DoE
&EPA
United States
De:partmen! of Energy
Division of Solid Fuel
Mining and Preparation
Pittsburgh PA 15213
U S Environmental Prnte<,t'on Agenrv
Office of Research and Development
Industrial Environmental Research
Laboratory
Res*-,net' InnnLjIe Park NC 27/1 1
FPA 600 7 /8 208
November 1978
High-gradient Magnetic
Separation for Removal
of Sulfur from Coal
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. Approval does not signify that the contents necessarily reflect
the views and policies of the Government, nor does mention of trade names or
commercial products constitute 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 $6.00
Microfiche S3.00
Frfmd In *• Ihlwrf *•** •• *•*•!€«
UJOOt T«*»k«l lirfoMMOT C«Mr, O+ IMp, T«
-------
FE-8969-1
(EPA-600/7-78-208)
November 1978
Distribution Category UC-90b
High-gradient Magnetic Separation
for Removal of Sulfur from Coal
by
F.E. Luborsky
General Electric Company
Corporate Research and Development
P.O. Box 8
Schnectady, New York 12301
EPA/DoE Interagency Agreement No. OXE685AK
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
-------
HIGH GRADIENT MAGNETIC SEPARATION
FOR REMOVAL OF SULFUR FROM COAL
F. E. Luborsky
FINAL REPORT
Period Covered: March 1, 1976 through January 31, 1977
Contract No. HO366008
BUREAU OF MINES
U. S. DEPARTMENT OF THE INTERIOR
PITTSBURGH, PENNSYLVANIA
February 1977
General Electric Company
Corporate Research and Development
Schenectady, New York
-------
ABSTRACT
In this project a Pennsylvania coal from the Upper Freeport seam was
thoroughly characterized physically, chemically, and magnetically. The
powdered coal was then subjected to high gradient magnetic separations,
as a function of magnetic field and fluid velocity, in both a water slurry
and an air dispersion.
Ash and pyritic sulfur reductions occurred with increasing magnetic
field intensities and decreasing fluid velocities. The best results were
obtained in water slurries where approximately fifty percent of the total
sulfur and fifty percent of the ash were removed. Air dispersions produced
insignificant results.
ii
-------
TABLE OF CONTENTS
Section Page
I SUMMARY 1-1
II INTRODUCTION 2-1
Objective 2-1
Background 2-1
Previous Work on Magnetic Desulfurization of Coal .... 2-1
III MAGNETIC SEPARATOR MODELING THEORY 3-1
Theoretical Analysis 3-1
Calculations for Ribbon-Like Fibers 3-1
Particle Trajectories with Buildup of Particles .... 3-2
Filter Performance for Ribbon-Like Fibers 3-4
Mechanical Capture of Particles 3-4
IV THE COAL USED 4-1
Source and Preparation 4-1
Magnetic Characterization 4-1
Structural and Chemical Characterization 4-11
Mossbauer Analysis 4-16
V MAGNETIC SEPARATION OF COAL AS A WATER SLURRY 5-1
The Separator System and Experimental Methods 5-1
Results and Analysis 5-1
Discussion and Conclusions 5-12
VI MAGNETIC SEPARATION OF COAL AS A DRY POWDER . . 6-1
Separator System and Experimental Methods 6-1
Results of HGMS on Dry Coal Powders 6-2
Conclusions and Discussions 6-5
VII RECOMMENDATIONS 7-1
References 7-2
Appendix A— TYPICAL COMPUTER PROGRAM TO CALCULATE
PERFORMANCE OF A HIGH GRADIENT MAGNETIC
SEPARATOR A'1
Appendix B— SUPPLEMENT TO FINAL REPORT (Period Covered:
February 1 through July 15, 1977) B-i
ill
-------
LIST OF ILLUSTRATIONS
Figure Page
1 Schematic Cross Section Through a Fiber Ribbon and
Coordinates Used, Showing Viscous and Magnetic Forces
Acting on a Particle Deposited in the nth Layer 3-2
2 Magnetization vs. Field at Room Temperature for Heavy
Fraction (p > 1.6) for the -200 Mesh Size Range from a
Delmont Coal Piece 4-8
3 Thermomagnetic Behavior, a vs. T, at 10 kOe in He Atmo-
sphere of a Heavy Fraction from -200 Mesh Sample of
Delmont Coal Piece 4-10
4 Scanning Electron Micrographs and Energy Dispersive
Analyses of (a) Overview of Strongly Magnetic Particles and
(b) Pyrite Particle from Powdered Delmont Coal 4-12
5 Scanning Electron Micrographs and Energy Dispersive
X-Ray Analysis of (a) Kaolinite Particle and (b) Calcite
Particle from Powdered Delmont Coal 4-13
6 Scanning Electron Micrographs in (a) and (b) and X-Ray
Maps in (c) and (d) of (b) for Fe and S, Respectively, on a
Polished Surface of a Piece of Delmont Coal 4-15
7 Mossbauer Spectrum from Delmont Coal at Room
Temperature 4-17
8 Mossbauer Spectrum from Fernglen Anthracite Coal at
Room Temperature 4-17
9 Mossbauer Spectra of a Sieved (100 x 200) Fraction of
Delmont Coal Compared to Its Light and Heavy Components
Separated by Sink-Float in CC14 4-18
10 Mossbauer Spectra of the Feed, Mags, and Tails from the
First Series of HGMS Runs: Sample #4 4-20
11 Sulfur Reduction as a Function of Field from Both Series
of Runs in the HGMS Apparatus 5-4
12 Ash Reduction as a Function of P'ield from Both Series of
Runs in the HGMS Apparatus 5-4
13 Sulfur Reductions as a Function of Velocity from Both
Series of Runs in the HGMS Apparatus 5-5
14 Ash Reduction as a Function of Velocity from Both Series
of Runs in the HGMS Apparatus 5-5
iv
-------
LIST OF ILLUSTRATIONS (Confd)
Figure Page
15 Magnetic Separation Apparatus for Second Series of Tests. . 5-6
16 Sulfur and Ash in the Product as a Function of Fluid
Velocity for a Variety of Applied Fields — Curves Drawn
Through Data for 20 kOe; Second Series of Runs 5-8
17 Sulfur and Ash in the Product as a Function of Applied
Field for a Variety of Fluid Velocities — Curves Drawn
Through Data for v « 2. 7 cm/s; Second Series of Runs .... 5-8
18 Calculated Curves and Experimental Data for Pyritic Sulfur
Reduction as a Function of Field 5-10
19 Calculated Curves and Experimental Data for Pyritic Sulfur
Reduction as a Function of Velocity, as Described in
Figure 15 Caption. '. , 5-10
20 Calculated Curves and Experimental Data for Ash Reduction
as a Function of Field 5-11
21 Calculated Curves and Experimental Data for Ash Reduction
as a Function of Velocity 5-11
22 Air System for Magnetic Separation of Dry Coal 6-1
23 Ratio of Ash Collected in Mags to Ash in Product
as a Function of (Field/Velocity) for Dry Powder
Separation 6-4
24 Ratio of Sulfur Collected in Mags to Sulfur in Product
as a Function of (Fie Id/Velocity) for Dry Powder
Separation 6-4
LIST OF TABLES
Table
I Analysis of the Complete Sample 4-2
II Detailed Washability Analysis; Raw Delmont Coal;
Size: 2 x 3/8-inch r. h 4-2
III Detailed Washability Analysis; Raw Delmont Coal;
Size: 3/8-inch r. h. x 28 mesh 4-3
IV Detailed Washability Analysis; Raw Delmont Coal;
Size: 2-inch r. h. x 28 mesh 4-3
-------
LIST OF TABLES (Confd)
Table Page
V Forms of Sulfur Analyses on Washability Fractions;
Size: 2 x 3/8-inch r. h 4-4
VI Forms of Sulfur Analyses on Washability Fractions;
Size: 3/8-inch r. h. x 28 mesh 4-4
VII Froth Flotation Analysis; Raw Delmont Coal;
Size: 28 mesh x 0 4-5
VIII Feed Analysis of Coal for the Two Series of HGMS Tests . . 4-5
IX Cumulative Summary of Magnetic Data on Selected Fractions
from Delmont Coal 4-7
X Ash Content After LTA from 100 x 200 Mesh Sample of
Chunk No. 2 of Delmont Coal 4-11
XI Energy Dispersive Analysis and X-Ray Diffraction Results
on Selected Samples of Delmont Coal 4-14
XII Mossbauer Analysis for Pyritic Sulfur in Feed, Mags, and
Tails of HGMS Run #4 from First Series of Tests, Compared
to Chemical Analysis 4-20
XIII Mossbauer Analysis of Samples at Various Stages in Their
Treatment During Sulfur Analysis 4-21
XIV Distribution of Ash and Sulfur in First Series of Runs 5-3
XV Second Series of HGMS Results on Water Slurry 5-7
XVI Input Parameters for Calculations 5-9
-------
Section I
SUMMARY
A Pennsylvania coal from an upper Freeport seam has been carefully char-
acterized physically, chemically, and magnetically. The measurements in-
cluded float sink separations; ash content, total sulfur, pyritic sulfur, and
sulfate sulfur as a function of size fraction; magnetic susceptibility, magneti-
zation, and Mossbauer spectroscopy; and optical and scanning electron micro-
scopy. High gradient magnetic separations were carried out as a function of
magnetic field and fluid velocity with the powdered coal as a water slurry and
as a dry air dispersion. In the water slurries both sulfur and ash were signif-
icantly reduced: 60 to 80$ of the pyritic sulfur, corresponding to ~50% of the
total sulfur, and ~50# of the ash were removed. In contrast to these good re-
sults, in the air dispersions the sulfur and ash reductions were much less
significant: the decrease in sulfur and ash was only marginally significant.
The behavior of the coal in the water slurry followed the behavior calculated
from the physical models, with the magnetic properties derived from our
magnetic measurements. The problem with the dry coal may be the result of
local air turbulence effects, or to changes in the magnetic and physical prop-
erties of the coal. We therefore recommend further work to try to understand
the difference in efficiency of separation in HGMS between water slurries and
dry dispersions. We further recommend that engineering development be
started on applications where HGMS can be applied to liquid slurries of coal
powders — for example, in beneficiating tailings of fines from conventional
coal cleaning processes, in cleaning coal for pipeline transport, or in cleaning
coal dispersed in oil.
The work on this project was accomplished by Corporate Research and
Development personnel of the General Electric Company (Schenectady, N. Y. )
with the leadership of F. E. Luborsky; by the personnel of the Francis Bitter
National Magnet Laboratory of the Massachusetts Institute of Technology
(Cambridge, Mass.) with the leadership of E. Maxwell; and by personnel of
the Eastern Associated Coal Corp. (Everett, Mass. ) with the leadership of
H.E. Harris.
1-1
-------
Section II
INTRODUCTION
OBJECTIVE
The objective of this program was to establish the technical feasibility tor
removing a substantial fraction of the inorganic sulfur from dry coal powders
at significant processing rates by high gradient magnetic separation (HGMS).
BACKGROUND
One of the United States' most vital current needs is an adequate supply
of clean fuel. Abundant reserves of coal lie untapped because no economical
means of reducing the sulfur content to acceptable levels exists. The U. S.
has between 30 and 40 percent of the known world reserves of commercially
producible coal, and our resources alone are estimated at 765 billion tons of
economically recoverable coal. This is roughly equivalent to at least 1. 5
trillion barrels of oil, which is about 2. 5 times the current proven worldwide
oil reserves. Coal is perhaps the most easily developed alternative energy
source of significance available to augment costly petroleum and natural gas
fuels. Among these untapped resources are large coal reserves east of the
Mississippi River which have a high percentage of pyritic sulfur but are low
in organic sulfur content. If sufficient pyritic sulfur were removed from
these coals, they could be reclassified as low-sulfur coals.
Because of the renewed interest in coal, there is now a growing effort to
create cost-effective, environmentally sound methods of utilizing coal as a
clean fuel. Almost all methods for cleaning coal require grinding to a size
where liberation of the sulfur-bearing phase occurs. Cleaning at the mine
site is generally applicable only for larger sizes so that shipping of fine dry
powders is avoided. However, for the major usage of coal (namely burning
in power plants), the coal is often ground to a fine powder before burning.
Thus, it is then already in a form particularly suited for physical separation
techniques. However, cleaning of coal fines in a water slurry may be impor-
tant in cleaning coal for pipeline transport, in cleaning coal from mine mouth
wet beneficiation processes to recover the fines, or in cleaning coal in oil
dispersions.
PREVIOUS WORK ON MAGNETIC DESULFURIZATION OF COAL
The first attempt to remove sulfides from finely ground coal by magnetic
means appeared in a 1957 German patent^ issued as an addition to a 1956
U. S. patent*2); Russian work(3> in 1958 also supports the results. This
early work used conventional magnets with shaped pole faces to achieve rea-
sonable field gradients and fields for the separating force. However, pre-
treatments of the coal at 200° to 300°C in air or steam were used to form
2-1
-------
films of ferromagnetic oxides on the surface of the paramagnetic sulfides to
increase their magnetic susceptibility and thus enhance their removal effi-
ciency. Studies of the same process had been reported in 1956 by Kester^.
A low-temperature roasting in an inert atmosphere has been proposed by Fine,
et alf5^ to enhance the magnetic susceptibility of the sulfur-bearing phases.
Benefaction of untreated powdered coals appeared to have been observed
(Ref. 6) in 1961, confirmed(T) in 1965, and studied in more detail*8) in 1967.
From 54 to 86 percent of the original pyritic sulfur was removed from four
different coals. Early efforts in magnetic separation of pyrite, such as
Leonard's work at West Virginia University, indicated technical feasibility;
but, as he pointed out, further commercial progress awaited the development
of more effective magnetic separators. In his work, four different U. S. coals
were pulverized and passed through a laboratory (Franz Isodynamic) separa-
tor as dry powders. As much as 84 percent of the pyritic sulfur was removed
from the Pittsburgh coal.
In contrast to these and other early efforts made with conventional mag-
netic circuit devices at relative low fields and trivial throughputs, recent de-
velopments in high gradient magnetic technology and in superconductivity open
an entirely new dimension to the potential usefulness of magnetic separation.
Trindade^-ll), working with the new high gradient magnetic separators'*2, 13)
developed at the Massachusetts Institute of Technology, achieved removal of
approximately 90 percent of the pyritic sulfur from one Brazilian coal using
water slurries. This substantial improvement over conventional processes
results in part from the ability of the high gradient magnetic separation pro-
cess to handle fine particles, the form of most pyritic sulfur left in coals by
conventional processes. Trindade's work was extremely promising but was
limited to water slurries of one type of coal and involved small-scale labora-
tory testing in a batch-type system. Further, for most applications dry coal
is required, and the removal of the water represents a significant loss in
energy. Thus, the performance of dry coal in the HGMS as well as the perfor-
mance of coals from a variety of coal fields is yet to be examined. However,
even for the wet slurries Trindade concluded that the HGMS process for desul-
furizing coal was economically viable.
High-gradient magnetic separation results on water slurries of other coals
have recently been reported. Murray**** used coals from Warrick County,
Indiana, containing over 4$ total sulfur, about half organic and half inorganic.
For the coal ground to 99$ below 74 ^im (200 mesh) using a stainless steel
roving packed in the canister, for 20 kOe applied field, he obtained reductions
of inorganic sulfur in two different samples of 55$ and 78$ for 120 s retention
time in the filter. Three passes through the filter at 30 s retention time in-
creased the removal of inorganic sulfur to 67$ and 85$.
Lin et al. " •*) aiso reported results for high gradient separation of water
slurries of coal. The coal used was an Illinois coal with 1. 11% pyritic sulfur,
0. 03$ sulfate sulfur, and 1. 90$ organic sulfur for a total of 3. 10$. This was
2-2
-------
ground to a mean particle size of 33 urn; 99$ less than 65 urn. A variety of
experimental variables was used. For a stainless steel packing, using 20 kOe,
28. 4$ of the total sulfur was removed. This would correspond to 75$ removal
of the pyritic sulfur. At the same time 45. 6$ of the ash minerals were removed.
An open field annular quadrupole magnet inside a corresponding slurry
channel has been used^S) on two Pennsylvania coals with very high pyritic con-
tents. This type of separator works by the development of a secondary circu-
lation transverse to the primary orbital circulation, which brings the particles
closer into the region of high field intensity. One of the coals yielded 81$ re-
covery of cleaned coal with 83$ removal of pyritic sulfur.
The only high gradient separation of dry coal powders was reported in 1976
by Murray(14) in the same work on water slurries described above. Stainless
steel screens were used in the canisters. After three passes of dry coal
through the filter, for the two different Indiana coals only 25$ and 39$ of the
inorganic sulfur was removed, in contrast to the 67$ and 85$ removed when
in the water slurry. There was no indication in the paper as to why the re-
moval of pyrite was less efficient for the dry powders.
These recent developments in the magnetic separation technology applied
to cleaning coal have been summarized in two recent papers by Liu and Lin'*''
and by Oder'^8). These have all been published since the work on the contract
started. This new work on cleaning water slurries of coal thoroughly confirms
the original publications of Trindade (based only on work using a Brazilian coal).
The one publication which has now appeared and is concerned with cleaning dry
coal powders in a high gradient separator has confirmed our suspicion that it
will not be as efficient as cleaning of water slurries but provides no answers
as to why.
2-3
-------
Section III
MAGNETIC SEPARATOR MODELING THEORY
The basic incentive of the modeling is the need to develop the conceptual
understanding, as well as the quantitative models, necessary to predict the
technical performance of a particular system so that the economics can be
predicted without extensive trial-and-error testing. In the next sections we
will outline the model.
The problem is to calculate the motion of a paramagnetic particle trav-
eling in a fluid stream in the presence of a magnetized fiber. First calcu-
lations'* ^* 20) assumed a fiber with circular cross section, with its long
axis perpendicular to the field and velocity vectors, and magnetized uni-
formly to saturation. These analyses followed the analogous analysis^21)
carried out for electrostatic filters. In order to arrive at an exact solu-
tion to the equations of motion, only the magnetic and drag forces were con-
sidered. Simplifying assumptions were made by neglecting fluid boundary
layer changes around the fiber. For the fluid flow, two cases were con-
sidered: (1) for high Reynolds numbers where the inertial forces are much
greater than the frictional forces, the potential flow equation is valid, and
(2) for Reynolds number «1 where the inertial forces are included in an ap-
proximate way.
THEORETICAL ANALYSIS
Calculations for Ribbon-Like Fibers
A model approximating the conditions used to obtain the experimental
results on CuO/Al2O3 particles^22* 2^) was next developed. A steel wool
fiber packing was used in the filter; thus the filter performance was calcu-
lated assuming an infinite ribbon instead of a rod. However, as a first ap-
proximation we assumed that the field and flow lines on the upstream side
of the fiber are the same as for the circular cylinder previously calculated
(Refs. 24-26).
Let us now compare the weight of CuO particles actually captured in the
filter, ~ 5 to 85 gms, compared to the maximum weight Wm that can be held
on the first monolayer in the active area. The active area is the area of the
one upstream edge of the fiber where particles are captured. From geo-
metric considerations based on Figure 1, assuming only 1/3 of the fibers
are now correctly oriented,
Wm = rrfPWpp (T +2P)/9STpf (1)
where W = total weight of fibers in the filter, pp and pf are the densities of
the particles and fibers, and f is the fraction of the fibers which are active
3-1
-------
Figure 1. Schematic Cross Section Through a Fiber Ribbon and Coordinates
Used, Showing Viscous and Magnetic Forces Acting on a Particle
Deposited in the nth Layer
in capturing particles, i. e., the capture efficiency. Using the values given
(Refs. 22, 23), Wm = 0. 52 gms for the first layer. Thus, it is apparent
that many layers of particles must build up on the fibers. We consider here
only the case for large Vm/V, where the buildup fans out as approximated
in Figure 1. (27, 28) Thus, it is readily shown that each layer r\ will hold
increasing weights
wm, n = TtfPWpp [T+2P(1 +2n)]/9STpf.
Particle Trajectories with Buildup of Particles
(2)
The fluid flow sees a changing radius as the particle layers build up,
given by
an = a + 2Pn.
Assuming potential flow as before, the flow potential is
Y(flow) = Vr cos9 + (Va&/r) cose.
(3)
(4)
The field potential is assumed to be unaffected by the buildup of particles,
i. e., the permeability of the particles is assumed to be near that of the
fluid. Then
(5)
Y (field) = -rH cos6 + [(ji - l)Ha2/(n + l)r] cos9
where n = permeability of the fiber.
3-2
-------
The fluid velocity components are then
Vr = 6 Y/6r = V cosG - (Van/r2) cos6
(6)
Ve = 6 Y/r66 = -V sin6 - (Va^/r2) sinG
and the field components
Hr = H cos6 + [(n - l)Ha2/(n + l)r2] cos6
(7)
He = -H sine + [(M - DHa2/(u + l)r2] sine J .
Since the experimental data for CuO covers fields from 5 kOe to 1 00 kOe
and fields > 2tTMs = 7500 Oe are required to saturate a cylindrical rod across
its diameter, we consider two regions. For H > 7500 Oe, the fiber is as-
sumed saturated. Then
+1) = 2rrMs/H (8)
for H < 7500 Oe we assume that p » 1 and then
(H- D/Oi + 1) = 1. (9)
The force on the particles
F = (4nP3/3) V (M-H)/2 (10)
for saturating fields is evaluated to be
Fr = (16TT2xMsP3a2/3r3)[(2nMa2/r2) + H cos 26] ~\
> (ID
F0 = -(16n2xMsP3a2/3r3) sin 26. J
The differential equations of motion for this system
dr/dt = Fr/6TrnP + Vr ^
) (12)
rde/dt = Fe/6TTTiP + V0 J
become
dra [1 - (an/ara)2] cos6 - (Vm/Vr|)[(2TTMs/Hrgl) + cos 26]
rad6 = [1 + (an/ara)2] sin 9 + (Vm/Vr|) sin26.
As for the case when n = 0, i.e., an/a = 1, this equation has an analytic
solution only when 2TiMs/H = 0; namely
3-3
-------
yS = ra[l - (an/ara)2] sine - (Vm/2Vr|) sin 26. (14)
The critical entering coordinate is now
C = (Vm/2V) (a/an)2, if (Vm/V) * 2^ (an/a)2")
, > (15)
= (constant)sing, if (Vm/V > 21/2 (an/a)2 J .
Some typical trajectories and the dependence of C on Vm/V are shown by
Luborsky(29) for various values of n. For smaller values of H, for ex-
ample when H 2rrMs, the complete equation (13) has been solved by numer-
ical integration, using the 4th order Runge-Kutta approximation. Only the
exact solution approximation is used in most of the calculations. With this
approximation, £ for an unsaturated fiber is also given by (15) but with Vm
replaced by Vm> u where
Vm,u = 4XP2H2/9iia. (16)
Filter Performance for Ribbon-Like Fibers
We have assumed that, on-average, only 1/2 as many ribbon-like fibers
are correctly oriented to interact with the particles in the stream as was
the case for cylindrical fibers. The fractional recovery of magnetic par-
ticles is now
R = 1 - exp (-fFLC/3S). (17)
MECHANICAL CAPTURE OF PARTICLES
Because of the very irregular nature of the packing of the steel wool
fibers in the filter, we expect a significant fraction of the particles to be
mechanically trapped. This has been found^30) to be true in electrostatic
filters. We therefore write, for the total fractional recovery of paramag-
netic particles
R = 1 - exp [-fFUC + S)/3S] (18)
where 2§ is a reduced cross section for mechanical capture.
The mechanical recovery fraction is
Rm = 1 - exp [-fFL?/3S]. (19)
We have found no means of calculating § from first principles. Instead we
say, a priori, that § should be proportional to S + 2P, and proportional to
1/V as suggested by the experimental data. The 1/V dependence is ex-
pected from the increasing kinetic energy of the particle, which will prevent
sticking or will knock off particles previously deposited. Thus
3-4
-------
§ = Q1 (S + 2P)/2Va (20)
where Q1 is a proportionality factor to be adjusted to fit for the entire
filter.
The purity, or "grade" G, of the captured material can now be calcu-
lated by assuming that the paramagnetic and diamagnetic particles are
mechanically captured with equal efficiency and that no other factors af-
fecting purity, such as agglomeration between A12O3 and CuO particles,
are significant. Then
G Rwp/(Rwp +Rmwd) (21)
where Wp = weight of paramagnetic and wj = weight of diamagnetic particles
entering. We next consider, in an approximate way, the effect of the bound-
ary layer, following the procedure of Luborsky. (31) jr0r impingement of a
fluid on the front of a circular cylinder, the boundary layer thickness, 699,
averaged over the region of interest: i. e., from 135° to 180°, is given ap-
proximately ^32) by
n1/2 (WV)1/2 (22)
°99
where r] and p are the fluid viscosity and density, respectively. This gives
a minimum value of 699 = 0. 0018 cm for n 0, and larger values as n in-
creases, for the maximum velocity of 20 cm/s used in the CuO-Al2O3 ex-
perimental work. Since the particle diameter of the CuO, 2P, is smaller
than 699, we use an entrance velocity, V, reduced by P/699 as the velocity
impinging on the particle. Thus
Fs, 6 6P2 (TTTiV3P/an)1/2 & + (an2/r2)] sin0 . (23)
As the layers buildup, the radial coordinate of the particles is assumed to
increase in steps given by
r = a +(2n + 1) P (24)
Setting Equation 11 for FQ equal to Equation 23, we can solve for the critical angle
9C for each layer of particles on the fiber structure. All particles in the region of
angles greater than 9C are thus swept off. From geometric considerations the
mass of particles removed from each layer per unit length of fiber is then:
WJ. = (TT - 49C) [2 (1 + 2n) P + T] PPp/2 (25)
where pp is the particle density. Applying the previously described corrections,
the weight of particles removed in each segment of the filter, I, is
Wr = (n- 49C) [2(1 +2n) P + T] PppW«/9pfSTL
(26)
3-5
-------
where we consider that only a fraction, f, of the fibers are active, only 1/3
are correctly oriented for capture, the packing fraction of particles in the
layer is 2/3, and W is the total weight of fibers, having a density of pf in
the total filter length, L.
In the previous calculation^ 9) the recovery of particles on the down-
stream edge of the fibers was not considered. This capture on the down-
stream edge is observed experimentally(27, 28, 33, 34) but cannot be accounted
for by tracing particle trajectories. We suggest that the particles arrive by
being swept off the front edge or by being caught up in the turbulent wake of
the fiber. Without defining the mechanism of arrival we now consider capture
on the downstream edge by assuming that capture of particles occurs with
the same geometry as on the upstream edge. We then apply criteria similar
to those used on the upstream edge to determine the fraction of the particles
remaining and their configuration. On the downstream edge the critical
angle occurs when the radial component of the magnetic force, Fr, is equal
to the radial component of the shear force, FS, r* where
Fr (16n2P3xMsa2/3r3)[(2TTMsa2/r2) +Hcos29] (27)
and
Fs, r = 6miPVr = 6miP [1 - (an2/r2)] VcosG. (28)
In this case the boundary layer thickness is taken as equivalent to that at
the end of a flat plate, namely
2 (29)
where Sn, the width of the ribbon, is a function of n given as Sn <=» S + 2nP.
Substituting
Fs> r = 1. 2TTP2 (TiV3p/Sn)1/2 [1 - (an2/r2)] cos6. (30)
Equating Equation 27 with 30 and solving for 9 gives the critical angle
for particles on each layer of the downstream edge of the fiber. The weight
of particles swept off is calculated from Equation 26 again.
The recovery and purity of the material captured in each element of the
filter is calculated as described previously, but the weight, Wr, is returned
to the stream to enter the next filter element together with the other par-
ticles in the stream. A typical computer program to carry out these calcu-
lations is listed in Appendix A.
The theoretical analysis developed above works well for separations in
liquids at reasonably low Reynolds numbers. However, in gas streams
inertia! forces on the particle should be considered. This has been discussed
3-6
-------
.
in some detail, for example by Lawson. v ' In addition, more complicated
fluid flow will probably exist, which will strongly affect capture efficiency and
the geometry of the buildup of the particles. These effects have not yet been
incorporated into the modeling equations.
3-7
-------
Section IV
THE COAL USED
SOURCE AND PREPARATION
The source of the coal to be used was decided on with the concurrence of
Dr. W. Spackman and Mr. Deurbrouck. It came from an Upper Freeport
seam in Westmoreland County, PA: the Delmont mine. An original sample
was rejected because of excessive roof clay. A more representative sample
of about 3000 Ibs labeled Delmont # 7945 was packed in drums for transit.
Prior to any processing, large pieces of rock were removed by hand, a
treatment that simulates commercial plant preparation, in which rock mate-
rial would be removed by a Bradford breaker before the raw coal was fed to
the preparation plant. The coal sample was screened at 2 inches and the
+2-inch material was crushed to minus 2 inches and then recombined with the
original minus 2-inch material. This sample was then thoroughly mixed and
a subsample of about 1200 Ibs was obtained from the larger sample. This
1200 Ib sample was subsequently subdivided into 6 samples, each containing
about 200 Ibs. One of these samples was air dried and sized; wet screening
was used for material less than 100 mesh. The size analysis is shown in
Table I. Ash and the forms of sulfur and their percentages for the various
size fractions are also given in Table I.
Float-sink separations were carried out and ash, total sulfur, and forms
of sulfur were made on the various size and specific gravity fractions. The distribu-
tion of ash and total sulfur in the float-sink separations are given in Tables II
and III for two different size ranges and in Table IV for the combined range of
sizes, representing 88. 1$ of the total sample. The percentages of the three
forms of sulfur (pyritic, sulfate, and organic) in the float-sink fractions are
given in Tables V and VI for the same two size ranges reported in Tables II
and III. Froth flotation tests were run on the 28 mesh x 0 size. All analyses,
except forms of sulfur, were run on the froth increments. These results are
shown in Table VII. Two independent series of separation experiments were car-
ried out with this coal but with different size fractions. The analyses of this feed
coal for the two series are shown in Table VIII. In the first series a 48x100
mesh fraction was used, but in the second series the entire coal was ground
to -60 mesh.
MAGNETIC CHARACTERIZATION
Magnetic susceptibility measurements were carried out on various sam-
ples of the Delmont coal supplied by EAC, using a vibrating sample
4-1
-------
Table I
ANALYSIS OF THE COMPLETE SAMPLE
DISTRIBUTION
Cum.
Weight Weight
FORMS OF SULFUR ANALYSES
ASTM Designation 2492-68
Sulfate Organic
Total Sulfur Pyritic Sulfur Sulfur Sulfur
ASH
2
1 1/2
1
3/4
1/2
3/8
1/4
1/8
14
28
48
100
200
Size
x 1 1/2-in. i.h.
x 1 -in. i.h.
x 3/4 -in. r.h.
x 1/2 -in. r.h.
x 3/8 -in. r.h.
x 1/4 -in. r.h.
x 1/8 -in. r.h.
In. i.h. x 14 mesh
x 28 mesh
x 48 mesh
x 100 mesh
x 200 mesh
x 325 mesh
minus 325 mesh
Percent Percent Percent
2.2
5.5
5.6
10. 1
6.4
15.3
19.9
13.2
9.9
3.4
3.8
1.8
0.6
2.3
2.
7.
13.
23.
29.
45.
65.
78.
88.
91.
95.
97.
97.
100.
DETAILED
2 2.84
7 1.84
3 1.89
4 1.64
8 1.58
1 1.54
0 1.51
2 1.77
1 1.54
5 1.73
3 2.23
1 2.10
7 1.63
1.03
Table II
Percent Percent Percent
2.48
1.37
1.50
1.26
0.97
1.07
<
0.36
0.47
0.39
0.38
0.61
0.47
u
1.04 -, 0.47
1.12 a? 0.65
1.10
1.00
1.25
0.87
0.88
0.51 '
0.44
0.73
0.98
1.23
0.73
0.52
Percent
27.8
38.8
39.0
26.9
18.8
15.2
12.6
10.7
10.1
10.3
10.8
11.2
11.8
16.7
WASHABILITY ANALYSIS
Raw Delmont Coal
Sink
1.30
1.35
1.40
1.45
1.50
1.55
1.60
Weight
Float t
i.30 .41.2
1.35 13.5
1.40 3.8
1.45 2.4
1.50 1.5
i.55 2.1
1.60 2.6
32.9
Ash
t
3.6
8.5
12.8
16.6
20.7
20.9
25.9
61.6
Size:
(29.
Sulfur
I
0.90
1.38
2.80
4.21
2.74
2.88
3.21
3.25
2 x 3/8-inch r.
h.
8 percent of total)
CUMULATIVE FLOAT CUMULATIVE SINK
Height A*h
Z I
41.2 3.6
54.7 4.8
58.5 5.3
60.9 5.8
62.4 6.1
64.5 6.6
67.1 7.4
100.0 25.2
Sulfur Weight Ash
Z Z Z
0.90 100.0 25.2
1.02 58.8 40.3
1.13 45.3 49.8
1.25 41.5 53.2
1.29 39.1 55.5
1.34 37.6 56.8
1.42 35.5 59.0
2.02 32.9 61.6
Sulfur
Z
2.02
2.82
3.23
3.27
3.21
3.23
3.25
3.25
4-2
-------
Table III
DETAILED WASHABILITY ANALYSIS
Raw Delmont Coal
Size: 3/8-inch r. h. x 28 mesh
Sink
1.30
1.35
1.60
1.45
1.30
1.55
1.60
Float
1.30
1.35
1.40
1.65
1.50
1.55
1.60
-
Weight
I
66.0
12.8
3.0
1.2
0.9
0.7
0.9
14.5
Ash
Z
2.6
7.6
13.0
16.7
20.6
23.5
31.0
66.4
(58.
Sulfur
I
0.69
1.22
2.46
2.97
3.24
3.30
3.68
5.12
DETAILED
3 percent of total)
CUMULATIVE FLOAT
Weight
I
66.0
78.8
81.8
83.0
83.9
84.6
85.5
100.0
Table IV
Ash
I
2.6
3.4
3.8
4.0
4.1
4.3
4.6
13.5
Sulfur
Z
0.69
0.78
0.84
0.87
0.89
0.91
0.94
1.55
CUMULATIVE SINK
Weight
Z
100.0
34.0
21.2
18.2
17.0
16.1
15.4
14.5
Ash
Z
13.5
34.8
51.2
57.4
60.3
62.6
64.3
66.4
Sulfur
Z
1.55
3.21
4.42
4.74
4.86
4.96
5.03
5.12
WASHABILITY ANALYSIS
Raw Delmont
Size: 2 -inch r. h.
Sink
1.30
1.35
1.60
1.45
l.SO
1.-.5
1.60
Float
1.30
1.35
1.40
1.45
1.50
1.55
1.60
.
Weight
z
57.6
13.1
3.3
1.6
1.0
1.2
1.5
20.7
Aflh
I
2.9
7.9
12.9
16.7
20.6
22.7
29.3
64.8
(88.
Sulfur
I
0.76
1.28
2.58
3.05
2.90
3.15
3.52
4.49
Coal
x 28 mesh
1 percent of total)
CUMULATIVE FLOAT
Weight
Z
57.6
70.7
74.0
75.6
76.6
77.8
79.3
100.0
Ash
Z
2.9
3.8
4.2
4.5
4.7
5.0
5.4
17.7
Sulfur
Z
0.76
0.86
0.93
0.98
1.00
1.04
1.08
1.79
CUMULATIVE
Weight
Z
100.0
42.4
29.3
26.0
24.4
23.4
22.2
20.7
Ash
Z
17.7
37.9
51.3
56.2
58.7
60.4
62.4
64.8
SINK
Sulfur
Z
1.79
31.9
4.04
4.22
4.30
4.36
4.42
4.49
-------
Table V
FORMS OF SULFUR ANALYSES
ON WASHABIUTY FRACTIONS
Size: 2 x 3/8-inch r. h.
(29. 8 percent of total)
Sink
1.30
1.35
1.40
1.45
1.50
1.55
1.60
Float
1.30
1.35
1.40
1.45
1.50
1.55
1.60
Height
Z
41.2
13.5
3.8
2.4
1.5
2.1
2.6
32.9
Aril Total
2 Sulfur. Z
3.6 0.90
8.5 1.38
12.8 2.80
16.6 4.21
20.7 2.74
20.9 2.88
25.9 3.21
61.6 3.25
Table VI
FORMS OF SULFUR
ON
Size
Pyritic Sulfate Organic
Sulfur. Z Sulfur, Z Sulfur. Z
0.54 '
1.02
2.29
(
^ 0.36
0.36
0.51
I
3.68 5 0.53
M
2.39 ^
(
2.76 9
2.82
3.21 *
ANALYSES
t 0.35
0.12
0.39
' 0.04
WASHABILITY FRACTIONS
: 3/8-inch r. h.
x 28 mesh
(58. 3 percent of total)
Sink
1.30
1.35
1.40
1.45
1.50
1.55
1.60
Float
1.30
1.35
1.40
1.45
1.50
1.55
1.60
Weight
Z
66.0
12.8
3.0,
1.2
0.9
0.7
0.9
14.5
Ash Total
X Sulfur, %
2.6 0.69
7.6 1.22
13.0 2.46
16.7 2.97
20.6 3.24
23.5 3.30
31.0 3.68
66.4 5.12
Pyritic Sulfate Organic
Sulfur, 7. Sulfur, 7. Sulfur, 7.
0.23
0.86
1.80
2.75 <
* 0.46
0.36
0.66
; 0.22
2.87 M 0.37
2.93 | 0.37
3.66
5.67*
L0.02
*This value is doubtful
4-4
-------
Table VII
FROTH FLOTATION ANALYSIS
Raw Delmont Coal
Increment
1
2
3
4
5
6
Tailings
Weight
I
31.9
42.8
13.0
4.1
3.1
2.4
2.7
Size:
.1.9 p<
Ash
%
3.1
4.8
11.5
28.5
41.6
57.3
75.5
28 mesh
srcent of
Sulfur
X
0.98
1.21
1.94
3.14
4.16
5.35
5.79
L X 0
total)
Weight
_*
31.9
74.7
87.7
91.8
94.8
97.3
100.0
-Cumulative--
Ash
J7.
3.1
4.1
5.2
6.2
7.4
8.6
10.4
Sulfur
%
0.98
1.11
1.24
1.32
1.41
1.51
1.62
Tine release method with MIBC
Table VIII
FEED ANALYSIS OF COAL FOR THE
TWO SERIES OF HGMS TESTS
Moisture
Ash
Total S
Pyritic S
Organic S
Size distribution
FIRST SERIES SECOND SERIES
0. 6toO. 8%
16. 9£
1.58
1. 30
0.28
60 x 0
12.6%
1.84
0. 94
0.90
48 x 100 44. 1
100 x 200 21. 2
200 x 325 7. 1
-325 27.0
4-5
-------
magnetometer (VSM) of the Foner design (PAR Model FM-1) at room tempera-
ture and at lower temperatures in a liquid helium dewar, in fields ranging from
-20 kOe to +20 kOe. For a few samples, thermomagnetic characterization,
i. e., magnetization at fixed field vs temperature, up to 600°C in a helium at-
mosphere was carried out.
As received, a cylindrical chunk of Delmont coal was found to be diamagnetic
at room temperature with Xg -0. 6 ± 0. 1 x 10~6 emu/g-Oe and with no dis-
cernible ferromagnetic component of magnetization. A sample from the same
piece (originally ~3 in. cube) was ground in a porcelain mortar and pestle.
No sieving or powder classification was made. The powdered coal also showed
a diamagnetic susceptibility of about the same value. These results agree
with literature values.
In order to focus on the inorganic minerals presumed to carry sulfur and
iron, a one-step float-sink separation was effected. Coal was hand ground
and sieved with Tyler sieves. The fractions between mesh sizes (-32, +100),
(500 nmto250 |jm), (-100, +200), (75 urn to 150 urn), and -200 (less than 75 jam)
were separately collected and processed through a pear-shaped separatory
funnel with CC14. The density of CCLj is p as 1. 6, so that coal (p as 1. 25-1. 3)
will float, and most pyritic minerals and ash materials will sink if liberated.
About 2* of the (-100, +200) and -200 fractions sank, while about 5% of the
(-32, +100) fraction sank. Initial magnetic measurements on the heavy
fractions of the two larger sized powders yielded paramagnetic susceptibili-
ties at room temperature, with values Xg = + 2. 4 ± 0. 1 x 10~6 emu/g-Oe and
+2. 7 ± 0. 1 x 10~6 emu/g-Oe, respectively, but again with no discernible fer-
romagnetic component. (Table IX; samples 1-V3H and 1-V3HA.
Low temperature measurements of the heavy fraction of (-100, +200)
material showed the presence of a nearly Curie-law component. At 30K the
susceptibility had increased by a factor of seven to Xg = 17. 7 x 10~6 emu/g-Oe
(valid to 20 kOe) with negligible ferromagnetic component. A rough prelimi-
nary analysis of data between 6K and 70K, plus the 300K point permits a de-
scription in terms of a Curie-Weiss behavior and a temperature independent
term (including diamagnetism), i. e.
Xg = Xc_w + XTIP (31)
= C
T-6 TIP
Roughly 6 = -5 to -10K and XTIP«S 0 -» 2. 3 x 10"6 emu/g-Oe. Treating
C = Ncwg2 p2S(S+l) as applicable to Fe2+ ions (S - 2, gas 2. 1 to 2. 2), we find
Ncw as 1 x 1020 Fe2+ ions/gram of heavy fraction. This Ncw is not total iron
but is only that portion exhibiting a localized spin magnetic behavior (Curie-
Weiss behavior) in the as-received state. This result may be compared to
observations on all significant iron species as seen in the Mossbauer spectro-
meter.
4-6
-------
Table IX
CUMULATIVE SUMMARY OF MAGNETIC DATA
ON SELECTED FRACTIONS FROM DELMONT COAL(a)
Size Wt o0(R-T.)
(mesh) Fraction^) emu/g
32x100
100x200
100x200
0.05
0.02
~0
0.009
heavy fraction
whole coal
light fraction
0.017
(0. 013)
0.045
(0. 003)
0. 118
(0. 004)
3. 5xlO"6 + Curie-law
5. 4xlO'6
a) Samples designated 1-xxx, 2-xxx, --are from individual chunks.
b) Fraction of heavy to total material of this size cut.
c) Reported previously; H=heavy - sinks in CC14.
d) Frantz Isodynamic Separator runs. Field current settings, IH = !. 0 and 1. 5 A.
Strongest magnetic fraction (0. 5 amp) previously removed.
e) Subjected to LTA and remeasured.
f) Sampling problem: average OQ from two runs with o0=0. 023, 0. 010.
g) Calculated as per gram of material before LTA.
For further identification/characterization the heavy fraction of (-100,
+200) was passed through the Frantz Isodynamic Separator (forward slope 30°,
side slope +10°) at field current strengths of 0. 5, 1. 0, and 1. 5 A. The mag-
netic fractions have considerably more translucent or transparent crystals
t~50#) when viewed at 60X than the "nonmagnetic" fraction (~10#). The major
components of the "nonmagnetic" heavy mineral fraction are FeS2, kaolinite,
and CaCOs, as found by X-rays. The major X-ray-determined components of
the magnetic fraction appear to be calcite-siderite intermediate compositions
[(Ca; Fe)COs]. Further aspects of chemical and structural analyses are given
in the next section.
4-7
-------
Additional magnetization measurements were made on other liberated
mineral fractions (heavy mineral particles which sink in CC14) for various size
classifications, after beneficiation in the Frantz laboratory magnetic separa-
tor, and on a sample of the "mags" from a preliminary water slurry HGMS
run. Several of these same samples, along with their associated materials,
e. g., feed, tails, light fractions, etc. , were also examined by SEM (Scanning
Electron Microscopy), energy dispersive analysis, and Mossbauer spectro-
scopy, as reported in the next sections.
The magnetization measurements are summarized in Table IX. The
curves, such as those shown in Figure 2, of total magnetization per gram,
a, are decomposed for H £ 5 kOe, so that ? a0 + Xg H. Among notable fea-
tures is the reproducible "Curie-law" behavior. From X-ray, SEM data, and
Mossbauer spectra, we shall associate this partly with a siderite-calcite com-
ponent and partly with iron in a clay-like component of the mineral matter.
The variability from sample to sample is illustrated by the changes between
samples from chunks #1 and #2 (e.g., 1-xxx, 2-yyy). Whereas, in the
(100 x 200) size fraction it took magnetic separation to bring out a small
spontaneous magnetization component, ?o, in the heavy fraction from chunk
#1, this appeared readily without magnetic separation in chunk #2 (compare
samples 1-V3HA, 1-V4H, 2-V2H). Another aspect of the variability relating
to the extent of liberation is the very marked increase in CTO on going to -200
mesh (heavy) material in the case of chunk f 1 (1-V5H), which may be con-
trasted with the lack of increase for chunk £2 (2-V3H vs. 2-V2H).
0.04
2-V3H(-200)
ROOM TEMP
Figure 2.
8 12 16
MAGNETIC FIELD, H, kOe
Magnetization vs. Field at Room Temperature
for Heavy Fraction (p > 1. 6) for the -200 Mesh
Size Range from a Delmont Coal Piece
4-8
-------
Size distribution of pyrite is obviously all-important for liberation. SEM
studies on Iowa coals by Professor R. T. Greer of Iowa State University show
very important concentrations of pyrites at very small size: e. g. , below
10 urn, in contrast to optical microscopy. In Delmont coal we have not seen
such small pyrite particles. It may be argued that yg does not change much
with the mineral fraction particle sizes herein, and that the contribution to total
magnetization at 20 kOe and above comes mostly from the ^gH term of a - a0 +
XgH. (In other words, ao is not really important.) However, the Curie-law be-
havior of \g and our proposed attribution of this to siderite-calcite minerals
and iron-bearing clays leaves us uncertain as to the relevance of these argu-
ments to the desulfurization goal.
With regard to the data from the HGMS test sample (MIT, Run #4, mags)
it may be of interest that OQ has roughly the same value as found in the heavy
fractions of sample #2 and the laboratory-magnetically separated #1-V4H. If
the same Curie-law behavior held for this MIT mags sample, we should have
found \g (R. T.) « 1. 3 x 10~6 emu/g-Oe. The rather less certain and lower
value we report is either a real effect or a manifestation of experimental
difficulties. Interesting results on this sample from Mb'ssbauer spectrometry
are given in the next section.
Another category of result of significance concerns the changes in coal
by handling. A light fraction of Delmont No. 2 (100 x 200) was aged in room
air for the humid eight weeks of June and July. It developed a spontaneous
moment, ao (R. T.) 0. 13 emu/g (i. e., per gram of light fraction.'), with no
measureable \g (R- T.). If this value were to be referred to the LTA ash
residue, it would be 30x higher (i. e., o0 = 3. 5 emu/g!). An effort is made
below to identify what has happened. Freshly ground light fraction is diamag-
netic.
In a related (and controlled) experiment, freshly ground whole coal (-200)
was aged 15 hours in distilled water with a pinch of Alconox, so that the re-
sulting slurry had a pH of 7. 4. The sample developed a spontaneous magne-
tization o0 = 0. 022 emu/g with Xg ** °- This finding could be rather important
for interpretation of HGMS data.
The presence of spontaneous magnetization (CTQ) components in the various
samples described above is a continuing curiousity as well as a tantalizing fact
of potential importance for the magnetic separation process. The "curiosity"
arises because spontaneous magnetization is not a property of any of the domi-
nant minerals identified in the coal fractions studied. In Figure 3 we show a
thermomagnetic curve, a vs T, at a constant field, H, of 10 kOe for sample
1-V5H (-200), a heavy mineral fraction selected because of its high value for
ao (R. T.). The sample magnetization a (10 kOe, T) was recorded continuously
during heating in a helium gas atmosphere to 600°C and back. About half of
the magnetization is associated with a Curie point near 300°C, a temperature
highly suggestive of the presence of a ferrimagnetic pyrrhotite (FexS, where
x is slightly less than unity). Its magnitude is very small, being about one or
two percent of the estimated pyrite, which, in turn, is about one to two percent
of the coal. Thus, the (presumed) pyrrhotite escapes detection in other methods.
4-9
-------
0.18
0.16
0.14
o.Q.12
3
E
o>
- 0.10
QJ
o
§ 0.08
b
0.06
OJ04
0.02
l-V5H-(-200), He
I
I
100
200 300 400
TEMPERATURE, °C
500
600
Figure 3. Thermomagnetic Behavior, a vs. T, at 10 kOe in He Atmo-
sphere of a Heavy Fraction from -200 Mesh Sample of
Delmont Coal Piece. Average heating rate, 5°C/min.
It is tempting to speculate from this evidence that the pyrrhotite occurs with
the pyrite (surface coating?) and facilitates the magnetic desulfurization.
The peak at 500°C is probably associated with a conversion to FeaO4 from
Fe§2 and perhaps from other iron-bearing minerals. The apparent Curie
temperatures of 5 50° to 580° Care appropriate for high-iron spinels. The high
value of a obtained on cooling is typical for such a conversion.
In order to obtain a larger magnetization signal, low-temperature ashing
(LTA) was used on some selected samples. The low-temperature ashing was
carried out in an oxygen plasma at 145°C at the ERDA Morgantown Energy
Research Center through the kind cooperation of W. F. Law son, Jr., and
J. J. Kovach. Samples of sieved coal, heavy fraction, and light fraction
were tested by Mossbauer spectrometry and magnetization before shipping
out for LTA. They were then reexamined on return.
There were three samples: whole coal, light fraction, and heavy fraction,
all originally from a particular chunk (No. 2) ground to 100 x 200 mesh. Mag-
netization of the heavy fraction, No. 2-V2H, is reported in Table IX. Moss-
bauer spectra for all three samples are described later. The ash amounts
after LTA are shown in Table X. These values are lower than expected,
4-10
-------
Table X
ASH CONTENT AFTER LTA FROM 100 x 200 MESH
SAMPLE OF CHUNK NO. 2 OF DELMONT COAL
Sample Fraction Weight (# ash)
Whole 7. 53
Heavy 75. 80
Light 3. 68
based on our previous chemical analyses of the homogenized Delmont coal.
The magnetization data after LTA are also reported in Table IX in terms of
grams of LTA ash and as calculated per gram of material before LTA.
The appearance of a significant QQ in the light-fraction post-LTA samples
is a surprise: either it was present in the whole coal before LTA but escaped
detection, or it represents a change produced by handling or by the LTA pro-
cess. It is not a major component (cf. Mossbauer). It is amusing (coinciden-
tal?) to note that the post-LTA values on light and whole samples are similar
in magnitude to those reported by Trindade (Thesis, MIT) on feed coal mea-
sured after LTA. Such data figured in his magnetic evidences for desulfuriza-
tion, but the present results are anticorrelated with pyrites (expected to be
more in heavy fraction). They do signify that a "paramagnetic" ash is concen-
trated by HGMS.
Thermomagnetic analysis of the light-fraction post-LTA material in
helium (as for pre-LTA heavy fraction shown in Figure 3) displays a hint of
a pyrrhotite Curie point near 300°C plus a larger fraction Tc near 490°C. The
cooling data show a conversion to a high Tc (585°C) material, probably Fe3O4,
and another lower Tc component (250° - 275 °C) suggestive of the cooling be-
havior of pyrrhotite.
STRUCTURAL AND CHEMICAL CHARACTERIZATION
The results of SEM and the associated energy dispersive analysis for
representative heavy fraction particles and for an overview thereof appear
as typical photographs and CRT traces in Figures 4 and 5, as well as in a
more complete report in Table XI. Results from X-ray diffraction are also
presented in the same table. The major work was on heavy fraction (100 x 200)
material from coal chunk #1, divided up by passage through the Frantz Iso-
dynamic (laboratory) Separator into Strongly Magnetic (IH 0- 5 A), Moder-
ately Magnetic (IH 1- °. 1- 5 A), and Nonmagnetic. Some microprobe work
on a light fraction from chunk #2 and on a polished chunk has also been done.
In the photographs there is an overview of the strongly magnetic fraction
at 100X and of three individual particles at 500X that typify the major minerals
4-11
-------
la) OVERVIEW
(b) Fe, S PARTICLE (PYRITE)
Ji
Na.MgCI K Ti Cr
Zn ADD!
Figure 4. Scanning Electron Micrographs and Energy Dispersive
Analyses of (a) Overview of Strongly Magnetic Particles and
(b) Pyrite Particle from Powdered Delmont Coal. The
lower analysis curve is for the circled cluster of particles
with only the additional elements identified.
present (identified by their energy dispersive analysis traces and corroborated
by overall X-ray results). These are as follows:
1. Fe, S particle; pyrite,
2. Al, Si particle; kaolinite, [(OH)8Si4Al4OioJ
3. Ca particle; calcite, CaCO3.
In the table we report peak height intensity (arbitrary units) from the
energy dispersive analysis for several individual pyrite particles, their ratio
with respect to iron, and similar data for the three overviews taken with
4-12
-------
(a)AI.Si PARTICLE(KAOLINITE)
m
x
•'. -
(b)Ca PARTICLE (CALCITE)
Figure 5. Scanning Electron Micrographs and Energy Dispersive
X-Ray Analysis of (a) Kaolinite Particle and (b) Calcite
Particle from Powdered Delmont Coal
equivalent conditions, so that comparisons among these latter are permitted.
We note that, within a factor of about two, the peak height intensity roughly
represents concentration. A number of comments may be made on these
results:
1.
3.
The S/Fe ratio on the individual (presumably) FeS2 particles sets a
chemical scale. The ratio agrees to within 10<# to 2(K of a crude
calculation of the intensity ratio for FeS2-
The S/Fe ratio in the non-mag and mod, mag overview samples approx-
imates that for FeSa. For the strong mag overview sample the iron
must be elsewhere (e. g. , in the calcite-siderite phase).
Calcite-like minerals increase in importance as we go from non-mag
to strong mag.
4. Clay minerals (Si, Al, K) have the inverse behavior to calcite-like
minerals, but there seems to be a nontrivial clay carryover in the
strongly magnetic portion.
5. For the cluster particle the high S/ Fe is interesting but not understood.
4-13
-------
Table XI
ENERGY DISPERSIVE ANALYSIS AND X-RAY DIFFRACTION RESULTS
ON SELECTED SAMPLES OF DELMONT (UPPER FREEPORT) COAL
-------
(o)SEM
(b)SEM
(c)Fe MAP OF(b)
(d)S MAP OF(b)
4.
Figure 6. Scanning Electron Micrographs in (a) and (b)
and X-Ray Maps in (c) and (d) of (b) for Fe and
S, Respectively, on a Polished Surface of a
Piece of Delmont Coal. The smallest specks
are extraneous dust particles. The pyrite par-
ticles are ~100 um up to ~300 urn in (b).
some pyrite elements. This agrees with X-ray data but tells us
about the nature of the distribution. Typical element groups appear-
ing are Al, Si, K, Ca, Fe and Al, Si, S, Ca, Fe, Cu.
X-ray analysis of the light fraction after LTA shows the minerals
listed in Table XI in order of decreasing abundance.
4-15
-------
MOSS BAUER ANALYSIS
Mossbauer spectroscopy has been used under this contract because of its
unique capabilities in characterizing the structural phases associated with
iron. Consequently, it is also used here to analyze quantitatively for FeS2.
The Mossbauer spectrum (room temperature) of the Delmont coal shows two
phases (see Figure 7). Phase A is conclusively identified as FeS2. The ab-
sorption spectrum is a quadrupole-split doublet (lines Al, A2) corresponding
to this weakly paramagnetic material. A second minor phase is identifiable.
Line B2 presumably corresponds to the right-most pair of a quadrupole-split
doublet. From the spectrum we know that the phase corresponding to lines
Bl, B2 is characteristic of a paramagnetic, strongly ionic, ferrous (Fe2+)
compound. Computer analysis of the spectrum would aid in the identification
of the B compound, and we would be able to extract the relative fraction of
the A and B phases (roughly proportional to line intensity).
A Fernglen anthracite coal sample was obtained for comparison. The
spectrum of this anthracite shows a third phase (Cl, C2) in addition to the
FeS2 and B phases evident in the Delmont coal (see Figure 8). Inasmuch as
the pyrite content of anthracite coals is much less than in bituminous coals,
the B and C phases appear to be present in larger relative quantity than the
B phase in the previous sample. Lines Cl, C2 correspond to another ferrous
(Fe2+) ionic, paramagnetic phase, more clearly identified below.
Three samples of (100 x 200) material were prepared from Delmont coal
chunk #2: i.e., whole (sieved) coal, and the heavy and light fractions (sink-
float in CC14). These same samples were sent for LTA at Morgantown.
Their Mossbauer spectra at room temperature appear in Figure 9, normalized
for equivalent pyrite absorption. The spectrum for whole coal closely resem-
bles that shown in Figure 7. It shows clearly the quadrupole-split doublet
A!-A2 characteristic of FeS2 and the secondary line B2 belonging to the as-
yet-unidentified doublet. To our surprise the B2 line almost disappeared in
the heavy fraction, leaving a very weak line, C^ whose existence on the
shoulder of B2 was suspected in some of our previous spectra. In comple-
mentary fashion, B2 was considerably enhanced in the light fraction. In later
work, carried out for quantitative evaluation, we observed a ten-fold increase
of pyrites in the heavy fraction and a ten-fold decrease of pyrites in the light
fraction, both with respect to the whole coal. Therefore, the enhancement
(depression) of B2 is largely apparent. In the light fraction B2 is less than in
the whole coal (about 1/4).
We are aware that the B2 line has been studied in some depth in the only
previously published Mossbauer investigation of coal we have been able to
find .(36) From those results, we have indicated the position of its mate,
BI, which occurs very close to Ax and consequently affects the "apparent"
intensity ratio of "A!": A^ Lefelhocz et al. (36) made a verv careful attempt
to identify the B!-B2 doublet. It appears to be high-spin Fe2+ ion octahedral
coordination. They leaned toward attributing its presence as "organic" iron
4-16
-------
Bl ? 82
Ml I
I I „
Al A2 /~"%-••'
o
en
DELMONT COAL
ENERGY -
Figure 7. Mossbauer Spectrum from Delmont
Coal at Room Temperature
BI?CI? C2B2
\ Al A2 .^\ ?
o
CO
^ FERN6LEN
E ANTHRACITE
ENERGY -
Figure 8. MSssbauer Spectrum from Fern-
glen Anthracite Coal at Room
Temperature
in the coal macerals, but they could not completely exclude its association
to a clay-like silicate mineral or gel. In our study we have tried to distin-
guish between these possibilities.
4-17
-------
Bl Cl C2
|
Al A2
\ Al A2
\
-
f
LIGHT \
.
*^ ^
fVUi^,^
' ^V
\ •
WHOLE
TT If lit Ttl
to
ENERGY ^
Figure 9. MSssbauer Spectra of a Sieved
(100 x 200) Fraction of DelmontCoal
Compared to Its Light and Heavy
Components Separated by Sink-Float
in CC14
In work outside this contract, we have found a strong enhancement of
this Bi-B2 line in a Pomeroy roof shale whose clay mineral content is higher
than in usual coals. However, the crucial test should follow from the low-
temperature ashing study. The Fe in a coal maceral would be profoundly
affected by the oxidation of the coal, but the clay mineral should, in principle,
remain unaffected by this process. Finally, if the Fe is organic and is oxi-
dized by LTA, its new state could well affect the magnetization characteriza-
tion of the coal fractions as well as their Mossbauer spectra. (As a high-spin
Fe2+ ion, it contributes to the paramagnetic susceptibility.) We remark that
the relative amount of Fe involved is not trivial (—10$) and that many previous
magnetization studies of coal have been preceded by LTA.
The Mossbauer spectrum of each of the three samples after LTA is
qualitatively similar to the pre-LTA results. (There are minor changes in
4-18
-------
relative intensities of Ai-A2, B2-B3, etc. spectra, but these may result from
absorption effects.) From this measure it appears that there was no signifi-
cant transformation of phases by the LTA process. This result is very im-
portant in the search for the origin of the Bj-B2 doublet in the spectrum. It
could not be organic Fe and remain unchanged. Hence, we lean more strongly
to the suggestion that it is a finely divided or gel-like clay mineral containing
Fe2+
In Section IV, "Structural and Chemical Characterization, " we noted sev-
eral SEM observations on Delmont coal of the coexistence of Fe with elements
normally belonging to clay minerals. These were in clay-like bands on a
polished coal piece and on light-fraction particles (100 x 200). Thus, an iden-
tification of the Bi-B2 doublet with a clay mineral containing Fe2+ is fully jus-
tified. Without going into detail on clay mineral compositions, our findings
point strongly to the illite and chlorite groups of clay minerals.
At the later stages of this work, R. Hucko of the Bureau of Mines kindly
helped us to locate references in contract reports and conference abstracts
of some additional Mossbauer 'and SEM studies regarding iron-bearing clays
in coal.'*'• °*°' These references were to work which followed that of
Lefelhocz and others^G) jn 1957 and which reached rather similar conclu-
sions.
We also examined the three components of a preliminary HGMS water
slurry run at MIT on Delmont Coal #7945, feed, mags, and tails; identified
as run #4, 5-20, 64 kOe. Their Mossbauer spectra are presented in Fig-
ure 10, normalized for equivalent pyrite absorption. Several results of inter-
est are contained herein.
First, we note that the C2 absorption line is present in the feed coal with
intensity equal to B2. In the mags, spectrum C2 is enhanced and well defined,
while it is suppressed in the tails, or product, spectrum. This absorption
corresponds well to one of the doublet lines belonging to siderite (FeCO3).
Its mate, GI, falls in the "valley" between A! and A& and can be inferred to
be present by looking at peak-to-valley ratios. The association is reinforced
by the SEM and X-ray evidence for the presence of calcite-siderite minerals
in the Frantz Separator magnetic fractions (Table XI). Thus this paramag-
netic component of the ash is also removed by magnetic separation. Finally,
we have completed a justification for the attribution of the Curie-law behavior
in the heavy-fraction magnetic susceptibility to a siderite-like mineral.
This series was also undertaken to attempt a semiquantitative analysis of
the pyritic sulfur. By using equal amounts of sample in the same configura-
tion, we hoped that absorption effects would be kept constant. As a rough
approximation we used the peak height intensity (of A^ as the measure, al-
though the procedure is capable of refinement (area under line, etc.). The
curves of Figure 10 have been normalized and are inapplicable in that form.
From the direct data, we obtain the results shown in Table XII for the
4-19
-------
Bl Cl C2 B2
'•• % Al A2 .,:,""V;''-
: -; TAILS
ii^r.-.Afc.vr.. r -...,........
1 --.--..-•.-v-:^V^
"• .S.'
' >• FEED
ENERGY
Figure 10. Mossbauer Spectra of the Feed, Mags, and Tails from
the First Series of HGMS Runs: Sample f 4
Table XII
MOSSBAUER ANALYSIS FOR PYRITIC SULFUR IN FEED,
MAGS, AND TAILS OF HGMS RUN #4 FROM FIRST SERIES
OF TESTS, COMPARED TO CHEMICAL ANALYSIS
Feed Mags Tails
Arbitrary units 0.034 0.059* 0.015
Normalized to QQ^
feed analysis in wt. %
Chemical analysis
in wt. %
*This value could be higher, owing to self-absorption
effects at that level.
4-20
-------
approximate relative concentrations of FeS2 in the captured materials (mags)
and in the product (tails). These are compared in the table to the direct
chemical analysis.
The Mossbauer technique has also been tried in order to check on pyrite
removal in the chemical analysis performed according to ASTM standards.
Questions have been raised along this line by studies that show important con-
centrations of pyrite particles in the 1 to 10-micron-size range in certain
coals. Our exposure to this question is from the work of Professor R. T.
Greer of Iowa State University, who questions whether pyrite particles that
are well encapsulated by coal can be leached by the dilute HNOs used for
pyritic sulfur determination. Samples of the Delmont coal were obtained
from EAC at various stages in their analysis for S, to see if the FeS2 removal
was as expected. Three samples were tested (1) untreated powder; (2) pow-
der after HC1 treatment, which should take out all of the Fe except that asso-
ciated with FeS2; and (3) powder after HNO3 treatment, which takes out the
FeS2- Two samples of each were run. With the appropriate peak heights as
an approximate measure of concentration, the results obtained are shown in
Table XIII. It is clear that most of the nonpyritic iron is removed by the
HC1 treatment as expected (but not all), but more than 20% of the pyritic iron
appears also to be removed.' The HNOa treatment does remove essentially
all of the remainder of the pyritic iron, as expected. We suggest that the
Mossbauer technique could be used for quantitative analysis for pyrite in coal.
Table XIII
MOSSBAUER ANALYSIS OF SAMPLES AT VARIOUS STAGES
IN THEIR TREATMENT DURING SULFUR ANALYSIS
Sample
Untreated
After HC1
After HN03
Relative Heights of Peak
A2 for FeS2 B2 for other Fe
First
Sample
4.2
3.4
~0. 3
New
Sample
4. 5
3.4
First
Sample
0.8
~0. 2
New
Sample
0. 9
~0. 15
4-21
-------
Section V
MAGNETIC SEPARATION OF COAL AS A WATER SLURRY
THE SEPARATOR SYSTEM AND EXPERIMENTAL METHODS
The 48 x 0 size fraction of the coal for the first series of experiments
was slurried in water to a slurry density of approximately 10%. About
25 ppm of Alconox was used as a dispersant. Because the sample con-
tained some rather large particles (~350|a), we used an expanded metal
stainless steel matrix, rather than steel wool, to avoid possible clogging.
A gravity feed system at constant head was used. In the first group of tests
the nominal flow velocity was 1. 5 cm/s and magnetic field strengths of 10,
20, 40, 60, and 80 kOe were used. In the second group of tests the field was
held constant at 20 kOe and nominal flow velocities of 1. 5, 3. 4, 4. 8, 5. 6,
and 11.6 cm/s were used. Two zero field runs were made at 3. 4 cm/s.
Both mags and tails were collected in each series of tests. The samples
were filtered to separate out the solids. They were then dried and analyzed.
A conventional canister and piping system with appropriate valving were used.
The tests were conducted by slurrying feed samples (150 to 200g) in a gal-
lon of water. The slurry was continuously stirred just before use. The
first five tests (samples 1 through 5) were run without a dispersant. The
later tests were run with approximately 0. Ig of Alconox added as the disper-
sant. The slurry was fed through the matrix by gravity; the flow was at con-
stant head and was limited by an orifice in the exit tube. Because of the
tendency of the slurry to settle out, water was added to the feed as the flow
proceeded, so that the actual slurry volume passed through the matrix was
usually 2 to 4 gallons. The net result is that, although the average superfi-
cial flow through the matrix was accurately determined, there were signifi-
cant fluctuations in the slurry density during any single test. These were
larger for the higher flow velocity measurements.
We also ran another set of tests on Delmont coal, tests similar in scope
to the earlier set but with an improved technique to minimize materials
losses in the system. (The general procedure is later illustrated in Figure 15.)
The feed slurry was contained in a chamber directly over the matrix and was
continually and vigorously stirred to promote sample uniformity. All of the
feed sample was forced to enter the matrix. Flow rate was controlled by
the size of the exit orifice and by maintaining constant head.
RESULTS AND ANALYSIS
In the first series of runs the material balance for both ash and sulfur
was unsatisfactory. The difficulty was traced to material losses in the sys-
tem. Two such losses could be accounted for in the following way.: (1) small
5-1
-------
fractions of the slurry feed that did not go through the separator and (2) a
so-called "clean-out" fraction obtained by scouring the matrix after each
magnetics product collection. Fortunately, these fractions had been saved.
When they were analyzed separately for ash and total sulfur, rather high
concentrations were found in the "slurry" fraction, evidently caused by
gravity settling of the heavier particles. With these additional data, a
better materials balance was obtained. Revised ash and sulfur concentra-
tions for each feed sample can then be calculated by summing the ash and
sulfur in all of the products, including the "slurry" and "clean-out" frac-
tions. These calculated sulfur and ash concentrations are listed in Table XIV
and are more realistic characterizations of the samples than those arrived at
by summing the sulfur/ash in the mags and product fraction.
The following dilemma arises in interpreting the results of the first runs.
If we calculate the sulfur/ash reduction using as the assumed feed composi-
tion, only that part of the sample which actually passed through the separator
(i. e., excluding the slurry remnant) we are not fairly evaluating magnetic
separation because the heavy slurry fraction, accidentally separated out,
contains a significant amount of pyritic sulfur and ash. This result will
yield an artificially low figure. An alternative calculation is to figure the
reduction on the basis of the total sulfur/ash in the sample, including the
slurry fraction in the feed. This is what would be obtained if all the sulfur/
ash in the slurry remnant passed through the separator. The true value lies
between these limits. In Table XIV we included the latter calculation.
To clean up the ambiguity arising from the presence of slurry remnants,
we ran the second series of tests which avoided the problem by an improved
technique of feeding the slurry. The results of these tests were in general
agreement into the results of the first series of tests using the alternative
method of calculation described above.
The numerical data of the analysis of ash and sulfur are shown in
Table XIV and plotted in Figures 11 through 14 with the square symbols.
The reduction in the pyritic sulfur is not plotted because of the uncertainties
in the feed composition described above.
In all of the tests significant reductions were observed in the tails, i. e.
the product, for both the pyritic sulfur and the ash. In most cases the total
sulfur content could be reduced to below 1%. It is particularly noteworthy
that the pyritic sulfur content of the tails goes down as the field is increased,
the lowest being 0. 22% at 85. 5 kOe.
Three consecutive tails fractions were collected for sample 5, and
these show the effect of loading. The first fraction has a pyritic sulfur
content of 0. 22%, the second 0. 52%, and the third 1.16%. The ash content
similarly increased from 5. 0% to 6. 3% to 8.6%.
5-2
-------
Table XIV
DISTRIBUTION OF ASH AND SULFUR
(FEED % CALCULATED) IN FIRST SERIES OF RUNS
RUN
1
2
3
4
5-1*
5-2
tn
1 5-3
CO
5t
6
7
8/9
10/12
11/13
14/15
16/17t
18/19
20/21
H
kOe
10.69
21.38
42.75
64.13
85.5
_
.
-
42.75
42.75
21.38
21.38
21. 38
21.38
21.38
0
0
V
cm/ s
2.80
2.80
2.80
2.80
2.80
_
-
-
2.80
2.80
1.51
3.26
4.55
5.86
9.80
1.74
2.98
Feed
wt.
g-
158.0
178.0
161.8
137.5
178.9
.
.
-
141. 1
145.8
213.0
203.6
216.9
199.3
155.0
202.2
226. 4
I prod.
wt.
g-
156.5
173.2
160.4
135. 1
174.3
_
.
-
140.4
142. 4
204.0
197. G
215. ' Ash
in
mags
21. 4
23.0
22. 4
20. 8
20. 4
20.5
-
20.5
27.0
2G. 1
19. 9
35.4
38. 1
32. 7
45. 6
13. 1
16. 2
< Red.
in
Ash
33.0
43.1
50.0
55. 2
-
_
-
47.9
49. 1
47.3
44.7
48. I,
1G. 7
10.8
8.5
3.7
4.5
* STOT
in feed
(calc)
1.07
1.57
1.59
1. 57
-
„
-
1.51
1.18
1. 44
1.58
1. 50
1. 3J
1.45
(1.28)
1.19
1.21,
< STOT
in
i SpYR
in
product product
0.
0.
0.
0.
0.
0.
1.
n.
0.
0.
0.
i.
i.
i.
i.
i.
96
92
78
70
74
97
47
84
70
77
77
81
08
14
04
04
12
0. 50
0. 4G
0. 33
0. 2G
0. 22
0. 52
1. 1G
0. 35
0. 26
0. 30
0. 36
0. 38
0. 71
0. 84
-
0. 68
0. 81
*SORG
in
product
0. 46
0. 46
0. 45
0. 44
0. 52
0. 45
0. 31
0. 49
0. 44
0. 47
0. 41
0. 43
0. 37
0. 30
-
0. 36
0. 31
t STOT
in
mags
1. 80
1 91
1. 70
1 87
l.'>5
.
.
1. 95
2. 51
2. 39
1. 80
3. 54
2. 11
2. 32
-
2. 22
2. 85
* SPYR
in
mags
1. 56
1. 54
1. 41
1 41
1. G3
_
-
-
2. 27
. 9G
1. 41
2. 97
2. 29
1. 47
-
1. 95
2. 53
* SORG
in
mags
0 24
0. 37
0.29
0. 4>>
0 32
_
-
-
0. 24
1 43
0. 39
0. 57
0 12
0. 85
-
0. 27
0. 32
t Re
in
STO:
10.3
41.4
50.9
55.4
-
.
-
•14. ••
40.7
4(i. 5
51. 3
49. 1
1(1.8
21.4
-
12. ..
11.1
* 3 consecutive samples
t Weighted average of the three
-------
I 1
v, cm/s
a FIRST SERIES 2.8
-•o SECOND SERIES 2.7+0.2.
40
FIELD, kOe
80
Figure 11.
Sulfur Reduction as a Function of Field from Both Series
of Runs in the HGMS Apparatus. Solid points are pyritic
S data.
100
80
2 60
a
UJ
cc
o
tr
40
20
I
v, cm/s
D FIRST SERIES 2.8
o SECOND SERIES 2.7±0.2
I
I
I
T
8
j_
i
'O
0
20
40
FIELD, kOe
60
80
Figure 12. Ash Reduction as a Function of Field from Both Series
of Runs in the HGMS Apparatus
5-4
-------
100
80
CO
o
o
o
60
40
£ 20
a.
I I I
H, kOe
D FIRST SERIES 21.4
'O SECOND SERIES 20.0
°XNXPYRITIC S
6 8 10
VELOCITY, cm/S
12
14
16
Figure 13. Sulfur Reductions as a Function of Velocity from Both Series
of Runs in the HGMS Apparatus. Solid points are pyritic S
data.
i i i
H, kOe
a FIRST SERIES 21.4
o SECOND SERIES 20.0
CO
6 8 10
VELOCITY, cm/S
12
14
Figure 14. Ash Reduction as a Function of Velocity from Both Series
of Runs in the HGMS Apparatus
5-5
-------
In the second series of tests the material balances were greatly im-
proved by using the system shown in Figure 15. The numerical results are
presented in Table XV. Because of the improved material balance, these
results show reasonable trends when plotted directly as total sulfur, pyritic
sulfur, and ash in the product (i. e., tails) as a function of field and velocity.
These results are shown in Figures 16 and 17. In addition, these results
are plotted as the reduction in total sulfur, pyritic sulfur, and ash in Fig-
ures 11-14 as they are for the first series of tests. The agreement between
the two sets of data is good.
-tXJ
WATER INLET FOR
CONSTANT HEAD
MIXER
COAL SLURRY
- TO MAGS
COLLECTION
MAGNET
EXPANDED
METAL
MATRIX
BACKWASH
CALIBRATED ORIFICE
PRODUCT.
COLLECTION
VESSEL
Figure 15. Magnetic Separation Apparatus
for Second Series of Tests
Calculations by means of the physical model described in an earlier sec-
tion were carried out. In order to apply this model to the case of coal sepa-
ration, we have chosen the set of parameters listed in Table XVI. In addition,
only collection on the front edge of the stainless steel fibers was considered
because of recent direct observations, (28» 33) which showed that collection
on the downstream side of the fibers was not significant. The dimensions
and saturation magnetization of the expanded metal used in the HGMS col-
lecting can were measured directly. The magnetic susceptibility of the coal
phases are average values deduced from our measurements as summarized
5-6
-------
Table XV
SECOND SERIES OF HGMS RESULTS ON WATER SLURRY
H
f kOe
1 20
2 40
3 40
4 10
5 GO
6 80
7 0
8 20
9 20
10 20
11 20
12 20
13 20
14 40
15 80
16 0
TABLE XV
V
cm/s
2.60
2.44
2.93
2.74
2.53
2.93
2.62
1.36
2.55
5.67
7.94
11.33
12.03
1.40
1.28
1.35
SECOND SERIES OF
% Ash
12.1
7.7
9.4
13.7
11.5
9.5
17.5
11.7
12.4
13.4
16.4
16.7
16.4
9.5
1 0. G
1G.2
PRODUCT
1.02
0.68
0.52
1.10
1.05
0.88
1.44
0.95
1. 1G
1.29
1.50
1.44
1.41
0.82
0.94
1. 21
HGMS RESULTS ON
0.69
0.29
0.42
0.76
0.50
0.51
1.24
0. G3
0. 7(i
0. 97
1.29
1.24
1. 19
0.40
0.78
0.83
WATER SLURRY
'oSQRG
0.33
0.39
0.40
0.34
0.25
0.37
0.20
0.32
0. 40
0. 32
0.21
0. 20
0.30
0.42
0. Ib
0. 38
% Ash
41. 3
43.9
44.0
37.0
36.6
40.8
27.6
37.3
34.4
55.1
61. G
47.5
5'). 0
42. 8
33.8
32. 1
Feed
Ash
Total S
Pyr. S
OrK. S
MAGS
% SXOT % SPYR
3.73
3.36
3. 92
3. 68
3. 18
3. 40
4.70
3.37
3. 48
4.92
4.52
2. 86
4. GO
2. GO
2. 94
5.22
Analysis
1 (!.!>%
1.58%
1 . 30%
0. 28%
2.55
3.21
3.62
3.29
2.56
2.87
4. 39
2.84
3. 17 '
4.81
3.78
1.92
3.53
3. 19
2. 32
4.54
%SORG
0.88
0.15
0.30
0.39
0.62
0.53
0. 31
0.53
0. 31
0.11
0.72
0. 94
1.07
0.41
0. (12
0. (18
%Red.
Ash
28.4
54.4
44.4
18.9
32.0
43.8
-3. G
30.8
2G. G
20. 7
3.0
1.2
3.0
43.8
37. 3
4. 1
% Red.
STOT
35.4
57.0
67. 1
30.4
33.5
44.3
8.9
39. 9
2G. G
18. 4
5. 1
8. 9
10.8
48. 1
40. 5
23. 4
%Red
SPYR
4G. 9
77.7
(17. 7
41.5
61.5
60.8
4.6
51.5
41.5
25.4
0.8
4. G
8.5
(19. 2
40.0
3(1.2
-------
i
00
V
1.3-1.4
• 02.4-2.9
a 5.67
07.94
vll.3-12
CO
o
ce
0
20
TOTAL S
PYRITIC S
H, kOe
A 0
+ 10
• 020
040
060
v80
10
o
cc
80
10
15
v-cm/s
Figure 16. Sulfur and Ash in the Product as a
Function of Fluid Velocity for a
Variety of Applied Fields—Curves
Drawn Through Data for 20 kOe;
Second Series of Runs
Figure 17.
Sulfur and Ash in the Product as a Func-
tion of Applied Field for a Variety of
Fluid Velocities—Curves Drawn Through
Data for v« 2. 7 cm/s; Second Series of
Runs
-------
Table XVI
INPUT PARAMETERS FOR CALCULATIONS
FEED
COAL SULFUR ASH
wd Total wt. , g 175
wt. , % 1.4 (S) 10
wp wt. , g 4. 59 (FeSj) 17.5
pp density, g/cms 1.3 5. 0 (FeSJ 2.1
Xm magn. susc. , emu/cm3 -0. 6 x 10'" [0. 075/H] + 1
-------
100
CO
(E
Q.
o
o
UJ
QE
50
cn
i-»
o
I 'I r
PYRITIC SULFUR
f=0.0028
Cffl/S
A 1.9 ±0.2
D 1.74
O 2.7 ±0.2
V2.98
• 3.26
• 4.95
A 9.8 ±0.1
O 7.94
9.8
11.7 ±0.4
cm/s
100
20 40 60
APPLIED FIELD, kOe
80
100
Figure 18. Calculated Curves and Experi-
mental Data for Pyritic Sulfur
Reduction as a Function of Field.
Solid curves calculated for parti-
cle size of 220 (am using the param-
eters shown in Table XVI - dotted
curve calculated for an assumed
distribution of liberated particles,
50% fully liberated and 50% half-
liberated; dashed curve calculated
assuming particle size distribution
given in the table.
CO
50
o
0
UJ
oc
PYRITIC SULFUR
f = 0.0028
kOe
85.5, 80
64.1, 60
42.8, 40
21.4, 20
10.7, 10
0
5.0
10
STREAM VELOCITY, cm/sec
Figure 19.
Calculated Curves and Experimental Data
for Pyritic Sulfur Reduction as a Func-
tion of Velocity, as Described in Fig-
ure 18 Caption
-------
cm/s
1.5 ±0.2
Dl.74
02.7 ±0.2
V2.98
• 3.26
• 4.55
A5.8 ±0.1
07.94
T9.80
-t-11.7 ±0.4
40 60 80
APPLIED FIELD, kOe
Figure 20. Calculated Curves and Experimental
Data for Ash Reduction as a Function
of Field
100
50
STREAM VELOCITY, cm/sec
kOe
• 85.5, 80
T64.I, 60
o 42.8, 40
D 21.4, 20
10.7, 10
0
I m 9
10
Figure 21. Calculated Curves and Experimental Data for Ash
Reduction as a Function of Velocity
5-11
-------
To see how sensitive the calculations are to the size distribution, we
included the simple size distribution listed in Table XVI without changing
_f or the active radius of curvature, ^ (or anything else). The results cal-
culated for only one H and v are shown by the dashed curves in Figures 18
and 19. The difference is not very dramatic but is in the right direction.
We then assumed an arbitrary function for the liberation of the magnetic
phases from the diamagnetic coal phase: 50% fully liberated and 50% half-
liberated. Again using the single particle size of 220 \jjrn. and without any
changes, we calculated the effect at one H and v, shown by the dotted curves
in Figures 18 and 19. Except for a small displacement, the functional de-
pendence appears to be unchanged.
DISCUSSION AND CONCLUSIONS
It is clear that the magnetic field and flow are affecting the removal of
pyritic sulfur and ash minerals in approximately the manner expected. In-
creasing field and decreasing flow both increase the recovery of the mag-
netic phase leading to a cleaner coal product. The model calculations can
be made to fit the data only approximately. This may be in part a result of
the large scatter in the results, especially for the pyritic sulfur. The effect
of the size distribution and degree of liberation assumed in the calculation
appeared to be minor.
In the case of the recovery of the ash, there appears to be a much
smaller observed field and flow dependence than calculated. This may be
accounted for by assuming that a large fraction of the ash minerals are non-
magnetic: approximately 30%.
The experimental scatter seen here is certainly very much larger than
observed in the CuO/Al2O3 model system^2, 23) and may be attributable to
the variability in the coal from sample to sample.
5-12
-------
Section VI
MAGNETIC SEPARATION OF COAL AS A DRY POWDER
SEPARATOR SYSTEM AND EXPERIMENTAL METHODS
The air system for the magnetic separation of coal is shown schematically
in Figure 22. Our first attempts to use a screw feeder for the coal were suc-
cessful with a distribution containing relatively coarse particles, but they were
unsuccessful on the 60 x 0 fraction, which contained many fines. The fine coal
had a tendency to pack, making the screw feeder inoperative. We therefore
devised a simple vibratory shaker type of feeder, which circumvented this
difficulty. Although the feed rate is not as uniform as it would be with an
operative screw feeder, this does not seem to be a problem. On the other
hand, because it is a simple batch device, we can determine the mass of the
feed sample very accurately and have been able to achieve an overall mass
material balance of the order of 1% or better. The first tests were made with
a canister 1 inch in diameter and 6 inches long.
VIBRATORY
THERMOMETER
AIR
PUMP
MANOMETER
VALVE
MATRIX
N
MAGNET
COLLECTION
BAG
Figure 22.
Air System for Magnetic
Separation of Dry Coal
In the preliminary tests it was shown that very good material balances
were achieved. However, the preliminary tests did not demonstrate any sig-
nificant reduction in the sulfur and ash content of the products. There is
definite evidence of magnetic differentiation between the mags and the product,
but it is overshadowed by another effect, which we at first believed to be electro-
static charging of the particles and subsequent agglomeration. Tests were made
both in a field and in zero field. Although there is an increase in the ratio
6-1
-------
in the presence of a field, the effect is small. Also the separation efficiency
does not improve with increase in H/v. However, there are sufficient data
points above the range of zero field separation to give us confidence that sul-
fur and ash are being separated magnetically in spite of the fact that some
other phenomenon is interfering. These tests were made mostly with an ex-
panded metal matrix in the canister; a few were run with stainless steel wool.
RESULTS OF HGMS ON DRY COAL POWDERS
In contrast to the successful magnetic separations performed on water
slurries of coal, the dry separations were disappointing. Although some de-
gree of magnetic selectivity was observed, we were not successful in this
limited test program in achieving separations at all comparable to those ob-
tained with the water slurries. Several hypotheses for this have been ad-
vanced, but in the limited time remaining for this work we could not test them
adequately. Following is a description of the tests, the results, and the ten-
tative conclusions we have drawn.
The first series of tests was run with a canister one inch in diameter
and 6 inches long with an expanded metal matrix similar to that used in the
water tests. The feed material was Delmont 60 x 0 reconstituted from sepa-
rate 2 x 1/4, 1/4 x 48, 48 x 0 fractions in the proportions that these fractions
had been derived from the original coal. Fifteen runs were executed with
fields which varied from 0 to 80 kOe and superficial air velocities of 32 to
915 cm/s. The reduction in the sulfur content of the product over that of the
feed was small and irregular, rarely more than 10 percent and sometimes
imperceptible. A more positive indication that some magnetic separation
was taking place was that the sulfur (and ash) content of the mags was always
definitely larger than that of the product. (These data will be presented
graphically later in this discussion.)
In the next series of 24 tests we substituted coarse steel wool for the
expanded metal in the matrix canister. Magnetic fields up to 50 kOe and
air velocities of 59 to 1, 018 cm/s were used. In several of these tests the
products were passed through the matrix anywhere from 2 to 4 times. No
substantial improvement in the sulfur or ash reduction was noted in these
tests either. We suspected that electrostatic charging effects were occur-
ring and grounded the matrix but with no improvement. In these tests, how-
ever, as in the preceding series, we again noted an enhancement in the sulfur
and ash content of the mags as compared to the product.
To further explore the possibility that electrostatic effects might be
present, we performed a simple experiment to see if the agitation of the coal
in the glass feed flask put a charge on the coal. After shaking in the flask we
placed a few grams of the coal on the electrode of an electroscope. A clearly
detectable charging effect was observed although the experiment was only
qualitative.
6-2
-------
We then replaced the glass flask by a metal flask and made sure that all
parts of the system were at the same potential by interconnecting and ground-
ing. Seven more runs were performed with this arrangement, again with no
significant change in the results. We concluded that the major problem was
not electrostatic charging.
A batch of 60 x 0 Delmont coal was next washed in methanol, filtered,
and dried. The assumption behind this was that the coal particles might be
coated with some foreign organic material causing them to agglomerate and
the methanol could possibly remove this coating. Again no substantial im-
provement was observed.
In the next series of tests we tested a batch of 60 x 100 Delmont coal
which had been produced by splitting off the -100 fraction. We also split off
a 60 x 400 fraction. Again the product was not significantly improved, but
the enhancement of sulfur and ash in the mags was somewhat greater than in
preceding runs.
We then ran a test using the mags that had been collected in some of our
earlier water slurry tests as'a feed material. These were, of course, dry.
The purpose of this test was to test the idea that some alteration of the pyrite
takes place in the water slurry enhancing the magnetic properties of the pyrite.
Because we had relatively little material to work with, we could not do exten-
sive testing. In these runs we found that in a field of 20 kOe and a velocity of
59 cm/s, 74 percent of the feed was trapped as mags. In zero field we ob-
served trapping of 23$ to 46$. This shows that the magnetic fraction we suc-
cessfully captured in a water slurry could also be captured in an air stream,
although not as completely. A sizable zero field capture also occurred. It
should be noted that this high trapping percent occurred with previously bene-
ficiated material, i. e., there were few coal particles to promote the unde-
sirable agglomeration hindering magnetic separation.
To determine whether or not the water had converted the pyrite in the
coal to a more magnetic form, we slurried some Delmont 60 x 400 coal in
water and let it stand for 121/2 hours in one case and 84 hours in another.
(This, of course, was a much longer period of time than the few minutes the
coal had been immersed in the original water slurry tests.) The coal was
then dried and injected into an air stream and fed into the magnetic separator
at a field of 20 kOe and a velocity of 59 cm/s. The results were disappointing.
No significant difference was observed between the mags and the product.
The result of all these tests is best summarized in the two accompanying
plots. Figures 23 and 24. We have plotted the ratio of the percent sulfur (or
ash) in the mags to the percent sulfur (or ash) in the product, for each test
run, vs. the parameter H/v. This separation parameter was easier to use
experimentally (in contrast to the product/feed ratio) because of the marginal
beneficiation. There is no significant correlation with H/v; however, the
plots are useful in setting off those data for which the ratio is noticeably
6-3
-------
V)
a*
ZERO FIELD
V///// LEVEL O
080kOe
Vs 915
SOkOe
V 900 -1000
I
(400 FRACTION
H'40kOe MIDDLINGS
AT HIGH VEL
inn
100
o o
o o
O
o
8
001
003
01
H/V
03
kO« -s/cm
1.0
Figure 23. Ratio of Ash Collected in Mags to Ash in Product as a Function
of (Field/Velocity) for Dry Powder Separation
5 3
I \
OlH-40 kOe, MIDDLINGS PROD
WASHED OFF AT 972 cm/s)
601400
FRACTION
(601100
FRACTION.
50 kOe •
V=900-I000\
,80 We
V=9I5
~ 080kOe
, ZERO FIELD
77777 LEVEL
I
o
o
8
o
OX) I
I
I
01 0.3
H/V kOe - «/cn
Figure 2 4. Ratio of Sulfur Collected in Mags to Sulfur in Product as
a Function of (Field/Velocity) for Dry Powder Separation
larger than the zero field background. These are, first, the runs in which
the fines were split off and a few others in which a high field in combination
with a high flow velocity was used. Generally the fraction of mags collected
ranged from 50$ to 10$, decreasing with increasing velocity, but showing
rather little field dependence as well as rather little correlation with sulfur
(ash) content.
6-4
-------
CONCLUSIONS AND DISCUSSIONS
Our conclusions at this stage are as follows:
(1) We have achieved only marginally observable desulfurization in air streams.
(2) The presence of fines -400 (or -100) impede magnetic separation,
possibly because they act to promote agglomeration of coal and pyrite.
(3) Our admittedly limited test gave no indication that the magnetic properties
of the pyrite were enhanced by water immersion, but these tests were not ex-
tensive enough to provide a final answer.
The question of why wet magnetic separation was successful and dry
separation thus far unsuccessful has not been definitively answered. It may
be that local turbulence of the air stream occurs in the vicinity of the indi-
vidual matrix fibers and that the viscous forces for this reason are much
greater in air than in water. We have also not definitively determined
whether or not chemical alteration of pyrite occurs in water. To answer
these questions some further testing is required.
*
To determine if there are inherent difficulties with dry separation be-
cause of exaggerated viscous forces arising from turbulence, we could look
at a system like a mixture of A^Oa and CuO which has been previously studied
in water slurries and for which data is available. Performing a dry magnetic
separation on this system, whose constituents have known magnetic properties,
will provide us with data we can compare directly to the water slurry results.
These tests should be supplemented with more extensive tests on coal
which has been treated in water of controlled pH. In addition, both magnetic
susceptibility and Mdssbauer measurements should be performed on coal so
treated, to decide more definitively if the magnetic properties are altered by
water treatment.
6-5
-------
Section VII
RECOMMENDATIONS
1. Given the background information we have developed on the perfor-
mance of water slurries of this coal, and the characterization of this coal, we
have a unique opportunity to try to understand the reasons for the poor perfor-
mance of this coal as a dry powder. We recommend further work to explore
this problem.
2. We have demonstrated, on another coal in a water slurry, the ability
of high gradient magnetic separation to remove most of the pyritic sulfur and
a substantial fraction of the ash. We recommend that applications develop-
ment work be initiated to apply HGMS to processes involving water or liquid
slurries of coal. There are many of these applications. For example water
slurries of coal fines are found in mine mouth coal cleaning operations and in
pipe line transport of coal. Coal in oil dispersions for firing boilers and coal
in various organic liquids are found in the initial steps of coal liquifaction
processes. .
7-1
-------
REFERENCES
1. S. Siddiqui, Desulfurization and Concentration of Coal, German Patent
1,005,012, March 28, 1957.
2. S. Siddiqui, Recovery of Coal Resins with Partial Desulfurization of
Coal, U.S. Patent 2, 272, 265, November 27, 1956.
3. A. Yurovsky and I. Remesnikov, "Thermomagnetic Method of Concen-
trating and Desulfurizing Coal," Coke and Chemistry, Vol. 12, 1958, p. 8.
4. W. Kester, "Magnetic Demineralization of Pulverized Coal," Mining
Engineering. Vol. 17, No. 5, 1965, p. 72.
5. A. H. Fine, M. Lowry, L. F. Power and G. H. Geiger, "A Proposed
Process for the Desulfurization of Finely Divided Coal by Flash Roasting
and Magnetic Separation," IEEE Trans, on Magn., Vol. MAG-12, 1976,
p. 523.
6. W. Kester, "The Effect of High Intensity Magnetic Cleaning on Pulverized,
High Sulfur, Bituminous Coals," Thesis, West Virginia University School
of Mines, 1966.
7. R. D. Harris, "Reducing the Sulfur Content of Steam Coal by Removing
Fine Iron Pyrite at the Power Station," Fall Meeting, Society of Mining
Engineers of A1ME, Phoenix, Arizona, October 7-9, 1965.
8. W. Kester, J. Leonard, and E. Wilson, "Reduction of Sulfur from
Steam Coal by Magnetic Methods," Mining Congress Journal, Vol. 53,
No. 7, 1967, p. 70.
9. S. C. Trindade, "Studies on the Magnetic Demineralization of Coal,"
Massachusetts Institute of Technology Dissertation No. 1329, April 1973.
10. S. Trindade and H. H. Kolm, "Magnetic Desulfurization of Coal," IEEE
Trans, on Magn., Vol. MAG-9, 1973, p. 310.
11. S. Trindade, J. B. Howard, H. H. Kolm, and G. J. Powers, Fuel, Vol. 53,
No. 3, 1974, p. 178.
12. Proceedings of the High Gradient Magnetic Separation Symposium,
edited by J. Oberteuffer and D. Kelland, Massachusetts Institute of
Technology, Francis Bitter National Magnet Laboratory, June 22, 1973.
13. J. Oberteuffer, "High Gradient Magnetic Separation," IEEE Trans, on
Magn., Vol. MAG-9, 1973, p. 303.
14. H. H. Murray, "Benefieiation of Selected Industrial Minerals and Coal by
High Intensity Magnetic Separation," IEEE Trans, on Magn., Vol MAG-12,
1976, p. 498.
15. C. J. Lin, Y. A. Liu, D. L. Vines, M. J. Oak, G. E. Crow, and E. L.
Huffman, "Studies on Sulfur Recovery from Coal Wastes and Prospective
Magnetic Separation Applications," IEEE Trans, on Magn., Vol. MAG-12,
1976, p. 513.
7-2
-------
16. E. Cohen and J. A. Good, "The Application of a Superconducting Magnet
System to the Cleaning and Desulfurization of Coal," IEEE Trans, on
Magn., Vol. MAG-12, 1976, p. 503.
17. Y. A. Liu and C. J. Lin, "Assessment of Sulfur and Ash Removal from
Coals by Magnetic Separation," IEEE Trans, on Magn., Vol. MAG-12,
1976, p. 538.
18. R. R. Oder, "High Gradient Magnetic Separation Theory and Applications,"
IEEE Trans, on Magn. , Vol. MAG-12, 1976, p. 428.
19. C.P. Bean, Bull. Am. Phys. Soc., Vol. 16, 1971, p. 350.
20. J. H. P. Watson, J. Appl. Phys., Vol. 44, 1973, p. 4209.
21. G. Zebel, J. Colloid Science, Vol. 20, 1965, p. 522.
22. J. A. Oberteuffer, IEEE Trans, on Magn., Vol. MAG-9, Sept. 1973,
pp. 303-306.
23. J. A. Oberteuffer, Proceedings of the High Gradient Magnetic Separation
Symposium, May 22, 1973, edited by J. A. Oberteuffer and D. Kelland,
Massachusetts Institute o£ Technology, Francis Bitter National Magnet
Laboratory, June 27, 1973.
24. C.P- Bean, Bull. Am. Phys. Soc., Vol. 16,1971, p. 350 and in more
detail in personal notes.
25. J.H. P. Watson, J. Appl. Phys., Vol. 44, Sept. 1973, pp. 4209-4213.
26. G. Zebel, J. Colloid Science, Vol. 20, 1965, pp. 522-543.
27. D. Himmelblau, "Observation and Modeling of Paramagnetic Particle
Trapping in a Magnetic Field," Massachusetts Institute of Technology,
Masters thesis, June 1973.
28. C. Cowen, F. Friedlaender, and R. Jaluria, IEEE Trans, on Magn.,
Vol. MAG-12, 1976, p. 898.
29. F. E. Luborsky and B. J. Drummond, IEEE Trans, on Magn., Vol.
MAG-11, 1975, p. 1696.
30. Electrochemistry of Cleaner Environments, edited by J. O'M. Bockris
(Plenum Press, N. Y., 1972), Chapter 5 by E. C. Potter, p. 138.
31. F. E. Luborsky and B. J. Drummond, IEEE Trans, on Magn., Vol.
MAG-12, 1976, p. 463.
32. H. Schlichting, Boundary Layer Theory (McGraw-Hill Co., Inc., New
York, 1960), Chap. XII.
33. C. Cowen, F. J. Friedlaender, and R. Jaluria, IEEE Trans, on Magn.,
Vol. MAG-11, 1975, p. 1600.
7-3
-------
34. E. Maxwell and D. Kelland, "Matrix Loading in High Gradient Magnetic
Separators," Digests of INTERMAG Conf. . 1975, IEEE Pub. 75CHO
932-4-MAG, Paper No. 24-5.
35. W. F. Lawson, Jr. "The Dynamics of Paramagnetic Particles Near a
Magnetized Wire," Thesis, West Virginia Univ., Morgantown W. VA.,
1976.
36. J. F. Lefelhocz, R. A. Friedel, and T. P. Kohman, "Mbssbauer Spectro-
scopy of Iron and Coal, " Geochimica Cosmochimica Acta, Vol. 31, 1967,
pp. 2261-73.
37. T. P. Kohman et al., USAEC Report NYO-844-81, II. A. 4., 30 June 1970,
pp. 31-34; USAEC Report COO-3236-3, II. A. 1, 31 May 1972, pp. 5-16.
38. L. F. Vassamillet, Proc. Seventh Nat. Conf. on Electron Probe Analysis,
San Francisco, CA., 17-21 July 1972.
7-4
-------
Appendix A
TYPICAL COMPUTER PROGRAM TO CALCULATE PERFORMANCE
OF A HIGH GRADIENT MAGNETIC SEPARATOR
1000 LET P-0
1010 LET H-3E+3
1020 SO TO 1130
1030 I? H>30E+3 THEM 3340
1040 If H>oOE+3 THEM 1170
1030 If H^40c»3 THEM 1150
1060 IF H>20E+3 THEM 1130
1070 IF HMOE»3 THEM 1110
1030 IF H>1E*3 THEM 1090
1093 LET H-10.69E*3
1100 30 TO 1130
1110 LET H-21.33E*3
1120 GO TO 1130
1130 LET H-42.73E»3
1140 60 TO 1130
1130 LET H-64.13E»3
1160 GO TO 1130
1170 LET H-33.3E+3
1130 LET V2-1.31
1130 IF V2-1.31 THEM 1330
1200 LET V2-1.74
1210 SO TO 1330
1220 LET V2-2.3
.' 0 50 TO 1330
IC.-.D LET V2-2.93
1250 50 TO 1330
1260 LET V2«3.26
1270 SO TO 1330
1230 LET V2-4.55
1290 50 TO 1330
1300 LET V2-3.36
!310 50 TO 1330
1320 LET V2«9.3
1330 IF P-0 THEM 1430
1340 IF P-.002 THEM 1430
1330 IF P-.011 THEM 1400
1360 IF P-.011 THEM 1370
1370 LET ?-.002
1330 LET S1-.004
1390 53 TO 1430
1400 LET P-.011
1410 LET S1-.02
1420 53 TO 1450
1430 LET P-.022
1440 LET *1-».04
1430 LET R6«0
14oO LET P3-1
1470 LET Dl-3
1-HO LET .I1..0032
I ) LET 5».01
1300 LET ,1-1416
1310 LET L-ll.?
1520 LET F«.1
1330 LET 32-149.3
1540 LET Ll-L'10
1330 LET Pl-1
1360 LET P2-0
1370 LET D2-7.36
1330 LET :-:0--.7E-6
1390 LET >:i"..073'H>*10E-6
1600 LET X2—.6E-6
0 LET X«?1»X1*P2»X2-XO
loiO LET T-.033
1630 LET S-.074
1640 LET 07-1
1630 LET V3*<3»3.14»X*H»H«P~2>'O»S«ai>
1660 IF H<7300 THEM 1630
1670 50 TO 1690
1630 LET V3»'<2»3.14»«)
1690 LET S9-0
1700 PPIMT
1710 PRIMT USIM5 1740."FIBER EFFIC.-"JC»» "MECH.aDJ.-MQ7
1720 PRIMT USIMS 1730. "FIBER . .ai-'lfli; "aiDrH.S«"JSJ
1730 PRIMT USIMS 1760. "THICKMESS-MT
A-l
-------
17402 ' Li_LuLi_LLi_LLLL3. -sass;: ' LLLLLLLLLLs. «S
1770; i-i-i-i_Li_LLLLLLLL". saaas • LLLLLLLL=. s=3=
1770 ??IMT
I73n PPIMT
ir?>i PPIMT -RPPLIED FIELD-"iH
1300 PPIMT UCIMG 1320i •EMTP.VELOCITY.«":V2:-;ia5.VELaCITY."":v3:
1310 PPIMT USIMJ5 1330«"V!fV-"V3'V2
1320: ' LLLLLLLLLLLLLL---. --" Li-Li_LLLLLLuLLs«a— — =3. «
I3?ri: " LLLLL-----3". -
1 = 4i.i PPINT
IrT.O .-ET ;.IO-ii
! " -') .-ET ?1-0
0 IF ?-. Oil THEM Ia10
15:0 I? ?-."02 THEM 1M"
15?" I.ET yi-i.i'---
1 -ii" .53 TO 1 ?20
1?30 IF ?-. Oil THEN 1 jrij
1?40 I? P-.003 THEM l?rn
If1?" tET 00-7A.4?
1?-JO '53 TO l?sn
l?rn ^ET 00- 1-5. £35
1 ??0 LET •:'£ = •''•••
1??" LET 1^-1
fOOO IF I 0 THEM £0?0
iOlO LET :.I4- .?I«JM. ».P1»J
ifiiO '30 TO i04n
iuii'1 LET ^I4'U4*-P1*U| •
i'i.'41'i LET ,.in.= ..il
;-050 LET ?'4«fi
iO-i" --ET -'1-00
in70 LET Fi-.''7» •>;••?•• •.ri
ipixn LET ?;-•-•: 1»F»L 1'Fc >
iiv-n ^ET si = 0
.ilO'i ,_£T 1 = 1*1
.110 ^ET :-»
0 IF I ? THEM 3J4i.
c.-O IF I- - THEM .JiVi
-•1*0 30 TO iiT'i
^17" PPIMT "I-:Tr»P-4.!-fr
-: : 30 ??IMT TflB- 1! > J "P":
.21 ?0 =?IMT TflB'l ?• ! -^2"!
iiwi'i P'PIMT Tf»P'i?> : -HI"!
iilO PPIMT TflB'3?>: u?-;
ii20 PPIMT TftB-4:.. : I":
£530 PPI.1T TflB • 57 >:?--:
^O LET N—l
0 LET M*M*1
70 ^ET R3-P4
LET P7-P6
LET Bl*V3"£»v£»i l*'£»N»P'ai > > •£•
2300 LET Y-f-Cl^r^Ll^Bl^^'J'S*
£310 LET P-M-EXP'V' >
2320 LET P4-ai*< '£»M*1>»?>
2330 LET ?5*S1+2»M»?
2340 LET Z8»e»P'2»TOP'3. 1416»5»V2'3»D3x?5>» . 1+ /
2330 LET Z4-<1»»3. 1416'2»P*3»
£3oO LET i3»22^i2»24>
2370 lr Z3-»l THEM 2400
2330 IF Z3.»l THEN 2400
2330 80 TO 2410
£400 LET 23-1
£410 LET Z1-
£430 LET y9»3. 14»?»'T*2»' 1»£»
2440 LET
2430 LET
2460 lr i!2.7334 THEM 2530
2430 LET
£500 LET
.' 0 LET P.O-U3
c.cO 50 TO 2340
2330 LET ?3«0
£340 LET y3*U2-P3
2330 LET P4-?3»P3
£3oO LET P6-R3+P7
2370 LET
2330 50 TO 2630
£5?0 PPIMT USIM5 2630. 1 ^Mi IMT <1 0^4*?*. 3< '1 0 '4. IMT • 1 0N4»W2*.3> - 1 0^
£oOO PPIMT USIM5 2660. IMT • 10"4»ai*.S"10"4!
£410 PPIMT USIM5 2670tINT<10^4»«9+.S/'10M!
2o20 PPIMT USIM5 2630. INT • 1 0"1»*.3> /I 0^1 5
£630 PPIMT US I MS 2670. IMT '1 0'4»»3».3 • '\ 0'4S
A-2
-------
£640 PRINT USING 2670.INT<10'4»y3».S>'\0'4
£^30>3 333 a.3333 3.3333
£6601 333.3333
2670: 3;:.03aa
i-j'-ft) LET i =. 1
£700 IF »l£ >-»y? THEN £720
£710 -50 TO 2260
2720 LET :.ii*ai-.Ri»yo>
c740 LET «(^-*yo
£730 LET -30* £
£320 PRINT TaB<33>;"tUM!iaRY-
2330 RPINT -L3ST N-";N
2330 PRINT Tf»B<37>: "INC.iia-5.PEC.aT. .»": - .?1»J5>
2360 PRINT -HECH.?EC.P?aC..Rl--;;»l!
2370 PPINT TaB(37);-INC.aECrl.REC.aT.,a-:Pl»y5
2380 Ir MO !
2900 PRINT TaB<37>;"TaT.aaS.aT.ReC.«"!
"tO 50 TO 2930
. JO PRINT "raT.i1ECr).WT.REC..y4--;iJ4;
2930 PRINT TaB<37>:-TaT.nas.aT.aec.»-s -y4
2940 PRINT -TaT.«aS.»rlECH.PEC.WT.--%>'< *100^Uoi
2960 PRINT Tr)B'37);"5Rr)DctSO--|50
3970 PRINT -Dianas. OUT. oo--;ao;
2930 PRINT TaBO7>;-?aRfW1fl6.aUT»Wl--|yi
2990 PRINT 'TDTBL SUEPT 0?r .R6-" IR6; T3B <37> ; • INC. SUcPT OrriR4-"»R4
3000 PRINT '
3010 LET Y3-
3080 IF PC. 022 THEN 3040
3030 63 TO 3030
3040 LET Y3-V3»Y6
3030 IF UOCJ6 THEN 3030
3060 LET Y9«
3070 60 TO 3090
3030 LET Y9»-U4
3090 IF P'.Oll THEN 3130
3100 IF P-.002 THEN 3130
3110 LET P9-100»Y9'17.5
3120 GO TD 3160
3130 LET ?3-100»Y9'17.3
3140 SO TO 3160
3130 LET P7-100»Y9'17.3
3160 IF I<9 THEM 3210
"0 PRINT -.3»aifV.R«-{P9;TaB<30)i".23»Wlt5'.R--;P3!TaB<33)f.25»W1.5'J»«":?7
3i30 PRINT -Rr.o«*.--»Y3.-v.ar--»100»Y3'17.3
3190 LET Y6-Y3
3200 IF I '9 THEN 3230
3210 RESTORE
3220 SO TO 2000
3230 IF P'. 002 THEM 3260
3240 IF P>. Oil THEN 1370
3230 IF P'. 022 THEM 1400
3260 IF V2-9.3 THEM 1030
3270 IF V2-3.36 THEN 1320
3230 IF V2-4.33 THEN 1300
3290 IF V2-3.26 THEM 1230
3300 IF V2-2.93 THEN 1260
3310 IF V2-2.3 THEM 1240
3320 IF V2-1.74 THEM 1220
3330 IF V2-1.31 THEN 1200
3340 END
•BYE
•••csaooccs o»«o t 1.21 1 •jifO TO c«r« I 1.21' 0'<
••n«« sHAotxa Off fir 10.237 ON 10^13^76
A-3
-------
HGMS PROGRAM PRINT-OUT MANIPULATIONS
GTSS SYSTEM
PROGRAM NAME:
The program as saved will print standard calculations for all
velocities and drive fields at I ** 9 and last N.
To print both 1=0 through I 9 and NO through last N,
at all velocities and drive fields, adjust program as follows:
(longest possible print-out)
LINE
NUMBER CHANGE FROM: CHANGE TO:
i\3° •n=-T ro3Q
To print any of the previous combinations at only one drive field, for
example, H = 62,000 Oe. Adjust the program as follows:
LINE
NUMBER CHANGE FROM: CHANGE TO:
IQIQ
1030
A-4
-------
Appendix B
HIGH GRADIENT MAGNETIC SEPARATION
FOR REMOVAL OF SULFUR FROM COAL
F.E. Luborsky
Supplement to Final Report
Period Covered: February 1 through July 15, 1977
Contract No. H0366008
BUREAU OF MINES
U.S. DEPARTMENT OF THE INTERIOR
PITTSBURGH, PENNSYLVANIA
September 19, 1977
General Electric Company
Corporate Research and Development
Schenectady, New York
B-i
-------
FOREWORD
The work on this project was accomplished by
Corporate Research and Development personnel of the
General Electric Company (Schenectady, N.Y.) with
the leadership of F.E. Luborsky; by the personnel
of the Francis Bitter National Magnet Laboratory
of the Massachusetts Institute of Technology (Cam-
bridge, Mass.) with the leadership of E. Maxwell;
and by personnel of the Eastern Associated Coal
Corp. (Everett, Mass.) with the leadership of
H.E. Harris.
B-ii
-------
TABLE OF CONTENTS
Section Page
I Summary B- 1
II Objective B- 1
III Background B- 1
IV Experimental Apparatus B- 1
V Coal Sample B- 3
VI Sample Size B- 3
VII Analysis Procedure B- 4
VIII Results B- 5
IX Conclusions B-14
<
X Recommendations B-15
LIST OF ILLUSTRATIONS
Figure Page
1 Dry Separation Apparatus B- 2
2 Eschka-Leco Correlation B- 5
3 (a) Dry Separation. Sulfur and Ash Reduction
as Function of Recovery, Delmont 7945 B-10
3(b) Dry Separation. Sulfur and Ash Reduction
as Function of Recovery, Delmont 8118 B-10
4 Wet Separation. Sulfur and Ash Reduction as
Function of Recovery for Delmont 7945 B-ll
5 Cumulative Reduction in Sulfur
and Ash for Run 11 B-13
6 Wet Separation. Sulfur and Ash Reduction
as Function of Field for Delmont 7945 B-14
B-iii
-------
LIST OF TABLES
Table Page
I Dry Separations B- 6
II Wet Separations B- 7
III Dry Separation - Delmont 7945 B- 8
IV Wet Separation - Delmont 7945 B- 9
B-iv
-------
I. SUMMARY
The removal of sulfur and ash from dry powdered coal by high
gradient magnetic separation has been successful. The pyritic
sulfur and ash reduction of the coal was equivalent to that ob-
tained in water slurries, but the percentage of coal recovered
was somewhat lower and lower throughputs were required. The
removal of the coal fines and the vibration of the canister
were both helpful in improving the separator performance.
Wet separations in methanol slurries were shown to be effect-
tive.
Tests on oxidized vs. fresh coal resulted in equivalent sepa-
rations.
II. OBJECTIVE
This report covers the supplemental HGMS tests on Upper
Freeport coal from the Delmont mine. The purpose of these tests
was to get additional data< on the dry separation of pyrite and
ash from coal in an effort to gain a better understanding of why
our earlier results on dry separation were unsatisfactory and
how they might be improved.
III. BACKGROUND
The earlier tests (described in the Final Report of Contract
No. H0366008, February 1977) were performed by injecting pulver-
ized coal into a relatively high-velocity air stream, which was
then passed through the separator in a manner analogous to that
done with water slurries. The poor separations were thought to
result from the agglomeration of coal and mineral particles. It
was observed that the separation was marginally better when the
fines were taken out, and it was hypothesized that fines promoted
agglomeration. We also suspected that there might be significant
turbulent flow in the neighborhood of individual matrix fibers,
which would result in large viscous forces on the particles, pos-
sibly making the retention of trapped particles on the matrix
very difficult. Electrostatic forces did not appear to be signifi-
cant factors.
IV. EXPERIMENTAL APPARATUS
In the dry separation tests carried out in this supplementary
program, we abandoned the use of an air stream for propelling the
coal through the matrix and instead used gravity feed, assisted by
vibration, to move the coal. A combination of mechanical and elec-
tromagnetic means was employed. These means are shown in Figure 1,
which is a sketch of the experimental apparatus. The matrix can-
ister and feed pipe were supported by a thick rubber diaphragm
fastened to the feed pipe at the top of the magnet opening. This
B-l
-------
COAL IN
MOTOR AND ECCENTRIC
RUBBER DIAPHRAGM
SUPPORT
a.c. COIL
w •vi .vvi
V///////A
Q.C. II,--! ' MATRIX
COLLECTION VESSEL
Figure 1. Dry Separation Apparatus
arrangement provided sufficient flexibility for both longitudinal
and lateral vibration.
Lateral mechanical vibration was provided by an eccentric
rotating shaft driven by a variable- speed motor and positioned to
strike the side of the feed pipe once per revolution. Longitudi-
nal electromagnetic vibration was provided by a coil on the matrix
canister energized with strong alternating current. A vibratory
force was exerted on the coil by the interaction of the dc field
of the magnet and the ac in the coil. No effort was made to quan-
tify the amplitude of vibration in absolute terms, but the vibra-
tion was adjusted to permit relatively easy motion of the coal
through the canister. The ac current through the coil was fixed,
so the amplitude of this component of the vibration did change
proportionately to the dc magnetic field.
B-2
-------
V. COAL SAMPLE
The coal sample we used for most of the tests was taken from
the same 60 mesh x 0 batch of Delmont 7945 coal used in the earlier
series of tests. Tests at EAC indicated that the original lot,
from which our sample had been prepared, had weathered signifi-
cantly, with measurable conversion of pyrite to sulfate. Eastern's
analysis as of June 27, 1977, was as follows:
Ash 14.6 %
Total sulfur 1.55%
Pyritic sulfur 0.75%
Sulfate sulfur 0.31%
Organic sulfur 0.49%
Late in the test period we did obtain a fresh sample of Del-
mont coal (#8118) from the mine and were able to carry out a few
dry separations on it. The EAC analysis of this material was
Ash 16.0 %
Total sulfur 3.42%
/
No analysis of the forms of sulfur was made.
In addition to the dry tests we performed some wet separa-
tions. In view of the fact that the coal had probably undergone
alteration since our earlier wet tests, this gave us a proper
baseline for comparison with the dry tests. It was also useful
to see, if possible, what effect sulfate conversion would have on
magnetic separation.
We also carried out a few wet separations using methanol and
methanol-water mixtures as the carrier fluid. The use of methanol
avoids the necessity of drying the coal after magnetic separation.
One can visualize a separation scheme in which a circulating load
of methanol is used to carry the coal through the separator. Pro-
vided that the methanol losses could be kept small, this should be
a workable technique which would combine the advantages of wet and
dry separations. Methanol-water mixtures could also be useful if
it turned out that drying would be significantly easier for some
methanol-water proportion than for water alone.
vi. SAMPLE SIZE
These tests were conducted with relatively small feed samples
of coal, of the order of 20 g in the dry tests and 80 g in the wet
tests. One reason for these small samples was that, because we had
a limited supply of the prepared coal on hand in our laboratory
and because of the short time available for testing, we could not
wait for the delivery of a fresh sample from the mine and its
B-3
-------
preparation by EAC. Second, it was easier to do the dry tests
with small samples; moreover, the mass materials balances were
very good — of the order of 1%. Inasmuch as our main purpose was
to investigate the factors which might inhibit or promote dry sep-
aration, rather than to simulate a large-scale process, there was
no advantage in working with large samples.
VII. ANALYSIS PROCEDURE
Most of the ash and sulfur analyses of our test products were
performed by EAC. In a few runs they determined the various forms
of sulfur present, but for the remainder only total sulfur analyses
were done. They analyzed some feed samples and the products (non-
magnetics) of each run. The magnetics (mags) were not analyzed by
them. Because of the tight time schedule, it was decided to con-
centrate on the product analyses.
We did carry out some total sulfur determinations in our own
laboratory using the Leco apparatus. Initially we tested a few
of the mags samples and later extended this to include non-mags
samples as well. In cross-checking with the EAC determinations we
found a systematic difference in total sulfur determinations be-
tween the EAC analyses by the Eschka method and ours by the Leco.
(EAC informs us that their Pittsburgh laboratory has also observed
systematic differences in the two methods.) In Figure 2 we show
the results of such a correlation for 17 samples. The correlation
between the two sets of observations, determined by a linear re-
gression, is
SEschka • ^O153 SLeco + °-2396'
If the slope of the line were assumed equal to one, the least
squares value of the intercept would be 0.2540.
A Leco standard sample (prepared by Leco) and rated as 2.45%
sulfur was analyzed by the Eschka method; the analysis yielded a
result of 2.68%.
Because we could hardly hope to completely resolve these dis-
crepancies within the scope of the present investigation, we have
referred all our data to the Eschka scale, which is the presently
accepted standard in the coal industry. Leco analyses were then
converted to equivalent Eschka values by means of the relation
SEschka = 1*0153 SL + O'23^6' Where we present data taken with
the Leco apparatus, we have already converted the values.
Because of the fact that our mass material balances were good,
we calculated the percentage of sulfur in each feed, using the
separate sulfur analysis data for mags and non-mags and assuming
perfect sulfur materials balance. Alternatively we could have
attempted to take a representative sample of each feed and ana-
lyzed it for sulfur. If we had been working with large samples,
this method probably would have been preferable, but because we
B-4
-------
Figure 2. Eschka-Leco Correlation
used small samples in these tests, and in view of the good mass
balance, the first approach probably has less error. In those
cases where a product was analyzed by both the Eschka and Leco
methods, we used the average figure, after adjusting the Leco
value to the Eschka scale as explained above. There was generally
good agreement between these separate determinations.
VIII. RESULTS
The scope of the experimental runs is set forth in Tables I
and II for the dry and wet runs, respectively. More detailed data
on these runs is given in Tables III and IV. In the dry separa-
tions the coal was fed into the system manually at a relatively slow
rate to avoid the accumulation of a slug of coal which could clog
the system. Tests were made at fields of 0 to 64.5. The coal sample
available to us was a 60 mesh x 0 size distribution, and it was used
in this form in several tests. We also separated some of this feed
material in +200 mesh and -200 mesh size fractions and tested these
separately. The non-mags fraction was captured by collecting the
material that passed through the matrix with the field on and with
B-5
-------
Table I - DRY SEPARATIONS
*
H
kOe
Matrix
% Reduction
Ash | S
%
Recov.
Vibra-
tion
Delmont 7945
2
3
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
20.0
20.0
20.0
50.0
64.5
64.5
64.5
64.5
20.0
64.5
64.5
64.5
64.5
64.5
50.0
64.5
20.0
64.5
20.0
64.5
0.0
64.5
0.0
64.5
64.5
64.5
64.5
64.5
64.5
64.5
ss wool
n
II
n
n
n
n
n
ii
n
n
x-met
x-met
ss wool
n
n
ii
n
ft
n
n
n
n
n
n
x-met
n
li
n
n
7.3
3.0
10.9
34.5
39.4
35.8
37.6
63.6
55.2
17.6
10.9
72.7
41.8
33.3
26.1
-1.5
-0.7
0.0
9.0
14.9
15.1
16.3
32.8
32.6
7.4
0.7
42.9
28.3
9.1
7.6
43.5
33.6
38.0
12.9
25.2
-10.7
19.0
0.0
29.7
23.7
21.2
34.9
18.2
90.8
94.7
92.4
84.8
83.7
73.5
44.0
70.0
25.8
85.5
93.5
52.0
21.0
61.0
81.5
53.8
54.9
67.5
87.6
49.0
88.1
61.4
98.0
68.3
28.7
52.9
21.8
43.2
both
n
n
it
ii
it
it
ii
M
II
II
e*m.
both
both
e.m.
e.m.
roech.
M
II
It
It
II
n
n
e.m.
both
e.m.
both
it
n
Delmont 8118
33
35
36
37
38
39
40
20.0
64.5
20.0
64.5
20.0
0.0
OiO
9.4
54.4
18.7
41.1
14.9
-0.7
-0.9
94.2
57.5
63.0
71.5
72.0
80.0
97.0
both
e.m.
e.m.
e.m.
e.m.
none
e.m.
Notes: 1. Expanded metal
B-6
-------
Table II - WET SEPARATIONS
I
H
kOe
V
cm/8
Size
% Reduction
Ash 1 S
%
Recov .
Medium
Delmont 7945
1W
2W
3W
4W
5W
6W
7W
8W
8W
10H
11W
12W
13W
14W
15W
16W
17W
18W
19W
64.5
50.0
40.0
20.0
0.0
64.5
64.5
20.0
0.0
64.5
64.5
20.0
64.5
20.0
64.5
20.0
64.5
20.0
64.5
60x0
60x0
60x0
60x0
60x0
60x0
60x0
60x0
60x0
60x0
+200
4-200
60x0
60x0
60x0
60x0
60x0
60x0
60x0
47.9
48.4
49.1
33.3
11.5
52.7
54.5
44.2
6.1
49.1
70.9
49.7
50.3
41.2
49.1
38.2
58.8
48.5
47.3
22.0
29.9
30.4
17.8
8.3
32.3
32.5
28.0
4.1
26.0
40.6
31.5
35.5
31.5
27.4
14.5
35.5
29.0
37.1
67.8
74.0
75.4
83.8
91.3
66.2
69.7
78.4
92.7
73.7
65.7
76.8
71.7
80.1
76.1
82.7
67.7
76.8
wl
W
W
W
W
W
M2
M
M
25% M
W
W
10% M
10% W
50% M
50% M
M
M
M
Notes:
1.
2.
Water
Methanol
B-7
-------
Table III - DRY SEPARATION - DELMONT 7945
1
2
3
5
6
7
a
9
10
11
12
13
14
IS
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Prod.
17.98
18.93
18.47
16.95
16.73
14.7
8.8
14.0
a14.9
^4.6
C18.6
17.1
18.7
10.2
4.2
12.2
16.3
6.4
5.0
8.1
10.6
5.0
8.9
6.2
9.8
6.9
2.9
3.7
7.2
3.7
1.6
Height i
Mags.
0.68
0.37
0.71
2. IS
2.30
4.0
10.6
6.0
0.8
2.4
0.6
9.3
14.6
7.6
3.0
5.6
4.3
3.7
1.4
5.1
0.0
3.8
0.0
2.7
7.2
2.7
24.7
1.9
2.2
n Grams
Residue
0.48
0.30
0.33
0.49
0.45
1.2
0.4
0.2
0.1
0.7
0.3
1.0
0.7
0.3
0.5
0.1
0.1
0.2
0.1
0.2
0.1
0.2
0.4
0.9
1.3
O.S
0.1
I
19.14
19.6
19.51
19.59
19.48
19.9
19.8
20.2
19.0
20.2
19.6
20.5
19.5
20.1
19.8
12.1
9.3
11.9
12.2
10.2
9.1
10.0
10.0
10.0
10.1
7.3
33.2
6.1
3.9
Feed
19.80
20.0
20.0
20.0
20.0
20.0
20.0
20.0
19.0
20.0
20.0
19.6
20.0
20.0
20.0
11.9
9.1
12.0
12.1
10.2
10.1
10.1
10.0
10.1
10.1
7.0
33.0
6.1
3.7
Eschka Leco
Product
1.38
1.38
1.36
1.22
1.14
1.18
1.16
0.84
0.94
1.32
1.68
1.34
1.43
0.78
1.02
1.19
1.29
0.76
0.94
0.73
1.10
0.84
0.92
1.31
1.53
1.29
1.29
0.74
0.93
1.32
1.15
0.87
1.01
0.88
1.28
1.13
1.55
1.32
1.64
1.09
1.16
0.91
1.04
0.84
t Total
Mags.
1.88
2.12
1.99
2.42
2.62
2.11
1.54
2.17
2.02
1.99
2.49
1.80
1.51
1.56
1.88
2.2S
2.05
2.64
2.85
1.88
2.13
2.70
1.67
1.21
1.66
1.17
0.98
Sulfur
Residue
1.60
1.79
1.76
1.94
2.27
1.76
1.58
1.28
1.47
1.65
1.79
1.53
1.44
1.41
1.86
1.43
1.49
1.27
1.41
1.48
1.71
1.46
2.10
1.18
1.64
1.47
1.29
Feed
1.36
1.37
1.36
1.34
1.31
1.39
1.35
1.25
1.38
1.42
1.37
1.38
1.36
1.38
1.32
1.54
1.52
1.42
1.47
1.51
1.40
1.63
1.64
1.55
1.52
1.06
1.52
0.96
% Red.
S
- 1.5
- 0.7
0.0
9.0
13.0
15.1
16.3
32.8
32.6
7.4
0.7
44.9
28.3
9.1
7.6
43.5
33.6
38.0
12.9
25.2
-10.7
19.0
0.0
29.7
23.7
21.2
34.9
18.2
t A
Prod.
15.3
16.0
14.7
10.8
10.0
10.6
10.3
6.0
7.4
13.4
24.0
13.6
14.7
4.5
9.6
11.0
12.2
sh
Feed
16.5
%Red.
Ash
7.3
3.0
10.9
34.5
39.4
35.8
37.6
63.6
55.2
17.6
10.9
72.7
41.8
33.3
26.1
t Rec-
covery
90.8
94.7
92.4
84.8
83.7
73.5
44.0
70.0
25.8
85.5
93.5
52.0
21.0
61.0
81.5
53.8
54.9
67.5
87.6
49.0
88.1
61.4
98.0
68.3
28.7
52.9
21.8
60.7
43.2
H
kOe
20.0
20.0
20.0
50.0
64.5
64.5
64. S
64.5
20.0
64.5
64.5
64.5
64.5
64.5
SO.O
64.5
20.0
64.5
20.0
64.5
00.0
64.5
00.0
64.5
64.5
64.5
64.5
64.5
64.5
Size
Fraction
60 x 0
60 x 0
60 x 0
60 x 0
60 x 0
60 x 0
60 x 0
+200
+200
-200
-200
+ 200
-200
60 x 0
60 x 0
+200
+200
+200
+200
60 x 0
60 x 0
-200
-200
60 x 0
60 x 0
60 x 0
60 x 0
60 x 0
60 x 0
w
I
oo
Notes: ). Consecutive products: see text
-------
Table III - DRY SEPARATION - DELMONT 7945 (Cont'd)
t
33
35
36
37
38
39
40
Prod
20.0
11.5
12.6
14.3
14.4
16.0
19.8
Weight
Mags
3.3
9.1
7.7
5.7
5.6
3.1
0.5
in Grams
E
23.3
20.6
20.3
20.0
20.0
19.1
20.3
Feed
21.24
20.0
20.0
20.0
20.0
20.0
20.42
» T
Prod
3.47
2.31
4.05
2.85
4.10
4.37
3.45
otal Sul
Mags
3.64
8.23
6.30
9.84
6.67
5.47
3.06
fur
Feed
3.83
5.07
4.98
4.84
4.82
4.34
3.42
% Red.
S
9.4
54.4
18.7
41.1
14.9
-0.7
-0.9
% Re-
covery
94.2
57.5
63.0
71.5
72.0
80.0
97.0
H
kOe
20.0
64.5
20.0
64.5
20.0
0.0
0.0
Size
Fraction
60 x 0
+ 200
+200
+200
+ 200
+ 20C
60 x 0
Table IV - WET SEPARATION - DELMONT 7945
1
IN
2W
3W
4H
5W
6H
7W
8W
9W
ION
11H
12W
13H
14H
15H
16H
17W
18W
19H
K
Prod
54.57
60.79
63.76
69.38
79.26
59.78
60.54
72.86
75.58
57.89
40.27
43.77
55.03
62.88
65.26
66.64
57.79
62.47
40.64s
eight i
Hags
24.85
21.68
19.25
12.28
6.42
29.11
25.47
19.07
5.65
17.41
19.52
12.09
20.02
13.01
19.63
13.59
25.62
18.40
12.73
i> Grams
I
79.46
82.47
83.01
81.66
85.68
88.89
86.01
91.93
81.23
75.30
59.79
55.86
75.05
75.89
84.89
80.23
83.41
80.87
53.37
Feed
80.5
82.1
84.6
82.8
86.8
90.3
86.8
92.9
81.5
78.6
61.3
57.0
76.7
78.5
85.8
80.6
85.3
81.3
67.2
% Total
Eschka Leco
Product
0.84
0.77
0.75
0.91
1.05
0.80
0.87
0.95
1.24
0.85
0.66
0.75
0.80
0.85
0.90
1.06
0.80
0.88
0.78
0.97
0.94
0.85
1.03
1.15
0.88
1.02
1.08
1.33
0.94
0.79
0.88
Sulfu
Hags
1.76
2.24
<2.39
2.47
2.65
2.12
2.53
2.84
2.10
2.50
2.35
2.68
r
Feed
1.16
1.22
1.15
1.18
1.20
1.24
1.40
1.41
1.34
1.21
1.22
1.19
1.241
I Red.
S
22.0
29.9
30.4
17.8
8.3
32.3
32.5
28.0
4.1
26.0
40.6
31.5
35.5
31.5
27.4
14.5
35.5
29.0
37.1
% i
Prod
8.60
8.52
8.40
11.0
14.6
7.8
7.5
9.2
15.5
8.4
4.8
8.3
8.2
9.7
8.4
10.2
6.8
8.5
8.7
tsh
Feed
16.52
% Red.
Ash
47.9
48.4
49.1
33.3
11.5
52.7
54.5
44.2
6.1
49.1
70.9
49.7
50.3
41.2
49.1
38.2
58.8
48.5
47.3
% Re-
covery
67.8
74
75.4
83.8
91.3
66.2
69.7
78.4
92.7
73.7
65.7
76.8
71.7
80.1
76.1
82.7
67.7
76.8
H
kOe
64.5
50.0
40.0
20.0
0.0
64.5
64.5
20.0
0.0
64.5
64.5
20.0
64.5
20.0
64.5
20.0-
64.5
20.0
64.5
Size
Fraction
60 * 0
60 x 0
60 x 0
60 x 0
60 x 0
60 x 0
60 x 0
60 X 0
60 x 0
60 X 0
+200
+200
60 X 0
60 x 0
60 X 0
60 X 0
60 x 0
60 X 0
60 X 0
Medium
W3
w
W
w
W
w
M«
H
M
25% M
H
H
10% M
10% M
50% M
50% M
M
M
M
Notes:
1. Average feed analysis for runs 1H-12W used for runs 13W-MW
2. EAC ash analysis of feed used throughout
3. Water
4.. Hethanol
5. Partial Product
vibration of the matrix. The mags fraction was collected by turn-
ing the field off and vibrating the matrix more vigorously. Fi-
nally, a residual mags fraction was collected by removing the
matrix canister and rapping out the remaining few tenths of a gram
of trapped material.*
The wet separations were carried out in the usual way (cf.
volume 1 of this report) using water, methanol, and water-methanol
mixtures. The product slurries collected were then filtered with
a pressure filter to recover the dry products.
Figures 3(a)f 3 (b), and 4 summarize the results of these
tests. In these figures we have plotted (on a weight basis) the
*This represents a normal equilibrium buildup, which would level
off in successive tests.
B-9
-------
1001 i 1 1 i
in
<
50
Q
Ul
CE
to
1
S 50
o
Figure 3(a).
o
a
0
o
a
X
H
50
SIZE
60 iO
64.5 -2OO -
«• 20 '20C
t64.5 -20C
20 -200J
* 0 -200
•6-
O
O Qf
50 100
% COAL RECOVERY
Dry Separation. Sulfur and Ash Reduction
as Function of Recovery, Delmont 7945
100
Figure 3(b)
H SIZE
• 642 •'200
* 28 *S22
« 5 '2OO
I O 6OIQ
a 20 eo>o
50 100
% COAL RECOVERY
Dry Separation. Sulfur and Ash Reduction
as Function of Recovery, Delmont 8118
B-10
-------
100
z
o
I-
o
o
ui
cr
50
UJ
H h
H SIZE FLUID
a
X
•
9
a
e
C
n
«
64.5 60x0
50
40
20
0
64.5
20
64.5
20
64.5 60 "0
20
64.5
645
20
0
* 200
+ 200
water
methanol
ii
water
10% methanol
25% methanol
50% methanol
II
methanol
D
a
I
50
% COAL RECOVERY
100
Figure 4.
Wet Separation. Sulfur and Ash Reduction as
Function of Recovery for Delmont 7945
percentage reduction in ash or sulfur in the non-mags, compared to
the feed, vs. the percentage of the feed recovered in the non-mags,
Percentage reduction =
% ash/sulfur in non-magsj
% ash/sulfur in feed
These plots are similar to the grade-recovery plots used in
mineral beneficiation. At 100% recovery there is zero improvement
in the grade (reduction of ash or sulfur), while, as 100% grade is
approached, the recovery becomes vanishingly small.
B-ll
-------
Figure 4 presents the results of the wet separations. The
separations on the +200 mesh fractions result in higher reductions
of ash and sulfur at some cost in coal recovery. The higher fields
also consistently yield products with greater reductions and lower
recoveries than lower fields. One could, however, draw single
curves which would be fair approximations to all the data points.
The results of the dry separations are shown in Figure 3(a)
and Figure 3(b). Here it is clear that a single curve would not
serve to represent the data. The +200 mesh data points yield re-
sults significantly better than those for 60 mesh x 0 fraction,
and these in turn are better than for -200 mesh fraction. Although
the data points for each set of conditions are few, it is clear that
they could not be even approximately represented by a single curve.
It is significant, however, that the best results (those obtained
with the +200 mesh fraction) are roughly the same for both the wet
and dry separations. This fact suggests that dry separations should
be feasible if the fines are split off and if a practical feeding
scheme can be devised which sifts the coal through a matrix with-
out using a high-velocity air stream (e.g., fluidized bed).
Fewer data were taken on the freshly mined Delmont 8118 coal.
These, however, are in general agreement with the data on the
weathered Delmont 7945. In fact, the results are a little better.
This is strong evidence that the success of pyrite removal in mag-
netic separation is not principally the result of oxidation of the
coal and would apply to freshly mined coal.
As mentioned earlier, we generally used a combination of
longitudinal vibration, induced electromagnetically, and lateral
mechanical vibration, generated by rapping on the feed pipe with
an eccentric rotating shaft. The longitudinal vibration was prob-
ably effective in moving the coal through the matrix; the lateral
mechanical vibration tended to prevent coal from accumulating at
the top of the matrix. There is some evidence that the mags frac-
tion may not be completely immobilized on the matrix but is slowed
down by the magnetic forces, while the clean coal moves more rap-
idly and accumulates more quickly in the collection vessel. In
Run II we collected three successive product fractions in the fol-
lowing manner:
Fraction a - no matrix vibration
Fraction b - electromagnetic vibration only
Fraction c - electromagnetic + mechanical vibration.
The sulfur and ash reductions for these fractions were as
follows:
B-12
-------
Fraction
a
b
c
Reduction, %
S Ash
32.6 55.2
4.7 18.8
-16.3 -45.5
a
a+b
a+b+c
Cumulative
% Reduction
S Ash
32.6 55.2
19.4 37.6
2.2 -1.8
Cumulative
Recovery, %
25.8
50.0
95.3
The cumulative reductions and recoveries for these fractions are
shown graphically in Figure 5.
100
g
o
o 50
UJ
IT
RUN II
ASH
0
Figure 5.
50
% COAL RECOVERY
Cumulative Reduction in Sulfur
and Ash for Run 11
^100
In Figure 6 we show the percentage reductions in ash and
sulfur for the wet separations vs. field strength. Higher reduc-
tions are obtained with +200 mesh fractions than with the 60 mesh
x 0 size distribution. The separations in methanol are better than
those in water; in methanol-water slurries the results fall in be-
tween. Methanol wets the coal easily and no dispersant is necessary,
B-13
-------
70
60
50
I
in
40
u
30
20
,o'
0
5O
I I
SIZE
FLUID
O 60«0 woltr
• « methonol
<» • 10% rtwthanol
e • 25% mtlhanol
• • 50% mclhonol
^ »200 water
30 40 50 60
80
Figure 6. Wet Separation. Sulfur and Ash Reduction
as Function of Field for Delmont 7945
even in the 10% methanol slurries. There is evidently less
tendency for particles to agglomerate.
IX. CONCLUSIONS
Dry magnetic separation by HGMS should be feasible if the coal
fines are separated out and if a suitable technique for grav-
ity feeding can be developed. Multiple passes may be desir-
able to increase the coal recovery. (In this work only single
passes were taken.)
Wet separation in methanol or in water-methanol mixtures is
an interesting alternative process deserving further investi-
gation. If the separations were done with a circulating load
of methanol, the drying problem would be eliminated or, in the
case of methanol-water mixtures, may be reduced. In the short
time available we could not investigate this problem further.
B-14
-------
3. Tests on oxidized coal and freshly mined coal indicate that
the pyrite removal is substantially the same in both cases.
X. RECOMMENDATIONS
Based on the satisfactory cleaning of sulfur and ash from a
water slurry of United States coal and the equally effective clean-
ing of dry powders of the same coal, we strongly recommend the con-
tinuation of this work. The following specific items of work
should be undertaken.
1. On dry coal cleaning, work should continue toward development
of a suitable feeding procedure to lead to maximum throughput
and maximum cleaning. Fluidized bed techniques or a develop-
ment of gravity feeding procedures are indicated as the direc-
tion for this work. This should lead to an initial evaluation
of the economics of the dry cleaning of this coal.
2. On water slurries of coal, a variety of coals mined in the
United States should be^ubjected to high gradient magnetic
separation, using the same equipment and procedures. This
will yield information on the variability of the cleaning of
coals from different sources, which is necessary to form the
basis for determining how generally this type of cleaning can
be applied.
3. The variety of coals to be tested should be the same as that
for dry powders.
4. Methanol and methanol-water slurries should be tested to see
in detail how well the high gradient cleaning works as com-
pared to water slurries. Further questions to be answered in-
clude the following: Are wetting agents or dispersants needed?
Under what conditions, how efficiently, and how much energy is
required to recover the methanol? This preliminary analysis
should be sufficient to evaluate whether or not further work
is justified.
5. Oil-coal slurries should also be evaluated because of their
potential in power plant applications.
6. All of the above work should be closely coupled with continu-
ing work on the fundamental understanding of the various as-
pects of the problem - for example: on developing better
physical modeling for the magnetic separation in both wet and
dry systems; on determining the mineral phases present, their
physical and magnetic properties, their changes during process-
ing, and means of enhancing their magnetic properties.
B-15
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
FE-8969-1 (EPA-600/7-78-208)
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE ANDSUBTITLE
High-gradient Magnetic Separation for Removal of
Sulfur from Coal
5. REPORT DATE
November 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
F.E. Luborsky
9. PERFORMING ORGANIZATION NAME AND ADDRESS
General Electric Company
Corporate Research and Development
PO Box 8
Schnectady. New York 12301
10. PROGRAM ELEMENT NO.
EHE623A
11. CONTRACT/GRAN F NO.
EPA Interagency Agreement
DXE 685 AK
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
0 PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/13
is. SUPPLEMENTARY NOTESJJERL-RTP project officer is David A. Kirchgessner, MD-61, 919/
541-2851. DoE project officer is R.E. Hucko, Div. of Solid Fuel Mining and Prepar-
ation. Pittsburgh PA 15213.
16. ABSTRACT
The report gives results of a thorough physical, chemical, and magnetic
characterization of a Pennsylvania coal from the Upper Freeport seam. The powdered
coal was then subjected to high-gradient magnetic separations, as a function of mag-
netic field and fluid velocity, in both a water slurry and an air dispersion. Ash and
pyritic sulfur reductions occurred with increasing magnetic field intensities and
decreasing fluid velocities. The best results were obtained in water slurries where
approximately 50 percent of the total sulfur and 50 percent of the ash were removed.
Air dispersions produced insignificant results.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEM ENDED TERMS
c. COSATi Held/Group
Pollution
Coal
Desulfurization
Magnetic Separators
Separation
Magnetic Properties
Pollution Control
Stationary Sources
High-gradient Magnetic
Separation
13B
08G/21D
07A/07D
131
20C
13. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (Tilts Report I
Unclassified
21. NO. OF PAGES
90
20. SECURITY CLASS (Thispage/
Unclassified
22. PRICE
$6.00
EPA Form 2220-1 (9-73)
------- |