fractions from the May 8, 1974 run is 8.13%
vs. the expected 10.71%. The material balance from the July 12,
1974 run shows that a total of about 25% of the material is escaping.
This is a serious discrepancy.
C. Purity of Compressed Air
The existing arrangement brought the air from the house line to a
paint .sprayer tank through an oil filter. While this filter demon-
strably remove considerable amounts of water and oil, some probably
escaped. These small liquid droplets are excellent agglomerating
sites for dust particles, and caused formation of large dust aggre-
gates which could not be fluidized. An oil filter with a high
efficiency for condensed moisture was ordered, and obviated this
difficulty.
31
-------
Table 9 - Analyses for Material Caught in Feed System
Ignition Loss
Sample _ Source _ % _ H O^Sol.
701A PPT Dust 16.29 6.99
as received
701B From Swirl Feeder 16.94 5.1
701C From Inside of Piping 18.79 4.7
to burner
32
-------
D. Velocity in System
At present, there is no convenient method of monitoring air flow in
the system. This was a drawback in our attempts to control the
system for optimum separation and collection. A pitot tube, subse-
quently ordered, allowed us to obtain useful velocity data.
E. Flow of Gas and Air to Burner
Optimization and characterization of flame conditions requires
knowledge of the flow rates of gas and air to the burner. A flow
meter valve assembly was purchased for this purpose, and aided in
optimizing the flame pattern, and reproducing it.
Operations and Product Measurements
A. Chemical Analysis
Table 10 shows the analytical results obtained on the several dust
fractions from a burn conducted May 8, 1974. Especially notable is
the considerable degree of beneficiation achieved with respect to
SO- and K?0. X-ray diffraction data obtained for these samples
(Table 11J confirm the substantial quality improvement in the
beneficiated dust. The intermediate level of beneficiation noted
with the dust fallout sample (FS-6), coupled with its high ignition
loss, suggests that the maintenance of slightly better fluidization
should avoid dropout in the dust and could improve the beneficiation
considerably.
B. Material Balance
Table 12 shows the material balance achieved on a bum conducted
for that purpose. Clearly, there is a substantial weight dis-
crepancy between input and output, amounting to nearly 25%. This
is a significant quantity, and hampered our attempts to interpret
the analytical data generated. Fortunately, K^O analyses, shown in
Table 15, indicate our potassium balance to be nearly as good as
recovery in general. A general tightening of the system was
attempted, but our fears that the alkali fume was too fine to be
collected were somewhat allayed. The discrepancies between total
K 0 recovery and water soluble K20 can thus be interpreted as a
chemical conversion in the flame, which is also a hopeful sign
borne out by the X-ray results.
C. Temperature Profile Studies
In Table 14 are shown the results of a longitudinal temperature
profile of the system. This information was valuable in rede-
signing alkali condensation surfaces for optimum collection. In
addition, transverse temperature measurements were made at thermo-
33
-------
Table 10 - Analytical Data - Run of 5/8/74
Description
Wt. Percentage of Product
Na20, %
K20, %
S03, %
Loss on Ignition* %
Na20, Water Sol., %
K20, Water Sol., %
Wt. % of Product, Ign. Basis
K20, %, Ign. Basis
K20* Beneficiation, %, Ign. Basis
Na20, %, Ign. Basis
Na20, Beneficiation, %, Ign. Basis
S03, %, Ign. Basis
S03 Beneficiation, %, Ign. Basis
Water Sol. K20 Beneficiation, Ign. Basis
Beneficiation, B, is defined as:
similarly applicable to other elements.
FS-0
as
Received
-
0.41
8.9
12.34
16.26
0.22
6.99
-
10.71
-
0.49
FS-2
Stack
Fallout
38.2
0.25
4.6
3.27
8.12
0.08
2.13
37.6
5.01
53.2
0.27
FS-4
Glass
Bag
20.6
0.52
4.5
3.21
3.48
not run
not run
21.3
4.66
56.5
0.54
44.9 -10.2
14.86
-
—
% K20
v
3.56
76.0
72.4
[as received]
% K20
3.33
77.6
—
- % K20
FS-5
Condensate
From Shell
18.1
0.70
18.8
20.0
5.96
0.57
14.71
18.3
19.99
-86.6
0.74
-51.0
21.27
-43.1
-85.7
(Benef iciated
[as received]
FS-6
Duct
Fallout
23.1
0.35
6.5
8.77
7.60
0.20
4.66
22.9
6.96
35.0
0.37
24.5
9.39
36.8
40.7
Dust5 and i
clUQ 1
-------
CM
Table 11 - X-ray Diffraction Results - May 8, 1974
Sample
No.
FS-2
FS-4
FS-5
FS-6
Compounds Identified and Quantity
Cao
large
amount
large
amount
medium
large
amount
medium
amount
CaCOs
large
amount
small
amount
medium
large
amount
medium
amount
KzSOn
small
amount
small
amount
very
large
amount
small
amount
Spurrite B-C2S
small Trace
amount
small Trace
amount
Trace
small
amount
small Trace
amount
Others
tr. a'C2S,
tr. a'C2S,
small amt.
tr. a'C2S,
-t-f Oi'P oQ
Ui 4 Vt £.J y
small amt.
C3S,
C3S,
Si02,
C3S,
C2AS,
Si02
C3A, Si02
Si02
Langbenite
CnAF, KOH
C3S
-------
Table 12 - Material Balance - Run of 7/12/74
Fraction
712A
712B
712C
712D
712E
712F
712G
712H
7121
712J
712K
712L
712M
712N
Description
Burner Pipe Condensate
T/C Deposit
Dust -200 Mesh
Stack Fallout
Glass Bag Sample
Baffle + Tee
Condensate Baffle + 2nd Duct
Duct Fallout Coarse
2nd Duct + Fallout
Burner Pipe
Swirl Feeder
Feed Screw
Main Condensate
Filter Paper
Weight
gram
0.1331
0.2851
3.893
1.9021
23.5
1.6354
0.7715
2.2314
1.0325
11.9913
13.3066
3.995
4.0063
0.4373
Ign. Loss
%
4.19
-0.78
1.44
1.41
2.17
1.70
1.93
8.82
1.59
16.68
16.05
17.26
1.24
Loss Free
wt . , gram
0.1275
0.2873
3.837
1.8753
22.990
1.6076
0.7566
2.0346
1.0161
9.9912
11.1709
3.3094
3.9566
0.4873
Total Solids (loss-free) Out
63.4474 gram
Total Loss Free Solids in grams 83.71
Recovery, % by wt., total 75.8
Recovery, % by wt., Material Entering Flame 65.8
36
-------
Table 13 - Potassium Balance - Run of 7/12/74
Fraction
As received
Alkali Concentrate
Beneficiated Dust
Classified Feed
Filter Paper
K20
Total
%
8.9
16.5
9.8
7.3
23.5
Weight of
Fraction
100 gram
6.5463 gram
32.8463
29.2929
0.4873
Total wt.
K?0, g
8.90
1.08
3.22
2.14
0.11
% K20 of
Fraction
(H?0 Sol.)
6.99
11.2
5.4
4.4
16.1
Weight K20
g
6.99
0.73
1.77
1.29
0.08
Net recovery of total
K20, %
Net recovery of
soluble K20, %
73.6
55.4
37
-------
Table 14
Temperature Profile of System
Date
5-3-74
5-8-74
Burning Condition
Burner 3-1/2" into 1st section
shell only, no dust feed,
free outlet
with 2nd shell section attached
no dust feed
free outlet
1st section shell only
with dust through burner pipe
with 2nd section attached
exhaust stack attached with
vacuum connection
1st, 2nd and exhaust stack
fully assembled with vacuum on
(burner pipe 3" into shell)
Air Temperature, °F
From
Manifold
125 @
0.27 psi
125 @
0.27 psi
100
Cyclone
Feeder
150 @
2.5 psi
150 @
2.5 psi
110 @
2 psi
Flue Gas Temperature, °F, Thermocouple No.
1
1870
1830
1700
1720
1830
1750
\
\
i
!
1
I
2 3
1910 1710
1950
1775
1750
1830
1650
1
2029
2028
2028
2028
1513
4
550 at center line
1350 1/4" from
top surface
1250 touching top
surface
1220 tip of dust
laden flame
1340 tip of dust
5*
875 1/4..
from
top
surface
890
840
840
laden flame ;
1310 1-1/2 from top
1230 1" from top
1100 1/2" from top
910 1/4" from top
; 760 touching top
1 surface
! 350 at center line
] 4" from top
top half of heat
shield open
520 @ center line 4" from top
450 1" below center line
i 320 2" below center line
1/4 heat shield open
' 410 3" below center line
• 300 3" below center line
without vacuum
00
Thermocouple No. 6
@350° F
-------
couple point #4, a likely location for the louver-type baffle planned
for alkali condensation. A plot of this temperature profile appears as
Fig. 7.
FURTHER IMPROVEMENTS
The next phase of the work was an attempt to refine the techniques
described above.
Instrumentation to measure many more operating conditions was installed.
These include temperature readings at the stack inlet and outlet, as
well as along the horizontal duct and duct wall, and gas flow rates of
the pilot and burner. Extensive modifications were made in the feeder
system with eventual elimination of the swirl feeder. Investigations of
alkali loss behavior of the kiln dust were conducted to determine ana-
lytically the initial and final temperatures of the reaction and maximum
rate of loss. These changes are described more fully as follows:
A. Cyclonic Dust Feeder
After continued trouble with feed accumulating in the bottom of the
cyclonic Cor swirl) feeder, it was temporarily abandoned in favor
of a direct feeding method.
B. Burner and Feed System (Figure 8)
A pilot burner using the house gas was installed which made light-
ing the burner easier. A gas-air premixer was installed immedi-
ately before the venturi; this accomplished a complete mixing of
the methane with the primary air. This mixture was then propelled
through the venturi, acting as the fluidizing gas for the feed.
This method of gas introduction was superior to bringing the meth-
ane in with the dust, which tended to choke the venturi. The
premixing also increased the operator's control over the flame
temperature.
An open feed system was established to provide a more uniform feed
to the flame. This involved using the old screw feeder to deliver
the dust into a funnel which is open to the atmosphere. This
arrangement eliminated the problem of flame extinction caused by
surges of the feed. It was also found that with this arrangement a
greater feed rate could be accommodated. Hot secondary air from
the condenser was introduced with the dust.
C. Condensate Separation
A five louver baffle was placed in the second stainless steel duct
(from the burner) near its outlet end. This provided a greater
collision surface on which the alkalies could impinge and condense.
Test runs showed that some alkalies were condensing, while the
baffle also knocked agglomerated particles out of the air stream.
These agglomerated particles tended to be moderately beneficiated.
39
-------
1600
200
2345
DISTANCE FROM TOP SURFACE, INCHES
FIG. 7-TEMPERATURE PROFILE OF DUCT CROSS SECTION
THERMOCOUPLE N0.4
40
-------
HEAT EXCHANGER HOT SECONDARY
.FEEDER
/SYSTEM
]|GAS
PRIMARY AIR
BURNER
HOUSE GAS
FIG. 8-BURNER AND FEED SYSTEM
-------
D. Product Collection
The fine weave glass cloth bag was treated with chemicals that
could be expected to tie up the bonding sites which inhibited dust
release. (Dimethyl dichlorosilane was chosen to inactivate Si-O-H
sites on the bag surface.) Test runs after the treatment, however,
revealed that the dust remained hard to remove.
E. Parameter Measurement
A flow meter was used to measure the volume of gas used in the
burner and pilot. A pitot tube and inclined draft gauge were used
to measure velocity heads and static pressure. The combustion
gases were analyzed for 02, CO-, CO and H20. Temperatures were
recorded in the flame and stacR areas.
Test Run Results
A. Operating Parameters
Table 15 shows the temperature variation through the system for
both the runs of 28 and 30 October. The runs were nearly iden-
tical, except that the run of 28 October had a closed feed system
(i.e., the screw feeder was directly connected to the tee above the
venturi, not allowing leakage of ambient air.) This difference had
its greatest effect on the hot secondary air from the heat ex-
changer. As can be seen in the table, the temperature of the air
was 610 F (with dust feed) when the pressure was -2.5 inches of
water (28 October) and 155°F when the pressure was -0.2 inches (30
October). Obviously the vacuum caused by the venturi pulled mainly
ambient air with the open feed system; this problem suggested the
need for future adjustments to decrease this ambient air intake and
thus increase heat exchanger air flow. Table 16 shows the gas
analysis of the combustion gases and the static pressures for 30
October 1974. The combustion gas flow of the system was in the
range of 425 ft/min., or 148 ft /rain, in the horizontal duct, as
determined by pitot tube measurements and estimates based on C02
content and burner gas flow.
B. Chemical Analysis
All samples were examined by X-ray diffraction analysis and, based
on this information, were composited. The sources of the five
composites can be seen in Figure 9. These composites were then
analyzed for loss on ignition, (LOI)* S03, Na20, and K20. The
results of these tests are given in Table 17. There is little
difference between the moderately beneficiated and moderate alkali
*The LOI represents that percent of the sample weight which is volatilized
during 15 minutes at 950 C.
42
-------
LOCATIONS OF:
0 BENEFICIATED DUST
(2) HIGH ALKALI DUST
© FEED MATERIAL
@ MODERATELY BENEFICIATEL DUST
(5) MODERATE ALKALI DUST
i FEEDER
HOT W
AIR_J
AIR
L
GASli
DUCT#I
WALLS
*-<5>-*LOOSE
>— *•
DUCT#2
WALLS 2
LOOSE
F1LJER pAPER
DUST
COLLECTOR
BAG
DUCT#3
I
! LOOSE
TO
VACUUM
AIR
INLET
FIG. 9- SOURCES OF COMPOSITE DUST
-------
Table 15 - Temperature Profile, Runs of 10/28/74 and 10/50/74
Operating
Parameters
28 Oct. 1974
Closed Feed
System
30 Oct. 1974
Open Feed
System
Temperature:
2" from burner
without dust
with dust
1870°F
1600°F
1990°F
1950°F
3" from burner
without dust
with dust
2100°F
1875°F
2120°F
2020°F
6" from burner
without dust
with dust
1880°F
1830°F
1520°F
1230 F
Hot secondary air
without dust
with dust
pressure (inches
310°F
610°F
- 2.5
155°F
155°F
- 0.2
T joint at stack
without dust
with dust
725°F
500°F
550°F
Exit gases to vacuum
without dust
with dust
Burner gas flow (ft,/min.)
Pilto gas flow (ft /min.)
Primary air pressure (psig)
310°F
140 F
0.781
0.015
0.5
370°F
0.826
0.015
1.0
44
-------
TABLE 16
Combustion Gas Composition, Run of 10/50/74
Operating
Parameters Ccont.)
Combustion gas composition
H20
C02
CO
02
Test Run of 30 October 1974
Open Feed System
Static pressure
2 ft. from T of stack
at tee of stack
stack exit
(toward burner)
(% by volume)
5.53
0.57
0
19.08
74.82
-0.06
-0.055
-0.29
45
-------
Table 17 - Beneficiation of Kiln Dust Samples
Ideal Kiln
Precipitator Dust
(as received)
Composite No.
Wt . percent of recovered
NazO, %
K20, %
S03, %
Loss on ignition @ 950 C, %
Wt . percent of recovered
ignited basis
Na20, %, ignited basis
NaaO, beneficiation, %,
ignited basis
K20, %, ignited basis
K^O, beneficiation, %,
ignited basis
SOa, %, ignited basis
503, beneficiation, %,
ignited basis
Percent relative calcination
using LOI's
Wt. percent of product
Wt. percent of product,
ignited basis
Beneficiation = % (as
High
Beneficiated Alkali
Dust Dust
(1)
-
0.
6.
12.
16.
-
0.
0.
10.
0.
14.
0.
0.
-
-
received)
41
99
34
26
49
0
71
0
86
0
0
34.
0.
4.
5.
5.
36.
0.
-30.
5.
52.
5.
62.
66.
38.
39.
79
60
8
27
53
29
64
6
08
6
58
4
0
73
94
(2)
4
0
14
16
6
4
0
-34
15
-47
17
-15
60
4
5
.43
.62
.8
.04
.42
.58
.66
.7
.82
.7
.14
.3
.5
.94
.04
Moderately
Feed Beneficiated
Material Dust
(3)
10.
0.
8.
7.
18.
9.
0.
4.
9.
6.
9.
36.
-15.
-
-
18
38
1
63
81
12
47
1
98
8
40
7
7
(4)
38.
0.
8.
7.
9.
38.
0.
-10
8
16
8
43
42
42
42
18
49
1
55
41
19
54
.2
.94
.5
.33
.9
.1
.51
.03
Moderate
Alkali
Dust
(5)
12.41
0.39
7.7
8.47
13.85
11.81
0.45
8.2
8.94
16.5
9.83
33.8
14.8
13.82
12.99
- % (composite)
% (as received)
-------
UJ
o
a:
UJ
a
M
O
o
O
if)
O)
20
19
18
17
16
15
14
13
CD
CO
CO
O
I I
10
9
8
COMPOSITE #3:
FEED MATERIAL o
COMPOSITE #5:
MODERATE ALKALI DUST
COMPOSITE #4:
MODERATELY BENEFICIATED DUST
COMPOSITE #2: HIGH ALKALI DUST
COMPOSITE #1: BENEFICIATED DUST
ot
1.0
2.0
AREA RATIO,
3.0 4.0
ACaC03
FIG. 10 - PLOT OF AREA RATIO vs. LOSS OF IGNITION
47
©AS IS
(LO.I. FROM TGA)
J
5.0
6.0
-------
dusts, while together they differ from the "as received" dust in
percent SO, and percent LOI. The beneficiated dust represents a
large portion of the recovered material; 34.79% of all recovered
material and 38.73% of all recovered material passing through the
burner. The high-alkali dust represented only a small part of the
material, 4.43% of the whole and 4.94% of the material passed
through the burner.
Table 18 shows the distribution of the recovered dust throughout
the system. The glass bag accounted for 30.16% of the material
which passed through the burner, thus illustrating its collecting
ability. The loose material in the horizontal ducts tended to
agglomerate into little balls; it is unlikely that it passed
through the burner as such; more likely it agglomerated while being
blown along the bottom of the duct.
Kiln Dust Properties
A. Thermogravimetric Study
"As received" cement kiln dust was tested in a thermogravimetric-
differential thermal analysis instrument to determine the temper-
atures of loss of combined water, carbon dioxide and alkali and the
rates and magnitudes of such losses. Samples were subjected to air
and nitrogen atmospheres and to various rates of heating. The
patterns were then analyzed to determine initial and final temper-
atures for the various losses and the amount of loss between those
temperatures.
Results revealed that there was a significant weight gain in the
range of 500-600°C when heating took place in an air atmosphere.
Then compared to the continuous loss seen in a nitrogen atmosphere,
this weight gain is in the range of 4.3-5.5% (see Table 20, LOI at
950 C). A semi-quantitative analysis revealed the presence of
sulfides, a trace being calcium sulfide and a majority a water-
insoluble mineral form of sulfide, probably pyritic. This weight
gain masks up to 25% of the combined water and carbonate losses,
introducing error into calculations of carbonate beneficiation.
Maximum rates of loss were found to occur around 770 C and 1360 C
in air and 740°C and 1154°C in N_ for carbonate and alkali, respec-
tively (Table 20). This confirms that alkali volatilization in-
creases markedly at lower tempertures when in a non-oxidizing
atmosphere. These temperatures are higher than that required in
our flame spray unit, since the large volume of moving air can
carry the volatilized alkalies away more efficiently.
B. X-Ray Diffraction Study
All samples taken were run on X-ray diffraction equipment to deter-
mine composition. An extensive study of the composite material was
made to determine if LOI and alkali levels could be predicated by
48
-------
Table 18 - Material Balance - Runs of 28 and 30 Oct. 1974
vo
Recovered material
distribution
(28 and 30 Oct. 1974)
Sample location
Screw feeder
Burner pipe, venturi § tee
Thermocouples
Loose material in duct #1
Loose material in duct #2
Loose material in duct #3
Walls of duct #1
Walls of duct #2 § baffle
Walls of duct #3
Baffles, bottom and tee
of stack
Dust collector bag
and surrounding duct
Filter paper and
surrounding duct
Top exhaust cap of stack
Composite
Number
3
3
4
5
4
5
2
4
1
1
1
4
1
percent of
recovered
weight
%
2.60
7.58
2.54
11.75
14.76
0.67
4.43
5.06
3.06
8.15
27.09
9.75
2.58
99.99
percent of
recovered
product
weight
%
-
2.83
13.08
16.43
0.74
4.94
5.63
3.40
9.07
30.16
10.85
2.87
100.00
percent of
recovered
weight
(ignited
basis.)
%
2.33
6.79
2.54
11.18
14.76
0.63
4.58
5.06
3.19
8.50
28.00
9.75
2.69
100.01
percent of
recovered
product
weight
(ignited
basis)
%
-
2.79
12.30
16.25
0.70
5.04
5.51
3.51
9.35
30.81
10.73
2.96
100.00
-------
Table 19 - Kiln Precipitator Dust (taken from can Oct. 14, 1974)
in
O
LOI Water Loss
@950°C
Sample Indentification
Carbonate "Alkali" Remaining Final
Loss Loss Loss Loss
(temp, range) (temp, range) (temp, range) (temp, range) (temp.)
2-152 17.2
packed sample, air,
5 K/min.
2-156 22.7
packed sample, N2,
5 k/min .
2-157 21.5
packed sample. N (glass
column), 5 K/min.
2-153 17.2
0.0
(20-582°C)
4.1
(20-576)
2.7
(20-570)
1.0
17.2
(582-950)
17.8
(576-895)
17.3
(570-852)
16.5
18.6 35.8
(950 - 1395) (1395°C)
10.9 5.9 38.7
(895-1242) (1242-1404) (1404)
11.1 7.5 38.6
(825-1215) (1215-1395) (1395)
19.5 37.0
losse pack, air 20 K/
min.
(20-615)
(615-1002)
(1002 - 1395)'
(1395)'
a2-153 was 1st heated to 1395°C at 20 °K/min., colled and reheated to 1302 C, held approximately constant
(to 1332 C) for 64 min., upon which constant weight was obtained with 37.0% weight loss. After initial
heating to 1395°C, loss was 25.5%.
-------
Table 20 - Rates of Weight Loss at Various Temperatures
Test
Atmosphere
Test, description
2-152
packed sample air
(heating rate:
5 K/min.)
2-153
loose pack sample air
(heating rate:
20 K/min.)
2-156
packed sample. N_
.08 1/min. N«
(heating rate:
5 K/min.)
2-157a
packed sample N«
0.8 1/min. N2
(heating rate:
5 K/min.)
Combined Water
maximum rate of
loss
(temp, of loss)
%/min. %/min.
0.10
(250°C)
0.23 0.26
(264°C) (423°C)
0.10 0.15
(240°C) (384°C)
0.15
(387°C)
Carbonate
maximum
rate of
loss
(temp . )
%/min.
0.50
(762°C)
1.90
(788°C)
0.55
(744°C)
0.60
(738°C)
Alkali
maximum
rate of
loss
(temp . )
%/min.
0.45
(1340°C)
0.76
(1395°C)
0.21
(1158°C)
0.25
(11SO°C)
a a glass column in the base of the sample chamber was used, providing
greater flow and mixing around the sample.
51
-------
area ratios of certain peaks characteristic of the materials. A
general "feel" for alkali concentration and losses was obtained and
used in determining how to composite the materials. A detailed
analysis of the area ratio of the CacO peak (39.40 , CuKotj) to a
CaO peak (53.85°, CuKa ) revealed a good means of predicting LOI
(see Figure 10). A reliable ratio for alkali was not to be found.
Conclusions
(1) Alkali Volatilization
Noting the tremendous amount of excess oxygen and the requirement
of maintaining the present flow rate to avoid excessive dropout the
recommendation mentioned previously to reduce excess oxygen was
judged unfeasible. This leaves the possibility of introducing
water vapor through the burner, entering between the premixer and
venturi, to cause metathesis of K2S04 to KOH- The water vapor
could be preheated by putting a coil around the first and/or second
horizontal duct. This would achieve two goals, a reduction of the
temperature of the shell to increase condensation of alkalies and
reduction of the gas exit temperatures. This would increase energy
efficiency and eliminate the problem of exposing the vacuum tubing
and motor to high temperatures.
(2) Product Collection
The present fine weave glass bag was collecting a larger percentage
of the product, but dust removal remained a problem even after
chemical treatment. Thus the use of a metallic mesh filter, a
"felt" layer made with extremely fine metal threads was considered
since it could withstand higher temperatures and dust released is
less likely to be contaminated by the filter media.
Installation of Metallic Mesh Filter
In consideration of the numerous difficulties of removing the kiln dust
material from the glass bag, it was decided to install a metal mesh
filter. This filter was expected to afford an opportunity for easy
release of the dust, and to tolerate higher temperature ranges than the
glass bag. The filter was installed in a vertical duct section down-
stream of the "T" shown in Figure 9 in such a way as to allow easy
removal. "Taps" were made available for measurement of pressure drops
and velocity heads throughout the system, particularly upstream and
downstream of the filter, since the efficiency of the filter is a strong
function of gas velocity and pressure drop. Provision was made, in the
reconstructed system, for measurement of CO-, 02, and CO concentrations
in the flue gas. A New York Blower Co. Fan, capable of withstanding
total system pressure drops of up to 25 cm water, was installed in place
of the vacuum cleaner, and a damper installed to permit regulation of
the air velocity. In case cooling was to be required for protection of
the filter, a fresh air bleed valve was installed upstream of the fan.
The swirl feeder, modified by shortening, was reinstated.
52
-------
Experimental Results
A. Initial Testing
Examination of Table 5 reveals that the preponderance of the alkali
in this dust sample is concentrated in the very fine particles ( 6
urn). Since the mesh openings in the fabric were 25 um, it was
judged that simple fluidization of the dust, without burning, in
conjunction with the use of the filter might be expected to result
in some beneficiation. Accordingly, the dust was passed through
the swirl feeder and drawn through the filter with the fan until
the pressure drop across the filter approached 12.5 cm HO. The
material adhering to the filter, which represented about 25% of the
total dust fed by weight, was analyzed for Na_0, K 0, and SO . The
results were as follows:
Original
Dust from filter Feed Beneficiation
K20 - 8.40% 8.90% 5.6%
S03 - 10.76% 12.34% 12.8%
Na20 - 0.39% 0.44% 11.4%
The beneficiation observed was minimal. This fact may be attributable
to smaller particles being trapped on the filter by a mat of larger
particles previously laid down during the early stages of the run.
B. Flame-Spray Runs with Metallic Filter
Four runs attempting to assess the feasibility of the metallic mesh
filter were made. Parameters were varied as required to optimize results.
The runs can be characterized as follows:
Run #1 - 6/3/75
- Filter horizontal in vertical stack section
- Primary air pressure - 0.5 psi
- Swirl feeder air pressure - 6.25 psi
- No steam addition
- Flame temperature - 1150°C (2100°F)
Run #2 - 6/11/75
- Filter horizontal in vertical stack section
- Primary air pressure - 0.5 psi
- Swirl feeder air pressure - 6.25 psi
- Steam added
- Flame temperature - 1072°-1129°C C1960°-2065°F)
53
-------
Run #3 - 6/16/75
- Filter horizontal in vertical stack section
- Primary air pressure - 0.6 psi
- Swirl feeder air pressure - 6.10 psi
- Steam added
- Flame temperature 1082°-1104°C (1980°-2020°F)
Run #4 - 6/20/75
- One 122 cm (4 ft) section of dust removed to raise filter temperature
- Primary air pressure - 0.5 psi
- Swirl feeder air pressure - 6.25 psi
- Steam not added
- Flame temperature 1046°-1115°C (1915°-2040°F)
The operating parameters for the four runs are shown in Table 21.
Velocity of the gas was observed to give maximum efficiency of collec-
tion values of about 350-500 fpm. Hence, insofar as was practical, the
velocity was maintained between these limits. The amount of secondary
air was a variable, difficult to control in our small system, and had an
appreciable effect on the excess oxygen level, as the same fuel rate was
maintained, so that C02 production was virtually constant during each
run. Only when the flame temperature showed fluctuation tendencies did
the fuel rate require adjustment. Similarly, only when necessary, the
primary air was adjusted.
The dust samples removed from the system after each of these runs were
subjected to X-ray diffraction examination. These X-ray patterns
revealed a degree of success in removing alkali salts, but the actual
extent of beneficiation was disappointing. Evaluation of the magnitude
of characteristic peaks is found in Table 22. It is interesting to note
that the most effective separation achieved was for run #3, which had a
low efficiency of collection. This observation is, unfortunately,
rather consistent.
The material balance for these four runs can be found in Table 23. The
total recovery figures are indicative of the large openings in the mesh
of the filter. Success in using metallic filters for separation of
alkali-rich fume from beneficiated kiln dust probably requires higher
temperatures, optimized velocities, and finer mesh size.
High-Temperature Run
Just prior to the completion of this project, a final run of the equip-
ment was made in which the filter was permitted to operate at increased
temperatures. The path between the burner and filter was greatly
shortened, so that the opportunity for dust and alkali to become reunited
prior to collection was reduced.
54
-------
Table 21 - Process Parameters for Flame-Spray Runs
Date
Mass of Dust in, gm
Duration of run, min.
Pressure Drop across filter,
initial
Pressure Drop across filter,
final
Filter Temp., max., C (°F)
Filter Temp., min., °C (°F)
Temp. Downstream of Filter,
Range °C (°F)
% 02, Range Dry Basis
% CCL, Range Dry Basis
Velocity M/min.
(Ft/Min.), Range
Gas Volume M /min.
(CFM), Range
1
6/3/75
45.60
124
5.1 cm
(2")
14.7 cm
(5.8")
235°(455°F)
110°(230°F)
33-143°
(92-290°)
14-21
0.8-4.5
(109-304)
(359-997)
3.5-9.9
(125-348)
Run
2
6/11/75
26.56
100
7.62 cm
(3»)
13.2 cm
(5.2")
196° (385°)
165° (329°)
99-124°
(210-255°)
17.5
2.6
None
None
No.
3
6/16/75
38.82
225
5.59 cm
(2. 2")
10.92 cm
(4.3")
(280°F)
(230°F)
(140-210°)
18.6
1.8
116 (381)
3.8 (133)
4
6/20/75
35
292
5.1 cm H_0
(2")
13.0 cm
(5.1")
277° (530°)
154° (310°)
—
10.5-20
0.75-6.0
121-162
(396-530)
3.9-5.2
(138-185)
55
-------
Table 22
tn
X-Ray Diffraction Results
Runs of 6/3, 6/11, 6/16, 6/20/75
CaO
Run #1
Coarse Fallout W*
Filter
Easily removed SS
Intermediate SS
Difficult SS
Condensate
First Section SS
Subsequent Sec. SS
Run «2
Coarse Fallout S
Condensate SS
Filter SS
Run #3
Condensate SS
Steam Leachate
Filter
Easily Removed SS
Difficult S
Run #4
Coarse Fallout MS
Condensate SS
1st Section Filter
Easily Removed SS
Intermediate SS
Difficult S
Dust Passing Filter MS
SS = Very strong
S = Strong
MS = Moderately Strong
M = Moderate
MW = Moderately Weak
W = Weak
WW = Very Weak
tr = Trace
CaCOj
SS
W
W
SS
W
W
SS
MW
S
W
SS
WW
SS
SS
MW
WW
MS
SS
S
(very
(large
etc
Ca(OH)2
--
__
WW
W
--
M
--
__
__
tr
tr
tr
tr
W
large amount
amount)
V
W
MS
MS
MS
SS
S
S
MS
W
--
S
MS
MS
MN
S
MS
MW
MW
BC2S C3S Spurrite
W
MS
MS — tr
S
MS tr
SS MW
MW W —
MW WW
MW tr? tr
—
__
S
MS
MS WW tr
MW tr tr?
MS W —
MS WN tr
MS
W
KC2JS12 K2S04 KC1 K2Ca(S04)2 K2Ca(S04)2-H20 K3N6(S04)2 CaS04
W
WW
W
S
SS
MW MS
W MW
MS SS
tr W
W SS
W
W
MW
MW
SS
W MW
W W
WW MW
W W
tr
tr?
tr
MW — tr
W
tr MW
W MS WW
W — K
tr MS MS
SS — W
tr -- tr
W
tr
MW MW
tr
WW
MW tr?
W
tr
W tr
WW WW
W
tr
—
W
—
—
tr WW
—
tr
tr
tr? tr?
tr
Si02
MS
W
W
S
W
MW
S
W
W
tr
W
tr
tr
SS
WW
tr
W
MS
MS
of material)
-------
Table 25 - Material Balance - Runs of 6/5. 6/11, 6/16, 6/20/75
Date
Dust in, grams
Dust Out
Feeder + Burner P.PE
Thermocouple Deposit
Coarse Dust Fallout
Condensate
1st Section
Subsequent Sections
Vertical Stack
Filter
Early Removed
Intermediate Removed
Late Removed
Condensate from Steamruns
Filter Paper Catch (where
applicable)
Dust caught beyond Filter
Total ("As is" Basis)
% Recovered (Ignited Basis)
1
6/3
50.0
4.40
0.0802
5.3400
1.0612
1.7650
0.3879
9.0549
2.2593
1.6858
Run
2
6/11
35.0
8.44
0.0414
1.9700
1.3204
0.9858
-
None
it
"
u
No.
3
6/16
50.0 .
11.62
0.2420
1.7552
2.0448
1.3021
0.5520
2.91
_
0.82
4
6/20
50.0
14.95
0.0759
3.1523
1.6431
1.1131
-
1.11
_
9.06
26.0343
55.53
2.4500
15.2076
43.84
0.7214
21.9675
44.72
0.2904
31.3948
63.15
57
-------
Table 24 - Process Parameters and Material Balance
High-Alkali Kiln Dust Run of 7/25/75
Total Dust In, grams
Dust Out, grams
Filter Catch
Condensate
Coarse Dust Fallout
Filter Paper
Feider £ Pipe
Collection Efficiency
Wt. in grams
As Is Basis Ignited Basis
35.00
15.00
0.1965
1.9900
0.23
14.31
28.71
14.57
0.19
1.72
0.17
11.66
98.6%
Duration, min. 140
Flame Temperature, range, °C (°F) 1043-1154°(1910-2110°F)
Filter Temperature - Upstream, range 393-527°C (740-980°F)
- Downstream, range 279-410°C (535-770 F)
Exit Gas Temperature
118-150 C (244-302"F)
Pressure Drop 0.3-8.9 cm HO (0.1-3.5" HO)
58
-------
The five louver baffle was installed between the flame and the filter as
a protective device, and a piece of filter paper was installed down-
stream of the filter to catch escaping alkali concentrate.
Process parameters for this experiment, together with a material balance
for the system are given in Table 24.
Alkali Removal Determination
The alkali contents of the two samples generated in sufficient quantity
to permit analysis were as follows:
Sample % Na20 (Ig. Basis) % ^Q (Ig. Basis)
Coarse Fallout 0.39 7.96
Metal Filter 0.27 4.36
Untreated 0.49 10.71
These figures permit calculation of the beneficiation (defined earlier
in Table 10), as follows:
Coarse Fallout
% B (K20) = 25.7% % B (Na20) = 20.4%
Filter
% B (K20) =59.3% % B (Na20) = 44.9%
These results show the first and only instance of simultaneous high
collection efficiency and good separation. The keys to the success of
this last experiment would appear to be the low velocities and high
filter temperature.
In summary, then, there is evidence that kiln dust can be freed of a
substantial portion of its alkali content, and at the same time be
collected with high efficiency, only when careful control is exercised
over the temperature, flame characteristics, and gas velocity parameters
in the system. Further study is in order, to assess more thoroughly the
realizable separation of alkalies from kiln dust, using better insulation
techniques, and efficient particulate collectors designed to trap the
alkali sulfate fumes.
59
-------
REFERENCES
1. R. Beetz and K. Lambecht,"Apparatus for Continuous Separation of
Alkalies from the Offgases of a Cement Kiln," West German Patent
Application 1 813 762, Nov. 6, 1969.
2. F. Keil, "Process and Apparatus for Minimizing the Alkali Content of
Cement Clinker," West German Patent Application 1 471 279, Nov. 28,
1968.
3. Portland Cement Association, "Energy Conservation in the Cement
Industry," Report to the Federal Energy Administration, in press.
4. J. J. Shideler, "Waste Kiln Dust as a Fertilizer and Soil Condi-
tioner, A Literature Review," Portland Cement Association, Mill
Session Papers M-197, Sept. 20-22, 1971, San Francisco, Calif.
5. W. J. McCoy and R. W. Kriner, "Use of Waste Kiln Dust for Soil
Consolidation," loc. cit.
6. R. Packard, Portland Cement Association, unpublished work.
7. A. A. Spinola, "Treatment of Acid Waste Waters with Cement Kiln
Flue Dust," Portland Cement Association, Mill Session Papers M-197,
Sept. 20-22, 1971, San Francisco, Calif.
8. T. A. Davis and D. B. Hooks, Southern Research Institute, Birmingham,
Alabama, 35205, "Disposal and Utilization of Waste Kiln Dust from
Cement Industry," Environmental Protection Agency Project No. R-801
872, May 1975.
9. Among others;
H. Ritzmann, "Cyclic Phenomena in Rotary Kiln Systems," Zement-Kalk-
Gips, 1971, Vol. 8, p. 338.
H. Ihlefeldt, "Measures for Reducing the Alkali Cycle in the Lepol
Kiln," Zement-Kalk-Gips, 1972, Vol. 1, p. 15.
P. Weber, "Alkali Problems and Alkali Elimination in Heat Economizing
Dry Process Kilns," Zement-Kalk-Gips, 1964, Vol. 8, p. 335.
60
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO.
'EPA-600/2-76-194
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
ELIMINATION OF WATER POLLUTION BY RECYCLING CEMENT
PLANT KILN DUST
5. REPORT DATE
July 1976 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
7.AUTHOR R> Greening) F. M. Miller, C. H. Weise,
H. Nagao
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Portland Cement Association
5420 Old Orchard Road
Skokie, Illinois 60076
10. PROGRAM ELEMENT NO.
1BB610: 01-04-02A
11. CONTRACT/GRANT NO.
S-802196
12. SPONSORING AGENCY NAME AND ADDRESS
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
PTNr.TNNATT OHIO 45?68
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
is. ABSTRACT The effortg Of this study have included determination of the feasibility
of separation of cement plant kiln dust, into fractions which are alkali-rich and
alkali-poor, with various pyroprocessing techniques. These have included fluidized
bed and flame-spray methods.
The study included the investigation of the effect of varying a number
of process parameters on the achievement of four goals: 1) Effective feeding of
the kiln dust raw material; 2) Maintenance of flame stability and of adequate
temperature to achieve alkali volatilization; 3) Achievement of separation of the
two aforementioned fractions until collection was complete; 4) Efficient collection
of the two kiln dust fractions.
The parameters varied were: The feeding system and fluidizing arrangement,
the portion of the system designed for alkali entrapment, the dust collection
mechanism, the temperature of the flame and. collection system, and the collecting
medium itself.
Although the first two objectives were generally met, there seemed to be a
degree of mutual exclusivity in the third and fourth objectives in some cases.
However, optimization of operational parameters resulted in simultaneous achieve-
ment of all four goals. A theoretical study of the operative chemical parameters wa
made, and suggestions for achievement of these goals in other ways have been pre-
pared. In light of the energy shortage which^is a problem now and for the forseeabl
—ruture, some ot these suggesgidr^ --—a^--" ...-.^--c. .—•—. ._
17. hy pvroprocessing
IS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
Flue Dust
Portland Cements
Water Pollution
Flame Spraying
Sulfates
Alkali Metal Compounds
13/B, 13/H,
11/D
18. DISTRIBUTION STATEMEI
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
69
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
61
------- |