University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Virginia Tech | Table 3.14 - Concentrate water/wash water ratio by cell for series 4 tests.
Concentrate gallons/Wash water gallons (dilution washes)
With Paddles No Paddles Low Level Very Low No Wash
Cell (B) (C) (D) (E) (A)
1 0.44 0.37 0.75 0.73 0
2 0.69 0.67 0.88 0.97 0
3 0.52 0.44 0.29 0.32 0
4 1.45 0.84 0.47 0.53 0
5 2.77 1.64 0.98 1.20 0
Weighted 0.51 0.50 0.61 0.67
The last line of Table 3.14 shows the weighted average of the wash water bias.
This wash water bias was weighted using the tons of concentrate produced by each cell.
As you can see the bias never fully approached 1 for the entire bank, although it exceeded
it for individual cells. The first two tests (4-B, 4-C) comparing with paddles and without
paddles show almost the same bias 0.51 vs. 0.5. This is understandable because no shift
in recovery was made by removing the paddles. However, for tests 4-D and 4-E (low
level and very low level respectively), the difference was made by reducing the pulp level
in each cell.
When compared at an equivalent combustible recovery of 90% for the tests on
Eagle Coal (series 4 tests), the addition of wash water reduced the ash content of clean
coal product from 9.6% to 8.2% (1.4 percentage points) for the low and very low pulp
levels. This reduction allowed this particular flotation circuit to reject an additional 1.3
tph of ash. A lower reduction of only 0.8 percentage points (9.6% to 8.8%) was observed
when the high pulp level was used.
The data shown in Figure 3.17 provides a partial explanation for the limited
effectiveness of froth washing for the conventional cells. In this figure, the mass rate of
56 |
Virginia Tech | 32
28
24
20
16
12
8
Cell #1 Cell #2 Cell #3 Cell #4 Cell #5 Bank
Flotation Cell
59
)%(
tnetnoC
sdiloS
No Wash
Wash (High Level)
Wash (Low Level)
Wash (Very Low Level)
Figure 3.18 - Clean coal solids content obtained for each cell under different operating
conditions for Eagle coal test (series 4).
In order to be effective, sufficient wash water must be added to completely
replace the pulp reporting to the froth launder. The number of dilution washes (defined
as the ratio of the volumetric flow of wash water to concentrate pulp) must be greater
than unity. Table 3.14 (shown earlier) provides these values for the existing circuit. As
shown, the dilution washes for the overall bank ranged from 0.51-0.67. In particular,
cells #1, #3 and #4 were operated with too little wash water. A back-calculation shows
that the wash water flow rate for the overall circuit would need to be increased from 1050
gpm to approximately 1800 gpm to achieve one dilution wash. Although this number
could be increased depending on how much carry over of wash water the froth has. At
some point the froth would no longer be able to hold any more water and any additional
wash water added would not dilute the froth. |
Virginia Tech | Figure 3.19 shows the effect of having the wash water ratio of water input to
water recovery approach 1. This plot took the wash water ratio and ash of the primary
cells for the Eagle coal series of tests (series 4). Notice that if the trend line is continued
to the ratio of 1, shown here by the square, the theoretical limit for this wash water
system would produce a primary ash of 5.4% significantly lower than 6.6% but still not
as good as the release analysis.
1.2
1
0.8
0.6
y = -0.1629x + 1.877
0.4 R2 = 0.9269
0.2
0
5.00 5.50 6.00 6.50 7.00 7.50 8.00 8.50 9.00
Ash
60
oitaR
retaW
hsaW
Figure 3.19 - Primary cell product ash for series 4 tests compared to wash water addition
ratio.
Using the 1800 gpm figure, the unit wash water flow rate would need to be
increased to more than 4.5 gpm/ft2 for the first three cells and to 2.0 gpm/ft2 for the last
two cells. Typically, column cells are operated with less than 3.5 gpm/ft2. However,
column cells are operated with a much deeper froth (3-4 ft). The deeper froth allows for
additional drainage and reduces the water demand. In most cases, conventional cells
cannot be operated with such a deep froth since this would substantially reduce the pulp |
Virginia Tech | volume. For a fixed volumetric flow, a smaller pulp volume reduces residence time and
lowers recovery. As was shown in section 3.1.2 the bank originally had a mean residence
time of 8.7 minutes, but was reduced to 6.3 minutes when wash water was added to all of
the cells. If the cells were operated with much deeper froths, the reduction in residence
time made by adding wash water would be more pronounced.
3.3.2 - Residence Time Predictions
In order to evaluate the effects of higher wash water additions on residence time,
mathematical calculations were performed as a function of froth depth and number of
dilution washes for the existing flotation bank. The results of these calculations are
summarized in Figure 3.21, with the plant results in Figure 3.20. As shown, the
maximum residence time that can be maintained with one dilution wash is 8.7 minutes
with ½ ft of froth depth, compared to 3.8 minutes for a froth depth of 4.0 feet. The 3.8
minutes is very close to the minimum residence time that would be required to maintain
acceptable coal recoveries. In addition, the data from the in-plant test work indicates that
water addition rates are very difficult to balance along the length of the bank. Normal
fluctuations in feed tonnage to the flotation bank create shifts in the volumetric demand
for wash water along the bank length. Therefore, higher wash water rates (e.g., 1.2-1.4
dilution washes) would probably be required in actual practice to compensate for these
fluctuations. These higher rates would further shorten the available residence time and
not allow acceptable coal recoveries to be obtained.
More importantly, the data obtained to date suggest that froth washing is not a
practical solution for most of the existing flotation circuits in the coal industry. For
61 |
Virginia Tech | 3.3.3 - Effects of Staged Frother
Staged frother addition was chosen in the hopes of both moving the froth to the
secondary cells and allowing for less frother addition due to more efficient placement. At
first the frother was added in a transfer tank from the first cells to the second. Due to a
lack of mixing in this tank the addition of frother seemed to produce almost no results.
The frother addition was transferred to the air intake of the third cell. This is the
recommended point of introduction according to Wemco. The theory was that the frother
would be added at the point where the most mixing and shearing action of the water
would occur. With the frother being added at a point of high shear for the water air
mixture, the frother should have created a very fine froth. When the frother was added at
first the effect was immediate. The froth became extremely fine and very loaded, and the
froth level dropped. But as time went on the froth went back to normal, and eventually
more frother was needed to maintain performance. Once the system reached equilibrium,
the frother needed to maintain the recovery was much greater than if the frother was not
added in stages. Thus, no frother savings could be achieved.
The placement as well as amount of frother, should have produced and
maintained a much more dramatic effect on the froth cell performance. This raises the
question as to why the savings to frother addition were not realized. One theory at the
plant was that the frother was not actually making its way to the air water mixture, but
was somehow going into the internal workings of the rotor assembly. When the frother
was added to the cell the first time, it was fed through the air intake port. In subsequent
tests the frother was added to a pipe that was supposed to connect to the air intake. The
only explanation for the frother changing the froth initially is that the cell saw a large
64 |
Virginia Tech | dose of it quickly. As of yet no explanation of why the frother was not being as effective
has been shown. Other methods of adding the frother were tried but with no better
results.
When comparing the total amount of frother used to the effectiveness of the wash
water at reducing ash, a loose correlation becomes apparent. Table 3.15 lists all of the
frother dosages as well as intake air and diesel addition for all of the tests conducted. If
you compare the total frother amount used the test series 3 stands out for having very
high frother addition rates. Notice also that the ash reduction was only 0.4 percentage
points, whereas the series 2 tests reduced the overall ash by over 1 percentage point. Test
3-B had wash water on all of the cells while 2-B test only had wash water on the last
three cells. This brings us back to water recovery. Looking at Table 3.16, one can see
that the 3-B test had better wash water distribution (liquid flow/wash water); however the
tph of liquid flowing from the cell was higher. With this higher recovery of water came a
higher recovery of entrained clays. High water recovery is more a function of frother
dosage than wash water effectiveness. With higher frother dosages, more froth is
produced for the same amount of air. Also a more stable froth is produced. Taking a
look at diesel addition, the series 3 tests also had the least amount added, in the low 200’s
(ml) vs. upper 300s (ml) for the series 4 tests. For metallurgical coals, the coal needs
very little diesel to act as a collector. The diesel is often used to aid in reducing water
recovery by matting the froth, i.e. diesel and frother work against each for froth stability.
Series 3 tests not only had the most frother, but also the least amount of diesel, translating
to very high water recovery, reducing the effectiveness of the wash water.
65 |
Virginia Tech | Table 3.15 - Comparison of frother addition for all tests.
Test Date 1-B 1-A 2-B 2-A 3-B 3-A 3-C
100% 100% 100%
Coal Type Powellton Powellton 80E/20P 80E/20P Tunnel Tunnel Tunnel
Eagle Eagle Eagle
Air (P) 0.25 0.25 0.25 0.25 0.25 0.25 0.125
Frother (P) 260 260 170 170 206 206 206
Air (S) 0.5 0.5 0.5 0.5 0.625 0.625 0.625
Frother (S) 0 0 120 120 210 210 170
Total frother 260 260 290 290 416 416 376
Diesel 320 320 320 320 232 232 232
Wash Water Y N Y N Y N Y
Test Date 4-B 4-C 4-D 4-E 4-A
W P N P L L V L L N W
100% 100% 100% 100% 100%
Coal Type
Eagle Eagle Eagle Eagle Eagle
Air (P) 0.25 0.25 0.25 0.25 0.25
Frother (P) 192 192 192 205 200
Air (S) 0.5 0.5 0.5 0.5 0.5
Frother (S) 160 160 160 132 0
Total frother 352 352 352 337 200
Diesel 380 380 380 335 335
Wash Water Y Y Y Y N
Table 3.16 - Water recovery effect on concentrate ash for 2-B and 3-B tests.
Liquid flow rate from Tons of ash/ton of
Liquid flow/ wash water
cell tph concentrate
2-B 3-B 2-B 3-B 2-B 3-B
1 83 168 0.073 0.073 0.00 0.45
2 120 179 0.082 0.088 0.13 0.42
3 52 60 0.063 0.101 0.72 0.63
4 77 42 0.084 0.127 0.49 0.90
5 60 14 0.119 0.165 0.63 2.68
total 392 462 0.081 0.086 0.33 0.57
66 |
Virginia Tech | 3.4 - Release Analysis
3.4.1 - Overview
Several series of laboratory release analysis tests were conducted to evaluate the
“ultimate” cleanability of the flotation feed. A second series of tests was performed on
the combined concentrates for each test. This testing used release analysis methods to
separate the coal from the recovered entrained particles. In a similar manner release
analysis techniques were used to confirm the quality and quantity of floatable coal left in
the tailings stream.
3.4.2 - Experimental
Release analysis is a commonly used technique for separating coal into fractions
for froth flotation feed. This technique is similar to washability testing used for coarse
coal fractions. The procedure used has been described in detail elsewhere (Killmeyer et
al., 2000), but a brief outline of the procedure will be made.
If the feed is in slurry form then enough slurry is used to ensure ~ 200 grams of
solids are tested. The slurry is added to a one-liter lab froth cell and the remainder of the
cell is filled with fresh water to increase the level of the cell. Diesel and frother are
added to the cell, and the entire amount of coal is floated out. The first stages of the
procedure can be time consuming the last coal to float is often hard to float and requires a
long residence time to be floated. The froth product is reslurried and diluted with fresh
water to fill the cell and refloated. This procedure is repeated three to five times until a
complete separation of coal and non-coal particles are made. From this point there are
two methods of separating the coal into fractions: the standard or forward method, or the
67 |
Virginia Tech | reverse method. Although the forward method is quicker, this method can entrain less
floatable coals with the more floatable coals, which can shift the results. For all release
analysis performed in this study the reverse method was used.
Once all of the products have been split into their respective fractions, non-
floatable material, slightly floatable, to increasing floatable products, then the products
are filtered and dried. The products are then weighed and ashed. This gives the weight
percentages of the different qualities present. Then the results are plotted as combustible
recovery vs. cumulative ash.
For the product release tests performed, the feed for the test had already been
dried to determine the percent solids in the froth. The dried product was split to 50 grams
and reslurried in enough water to fill a one liter froth cell. The coal was mixed with
diesel in a mixer for several minutes to ensure that all of the particles were re-wetted.
Then the coal was floated in the same manner as the first steps of a normal release
analysis. The amount of diesel and frother used to make this separation were double to
triple what a normal release analysis would take; this was because the product was not
still in a slurried form. Because the release analysis of the feed had already been done,
the product coal did not need to be split into its components; rather the composite was all
that was needed.
For some of the tests a tailings release was also conducted. This was conducted in
the same manner as a normal release analysis would be, only instead of feed, the plant
tailings were used. This test work was used to show what quality of coal was being left
by the circuit and what was the quantity of that coal.
68 |
Virginia Tech | 3.4.3 - Results (Run-of-Mine Feeds)
For this section the release analysis will be discussed in the same order as the
plant testing was conducted. Also as a reference to the work in the plant, the coal
recovery vs. ash was plotted along with the release analysis.
3.4.3.1 - Powellton Coal (1-A, 1-B)
For the first test series using Powellton coal, the release analysis of the feed
shows that at a 90% combustible recovery the total product ash would be 6.0% quite a
difference from 10.48% actually recovered. Several observations can be made about the
two tests. The first is that the two tests varied little in the recovery and ash from each
cell. Looking at the two tests in the plant, Figure 3.23 shows that about 90% of the coal
was recovered in the first two cells. One other unique difference is that in this case the
curve with the wash water is further from the release analysis curve than the curve
without wash water. This is probably from slight differences in the feed between the two
tests. It is unlikely that the wash water played any significant role in this shift because
the wash water was only used on the last three cells for 6% of the total product.
69 |
Virginia Tech | 100
90
80
70
60
50
40
30
20
10
0
0 5 10 15 20
Ash (%)
72
)%(
yrevoceR
elbitsubmoC
Release
With Wash
Without Wash
Figure 3.25 - Comparison of plant performance to release analysis for series 3 tests.
3.4.3.4 – Eagle Coal (4-A, B, C, D, E)
For the Eagle coal tests, the release analysis showed that a 4% ash could be
achieved with a 90% coal recovery. See Figure 3.26 for a graphical summary. Although
the final concentrate ash of the very low pulp level test was 8.3%, much closer than all of
the other tests in approaching release analysis, the product ash was still more than 100%
higher than it should be. In this series of tests the difference in pulp level, in concert with
the wash water, was made evident. Both the low level (4-D) and very low level (4-E)
tests were significantly better than the tests where only wash water was added. This was
the first series of tests where the entire curve shifted from the curve without wash water.
Also note the shift in the first few cells downward in recovery. This is attributed to the |
Virginia Tech | For example column flotation cells that are well maintained are generally capable
of producing products at acceptable recoveries at or close to the release analysis.
3.5 - Parametric Study
Throughout the testing, several questions were raised about how the wash water
should be added. How should the wash water system be used? What are the critical X’s
controlling washing out clay? Here are some of the specific questions that the parametric
study attempted to answer: What does stopping the paddles so that all of the froth must
pass under the paddle do? What is the effect of running the paddles vs. not using paddles
do? How does froth depth affect product ash? And are the wash boxes too high above
the froth etc.
The testing was broken up into three days. The first two days dealt primarily with
the effect of paddles and froth depth, while the last day dealt with wash box height and
addition method as well as froth depth. Statistical methods of analyzing means and
distributions were used to determine the critical X’s. Throughout the testing, only cell #2
was used, and the comparison was between the two sides. This method was chosen
because it reduced the number of samples required, and it increased the clarity of the
effects.
3.5.1 - Parametric Study Results
3.5.1.1 - Paddle Usage at Various Froth Height (5-A)
Purpose was to test the difference in wash water effectiveness for different froth
height levels. Also, the effects of having the paddles stopped versus running were tested.
77 |
Virginia Tech | Six tests were completed with two tests comparing paddles stopped. For all of the tests,
only two froth depths were used. For each of the three test series, a blank indicates a test
condition that has the same settings for each sample. This was used to see if the
differences observed were from the varied conditions or the normal plant variation (Table
3.17).
Table 3.17 – 5-A test summary.
Description Coal Tunnel Eagle
Paddles on and running tph 2000 Diesel 220
Test order /
Cell set TPH
Description point Sample Ash % Solids Solids
no wash low level blank 4 T1R 7.75 28.76 19.2
no wash low level blank sp 14.15 T1H 7.43 28.25 18.6
w/ wash low level 3 T2R 5.98 22.44 9.5
no wash low level sp 14.15 T2H 8.11 26.51 17.6
w/ wash high level 1 T3R 8.09 19.50 11.8
no wash high level sp 12.00 T3H 9.66 22.45 20.1
no wash high level blank 2 T4R 9.48 26.09 15.8
no wash high level blank sp 12.00 T4H 9.84 24.69 15.6
w/ wash low level paddles stopped 5 T5R 6.64 20.75 12.7
no wash low level paddles stopped sp 14.15 T5H 7.44 29.63 12.4
no wash low level paddles stopped 6 T6R 7.51 29.67 13.9
no wash low level paddles stopped sp 14.15 T6H 7.39 29.76 16.1
3.5.1.2 – Paddles vs. No Paddles at Various Froth Heights (5-B)
The purpose of this test series was to find the effect a running paddle had on froth
washing. For this test, a paddle was removed on one side of cell 2, while the on the other
side, the paddle remained. Samples were taken simultaneously on both sides at different
conditions. The tests were repeated at several froth heights to see how froth height might
vary any differences between the two sides (Table 3.18).
78 |
Virginia Tech | Table 3.18 – 5-B test summary.
Description Coal Eagle
paddles cut off right side tph 2000 Diesel 272
Frother (Primary) 224
Test order
and Cell TPH
Description set point Sample Ash % Solids Solids
w/wash low level no paddle on 7 T7R 6.165 23.14 10.48
w/ wash low level paddle running sp 11.6 = 10 in. T7H 7.93 21.99 22.13
no wash low level l lank no paddle on 9 T8R 7.745 28.59 25.21
no wash low level blank paddle flat sp 11.6 = 10 in. T8H 7.965 27.54 28.07
w/ wash low level no paddle on 8 T9R 6.705 23.15 13.90
no wash low level paddle flat sp 11.6 = 10 in. T9H 8.14 27.58 24.30
w/wash high level no paddle on 10 T10R 8.635 19.90 15.20
no wash high level paddle flat sp 13.25 = 8 in T10H 10.19 23.19 30.71
no wash high level blank no paddle on 11 T11R 9.88 24.89 22.45
no wash high level blank paddle flat sp 13.25 = 8 in T11H 10.005 24.58 19.61
w/ wash high level no paddle on 12 T12R 8.345 19.44 14.44
w/ wash high level paddle running sp 13.25 = 8 in T12H 9.285 19.42 21.52
3.5.1.3 – Wash Box Position at Various Froth Heights (5-C)
This test was conducted to check various froth water addition methods (Table
3.19). Should the wash boxes be right over top of the froth, or kept in original positions?
For one test, a wash box was tipped on its side to simulate water weiring over in a sheet
like fashion. This was to see if there was a better way to distribute the wash water. The
thought was that by making a sheet of water all of the bubbles would have to pass
through the clean water. This idea may work, but the way that it was implemented was
not very good. Unfortunately, laminar flow conditions could not be achieved; the flow
was also intermittent and not very well distributed across the whole bed. It ended up not
washing the froth, but more or less diluting it. A second attempt of changing the wash
water addition method was made by one of the operators. The operator added a sheet of
79 |
Virginia Tech | hard plastic at an angle to the wash boxes. The wash water would hit the plastic and slide
off the end of the sheet. This provided much better distribution than the first attempt but
didn’t seem to have any better results. The results seemed to be sporadic at best. The
water flow in the downward direction was not enough for the amount of water being
carried up and over the weir, and so the effect was more of dilution than washing. (See
Figure 3.33)
Table 3.19 – 5-C test summary.
Paddles off Water addition methods
TPH Cell set Froth
Description Sample Ash % Solids Solids point height
Low level wash box T13H 5.44 23.28 6.73 8.10 9.0
High level wash box T13R 5.18 25.59 2.87 8.10 9.0
Low level wash box T14H 9.55 17.13 16.23 10.00 4.5
High level wash box T14R 9.54 14.92 10.71 10.00 4.5
Low level wash box T15H 8.77 17.34 12.50 9.05 6.5
High level wash box T15R 8.83 17.46 10.37 9.05 6.5
High level wash box with weir water T16H 7.33 20.03 14.08 8.75 7.0
High level wash box normal T16R 7.27 20.48 8.13 8.75 7.0
3.5.2 - Discussion
For the first set of tests, the effect of stopped paddles verses paddles running
contributed vary little to the overall effectiveness of the system. Froth depth had a much
greater effect on the whole system. Comparing the populations of stopped paddles to
running paddles, tests 1,2,5, and 6 (T 1,2,5,6) the means are 7.24% and 7.31% ash
respectively. A two-sample T test was conducted to see if this difference was significant,
and the difference is not statistically significant for the sample size taken.
Tests T1-4 were combined with T7-12 for statistical analysis of the effects of
running paddles vs. no paddles. In this case, the means between both data sets were
80 |
Virginia Tech | 8.35% and 8.38% ash with paddles on and no paddles respectively. A two sample T test
proved that there was no statistically significance between the two populations, for the
given sample size. This was also shown with the tests conducted on 6/16 with the
primary cells with and without paddles test. In that case the two tests acted almost the
same, showing no difference in performance.
Similarly, the differences between the addition methods, high and low wash
boxes, as well as weiring or sheeting of the wash water are very slight in comparison
between other variables (T13-17). Their means are 7.70% for the high wash box and
7.77% ash for the low wash box. Of the X’s controlling product ash, froth depth, and
adding wash water had the greatest effect on lowering product ash. Tons per hour of
concentrate also played a significant contributor but was really more of a result of other
variables than a contributor. As the froth depth increased so the points on a water ash
curve shifted downward, as in Figure 3.31. This is due to letting more froth water drain
back into the pulp. Also as the froth depth increases, the tons per hour of product
decrease increasing the number of dilution washes. Note: for the tests with the low wash
boxes, the greatest effect seen was the amount of water recovery and not the over all ash.
This is due to less bubble coalescence by adding the water more gently, but not
increasing the froth depth to allow for better drainage. So the net affect is more water
carryover.
81 |
Virginia Tech | 3.5
3.0
y = 0.0185x
2.5
2.0
1.5
1.0
0.5
0.0
0 20 40 60 80 100 120 140 160 180
Water Flow (tph)
82
)hpt(
hsA
fo
etaR
ssaM
9.0 inches
7.0 inches
6.5 inches
4.5 inches
Figure 3.31 - Effect of froth depth on water ash plot for 5-C tests.
3.5.2.1 - Explanation to Effectiveness of Froth Washing.
During the parametric study, the critical X’s were discussed. Both froth depth
and the number of dilution washes were determined to be the controlling factors in the
process. Other questions, such as how wash water can be added, why its effects seem to
be minimized, and why a deeper froth works better, are answered in this section.
For all of the tests in this study, the froth washing always took place at the cell lip.
This was to ensure that all of the froth would be washed. However, looking back a better
placement might have been further back from the edge of the cell. This is due to the
short residence time of the froth leaving the cell after the wash water was added between
0.25 seconds and 2 seconds (deeper froths). This is not enough time for the wash water
to effectively wash and drain the pulp water out of froth. So rather than washing the
clays out of the froth, the clays are only diluted. As froth depth increases, product ash |
Virginia Tech | decreases. This is due to froth water draining back into the pulp, carrying entrained clays
with it. Deeper froths not only allow for more drain back of the clays, but the flow past
the wash water is much slower. This increases the froth’s residence time, allowing the
wash water to drain rather than absorbing the extra water.
One of the differences of wash water systems for columns is that the wash water
is distributed evenly across the entire area of the cell. This allows the froth more time to
drain as it travels to the lip and into the launder, Figure 3.32.
Deep Froth
Wash Water
Pulp
Figure 3.32 - Column cell arrangement.
Froth residence time is dictated by the time it takes for a bubble to rise to the top
of the froth and travel to the lip. Because froth cell geometry is fixed, the time for the
froth to travel to the cell lip is dictated by the length that bubble must travel. Therefore,
the way to increase the residence time is to increase the froth depth. Froth that has
83 |
Virginia Tech | formed in the center of the cell will have a longer distance to the cell lip; therefore, it will
have a longer residence time, and a greater froth liquid drain-back.
Weir bars are an effective way to increase the froth depth, especially when greater
cell residence time is needed to increase recovery. Weir bars increase the froth depth
without changing the cell volume that the pulp is occupying. Lowering the pulp level in
the froth cell also increases the froth depth; however, the volume that the pulp now has to
fill is smaller, which decreases pulp residence time and adversely effects recovery.
The following Figures 3.33 and 3.34 are graphic illustrations of the effects of time
and depth of froth on wash water effectiveness. With shallow, fast-flowing froths, the
wash water does not have time to adequately drain and effectively remove the clays,
Figure 3.33. Instead, the froth water is merely diluted and clays are swept over the cell
lip with the product. Figure 3.34 shows that deep, slow-moving froth increases the
natural froth drain-back, allowing better separation of froth from entrained water, and
reduces the amount of water carryover. The reduction in hydraulic entrained clays is
enhanced when a deep froth is used in combination with wash water; in this case, this
reduction is greater than that which could be achieved by the sum of its parts (the effects
of deep froth and wash water).
84 |
Virginia Tech | Wash Box
Wash
Deep
Slow
Froth
Heavy Pulp Effective
Froth Froth
Drain
Washing
Cell
Figure 3.34 - Effective froth washing with slow deep froth.
The effect of one hole in the wash box is to produce a single stream of water
pouring onto the froth. This stream of water is deflected around the various froth
bubbles. As the water is deflected, the natural shape of this stream becomes a cone,
whose base widens as depth increases. The spacing of the cones is dictated by the hole
spacing of the wash box. The desired effect is to have the froth pass through the
overlapping sections of the cones, where the washing would be the most complete
(Figure 3.35). If the froth is too shallow and the cones do not overlap, parts of the froth
will not pass through the area of wash water influence. Deeper froths increase the
overlap; however, if the froth is too deep, it will become unstable. Froth that has wash
water added can support deeper froths; unfortunately, this system adds wash water at the
86 |
Virginia Tech | cell lip so the total froth does not benefit from this stabilizing effect. Column cells are
able to take advantage of this effect because the wash water is added over the entire froth,
rather than only at the cell lip.
A secondary cause of reduced water effectiveness is the velocity at which the
wash water is added. If the wash water is added too forcefully, then the area of effective
washing will be greatly diminished, i.e. the cone of effective washing does not start at the
surface, but at a point in the froth where the water velocity is slow enough to be
dispersed. Heavier-laden froths tend to spread wash water out faster than lightly loaded
froths, thus wash water needs to be added more gently to lightly loaded froths. If wash
water is added too forcefully, a jet of water will penetrate the froth rather than dispersing
within the froth and replacing the pulp water that makes up the bubbles. When this
happens, the effectiveness of the wash water is mostly due to the increased bubble
coalescence from the forceful water. As a general guideline, wash water needs to be
added as gently as possible; however, as water is added more gently, the froth residence
time needed for adequate water drainage increases. Hence, adding wash water over the
entire cell is preferential to adding it only to the lip.
87 |
Virginia Tech | sides were tested to quantify the residence time the two banks showed a lack of balance
in flow as well as high variability in residence time.
Besides the distribution and residence time problems, the banks had various
mechanical problems. Other problems included a lack of response to frother and diesel
addition rates. All of these problems indicated that the froth cells were overloaded and
adding a wash water system would have a far greater impact on recovery than on
cleaning the froth product. Because the froth product ash ranged between 9 and 18%,
even though previous release analysis showed a >90% recovery at ash values less than
5%, there was great incentive to find a solution. For this reason a two staged flotation
system study was conducted.
3.6.2 - Plant Conventional
The following tests were conducted to see what kind of effects various chemical
addition rates had on the recovery of the system. The tests are summarized here in Table
3.20. Due to the location of the feed sample port, the effects of dilution water are not
seen in the feed percent solids. The actual feed percent solids for all of the tests with
dilution water added are lower than reported, no good method for collecting feed percent
solids information was available. Unfortunately, no corresponding residence time data is
available for the two test conditions, with and without dilution water.
89 |
Virginia Tech | Table 3.20 – Effects of reagent dosage and dilution water for Plant 2 froth cells.
Test # Description Chemicals ml Feed Con Tails
1 Normal reagen dosage Frother 110 Ash 42.86 17.85 50.38
Dilution water added Kerosene 150 % Solids 24.34 31.46 9.35
2 Frother 110 Ash 42.25 17.68 50.72
Dilution water added Kerosene 196 % Solids 24.12 30.61 9.38
3 Frother 132 Ash 43.36 17.94 47.89
Dilution water added Kerosene 150 % Solids 23.61 30.2 10.35
4 Frother 150 Ash 42.4 17.36 75.84
No dilution water added Kerosene 150 % Solids 24.09 33.87 10.33
Cumbustible Froth
Test # Description Yeild Recovery depth (in.)
1 Normal reagent dosage 23.12 33.24 3
Dilution water added
2 Dilution water added 25.64 36.54 3
3 Dilution water added 15.13 21.91 3
4 No dilution water added 57.18 82.04 8
As mentioned earlier in section 3.5.1, the froth cells at Plant 2 are very unreactive
to variation in frother and kerosene addition for certain feed conditions. This information
coupled with the fourth test where the dilution water was eliminated, indicates that the
froth cells are overloaded. Often the first cell would be unreactive to any changes made
to the system, and no froth would form, indicating that the first cell was acting more like
a conditioner than a froth cell greatly reducing the effective residence time. Normally
reducing the amount of dilution water would decrease the recovery of the system, as well
increase the product ash; however, in this case the opposite effect was seen.
90 |
Virginia Tech | 3.6.3 - Two Stage Float
3.6.3.1 - Two Stage Float Experimental
For this study a combined concentrate sample was collected Table 3.21. All work
was done from this sample, rather than from fresh feed. This was done to keep from
biasing the first stage results. The froth product sample was split via a slurry splitter into
five equal samples. Each sample was then diluted with water and floated using a two-
liter laboratory flotation cell. The target dilution rates were 5, 10, 15, and 20% solids,
however actual tests produce feeds of the following percent solid ratios: 4.2, 8.8, 13.1,
and 17.3. The purpose of this testing was to show what kind of product could be
obtained while maintaining good recovery.
Table 3.21 – Plant performance for corresponding test samples.
USX Product Run
Ash % Solids Yield Combustible Recovery
Feed 47.03 37.78 51.5 87.9
Con. 9.535 22.31
Tails 86.81 9.9
The tests conducted as follows: Each sample was conditioned for one minute after
one drop of frother was added, and before any air was added to the system. The air was
then added to the cell and product was paddled off as rapidly as it was produced. Care
was taken that the froth paddled off of the cell did not contain any pulp; this was to
minimize any clay entrainment. The product sample was divided by how much time was
used in making the product. For all samples other than the 4.2 percent solids dilution,
times of 1, 2, 3, and 5 minutes were used as the demarcation times. For the 4.2 percent
91 |
Virginia Tech | solids dilution product was grouped by half-minute intervals, through one and a half
minutes and the final product sample at 3 minutes.
3.6.3.2 - Two Stage Float Results
All of the tests produced recoveries above 93% by three minutes of residence
time, and over 99% by 5 minutes as shown in Figure 3.36. Because all of the coal in this
second stage of flotation had already been floated, the coal floated quickly. This also
explains the high recovery in the second stage of flotation.
100.00
80.00
60.00
40.00
20.00
0.00
0.00 1.00 2.00 3.00 4.00 5.00
Time (minutes)
92
)%(
yrevoceR
elbitsubmoC
T- 8.83%
T- 13.14%
T- 17.26%
T- 4.23%
Figure 3.36 - Effect of time on recovery for various repulped % solids.
Each one of the second stages of flotation tests provided excellent reduction in
ash. The ash reduction ranged from 3.8 points (41%) for 17.3% solids, and 4.9 points
(53%) for the 4.2% solids dilution, Figure 3.37. When compared to the ash reduction
from froth washing, this method is far superior in effectiveness. The drawback from this |
Virginia Tech | 3.6.3.3 - Two Stage Float Discussion.
Several conclusions can be drawn from this series of tests. The first conclusion is
that as dilution water increases the product quality approaches that of release analysis.
This is because the concentration of clays in the pulp decreases with increasing dilution.
All of the clays recovered are from hydraulic entrainment.
The second conclusion that can be drawn is that as dilution is increased the time
required to recover the same amount of coal decreases. This is partly due to increased air
to coal particle ratio, increasing the probability of coal attaching to bubbles. However as
dilution rate increases so does the volumetric flow rate through the cells. So there is a
trade off between quicker recovery rates and decreased residence time.
The third conclusion that can be drawn from these tests is similar to the first: that
the selectivity of the circuit increases with increasing dilution. This very simply put
means that as dilution of clay in the pulp is increased, hydraulic entrainment is decreased,
decreasing the recovery of clay. Selectivity is the recovery of desired material over
recovery of gangue. Figure 3.38 shows that as residence time is increased selectivity is
increased. This trend would continue only until all of the coal is recovered. Also the rate
at which selectivity increases grows as dilution is increased. These effects are amplified
when looking at the total system, as shown in Figure 3.39. Note that the expected
selectivity for a column flotation cell is similar to the selectivity of the high dilution
scenario, however it takes half of the residence time that the two-stage flotation system
does.
94 |
Virginia Tech | Chapter 4
CONCLUSIONS
Wash water systems like the system installed at Plant 1 provide an inexpensive
solution to reducing hydraulic entrainment in conventional froth flotation systems. This
study has shown that reductions in ash by one percentage point can be achieved through
this system. Critical to the operation of these systems are the following:
Ample residence time (above 6 minutes)
Bias flow of wash water (1.2-1.4 preferred)
Wash water should be added in a gentle manner
Even wash water distribution
As deep a froth as can be run and still be stable enough for recovery over 90%
Only enough frother to produce a stable froth
Feed that does not have a high coarse fraction (>48 Mesh)
Even with all of the above parameters optimized, the wash water system does not
compete with either two stages of flotation or column cell flotation for comparing
performance to release analysis.
Two stage flotation systems can achieve high degrees of cleaning approaching
that of column cells and even release analysis if the second stage is diluted to the five
percent solids range. Drawbacks include potential loss of recovery when compared to
column cells and a guaranteed loss of recovery to single stage conventional flotation.
97 |
Virginia Tech | Mass Balance For Froth Cells Using Complete
Streams
Plant 1 Measured or Calculated
Value
With Wash Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
Test 2-B F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 24.79 8.21 76.15 85.00 7.10 7.88 6.22 8.26 11.78 0.00 85.00 85.00 85.00 85.00
Mass (tph):
Total 174.30 131.76 42.57 0.03 36.41 44.90 17.14 20.74 12.57 0.000 0.004 0.010 0.010 0.010
Ash (tph):
Minus 325 M 43.21 10.82 32.42 0.03 2.58 3.54 1.07 1.71 1.48 0.00 0.00 0.01 0.01 0.01
Percent Solids: 17.19 25.97 4.00 0.027 29.98 26.73 24.42 20.97 17.09 0.000 0.027 0.027 0.027 0.027
Solid SG: 1.48 1.34 2.20 2.40 1.33 1.34 1.32 1.34 1.37 1.28 2.40 2.40 2.40 2.40
Flow Rate:
Slurry (tph) 1014 507 1065 129 121 168 70 99 74 0 16 38 38 38
Liquid (tph) 840 376 1023 129 85 123 53 78 61 0 16 38 38 38
Slurry (gpm) 3827 1895 4165 515 449 626 264 374 281 0 65 150 150 150
Liquid (gpm) 3359 1502 4091 515 340 492 212 313 244 0 65 150 150 150
Estimated or Calculated
Value
Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 23.88 8.08 81.01 81.01 7.28 8.16 6.29 8.40 11.95 0.00 81.01 81.01 81.01 81.01
Mass (tph):
Total 167.73 131.40 36.37 0.03 36.31 44.88 16.90 20.61 12.70 0.000 0.004 0.010 0.010 0.010
Ash (tph):
Minus 325 M 40.05 10.62 29.46 0.03 2.64 3.66 1.06 1.73 1.52 0.000 0.003 0.008 0.008 0.008
Percent Solids: 13.72 25.10 4.39 0.027 30.32 27.14 24.58 21.17 17.54 0.000 0.027 0.027 0.027 0.027
Solid SG: 1.47 1.34 2.30 2.30 1.33 1.34 1.32 1.34 1.37 1.28 2.30 2.30 2.30 2.30
Flow Rate:
Slurry (tph) 1223 524 828 129 120 165 69 97 72 0 16 38 38 38
Liquid (tph) 1055 392 792 129 83 120 52 77 60 0 16 38 38 38
Slurry (gpm) 4672 1960 3227 515 442 616 258 368 276 0 65 150 150 150
Liquid (gpm) 4220 1569 3167 515 334 482 208 307 239 0 65 150 150 150
126 |
Virginia Tech | Mass Balance For Froth Cells Using complete
streams
Plant 1 Measured or Calculated
Value
With Wash Water Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
Test 3-B F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 22.19 8.47 86.06 85.00 7.45 9.09 10.43 12.73 16.52 85.00 85.00 85.00 85.00 85.00
Mass (tph):
Total 150.81 124.14 26.75 0.07 50.87 49.68 16.00 6.14 1.45 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 33.47 10.51 23.02 0.06 3.79 4.52 1.67 0.78 0.24 0.02 0.02 0.01 0.01 0.01
Percent Solids: 15.66 27.22 3.07 0.027 22.55 20.27 18.15 12.30 8.32 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.45 1.34 2.42 2.40 1.33 1.34 1.35 1.37 1.40 2.40 2.40 2.40 2.40 2.40
Flow Rate:
Slurry (tph) 963 456 872 263 226 245 88 50 17 75 75 38 38 38
Liquid (tph) 812 332 845 263 175 195 72 44 16 75 75 38 38 38
Slurry (gpm) 3660 1697 3421 1050 851 928 336 193 68 300 300 150 150 150
Liquid (gpm) 3249 1328 3380 1051 699 781 289 175 64 300 300 150 150 150
Estimated or Calculated
Value
Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 21.54 8.56 86.26 86.26 7.27 8.84 10.07 12.67 16.46 86.26 86.26 86.26 86.26 86.26
Mass (tph):
Total 148.21 123.45 24.83 0.07 53.21 49.38 13.62 5.94 1.30 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 31.93 10.57 21.42 0.06 3.87 4.37 1.37 0.75 0.21 0.017 0.017 0.009 0.009 0.009
Percent Solids: 13.74 21.10 3.28 0.027 24.07 21.66 18.57 12.49 8.48 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.45 1.34 2.43 2.43 1.33 1.34 1.35 1.37 1.40 2.43 2.43 2.43 2.43 2.43
Flow Rate:
Slurry (tph) 1079 585 756 263 221 228 73 48 15 75 75 38 38 38
Liquid (tph) 931 462 732 263 168 179 60 42 14 75 75 38 38 38
Slurry (gpm) 4127 2213 2965 1050 830 861 279 184 60 300 300 150 150 150
Liquid (gpm) 3722 1847 2926 1051 671 714 239 166 56 300 300 150 150 150
134 |
Virginia Tech | Mass Balance For Froth Cells Using complete
streams
Plant 1 Measured or Calculated
Value
With Wash Water Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
Test 4-B F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 33.63 9.22 89.35 85.00 8.91 7.84 10.57 13.58 19.55 85.00 85.00 85.00 85.00 85.00
Mass (tph):
Total 108.56 75.49 33.15 0.07 36.62 21.83 12.68 3.12 1.25 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 36.51 6.96 29.62 0.06 3.26 1.71 1.34 0.42 0.24 0.02 0.02 0.01 0.01 0.01
Percent Solids: 12.67 18.97 3.64 0.027 19.95 18.72 16.38 11.52 8.96 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.57 1.35 2.51 2.40 1.34 1.33 1.36 1.38 1.43 2.40 2.40 2.40 2.40 2.40
Flow Rate:
Slurry (tph) 857 398 911 263 184 117 77 27 14 75 75 38 38 38
Liquid (tph) 748 322 878 263 147 95 65 24 13 75 75 38 38 38
Slurry (gpm) 3268 1512 3562 1050 696 444 296 105 54 300 300 150 150 150
Liquid (gpm) 2994 1289 3512 1051 588 379 259 96 51 300 300 150 150 150
Estimated or Calculated
Value
Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 32.45 9.23 89.02 89.02 8.89 7.83 10.55 13.57 19.55 89.02 89.02 89.02 89.02 89.02
Mass (tph):
Total 105.89 75.07 30.89 0.07 36.48 21.69 12.56 3.10 1.25 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 34.37 6.93 27.50 0.06 3.24 1.70 1.33 0.42 0.24 0.018 0.018 0.009 0.009 0.009
Percent Solids: 10.85 18.30 3.73 0.027 20.36 18.97 16.53 11.55 8.98 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.55 1.35 2.50 2.50 1.34 1.33 1.36 1.38 1.43 2.50 2.50 2.50 2.50 2.50
Flow Rate:
Slurry (tph) 976 410 828 263 179 114 76 27 14 75 75 38 38 38
Liquid (tph) 870 335 797 263 143 93 63 24 13 75 75 38 38 38
Slurry (gpm) 3748 1562 3235 1050 679 435 290 104 54 300 300 150 150 150
Liquid (gpm) 3479 1341 3189 1051 571 371 254 95 51 300 300 150 150 150
142 |
Virginia Tech | Mass Balance For Froth Cells Using complete
streams
Plant 1 Measured or Calculated Value
With Wash Water Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
Test 4-C F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 31.20 9.20 89.47 85.00 9.26 7.50 10.25 13.59 19.37 85.00 85.00 85.00 85.00 85.00
Mass (tph):
Total 120.87 87.75 33.19 0.07 41.27 21.16 17.34 5.87 2.11 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 37.70 8.07 29.69 0.06 3.82 1.59 1.78 0.80 0.41 0.02 0.02 0.01 0.01 0.01
Percent Solids: 13.28 18.97 3.81 0.027 18.80 17.85 19.17 12.43 8.82 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.54 1.35 2.51 2.40 1.35 1.33 1.35 1.38 1.43 2.40 2.40 2.40 2.40 2.40
Flow Rate:
Slurry (tph) 910 462 870 263 220 119 90 47 24 75 75 38 38 38
Liquid (tph) 789 375 837 263 178 97 73 41 22 75 75 38 38 38
Slurry (gpm) 3467 1758 3398 1050 835 453 343 182 93 300 300 150 150 150
Liquid (gpm) 3156 1499 3349 1051 713 390 292 165 87 300 300 150 150 150
Estimated or Calculated Value
Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 30.44 9.32 89.03 89.03 8.92 7.39 10.08 13.49 19.30 89.03 89.03 89.03 89.03 89.03
Mass (tph):
Total 119.36 87.73 31.70 0.07 41.30 21.36 17.23 5.77 2.08 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 36.33 8.18 28.22 0.06 3.68 1.58 1.74 0.78 0.40 0.018 0.018 0.009 0.009 0.009
Percent Solids: 11.88 17.90 4.08 0.027 19.31 18.12 19.39 12.51 8.85 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.53 1.35 2.50 2.50 1.34 1.33 1.35 1.38 1.43 2.50 2.50 2.50 2.50 2.50
Flow Rate:
Slurry (tph) 1005 490 778 263 214 118 89 46 23 75 75 38 38 38
Liquid (tph) 886 402 746 263 173 97 72 40 21 75 75 38 38 38
Slurry (gpm) 3850 1868 3032 1050 812 450 337 178 91 300 300 150 150 150
Liquid (gpm) 3543 1609 2984 1051 690 386 287 161 85 300 300 150 150 150
146 |
Virginia Tech | Mass Balance For Froth Cells Using complete
streams
Plant 1 Measured or Calculated Value
With Wash Water Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
Test 4-D F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 31.44 8.53 84.42 85.00 6.55 6.19 9.01 12.54 13.07 85.00 85.00 85.00 85.00 85.00
Mass (tph):
Total 128.67 89.83 38.91 0.07 24.64 19.72 27.19 12.83 5.45 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 40.45 7.66 32.84 0.06 1.61 1.22 2.45 1.61 0.71 0.02 0.02 0.01 0.01 0.01
Percent Solids: 12.51 20.13 3.41 0.027 23.00 21.54 19.42 15.38 13.47 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.54 1.34 2.38 2.40 1.32 1.32 1.34 1.37 1.38 2.40 2.40 2.40 2.40 2.40
Flow Rate:
Slurry (tph) 1028 446 1140 263 107 92 140 83 40 75 75 38 38 38
Liquid (tph) 900 356 1101 263 82 72 113 71 35 75 75 38 38 38
Slurry (gpm) 3928 1692 4466 1050 404 347 532 320 156 300 300 150 150 150
Liquid (gpm) 3599 1426 4405 1051 330 287 451 282 140 300 300 150 150 150
Estimated or Calculated Value
Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 30.45 8.51 85.73 85.73 6.57 6.21 9.07 12.59 13.09 85.73 85.73 85.73 85.73 85.73
Mass (tph):
Total 124.70 89.27 35.50 0.07 24.47 19.55 27.02 12.79 5.43 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 37.97 7.60 30.44 0.06 1.61 1.21 2.45 1.61 0.71 0.017 0.017 0.009 0.009 0.009
Percent Solids: 10.85 19.59 3.71 0.027 23.20 21.69 19.64 15.48 13.64 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.53 1.34 2.42 2.42 1.33 1.32 1.34 1.37 1.38 2.42 2.42 2.42 2.42 2.42
Flow Rate:
Slurry (tph) 1149 456 956 263 105 90 138 83 40 75 75 38 38 38
Liquid (tph) 1024 366 921 263 81 71 111 70 34 75 75 38 38 38
Slurry (gpm) 4419 1731 3738 1050 398 341 522 316 153 300 300 150 150 150
Liquid (gpm) 4098 1466 3683 1051 324 282 442 279 138 300 300 150 150 150
151 |
Virginia Tech | Mass Balance For Froth Cells Using complete
streams
Plant 1 Measured or Calculated
Value
With Wash Water Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
Test 4-E F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 31.33 8.29 82.83 85.00 6.58 7.02 8.87 11.72 12.01 85.00 85.00 85.00 85.00 85.00
Mass (tph):
Total 126.92 87.69 39.30 0.07 27.67 19.07 24.65 11.54 4.76 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 39.76 7.27 32.55 0.06 1.82 1.34 2.19 1.35 0.57 0.02 0.02 0.01 0.01 0.01
Percent Solids: 12.40 20.60 3.52 0.027 25.10 23.10 19.56 15.60 14.64 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.54 1.34 2.35 2.40 1.33 1.33 1.34 1.37 1.37 2.40 2.40 2.40 2.40 2.40
Flow Rate:
Slurry (tph) 1024 426 1116 263 110 83 126 74 32 75 75 38 38 38
Liquid (tph) 897 338 1077 263 83 63 101 62 28 75 75 38 38 38
Slurry (gpm) 3912 1612 4372 1050 413 311 478 283 125 300 300 150 150 150
Liquid (gpm) 3586 1352 4309 1051 330 254 405 250 111 300 300 150 150 150
Estimated or Calculated
Value
Feed Clean Reject Wash Clean #1 Clean #2 Clean #3 Clean #4 Clean #5 Wash #1 Wash #2 Wash #3 Wash #4 Wash #5
F1 C1-C5 R5 W1-W5 C1 C2 C3 C4 C5 W1 W2 W3 W4 W5
Ash (% stream):
Minus 325 M 30.23 8.30 83.43 83.43 6.58 7.02 8.88 11.73 12.01 83.43 83.43 83.43 83.43 83.43
Mass (tph):
Total 123.01 87.10 35.98 0.07 27.47 18.96 24.43 11.49 4.75 0.020 0.020 0.010 0.010 0.010
Ash (tph):
Minus 325 M 37.19 7.23 30.02 0.06 1.81 1.33 2.17 1.35 0.57 0.017 0.017 0.008 0.008 0.008
Percent Solids: 10.64 20.65 3.61 0.027 25.16 23.14 19.61 15.62 14.65 0.027 0.027 0.027 0.027 0.027
Solid SG: 1.53 1.34 2.36 2.36 1.33 1.33 1.34 1.37 1.37 2.36 2.36 2.36 2.36 2.36
Flow Rate:
Slurry (tph) 1156 422 997 263 109 82 125 74 32 75 75 38 38 38
Liquid (tph) 1033 335 961 263 82 63 100 62 28 75 75 38 38 38
Slurry (gpm) 4448 1597 3901 1050 409 309 473 282 125 300 300 150 150 150
Liquid (gpm) 4131 1338 3843 1051 327 252 401 248 111 300 300 150 150 150
155 |
Virginia Tech | Methods of Improving Oil Agglomeration
Sarah Ann Smith
Abstract
A simple thermodynamic analysis suggests that oil can spontaneously displace water
from coal’s surface if the coal particle has a water contact angle greater than 90°. However, the
clean coal products obtained from laboratory-scale dewatering-by-displacement (DbD) test work
assayed moistures substantially higher than expected. These high moisture contents were
attributed to the formation of water-in-oil emulsions stabilized by coal particles. Four different
approaches were taken to overcome this problem and obtain low-moisture agglomeration
products. These included separating the water droplets by screening, breaking emulsions with
ultrasonic energy, breaking agglomerates with ultrasonic energy, and breaking agglomerates
using vibrating mesh plates. On the basis of the laboratory test work, a semi-continuous test
circuit was built and tested using an ultrasonic vibrator to break the water-in-oil emulsions.
The most promising results were obtained agglomerates were broken using the ultrasonic
probe and the vibrating mesh plates. Tests conducted on flotation feed from the Kingston coal
preparation plant gave a clean coal product containing 1% by weigh of moisture with a 94%
combustible recovery. The separation efficiency of 93% is substantially higher than results
achievable using froth flotation. When agglomerates formed from thermal coal from the Bailey
coal preparation plant were broken using either ultrasonic energy or vibrating mesh plates, the
obtained results were very similar: clean coal products assayed less than 5% moisture with
separation efficiencies of 86% in average.
ii |
Virginia Tech | Chapter 1: General Introduction
1.1 Preamble
In 2011, 42% of the electricity in the United States was generated from coal (EIA 2011).
Additionally, projections show that 39% of US electricity will still come from coal in 2035 (EIA
2012). Over the next 23 years it is expected that electricity generation will increase by
approximately 1 trillion kilowatt hours (EIA 2012). With the majority of electricity coming from
coal, it is more important than ever that coal processing becomes as economic and efficient as
possible.
In older coal preparation plants, the recovery processes for fine and ultrafine coal were not
as efficient as today; therefore, many impoundments contain recoverable coal as large as 600 µm
(National Research Council 2002). The larger material (+44 µm) can be re-mined and
recovered using current operations in the plant; however, some ultrafine material is still lost in
reprocessing.
While improvements in coal cleaning technology have led to an increased recovery of fine
particles within the preparation plant, plant refuse still contains a large number of particles finer
than 44 µm (National Research Council 2002). Because the -44 µm size class is so well
liberated, its lower ash content makes it an extremely desirable product. These particles have a
large surface area, and therefore have a higher moisture content. Thermal dryers afford the
opportunity to dewater and dry this ultrafine coal; however, dryers also create an array of
problems including high operating cost, potential safety concerns from explosions, and
regulatory restrictions. As a result, most ultrafine particles are discarded with the tailings stream
and ultimately end up in impoundments. A new and innovative process to recover these ultrafine
particles has two main benefits: the ultrafine coal is a sellable product that will pay for the
recovery process and recovering the coal decreases the amount of material being sent to the
impoundment.
1 |
Virginia Tech | 1.2 Objectives
The goal of the project is to improve the oil agglomeration process in order to replace
current technologies used to clean and dewater fine and ultrafine coal. An energy source is used
to disperse coal particles into a phase comprised of hydrophobic liquid, while water and ash
remain in a separate phase. The hydrophobic phase will be recovered, and the liquid recycled.
The resulting clean coal product should contain less than 10% moisture. All experiments will use
Pentane as the hydrophobic liquid since it is affordable and can be easily recycled through
evaporation and condensation.
This project focused on conducting laboratory-scale batch tests using four separation
methods to determine the best method for scale-up. The separation methods focused on cleaning
coal via agglomeration or emulsification. During the experimental processes, the produced
agglomerates or emulsions were dispersed into hydrophobic liquid using mechanical vibrations
from an ultrasonic probe or vibrating mesh plates. Additionally, a laboratory-scale continuous
testing unit was developed and constructed to test the feasibility of the process over a long
operating period.
1.3 Organization
This thesis is divided into five chapters. The first chapter discusses the need and benefits of
the Dewatering by Displacement method and the objectives of the laboratory testing methods
employed.
Chapter 2: Literature Review is designed to bring readers up to date on the necessary
background needed to understand the scope of the project. The review discusses prior research
that evolved into the current project, thermal drying, the theory of Pickering emulsions, and the
theory and practice of oil agglomeration.
The third chapter reviews the laboratory batch testing methods tried throughout the project.
Each subsection of chapter three is dedicated to one of the four methods tried. For each of the
testing methods, subsections will be further divided to discuss the experimental apparatus,
experimental methods, and results obtained using the method. The fourth chapter is organized
similar to the third chapter, but focuses on a continuous laboratory testing unit. Finally, the fifth
2 |
Virginia Tech | Chapter 2: Literature Review
2.1 Prior Work
2.1.1 Hydrophobic Displacement
Virginia Tech began developing the dewatering by displacement process in 1995 by
researching hydrophobic displacement. Hydrophobic displacement uses a hydrophobic liquid
with a higher affinity for the coal’s surface to displace water presently on the surface of the coal.
Figure 2-1a shows that when dewatering by hydrophobic displacement, a hydrophobic solid
particle (in this case coal) of state 1, must leave an aqueous state, 3, and cross an interface into a
non-polar hydrophobic (oil) phase, 2 (Yoon). For this displacement to occur, the change in Gibbs
free energy, G, with respect to the particle surface area, A, must equal the difference in the
surface free energy at the coal/oil interface and the surface free energy at the coal/water
interface, represented by γ and γ respectively. Additionally, for a reaction to be
12 13,
thermodynamically spontaneous, the change in Gibbs free energy must be less than zero.
Therefore:
dG/dA= γ - γ < 0 [1]
12 13
Young’s equation can be used to define the relationship between γ , γ , and the surface
12 13
free energy at the oil/water interface, γ (Yoon):
23
dG/dA= γ - γ = γ cos θ [2]
12 13 23
where θ represents the contact angle between a droplet of oil on a coal surface and the surface
itself, measured through the water phase (Figure 2-1b). Equation 2 can be substituted into
Equation 1, so that:
dG/dA= γ cos θ < 0 [3]
23
Therefore, when θ > 90°, the hydrophobic oil phase will displace water on the surface of the
coal.
4 |
Virginia Tech | Figure 2-1: a) Schematic showing the displacement of coal, 1, from the water phase, 3, into the oil phase, 2;
and b) the three-phase equilibrium between interfacial tensions (Redrawn from Yoon).
2.1.2 Low Temperature Drying
In 2010, a low temperature drying process was developed at Virginia Tech to reduce the
moisture of displaced coal. The technology was applicable to coal agglomerates and filtered
flotation concentrate with less than approximately 22% moisture (Freeland 2010). Three devices
were developed to explore the process: a static breaker, air jet conveyor, and centrifugal fan. In
each device, the coal agglomerates or cake were subjected to a high, mechanical shearing force.
Compared to the other two methods, the centrifugal fan consistently produced a low-moisture
product (less than 2%) without plugging.
Low temperature drying requires a high amount of airflow to dry the particles. The relative
humidity and temperature of the ambient air have a large impact on the water carrying capacity
of the air. It was discovered that the process worked best by heating the air to at least 48.89 ºC
(120ºF) (Freeland 2010). Unfortunately, heating the air adds an additional cost to the process.
Based on a mathematical model to calculate the cost of the an industrial scale low temperature
dryer, it was discovered that a thermal dryer requires $0.18/ton less energy than a low heat dryer
(Freeland 2010).
5 |
Virginia Tech | 2.2 Thermal Drying
For fine and ultrafine coals, mechanical dewatering is able to lower the surface moisture to
approximately 15 to 25%, respectively (Korte and Mangena 2004). Due to customer
specifications and difficulty in handling, these values are unacceptable. Thermal drying is
currently the only operation which can give fine and ultrafine coals moistures in the single digit
range.
2.2.1 Operation
All thermal dryers operate under the same condition: in order for moisture to evaporate
from the surface of the coal, the coal must be brought into contact with a heat source. There are
many different types of thermal dyers used in the coal industry including: rotary dryers, flash
dryers, fluidized beds, and conveyor type dryers.
Rotary thermal dryers are one of the most common dryers used in industry. Of the two
types of rotary driers, parallel flow and counterflow, parallel flow rotary driers are more
commonly used due to higher fuel efficiency rates and larger capacity. Figure 2-2 shows a basic
parallel flow rotary drier. The dryer consists of long, rotating, cylindrical shell. In direct heating
dryers, feed and heat enter at the upper end of the dryer and flow downward by gravity due to the
slight incline of the dryer drum. In indirect heating dryers, heat circulates around the rotating
cylinder drum. At the end of the process, dry product is deposited onto a conveyor belt while the
remaining hot air and fugitive dust particles are exhausted into a cyclone (Wills and Napier-
Munn 2006).
Figure 2-2: Example of a parallel flow rotary dryer (Wills and Napier-Munn 2006).
6 |
Virginia Tech | 2.2.2 Operation Issues
When using thermal dryers, it is possible for filter cake to clump within the dryer,
creating a risk for a fire or explosion. In order for a fire to occur, fuel, heat, and oxygen must be
present. For an explosion, two additional elements are required: dust suspension and confinement
(Korte and Mangena 2004). To avoid suspension and clumping, coarser material (6.35 mm x
0.595 mm) is often mixed with fine coal in rations of 2:1 to 4:1 (Luckie 1991). While this
method keeps material from clumping in the dryer, the larger size class can normally be dried
using cheaper methods, thus creating inefficiencies in the drying process.
2.2.3 Permitting Issues
Sulfur dioxide (SO ), nitrogen oxides (NO ), and carbon monoxide (CO) are all gaseous
2 x
byproducts of operating a thermal dyer. In addition to these gases, several other volatile
compounds released by the coal when heating are considered to have carcinogenic effects. For
thermal dryers built after 2008, the EPA has implemented very strict emission and monitoring
guidelines for these gaseous byproducts. The guidelines are so strict that is very difficult for a
mine operator to obtain a new permit for a thermal dryer and operate below maximum emission
levels (2012).
2.3 Pickering Emulsions
In 1907, S.U. Pickering discovered that solid particles could be used to stabilize
emulsions (Giermanska-Kahn, Schmitt et al. 2002). Finkle et al. were the first to make a
distinction between the two types of emulsions formed (oil-in-water versus water-in-oil). His
team discovered that in a particle-stabilized emulsion, one liquid will wet the particles more than
the other liquid. The liquid with the poorer wetting properties will become dispersed in the phase
that better wets the particles. The following sections provide a summary of the theories behind
the formation Pickering emulsions and factors that affect the determination of the type of
emulsions formed.
2.3.1 Emulsion Formation
When a group of particles are placed at an air-water interface, the particles are disordered
with no geometric arrangement. However, when an alkane is layered onto the interface, the
particles rearrange into an ordered monolayer displaying a hexagonal pattern. When electrolyte
concentrations of up to 0.1M NaCl, high enough to cause particle aggregation at an air-water
7 |
Virginia Tech | interface, were added to the aqueous phase of the oil-water system, the layer maintained its
geometric arrangement (Binks 2002). Therefore, this arrangement is thought to be due to long-
range repulsive forces acting over several micrometers through the oil phase (Binks 2002). It is
believed that residual surface charges on the particle surface are responsible for the repulsion.
The minimum charge required to cause repulsion was calculated to be only 1% of the total
possible surface charge.
When the monolayer is compressed, the most hydrophobic and hydrophilic particles are
expelled from the layer and report to the oil and water phases, respectively. Particles of
intermediate hydrophobicity remain on the interface. Upon compression by additional particles,
the monolayer rearranges from a hexagonal array to a rhombohedral array. Further compression
of the rhombohedral array will first lead to the layer folding over on itself and upon even further
compression the layer will create a wave shape (Binks 2002).
2.3.2 Emulsion Type
There are two possible types of emulsions, oil-in-water (o/w) or water-in-oil (w/o). As
shown in Figure 2-3, the contact angle, θ, formed between the particle and the oil-water interface
determines which type of emulsion is formed. For hydrophilic particles, the contact angle formed
is normally less than 90°; therefore, an oil-in-water emulsion will be formed. A larger portion of
the particle surface area will exist in the water phase and particles will encapsulate the oil or air
droplet. Similarly, if the particle is hydrophobic, the contact angle formed between the particle
and the interface will be greater than 90°. In this case, the majority of the particle surface area
will reside in the oil phase and water will be stabilized by particles. Only in cases when θ is
exactly 90°, will no emulsion form. The particle will have an equal surface area in both the oil
and water phases and the net curvature will be zero.
8 |
Virginia Tech | Figure 2-3: Emulsion type based on contact angle, θ. For
θ<90°, oil- in-water emulsions form (left). For θ>90°, water-in-
oil Emulsions form. Used with permission of Binks. (Binks
2002).
Once an emulsion is formed, it is possible to cause inversion and change the emulsion
type. This can be accomplished by changing the oil:water ratio or by changing the average
wettability of particles at the interface. By changing the ratio of oil to water, inversion can occur
so that an o/w emulsion will change to a w/o emulsion or vice versa, despite particle
hydrophobicity. Figure 2-4 shows how the fraction of water, φ , present in the system affects the
w
conductivity of the emulsions and the type of emulsions formed. At low water volumes,
conductivity of emulsions is low and emulsions disperse in the oil phase (w/o emulsions
formed). As the volume of water increases, the type of emulsions formed change-to-oil in water
emulsions and conductivity greatly increases. The experiment was conducted using particles
with 50% SiOH. Open points on the graph represent cases where the ratio was changed by
adding water to oil while filled points represent oil added to water. Based on this study, it was
determined that the most stable emulsions were formed at points near inversion (Binks 2002).
9 |
Virginia Tech | 2.3.3 Energy of Detachment
For particles small enough so that gravity effects are negligible (less than a few microns
in diameter), the energy of detachment, E, to remove the particle from the oil water interface can
be calculated using Equation 1 (Binks 2002):
(cid:1) = (cid:3)(cid:4)(cid:5)(cid:6) (cid:9)1±(cid:12)(cid:13)(cid:14)(cid:15) (cid:16)(cid:5) [4]
(cid:7)(cid:8) (cid:7)(cid:8)
where r equals the particle radius and ɣ is the tension at the oil/water interface. For removal
ow
into the water phase, a negative sign is used inside the bracket while a positive sign is used for
removal into the oil phase. A series of calculations to determine the effect of contact angle on the
energy of detachment has been conducted by Binks. It was found that particles are most strongly
held at the interface when θ = 90°. As the contact angle moves away from 90°, the energy
ow
necessary to hold particles at the interface greatly decreases. For angles between 0 to 20° and
160 to 180° the energy of detachment was nearly thirty times less than the energy required at
90°.
As shown in Equation 4, the detachment energy is dependent on the square of the particle
radius. Figure 2-6 shows the relationship between particle radius and energy of detachment at an
alkane-water interface when θ = 90° and ɣ = 50 mN/m. The graph shows that low energy is
ow
required to detach very small particles (less than 0.5 nm), leading to the conclusion that smaller
particles create less stable emulsions compared to larger sized particles.
Figure 2-6: Required energy of detachment as a function of
particle radius. Used with permission of Binks.
11 |
Virginia Tech | 2.4 Oil Agglomeration
An increase in low-seam mechanized mining methods has resulted in an increased
production of coal fines mixed with high ash tailings (Hazra, Rao et al. 1988). Due to the small
size of the particles (-44 µm), froth flotation is not always a suitable cleaning method. Compared
to oil agglomeration with an oil recovery system, research has shown that using froth flotation to
decrease the ash in ultrafine coal decreases yield and increases processing cost (Hazra, Rao et al.
1988). The main factors affecting the cost of the agglomeration process are the cost of the coal,
the cost of the oil, and the selling price of the final product (Hazra, Rao et al. 1988). Oil
agglomeration is most economical when performed on refuse material leaving as effluent since
the material is essentially “free” and would otherwise be sent to impoundments (Nicol 1979).
2.4.1 History
Oil agglomeration for coal cleaning first began in the early 1920’s as the Trent Process.
This first process used a 40% solid, -100 mesh (150 µm) slurry mixed with 30% fuel oil (by
weight of dry coal). Mixing time was in the range of 6-24 minutes and oil loss was around 10 to
50%. The costs of creating the agglomerates were so high that the agglomerates could not be
sold to the steam market and could only be sold as home “Superfuel” after briquetting,
Therefore, the process was abandoned after only a few years of operation (Mehrotra, Sastry et al.
1983).
Oil agglomeration was heavily researched and developed during the 1970’s due to a
sharp increase in oil prices. Though all oil agglomeration processes operate under the same
theory, a large number of oil agglomeration patents were filed during the 70’s and 80’s due to
significant process variations. During this time period, the oil agglomeration process was
improved, oil dosages, and mixing times were decreased while combustible recovery remained in
the upper 90th percentile. Several pilot plants were constructed to test the feasibility of the
process as a method of cleaning fine coal. As oil prices began to decrease in the late 1980’s the
process once again became uneconomical as an oil substitute was no longer needed (Mehrotra,
Sastry et al. 1983).
2.4.2 Factors affecting agglomeration
Oil agglomeration depends on the surface hydrophobicity of a coal particle and its ability
to be preferentially wetted by hydrophobic oil (Mehrotra, Sastry et al. 1983). The interaction that
12 |
Virginia Tech | occurs between the hydrophobic particles and hydrophobic liquid is controlled by three factors:
the free energy at the three phase interface (the interface between water, coal, and the
hydrophobic liquid), the amount of hydrophobic liquid used, and the mixing intensity (Capes and
Darcovich 1984). These interactions allow the oil to wet the coal and suspend it in a hydrophobic
phase while the ash and oxide matter remain in the aqueous phase. With sufficient oil and
mechanical agitation, coal particles collide and an oil bridge is formed between particles. The
interfacial tension of the oil and capillary attraction of the bridges helps keep the agglomerates
stable so that they may continue to grow and attract new particles (Mehrotra, Sastry et al. 1983).
A wealth of research has been conducted on the type of oil used for agglomeration;
however, results vary over researchers. Oils are divided into light and heavy classes depending
on viscosity. One set of research shows that oils with specific gravities between 0.7 and 0.85 are
most effective in agglomeration coals while oils with gravities below 0.64 or above 0.97 are
ineffective. The same research states that heavy oils are too viscous to disperse in the slurry
while light weight oils were unable to make the coal hydrophobic enough; however, this research
was later disputed when it was discovered that heavy oils could produce agglomerates and high
recoveries if long mixing times were used. Multiple researchers have confirmed that heavy oils
result in a higher combustible recovery, but also give a poor ash and sulfur rejection. In the
1970’s, it was discovered that oils with medium surface tension and medium viscosity were
most effective in agglomerating and that oils with lower or high viscosities and surface tensions
were ineffective at producing agglomerates. Due to such variations in research, an appropriate oil
can only be selected after the properties of the coal surface are known and through
experimentation to understand how the oil will interact with the surface (Mehrotra, Sastry et al.
1983).
2.4.3 Dewatering Agglomerates
The moisture content within the agglomerates comes from two sources: internally trapped
moisture and surface moisture. The internal moisture which heavily depends on the mixing
method, intensity, and oil dosage, commonly makes up 5 to 10% of the total agglomerate weight.
Therefore, most of the agglomerate moisture is on the surface of the agglomerate. For
agglomerates larger than about 2 mm, gravity drainage on a dewatering screen can reduce the
total agglomerate moisture to 10% by weight or less. Since most moisture comes from the
13 |
Virginia Tech | Chapter 3: Batch Testing Methods
Batch tests were conducted using four different methods to clean and dry coal: breaking
agglomerates via screening, breaking emulsions with ultrasonic energy, breaking agglomerates
with ultrasonic energy, and breaking agglomerates using vibrating mesh plates. This chapter is
divided into four subsections, one for each method. The subsection for each experimental
process is further to divided to include a description of the apparatus used for the process, the
methods followed, the results obtained, and a discussion of the results.
For each testing method, n-Pentane, produced by Alfa Aesar, was used to produce either
agglomerates or emulsions. Pentane is a colorless, immiscible liquid with a density of 0.631
g/cm3. The liquid and vapors are highly flammable. Pentane has a boiling point of 36oC. When
pentane enters the air and concentrations of 1.8% to 8.0% by volume are reached, an explosive
environment is created. The pentane used in the following experiments was HPLC grade and
was composed of a minimum of 99% pentane by volume (Alfa Aesar 2009). Additional supplies
and apparatuses used for each individual method will be discussed in the method’s individual
section.
15 |
Virginia Tech | 3.1 Breaking Agglomerates via Screening
3.1.1 Experimental Apparatus: Breaking agglomerates via screening
To form the agglomerates for the process, a Black and Decker kitchen blender was used.
The blade of the blender “cut” the agglomerates and was therefore replaced with a custom-made
flat paddle. A variable speed control was used with the blender so that both high and low shear
mixing environments could be created. To dewater and break agglomerates, 8 inch laboratory
sieves of varying apertures were used.
3.1.2 Experimental Methods: Breaking agglomerates via screening
To form the agglomerates, a volume of coal slurry was poured into the blender and mixed
on a high speed setting for 30 to 50 seconds after the pentane was added. The volume of pentane
added varied based on the requirements of each sample to form the agglomerates. Immediately
following the addition of the pentane, an obvious phase separation could be observed. Heavier,
ash-containing water remained in the lower portion of the blender while black, while less-dense
coal agglomerates rested on the top of the blender.
Larger agglomerates were desired to enhance the dewatering stage; therefore, the blender
speed was reduced using a variable speed controller to create a low-shear mixing environment.
The agglomerates were then mixed in the low shear environment for an additional 4.5 minutes or
until the total mixing time was five minutes. The long mixing time allowed large (approximately
0.5 cm diameter) agglomerates to form.
The agglomerates and ash-containing water were poured across a large-mesh screen
(ranging between ten and thirty mesh depending on agglomerate size) to dewater the
agglomerates. The dewatered agglomerates were then transferred to a larger mesh (smaller
opening) screen. Next, the screen was shaken by hand so that individual dry coal particles or
small groups of particles sheared off the agglomerates and fell through the screen and wet coal
remained on the screen. Shaking times were varied based on the amount of coal that fell through
the screen. Screens were shaken until the majority of original material fell through.
16 |
Virginia Tech | To further reduce the product moisture, the dry coal particles were placed on another,
smaller-opening screen and shaken. This step was repeated for one to two additional stages using
increasingly smaller screens than the previous stage. Depending on the moisture of the oversized
material, the oversized material was either added into the final product or recycled and used to
form new agglomerates.
3.1.3 Results: Breaking agglomerates via screening
Screening was originally tried on sample from the Blue Creek preparation facility. Size
analysis done on the sample showed that 71% of the sample was smaller than 500 Mesh.
Agglomerates were made via handshaking in a separatory funnel using a 6% solids (by weight)
slurry containing 39.1% ash. The agglomerates were dewatered by opening the valve at the
bottom of the funnel to remove ash. After dewatering, the ash of the product was reduced to
4.65%. The agglomerates were shaken by hand on 50, 80 and 100 mesh screens until sample
would no longer pass through the screen. By combining the -50 + 80, -80 + 100, and -100 x 0
size classes, the overall moisture was reduced to 2.10 % and cumulative ash was reduced to
3.88%. Table 3-1 shows the results obtained by screening the Blue Creek agglomerates. With
each new screen, the moisture of the particles passing continued to decrease. Figure 3-1 shows
the steady decrease as screening progressed.
Table 3-1: Results of agglomerating and screening Blue Creek Coal.
Screen Size, Mesh (micron) Dry Weight, g Moisture, % Ash, % Yield, % Recovery, %
+50 6.06 40.95% 4.65 - -
-50 + 80 0.20 10.00% -
-80 + 100 1.16 1.19% 3.85 50.36% 79.06%
-100 x 0 0.73 1.08% 3.96
Tails 8.40 - 74.31 - -
Feed 19.27 - 39.1 - -
17 |
Virginia Tech | Figure 3-1: The effect of screen size on moisture.
In an attempt to lower the moisture and increase the recovery, different screen
configurations were tried. In the next experiment, agglomerates were created in the blender using
flotation feed from CONSOL’s Bailey preparation plant. The agglomerates were dewatered on a
20 mesh screen, and then poured on a set of 50 and 70 mesh stacked screens. The screens were
shaken until no more material fell through the openings (due to screen blinding). Table 3-2
shows the results obtained from the experiments while Figure 3-2 shows a flow sheet of the
process followed. Approximately 22% (by weight) of the original feed remained on the 50 mesh
screen and had a moisture content of 21%. The material which passed through the screen fell
onto a 70 mesh screen. The 70 mesh screen was hand shaken until no material would pass the
screen. The majority of the material passed through the screen very quickly. After shaking,
nearly all of the material passed through. The material remaining on the screen had a very high
moisture content of 33%. The majority of the original feed (35.24% by weight) was able to pass
the screen and report to the -70 x 0 mesh size class. This material had the lowest moisture,
4.85%.
18 |
Virginia Tech | Table 3-3: Cumulative product ash, moisture, and combustible recovery for screening process using Bailey coal.
Dry Weight Cumulative Combustible
Sample Ash, %
Weight, g Percent, % Moisture, % Recovery, %
Screen
Lost 2.44 13.27% 14.56%
Losses
+50 M
Product -50 + 70 M 10.49 57.04% 3.88% 11.85% 78.04%
-70 x 0 M
Tailing Tails 5.46 29.69% 82.28% 7.40%
Feed Feed 18.39 100.00% 29.74%
A second experiment was tried using flotation feed CONSOL’s Bailey preparation plant.
Larger screen sizes were used to see if combustible recovery could be increased. The
agglomerates for this experiment were much larger in diameter than in the previous experiment.
The agglomerates were drained on a 40 mesh screen and then transferred to a 10 mesh screen.
The agglomerates were shaken until all coal particles had passed through the screen and no
material remained on top of the screen. Next, the coal particles were placed on a 45 mesh screen
for shaking. After just over half of the material had passed through the screen, shaking was
stopped because the +45 mesh material appeared dry. Moisture and ash analysis was conducted
on the 45 mesh overflow and underflow, as well as the material which remained plugged in the
screen openings. Analysis results are shown in Table 3-4. Figure 3-3 shows the generalized flow
sheet for the process with the results for each size class. Please note in Table 3-4 that the samples
containing “screen” in the name are the dry material which was left in the screen openings due to
blinding. The screens were dried weighed wet and then placed in the oven so that the moisture of
the product left on the screen could be calculated. After the screens were dry, each screen was
tapped over a sample pan to dislodge the blinded material.
20 |
Virginia Tech | Table 3-5: Screening results using Bailey sample after combining material to create final product.
Weight Average Cumulative Combustible
Sample Dry Weight, g
Percent, % Ash, % Moisture, % Recovery, %
Screen
Lost 6.33 18.23% - - 22.77%
Losses
-10 + 45 M
Product 17.39 50.07% 3.39% 2.36% 69.13%
-45 x 0 M
Tailing Tails 11.01 31.70% 82.02% - 8.10%
Feed Feed 34.73 100.00% 29.87% - -
Table 3-6: Screening results using Bailey sample with screen losses neglected.
Weight Cumulative Combustible
Sample Dry Weight, g Ash, %
Percent, % Moisture, % Recovery, %
-10 + 45 M
Product 17.39 61.23% 3.39% 2.39% 84.35%
-45 x 0 M
Tailing Tails 11.01 38.77% 82.02% - 9.94%
Feed Feed 28.4 100.00% 29.87% - -
to the tails. Nearly a quarter of the combustible material was lost due to screen blinding. As
shaking progressed, the screen openings plugged, leaving behind 18 % of the total material fed
into the system and 23% of the total combustible material in the system.
If screen losses are neglected (assuming a screen cleaning method would be available in a
continuous process), the combustible recovery increases by 15 percentage points to nearly 85%.
The percentage of combustible material in the tailings increases to 10%. Table 3-6 shows the
results of the experiment if the screen losses are neglected.
In the next experiment, Agglomerates were created with screen bowl effluent from the
Bailey preparation plant and screened using five different sieve sizes: 20 mesh for dewatering,
and 10, 45, 60, and 80 meshes for drying. More screen sizes were used than in previous
experiments to test the theory that more screening stages would decrease the product moisture. A
flow sheet of the procedure is shown in Figure 3-4. The smaller, 20 mesh sieve was used for
dewatering to ensure that no whole agglomerates passed through to the tailings. A large 10 mesh
screen was used as the first stage of drying. After shaking, no material was left on top of the
22 |
Virginia Tech | screen; however, 5.5% of the initial feed weight remained in between the mesh openings. The
underflow of the 10 mesh screen was poured onto a 45 mesh screen. All screens in the process
were shaken until material stopped passing through the openings. The 45 mesh screen became
completely plugged, preventing material from passing. Nearly 5.5 grams, or 24% of the feed by
weight, remained on the screen. The -45 mesh coal was transferred onto a 60 mesh screen and
then an 80 mesh screen. All of the material passed through both screens at a very fast rate. The
-10 + 45, -45 + 60, -60 + 80, -80 x 0 mesh materials were all combined to create the final
product. Table 3-7 shows the individual results for each step in the process. The first 5 rows,
which include the word “screen” in the sample name, represent the coal that was left on the
screen due to blinding.
Figure 3-4: Experimental flowsheet for multistage screening processes.
23 |
Virginia Tech | Table 3-7: Induvidual sizeclass results for multistage cleaning process.
Weight
Sample Dry Weight, g Percent, % Moisture, % Ash, %
20M screen 1.12 4.94% - 3.34
10M screen 1.24 5.47% 38.00% 3.58
45M screen 0.67 2.96% 22.99% -
60M screen 0.00 0.00% - -
80M screen 0.02 0.09% 0.00% -
-10 + 45M material 5.44 24.01% 17.58% 3.46
-45 + 60M material 0.1 0.44% 0.00% 8.14
-60 + 80M material 0.14 0.62% 6.67% 4.49
-80 x 0 M material 6.13 27.05% 5.26% 3.9
Tails 7.8 34.42% - 78.67
Feed 22.66 100.00% - 30.50
Based on the screen data, most of the moisture in the sample was lost on the 10 and 45
mesh screens. The material left in between the openings of the 10 mesh screen had a moisture
content of 38% while the moisture content of the blinded 45 mesh material was 23%. The -10 +
45 mesh material that fell in between these two screens had a moisture content of 17.5%. The 60
and 80 mesh screens had no material left between the openings and the material that passed the
screens had low moisture contents of 6.6% and 5.3% for the -60 and -80 mesh size classes,
respectively.
Though the -10 + 45 mesh size class had high moisture content, the large amount of dry
material that fell into the -80 x 0 mesh class was able to “balance” out the moisture contents so
that both materials could be used in the final product. The -45 + 60 and -60 + 80 mesh materials
were also combined into the final product. Cumulatively, the final product had a moisture value
of 11.3%. Table 3-8 shows the specifications of the product, screen losses, tailings, and feed.
Seventeen percent of the combustible material was lost on the screens, with the majority being
lost on the 20 and 10 mesh screens (Table 3-7). Ten percent of the combustible material passed
through the dewatering screen, into the tailings. Of the combustible material fed into the process,
72% reported to the final product. Using a recovery of 72%, the separation efficiency for the
process is 62.52%. If screen losses are neglected (assuming screens will be cleaned and the
material will be recovered), the combustible recovery increases to 87.23% and the separation
24 |
Virginia Tech | Table 3-8: Multistage screening results for cummulative product and tailings.
Dry Weight Average Cumulative Combustible
Sample
Weight, g Percent, % Ash, % Moisture, % Recovery, %
Screen
Lost 3.05 13.46% - - 17.24%
Losses
-10 + 45M
-45 +60 M
Product 11.81 52.12% 3.74% 11.34% 72.16%
-60 + 80 M
-80 x 0 M
Tailing Tails 7.8 34.42% 78.67% - 10.60%
Feed Feed 22.66 100.00% 30.50% - -
efficiency rises to 77.59%. Neglecting screen losses will also raise the combustible recovery of
the tailings to 12.77%. However, during the dewatering stage, any coal which falls through the
dewatering screen forms a thin layer on the surface of the tailings. This material could be
skimmed off and recycled into increase combustible recovery in the product and decrease
combustible recovery in the tailings.
Using multiple stages of screening seemed to have no effect on the product moisture.
Therefore, an experiment utilizing only two shaking stages was tried to see how the moisture
would be affected. A 30 mesh screen was used as the dewatering screen. In all previous
experiments, the dewatered agglomerates were removed from the screen so that the majority of
agglomerates were sent to the subsequent shaking stages. Most of the agglomerates directly
against the screen mesh have higher surface moisture content compared to agglomerates near the
top of the material on the screen. As the water from the upper levels of agglomerates drain
through the screen, some water collects along the bottom of the screen and between the
openings, wetting the agglomerates. These wetter agglomerates tend to “stick” to the screen due
to the surface tension between the water on the coal surface and the water on the screen.
In this experiment, the agglomerates created with Bailey flotation feed were dewatered,
and then the dewatering screen was tilted 90°. All of the agglomerates that fell from the screen
were placed onto the 50 mesh screen. Agglomerates that were stuck to the screen with water
were not manually removed from the screen, instead, they were assumed to be recycled back to
the agglomeration stage in a continuous circuit because of their higher moisture content. The
25 |
Virginia Tech | Figure 3-5: Process flow sheet for screening process with recycle stream used to dewater Bailey flotation feed.
agglomerates which went onto the 50 mesh screen were shaken until the screen blinded. The
material passing the screen fell onto a 70 mesh screen. A flow sheet for the process followed is
shown in Figure 3-5.
Table 3-9 shows the moisture, ash, and weight percent for each size class. The
agglomerates which did not freely fall from the angled 30 mesh dewatering screen were
manually removed for moisture and ash analysis. The +30 mesh agglomerates that were not
removed from the screen had a moisture content of 28.62% and comprised nearly 17% of the
original feed. The +50 mesh sample had a moisture content of 9.83% and was combined with the
-50 + 70 and -70 x 0 mesh samples which had moisture contents of 3.45% and 3.21%,
respectively, to create the final product. The specifications of the combined product, tailings,
feed, and internal recycle streams are shown in Table 3-10 .
26 |
Virginia Tech | 3.1.4 Discussion: Breaking agglomerates via screening
The screening method for breaking agglomerates can give a low moisture product;
however, the process is very sensitive and has many drawbacks. Separation efficiencies were
calculated for each of the screen configurations tested. The multiple stages of screening and the
coal losses due to blinding make the process very inefficient. The average separation efficiency
for all of the process was 66.3% with a standard deviation of 3.7%. If the screen losses are
neglected, assuming the coal can be recovered, the average separation efficiency increases to
78.3% with a standard deviation of 1.9%; however, the process is still less efficient than other
cleaning methods tested.
As mentioned, One of the largest problems with the process is screen blinding. As
agglomerates are shaken on the screen, particles begin to shear off the surface of the
agglomerate, exposing internal droplets of water. Some of the extremely small droplets evaporate
into the air. When using traditional stainless steel screen mesh, the water droplets too large to
evaporate will stretch across the screen opening and collect some loose particles until the screen
is completely blinded. If a large amount of water was present in the agglomerates, the water in
the screen openings would eventually become heavy enough to overcome the tension between
the opening and fall into the product below. The trapping of moisture in the screen mesh is
shown in Table 3-4. The material left in between the openings of the 45 mesh screen had 7%
moisture. The +45 mesh and -45 mesh materials had moistures of just over 2%.
A Teflon coated mesh was tried to prevent the water droplets from stretching across the
screen openings. The screen worked very well when shaking was first started. As shaking
continued and water droplets became exposed, the droplets started to roll around on top of the
screen with the agglomerates. Initially, the droplets would coalesce and look like a coal coated-
water drop. As shaking continued, the droplet picked up more coal particles and become thicker
in consistency. The water continued to pick up coal particles until everything on the screen was
coalesced into a sticky mass. In industry, this wet mass could be recycled and sent back to the
agglomeration tank. These blinding issues mean the process could never be truly continuous
using traditional sieves. Screens would have to go through cycles of shaking/breaking
agglomerates then be cleaned to remove dry material.
28 |
Virginia Tech | 3.2 Breaking emulsions with ultrasonic energy
3.2.1 Experimental Apparatus: Breaking emulsions with ultrasonic energy
During separation, the plugging of feed lines by emulsions was a common problem;
therefore, many different experimental setups were explored using the concept of breaking
emulsions with ultrasonic energy. In all setups, a Qsonica Q700 ultrasonic probe operating at
20kHz was inserted into a separation column and used to break the emulsions. Different methods
of creating and feeding the emulsions into the separation column were tried in an effort to
prevent lines from plugging. The following sections will discuss each experimental setup in
further detail.
3.2.1.1 Experimental Apparatus: Emulsion formation with a traditional
mixer
In the initial experimental setup shown in Figure 3-6, two Masterflex peristaltic pumps
were used to pump feed slurry and pentane into an Oster kitchen blender. A variable speed
controller was used with the blender to control the rotational speed of the blender’s mixing
paddle. Two ports were located on the side of the blender pitcher. A lower port was used for
incoming feed while the upper port was used as an overflow to the separation column.
The ultrasonic probe was mounted on the bottom of a custom-made, 1.5 inch diameter, 9
inch long, glass separatory column. Two ports were located at the bottom of the column directly
above the tip of the probe, one for feed, and one for tailings. Originally, three ports were at the
bottom of the column. Due to damage, one port was sealed off. Note that in Figure 3-6, the
original third port is being used for feed and is located on the back side of the column.
30 |
Virginia Tech | Separatory
Column
Column
Feed
Pump
Pentane/Coal
Feed Pumps
Mixer
Ultrasonic
Probe
Figure 3-7 : Experiemental setup consisting of feed pumps, kitchen blender for emulsion formation,
column feed pumo, and separatory column.
3.2.1.2 Experimental Apparatus: Emulsion formation with static mixer
In the second experimental configuration, shown in Figure 3-8, two peristaltic pumps
were used to feed coal slurry and pentane into the mixing system. The two feed lines joined into
a single line and were directed into a mixing loop. A 1/16 hp chemical resistant, magnetically
coupled, pump was used to pump slurry and pentane across a 0.25 inch in-line static mixer. The
rest of loop was comprised of 0.375 inch steel tubing and 0.625 inch hosing. A tube was
connected to a 0.25 inch opening at the top of the static mixer so that a small amount of material
could exit the loop, while the majority continued mixing. The same separation column and
ultrasonic probe used with in section 3.2.2.1 was used in this experiment. Feed leaving the
mixing loop entered the column through one of the two bottom ports. The tailings port was
attached to an overflow system so that the tailings would exit the column at a rate necessary to
maintain the pentane-water interface at a constant level.
32 |
Virginia Tech | Ultrasonic
Mixer
Probe
Overflow
Static
Coal / Pentane
Mixer
Feed
Concentrate
Figure 3-9: Experiemental setup consisting of static mixing system and top-mounted ultrasonic probe.
3.2.2 Experimental Methods: Breaking emulsions with ultrasonic energy
3.2.2.1 Methods: Emulsion formation with a traditional mixer
In the initial experimental setup shown in Figure 3-6, coal and pentane were pumped into
an Oster kitchen blender at equal rates. The blender operated at high speed in order to form the
emulsions. Initially, the blender operated at 2 minutes before opening the overflow port. The
mixture of emulsions and water overflowed the blender from a side port and fed into the
separation column via gravity.
Feed entered the column through one port while the second port was used to drain off
tailings. After the initial feed entered the column, the feed port was closed for two minutes to
allow for initial emulsion breakage. The feed pumps were turned off during the initial two
minutes of operation to prevent a buildup of material in the mixer. The pumps were turned back
on after the feed port was opened and the initial two minute sonication period had passed. As
emulsions continued to enter the column, pentane and coal overflowed the weir and was
collected for drying. In initial tests, coal was filtered on filter paper under the fume hood. In later
34 |
Virginia Tech | tests, the coal/pentane concentrate was evaporated in the evaporation circuit of the continuous
unit discussed in Chapter 4.
3.2.2.2 Methods: Emulsion formation with static mixer
Initially, the coal and pentane feed pumps were turned on to allow the system to fill. The
feed valve on the column remained closed during this time. Once the pump and pipes comprising
the mixing loop were full, the centrifugal pump was turned on and the feed pumps were turned
off. The centrifugal pump ran for two to three minutes, giving time for the static mixer to form
emulsions from the slurry and pentane.
After initial mixing, the feed valve to the column was opened and the slurry and pentane
feed pumps were turned back on. Slurry entered the column at the same flow rate which fresh
pentane and coal were being pumped into the mixing system. A gravity overflow system was
added to the overflow valve. The height of the t-joint in the overflow system controlled the
height of the pentane-water interface. This prevented the system operator from having to control
the level manually by sporadically opening and closing the overflow valve.
As with the previous process, new feed entering the column caused the dispersed coal in
the pentane phase to overflow. The dispersed coal was poured over a filter paper so that pentane
could drain out of the coal. The tailings from the column were also filtered and dried.
3.2.3 Results: Breaking emulsions with ultrasonic energy
3.2.3.1 Results: Emulsion formation with Laboratory Mixer
The ultrasonic probe was consistently able to dewater the emulsions so that the final
product had less than 2.0% moisture. The mean, standard deviation and confidence interval were
calculated for the moisture values obtained. The moisture values were converted to a normalized
Z value. Any samples more than four standard deviations away from the mean moisture value
were discarded from the data set. The average moisture value obtained using the data in Table
3-11 was 0.96% ± 0.15% at a confidence interval of 50%. Please note that two samples do not
have moisture values listed. For these samples, so little sample was collected that it was
impossible for the scale to accurately read the change in weight after the sample was removed
35 |
Virginia Tech | Table 3-11: Results obtained by breaking emulsions in the ultrasonic separator.
Product Separation Ash
Feed Ash Tailing Recovery
Sample Moisture Efficiency Rejection
(%) Ash (%) Ash (%) (%)
(%) (%) (%)
Flotation Feed 9.91% 86.24% 3.30% 1.32% 98.80% 68.14% 68.14%
Flotation Feed 41.11% 82.66% 3.36% 0.28% 85.98% 81.70% 81.70%
Flotation Feed 42.73% 81.86% 3.27% 1.30% 84.10% 80.29% 80.29%
Flotation Feed 36.65% 84.43% 3.94% - 90.32% 83.63% 83.63%
Screen Bowl
19.22% 73.06% 3.38% - 92.42% 78.83% 78.83%
Effluent
from the oven; therefore, the moisture content could not be calculated. These samples were not
factored into statistical moisture calculations, but were included in recovery calculations.
The recovery values ranged between 84.10% and 98.80% with an average of 90.25%.
The separation efficiency increased compared to the agglomerates screening method. The
separation efficiency averaged 78.50%. Many experiments were performed to attempt to break
the emulsions; however, throughput of the system was so low that for many tests there was not
enough dry coal collected to properly measure the moisture and recovery.
3.2.3.2 Results: Emulsion formation with static mixer
Experiments tried using the static mixer were largely unsuccessful. Initially, as sample
circulated through the mixing loop, emulsions exited through the discharge port at the top of the
loop. As the material in the mixing loop continued to circulate, the discharge rate of the
emulsions began to slow down. After running in the range of 5-10 minutes, emulsions
completely stopped exiting the discharge port and only pentane was leaving the loop. As the
pump continued to run, the coal plugged the static mixer and water traveled through the mixer
via capillaries formed through the coal.
The plugging of the mixer could be prevented by drastically decreasing the feed ratio of
coal to pentane. When the coal ratio was decreased, there was insufficient coal in the system to
feed the separation column. As the coal in the system built up, the same plugging problem would
be encountered, though the mixer was able to operate for a longer time period before plugging.
36 |
Virginia Tech | For coal that entered the separation column with the bottom mounted probe, the
ultrasonic probe did a very poor job of breaking the emulsions. The concentration of coal in the
pentane phase overflowing was extremely low, so that the overflow was practically clear. To
combat this, a column with a flat bottom was fabricated and the probe was mounted downward,
through the top of the column. It was theorized that the waves from the probe would reverberate
off the bottom of the column, helping to better suspend coal particles. As the emulsions entered
the column, the waves from the probe forced the particles downward. The pentane phase
remained completely clear. As new emulsions entered the column, the pentane phase
overflowed. After all of the pentane had left the column, a mixture of water and emulsions began
to overflow. The energy from the probe caused the emulsions to remix with the tailings and
form new oil-in-water emulsions. Figure 3-10a shows the remixing effect caused by the probe.
When the probe was bottom-mounted a distinct color separation between coal and tailings could
be seen (see Figure 3-6); however, with the top-mounted probe, the material in the column is
uniform in color. Figure 3-10b shows an up close view of the emulsions overflowing the column
during operation.
Figure 3-10: (a) Picture showing the remixing of the sample caused by the ultrasonic
probe. (b) Picture showing wet emulsions overflowing the separatory column
37 |
Virginia Tech | 3.2.4 Discussion: Breaking emulsions with ultrasonic energy
Using the ultrasonic probe to break emulsions created a low moisture product at a high
recovery; however, the process was inefficient in terms of throughput. The percentage of solids
in the overflow was very low, less than 0.5%. The low solids concentration could be attributed to
the ultrasonic probe. Instead of breaking the emulsions and sending all of the surface coal to the
pentane phase, it is believed that the vibration waves from the probe actually split the majority of
emulsions, so that one emulsion would be divided into multiple emulsions by the waves. Any
coal into the pentane phase was “thrown” off the surface of the emulsion when the emulsion was
split.
The rate of emulsion breakage was extremely slow. As the probe operated and new
emulsions fed into the column, the thickness of the emulsions layer within the column
continually increased. Eventually, emulsions would fill almost the entire length of column. The
amount of time necessary for the emulsions to fill the column depended on the feed rate into the
column, but the build-up of the emulsion layer was observed in every test, regardless of feed
rate.
Another operating issue encountered in the process was heating of the probe tip. The
ultrasonic probe operated by vibrating the steel tip at a frequency of 20 kHz, causing the tip to
heat. The probe was always placed in the water phase to prevent pentane from exploding;
however, after approximately 15 minutes of operation, the probe would generate enough heat to
heat the water and cause the pentane layer to boil. After operating approximately 20 minutes,
large cavities could be seen forming at the tip of the probe and the glass column was warm to the
touch. It is believed that the water directly against the tip was hot enough to boil, causing the
cavities. At this point, the tip of the probe was very hot. Experiments were never conducted for
longer than 25 minutes due to the heat.
38 |
Virginia Tech | 3.3 Breaking agglomerates with ultrasonic energy
3.3.1 Experimental Apparatus: Breaking agglomerates with ultrasonic energy
3.3.1.1 Experimental Apparatus: Preliminary Testing
Due to the low throughput and inability to continuously run the emulsion process
described in Section 3.2, other coal/hydrophobic liquid products that could be fed into the
separatory column were explored. Once again the kitchen blender was used to create the
agglomerates for the process. The ultrasonic probe was mounted at the bottom of a 1.75 inch
diameter, 5 inch tall glass separatory column with a single port at the top of the column for
product overflow. To overflow the coal/pentane product, liquid pentane was manually poured
into the top of the column by hand. The tailings were removed from the column using an
overflow system designed to maintain the interface at a constant height within the column.
3.3.1.2 Experimental Apparatus: Moisture Testing
Tests were performed in a custom made, 1.5 inch diameter, separatory column. The
ultrasonic probe mounted to the bottom of the column so that the distance between the tip of the
probe and the overflow spout was 3 inches. A tube entering the top of the column was connected
to a peristaltic pump set to pump pentane into the column at a rate of 48 mL/min. Figure 3-11
shows the experimental setup for the process.
39 |
Virginia Tech | Separatory
Column
Pentane
Feed
Pentane
Storage
Ultrasonic
Probe
Figure 3-11: Experimental setup used to determine the effect of interface distance
from probe tip on moisture.
3.3.2 Experimental Methods: Breaking agglomerates with ultrasonic energy
3.3.2.1 Experimental Methods: Preliminary Testing
Initially, the screening process followed in Section 3.1.2 was modified so that the
overflow agglomerates were poured into the ultrasonic separatory column to be cleaned and
dried. Figure 3-12 shows the flow sheet of the process followed. Coal agglomerates were made
by mixing 600mL of slurry (6% solids) from CONSOL’s Bailey preparation plant and 10mL of
hydrophobic liquid in a kitchen blender. For the initial 60 seconds of mixing, during which the
agglomerates formed, the blender was operated at a high speed to ensure a high shear mixing
environment. To increase the diameter of the agglomerates, the blender was turned down and
operated at low shear for 30 seconds. The agglomerates floated on top of the water and ash
phase. The blender contents were poured across a 30 mesh screen to remove water (the “Primary
Dewatering Screen” in Figure 3-12). The water stream was collected, dried, and assayed.
40 |
Virginia Tech | The agglomerates remaining on the screen were poured onto a second screen, smaller
mesh screen and shaken so that dry particles fell through the mesh (the “Secondary Dewatering
Screen” in Figure 3-12). For two of the three tests, a 50 Mesh screen was used as the secondary
dewatering screen. In the final test, the secondary dewatering screen had 70 mesh openings. The
screens were shaken until no new material fell through. The agglomerates remaining on the
screen were hand-fed into the ultrasonic 1.75 inch diameter by 5 inch separatory column used in
previous sections. Unlike emulsions which were fed into the column in the water phase,
agglomerates were poured into the top of the column, into the pentane phase.
Additional pentane was poured into the top of the column to cause the dispersed particles
to overflow the column. The pentane was separated from the coal through evaporation. The
column overflow was combined with the secondary dewatering screen underflow to compose the
final coal product.
Figure 3-12: Flow sheet of process combining screening and ultrasonic separation to clean and dry coal.
41 |
Virginia Tech | The majority of preliminary tests were conducted modifying the process described above
so that no secondary dewatering screen was used. Agglomerates were formed in the same
manner discussed above using flotation feed from Alpha Natural Resources’ Kingston
preparation plant and flotation feed from CONSOL’s Bailey preparation plant. After dewatering
the agglomerates on the primary dewatering screen, the separatory column and ultrasonic probe
described in the previous process were used to break the agglomerates. The column was filled
with a small volume of water so that the water level was approximately 1 inch above the probe
tip. The remainder of the column was filled with hydrophobic liquid. Agglomerates were
removed from the screen and dropped into the hydrophobic liquid phase. The agglomerates
broke up and coal dispersed into the hydrophobic liquid phase almost immediately. Additional
hydrophobic liquid was poured into the top of the column to overflow the liquid and coal. The
hydrophobic liquid was evaporated, leaving behind dry coal.
3.3.2.2 Experimental Methods: Moisture Testing
These experiments were focused on determining the effect of the interface distance from
the probe tip on the moisture, recovery, and percent solids in the final product. The distance of
the pentane/water interface from the probe tip was varied at 4 separate experimental distances.
The first sample was taken when the interface was 1 inch away from the probe tip subsequent
samples were taken at 0.5 inch increments until the interface distance was 2.5 inches away from
the probe’s tip.
Agglomerates were produced by mixing 500 mL of flotation feed from Alpha Natural
Resource’ Buchanan preparation plant with approximately 10mL of pentane. The sample was
agitated at high speed for 30 seconds then low speed for 30 seconds. After the agglomerates had
formed, the contents of the blender were poured across a 100 mesh screen and allowed to drain
for 10 minutes. The tailing material was dewatered and dried for ash analysis.
A portion of the agglomerates were weighed and dried so that an initial moisture value
could be calculated. The remaining agglomerates were weighed in aluminum dishes to serve as
the feed for the column. One gram of agglomerates was weighed into the first dish to serve as the
initial feed to the column. Half a gram of agglomerates were weighed into the remaining six
dishes.
42 |
Virginia Tech | For all experiments, the ultrasonic probe was set to operate at 15 amps. Initially, the
column was filled with enough water to reach the desired interface height. The remaining
column volume was filled with pentane until pentane just started to exit through the overflow
spout. The probe was turned on and the 1 gram sample was poured into the column. As soon as
the sample entered the column, a timer was started. After 30 seconds of operation, the pentane
pump was turned on and one of the 0.5 gram samples was added. Half a gram of agglomerates
were added every 30 seconds. The last sample was added at a running time of 3:30. After the last
sample was added, the pentane pump continued running for 2 minutes to allow the coal in the
system to continue to overflow. After a total running time of 5:30, the collection beaker was
removed from below the overflow spout and the pump and probe were turned off.
The total volume of sample collected was measured and the pentane/coal overflow was
poured into the evaporation chamber of the continuous testing circuit. The continuous circuit was
used to evaporate the excess pentane. After all pentane had evaporated and condensed, the coal
was removed from the chamber and weighed for moisture analysis.
3.3.3 Results: Breaking agglomerates with ultrasonic energy
3.3.3.1 Results: Preliminary Testing
For initial tests, the agglomerates were dewatered on a sieve, and then shaken on a screen
similar to the process used in section 3.1. Results of the process are shown in Table 3-12. All
tests used 30 mesh sieves to remove the tailings water from the agglomerates. The first two tests
used a 50 mesh sieve as the secondary dewatering screen, while Test 3 used a 70 mesh sieve. The
underflow moistures from the 50 mesh sieves were 16.53% and 15.68%, respectively. The
underflow of the 70 mesh screen was much dryer, containing only 2.63% moisture.
Table 3-12: Results obtained by screening agglomerates then dewatering overflow in ultrasonic separator.
Underflow Agglomerate Moisture Product Recovery,
Test
Moisture
Moisture Before Sonication After Sonication %
1 16.53 - 8.42 15.35 85.61%
2 15.68 30.98 1.01 13.97 84.09%
3 2.63 19.90 0.60 1.10 84.58%
43 |
Virginia Tech | The agglomerates which remained on top of the sieve were removed and poured into the
separatory column. The column was very effective at the reducing the agglomerate moisture. For
the second test, the agglomerates entering the column contained nearly 31% moisture while the
overflow material had a moisture content of 1%. The +70 mesh material had a moisture content
of nearly 20% when entering the column, and a product moisture of less than 1%.
The final product moisture was calculated by combining the screen underflow with the
overflow from the separatory column. The screen underflow comprised a high percentage of the
final product; therefore, the low moisture column overflow was unable to reduce the moisture for
the first two tests below the target 10% value. For the first two tests with high moisture
underflow samples, the total product moistures were approximately 15% and 14%, respectively.
The third test had very promising results, yielding a final product with containing 1.10%
moisture.
The recovery for all three tests averaged 84.76%. It was theorized that the low recovery
was due to screen loses and entrained ash present in the small water droplets carried through with
the screen underflow. Because the probe was so effective at reducing the moisture present in the
overflow agglomerates, it was determined that screening the agglomerates before sonication to
was not necessary. Therefore, no additional tests using this method were performed.
All other preliminary tests were performed by feeding dewatered and de-ashed
agglomerates into the separatory column with no pre-screening. Table 3-13 shows the results for
the individual preliminary tests performed using flotation feed from the Kingston preparation
facility. Tests 1 - 4 were used to calculate the moisture and recovery of the overall process;
therefore, the agglomerate moisture was not calculated in these initial tests. These tests were
performed in the 1.75 inch diameter, 5 inch tall column. Tests 5 - 14 were used to assess the
probe’s effectiveness at reducing the moisture and ash within the agglomerates using the 1.5 inch
diameter, 3 inch high column.
Pouring the agglomerates into the ultrasonic separatory column proved to be an
extremely effective cleaning process. The flotation feed contained an average of 54.6% ash.
After agglomeration, the ash was reduced to an average value of 7.13%. Pouring the dewatered
agglomerates into the separation column reduced the ash to below 5%. The process was
44 |
Virginia Tech | extremely consistent in its separation and produced high recovery rates. The average recovery in
the process was 94.09 % ± 1.22%. The average ash rejection of the process was 99.23% with a
standard deviation of 0.003%. The separation efficiency these tests were the highest off all the
other experiments conducted. The average separation efficiency was 93.33% with a standard
deviation of 0.04%.
The sonication process was also effective in creating a dry coal product. The largest
moisture reduction can be found in Test 5. The agglomerates fed into the separatory column
contained 58% moisture. After sonication, the sample contained 0.60% moisture. The average
agglomerate moisture in the testing data was 42.22%. This number was reduced over 38 times so
that the concentrate sample had an average moisture content of 1.10%. The histogram in Figure
3-13 shows that of the 14 tests run with the Kingston sample, 90% of the product samples had
moisture contents less than 1.75%. The products most frequently contained between 1.00% and
1.25% moisture. All of the product samples had moistures below 2.75%.
Due to slight differences in testing methods, the percent solids of the concentrate samples
are not included in Table 3-13; however, compared to breaking emulsions with the ultrasonic
probe, the column overflow contained a much higher percentage of coal. When breaking
agglomerates in the column, the percentage of solids in the overflow pentane was as high as
2.26% (compared to less than 0.5% when breaking emulsions). The percent solids depended
heavily on the flow rate of the overflow pentane fed into the column.
45 |
Virginia Tech | Table 3-14: Data obtained by breaking agglomerated CONSOL coal with ultrasonic energy.
Product Reject Recovery Separation Ash
Feed Ash % Product Moisture%
Ash % Ash % % Efficiency % Rejection %
41.11 1.15 3.06 86.94 89.94 88.45 98.51
41.11 8.42 2.43 84.19 87.30 75.65 88.35
41.11 0.60 3.36 86.97 90.01 89.24 99.23
41.11 1.01 3.60 87.41 90.43 89.11 98.68
Four preliminary tests were performed using flotation feed from CONSOL’s Bailey
Mine. The results are shown in Table 3-14. The average combustible recovery was 89.4%,
slightly lower than the Kingston sample. With the exception of one data point, the process was
able to reduce the moisture to below 2.0%. It is believed that the 8.42% moisture is due to
improper agglomerate breakage; however, the data was included in the set due to the limited
number of tests that were performed before sample. When excluding the high moisture point
from the data set, the average moisture content was 0.9%. The ultrasonic separator was not as
efficient at breaking the agglomerated Bailey coal as it was with the Kingston sample. With the
separation efficiency for the 8.42% moisture point excluded, the average separation efficiency
was 88.9% with a standard deviation of 0.35%
3.3.3.2 Results: Moisture Testing
The interface distance from the probe’s tip was varied so that samples were taken when
the interface was 1, 1.5, 2, and 2.5 inches from the tip, creating interface thicknesses of 2, 1.5, 1,
and 0.5 inches, respectively. Figure 3-14 shows the results for the moisture determination tests.
To calculate the data found in the plot, multiple tests were performed at each distance, and the
results were averaged.
As the distance from the tip increased (decreasing interface thickness), the product
moisture decreased. The total range of moistures was very narrow, ranging from an average low
moisture of 1.46% at 2.5 inches away from the tip to a high average moisture of 1.86% at 1.0
inch away from the tip. Agglomerates were fed into the top of the column. It is believed that as
the water droplets within the agglomerate fell through the pentane phase, heading towards the
interface, the upward vibrations produced by the ultrasonic probe “pushed” the droplets further
47 |
Virginia Tech | Figure 3-14: Plot showing the effect of interface distance from energy source and interface thickness on
moisture content, ash content, and recovery of final product.
upward in the pentane phase. It is believed that the cycle of the droplets falling, then getting
pushed upward by vibrations is responsible for the slight increase in moisture in the thicker
pentane layers. When the pentane layer is thinner, the droplets have less distance to travel and
are not impacted by the probe’s upward waves before reaching the water as many times as with a
thicker pentane layer.
While the moisture of the sample decreases with increasing interface distance, the percent
solids in the concentrate increases. When the pentane layer is thicker, there is a large volume
through which the hydrophobic particles can disperse. Therefore, since the same amount of coal
was added during each test, it is expected for the solids concentration to increase as the volume
of pentane decreases. The concentrate contained 1.05% solids when the pentane thickness was 2
inches. When the thickness of the level was decreased to 1 and 0.5 inches, the percent solids
increased to 1.64% and 1.63%, respectively. Thermodynamically, the coal will always prefer to
be displaced into the hydrophobic phase as long as there is enough volume of hydrophobic liquid
to accommodate all of the coal. It is believed that the percent solids will keep increasing, until
the layer is so thin that water is being thrown into pentane phase, causing the overflow to be
diluted and the moisture to increase.
48 |
Virginia Tech | Due to a malfunction of the ash analyzer, the ash contents of the concentrate and tailings
samples of the latter sets of experiments could not be measured. Before the equipment
malfunction, recovery rates and ash values were calculated for the first set of experiments and
then used to calculate the data in Figure 3-14. The effects of interface distance on recovery and
concentrate ash are show in Figure 3-15. As the interface distance from the probe increased, the
concentrate ash also increased; however, the increase was very slight. At a distance of 1 inch, the
concentrate contained 1.98%. At 2.5 inch, the ash content had increased to 2.09%. Overall, the
process produced an average concentrate moisture of 2.02% with a standard deviation of 0.06%.
The recovery of the sample peaked at distance of 1.5 inches; at which point the process
recovered 84.54% of the combustible material. Over the four tests, the process produced
consistent recoveries, ranging from 83.58% to 84.54%, depending on the interface level. The
average recovery was calculated to be 84.14% with a standard deviation of 0.45%.
2.25% 84.80%
84.60%
2.15%
84.40%
2.05%
84.20%
84.00%
1.95%
83.80%
1.85%
83.60%
1.75% 83.40%
0 0.5 1 1.5 2 2.5 3
Figure 3-15: The effect of interface distance on recovery and concentrate ash.
49
%
,hsA
etartnencoC
%
,yrevoceR
Concentrate Ash
Recovery
Interface Distance From Probe Tip, in. |
Virginia Tech | 3.3.4 Discussion: Breaking agglomerates with ultrasonic vibrations
Breaking agglomerates using ultrasonic vibrations was successful in both cleaning and
dewatering coal. The main advantage of using ultrasonic energy to break agglomerates over
using ultrasonic energy to break emulsions is higher throughput. The probe was able to break the
agglomerates at a much faster rate than the emulsions, causing more coal to be present in the
column overflow. As previously mentioned in section 3.2.4: Discussion, it was believed, through
visual observations, that the ultrasonic vibrations did not actually break the emulsions, but
instead, split the emulsions into smaller and smaller emulsions. However, when agglomerates
were introduced into the column, they broke immediately and coal dispersed throughout the
pentane phase. The pentane phase in the column turned black, with no transparency, compared to
being nearly clear at times when attempting to break emulsions.
While breaking the agglomerates was successful in drying the coal, the process also has
some drawbacks. As previously discussed in section 3.2.4: Discussion, heating of the ultrasonic
probe’s tip proved to be a problem. It was impossible to operate the column for longer than 25
minutes due to the heat generated by the probe. Heating caused pentane to evaporate at a faster
rate and eventually led to boiling in the water phase. Once the water began to boil, selectivity of
the process was greatly reduced as the boiling water caused water and wet particles to be
displaced into the pentane phase.
With most coal samples, emulsions began to form at the pentane/water interface after
running for a certain time period. Coal from Alpha Natural Resources’ Buchanan preparation
plant was more hydrophobic than other samples used, and no emulsions formed. For all other
samples, the emulsion layer which formed was very thin and did not seem to interfere with rate
of agglomerate breakage. However, the probe was unable to run for a sufficient time period (due
to overheating), to determine if the emulsion layer could eventually interfere with breakage.
Overall, the process was successful in meeting the goals of the project; however, one
main concern regarding the process is the scalability of the ultrasonic probe. The largest
ultrasonic probe which is currently being produced has a tip diameter of 2 inches, which will not
allow for the throughput required to have an industrial scale process. Other industrial options
which could be used in place of the ultrasonic probe were explored. One option considered
included using ultrasonic transducers outside of a horizontal pipe and feeding agglomerates and
50 |
Virginia Tech | 3.4.2 Experimental Methods: Breaking agglomerates with vibrating mesh
Before the testing began, water was poured into the lower portion of a glass column and
pentane was poured into the upper portion, creating an interface. The mesh discs were lowered
into the column. The discs were positioned so that the lower disc sat at the pentane/water
interface.
Before testing, a Ninja brand kitchen blender with a flat, three tiered paddle was used to
create agglomerates. Pentane dosages were determined based on the coal sample used to create
the agglomerates. Six hundred milliliters of coal slurry was poured into the blender. High shear
mixing was started immediately after 20 ml of pentane was added to the slurry. The pentane and
slurry were mixed for 30 seconds so that small, powder-like agglomerates could form. The
agglomerates were dewatered on an 80 mesh screen.
The mechanical shaker was turned on and set to operate at 30 Hz. The agglomerates were
removed from the screen on a laboratory spoonula and dropped into the top of the glass column.
Immediately, the agglomerates dispersed into the pentane phase and ash could be seen falling
into the water phase. The initial coal dispersion is shown in Figure 3-17. Pentane was slowly
poured into the top of the column to overflow the coal-pentane mixture. The process of adding
more coal particles and overflowing the column was repeated until enough product for ash
analysis had been collected. If needed, ports at the bottom of the column were used to drain the
ash containing water so that the interface could be maintained at the same level as the lower disc.
53 |
Virginia Tech | Pentane /
Water
Interface
Figure 3-17: Complete dispersion of coal in the
pentane phase (above the pentane/water
interface)
The coal/pentane product was then poured into the evaporation circuit of the continuous
separation unit discussed in Chapter 4: Continuous Testing Unit. The pentane was evaporated
and condensed into a separate chamber, leaving behind dry coal. The dry coal was removed from
the system and the final product moisture was calculated. Results from these experiments are
discussed in section 3.4.3.
3.4.3 Results: Breaking agglomerates with vibrating mesh
Breaking agglomerates with the vibrating mesh was successful in both cleaning the coal
and creating dry final product. The process produced results similar to the ultrasonic probe used
to break agglomerates. Flotation feed samples from both the Kingston preparation plant and the
Bailey preparation facility were used to test the process. Screen bowl effluent samples from
Bailey were also tested. Results of the experiments are shown in Table 3-15.
The Kingston sample produced an average recovery of 89.8%, approximately 5% lower
than the average recovery obtained by breaking agglomerates from the same sample with the
ultrasonic probe, with an average moisture content of 3.63%. The separation efficiency of
54 |
Virginia Tech | Table 3-15: Results obtained for Kingston and Bailey samples using vibrating mesh to break agglomerates.
Feed Tailings Concentrate Concentrate
Sample Recovery, %
Ash, % Ash, % Ash, % Moisture, %
Kingston Floatation Feed 51.00 90.00 3.56 4.20 88.87
Kingston Floatation Feed 52.60 91.60 4.01 3.20 90.10
Kingston Floatation Feed 52.60 91.40 3.77 3.50 89.87
Kingston Floatation Feed 52.60 91.70 3.73 3.60 90.26
Bailey Flotation Feed 44.90 87.10 5.50 1.10 88.70
Bailey Flotation Feed 44.90 87.70 4.60 0.70 89.20
Bailey Screen Bowl Effluent 40.40 87.60 4.14 3.80 90.91
Bailey Screen Bowl Effluent 40.40 87.90 4.79 3.90 91.18
86.67% was also lower than obtained with the ultrasonic probe. The samples from the Bailey
plant had an average recovery of 88.9% with a separation efficiency of 84.39%. Lower moisture
contents were obtained with the flotation feed compared to the screen bowl effluent. The screen
bowl effluent product moisture content was an average of 3.9% compared to an average of
0.90% for the flotation feed sample.
3.4.4 Discussion: Breaking agglomerates with vibrating mesh
Using the vibrating mesh to break agglomerates is a very effective process. The recovery
and moisture results obtained using the vibrating mesh plates are not as good as the results
obtained by breaking the agglomerates with the ultrasonic probe; however, the two processes are
very comparable. As previously mentioned, there is currently no large-scale ultrasonic device to
act as a substitute for the ultrasonic probe. The vibrating mesh plates are capable of being scaled
up for use in pilot-scale and industrial-scale processes, giving this method of agglomerate
dispersion a competitive advantage over the ultrasonic dispersion method.
The vibrating mesh separator tended to produce higher moisture contents than the
ultrasonic probe for the Kingston sample, but produced an equivalent moisture content for the
Bailey flotation feed sample. The screen bowl effluent was not tested in the ultrasonic separator;
55 |
Virginia Tech | therefore, the product moisture could not be compared. Though the Kingston product had higher
moisture, the average moisture values of 3.6%, 3.9%, and 5.1% obtained for the Kingston and
two Bailey samples, respectively, are well below the target 10% moisture and are acceptable
values for a final product.
The vibrating mesh produced a lower average recovery for the Kingston flotation feed by
nearly 5% compared to the ultrasonic probe. This lower recovery is most likely due to two
factors: sample age and coal loss through the dewatering screen. Though both samples came
from the Kingston preparation facility, the sample had aged for approximately 2 months before
being agglomerated for use with the vibrating mesh separator. Aging of the sample leads to
oxidation and decreased surface hydrophobicity. Particles with a high degree of decreased
hydrophobicity will not attach to agglomerates and will therefore fall through the agglomerate
dewatering screen. Any coal which passes through the screen will float on the surface of the
tailings water. Therefore, any sample lost through the screen could easily be skimmed from the
surface and recovered for additional conditioning.
Besides scalability, the vibrating mesh offers three other advantages over the ultrasonic
process: decreased operating cost, lower operating temperatures, and higher throughput. The
vibration generator used to move the discs uses less energy than the ultrasonic probe, resulting in
a decreased energy cost. Unlike the ultrasonic probe which has the potential to overheat the
sample in the separatory column, the vibrating mesh discs do not generate a significant amount
of heat and no sample-overheating issues were ever encountered; therefore, the risk of boiling
and subsequently displacing water particles into the product is greatly diminished. From a safety
aspect, the risk of creating a pentane explosion is also virtually eliminated.
56 |
Virginia Tech | low shear. No pentane was added to the mixing tank. The pump had a difficult time moving the
dry agglomerates; therefore, in some trials, the agglomerates formed in the blender were poured
into the tank. Emulsions or agglomerates in the mixing tank overflowed from a vertical pipe into
the ultrasonic separatory column.
The column was mounted on top of the ultrasonic probe used in previous experiments. A
cooling jacket was built around the column to prevent the pentane and water within from boiling.
An overflow port at the bottom of the column was attached to flexible hosing and a t-joint open
to the atmosphere to create interface level-control system used in previous experiments. Pentane
from the reagent tank was pumped into the column to overflow the suspended coal through a port
in the upper portion of the column. The overflow was pumped to a settling tank to give coal
particles time to settle out of the pentane phase.
The settling tank had two ports, an adjustable overflow port so that clear pentane at the
top of the settling tank could overflow into the reagent tank and an underflow port for thickened
coal to flow into the evaporator. A valve was placed on the underflow tube so that the operator
could control the rate at which thickened coal entered the evaporator. A double boiler was
utilized in evaporating the pentane. The ignition temperature of pentane is 260°C (Alfa Aesar
2009). To fabricate the double boiler, a large beaker of water was placed on a hot plate. The
sealed pentane evaporator was placed into the water. The maximum temperature the water would
reach was 100°C before boiling. This system ensured that pentane would never reach its ignition
point, only a maximum of 100°C. As the pentane evaporated, it traveled upward and was
condensed by two condensers. The liquid pentane fell into the pentane reagent tank. In order for
the pentane to evaporate and condense properly, a pump was used to pump the displaced gas
from the reagent tank into the evaporation beaker; otherwise, the pentane would not condense.
58 |
Virginia Tech | 4.2 Experimental Methods: Continuous Testing Unit
Separate experiments were performed using emulsions and agglomerates as the feed
material. Emulsions were formed by pumping coal slurry from the feed sump and pentane from
the reagent tank into the mixer. The mixer was operated at high speed to form the emulsions. The
level within the phase separator was controlled by hand using the overflow assembly discussed
in Chapter 3. Once the sample in the evaporator was completely dry (no more pentane was
condensing and falling into the reagent tank), the coal was removed and weighed for moisture
analysis.
Through experimentation, it was discovered that the sealed mixer built into the
continuous reactor was not capable of producing a high enough shear force to create the
agglomerates. Therefore, the agglomerates were created outside of the system in a kitchen
blender. Originally, it was intended to create the agglomerates via high shear mixing in the
blender, then pump the agglomerates into the built-in mixer for low shear mixing. Once the
pumps were turned on, it was discovered that the agglomerates collected at the mouth of the
pumps inlet tube, creating a filter, so that only ash and water passed. Therefore, it was necessary
to form the agglomerates in the blender, and pour the agglomerates into the mixer built-in to the
reactor, creating a semi continuous process as opposed to a true continuous experiment. The
pentane feed line from the reagent tank was rerouted so that it fed into the phase separator
instead of the mixer. As with the emulsion method, the pentane level within the separator was
controlled using the overflow system. Once the sample had completed drying, the moisture
content and recovery rate was then calculated.
4.3 Results: Continuous Testing Unit
Breaking emulsions in the continuous reactor was unsuccessful. The rate of breakage in
the phase separator was unable to keep up with the flow rate of material entering from the mixer.
After the flow rates of the feed were decreased, the phase separator was still extremely slow at
breaking the emulsions. The layer of pentane was nearly clear. At times, the feed was completely
turned off; however, in nearly all of the experiments performed, the phase separator completely
filled with emulsions. Eventually, the emulsions began to exit the separator through the overflow
59 |
Virginia Tech | Table 4-1: Agglomerate results from ultrasonic continuous reactor.
Product Separation Ash
Feed Reject Combustible
Sample Moisture Efficiency Rejection
Ash % Ash % Ash % Recovery %
% %
%
Bailey Screen
40.40 80.10 3.30 3.80 83.90 79.64 95.78
Bowl Effluent
Bailey Screen
40.40 81.40 5.00 5.50 85.50 78.90 93.36
Bowl Effluent
Bailey Screen
40.35 83.23 5.46 2.99 87.39 79.92 92.54
Bowl Effluent
Bailey Screen
40.35 80.10 4.97 5.51 84.29 77.77 93.48
Bowl Effluent
Kingston
50.70 92.00 3.60 2.90 91.40 88.08 96.68
Flotation Feed
system, into the tailings. Of the samples that were collected, there was not enough weight to
accurately measure the moisture content or recovery of the sample.
The reactor was very successful in creating dry product from the agglomerate feed. Table
4-1 shows the results obtained using flotation feed from CONSOL’s Bailey preparation plant as
feed. Due to limited sample, one test was performed using flotation feed from Alpha Natural
Resource’s Kingston preparation plant. The results from this single test are also included in
Table 4-1.
For the Bailey samples, the continuous experiments produced slightly higher moisture
values than in the batch tests. The average moisture content was 5.5%. Recovery and separation
efficiency values were also slightly lower than the batch tests, with an average recovery value of
85.3% and an average separation efficiency of 79.06%.
60 |
Virginia Tech | 4.4 Discussion: Continuous Testing Unit
The continuous reactor offered promising results for removing the moisture from the coal
agglomerates. Even when running the reactor with the agglomerate feed, emulsions still formed
at the pentane/water interface. The formation of emulsions was faster than the rate of breakage
and eventually, the column filled with emulsions.
The recovery was slightly lower than in the batch tests; however, a dewatering screen
was not utilized to separate the ash from the agglomerates. The addition of a dewatering screen
along with a recycle stream to recover any coal passing through the screen could help improve
the recovery. The lower reject ash is likely responsible for the lower recovery and it is expected
that the value was decreased due to coal emulsions exiting to the tailings through the column’s
overflow system. If a dewatering screen had been incorporated, the tailings from the column
could then be recycled with the skimmed layer of coal taken off the top of the tailings which
passed through the dewatering screen. This process would prevent coal leaving the reactor from
reporting to the tailings.
In Table 3-14 and Table 3-15, where agglomerates were being broken by the ultrasonic
probe and vibrating mesh, average ash values for the tailings were 86.4% and 87.4%,
respectively. In the continuous testing unit, the average ash value for the Bailey sample was
81.2%. If the separation efficiency could be improved so that the tailings were comprised of 86%
ash, the average combustible recovery would be increased to 89.7%, competitive to froth
flotation circuits.
While the average moisture content was slightly higher than that obtained in the batch
tests, it is believed that some modifications to the system could help decrease the moisture
content. Originally feed tubes going into the pentane phase were included on the ultrasonic
separator. However, the diameter of the tubes were too small for agglomerates to flow through
without plugging; thus, when the tubes were removed. While the mixer was feeding the
ultrasonic separator, some of the water entering the separator ran down the walls of the vessel. A
small amount of the water flowed into the concentrate overflow spout. Enlarging the feed tubes
to allow for agglomerates to flow without plugging will help decrease the moisture content in the
product.
61 |
Virginia Tech | Chapter 5: Conclusions and Recommendations
Breaking coal agglomerates with the ultrasonic probe and vibrating mesh yielded very
similar results, especially when testing the Bailey sample. Figure 5-1 and Figure 5-2, given on
pages 63 and 64, respectively, show the separation efficiencies for all of the tests conducted in
the present work. The black diagonal lines on the plot represent the different areas of separation
efficiencies. Breaking agglomerates formed with the Kingston flotation feed was the only
method to achieve a separation efficiency of over 90%.
All of the final products obtained using the vibrating mesh plates to break agglomerates show
separation efficiencies near 85%. Additionally, the Bailey agglomerates broken by the ultrasonic
probe also show separation efficiencies near 85%. The average product moistures obtained using
both breakage methods are approximately the same (2.05% and 3.63% average moisture for
Bailey and Kingston agglomerates broken with mesh, respectively, versus 2.80% average
moisture for Bailey agglomerates broken with the probe). The recovery values for the Bailey
samples are also similar: 89.4% with the ultrasonic probe, and 88.9% with the vibrating plates.
Therefore, it is determined that the vibrating mesh plates provide the best method for
breaking agglomerates and thus cleaning and dewatering fine coal. Since the results obtained
with both methods are about the same, the decision as which dispersion method is best may be
based on operational issues. The ultrasonic probe has four main drawbacks: overheating,
emulsion formation at the interface, limited throughput, and higher energy consumption. These
shortcomings can be overcome by using the vibrating mesh plates.
It is recommended that agglomerate dispersion by the vibrating mesh plates be further
explored. It is believed that the separation efficiency of the Kingston agglomerates broken with
the vibrating mesh was lower, in part, due to an older sample which had been slightly oxidized.
Additional tests should be performed on this sample to determine exactly why the separation
efficiency was nearly 10% lower than when using the ultrasonic probe to break the agglomerates.
Additional sensitivity analyses should be performed to determine how certain parameters such as
plate vibration frequency, plate position in regards to the pentane/water interface, agglomerate
feed size and moisture, and column feed rate effect the final product. These findings could then
be used to optimize the separator’s performance.
62 |
Virginia Tech | Nomenclature and Symbols
CFD – Computational Fluid Dynamics
DLVO – Derjaguin and Landau, Verwey and Overbeek
MIBC – Methyl Isobutyl Carbinol
PPG 400 – PolypropyleneGlycol 400
DOL– Degree of Liberation
QEM*SEM – Quantitative Evaluation ofMinerals by ScanningElectron Microscopy
d – Particle diameter
1
d – Bubble diameter
2
d –Collision diameter
12
d – Diameter of bubbles entering the froth phase
2-0
d – Diameter of bubbles at the top
2-f
E – Kinetic energy of attachment
k
E’ – Kinetic energy of detachment
k
h – Height of the froth
f
K – Hydrophobic force constant between the bubble and particle
132
K – Hydrophobic force constant between two particles
131
K – Hydrophobic force constant between two bubbles
232
m – Mass of the paticle
1
m – Mass of the bubble
2
n – Number of cells in the bank
N – Number of particles attached to each bubble
P – Probability of attachment
a
P – Probability of collision
c
P – Probability of detachment
d
P – Probability of bubble-particle aggregates transferring from the pulp to the froth
f
r – Radius of the particle
1
r – Radius of the bubble
2
R – Bank recovery
v |
Virginia Tech | 1 Introduction
1.1 General Introduction to Flotation
Froth flotation, often referred to simplyas ‘flotation,’was first used commercially in 1877
fortreating graphite orein Germany. Widespread use of the technology did not occur until the
turn of the twentieth century, and mineral production began to expand rapidly around mid-
century(Fuerstenau, Jameson, & Yoon, 2007). Froth flotation is a techniqueused to separate
materials based on differences in theirsurface properties. Originally developed as a mineral
processing technique, there are numerous otheruses offlotation. Today it is used to treat billions
oftons of materials yearlyin the mining, recycling, and wastewater treatment industries (Rubio,
Souza, & Smith, 2002).
In froth flotation, particles are selectively attached to air bubbles. These bubble-particle
aggregates riseto the surface of the flotation cell and are removed from the system. Particles are
selected to attach according to their level ofhydrophobicity, or fear of water. In theory, flotation
occurs onlywhen hydrophobic particles attach to air bubbles, while hydrophilic particles stay in
the system. In reality there are threemechanisms by which flotation occurs: attachment,
entrainment, and agglomeration. Bubble-particle attachment is the most important of the three
mechanisms, and it is the onlymode of recoverythat is selective. Entrainment occurs when
particles are recovered by entrapment in the water films formed between bubbles.
Agglomeration, sometimes called coagulation, takes place when small particles attach to each
other and act as a single larger particle. These agglomerates have the potential to trap
hydrophilic particles within them. Recovery by agglomeration is sometimes referred to as
entrapment
In the past, the flotation process was often modeled as a first-order process with a single
rate constant for therecovery processes occurring in both the pulp and froth phases of a flotation
cell. In effect, flotation was viewed as a single-phase process. However, the cell consists of two
distinctlydifferent phases, each havingdifferent mechanisms of particle recoveryand roles in the
production of a concentrate. More recently, flotation is modeled byconsideringthedifferences
between and determining therate constants for the pulp and froth zones.
1 |
Virginia Tech | Flotation cells arenearly always arranged in a series called abank. The product of one cell
becomes the feed to the next cell. This setup helps to give the particles not recovered in the first
few cells additional opportunities for recovery in rougher and scavenger cells and reduces the
effects of entrainment and entrapment, to achieve higher quality products at maximum
recoveries. The effect of entrainment can also be reduced through the use of froth wash water.
There aretwo basictypes of flotation cells: column and mechanical. Mechanical flotation
is performed under turbulent conditions in a stirred tank. Column flotation is performed under
relatively quiescent conditions in a tall narrow cell. Mechanical flotation cells are more common
than column cells because theycan process high tonnages and are more flexible. Column
flotation cells can achieve moreefficient separations, but arelimited by relatively low capacity
due to theirsmaller cross-sectional area.
As the name would suggest, mechanical cells are mechanically agitated by a spinning
rotor. The rotor generates turbulence, which serves as a mechanism for bubble-particle
collisions, particle suspension, and air dispersion. As can be seen in Figure1, mechanical cells
generally have a low height to diameter ratio and hence a large cross-sectional area to volume
ratio.
Figure1:Mechanical Flotation Cell
2 |
Virginia Tech | Slurry is generallyfed into the tank near the middle, and air is injected through the center
of theshaft that spins the rotor. Hydrophobic particles are recovered by a launder at the top of
the cell, while the tails exit the cell through a pipe at the bottom. In addition to the rotor, a stator
ring is often used to induce additional turbulence, and prevent the formation of a vortex within
the cell. Many cells also employ a beveled edge along the bottom of the cell. This forces
particles towards the rotor, and prevents the buildup of sediment on the cell floor.
Column flotation cells have a high height-to-diameter ratio, as shown in Figure2 on the
following page. Air is injected at the bottom of the cell, making use of either in-line mixers or
spargers to reduce the bubble size. The feed, introduced at the top of the cell, flows downward
while the air bubbles rise. This creates a mixing action and eliminates the need for a rotor to mix
the pulp. As with mechanicals cells, the froth overflows into a launder and the tails exit through
the bottom of the cell. Column cells typically use wash water to reduce the effects of
entrainment and improve the product grade(Finch, 1994).
Figure2:Column Flotation Cell
Both mechanical and column cells employ a series of chemical reagents to improve
flotation performance. The three main groups of reagents arefrothers, collectors, and modifiers.
Surfactants lower the surface tension of water and promote the buildup of a stable froth.
3 |
Virginia Tech | Collectors render the target minerals hydrophobic, enabling them to attach to air bubbles. There
are numerous varieties of collectors, which arebroken into two main groups; thiol-typeand non-
thiol-type reagents. Each collectoris tailored to treat a specific mineral or group of minerals.
Modifiers are categorized into activators, depressants, and pH regulators. These particular
additives are used eitherto enhance the adsorption of collectors or reduce the floatability of
undesirable minerals.
The mining industry utilizes flotation to upgrade the finest fraction of run-of mine(ROM)
ores. Particles that arelarger than 100 mesh can often berecovered by other methods, making
flotation most useful in the 10 to 150 µm range. In the U.S. coal industry, many companies
discard -44 µm materials due to the difficulty in floating and dewatering finer particles. Now
this fraction can be treated using flotation with advanced dewatering techniques, such as
hyperbaric centrifugation in thecoal industry(Keles, 2010).
The importance of improving flotation performance has become more pronounced as the
industry has expanded. An increase in performance as small as a fraction of a percent can have a
vast financial and environmental impact when billions of tons of material are treated each year.
Thepresent work focuses on the simulation of flotation, with the hope of improving the general
understanding of flotation and aiding in the advancement of flotation technology. These
simulations will focus on threeimportant industrial minerals: chalcopyrite, coal, and phosphate.
Copper is predominantly found in porphyry deposits containing < 1% copper. The
principal mineral being mined forcopperis chalcopyrite (CuFeS ), which contains 34.5% copper
2
when pure. As the grade of theoreis so low, it is often necessary to grind the ore to finersizes
in order to achievea high degree of liberation (DOL). Greater than 50% DOL is achieved at
particle sizes less than 100 mesh (Subrahmanyam & Forssberg, 1995). For such small particles,
flotation is regarded as the most efficient beneficiation method. Short chain collectors, such as
xanthates, are frequently used for the flotation of copper minerals.
Coal processing, also known as coal preparation, is used to increase heating value and
reduce transportation costs. Flotation is often used forthebeneficiation of fine coal. Coal is
naturally hydrophobic; therefore, it can often be floated without using a collector. In general, the
hydrophobicity of coal increases with rank and vitrinite content within the coal (Ding, 2009).
4 |
Virginia Tech | Much of theworld’s phosphate reserves arefound in sedimentary deposits. Nearly half of
theworld’s phosphate is cleaned using froth flotation to remove silicates, carbonates, and clays
from the ore. There are two methods of flotation commonly used in the phosphate industry:
direct and reverse flotation. Direct flotation is a one-step process in which the phosphate is
directly floated. Reverse flotation is a two-step process, in which gangue materials areremoved.
Long-chain fatty acids are ordinarily used as collectors for phosphates (Sis & Chander, 2003).
1.2 Literature Review
Although extensive research is being conducted on flotation, both in industry and
academia, most investigators agree that flotation is the least understood of all mineral processing
techniques. This lack of understanding stems from the large number of variables in flotation,
and the fact that it is a three-phase process which is difficult to model mathematically. It is
widely believed that flotation may be described as a first order process in the pulp, and often in
the froth as well (Fichera & Chudacek, 1992). Flotation models generally consist of a bubble-
particle collision rate term and aprobability of flotation term.
The theoretical basis forthe collision term in the majority of current flotation models is
Abrahamson’s model for the collision rate of small particles in a turbulent fluid:
[1]
= 5 +
whereZ is thecollision rate between two types of particles, N and N are the number
12 1 2
concentrations of two types of particles, d is the sum of colliding particle radii, and and
12
are the root-mean square (RMS) velocities of the particles and bubbles. This model assumes that
particles move randomly under infinite Stokes number conditions as if no other particles are
present and that theirvelocities are fully independent of each other(Abrahamson, 1975).
The advances made in computingenergy dissipation rates duringthe past decade have
made it feasible for researchers to develop models of flotation cells using computational fluid
dynamics (CFD). As it applies to flotation, CFD is used to determine local turbulence by
breaking the flotation cell into many small, finite volume elements. These local turbulence
values are then utilized to determine collision, attachment, and detachment rates in the cell by
using a flotation model (Koh & Schwarz, 2006). Koh and Schwarz found that the volume
5 |
Virginia Tech | neighboring the impeller and stator is more turbulent than in the bulk of thepulp, the collision
probability decreases as particle size decreases due to streamlining effects, and the attachment
rate decreases as particle size decreases (Koh & Schwarz, 2003).
Schubert found that it is impossible to attain optimum hydrodynamics for all particle sizes
simultaneously. Thespecificpower input must be minimized while maintaining particle
suspension for optimum flotation ofcoarse particles, while a much higher power input is
required for optimum flotation offine particles (Schubert, 1999). Yang and Aldrich found that
the water recoveryincreases with aeration rate and power input. It was also found that solids
entrainment in the froth can be linearly related to the water recovery, independent of aeration rate
and power input (Yang & Aldrich, 2006).
The mineralogical characteristics of particles also play a role in flotation performance.
Lastra(2007), Savassi (2006), and Sutherland (1989)all found that mineral liberation has an
effect on particle recovery. Through the use ofquantitative evaluation of minerals by scanning
electron microscopy (QEM*SEM), these researchers were able to determine that particles with a
higher degree of liberation had a greater probability of ending up in the flotation concentrate.
QEM*SEM and other SEM techniques have become an important tool, allowing mineral
processors to better characterize their flotation feeds rapidly.
Thecomplexity of the flotation process has severely limited the availability ofworkable
flotation simulators. Whilea vast amount of work has gone into understanding the fundamental
principles of flotation, relativelyfew researchers have taken on the task of developing aflotation
simulator, and even fewer have accomplished the goal of producing a useful product. The
simulators that are available in industry and from the literature include JKSimFloat, USIM PAC,
MODSIM, and asimulator developed at University of Petrosani, Romania.
JKSimFloat is a commercially available flotation simulation program that is under
continuing development. The simulator is a collaborative effort between the Julius Kruttschnitt
Mineral Research Centre (JKMRC) at the University of Queensland, the University of Cape
Town, and McGill University(JKTech Pty Ltd, 2010). JKSimFloat allows users to input data for
each object on the flow sheet, and solves the flow sheet using a mass balance. Thebasis for the
model behind JKSimFloat is the following equation:
6 |
Virginia Tech | [2]
= ∙ ∙
whereP is the ore floatability, Sb is the bubble surface area flux, and Rf is the loss in recovery
due to the froth phase (Harris, Runge, Whiten, & Morrison, 2002). The ore floatability is
determined by either the Distributed Property Floatability Component Model (DPFC-Model) or
the Empirical Floatability Component Model (EFC-Model). These approaches involve
determining the floatability of a mineral experimentally, and using the results as an input to the
simulator. The drawback of this model is that it is dependent on the collection of good
experimental data from operating plants. The floatability here refers to size-by-size and class-by-
class flotation recoveries. The term class refers to degree of liberation, which increases with
decreasing particle size. Many researchers using models of this form also include a ¼ term in
Eq. [2].
USIM PAC© is a steady state simulator that can model over 100 unit operations, including
both column and mechanical flotation machines. It is currently used in industry fordata
reconciliation, plant simulation and design, flow sheet development, and cost estimation (Metso
Minerals). The USIM PAC© simulator contains two separate models for flotation. The first
model takes an approach that uses three sub-populations for each mineral, i.e., non-floating, fast-
floating, and slow-floating components, each have their own rate constants. The second model
uses a distribution of rate constants that are dependent on particle size:
[3]
.
.
= 1− −
Wherex is the average size in fraction i, a is an adjustment parameter, xl is thelargest floating
i j j
size for mineral j, and xe is the easiest floating particle size for mineral j (Villeneuve,
j
Guillaneau, & Durance, 1995). The model also accounts for the effects of entrainment of
particles in the froth based on water recovery. Like JKSimFloat, this simulator requires the
collection of large amounts of accurate experimental data.
TheMODSIM mineral processing simulator can simulate crushing, classification, flotation
and several other processes. MODSIM is based on a population balance model and can account
for changes in size and mineral liberation. The model used for flotation simulation is a form of
7 |
Virginia Tech | the distributed rateconstant model. The simulator provides integrated flow sheets, allowing the
user to easily simulate an entire processing plant (Mineral Technologies).
The simulator developed at the University of Petrosani is used to simulate flotation
circuits. It is apparent that the simulator uses both population and mass balances to design flow
sheets. However, the usefulness and accuracy of this tool cannot be verified becausethe details
of the model that drives the simulator are not revealed in the literature (Samoila & Marcu, 2010).
1.3 Flotation Model
A comprehensive flotation model was developed by Do. The model is derived from first
principles and is used as the basis for the flotation simulator described in Chapter 2. The model
accounts for both the surface chemistryand hydrodynamic properties of the system, which
allows it to predict real world flotation results (Do, 2010). Unlike themodels used in the
flotation simulators described in Section 1.2, this flotation model does not require the input of
experimental data. Thekey advantage of this model is that it can predict flotation results from
contact angle, particle diameter, and ζ -potential which arethekey parameters affection flotation.
This section presents the keyanalytical equations of the model.
In order for flotation to occur, a bubble must collide with a particle, attach to it, enter the
froth phase, and then overflow to a launder without becoming detached. Therate of flotation, k,
can be given as follows,
[4]
in which Z is th= e c− ollision frequency given in units of s-1 and P is a probability of flotation
12
which has no unit. Thus, khas a unit of s-1.
The collision frequencyis determined by usingtheAbrahamson’s equation [1] for random
collisions:
[5]
/ /
= 2 +
whereZ is the collision frequency between particles and bubbles, N is the number of particles,
12 1
N is the number of bubbles, d is the sum of radii ofone bubble and one particlewhich is
2 12
8 |
Virginia Tech | referred to as collision diameter, and and are the RMS velocities of the particles and
bubbles, respectively.
The diameter of the bubbles generated in the cell is calculated using the following
relationship derived by Schulze(1984):
[6]
.
.
.
=
where γ is the surface tension of thewaterin a flotation cell, ρ is the density of the water, and
lv 3
ε is the energy dissipation ratein the bubble generation zone. It has been reported that the high
b
energy zone around the impeller and statortypically has a dissipation rate of 5-30 times larger
than the mean (Schulze, 1984). In the present work it is assumed that the energy dissipation rate
in the bubble generation zone is approximately 15 times larger than the mean energy dissipation
rate in the cell.
The RMS velocity of the particles is calculated using the following relationship:
[7]
/ / /
/
= 0.4
whereε is the energy dissipation rate, d is the particle diameter, ν is the kinematic viscosity of
1
water, ρ is the density of the particle, and ρ is the density of water (Schubert, 1999).
1 3
The bubble RMS velocity is calculated using the equation derived by Lee and Erickson:
[8]
/
/
= (ε )
whereC is a constant given as 2 and d is the bubble diameter (Lee & Erickson, 1987).
0 2
Thetotal probability of flotation, P, is given by
[9]
= (1− )
whereP is the probability of attachment, P is the probability of collision, P is the probability
a c d
of detachment, and P is the probability of bubble-particle aggregates transferring from the pulp
f
phaseto the froth phase.
9 |
Virginia Tech | The probability of attachment is calculated as follows,
[10]
= exp
whereE is the energy barrier as calculated usingtheextended DLVO theory, and E is the
1 k
kinetic energy of attachment (Yoon & Mao, 1996).
The extended DLVO theory states that
[11]
= + +
whereV is the electrostatic interaction energy, V is the van der Waals dispersion force, and V
E D H
is the hydrophobic force.
The electrostatic interaction energycan be obtained from the following relation,
[12]
ζ ζ ζ ζ
= ζ ζ ln +ln(1+ )
where ϵ is the permittivity in a vacuum, ϵis the dielectric constant of the medium, ζ is theζ -
0 1
potential of the particle,ζ is ζ -potential of the bubble, κ is the inverse Debye length, and H is the
2
separation distance between the bubble and particle (Hogg, Healy, & Fuerstenau, 1966)and (Do,
2010).
The van der Waals dispersion energycan becalculated using the following relationship,
[13]
= − ( ) 1− /
whereA is the Hamaker constant for the bubble-particle interaction in the medium, b and l are
132
characterization parameters for the materials involved, and cis the speed of light (Rabinovich &
Churaev, 1979).
The hydrophobic force can be expressed as:
[14]
= − ( )
10 |
Virginia Tech | whereK is the hydrophobic force constant between the bubble and particle (Rabinovich &
132
Churaev, 1979), which can beobtained using the following relationship
[15]
=
whereK is the hydrophobic force constant between two particles and K is the hydrophobic
131 232
force constant between two bubbles (Yoon, Flinn, & Rabinovich, 1997). This relationship, which
is referred to as geometric mean combining rule has recently been proven in wetting film studies
(Pan and Yoon, 2010).
The hydrophobic force constant between two particles may be found by:
[16]
=
wherea and b are fitting parameters shown in Table1 (Pazhianur & Yoon, 2003). The
k
hydrophobic force constant between two bubbles is 2.5x10-18 (Do, 2010).
Table1:Fitting Parameters forK
131
Θ a b
k
> 92.28° 6.327x10-27 0.2127
92.28°>θ > 86.89° 4.888x10-44 0.6441
< 86.89° 2.732x10-21 0.04136
The kinetic energy of attachment is calculated using the followingrelation,
[17]
= 0.5
where m is the mass of the particle, and U is the velocity of a particle approaching a bubble at
1 Hc
the critical rupture distance. This velocity may be found by the following equation:
[18]
=
whereβ is the drag coefficient in the boundary layer of the bubble (Goren & O'Neill, 1971).
The drag coefficient may be expressed as (Luttrell & Yoon, 1992):
[19]
.
= 0.37
11 |
Virginia Tech | which has been derived from the Reynolds lubrication theory.
The probability of collision equation used in this model was derived byLuttrell and
Yoon, and modified to ensure the probability may not be greater than 1 (Do, 2010),
[20]
.
= tanh 1+ .
whereReis the Reynolds number (Weber & Paddock, 1983).
Probability of detachment, as suggested by Yoon and Mao, is given as follows:
[21]
= exp
where W is the work of adhesion (a function of contact angle), and E’ is the kinetic energy of
a k
detachment (Yoon & Mao, 1996). The work of adhesion can beobtained by:
[22]
= γ π (1−cos )
where γ is the surface tension of water, r is the radius of the particle, and θ is the angle of
lv 1
contact between water and the particle(Yoon & Mao, 1996). The kinetic energy for detachment
is calculated using the following equation (Do, 2010):
[23]
′ = 0.5 ( + ) /
where is the energy dissipation rate and is the kinematic viscosity.
The probability of bubble particle aggregates transferring to the froth phase accounts for
instances in which the aggregates may bounce off the pulp-froth interface:
[24]
= (1− )
whereP is the probability that the aggregate will remain at the interface after bouncing n times,
i
and P is the probability of aggregate rupture. The first term, P, is represented as:
r i
12 |
Virginia Tech | [25]
= 13
where μ is the dynamic viscosity of water. The second term, P is calculated by:
r
[26]
= −
whereE is the kinetic energy transferred to the bubble-particle aggregate by the motion of the
iw
pulp-froth interface, and E is the kinetic energy of the aggregate after bouncing off the interface
ka
(Do, 2010).
The kinetic energy that is transferred to the bubble-particle aggregate can be expressed as:
[27]
/
=
whereg is the acceleration of gravityand ν is the kinematic viscosity of water (Sanada,
Watanabe, & Fukano, 2005).
The kinetic energy of the bubble-particle aggregate after bouncing is determined using
the following equation:
[28]
=
where m is the mass of the bubble (Do, 2010).
2
Fractional recovery(R ) ofparticles in the pulp phaseis given by
c
[29]
= 1−(1+ )
where kis the flotation rate constant in the pulp phase, as determined from the preceding
equations, and t is the retention timeof the particles within the pulp.
As discussed in the foregoing section, the fractional recovery(R)in the froth phaseis the
f
sum oftherecovery by attachment and therecovery by entrainment,
13 |
Virginia Tech | [30]
.
= +
whered is thediameter of thebubbles entering the froth phase, d is the diameter of the
2-0 2-f
bubbles at the top, N is the number of particles attached to each bubble, h is the froth height, R
f w
is the maximum theoretical water recovery, ρ is the density of water, and ρ is the particle
3 1
density. The first term of Eq. [30] represents therecoverydue to attachment, while the second
term represents therecoverydue to entrainment (Do, 2010).
Entrainment is closely related to the water recovery, which can becalculated usingthe
followingrelation,
[31]
=
where is thevolumetric flow rate of air leaving the cell and is equal to thesuperficial gas
velocity, is thevolume flow rate of pulp entering the cell, and is thefraction of water in
the froth phase (Kelley, Do, Keles, Luttrell, & Yoon, 2011).
Assuming that the flotation rateis constant over all the cells in abank, recoveryfor the
bank can be expressed by:
[32]
= 1− 1−
whereR is the pulp or collection zone recovery, R is the froth zone recovery, and n is the
c f
number of cells in the bank (Finch & Dobby, 1990).
1.4 Research Objective
The objective of the research presented in this communication is to develop a software tool
that can easilyand accuratelysimulate froth flotation based on the model presented in Section
1.3. This tool will be used to improve upon theunderstanding ofthe fundamental principles of
flotation, allowingresearchers to improve flotation processes and design more efficient flotation
machines.
14 |
Virginia Tech | 2 Simulator Development
2.1 Introduction
Theaim of this research is to develop a user-friendly simulation tool for predicting
flotation recovery. This was accomplished using the programming language Visual Basic in
conjunction with Microsoft Visual Studio 2008. Since the model is based on first principles, the
simulator has predictive and diagnosticcapabilities, differentiating it from other currently
available flotation simulators that are based on empirical models. The following sections will
detail the development of theflotation simulator called SimuFloat and provide an overview of
how the program functions.
2.2 Physical Parameters
The simulator requires theinput ofa number ofhydrodynamicoperatingparameters,
including specific power, superficial gas rate, particle specific gravity,particle size distribution,
air fraction (air holdup), slurry fraction (% solids), and froth height. The user may also elect to
manually input abubble size distributionif it is known. Use of a bubble size distribution can be
beneficial in achieving effective simulations. Other physical parameters that effect recovery
such as number of cells and retention time per cell are also needed to perform the simulation.
These parameters, along with the chemical parameters discussed below, represent the majority of
the operator-controlled flotation parameters in a flotation plant.
2.3 Chemical Parameters
SimuFloat requires user input of contact angle, type of frother(surfactant), frother
concentration, and particle zeta()-potential. Bubble-potential, permittivity of air, and
dielectric constant ofthe medium are given in the simulator by default but may be changed if
necessary. The user may also elect not to select a frother, in which case the surface tension of
pure water will be used in the calculations.
Contact angle determines the strength of the bubble-particle attachment, as defined in [14]
and [20]. A greater concentration of collector in the system will render the particles more
hydrophobic, thus increasing the probability of attachment and the work of adhesion.
15 |
Virginia Tech | 60
40
)
V 20
m
(
al
ti 0
n
e
t
o
p
--20
-40
-60
2 4 6 8 10 12
pH
Figure 3: -potential vs. pH. This figure shows the general effect of pH on the
ζ -potential but is not specific to any one mineral.
Plots showing the -potential versus pH for common minerals treated by flotation may be
available in the literature. However, discussion of such plots is beyond the scope of this work, as
SimuFloat does not yet allow for input of pH.
2.4 Simulator Overview
Themain input form for SimuFloat is shown in Figure4. The user may input the desired
parameters on the left hand side of the form, and all simulation outputs are shown on the right
hand side of the form. Greyed input fields and buttons may be enabled by placing a check in the
adjacent checkbox. For the user’s convenience, some input fields contain preset values for
properties of water or air at or near23 °C. These values can change with large changes in
temperature, and may be modified by the user. The simulator allows for the input of both single
component and multi-component feeds.
17 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.