DoE
EPA
United States
Department of Energy
Division of Solid Fuel
Mining and Preparation
Pittsburgh PA 15213
FE 8887 1
U S Environmental Protection Agency Industrial Environmental Research EPA 600
Office of Research and Development Laboratory September 1978
Research Triangle Park NC 2771 1
Magnetite Recovery in
Coal Washing by High
Gradient Magnetic
Separation
Interagency
Energy/Environment
R&D Program Report
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects; assessments of, and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
EPA REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved
for publication. Mention of trade names or commercial products does not con
stitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
Price: Paper Copy $5.25
Microfiche $3.00
USOOE TocbUcol Intonation CM*, Odk ndo.,
-------
FE-8887-1
(EPA-600/7-78-183)
September 1978
Distribution Category UC-90b
Magnetite Recovery in Coal
Washing by High Gradient
Magnetic Separation
by
E. Maxwell and D.R. Kelland
Massachusetts Institute of Technology
Francis Bitter National Magnet Laboratory
Cambridge, Massachusetts 02139
EPA/DoE Interagency Agreement No. DXE685AK
Program Element No. EHE623A
EPA Project Officer: David A. Kirchgessner DoE Project Officer: Richard E. Hucko
Industrial Environmental Research Laboratory Division of Solid Fuel Mining and Preparation
Research Triangle Park, NC 27711 Pittsburgh, PA 15213
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
and
U.S. DEPARTMENT OF ENERGY
Division of Solid Fuel Mining and Preparation
Pittsburgh, PA 15213
-------
MAGNETITE RECOVERY IN COAL WASHING BY
HIGH GRADIENT MAGNETIC SEPARATION
Final Report
October 1977
Contract ET-76-C-01-8887
E. Maxwell
D.R. Kelland
Prepared by
Francis Bitter National Magnet Laboratory
Massachusetts Institute of Technology
Cambridge, Massachusetts 02139
Prepared for
Coal Preparation and Analysis Group
United States Bureau of Mines
Department of the Interior
Pittsburgh, Pennsylvania 15213
-------
ABSTRACT
A research program at the Francis Bitter National Magnet Laboratory has de-
monstrated successful recovery of magnetite from mixtures of magnetite and
coal — like those found in a coal-washing circuit — by High Gradient Magnetic
Separation. High values of magnetite recovery were achieved at reasonably high
material throughput rates with little coal found reporting to the magnetics.
A single-stage separator incorporating a new matrix design was used at rates
up to 4.4 tons of solids per hour per square foot of matrix cross-section
(300 gpm/ft2). At this throughput rate, more than 99% of the magnetite was
trapped along with less than 5% of the coal. Magnetic field values no
higher than 6 kilogauss were used to achieve these results, a value well
within the range of present commercial magnet designs.
-------
ACKNOWLEDGEMENTS
The investigators acknowledge the support of the U.S. Bureau of Mines,
Coal Preparation and Analysis Group, particularly Mr. Albert W. Deurbrouck,
who encouraged us and provided much information. Also, Mr. Richard E. Hucko,
who served as contract monitor, has helped with analysis work and supplied
technical assistance. Mr. Charles W. Statler of the Penelec Co. provided
the coal samples and flow sheets for the cyclone circuit of the Homer City
Coal Cleaning plant. Mr. H. Eugene Harris and Mr. Luis Riva of Eastern As-
sociated Coal Corp. Research Laboratory generously offered their advice,
prepared the float-sink products from a coal sample, and performed several
analyses. The magnetite was provided by the Reiss-Viking Corp. of Bristol,
Tennessee.
-------
TABLE OF CONTENTS
PAGE
INTRODUCTION AND SUMMARY 1
FEED MATERIAL 6
Coal 6
Magnetite 11
APPROACH 19
APPARATUS AND EXPERIMENTAL METHOD 24
ANALYSIS METHODS 27
Magnetics 27
Iron 28
Ash 28
Particle Size Analysis 28
RESULTS 30
Magnetite Capture and Coal Recovery as a Function
of Magnetic Field 30
Matrix Loading When Samples Consist of Both Coal
and Magnetite 33
Magnetite Recovery at Cyclone Input Slurry Density 35
Cyclone Products Tests 37
Effect of Dispersant 40
Restricted Flow Rates 41
Effect of Dispersant and Other Effects 41
Particle Size Analysis 45
CONCLUSIONS AND RECOMMENDATIONS 53
Conclusions 53
Recommendations 54
REFERENCES 56
APPENDIX 1: Ash Balances for Two Sets of Runs
APPENDIX 2: Typical Iron Balances
APPENDIX 3: Analysis Correlations
-------
LIST OF FIGURES
Figure Page
1 Schematic of continuous Carousel High Gradient Sepa-
rator. 5
2 The Carousel High Gradient Magnetic Separator (Photo-
graph) . 5
3 Magnetics capture (%) from the feed coal as a function
of flow rate. 10
4 Magnetite capture (Davis Tube) as a function of mag-
netic field. 18
5 Matrix loading (magnetite) as a function of feed weight. 18
6 Schematic of the Homer City Coal Cleaning Plant. 20
7 Homer City Station Coal Preparation Plant heavy medium
cyclone circuit flow rates. 21
8 Schematic of experimental apparatus. 25
9 Calibration curve for the integrating magnetometer. 27
10 Ash and magnetics measurements for the non-mags as a
function of magnetic field. 32
11 Magnetics capture as a function of field. 32
12 Magnetics capture at 3 kG as a function of feed weight. 33
13 Magnetics recovery from samples twice the usual feed
weight. 36
14 Magnetics capture as a function of feed slurry density. 40
15 Magnetics capture as a function of magnetic field with
no dispersant used. 44
16 Magnetics recovery as a function of magnetic field for
the runs without dispersant. 44
17 Ash and iron values for non-mags and mags as a function
of magnetic field. 45
18 Electron microscope photographs of non-mags particle
and chained magnetite particles. 47
iv
-------
LIST OF FIGURES CONT.
Figure Page
19 Cumulative particle size distribution of magnetite
before and after HGMS. 49
20 Magnetite particle size distribution with H = 3.0 kG. 49
21 Magnetite particle size distribution with dispersant
and with H = 0.5 kG. 51
22 Particle size distribution for magnetite at 3.0 kG with
dispersant. 51
23 Magnetite particle size distribution at 3.0 kG with dis-
persant and matrix filled before the run. 52
-------
LIST OF TABLES
Table Page
1 Feed Coal Size Analysis 7
2 Peed Coal Analysis 7
3 Magnetic Separation of Feed Coal 9
4 Magnetite Grades 12
5 Magnetite Size Distribution, HIAC Analysis 13
6 Analysis of Reiss-Viking Magnetite 14
7 HGMS Results With Reiss-Viking Magnetite 16
8 Effect of Loading Matrix With Reiss-Viking Magnetite 17
9 The Effect of Magnetic Field 31
10 Effect of Matrix Loading 34
11 Effect of High Slurry Density 36
12 Separation Tests on Sink Product (1.30) 38
13 Separation Tests on Float Product (1.30) 39
14 Effect of Restricting Flow Rate 42
15 Magnetite Capture Without Dispersant 43
16 Average Particle Size and Weight - HIAC 46
-------
-1-
INTRODUCTION AND SUMMARY
The objective of this research program is to apply High Gradient Mag-
netic Separation (HGMS) to the problem of the recovery of fine magnetite from
dense-medium circuits in coal preparation plants. Fine magnetite mixed into
a water suspension is commonly used in coal preparation to create a pseudo-
specific gravity great enough to effect separations between coal and associ-
ated impurities. The magnetite must not only be reclaimed but must also be
cleaned to prevent coal and impurities from building up in the washing cir-
cuit. The reclaimed coal fines have economic value. Studies have been made
of the performance of dense-medium cyclones treating coals down to 100 mesh
at the Bureau of Mines. Sharp separations have been achieved by reducing the
top size of the feed to the cyclone. Such recent advances in the cleaning
of finer sizes of coal required that entire coal-magnetite streams, which are
normally run over drain and rinse screens for separation of a large portion
of the magnetite, be fed directly to magnetic separators. The capacity of
conventional drum-type separators is such that multiple units are required
to handle the flow. HGMS may afford the same capability in fewer separating
units.
High gradient magnetic separators are either of the batch or continu-
ous type. The batch separator consists of a canister of appropriate size,
containing the matrix, enclosed by a solenoid which produces the required
magnetic field. Commercial units of this type capable of handling 60 tons
of solids per hour have been made for the kaolin industry. In our labora-
tory we have several small experimental separators which can be used in a
variety of magnets at fields up to 100 kG. For handling magnetite, however,
fields of the order of 1-20 kG are adequate, and can be realized with good
conventional magnet technology using iron bound electromagnets.
Continuous separators of the high gradient type have been built on
an experimental basis. We have used one such unit, the prototype Carousel
Separator which can handle about 3 tons (dry wt.) of feed per hour. This
device has been used to demonstrate the applicability of high gradient
-------
-2-
separation to the beneficiation of oxidized taconite in a cooperative pro-
gram with the Hanna Mining Company under the sponsorship of the RANN Division
of the National Science Foundation. Other pilot-plant size continuous sep-
arators have been built with matrix size up to 15 feet in diameter and cap-
acities of several hundreds of tons of feed per hour. Figure 1 is a schematic
of a Carousel Separator and Fig. 2 shows the prototype set up as a pilot
plant device in an iron ore laboratory.
Batch separators are practical and are used on a commercial scale
where the magnetics fraction to be removed from the slurry amounts to only
a few percent of the total throughput. The matrix is periodically back-
washed to remove the magnetics from the matrix. During the backwash in-
terval the slurry is either diverted to a surge tank or to another magnetic
separator.
If the magnetics fraction is high, a continuous separator may be
called for. In the continuous separator the matrix is in annular form,
(resembling the carousel in the carousel slide projector) and rotates in
and out of the magnetic field. The washing is done on that portion of the
matrix which is outside of the field. The magnetics are then continuously
collected and continuously washed.
The throughput capacity of HGMS separators which have been used at
j
this laboratory is as high as 500 gpm/ft - This flow was used to strip ka-
olin clay from magnetite following agglomeration of these materials in water
purification work. By contrast, multiple drums have possible feed rates of
50-100 gpm/ft of width. There is a definite economic advantage of higher
flow rates in reducing the machine size necessary for use in a coal washing
plant.
Drum separators are affected by the amount of magnetics in the slurry.
The recovery decreases in dilute slurries — an effect not seen in high
gradient separators. Also they are limited to 20% solids or less in the
feed and 60-70 gpm/ft of drum width if the non-magnetics content is re-
quired to be less than 10%. This corresponds to a limit of 2.9 tons per
-------
-3-
hour per foot of drum width. For a 30" drum this is 1.15 tph/ft width per
foot diameter.
Two hundred twelve experimental runs were made in this study using
an Upper Freeport coal (9 x 100 mesh) slurried in water together with
(96% -325 mesh) commercial magnetite. A new matrix design, which was used
for all of the work described in detail in this report, was developed as
part of this project. Single passes were made through the separator with
a feed equivalent to the feed to a heavy-medium cyclone. Some separations
were performed on simulated cyclone overflow and underflow streams, and
both the coal and magnetite were fed alone to the separator to determine
performance characteristics.
2
The throughput rates in this study varied from 1 to 4.4 tph/ft of
matrix with the best results at the highest rates. The least amounts of
non-magnetics in the magnetic concentrate occurred at the highest through-
put values and high levels of magnetite recovery were maintained.
Typical figures in a series of runs in which the magnetic field was
varied from 2 to 6 kilogauss were 99 to 100% magnetics recovery with an
average value of 4.8% of non-magnetics in the magnetics. In these runs
the flow rate was roughly 300 gpm/ft and the feed was a slurry with a
solids density of 15.6%. The throughput average value was 4.1 tph/ft of
matrix calculated from the total solids passed in the time taken to col-
lect the total non-magnetic fraction of the slurry. This fraction includ-
ed some extra water following the slurry. This was done to maintain a
head for constant flow. If the time was known for the slurry alone, the
throughput calculation would result in a higher figure. In further work
larger samples will be used and this throughput value obtained.
The head mentioned above was only a few inches above the matrix.
Higher heads will be used in the future which will result in higher through-
put and flow rate values. Continuous rotating HGMS separators have separate
slurry and wash water flows so that, in practice, if these were used for
this application, the throughput would be exactly determined by the slurry
flow rate.
-------
-4-
One possible application of HGMS in coal beneficiation would be as
a secondary magnetic separator following a primary drum for magnetite re-
covery. Since HGMS is particularly well suited to the recovery of particles
with sizes ranging down to below 1 micron, the magnetite slimes and fine
particles that cannot be recovered by drums would be retained in the High
Gradient Separator. While a magnetic field strength of 5 or 10 kilogauss
or more is required for the capture of paramagnetic particles such as those
found in oxidized taconite iron ores, previous work by us has shown that
1 kG is sufficient to trap 100% of laboratory samples of magnetite. The
size range of these samples was 0.6-2.5 microns.
An alternative application would be the use of HGMS as a primary mag-
netite separator. Used this way it would replace drums altogether. The
amount of magnetite retained would be much greater than it would be in its
use as a secondary separator but it has been shown that HGMS separators can
retain 80% or more of the solids in a 40% density stream of iron ore without
plugging. Separation was achieved even at a high recovery level (96% Fe
recovery). It is indicated, then, that high recovery of magnetite would be
possible with the required cleaning even for the wider range of particle
sizes found in the feed to a primary drum.
It remains to be seen whether a batch or continuous separator would
be desirable in an industrial magnetite separation process. The batch sep-
arators would be simpler and cheaper and if the wash water containing the
magnetite is dilute enough they would be practicable. In this introductory
study we used batch separators. Past experience in iron ore processing, with
the Carousel Separator, has shown that it is possible to correlate batch sep-
arator performance with Carousel Separator performance under corresponding
conditions of magnetic field and flow rate.
-------
-5-
FEED ISLURRY)
FLUSH IWATIRi RINSE WATER)
MAIHIX
COMPARTMENT
MORE «AGNET,C
PARTICLES
IOISCHARGEI
SIEEL
RETURN FRAME
ENERGISING COILS
Figure 1. Schematic of continuous Carousel High Gradient Magnetic Sepa-
rator.
Figure 2. The Carousel High Gradient Magnetic Separator with its power
supply at a mining company research laboratory. Feed and
wash pipes enter the Carousel at the top.
-------
-6-
FEED MATERIALS
Coal
The feed coal for this project consisted of a blend of two Upper Free-
port coals (Indiana Co., Pa), blended in the ratio expected to be used at
the Homer City Coal Cleaning Plant. The coals were collected over several
hours to avoid variations from such causes as differences in mining machine,
sharpness of bits, etc. They were put together by the R&P Coal Co. in the
ratio of 575-600 pounds Lucerne #6 Mine coal (Helvetia Mining Co.) and
275 pounds Helen Mining Co. coal. A 200-pound sample of 2 mm x 100 mesh
(149 y) was made from these coals and shipped directly to us on September
20, 1976.
A screen analysis was made by Eastern Associated Coal Corp. Research
Laboratory in Everett, Massachusetts and is shown in Table 1. The
largest weight fraction is between 14 and 28 mesh (1-19 mm x 595 y ) (Tyler)
which means that the separation process and the matrix design must be such
that particles this large can pass the matrix easily. Of course, the same
matrix must trap magnetite which contains particles even smaller than 1
micron.
A 20-pound sample of the coal was split at 1.30 specific gravity by
Eastern for use as discussed later. The results of this split were 48.4%
float and 51.6% sink. Analyses of the feed coal were performed to deter-
mine percent ash, % moisture, % iron, and % Davis Tube magnetics. Re-
sults are shown in Table 2. The moisture was taken on several samples by
the standard ASTM method. All other analyses are on an as-received basis.
Magnetics were measured by us using a magnetometric method and Davis
Tube magnetics were measured by Mr. Richard Hucko at the Bureau of Mines
and the value is exactly zero for the feed coal.
-------
-7-
TABLE X
Feed Coal Size Analysis
Screen Size
1/8" x 14 mesh
14 x 28 "
28 x 48 "
48 x 100 "
100 x 200 "
- 200 "
Wt. %
21.0
34.6
24.6
16.9
1.5
1.4
As Received
Float Product*
Sink Product*
TABLE 2
Feed Coal Analysis
MAGNETICS
FLOAT* SINK* MOISTURE ASH IRON MAGNETOMETER (DAVIS TUBE)
48.4 51.6
1.3 13.2 2.33
2.4 0.28
29.6 5.0
0.0
(0.0)
•Specific Gravity =1.30
-------
-8-
Magnetics measurement is one way of locating magnetite lost in the non-
mags stream and, as will be seen in the results section, this loss can be
very low. We believe that the Davis Tube does not necessarily indicate all
of the magnetics which might be trapped in a HGMS device, so a more sensitive
magneto-metric method was devised under this contract.
The feed coal contained no magnetics discernible by either of these
methods. However, the HGMS results on the feed coal at two flow rates and
two magnetic field values showed that some material was trapped by the matrix.
The results are shown in Table 3 where, for full flow, about one-third of a
percent by weight was trapped (Runs 201, 202). The amount was essentially
independent of field and therefore is considered to be caused by mechanical
trapping.
At the fast flow rate, material trapped magnetically would be the more
strongly magnetic particles or larger ones with a large magnetic moment. The
higher field should result in trapping some smaller or more weakly magnetic
particles if the trapping were caused by magnetic forces. But there is no
increase between 3 and 6 kG.
At the slower flow rate a small increase is seen in magnetics (from
2.9 to 3.5 percent weight) but there is some material balance error beyond
the average for these runs so that little can be concluded except that we
see little or no field effect and, hence, attribute this trapping to me-
chanical causes. This is not strange to those with experience in HGMS. In
fact, the low level of material trapping when coal alone is fed to the sep-
arator is important because it indicates the ability of the separator to
successfully pass large particles, a necessary but not sufficient condition
for success in making the coal-magnetite separation. The slower flow runs
were made to provide information of this kind, even though faster flow rates
are more desirable and were generally used in our tests.
-------
-9-
TABLE 3
Magnetic Separation of Feed Coal
% WEIGHT % IRON FLOW
RUN MAGNETIC VELOCITY SAMPLE RESTRICTOR
FIELD (kG) (cm/sec) non-mags mags non-mags mags (gal/min)
199 3.0 32.6 300.00 gm coal (feed) 97.1 2.9 3
1305 ml H20
18.7% solids by weight
200 6.0 32.5 SAME 96.5 3.5 2.01 6.71 3
201 3.0 117.4 SAME 99.7 0.3
202 6.0 148.4 SAME 99.7 0.3
-------
-10-
Figure 3 is a plot of the percent weight capture in these runs. The
straight lines drawn allow an estimate to be made of the capture due to coal
alone at several fields if one dares to interpolate. We should note, however,
that the capture characteristics of the magnetite differ substantially from
runs for which it is the sole feed material and those in which it is mixed
with coal. One might expect the same to be true for the impurities in the
coal, so we cannot place much emphasis on this attempt to keep track of ma-
terials that pass through the separator. We will make limited use of this
information later on, however.
4.0
3.0
LU
2.0
1.0
10
COAL
i i i i i i i I
i i i i i i i
100
FLOW RATE/ CM/SEC
1000
Figure 3. Magnetics capture (%) from the feed coal as a function of
flow rate. Feed = 300 gins.
-------
-11-
Magnetite
The magnetite for this project was obtained from the Reiss-Viking Corp.
in Bristol, Tennessee. The first sample received was 40 pounds of Grade E
(96-97 % -325 mesh, 90 % -30y, 36 % -5y) and the second was 100 pounds of
the same material. This grade of magnetite is required for washing coal re-
duced to 100 mesh and is the grade expected to be used at Homer City. Only
one brand of magnetite was used on the advice of Bureau of Mines personnel.
The magnetite is mined, crushed, and ball mill ground until the 325 mesh re-
quirement is met. It is dried by the supplier. This grade is stated to have
91% magnetics by Davis Tube.
We include a table of magnetite grades and typical size composition of
these (Table 4) . This was provided by Mr. A.W. Deurbrouck of the Bureau of
Mines. The largest fraction is between 10 and 20U in particle size (29.5%)
with the 5-10y range close behind (.28.5%). Along with this, Table 5 shows a
typical size analysis by weight from our HIAC particle size analyzer. This
shows a similar distribution except that the largest fractions are in the
S-lSp and 5-8y range. This indicates more fine particles than the screen
analysis. The density, assumed shape, and method of determining weight frac-
tion may differ among types of analyses and account for variations between
distributions .
Other analyses of the Reiss-Viking (R-V) magnetite included iron, ash,
sulfur, float-sink, magnetometric and photographic. In addition, a number
of runs were made on the HGMS with various matrices and conditions to char-
acterize the behavior of the magnetite before it was used to make feed samples.
These analyses are summarized in Table 6. It is interesting to calculate the
weight of the ash if we assume the simplest conversion of Fe3O4 to Fe203 ac-
cording to the reaction
2 Fe3°4 + \ °2
the weight ratio would be 1.035, close to our experimental value of 1.024
(1% difference) . This calculation assumes, of course, that the material
which is not Fe,O. does not change weight and we do not know that.
-------
-12-
TABLE 4
Magnetite Grades
PARTICLE SIZE
PERCENT MINUS
325 MESH
55 to 70
90
96
98
Average Size
FOOTE
MINERAL
CO.
A
B
E
F
Composition of
MINERAL
MILL
CO.
2
1
6
-
Magnetite Used
VIKING
CORP.
55/70
90
96
989
in Dense-Medium Separators
SIZE COMPOSI-
TION, MICRONS
COMPANY X COMPANY Y COMPANY Z
GRADE B GRADE E GRADE B GRADE E GRADE B GRADE E
90
74 =
(200 mesh)
60
44 =
C325 mesh)
30
20
10
5
97.0
95.0
92.5
87.5
74.0
62.0
36.0
8.0
98.0
97.5
97.5
96.0
84.5
74.5
44.5
12.0
PERCENT
99.0
98.0
97.0
90.0
68.0
53.5
32.5
7.5
PASSING
100.0
100.0
100.0
97.5
86.5
66.5
37.5
11.0
99.0
97.0
94.5
85.0
67.5
53.5
32.0
7.0
99.5
99.5
99.0
96.5
84.0
69.0
39.5
11.0
-------
-13-
TABLE 5
Magnetite Size Distribution, HIAC Analysis
PARTICLE WEIGHT % BELOW
SIZE, y STATED SIZE
60 100
20 91.6
15 88.7
8 58.2
5 30.0
3 7.0
-------
-14-
TABLE 6
Analysis of Reiss-Viking Magnetite
ANALYSIS % BY WEIGHT
Iron 63.3
Ash 102.4 (± 0.1)
Sulfur 0.18
Float (Sp. Grav. =1.30) 0.0
Sink (Sp. Grav. = 1.30) 100.0
Magnetics (Davis Tube) 90-91
Retention (HGMS) 96
-------
-15-
In our magnetometric analysis we took the magnetite as a standard at
100% magnetics and in the HGMS we repeatedly collected about 96% magnetics
when we ran the magnetite alone at sufficient field (2-6 kG) and so believe
the Fe,0 content to be higher than that indicated by chemical analysis.
The results of the HGMS Runs 101-110 using a new double screen
matrix are shown in Table 7 and in Fig. 4. Note the concentration of mag-
netics from 91% in the feed to 96% in the magnetic product.
Further, the Reiss-Viking magnetite was subjected to a Mossbauer an-
alysis to determine its molecular structure. This was performed by Dr.
Richard Frankel of this laboratory and the results are straightforward.
The spectrum could be read to within 5%. All of the iron in this sample
appeared to reside in sites characteristic of the Fe3O^ structure/ so that
at least 95% of the iron is contained in the Fe3O^ form to within experi-
mental error.
Table 8 and Fig. 5 show the results of loading tests on the double
screen matrix with magnetite alone. The sample weights range from 75 to
600 gms. The first six runs in the table cover this range and show that
recovery is independent of loading. A curve has been drawn for these
points which represent runs in which a dispersant was used. The next two
runs (150 and 151) were done at 25.6% solids density, higher than the pre-
vious series except for the last point (113), and still there is no ap-
parent loading effect. Repeating these two runs, however, without the
dispersant (153, 154) shows a higher capture for the high loading point and
about the same for the 300 gm sample as with the dispersant.
Looking at the percent iron in the magnetic fraction and the Davis Tube
data, these do not differ from the magnetite itself by very much except for
the Davis Tube readings which are higher than the 91% for the feed. So, the
separator concentrates the magnetite and does so independently of the load-
ing in this range.
-------
-16-
TABLE 7
HGMS Results With Reiss-Viking Magnetite
RUN
103
108
104
109
105
101
106
102
107
110
MAGNETIC
FIELD (kG)
0.2
0.2
0.5
0.5
1.0
2.0
2.0
3.0
3.0
3.0
THROUGHPUT
(tph/ft2)
1.8
1.2
2.2
1.1
1.7
2.5
1.8
3.2
3.0 '
1.5
% HEIGHT
SAMPLE
NON-MAGS MAGS
150.00 gm R-V magnetite
1305 ml H2O
20 ml dispersant
10.3% by weight solids
SAME
SAME
SAME
SAME
SAME
SAME
SAME
SAME
75.00 gm R-V magnetite
91.7
88.9
43. 0
40.9
8.3
4.8
5.3
4.5
5.0
5.7
8.3
11.1
57.0
59.1
91.7
95.2
94.7
95.5
95.0
94.3
% MAGNETICS
IN MAGS
(DAVIS TUBE)
(96.6)
(96.2)
(95.1)
(95.6)
(96.4)
1305 ml H2O
20 ml dispersant
5.4% solids by weight
-------
TABLE 8
Effect of Loading Matrix With Reiss-Viking Magnetite
RUN
102
107
110
111
112
MAGNETIC
FIELD (KG)
3.0
3.0
3.0
3.0
3.0
THROUGHPUT
(tph/ft2)
3.2
3.0
1.5
2.3
3.6
SAMPLE %
(by
150.00 gro R-V magnetite
1305 nl H20
20 ml dispersant
SAME
75.00 gm R-V magnetite
1305 JD! H20
20 jnl dispersant
100.00 gra R-V magnetite
1305 ml R2O
20 ml dispersant
300 gm R-V magnetite
1305 ml H2O
SOLIDS
weight)
10.3
10.3
5.4
7.1
18.7
% WEIGHT
non-mags
4.5
5.0
5.7
5.5
4.4
mags
95.5
95.0
94.3
94.5
95.6
% MAGNETICS % IRON % Fe304*
non-mags mags non-mags mags (calculated)
(Davis Tube) Mags
(95.6)
(96.4) 21.57 65.37 90.3
20.24 65.29 90.2
21.62 64.81 89.6
1
h-1
-J
1
113
150
151
153
154
3.0
3.0
3.0
3.0
3.0
20 ml dispersant
6.4 600.00 gm R-Vmagnetite 31.5 2.6 97.4
1305 ml H20
20 ml dispersant
2.3 300.00 gm R-Vmagnetite 25.6 2.7 97.3
870 ml H20
20 ml dispersant
0.3 600.00 gm R-Vmagnetite 25.6 2.0 98.0
1740 ml H20
40 ml dispersant
2.3 300.00 gm R-Vmagnetite 25.6 2.8 97.2
870 ml H20
0.2 600.00 gm R-Vmagnetite 25.6 0.9 99.1
1740 ml H20
15.82 64.20 88.7
(93.0)
72.36% Fe
-------
-18-
100
LU
50
0.1
I I I I I I
J.
1.0
MAGNETIC FIELD/ KG
J L
10
Figure 4. Magnetite capture (Davis Tube) as a function of magnetic field.
Feed = 150 gms. with a dispersant (Alconox).
100
ID
90-
80
0 100 200 300 400 500 600 700
FEED WT/ MAGNETITE (GMS)
Figure 5. Matrix loading (jnagnetite) as a function of feed weight
Field = 3.0 kG.
-------
-19-
APPROACH
Magnetite recovery in a heavy-medium coal-washing plant is generally
performed by drain and rinse screens and drum magnetic separators. As finer
sizes of coal are washed, new techniques will be used. In the accompanying
flow sheet for the Homer City coal-washing plant of the Penelec Co. shown as
Fig. 6, there is a heavy-medium cyclone which receives the underflow of a
2 mm x 0 classifying cyclone which makes a split at 100 mesh. Thus, the
cyclone input is sized at 2mm x 100 mesh. The specific gravity is maintained
at 1.30 in the heavy-medium cyclone. The cyclone overflow is a clean coal
product containing part of the magnetite medium, and the underflow is refuse
and middlings also containing part of the medium. Both of these streams are
pumped to magnetic separators for magnetite recovery.
Normally, we might have approached the problem of magnetite recovery
by securing samples of the overflow and underflow products from the cyclone
and processing these by HGMS, but these were not available. We considered
the problem of handling the wide range of particle sizes (2 mm by 0) to be
a major problem in operating the separator because a HGMS is a matrix-type
separator. Therefore, rather than try to split two output streams at some
arbitrary particle size, the input coal size range was chosen with the
thought that we might make a particle size split at, say 20 mesh and try to
handle these fractions separately. A quick test was tried using the coal
as received and an attempt was made to design a matrix to handle this wide
size range. About 100 experimental trials were conducted with various com-
binations of matrices and magnetic field configurations. Much information
about operating parameters was obtained. Screen spacing, flow rates and
distribution of magnetics in the matrix were all varied.
A matrix design was developed which could accept a wide particle size
range, retain essentially all the magnetite entering it and pass the coal
with low loss values. Clogging the matrix was a problem with earlier de-
signs but the final form escapes this difficulty. After this matrix was de-
signed, most of the experimental runs used the coal as received mixed with
magnetite in the proportions indicated in Fig. 7 for the case of specific
gravity = 1.30. Some runs were made on the magnetite and coal by themselves
-------
Raw coal
Concentrating
table
Refuse
Screen 1 in by in
Dense medium
cyclone 1.60 S.G.
jin by 2mm screen
R Classifier
cyclone
Dense medium
cyclone 1.80 S. 6.
Drain and rinse
screen
I [Centrifuge •
Middling
i
Is)
O
Figure 6. Schematic of the Homer City Coal Cleaning Plant (simplified
circuit). Feed samples are similar to feed (2 mm x 100 mesh)
to cyclone on left.
-------
-21-
Condition
Average Sp. Gr. of Separation
% Coal Solids (by wt.)
% Magnetite Solids (by wt.)
% Total Solids (by wt.)
Specific Gravity of Feed
Underflow
/Refuse or Middling
I Product Depending on
\ Application
Condition
wt.% Coal Solids
wt.% Magnetite
wt.% Total Solids(D
Tons/hr. per
Linear Foot
of Separator
Coal
Mag.
TOTAL
1.275
13.7
20.1
33.8
1.249
1.300
13.4
22.2
35.6
1.274
^
/
1.
13
24
325
.14
.24
37.4
1.
299
.^^, Over flow (Clean Coal)
11.54
17.86
29.4
2.70
4.19
6.89
19.05
20.25
29.3
2.11
4.72
6.83
7.19
22.32
29.5
1.69
5.30
6.95
B
9.77 12.13 13.74
13.35 14.47 15.73
23.1 26.6 29.5
2.05 2.64
2.80 3.15
4.84 5.79
Mag
3.06
3.50
6.56
Mag
Coal
Coal
NOTEi (1) wt.% total solids leaving H.M. cyclone underflow is 50%. Values
shown are after dilution by the water addition indicated to pro-
vide a more acceptable slurry mix to the magnetic separators.
Figure 7. Homer City Station Coal Preparation Plant fine coal (2 nun x 100 mesh)
heavy medium cyclone circuit flow ratios to heavy medium cyclones and
to magnetic separators for magnetite recovery.
-------
-22-
and the results are indicated under the section on feed.
In order not to ignore the fact that the separator would operate in
practice on the output streams of a 1.30 specific gravity, heavy-medium
cyclone, 10% of our 200 pounds of coal was split in an organic liquid at
1.30 specific gravity by Eastern Associated Coal Corp. The products of
this split were run through the separator with magnetite after the para-
metric studies on the coal-magnetite samples had been completed. The re-
sults are compared with the runs on coal as received in the results sec-
tion of this report.
Not all samples were run at the slurry density indicated in Fig. 7.
That was somewhat inconvenient when working with small samples. But
enough were run at the correct slurry density to indicate the effect of
density changes on the separation. Also, the separations were performed
at the fastest flow rates possible with the slurry tank located directly
above the matrix. Gravity feed was used exclusively.
No water preceded the samples through the matrix as is customary
in HGMS testing although some was used following the slurry to maintain
what head there was. Some runs were done at slower flows to see the ef-
fect of flow rate and to determine the effect of filling the matrix before-
hand. It is inevitable in research on an application of a relatively new
technology that machine design and process design get done together to
some extent. This was not a testing program of a standard commercial de-
vice, so much of the work here was done to give information about the pro-
cess and machine requirements.
All experimental work on this project consisted of single passes
through the separator. Multiple passes, and their contribution to the
-------
-23-
process, were left to a second phase of this work. Single pass means one
machine in practice, and if coal and magnetite can be separated by one ma-
chine, capital costs are held to a minimum. We decided to follow this plan
as far as we could and, as it turned out, we never felt the need nor had
time to investigate multiple passes. We will note in a section on recom-
mendations for future work, though, that this possibility should not be
ignored.
-------
-24-
APPARATUS AND EXPERIMENTAL METHOD
A dipole magnet consisting of two water-cooled coils surrounded with
iron — forming pole faces 7" x 24" in face area and 2" apart _ was used for
most of the experiments. This magnet is capable of producing a magnetic
field of 21,300 gauss, partly because of the massive iron structure. The
current required to produce this field is 650 amperes at 70.5 volts. The
field values were measured with a transverse probe used on a Bell gauss-
meter. Some earlier runs were done on a 10,000 gauss solenoid magnet
driven by the same power supply. The current for the dipole magnet was
measured by a millivoltmeter placed across a shunt.
The field values used in this experimental work were in the range of
0-6000 gauss. This range was determined by experimentation with magnetite
trapping as a function of field. The residual field in the iron structure
was of the order of 100 gauss, depending on the magnetization history of
the iron. It seemed to have no effect on the cleaning of the magnetite
from the matrix. Therefore, we took no precautions to reduce this to zero
although this is easy to do.
The separator itself consisted of an aluminum tube of 4.8 sq. in.
cross-section filled with stainless steel expanded-metal screens. This
material was a necessary substitute for steel wool used so often in HGMS
experiments and in commercial practice because of the wide particle size
range of the feed samples. It also is more difficult to plug this open
structure with magnetite. The outlet of this tube was open for maximum
flow rates and later restricted by a valve for controlled flow and attach-
ment of further flow restrictors.
The samples were slurried in a head tank with adequate stirring pro-
vided by two propellers on a laboratory stirrer. Baffles prevented vortices
and a plug was used to prevent premature solids entry into the separator.
Details of the apparatus are shown in Fig. 8.
-------
-25-
HEAD WATER
-
*s
\
\
\
\
\
\
\
\
\
-
1 I
-•*"
rr
• ^-
-
' ""
—
1
FEED SLURRY
POLE PIECE
Figure 8.
PRODUCTS
Schematic of experimental apparatus. The magnet gap is 2" by
7" between the pole pieces. Fields of 21 kG can be obtained.
A dispersant (Alconox) was used in some of the experimental runs to
reduce magnetic agglomeration but later runs were made without it for com-
parison .
A Denver 8" pressure filter was used to dewater the products. Samples
were air-dried and weighed before further analysis. A HIAC Particle An-
alyzer was used to analyze samples taken directly from the feed and product
slurries. The filter changes the particle size distribution radically/
therefore samples were taken from the slurries.
The method of running tests consisted of sampling both magnetite and
coal in a careful way and mixing them in the slurry for several minutes. In
the mode of unrestricted flow the slurry was allowed to flow by gravity
through the separator and the products collected at the outlet. The time
was recorded for the total products recovery and in some cases for differenti-
al volumes. Magnetics were washed out with water when the field was off.
No backwashing was found to be necessary although the distribution of material
in the matrix usually would make backwashing advantageous.
-------
-26-
Flow restrictors were used for several sets of tests so that values
for the throughput of solids and superficial slurry velocities would be
accurately known, but generally this reduction in flow rate degraded the
f)
separator performance. The throughput values range up to 4 tph/ff4 and
the flow rates up to about 300 gpm/ft2. These values are the lowest of
a range of throughputs to be investigated in further work.
Magnetic field values were kept as low as the data on magnetite re-
covery vs. field would allow because low field values mean cheaper magnets.
However, we find that work in the future might well be done at slightly
higher fields.
Other analyses obtained were iron, sulfur, and ash. A few float-
sink determinations on a number of the products to detect a loss of coal
(defined as 1.30 float product) were done quickly at the end of the pro-
ject. They indicated very low losses (1%) of float product in a number
of tests but suffered from poor material balances.
Although the length of the matrix was dictated somewhat by the magnet
configuration and we made no further attempts at improving matrix design,
loading was investigated because it would be important to know if there
were serious limitations in this regard.
-------
-27-
ANALYSIS METHODS
Magnetics
A Magnos MXM-3 integrating fluxmeter was obtained and a coil designed
for it. Capsules filled with samples were placed in the coil, the magneto-
meter was zeroed, and the capsule removed thereby changing the total flux
through the coil. This change is integrated by the fluxmeter as a voltage-
time product. Calibration samples were made up from the coal and magnetite,
and the calibration curve is shown as Fig. 9 to indicate that the curve
passes through zero and that there is very little scatter in the points. The
method is sensitive to about 0.1% magnetite.
Magnetics in some of the run products were measured by Davis Tube at
the Bureau of Mines Coal Preparation and Analysis Laboratory. These values
are given in parentheses in the data tables.
600
LU
20
40 60
% MAGNETITE
80
Figure 9.
Calibration curve for the integrating magnetometer.
were made of the feed coal and magnetite.
100
Standards
-------
-28-
Iron
All iron analyses were done by Lerch Bros. Inc. by wet chemistry or
by flame spectrometry (atomic absorption).
Ash
Ashing was done by the standard ASTM method at M.I.T. Materials
balances are shown in the Appendices for ash and iron analyses. Also in
an Appendix is a correlation between magnetics, ash and iron results.
Particle Size Analysis
We obtained a HIAC Particle Analyzer for this project to investigate
the effectiveness of HGMS on recovery of fine magnetite. The HIAC PC-320
works on the light blockage principle. Particles are carried through a
sensor by a fluid under pressure with the flow rate of the transport fluid
well controlled by means of a pressure valve, and thus the possibility of
coincidence, i.e., of two particles passing through the sensor simultane-
ously, can be minimized. The particle is described in terms of its equiva-
lent sphere diameter which is the diameter of a sphere with a cross-section-
al area equal to the projectional area of the particle.
Originally we sampled the magnetite solutions using the "volume sam-
ple" option of the HIAC PC-320 which would count the number of particles
in each size channel in a given amount of solution. We used a method of
sampling as much of the total sample as possible to minimize the problem
of settling. Stock solutions were made up of dry material shaken well
in 250 ml of water. Then approximately 20 ml of the stock solution was
diluted by an equal amount of distilled water and stirred well. The HIAC
was run such that it would count the number of particles from the first
drop of new solution until stopped when the meniscus of the solution hit
a mark on the sampling tube. Thus it would run through 35 ml of solution
leaving 5 ml in the beaker.
-------
-29-
In some suspensions Alconox was used to see if its presence could
inhibit agglomeration. Methanol was also tried to "wet" the material.
Neither appeared to be significant. Nor was glycerin which was used to
try to slow the settling rate. The figures were all corrected by sub-
tracting baseline values obtained by sets of distilled H_O runs done at
the beginning and end of each determination.
-------
-30-
RESULTS
Magnetite Capture and Coal Recovery as a Function of Magnetic Field
In Table 9, data is presented for a series of runs made at increasing
magnetic fields up to 6000 gauss. The samples were coal and magnetite in
the proportions found in the cyclone feed and a dispersant was used to re-
duce magnetic agglomeration. The solids density is less than specified for
the cyclone feed but we will present data on the effect of solids density
further on. The throughput is the weight of total solids in the feed di-
vided by the time taken to pass through a cross-section of the separator.
Runs 114 through 119 were preliminary runs made on the newly con-
structed double screen matrix and show some very good results particularly
run 116 at 2 kG. Runs 120-126 are a more carefully controlled set of runs
done with increasing values of field.
The values for magnetics contained in both separator products were
determined both by our magnetometer and by Davis Tube. At the higher fields
the loss of magnetics to the non-mags fraction is of the order of tenths
of a percent. In Fig. 10, we show this data to give an idea of the cor-
respondence of these methods, and the effect of magnetic field. Our magneto-
meter gives higher readings because we define the as received magnetite to
be 100% magnetics.
The values of magnetics loss to the non-mags are corroborated by the
values of ash which are asymptotic to the ash value in the coal (13.2%) as
the field increases. Coal loss is shown in Table 9 by percent ash in the
mags, percent magnetics .or percent iron (coal = 2.33% iron).
In Fig. 11 is shown the percentage of the feed captured vs. magnetic
field in order to determine the field necessary for successful separation.
The curve is essentially flat above 2 kG. The feed is 62.4% Reiss-Viking
magnetite and this curve seems to be asymptotic to 64% capture. As was
shown earlier, only a few tenths of a percent of the coal should be cap-
tured and about 4% of the magnetite should escape capture so there is an
-------
TABLE 9
The Effect of Magnetic Field
RUN MAGNETIC
FIELD UcG)
114 0.5
115 1.0
116 2.0
118 5.0
119 6.0
120 0.5
121 1.0
122 2.0
123 3.0
124 4.0
125 5.0
126 6.0
THROUGHPUT SAMPLE
Ctph/ft2)
2.3 150.00 gm R-V magnetite
90.54 gm coal
1305 ml H20
20 ml dispersant *
15.6% solids by weight
3.4 SAME
3.9 SAME
3.3 SAME
3.9 SAME
1.2 SAME
3.0 SAME
3.8 SAME
4.4 SAME
4.1 SAME
4.2 SAME
3.9 SAME
% WEIGHT % MAGNETICS % IRON % ASH
non-mags mags non-mags mags non-mags mags non-mags mags
(Davis Tube)t
56.6 43.4 37.2
C32.1)
40.0 60.0 13
( 6.9)
36.8 63.2 0.2
( 0.4)
31.3 68.7 0
31.8 68.2 0.1
72.4 27.6 46.4
(42.9)
44.7 55.3 18.2
(14.3)
36.3 63.7 1.1
( 0.2)
36.0 64.0 0.8
( 0.1)
36.4 63.6 0.5
( 0.3)
36.3 63.7 0.3
( 0.2)
35.2 64.8 0.4
( 0.2)
100.3 24.25 61.67
C91.9)
95.8 9.42 60.82
(89.7)
100.0 3.70 60.82 22.4 96.3
(88.5)
88.2 3.02 56.15 19.3 92.7 ,
lx
91.1 2.82 56.11 93.0 7
103.1 29.79 64.20 58.0 100.0
98.7 17.23 69.43 33.5 97.9
(91.8)
95.8 3.58 75.63 20.7 96.2
(88.2)
96.5 3.70 60.27 22.3 97.5
(88.6)
95.7 3.25 59.47 20.1 98.0
(88.4)
101.0 3.25 60.55 15.6 97.5
(88.8)
97.4 3.04 60.60 15.0 97.8
(89.0)
* Dispersant: 10 gm Alconox/liter
t Davis Tube values for magwttite are 90-91%, for coal 0
-------
-32-
100
I 50-
% MAGNETICS/ MIT
DAVIS TUBE
MAGNETIC FIELD/ KG
Figure 10. Ash and magnetics measurements for the non-mags as a function
of magnetic field.
100
50
MAGS
0.1
1
MAGNETIC FIELD/ KG
10
Figure 11. Magnetics capture (% of feed) as a function of field.
Feed = 150 gms magnetite, 90.5 cms coal at 15.6% solids.
Average throughput =3.5 tph/ft- (Table 9)
-------
-33-
uncertainty as to the path followed by the impurities. This curve indicates
that part of them reports to the magnetics.
Matrix Loading When Samples Consist of Both Coal and Magnetite
In Table 10 and Fig. 12 the percent capture at 3 kG is compared to the
weight of magnetite in the feed. In each case, the feed contained 37.6%
coal by weight as in almost all other runs. In fact, this data was selected
from several series of runs made for other purposes. The conditions are
somewhat different for different series of runs, therefore they are plotted
separately. Runs 123, 127-130 have the highest throughput values because a
valve was added to accommodate flow restrictors just after these runs were
made and,even though no flow restrictors were used in any runs, the later
ones had lower throughputs on the average with the valve in place.
100
80
LU
5
60-
n
o
AVG THROUGHPUT
A 1 .0 NO DISP
D 2.4 NO DISP
2.5 W/ DISP
3.0 W/ DISP
O
O
i
100 200 300
MAGNETITE VfT (QMS)
400
500
Figure 12. Magnetics capture at 3 kG as a function of feed weight ex-
pressed as weight of magnetite.
-------
TABLE 10
Effect of Matrix Loading
RUN MAGNETIC
FIELD (KG)
148
129
163
127
158
123
149
147
128
130
165
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
3.0
THROUGHPUT
Ctph/ft2)
1.2
1.9
1.9
2.9
2.5
4.4
0.8
0.9
2.7
3.5
0.5
R-V*WAGNETITE
FEED NT. (gm)
75
75
75
75
150
150
300
300
300
300
450
% SOLIDS
8.4
8.4
8*4
8.4
15.6
15.6
27.0
27.0
27.0
27.0
35.6
DISPER- % WEIGHT
SANT t non-mags mags
^
V
V
V
v
V
35.5
39.5
37.0
38.5
35.3
36.0
17.4
17.4
29.1
27.4
16.5
64.5
60.5
63.0
61.5
64.7
64.0
82.6
82.6
70.9
72.6
83.5
% MAGNETICS % IRON % ASH
non-mags mags non-mags mags non-mags mags
(Davis Tube)
0.6
(0.1)
1.3
1.2
0.6
2.3
0.8
(0.1)
0.6
3.0
1.0
1.1
1.2
96,8
(87.5)
102.2
96.. 7
99.7
94.8
96.5
(88.6)
77.0
74.6
87.9
87.3
74.4
3.61
3.33
3.90
3.81
3.42
3.70
3.25
8.27
3.61
2.97
3.50
60.60
63.60
61.81 18.7 98.9
62.08
58.91 20.9 95.8
60.27 22.3 97.5
46.27
47.96
55.34
54.06
45.53
1
UJ
1
t Dispersant: 20 ml of 10 gm Alconox/liter
* Feed: 62.4% R-V magnetite, 37.6% coal
-------
-35-
In all cases the total capture increases with loading. Coal or impuri-
ties or both report to the magnetics as the amount of magnetite trapped on
the matrix increases. The runs at the highest throughputs exhibit the lowest
entrainment and it remains to be seen how little coal loss can be achieved as
throughputs are increased in future work.
All runs at the 75 gm loading have less than 5% of the non-mags report-
2
ing to the magnetics and the high throughput run 123 (4.4 tph/ft ) has a non-
mag loss at 150 gins loading of less than 5%. We consider this throughput value
to be at the bottom of the range of values possible with HGMS operating in the
mode of unrestricted flow-
Magnetics recovery seems unaffected by loading as shown by the percent
magnetics in the non-mags. Even at 450 gms only 0.33% of the 450 gms of mag-
netite ends up in the non-mags. This seems to agree with the loading data
obtained for magnetite alone (see the section on feed materials, Fig. 5).
Magnetite Recovery at Cyclone Input Slurry Density
The effect of high slurry density (35.6 solids) and high loading (300
gms magnetite) can be seen in Table 11. Some magnetite is lost to the non-
mags as evidenced by the magnetics and iron values although it is evident
from Fig. 13 that magnetite recovery is increasing sharply with magnetic
field and presumably will reach an acceptable value at the 6 kG field used
in several previous runs.
-------
-36-
TABLE 11
Effect of High Slurry Density
RON MAGNETIC THROUGHPUT
FIELD (kG) (tph/ft2)
166 1.0 0.8 300.00
181.08
870 ml
35.6%
168 1.0 1.4
167 2.0 1.0
169 2.0 .1.4
170 4.0 1.1
RECOVERY
O
0
850
fc
<•
^
-
—
•_
SAMPLE % WEIGHT % MAGNETICS* % IRON
non-mags mags non-mags mags non-mags mags
gm R-V magnetite 67.1 32.9 64.3 82.5 35.32 47.18
gm coal
H20
solids by weight
SAME 76.3 23.7 55.0 86.3 34.95 54.30
SAME 52.9 47.1 45 83 27.98 47.61
SAME 58.0 42.0 55 75 30.74 49.30
SAME 36.3 63.7 32.0 85.5 18.04 51.98
* By magnetometer
V
x\
a/ \
\
O NON-MAGS V
A MAGS
u
\l\ll\ \ 1 1 t 1 I t 1
0.1
Figure 13.
1
MAGNETIC FIELD/ KG
10
Magnetics recovery from samples twice the usual feed weight
and at 35.6% solids (= density of cyclone feed). Recovery
is defined as ratio of magnetics weight in products to that
in the feed. (Table 11)
-------
-37-
Cyclone Products Tests
The preceding tests were all conducted on cyclone feed. In order to
test the effectiveness of HGMS on the products of the heavy-medium cyclone
where the magnetic separators would most likely be placed, a coal sample
was split at 1.30 specific gravity in an organic liquid. The products of
the split were 48.4 percent float by weight and 51.6% sink. These were
tested by mixing with magnetite (in the proportions shown on the Homer City
flow sheet) and subjected to HGMS under conditions similar to the previous
testing.
1. Sink Product (1.30)
This simulates the cyclone underflow and feed samples were
made up as stated above. The solids density was 29.3% as
specified in the cyclone flow sheet provided by Pennelec
(shown as Fig. 7). Runs were made at 2 kG and 4 kG with
no flow restrictor and also restricted to 4 gpm. All runs
were repeated. The results are shown as Table 12. The
total material capture is higher at the slower flow rates
and, at the higher field values, the percent magnetics mea-
sured in the non-mags reaches a low value of 0.2%.
2. Float Product (1.30)
The feed for these tests approximates the cyclone overflow
in the proportions of magnetite and coal. The solids den-
sity was 23.3%, about 3.3% less than called for on the
Pennelec flow sheet. Results are shown in Table 13.
At a restricted flow rate we retain essentially all the magnetite (see
percent magnetics in non-mags, percent iron, or ash) but apparently sustain
a coal loss as seen in the figures for ash and iron in the mags (for magne-
tite 63.3% iron, 102.4% ash). In the case of unrestricted flow the coal loss
-------
TABLE 12
Separation Tests on Sink Product (1.30)
RUN
183
188
185
189
184
186
187
190
% WEIGHT
MAGNETIC VELOCITY SAMPLE
FIELD (KG) .Con/sec) non-mags mags
2.0
2.0
2.0
2.0
4.0
4.0
4.0
4.0
87.4
87.4
47.1
52.1
112.4
87.4
46.3
49.8
150.00 gra R-V magnetite
67.04 gra coal (sink 1.30)
523.7 ml H20
29.3% solids by weight
SAME
SAME
SAME
SAME
SAME
SAME
SAME
32.7
29.6
22.7
21.3
31.7
28.9
20.2
21.6
67.3
70.4
77.3
78.7
68.3
71.1
79.8
78.4
% MAGNETICS*
non-mags mags
19.4
4.8
0.8
0.4
6.:
0.6
0.6
0.2
88
96.1
89.9
89.1
96.4
94.7
88.2
87.3
% IRON % ASH FLOW
RESTRICTIO
non-mags mags non-mags mags (gpm)
13.99
8.18
5.53
5.01
8.58
5.65
5.05
4.73
59
60
53
54
61
60
55
54
.63 36.9 98.8
.11 34.1 98.7
.62
.71
.76 34.6 99.9
.92 30.7 99.5
.54
.54
-
-
4
4
-
-
4
4
I
LO
oo
* By magnetometer
-------
TABLE 13
Separation Tests on Float Product (1.30)
% WEIGHT % MAGNETICS* % IRON
RUN MAGNETIC
FIELD CkG)
VELOCITY
Con/sec)
SAMPLE
% ASH FLOW
RESTRICTION
non-mags nags non-mags mags non-mags mags non-mags maas Igprn)
191
2.0
88.4 150.00 gm R-V magnetite 55.7 44.3
125.7 gm coal Cfloat 1.30)
907.7 ml H20
23.3% solids by weight
22.8 92.2
193
192
194
2.0
4.0
4.0
49.
112.
49.
5
4
5
SAME
SAME
SAME
34.
44.
35.
7
9
6
65.
55.
64.
3
1
4
0.
13.
2
5
84.5
91.2
14.95 58.35 24.2 92.5
0.56 49.97 4.3 84.1
6.93 57.23 10.3 92.5
51.90 3.6 85.8
vO
I
* By magnetometer
-------
-40-
drops but the magnetite recovery remains a function of field. A higher field
— say 6 kG — would retain more of the magnetite.
Effect of Dispersant on Entrainment at High Solids Density
In Fig. 14, we have plotted the percent mags capture against slurry den-
sity for magnetite alone with and without the use of a dispersant. All of
these points are at 3 kG and there seems to be little or no effect on the cap-
ture. However, there is a significant effect on the coal and magnetite mix-
tures. The total capture increases from ~63% to ~86% with an increase in per-
cent solids from 8.4 to 35.6. Without a dispersant, the capture increases
with an increase in solid density hut the effect is reduced considerably with
a dispersant.
100
80
60
0
10
A
D
A MAGNETITE
O MAGNETITE W/0 DISPERSANT
O MAGNETITE PLUS COAL
D MAGNETITE PLUS COAL W/0 DISP,
I i
20
SOLIDS
30
40
Figure 14.
Magnetics capture as a function of feed slurry density for
samples of magnetite alone and the same mixed with coal.
(H = 3.0 kG). Magnetics capture is % of feed by weight
which reports to magnetic fraction.
-------
-41-
Restricted Flow Rates
We list in Table 14 several runs made at slower flow rates for the pur-
pose of rate control. Also, the mode of operation of the separator was changed.
The matrix was filled with water so that the sample flowed into a matrix al-
ready filled. Unfortunately, slower flows have to be used to achieve this
condition. The mags fraction increases with decreasing flow rate at each field
value and the recovery of magnetic material is affected strongly at the lower
field values, but the effect is smaller above 2 kG.
Effect of Diapersant and Other Effects
In Runs 155-162 (Table 15), we repeat the magnetite capture vs. field
tests, but use no dispersant. We still have the flow controlling valve in
place although no other flow restrictions are used. In these runs we must
substitute a higher field to achieve adequate magnetite recovery. This
set of runs provides some interesting data since the operating conditions
were somewhat better controlled.
In Fig. 15, the percent of feed captured vs. field seems to level off
at 65% (62.4% of the feed is R-V magnetite). Figure 16 gives the magnetics
recovery in the mags and non-mags.
The percent iron and percent ash are plotted in Fig. 17 for both mags
and non-mags. As the field increases, both of these figures increase slow-
ly in the mags as expected but the reduction in the non-mags seems to ac-
celerate until the major part is removed at about 3000 gauss. This may be
a magnetic agglomeration effect. Ash and iron contents of the feed coal
are shown by arrows on the margin of the graph. As the field increases the
ash is reduced to a value higher than that which could be accounted for by
the coal so some magnetite or magnetite impurities report to the non-mags.
-------
TABLE 14
Effect of Restricting Flow Rate
RUN MAGNETIC
FIELD CkG)
173 1.0
176 1.0
174 2.0
177 2.0
175 4.0
178 4.0
% WEIGHT * MAGNETICS* % IRON FLOW
VELOCITY SAMPLE RESTRICTION
(cm/sec) non-mags mags non-mags mags non-mags mags (gal/min)
43.7 300.00 gra R-V magnetite 61.1 38.9 47.4 81.0 30.56 51.26
181.08 gm coal
870 ml H2O
35.6% solids by weight
29.3 SAME 36.0 64.0 41.9 87.7 21.96 54.06
53.1 SAME 25.4 74.6 1.3 81.9 3.13 50.58
28.1 SAME 18.2 81.8 0.5 79.7 2.28 48.97
43.0 SAME 25.0 75.0 0.4 81.1 2.65 50.58
25.4 SAME 21.0 79.0 0.4 76.7 2.16 48.61
4
3
4
3
4
3
By magnetometer
-------
TABLE 15
Magnetite Capture Without Dispersant
RUN MAGNETIC
FIELD UtG)
155 0.5
156 1.0
157 2.0
158 3.0
160 4.0
161 5.0
162 6.0
THROUGHPUT SAMPLE % WEIGHT
Ctph/ft^) non-nags mags
1.1 150.00 gm R-V magnetite 68.9 31.1
90-. 54 gm coal
1305 ml H20
15.6% solids by weight
1.7 SAME 56.1 43.9
2.4 SAME 40.8 59.2
2.5 SAME 35.3 64.7
2.7 SAME 35.1 64.9
2.9 SAME 34.9 65.1
3.1 SAME
% MAGNETICS % IRON % ASH
non-nags mags non-mags nags non-nags mags
(Davis Tube)
52.8 94.9 31.58 58.83 57.4
185.6)
40.1 94.0 23.27 59.19 49.3
20.6 94.3 13.14 58.67 33.2
2.3 94.8 3.42 58.91 20.9
0.5 96.0 3.26 59.96 19.2
0.5 95.3 2.41 59.07 19.6
97.2 60.84
94.7
95.2
95.0
95.8
96.7
96.2
99.3
I
•p-
OJ
I
-------
-44-
100'
LU
ce
50
0.1
1
MAGNETIC FIELD/ KG
10
Figure 15. Magnetics capture (% of feed wt.) as a function of magnetic
field. No dispersant was used. The feed density was 15.6%
solids. Normal feed weight of 240 gms total solids.
(Table 15)
100
I
9 so-
MAGNETIC FIELD/ KG
Figure 16. Magnetics recovery (% of magnetics in feed) in both mags and
non-mags as a function of magnetic field for the runs without
dispersant. (Table 15)
-------
-45-
100
01
lo
50-
ASH O
IRON A
MAGS
0.1
NON-MAGS
FEED
COAL
1
MAGNETIC FIELD/ KG
10
Figure 17. Ash and iron values for non-mags and mags as a function of
magnetic field. (Table 15)
Particle Size Analysis
Ten experiments were run in which the feed and products were sampled
directly from the slurry for the purpose of particle size analysis. The
products of the experimental runs which had been filtered were not suitable
for this analysis because filtering alters the particle size distribution.
The samples for these runs were all 150 grams of Reiss-Viking magnetite.
Eight of these were slurried in tap water which had been used throughout
the project. Two samples were slurried in distilled water to reduce the
background particle count from the water. A dispersant (Alconox) was used
in six runs and left out of the other four. Runs were made to investigate
the effect of increasing the magnetic field and in two runs the matrix was
not filled with water first.
The data were obtained using the HIAC Particle Analyzer. Numbers of
particles per channel (6 channels 1-3, 3-5, 5-8, 8-15, 15-20, 20-60 microns)
were recorded, an average particle size diameter per channel chosen and the
average weight percentage less than the upper limit of each channel plotted.
-------
-46-
The average particle size per channel was calculated by taking the
average value of the weight as a function of particle radius (spherical
shape is assumed, the HIAC measures equivalent circular area). The average
is given by:
^ 4 3
wt = / — irpr dr
avg r. - r., I 3
211
ri
For the purpose of calculation p was assumed to be 5 gm/cm . The values for
average particle diameter and average particle weight per channel are given
in Table 16.
In Fig. 18 we show two electron microscope photographs of products of
magnetic separation. The first is a particle from the non-mags and the second
is a chain of agglomerated magnetized magnetite particles collected on a mem-
brane filter. In each photograph the deviation from the ideal of a spherical
TABLE 16
Average Particle Size and Weight - HIAC
CHANNEL (P) AVERAGE DIAMETER (y) AVERAGE WEIGHT (gm)
1
3
5
8
15
20
-3
-5
-8
-15
-20
-60
2
4
6
11
17
43
• 15
.08
.61
.8
.6
.1
2
1
7
4
1
2
.62
.78
.57
.35
.43
.09
x 10
x 10
x 10
x 10
x 10
x 10
-11
-10
-10
-9
-8
-7
-------
-47-
Figure 18. Electron microscope photographs of non-mags particle (top)
(Magnification = 11,050) and chained magnetite particles
(Magnification = 2,100).
-------
-48-
particle assumed for the purpose of particle size measurement is evident.
One of the objects of this analysis was to determine if the separator
discriminated in its collection of magnetite on the basis of particle size.
The HGMS technique is suited for small particle trapping and we did not ex-
pect to lose magnetite fines in any large amount.
Figure 19 shows the results of a distilled water run at 1000 gauss.
The feed distribution is considerably above both mags and non-mags. In
each size range the weight fraction below a given size is higher in the
feed. In other words the feed has more small particles than the products.
This is the effect of magnetic agglomeration which shifts the size distri-
bution towards the larger sizes. Note that there is hardly any difference
between the mags and non-mags. If the separator failed to attract fine
particles but trapped only larger ones, these distributions would be dif-
ferent.
Figure 20 shows results of a run at the same conditions but with a
field of 3000 gauss. Here the products have decidedly different distribu-
tions. The non-mags have more larger particles and the magnetics contain
a larger fraction of smaller ones. For example, at 1000 gauss (Fig. 17)
the magnetics have 75% wt. below 20)1 and at 3000 gauss (Fig. 18) 89% is
below 20y. This result is certainly caused by an effect other than the
loss of fines to the non-mags. An explanation of this effect may be found
in the fact that the non-mags are 38.6% of the weight at 1000 gauss and
only 3.32% at 3000 gauss and probably consist therefore, at the high field,
entirely of impurities with a different size distribution. This may be due
to hardness differences which affect the results of grinding. At any rate
we do not see at either field any discrimination between mags and non-mags
at the smaller particle size.
-------
-49-
100
O FEED
A NON-MAGS
D MAGS
10
PARTICLE SIZE/
100
MICRONS
Figure 19. Cumulative particle size distribution of magnetite before
and after HGMS. The field was 1.0 KG. No dispersant was
used and the matrix was filled with water before running.
100
O FEED
A NON-MAGS
D MAGS
10
PARTICLE SIZE/ MICRONS
100
Figure 20. Magnetite particle size distribution. Same conditions as
Fig. 19 except H = 3.0 kG. Mags represent 96.7% of the
feed weight.
-------
-50-
Figures 21 and 22 represent runs at 500 and 3000 gauss respectively. A
dispersant was used in these runs and/ like most of the coal-magnetite work,
these samples were dumped through the matrix without first filling it with
water. The effect of the dispersant is not evident at the low field. These
curves are essentially identical with those in Fig. 19. Also, the mode of
operation seems to make little difference.
At 3000 gauss, however, the difference (Figs. 20, 22) is striking.
The relative position of the non-mags and mags distribution curve is rever-
sed. In the latter experiment many more fine particles show up in the non-
mags. The dispersant which may help prevent agglomeration of fines with
larger particles and the more ordered flow seem to interfere with the trapping
of the finer particles. An important point to be kept in mind is that the
mags collection at 3000 gauss is at least 95% of the feed weight so the dif-
ferences seen in these graphs may be entirely due to impurities in the mag-
netite rather than magnetite itself. The Davis Tube recovers only 91% of
the magnetite and the HGMS collects only 95-96% of the feed magnetite even
at 6000 gauss.
One more result may illuminate separately the effect of dispersant
and mode of operation. In Fig. 23 the results of a run with dispersant but
with filling the matrix beforehand are shown. In comparison to the results
in Fig. 20 (without dispersant) there is a shift of fine particles to the
non-mags (e.g., 22% < 5y vs. 16% < 5\i) but not to the extreme of the results
in Fig. 22 (.unflooded operation, 30% < 5y in non-mags) . Apparently, both
dispersant and the flow characteristics have an effect on fines collection.
It may be, as we have noted above, that the non-mags consist entirely of
much more weakly magnetic material than magnetite so that these effects
are to be expected and the question of magnetite fines is not yet resolved.
-------
-51-
100
LU
M
CO
CO
CO
CO
s
O FEED
A NON-MAGS
D MAGS
100
PARTICLE SIZE/ MICRONS
Figure 21. Magnetite particle size distribution with dispersant. All
distributions were obtained using the HIAC Particle Size An-
alyzer. Matrix was not filled before run. H = 0.5 kG.
100
O FEED
A NON-MAGS
D MAGS
10
PARTICLE SIZE/ MICRONS
100
Figure 22. Particle size distribution for magnetite at 3.0 kG with dis-
persant. Matrix not filled before run.
-------
-52-
100
c/)
C/J
O FEED
A NON-MAGS
D MAGS
i i i i i i
10
PARTICLE SIZE/ MICRONS
100
Figure 23. Magnetite particle size distribution at 3.0 kG with disper-
sant and matrix filled before the run.
-------
-53-
CONCLUSIONS AND RECOMMENDATIONS
Conclusions
The main result of this project was the separation of magnetite from
coal using an experimental High Gradient Magnetic Separator which achieved
high throughput rates, high values of magnetite recovery, and small amounts
of coal reporting to the magnetics.
The throughput rates, calculated as weight of feed divided by the time
to pass through the matrix, vary from 1 to 4'.4 tph/ft . The actual rate
of passage of feed 'material through the matrix is higher because we include
some process water in this calculation. Much higher rates are possible than
this and we plan to initiate work to explore higher throughput values.
The results quoted here for magnetite recovery (99.9% or higher at
highest throughput values) are at least comparable to values claimed for
drums and the value for coal reporting to the magnetics is probably lower.
Magnetic field values of 6 kG were the highest used in this study.
A new matrix design proved successful at recovering fine grade mag-
netite, passing the coarser coal and allowing easy recovery of the magnetics.
Washing off the magnetics was accomplished by passing water through the
separator in the same direction as the slurry. In a typical set of runs,
the feed time was 3.5 seconds for a rate of 15.6 gpm and the wash time was
2.5 se.conds at a rate of 48 gpm. Most of the magnetite was removed in the
first half of the wash time, but using these feed and wash times, the duty
cycle for a batch separator would be 58% not counting switching. Continu-
ous separators like one tested at M.I.T. on iron ore (with magnetics con-
tent similar to our samples in this project and at slurry densities up to
40% solids) might be used in place of batch separators.
We have seen that the use of a dispersant and the flow characteristics
affect fines collection. Particle size analysis shows that magnetic agglo-
meration occurs and that HGMS is effective in fines collection as far as we
have gone.
-------
-54-
Recoinmendations
The flow rates used in this study have been such that a very substantial
throughput has been achieved. However, improvements in this rate have not
been explored and, as with loading, we do not know the limits of throughput.
This is an important piece of information which should be obtained.
Among a number of tasks which need to be followed through, an import-
ant one is the question of how compact the matrix can be made to allow for
reductions in the necessary magnetic field volume. Less field volume would
mean lower magnet capital cost. The length of the matrix might be reduced
without loss in efficiency of magnetite capture, and the matrix material and
cross-sectional loading changed without increasing coal loss.
We have used one kind of stainless steel expanded metal and have not
studied the effect of the use of other materials on fine particle capture.
This clearly needs to be done, and done in conjunction with improvements in
matrix design.
Since the matrix has never been loaded to its capacity, a further
study of loading should be carried out to give an idea of the duty cycle
required for a batch-type device.
Multiple passes have not been attempted in this study but could make
possible more efficient collection of different size ranges of magnetite,
possibly in part at lower field values, the use of which would reduce capital
cost. Matrix materials should be studied as to their effectiveness on
fine particle collection and different matrices might be used on succes-
sive passes, each tailored to the particle size and average magnetic moment
in each size range.
Improvement is needed in the identification of impurities as they
pass through the separator both to characterize the process and to study
impurity build-up in a coal-cleaning circuit.
Flexibility in the application of HGMS to magnetite recovery in coal
cleaning would be achieved by doing more work on the overflow and underflow
-------
-55-
products from the heavy-medium cyclone including products from the cyclone
when operated at varied values of specific gravity. Also, other coals or
other samples of the same coal should be tested.
It would be helpful in future work to have a full coal analysis done.
This was beyond the scope of this project but would help identify impuri-
ties and, if a BTU analysis were performed on the separator products and
feed, one could know the real loss of what is important to a boiler operator.
-------
-56-
REFERENCES
1. Sealy, G.D. and Howell, W.F., "Magnetic Recovery of Medium from Heavy
Media Circuits," World Coal, June 1977.
2. deLatour, C., IEEE Transactions on Magnetics, Vol. MAG-9, No. 3, September
1973, pp. 314-316.
3. Silverman, L., Billings, C.E. and First, M.W., Particle Size Analysis
in Industrial Hygiene, Academic Press, New York, 1971.
4. Kelland, D.R., and Maxwell, E., "Pilot Investigation of High Gradient
Magnetic Separation of Oxidized Taconites," Paper presented at AIME
Annual Meeting, New York, February 1975.
5. Mitchell, D.R., Coal Preparation, AIME, New York, 1950.
6. Cavallaro, J.A. and A.W. Deurbrouck, Reclaiming Magnetite in Dense-Medium
Circuits by Froth Flotation. Bureau of Mines RI 6821, 1966.
7. Deurbrouck, A.W., Washing Fine Size Coal in a Dense-Medium Cyclone,
Bureau of Mines RI 7982, 1974-
8. Taggart, "Handbook of Mineral Dressing," Wiley, New York, 1945.
9. Perry, "Chemical Engineers Handbook," McGraw-Hill, New York, 1963.
10. Palazzari, M.R., "How Do You Choose a Magnetic Separator," Coal Mining
and Processing, February 1967.
11. Topsoe, H., Dumesic, J.A., and Boudart, M., "Mossbauer Spectra of
Stoichiometric and Non-stoichiometric Fe3O4 Microcrystals," Journal
de Physique, Colloque C6, supplement an No. 12, Vol. 35, December
1974 pp. C6-411.
12. Bartnik, J.A. and Shively, R.W., "Separation of Solids and Fluids by
Magnetic Flocculation," paper presented at AIME Annual Meeting, New
York, February 26 - March 4, 1971.
13. Voet, A. and Suriani, L.R., "Dielectrics and Rheology of Dispersed
Magnetized Particles," paper presented at Annual Meeting of the
Society of Rheology, New York, November 3-4, 1950.
-------
-57-
APPENDIX I
Ash Balances for Two
RUN
116
118
120
121
122
123
124
125
126
155
156
157
158
160
161
162
ASH WT.
PROD. MAGS
(gin)
19.76
14.46
100.32
35.86
17.98
19.23
17.56
13.55
12.68
95.04
66.32
32.42
17.71
16.15
16.40
9.92
146.03
152.23
65.81
129.74
146.80
149 . 18
149.65
148.51
152.13
70.85
100.21
134.87
148.99
150.75
150.31
150.56
Sets of Runs
TOTAL
ASH WT.
165.79
166.69
166.13
165.60
164.78
168.41
167.21
162.06
164.81
165.89
166.53
167.29
166.70
166.90
166.71
160.48
CLOSURE
(%)
100.3
100.8
100.5
100.1
99.6
101.8
101.1
97.99
99.66
100.3
100. .7
101.2
100.8
100.9
100.8
97.04
Feed: Coal 13.0% Ash
Magnetite 102.4% Ash
For,these runs wt. ash in Feed: 165.37 gm
-------
-58-
APPENDIX 2
Typical Iron Balances
RUN
139
140
141
142
143
144
145
WT. FE
PROD.
30.09
4.63
2.96
2.66
2.57
2.26
2.53
(gm)
MAGS
66.80
91.69
92.14
93.15
92.36
91.81
90.36
TOTAL
96.89
96.32
95.10
95.81
94.93
94.07
92.89
CLOSURE
%
99.8
99.2
98.0
98.7
97.8
96.9
95.7
Feed: Coal 2.33% Fe
Magnetite 63.30% Fe
For these runs wt. Fe in Feed: 97.06 gins
-------
-59-
APPENDIX 3
0
10
20 30
MAGNETITE (CALC.)
40
= °/oASH
50
Appendix 3.
Correlation between pairs of analyses for non-
mags products of four runs, (155-158) for which
field values range from 0.5 to 3.0 kG. The cor-
relation coefficients (r) and 0.999 (Q) and
0.999 (o). / m2 gx2 \ 1/2 where m = slope and
a = standard deviation.' For runs 120-122 the
corresponding values of r would be 0.987 and
0.999. Using Davis Tube magnetics values in place
of magnetometer measurements for these runs
r = 0.976, 0.999 respectively.
-------
-60-
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
FE-8887-1 (EPA-600/7-78-183)
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Magnetite Recovery in Coal Washing by High Gradient
Magnetic Separation
5. REPORT DATE
September 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
E. Maxwell and D. R. Kelland
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Massachusetts Institute of Technology
Francis Bitter National Magnet Laboratory
Cambridge, Massachusetts 02139
10. PROGRAM ELEMENT NO.
EHE623A
11. CONTRACT/GRANT NO.
EPA/DOE Interagency
Agreement No. DXE685AK
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final: 3/76-12/77
14. SPONSORING AGENCY CODE
EPA/600/13
15. SUPPLEMENTARY NOTES prOject officers are D. A. Kirchgessner (EPA/IERL-RTF) and
R.E.Hucko (DOE).
16. ABSTRACT
repOrt describes a demonstration of the successful recovery of magne-
tite from mixtures of magnetite and coal, like those found in a coal- washing circuit,
by High Gradient Magnetic Separation. The demonstration was part of a research
program at Francis Bitter National Magnet Laboratory. High values of magnetite
recovery were achieved at reasonably high material throughput rates with little coal
found reporting to the magnetics. A single-stage separator incorporating a new matrix
design was used at rates up to 4.4 tons of solids per hour per square foot of matrix
cross section (300 gpm/sq ft). At this throughput rate, more than 99% of the magnetite
was trapped along with less than 5% of the coal. Magnetic field values no higher than
6 kilogauss were used to achieve these results , a value well within the range of pre-
sent commercial magnet designs.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATl Field/Group
Pollution
Coal Preparation
Magnetite
Magnetic Separators
Magnets
Magnetic Properties
Pollution Control
Stationary Sources
High Gradient Magnetic
Separation
13B
081
10A
131
20C
13. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
68
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
$5.25
EPA Form 2220-1 (9-73)
------- |