United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S7-83-057 Feb. 1984
&ERA Project Summary
Evaluation of Low Gravity Dense
Media Cyclone Performance in
Cleaning Fine Coal
John L Zalar
Thirty-six pilot plant tests were
conducted to evaluate the performance
of dense medium cyclones in cleaning
fine coal (9x100 mesh)* at a low
specific gravity of separation (1.3).
(NOTE: These cyclones, consisting of
finely ground magnetite and water, are
commonly used in the coal industry to
separate coal and mineral particles.)
Test variables included two orifice
sizes, three medium-to-coal weight
ratios, and six flow rates. The clean coal
(overflow) and refuse (underflow) from
these tests were separated into three
size fractions (9x14, 14x28, and
28x100 mesh) and float/sink tested at
eight specific gravities. This report
gives results of statistical analyses
which were performed to evaluate
these tests. Performance criteria used
in these evaluations included: four
dependent criteria (recovery efficiency,
misplaced material, yield error, and ash
error); three independent criteria (prob-
able error, error area, and imperfection);
and three sulfur-based criteria (percent
sulfur in coal, percent sulfur reduction,
and percent reduction in pounds per
million Btu). Final results of the statis-
tical analysis were regression equations
relating each performance criterion to
particle size, orifice size, flow rate, and
medium-to-coal ratio. Particle size was
statistically related to cyclone perform-
ance for all criteria examined, except
weight recovery, efficiency, and percent
reduction in pounds per million Btu.
Generally, the best performance oc-
curred at the 9x14 mesh size range and
"Readers more familiar with metric units may use the
conversion figures at the back of this Summary-
deteriorated as particle size decreased.
Cyclone performance improved with
increasing flow rate up to 120 gpm,
then decreased beyond this value.
Cyclone performance, as measured by
the dependent criteria, improved slightly
when a 1.5 in. orifice was used. Percent
coal in slurry was a statistically signifi-
cant predictor of all measures of
performance criteria, except weight
recovery efficiency. In general, best
performance occurred at a nominal
medium-to-coal ratio of 7:1 and dete-
riorated as the percentage of coal in the
slurry increased. The cyclone operating
parameters were, in general, more
highly correlated with the independent
measures of cyclone performance than
with dependent or sulfur-based criteria.
This Project Summary was developed
by EPA's Industrial Environmental
Research Laboratory, Research Triangle
Park, NC, to announce key findings of
the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
The Multi-stream Coal Cleaning Strategy
(MCCS) is being tested by the Pennsylvania
Electric Company (PENELEC) and the
New York State Electric & Gas Corporation
(NYSEG) to control emissions of sulfur
oxides from coal-fired boilers at the
Homer City, PA, Generating Station. The
MCCS is based on physical coal cleaning
using dense medium cyclones in which
mineral impurities, including pyritic
-------
sulfur, are removed from raw coal. Dense
medium cyclones are used because they
are very efficient in separating organic
coal particles from the more dense
mineral particles. The apparent density of
the medium is controlled by mixing finely
ground magnetite with water.
In a dense medium cyclone, centrifugal
force is used to separate particles of
different specific gravities. The specific
gravity (s.g.) of organic coal particles may
range from 1.3 to 1.9, while that of coal
mineral particles may range from 2.2 to
5.0. To remove most of the mineral
contaminants from coal it is necessary to
crush the raw coal to fine particle sizes
and separate these particles at a low
specific gravity. Prior to this project there
was little data on the performance of
dense medium cyclones operating on fine
size coal particles at low specific gravity.
To provide additional data on dense
medium cyclone performance a coopera-
tive pilot plant test program was conducted
by the Department of Energy (DOE), the
U.S. Environmental Protection Agency
(EPA), and PENELEC. Thirty-six tests
were conducted at DOE's dense medium
pilot plant at the Pittsburgh Energy
Technology Center (PETC). These tests
are documented in DOE/PECTC/TR-
83/1 (DE83009250), Dense-Medium
Cycloning of Fine Coal at Low Specific
Gravities.
All of the dense medium tests were run
with an 8-in. diameter dense medium
cyclone operated at a nominal 1.3 s.g.
with 9x100 mesh coal. The operating
variables included two orifice sizes (1-
1 /2 and 1 -3/4 in.); three medium-to-coal
ratios (7:1, 5:1, and 3:1 (and six flow rates
(92, 100, 110, 1 20, 140, and 160 gpm).
The overflow (clean coal) and underflow
(refuse) from each test were divided into
the three size fractions (9x14,14x28, and
28x100 mesh) and float/sink tested at
eight specific gravities. Each size and
specific gravity increment was analyzed
for moisture, ash, and total sulfur.
Bituminous Coal Research Inc. (BCR)
was awarded a contract to perform
statistical evaluations of the 36 dense
medium pilot plant tests. Performance
criteria used in these evaluations included:
four dependent criteria (recovery effi-
ciency, misplaced material, yield error,
and ash error); three independent criteria
(probable error, error area, and imperfec-
tion); and three sulfur-based criteria
(percent sulfur in coal, percent sulfur
reduction, and normalized percent sulfur
reduction in pounds per million Btu).
This report documents the results of
these statistical analyses.
Conclusions, Limitations, and
Recommendations
The mathematical and statistical
evaluations of dense medium cyclones
conducted during this project have
yielded several important conclusions
regarding the relationships between
cyclone operating parameters and the
various performance criteria examined.
These conclusions are presented and
discussed in this section. In order that
they may serve as operating guidelines,
all conclusions are stated in terms of the
relationship of the operating parameters
to measured cyclone performance. Limit-
ations of the stated conclusions are
discussed, and recommendations are
made for further work in evaluating
cyclone performance.
Conclusions
Dense medium cyclone performance
improves as particle size of the feed coal
increases. Particle size was statistically
related to cyclone performance for all
criteria examined, except weight recovery
efficiency and normalized percent sulfur
reduction. Generally, best performance
occurred for the 9x14 mesh size range
and deteriorated as particle size decreased.
Only for percent sulfur reduction was this
tendency reversed; i.e., slightly higher
percent reductions occurred as particle
size decreased. When percent sulfur
reduction was normalized for Btu value,
this particle size effect was not observed.
Cyclone performance, as measured by
the dependent criteria and percent sulfur
reduction, was statistically related to flow
rate. Flow rate is a statistically significant
predictor of cyclone performance as
measured by the four dependent criteria
and percent sulfur reduction. In all cases,
this relationship is best characterized
using a second-order polynomial expres-
sion for flow rate, because cyclone
performance improved with increasing
flow rate up to a value of 120 gpm, then
deteriorated as the flow rate increased
beyond this value. Based on the flow-rate
levels examined in the test matrix and the
statistical models derived in this investi-
gation, 120 gpm appears to produce
overall best results for the criteria
mentioned above. It is possible, however,
that the true optimal flow rate is slightly
higher or lower than 120 gpm.
Cyclone performance, as measured by
the dependent criteria, was related to the
size of the inlet orifice. In all cases,
performance improved slightly when a
1.5-in. orifice was used.
Dense medium cyclone performance
improved as the percent coal in slurry
decreased. Percent coal in slurry was a
statistically significant predictor of all
measures of cyclone performance criteria
examined, except weight recovery effi-
ciency. In general, best performance
occurred at a nominal medium-to-coal
ratio of 7:1 and deteriorated as the
percentage of coal in slurry increased.
This tendency was most pronounced for
independent measures of cyclone per-
formance.
The cyclone operating parameters
investigated are, in general, more highly
correlated with independent measures of
cyclone performance than with dependent
or sulfur-based criteria. A comparison of
the results obtained for the various
performance criteria reveals that most of
the variability (over 80 percent) in the
independent measures can be accounted
for through a knowledge of the operating
parameters being considered. In contrast,
among the dependent and sulfur-based
measures, variability explained by the
operating parameters generally ranges
from about 25 to 70 percent. For sulfur in
clean coal, less than 10 percent of the
variability is explained; however, this is a
poor measure of sulfur performance.
These results are not surprising when
one considers that error area, probable
error, and imperfection are essentially
independent of the washability charac-
teristics of raw coal. As such, these
criteria are more characteristic of the
performance of the coal-cleaning unit
itself. However, some important and
consistent relationships between operat-
ing parameters and dependent criteria
have emerged from this evaluation. The
observed correlations between cyclone
operating parameters and the performance
measures do indicate particular operating
conditions under which cyclone perform-
ance is likely to be improved.
For the test matrix under investigation,
cyclone operating conditions correspond-
ing to Test No. 84 appear to produce best
overall cyclone performance. The results
of all statistical analyses performed in
this investigation suggest that one area of
the test matrix is optimal or very near
optimal for all cyclone performance
criteria examined. The operating condi-
tions of Test No. 84 (i.e., 1.5-in. orifice
size, 120 gpm flow rate, and 7:1 medium-
to-coal ratio) seem to produce improved
cyclone performance as compared to
oiher cyclone operating configurations.
While true optimal performance may not
occur exactly at the conditions of Test No.
84, generally this is the best set of
operating conditions in the test matrix
under consideration.
-------
Limitations
The results and conclusions of this
evaluation are strictly applicable to 8-in.
cyclones. Generalizations of these results
to larger or smaller cyclones should be
made with caution. Values of specific
operating parameters used in this investi-
gation, such as inlet-orifice sizes andflow
rates, may not be appropriate for larger or
smaller cyclones.
Extrapolations and interpolations of the
relationships and prediction equations
derived in this evaluation should be made
with caution. The results of this analysis
do not necessarily apply beyond the
ranges of values for the operating
variables that were used in this investiga-
tion. Further, interpolation of these
results is limited by the number of levels
of each operating parameter actually
incorporated in the test matrix, particularly
inlet-orifice size (which had only two
possible values).
The results and conclusions of this
evaluation, pertaining to dependent and
sulfur-based criteria, are not necessarily
generalized to other coals with different
washability characteristics. The indepen-
dent performance criteria, however,
should apply to other coals.
The actual performance curves and
performance criteria derived from these
curves may be somewhat different from
those derived using other curve fitting
procedures; e.g., hand-drawn or other
curve fitting routines.
Recommendations
Verification of the relationships and
prediction equations derived from this
analysis, carried out through additional
cyclone testing, would be valuable. These
tests could include other levels of the
operating parameters, such as additional
flow rates and inlet-orifice sizes. Exami-
nation of additional cyclone operating
parameters, through matrix testing,
would also be valuable. Emphasis could
be on improving correlations for the
dependent performance criteria; in
particular, the sulfur-based measurers.
Extending results of the present
investigation to larger cyclones, as well
as verifying results for other coals, would
also be of value.
Test Conditions
Thirty-six tests, using an 8-in. dense
medium cyclone, were conducted at DOE's
pilot plant in Bruceton, PA. The tests
provided performance data for the
cyclone operating under various condi-
tions on coal particle mesh sizes of 9x14,
14x28, 28x100, and 9x100. The operating
variables investigated included two inlet
orifice sizes (1-1/2 and 1-3/4 in.); three
medium-to-coal ratios (7:1, 5:1, and 3:1);
and six flow rates (92, 100, 110, 120,
140, and 160 gpm). Operating conditions
for each dense medium cyclone test are
given in Table 1.
Test Equipment and
Procedures
The pilot plant circuit contained an 8-
in. diameter Heyl & Patterson cyclone
with a 14° included angle, a full-stream
slurry sampler, a 3-x2-in. centrifugal
pump with a variable speed drive, a
nuclear density gauge, a magnetic
flowmeter, and a full-stream pressure
sensor and gauge. A schematic of the
dense medium pilot plant is shown in
Figure 1.
The circuit was a closed-loop batch
system that recirculated the slurry. Each
test began with 60 gal. of water mixed
with enough magnetite to produce the
desired medium density, as indicated by
the density gauge. The flow rate was set
using a flowmeter and variable speed
pump. The inlet pressure was read from
the pressure sensor near the inlet. Once
steady state was obtained, full-stream
samples of the recirculating medium
were taken to measure the medium
distribution between the overflow and
the underflow.
At this point, a measured weight of
9x10 mesh coal (100-200 Ib), based on
the recirculating medium density, was
added to obtain the desired medium-to-
coal ratio. The coal slurry was recirculated
for only about 2 minutes, in order to
minimize degradation while still allowing
time for good mixing. Between 8 and 16
consecutive full-stream cuts through the
product streams were then taken, de-
pending on the flow rate and concentra-
tion. The sampler was designed to take
identical simultaneous cuts of the clean
coal (overflow) and refuse (underflow)
Table 1. Dense Medium Cyclone Operating Conditions
Test
No.
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
Inlet
Orifice
in.
1.75
1.75
1.50
1.50
1.50
1.75
1.75
1.50
1.50
1.75
1.75
1.75
1.50
1.50
1.50
1.75
1.75
1.75
1.50
1.50
1.75
1.75
1.50
1.50
1.75
1.75
1.50
1.50
1.50
1.75
1.75
1.75
1.75
1.50
1.50
1.50
Inlet
Pressure
psi
7.0
5.0
10.0
5.5
4.5
5.0
9.0
6.5
7.5
4.0
5.5
6.5
10.0
8.0
6.0
5.0
8.5
5.0
6.5
5.0
6.0
4.5
7.0
5.0
6.5
6.5
6.5
6.0
10.0
8.5
11.0
11.0
11.0
12.0
12.0
12.0
Flow
Rate
gpm
120
100
140
100
92
100
140
110
120
92
110
120
140
120
100
100
140
92
110
92
110
92
120
92
120
110
110
100
140
140
160
160
160
160
160
160
Feed
Coal
tph
4.2
7.5
4.7
7.3
3.6
3.8
10.8
5.0
8.9
3.7
8.4
6.1
7.5
4.5
3.9
5.2
5.2
6.6
8.4
4.9
4.1
4.2
4.6
7.0
9.6
4.8
4.2
4.8
11.1
6.6
6.5
7.3
12.5
6.1
8.1
12.9
Medium-
to-coal
Ratio
7:1
3:1
7:1
3:1
7:1
7:1
3:1
5:1
3:1
7:1
3:1
5:1
5:1
7:1
7:1
5:1
7:1
3:1
3:1
5:1
7:1
5:1
5:1
3:1
3:1
5:1
7:1
5:1
3:1
5:1
7:1
5:1
3:1
7:1
5:1
3:1
Coal in
Slurry, wt%
11.2
23.3
10.9
22.7
12.3
12.0
24.1
14.6
23.3
12.8
24.0
16.3
17.3
12.1
12.5
16.5
12.2
22.7
24.4
17.2
12.1
14.6
15.2
24.0
25.5
14.2
12.5
15.5
25.2
15.3
13.3
15.0
24.8
12.3
16.5
25.6
-------
streams so that the yield could be based
on the weights of the samples. About 10
Ib of each product (clean coal and refuse
solids) was necessary for subsequent
washability analysis.
The weight and volume of slurry
samples were measured, and the samples
were wet-screened to remove magnetite.
The products were dried and then checked
with a hand-held magnet to remove any
magnetite before the final weighing.
Each whole sample was screened into
9x14, 14x28, 28x100, and 100 mesh x 0
size fractions (Tyler screens). The three
coarsest fractions were float/sink tested
in organic heavy liquid? at specific
gravities of 1.27, 1.29, 1.32, 1.35, 1.40,
1.50, 1.60, and 1.80. Each float/sink
increment was analyzed for moisture,
ash, and total sulfur.
Performance Criteria
The cyclone performance criteria used
in this statistical evaluation were depen-
dent, independent, and sulfur based.
Dependent Criteria
Equipment performance criteria that
depend on the properties of the coal bei ng
cleaned7 are called dependent criteria.
Four dependent criteria were used.
Recovery efficiency - the ratio, ex-
pressed in percentage, of the yield of
cleaned coal to the yield of float coal of the
same ash content shown, by specific
gravity analysis, to be present in the feed.
Misplaced material - the sum of the sink
material in the cleaned coal and the float
material in the refuse, expressed as a
percentage of the raw coal.
Yield error - the difference between the
yield of coal actually obtained and the
theoretical yield at the ash content of the
cleaned coal.
Ash error - the numerical difference
between the actual and theoretical ash
content of cleaned coal at the yield of
cleaned coal obtained.
Independent Criteria
Some equipment performance criteria
are substantially independent of the coal
properties. Independent performance
criteria used in this study are determined
from the separation distribution curve.
The distribution curve is a plot of the
percentage of each specific-gravity
fraction of the coal feed recovered in the
clean-coal product versus the median of
the specific-gravity fraction. The specific
gravity of separation is defined as the
specific gravity of the material in the raw
feed that is divided equally between clean
coal and refuse. Three independent
criteria were used.
Automatic
Full Stream
Sampler
Overflow
Sample
100 gal. Slurry
Sump
8-in. Dense-Medium
Cyclone
Pressure
Sensor and
Gauge
Magnetic
Flowmeter
Nuclear
Density Gauge
Variable Speed
Pump
Figure 1. DOE dense-medium cyclone pilot plant.
Probable error - the slope of the
distribution curve equal to half the
specific gravity difference between the
25- and 75-percent ordinate values on
the curve.
Error area - the area between the
actual distribution curve and the theoreti-
cally perfect distribution curve. The
theoretically perfect curve is a step
function at the specific gravity of separa-
tion.
Imperfection - probable error divided by
the specific gravity of separation.
Sulfur-Based Criteria
The performance of coal cleaning
equipment in removing sulfur has not
been traditionally evaluated. Sulfur
based criteria are all dependent on coal
properties. Three sulfur-based perform-
ance criteria were used.
Percent sulfur in clean coal - total
sulfur content of clean coal determined
from laboratory analysis.
Percent sulfur reduction - the difference
between percent sulfur in feed and
percent sulfur in clean coal, divided by the
percent sulfur in feed.
Normalized percent sulfur reduction -
computed using sulfur values for clean
and feed coal expressed as pounds per
million Btu. The Btu value of each specific
gravity fraction was estimated from the
ash content since the Btu value was not
measured.
Analysis Methodology
Dense medium cyclone performance
data for the 36 tests conducted at DOE's
pilot plant in Bruceton, PA, were supplied
to BCR by PENELEC. Computer software
for data entry and verification was
developed. Data files, created on a test-
-------
by-test basis, included all information
provided by PENELEC (i.e., operating
parameters and cyclone performance
measures). Additional software was
developed to generate long- and short-
form data summaries for each test, as
well as summary statistics (means and
standard deviations) for any desired
combination of tests.
Four distribution curves were generated
for each cyclone test: one for each of the
three particle-size ranges of coal and one
for the composite size range. The
technique used to fit curves to the
distribution data is a mathematical
procedure known as the cubic spline
method. This curve fitting was accom-
plished using a computer program
developed by BCR. One of the 108
computer generated distribution curves
developed during the study is shown in
Figure 2. The resulting smoothed distribu-
tion curves were evaluated by a second
BCR computer program in order to
compute values for the specific gravity of
separation, probable error, error area,
and imperfection. One of the 36 test data
summary sheets developed during the
study is given in Table 2.
Statistical analysis of the test data was
performed to determine the effects of
operating parameters on cyclone per-
formance. Specifically, the objective of
this analysis was to quantify the relation-
ships between the operating variables
under investigation (particle-size range,
inlet-orifice size, flow rate, and medium-
to-coal ratio) and the various cyclone
performance criteria. The nature of these
relationships suggests the most efficient
set of operating conditions for each
criterion. The statistical procedure used,
known as multiple regression analysis,
involved a general purpose computer
program developed by BCR to perform
multiple regression. Various functional
forms of the operating parameters were
examined in the course of the statistical
analysis, as well as combination (or
interactive) effects of these variables on
cyclone performance.
All multiple function statistical analyses
were conducted on a criterion-by-
criterion basis. Stepwise regression
models for predicting each criterion from
the relevant cyclone operation conditions
were developed. Relevant operating
conditions, statistically related (at the 95-
percent confidence level) to the given
performance criterion, explained (accounted
for) a significant portion of the observed
variability in the criterion. The amount of
explained variability was measured by
the coefficient of multiple determination
(MCC2). Specifically, the value MCC2 (the
Specific Gravity
of Separation = 1.297
Probable Error = 0.022
Imperfection = 0.076
Error Area = 25.
1.0 1.2 1.4 1.6 1.8
-10. Specific Gravity
Figure 2. Distribution curve, test No. 83 (9 x 100).
2.0
Table 2. Test 84 Data Summary
9x14
14x28
28x100
Screen Analysis. Percent:
Feed
Clean Coal
Refuse
Ash. Percent:
Feed
Clean Coal
Refuse
Total Sulfur, Percent:
Feed
Clean Coal
Refuse
Weight Recovery (Yield)
Theoretical Weight Recovery
Weight Recovery Efficiency .
Ash Error
Float in Refuse ...% of Product
Sink in Clean Coal
Total Misplaced Material..% of Feed
Near Gravity Material
Specific Gravity of Separation
Probable Error
Imperfection
Error Area
Distribution, Percent to Cleaned
Coal (Specific Gravity Fraction):
28.9
19.1
35.9
24.5
2.8
32.6
3.61
1.11
4.56
27.4
23.3
100.0
0.0
9.2
14.0
10.5
59.7
1.292
0.018
0.063
11.
29.8
30.5
29.3
24.5
3.4
40.2
3.73
1.16
5.63
42.6
44.0
96.8
0.1
16.1
9.4
13.2
65.7
1.306
0.021
0.068
13.
35.6
43.3
30.2
23.6
3.7
43.9
5.08
0.90
9.33
50.4
57.8
87.3
0.7
22.7
5.9
14.2
65.4
1.351
0.071
0.203
40.
9x100
94.4
92.9
95.4
24.2
3.4
38.5
4.20
1.03
6.40
40.9
43.2
94.7
0.2
15.9
13.9
15.1
64.0
1.308
0.027
o.oaa
23.
Float 1.27
1.27-1.29
1.29-1.32
1.32-1.35
1.35-1.40
1.40-1.50
1.50-1.60
1.60-1.80
Sink 1.80
85.7
70.0
30.2
4.6
0.6
0.0
0.0
0.0
0.0
87.6
80.5
53.2
14.9
0.4
0.0
0.0
0.0
0.0
85.2
86.8
77.3
58.7
37.2
23.4
12.1
0.0
0.0
86.1
80.0
53.9
24.6
13.6
7.2
3.6
0.0
0.0
-------
square of the correlation coefficient)
represents the fraction of total variability
in the given performance criterion that is
accounted for by the regression model.
The final prediction equation (regression
model) was developed for each criterion,
and comparisons were made between
the predicted and observed values for
each model. These values were used to
indicate the overall fit of the regression
model to the data as well as to select the
set of cyclone operation conditions that
would most likely produce the best
results for the given performance criterion.
Table 3 gives an example of thestepwise
development of a regression model.
Results
General Statistical Properties
The mean values, standard deviations,
and relative standard deviations for each
performance criterion are given in Table
4. These statistical parameters are
computed from test values in three size
ranges for 36 tests (N=108). Data are also
presented for the simple correlations
between each criterion and each cyclone
operating parameter.
The relative standard deviation (stand-
ard deviation + MEAN) provides a relative
measure of the variation in each perform-
ance criterion during the 36 tests. The
weight recovery efficiency, percent sulfur
in clean coal, percent sulfur reduction,
and normalized percent sulfur reduction
are not very sensitive to changes in the
operating variables. The misplaced
material and yield error show moderate
sensitivity. The ash error, error area,
probable error, and imperfection are the
most sensitive.
The simple correlations provide an
estimate of the strength and direction of
the relationship between a performance
Table 3. Results of Regression Analysis for Yield Error
Step
1
2
3
4
Independent
Variable
P
F.F2
0
R
MCC
0.496
0.639
0.706
0.722
Final Prediction Equation
MCC2
0.246
O.408
0.498
0.522
F-ratio
34.51*
28.49*
18.58*
5.01*
Y,= -1.09 F, + 0.004 F? - 5.13 P,+ 12.56 0, + 0.16 /?, +58.27
MCC =0.722
MCC2 = 0.522
SEE =3.58
0 - Orifice size
F- Flow rate
R - Percent coal in slurry
P - Particle size
MCC - Multiple correlation coefficient
MCC2 - Coefficient of multiple determination
F-ratio - Statistic for test of significance of regression model
* - Significant at 95-percent confidence level
Y - Predicted value for criterion variable
SEE - Standard error of estimate
criterion and single operating variable.
The only operating variable which shows
a strong correlation with the performance
criteria is particle size. The large negative
correlations for misplaced material, ash
error, yield error, error area, probable
error, and imperfection show that the
value of these performance criteria
increase significantly with decreasing
particle size.
Statistical Evaluation of
Dependent Performance
Criteria
A preliminary examination of the
correlations between the cyclone operat-
ing parameters and the dependent
performance criteria indicated that all
four operating parameters should be
tested for inclusion in the final prediction
models for the dependent criteria. First-
order expressions for orifice size (0),
percent coal in slurry (R), and particle size
(P) were selected as potential predictor
variables. A second-order polynomial
expression was used in the case of flow
rate (F,F2).
The final models which were developed
include those independent variables that
are statistically significant at the 95-
percent confidence level. The final
models of three of the dependent per-
formance criteria (misplaced material,
ash error, and yield error) include all four
independent variables (0, R, P, and F,F2).
Only 0 and F,F2 are significant for weight
recovery efficiency. The coefficients of
multiple determination (MCC2) for the
final models range from 0.274 for weight
recovery efficiency to 0.658 for ash error.
Thus, these models are able to account
for about 27.4 to 65.8 percent of the
variability in the various dependent
performance criteria. The final statistical
models for the dependent, independent.
Table 4. Means, Standard Deviations, and Correlations for Cyclone Performance Criteria (N = 108)
Relative
Correlations
Criterion
Weight Recovery
Efficiency
Misplaced Material
Ash Error
Yield Error
Error Area
Probable Error
Imperfection
Percent Sulfur
in Clean Coal
Percent Sulfur
Reduction
Normalized Percent
Sulfur Reduction
Mean
77.96
14.62
1.14
11.44
32.71
0.042
0.129
1.00
68.50
75.65
Standard
Deviation
8.22
3.75
0.73
5.18
22.81
0.03
0.08
0.15
9.30
7.90
Standard
Deviation
0.11
0.26
0.64
0.45
0.70
0.71
0.62
0.15
0.14
0.10
Orifice
Size
-0.342
0.201
0.222
0.301
0.091
0.053
0.055
-0.002
0.059
0.065
Flow
Rate
-0.108
-0.173
0.103
0.102
-0.036
-0.043
-0.049
0.019
-0.504
-0.515
Percent Coal
in Slurry
-0.132
0.244
0.240
0.161
0.176
0.233
0.203
0.196
-0.215
-0.223
Particle
Size
-0.008
-0.615
-0.649
-0.496
-0.826
-0.799
-0.803
-0.200
-0.249
-0. 120
-------
and sulfur based criteria are summarized
in Table 5.
Statistical Evaluation of
Independent Performance
Criteria
Distribution curves were generated
using the cubic spline method on a test-
by-test basis. Four curves are drawn for
each test: one for each particle-size range
and one for the composite. Values for the
three independent criteria, along with the
specific gravity of separation, are reported
for each curve. These values were read or
computed from the distribution curve
shown.
A preliminary examination of the corre-
lation between the cyclone operating
parameters and the independent per-
formance criteria indicated two statisti-
cally significant parameters to be tested
for inclusion in the final prediction
models for the independent criteria. A
second-order polynomial expression for
particle size (P,P2) and a first-order
expression for percent coal in slurry (R)
were selected as potential predictor
variables.
Final models for all three independent
performance criteria (error area, probable
error, and imperfection) include both P,P2
and R (see Table 5). The coefficients of
multiple determination (MCC2) for the
final models range from 0.828 for
imperfection to 0.881 for error area.
Statistical Evaluation of
Sulfur-Based Performance
Criteria
As for the dependent performance
criteria, a preliminary examination of the
correlations between the cyclone operat-
ing parameters and the sulfur-based
performance criteria indicated that all
four operating parameters should be
tested for inclusion in the final modelsfor
sulfur-based criteria. First-order expres-
sions for orifice size (0), percent coal in
TableS. Predicted Regression Equations for Various Performance Criteria*
slurry (R), and particle size (P) were,
again, selected as potential predictor
variables. A second-order polynomial
expression was used for flow rate (F,F2).
The final model for sulfur in clean coal
includes two significant terms, namely, P
and R (see Table 5). For percent sulfur
reduction, P and R are significant in the
final model. For normalized percent
sulfur reduction, F,F2 and R are significant.
The coefficient of multiple determination
(MCC2) is only 0.078 for percent sulfur in
clean coal, as compared to 0.535 and
0.491 for percent sulfur reduction and
normalized percent sulfur reduction,
respectively. This large difference is due
to variations in sulfur content in the feed
coal. Percent sulfur in clean coal includes
the variability in feed coal sulfur content,
making it a less desirable measure of
cyclone performance than either percent
sulfur reduction or normalized percent
sulfur reduction, both of which adjust for
differences in the sulfur content of the
feed. Normalizing percent sulfur reduction
for Btu value had little effect on the
magnitude of the overall correlation;
however, it did remove the particle size
effect (P) from the final model.
Metric Equivalents
EPA policy includes use of metric units
in all its documents. Although this
summary uses nonmetric units for
convenience, readers more familiar with
the metric system are asked to use the
equivalents below:
Nonmetric
Btu
gpm
in.
Ib
lb/106 Btu
psi
tph
Multiplied by
1055.06
3.785
2.54
0.454
429.91
70.307
0.907
Yields
Metric
J
l/min
cm
kg
ng/J
g/cm2
tonne/hr
Tyler Screen Size Mesh Openings
Sieve Opening
Mesh Size in. mm
9
14
28
100
200
0.0787
0.0469
0.0234
0.0059
0.0029
2.00
1.18
0.60
0.15
0.075
Performance Criterion
Predicted Regression Equations
MCC
MCC*
SEE
Weight Recovery Efficiency
Misplaced Material
Ash Error
Yield Error
Error Area
Probable Error
Imperfection
Percent Sulfur in Coal
Percent Sulfur Reduction
Normalized Percent Sulfur
Reduction
, = 1.71 Fr0.007F?-22.370,+ 12.74
K =-0.53 F, + 0.002 F?-4.61 P, + 0.18 R> + 6.180t
K, =-0. 14 f, + 0.0006 F?-0.95P, + 0.03 Ri+1.320t +
£ = -r. 09 F, +0.004 f,2-5. 13 P,+ 12.560, + 58.27
fc, = -0.0001 P, + 54.73 P? + 0.78 R, +95.70
£ = -0.188 P, + 0.070 P? + 0.0014 R,+ -0.1 17
£, = -0.488 P, + 0.1 79 P? + 0.0032 R,+ 0.331
£, = -0.06 P, + 0.006 R, + 0.95
% = 2.02 F.-0.009 F?-4.62 P,-0.32 R,-32.59
V, = 7.73 f, -0.0075 F?-0.287 R.-14. 44
7.52
0.624
0.757
0.811
0.722
0.938
0.921
0.910
0.280
0.731
0.701
0.274
O.673
0.658
0.522
0.881
0.849
0.828
0.078
0.535
0.491
7.01
2.45
0.43
3.58
7.88
0.012
0.034
0.15
6.34
5.63
Terms are defined in Table 3.
-------
John L. Zalar is with Bituminous Coal Research, Inc.. P. 0. Box 278. Monroeville.
PA 15146.
James D. Kilgroe is the EPA Project Officer (see below).
The complete report, entitled "Evaluation of Low Gravity Dense Media Cyclone
Performance in Cleaning Fine Coal," (Order No. PB 84-122 936; Cost: $22.00,
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
PS 0000329
U.S. GOVERNMENT PRINTING OFFICE: 1984-759-102/865
------- |