Professional Documents
Culture Documents
net/publication/301975838
CITATIONS READS
0 536
4 authors, including:
Corby Anderson
Colorado School of Mines
461 PUBLICATIONS 986 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Corby Anderson on 06 May 2016.
05/02/2016
Executive Summary
The crushing and grinding circuit in mining is the most energy intensive process in a
mining project. Improvements in predicting the energy required to crush a given material, and
therefore the cost of comminution, would help to mitigate cost overruns in mining. The Bond
Work Test is an empirical way to predict this energy cost. Sepor has developed a mill that is
supposed to more accurately predict the Bond Work Index of any given material.
The new Sepor mill was tested against an older BICO mill with two reference materials,
one used by Dr. Husni Usman for his thesis and the other one was from JKTech in Australia and
was sent around the world to compare the Bond Work Indices. The Bond Work Test was run in
duplicate on each mill for each material. The Bond Work Indices from Dr. Usman’s material
were calculated to be, 13.39 kWh/mt and 14.87 kWh/mt for the BICO and Sepor mills
respectively. The Bond Work Indices for the Australian material were calculated to be 13.01
kWh/mt and 14.71 kWh/mt for the BICO and Sepor mills respectively.
A student-t test was calculated for each set of tests with a 95% confidence interval. The
data from Dr. Usman’s material suggested that there was no statistical difference between the
two mills, but the data from the Australian material suggested that there was. The data from the
Sepor mill was still lower than the world average, which was 15.39 kWh/mt, but was closer than
the BICO mill.
An economic analysis was conducted and it was found that the calculated Bond Work
Indices resulted in large discrepancies in capital and operating cost. The difference between the
values from the BICO and Sepor mills from the Australian material resulted in a $5.1 million
dollar difference in power operational costs for a year for 80,000 tonnes feed per day.
This underestimation of the BICO mill could have a disastrous effect on predicted costs
in mining projects. While the Sepor mill reported a value that was still lower than the world
average, it more accurately predicted the Bond Work Index than the BICO mill. It is
recommended that a larger test program be conducted in order to refine the averages of the two
machines and increase the certainty that the two machines produce a statistically different result.
Table of Contents
Executive Summary ........................................................................................................................ 2
Figures and Tables .......................................................................................................................... 4
1.0 Introduction ............................................................................................................................... 5
2.0 Background ............................................................................................................................... 5
2.1 Bond Work Index .................................................................................................................. 5
2.2 Experimental Material ........................................................................................................... 8
2.3 Equipment ............................................................................................................................. 9
3.0 Scope ....................................................................................................................................... 11
4.0 Experimental Methods ............................................................................................................ 12
5.0 Results and Discussion ........................................................................................................... 15
5.1 Usman Aggregate ................................................................................................................ 15
5.2 Australian Material .............................................................................................................. 18
6.0 Economic Analysis ................................................................................................................. 21
7.0 Conclusion .............................................................................................................................. 26
Appendix A: Ball Charge Analysis............................................................................................... 27
Appendix B: Usman Material Feed Size Analysis ....................................................................... 29
Appendix C: Australian Material Feed Size Analysis .................................................................. 34
Appendix D: Husni Usman Material BICO Mill Tests ................................................................ 40
Appendix E: Husni Usman Material Sepor Mill Tests ................................................................. 43
Appendix F: Husni Usman Material BICO Mill Product Size Analysis ...................................... 46
Appendix G: Husni Usman Material Sepor Mill Product Size Analysis ...................................... 50
Appendix H: Statistical Analysis of Dr. Husni Usman’s Material ............................................... 54
Appendix I: Australian Material BICO Mill Tests ....................................................................... 55
Appendix J: Australian Material Sepor Mill Tests ....................................................................... 58
Appendix K: Australian Material BICO Mill Product Size Analysis ........................................... 61
Appendix L: Australian Material Sepor Mill Product Size Analysis ........................................... 64
Appendix M: Statistical Analysis of Australian Material ............................................................. 67
Appendix N: Economic Analysis.................................................................................................. 68
References ..................................................................................................................................... 74
4
Table 1: The size and weight analysis of the BICO ball mill charge. .......................................... 12
Table 2: The size and weight analysis of the SEPOR ball mill charge......................................... 13
Table 3: The results of the feed size analysis for both materials. ................................................. 13
Table 4: The Bond Work Index data for Usman’s material using the BICO mill. ....................... 15
Table 5: The Bond Work Index data for Husni Usman’s material using the Sepor mill. ............. 16
Table 6: Summary of results for t-test assuming unequal variance. ............................................ 16
Table 7: The t-test for the two samples assuming equal variance. ............................................... 17
Table 8: The Bond Work Index data for the Australian material using the BICO mill. ............... 18
Table 9: The Bond Work Index data for the Australian material using the Sepor mill. ............... 19
Table 10: The summary of the results from the t-test of the two samples assuming equal
variance. ........................................................................................................................................ 19
Table 11: The summary of the results from the t-test of the two samples. ................................... 20
Table 12: The variation in a simple payback period in years with over or underestimates of
CAPEX and OPEX models. .......................................................................................................... 21
Table 13: The power operational cost for each of the test runs from the different mills based on
20, 5000, and 80000 tonnes of feed per day. ................................................................................ 23
Table 14: The capital cost estimates of ball mills for various size plant scales using the test
results from the BICO, Sepor and Round Robin. ......................................................................... 25
5
1.0 Introduction
The purpose of this investigation is to compare the BICO ball mill to the Sepor ball mill
in regards to the measurement of Bond Work Indices. Both ball mills serve the purpose of
measuring the energy required to grind a material to a particular size, usually a target of 150
microns, according to the empirical test based on Fred Bond’s Third Theory of Comminution.
Using this test procedure, a previously characterized aggregate material and a reference material
from a global test set were tested using the two ball mills to determine the Bond Work Index of
the materials. The results of these tests were then analyzed using statistical methods to
investigate any differences between the experimental results. These results were also analyzed
using economics to see how the accuracy of the Bond Work Index can affect the capital and
2.0 Background
2.1 Bond Work Index
The Bond Work Index is a measure of how much energy is required to grind a given
material down to a set size. This material property is useful in determining energy costs and
capital requirements for comminution circuits, namely the grinding portions such as the ball mill
According to a survey by EY, “Mining projects are showing an average budget overrun
of a staggering 62%.” [2] One of the main causes of this problem is inadequate planning. The
determination of the Bond Work Index is part of the planning stage and can greatly affect the
projected cost associated with a grinding circuit. As shown in Figure 2, the grinding stage in
mining is one of the most energy intensive and also is an area with the largest energy saving
opportunity. [3] An accurate Bond Work Index is essential for the proper design and estimation
Figure 2: A graph depicting the top ten energy intensive processes and their energy saving opportunities
within the mining industry. [3]
Improvements in the accuracy of the Bond Work Index value allow for more accurate
engineering design criteria for grinding circuits which will lead to reductions in cost overruns in
The Bond Grindability Test is a procedure that attempts to quantify the amount of energy
required to grind a particular material. Based off of Fred Bond’s Third Theory of Comminution,
the test procedure produces an index, in units of kWh/mt, which describes the amount of energy
required to grind rock from a theoretically infinite size down to 150 microns. [4] Using a smooth
ball mill and a specified ball charge distribution, the test’s empirical procedure attempts to
While the Bond Work Index and procedure have become a world standard for measuring
the energy required to grind material and designing grinding circuits, the test is highly empirical
and can vary significantly as a result. According to one study, the Bond Work Index measured by
three different mills that were side by side varied as low as 4.2% and up to 8.6%.[5] Even
variations within the distribution of the ball charge will result in differences between 2.9% and
8
6.9 %. [5] Even though there are difficulties in reproducing Bond Work Indices and there is
variability in the results, the quantity is one of the better measures of the energy required in a
44.5
𝑊𝑖 =
10 10
𝑃10.23 ∗ 𝐺𝑃𝑅 0.82 [( 0.5 ) − ( 0.5 )]
𝑃80 𝐹80
(1)
P1 is the 100% passing size of the product in microns, GPR is the average
grams/revolution of the last three cycles, P80 is the 80% passing size of the product in microns
and F80 is the 80% passing size of the feed material in microns.
material was an aggregate used in a previous doctoral study by Dr. Husni Usman. During his
thesis, he measured the Bond Work Index of the material using the BICO mill used in this
investigation. Using a similar test procedure, Dr. Usman determined that the Bond Work Index
for the aggregate was 13.03 kWh/mt. [6] His material needed to be pre-crushed to >3.36mm.
In addition to this aggregate, material was secured from JKTech labs as a part of a global
investigation into the reproducibility of Bond Work Indices. This material was sent out to
various labs around the world in order to collect a wide set of data concerning the Bond Work
Index of the material. Upon the writing of this report, the full study had not been completed.
However, the preliminary results from the study were released in order to allow for comparison
2.3 Equipment
A roll crusher was used to crush Dr. Usman’s material to >3.36mm. Once the material
was crushed a Jones splitter was used to get representative samples. The Jones splitter was also
used later in the investigation to split the product material; it is shown in Figure 3. A Ro-Tap,
pictured in Figure 4, was used to separate the product material from the material that still needed
to be crushed. The Ro-Tap was also used to characterize the feed material and the product
material.
Figure 3: The Jones splitter used to increment the samples of the crushed material.
Figure 4: A Ro-Tap used for size analysis of the feed, crushed and product materials.
10
The Bond Work Index procedure calls for the use of a 30.5 cm x 30.5 cm smooth ball
mill, such that no part of the mill may act as a lifter, which would increase the energy imparted
on the ore. The BICO machine used in this experiment is shown in Figure 5. As shown in Figure
6, the barrel of the BICO is not particularly smooth and the seam along the back protrudes from
the surface. The Sepor machine used measures 30 cm by 30 cm and is shown in Figure 7. From
Figure 8, the Sepor machine has a much smoother barrel, ensuring that nothing no part of the
Figure 6: The inside of the BICO ball mill, showing the protrusions that could act as lifters.
11
3.0 Scope
As mentioned previously, the Bond Work Index is currently one of the most used factors
to determine the energy required for an ore to be ground to a set size and this factor is used in
many facets of grinding circuit design. Due to its importance in design, it is equally important
that the tools used to measure the value, the Bond ball mills, should adhere to the original
procedure as much as possible. The Sepor machine claims that it adheres more closely to the
12
ideal Bond procedure through the careful design of its mill and thus produces a more accurate
value for the Bond Work Index. In particular, it is suggested that the Sepor machine produces a
higher, but more accurate Bond Work Index, as it does not have any surfaces that may serve as
lifters. This investigation looks to compare the Bond Work Indices for two sets of material, the
aggregate from Dr. Usman’s thesis and the reference material from JKTech. By comparing the
values produced by the two machines, statistical tests, namely the Student t-test will be used in
order to determine if the indices produced are statistically different. Using the Bond Work
Indices found through these experiments, a capital and operating cost model will be built using
the values from the BICO and Sepor machines in order to highlight the economic implications of
Bond Procedure was followed as closely as possible. Unfortunately, the prescribed number of
balls and the total weight of the ball charge were unable to be met simultaneously, given the ball
charge that was provided. Professor Erik Spiller stated that the weight mattered the most in the
test, so the prescribed weight of 20.125 kg was obtained as closely as possible. [7] A size and
weight analysis of the two charges can be found in Tables 1 and 2(See Appendix A).
Table 1: The size and weight analysis of the BICO ball mill charge.
Table 2: The size and weight analysis of the SEPOR ball mill charge.
The material used by Dr. Usman needed to be crushed down to under 3.36 mm (<6 Tyler
Mesh) before the Bond Work Index test could be conducted. This was done by roll crushing in
stages and screening at the end of each stage to eliminate the material that was already under
3.36 mm. Once the material was under the prescribed size, the material was split using a Riffle
Type Jones Splitter. The material was run through several times to make sure that is was all
homogenous for testing. The material was split into 200-500 g increments. The Australian
material was already passing 3.36 mm, and had been pre-split, so this process was unnecessary
Both materials’ weights of 700 cc needed to be determined for the test. This was done by
placing the material in a graduated cylinder and packing it down. A size analysis of the feed
material also needed to be run before the Bond Work Index test could be run. The 80% passing
size and the percent of the product in the feed needed to be found. Three size analyses were run
for each material (See Appendix B and C); the results can be seen in Table 3.
Table 3: The results of the feed size analysis for both materials.
The material was placed into the ball mill and run, initially, for 100 revolutions. The
material is then removed and placed into a Ro-Tap to separate the product, -100 Mesh, from the
remaining material. The oversized material is weighed and new material is added to match the
initial weight of the sample. This completes one cycle for the test. A spreadsheet of the data
needed for the test was tabulated so that the grams per revolution could be analyzed and the
number of revolutions required for the next cycle could be calculated. The full Bond Work Index
Test can take anywhere from 5-9 cycles. This process was done until closure was achieved.
There were two conditions for closure from the literature. The first was from Prof. Erik
Spiller and stated, “A reversal in the grams of undersize product per revolution; a reversal is a
trend of up/down/up (e.g. 1.89, 1.88, and 1.90) or down/up/down (e.g. 1.75, 1.71, 1.72) in the
last three cycles.” Also, there must be less than a three percent difference between the highest
and lowest values. [8] The condition for closure that we used was from a procedure from Phillips
Enterprises LLC. and stated, “The reversal value should be within 3% of the previous number.”
[9]
Once the test achieved closure a composite of the last three cycles’ products was formed
and weighed out to around 250 g. This was wet screened with 400 mesh then dried. The dried
sample was then Ro-Tapped to form a product size distribution to calculate the 80% passing size
of the product. This number was used in the final Bond Work Index Calculation.
15
to determine its Bond Work Index using both mills (See Appendix D and E). The P80 of the
product was determined using polynomial plots (See Appendix F and G). The results of each test
Table 4: The Bond Work Index data for Usman’s material using the BICO mill.
The values produced by the BICO for the Usman material compare well with the value
Dr. Usman reported in his thesis. The average difference between the measured Bond Work
Index and the Usman value is 0.35 kWh/mt, a difference of 2.7%. Any differences may have
been the result of using a difference charge distribution than the one used by Usman to determine
the Bond Work Index for the aggregate. This suggests relatively good reproducibility for the
Table 5 below summarizes the test results using the Sepor mill. From the average value
for the Bond Work Index, the Sepor machine produced a value higher than that of the BICO mill
for the same material. The average value for the Sepor machine was 1.48 kWh/mt more, a
difference of 9.9%. As the ball charge was comparable between the BICO and the Sepor
machines and the procedure was the same for both machines, the difference between the two
values would appear to be the result of an inherent difference between the machines.
16
Table 5: The Bond Work Index data for Husni Usman’s material using the Sepor mill.
While there is a difference between the two averages, one must ask if this difference is
statistically significant enough to confirm that the two averages are, in fact, different. In order to
test this, a Student t-test was performed, first assuming equal variances between the machines
and the second assuming unequal variances between the machines. The t-test was performed
using both assumptions concerning the variance of the populations as it was uncertain that the
variance of each of the machines was equal. However, considering the sample variances, it may
be suggested that they are relatively equal and, as a result, the t-test using equal variances may be
Using the null hypothesis that the difference between the population mean for the BICO
and the population mean for the Sepor was zero, Table 6 was generated to summarize the t-test,
Calculated t-stat t-Critical P(T<=t) Low Error Bound High Error Bound
-3.347 4.303 0.07884 0.708 2.242
From this t statistic, it is not possible to reject the null hypothesis. As a result, it cannot be
said that the measures of the Bond Work Index are statistically different between the machines.
17
While it is not possible to reject the null hypothesis outright, bounds can be put on the range into
which the difference between the mean lies. Using the above t statistic and equation 2 below
[10], the difference between the means may range from 0.708 and 2.24 kWh/t.
𝑠12 𝑠2 2 𝑠12 𝑠2 2
x̅1 − x̅2 − 𝑡𝛼,𝑣 √ + ≤ 𝜇1 − 𝜇2 ≤ x̅1 − x̅2 + 𝑡𝛼,𝑣 √ +
2 𝑛1 𝑛2 2 𝑛1 𝑛2
(2)
Considering the same null hypothesis as previously described, Table 7 summarizes the t-
test for the Usman data assuming a 95% confidence interval and equal variances.
Table 7: The t-test for the two samples assuming equal variance.
Calculated t-stat t-Critical P(T<=t) Low Error Bound High Error Bound
-3.347 4.303 0.07884 0 3.37
Similarly to the t-test assuming equal variances, the t statistic from this test does not
allow for the rejection of the null hypothesis. However, using equation 3 [10] below, bounds may
be set on the differences between the means. The difference in means may statistically range
1 1 1 1
x̅1 − x̅2 − 𝑡𝛼,𝑛+𝑛
𝑠 √
−2 𝑝
+ ≤ 𝜇1 − 𝜇2 ≤ x̅1 − x̅2 + 𝑡𝛼,𝑛 +𝑛 −2𝑠𝑝 √ +
2 1 2 𝑛1 𝑛2 2 1 2 𝑛1 𝑛2
(3)
Statistically, it cannot be said that there is a difference between the two data sets.
However, there is an apparent difference between the Bond Work Indices of these two Bond
mills. Such a difference could have significant effects on any future design based on them. More
18
data points using the two mills should be collected in order to better refine the respective
determine the Bond Work Index using both the BICO and Sepor mills (See Appendix I and J).
The P80 of the product was determined by a polynomial plot (See Appendix K and L). Table 8
below summarizes the test results from the Australian material measured using the BICO mill.
Table 8: The Bond Work Index data for the Australian material using the BICO mill.
When compared to the global average value of 15.44 kWh/mt for the Australian material,
the BICO produced a value 2.43 kWh/t less, a difference of 15.7%. This is a substantial
difference between the global averages but still sits just outside of one standard deviation from
the mean. The differences between these measures and the global mean may be due to
differences in laboratory procedure, the ball charges used, and differences between machines
Table 9 below summarizes the test data of the Australian material measured using the
Sepor mill.
19
Table 9: The Bond Work Index data for the Australian material using the Sepor mill.
Comparing the average value of these tests to the global average, the Sepor machine is
much closer. In magnitude, the Sepor measurement is 0.73 kWh/mt less than the average, a
difference of 4.7%. As with the BICO differences between the global average and the Sepor
machine may be due to procedural differences, the ball charge distributions used, and the
As with the Usman material, student t-tests were performed on the data in order to assess
the statistical difference between the two data sets. Using the null hypothesis that there is no
difference between the data sets, the t-tests were run assuming the same variance and non-equal
variances. Table 10 below summarizes the t-test assuming equal variances. (See Appendix M).
Table 10: The summary of the results from the t-test of the two samples assuming equal variance.
Calculated t-stat t-Critical P(T<=t) Low Error Bound High Error Bound
-7.298 4.303 0.0183 0.71 2.71
As the t statistic is larger than the critical value for the two-tail t-test, it can be said that the two
populations are statistically different. Based on this result, it can be said that the Sepor machine
produces a value that is higher than and statistically different from the BICO. As all other
variables were controlled as much as possible between tests in the two machines, this may
suggest that such a difference may be attributed to the machine itself. The higher value produced
by the Sepor machine may be the result of a lack of any surface that could act as a lifter. Without
20
any lifters, it would require more energy to grind the material to the same size as a machine that
did experience the advantage of lifters. Using equation 3 previously described, the bounds on the
difference in the means ranges from 0.71 to 2.71 kWh/mt between the machines.
Table 11 below summarizes a t-test assuming that the variances between the two
Table 11: The summary of the results from the t-test of the two samples.
Calculated t-stat t-Critical P(T<=t) Low Error Bound High Error Bound
-7.298 12.71 0.0867 0.84 2.58
Based on the assumption of unequal variances, the t statistic produced is not greater than the
critical value. As a result, it is not possible to reject the null hypothesis; it cannot be said that the
two data sets are statistically different. Using equation 2 previously described, bounds on the
difference in the means may be set, ranging from 0.84 to 2.58 kWh/mt.
In summary, this investigation must consider the differences between the machines both
qualitatively and quantitatively. Qualitatively, the Sepor machine attempts to hold to the ideal
Bond mill through its higher quality construction. The completely smooth barrel does not allow
for a lifter to be formed that would otherwise influence the energy required to grind the material.
The BICO machine, on the other hand, has design issues that deviate from the Bond procedure
and may produce a value that is inaccurate as a result. While machines such as the BICO have
been the standard models used in Bond Grindability Tests in the past, the values produced may
have deviated from values found in the field and values suggested by Bond’s Third Theory of
Comminution.
As to the quantitative analysis, the t-test of equal variance will be considered as the
differences between the sample variances were relatively small. Based on these test, the Usman
21
material does not suggest that there is a statistical difference between the two mills. However,
considering the Australian material, there is a statistically significant difference between the two
data sets. From this result, it may be said that the Sepor mill produces a statistically different and
higher value for the Bond Work Index. Such a difference has significant implications concerning
economic evaluation can be done to determine capital (CAPEX) and operational (OPEX) costs.
Creating accurate CAPEX and OPEX models early in feasibility studies for projects can allow
for more accurate values for net present value, rate of returns and simple paybacks. (See
Appendix N)
Operating costs for concentrating are typically set about 13% too low which can greatly
affect the net present value, rate of returns and simple payback for the project. [11] For example,
a project with correct CAPEX and OPEX estimations, a simple payback period would be about 6
years. If the OPEX estimation for this project was low by 13%, Table 12 shows the payback
Table 12: The variation in a simple payback period in years with over or underestimates of CAPEX and
OPEX models.
OPEX models for mining related projects typically include costs from crushing, grinding
and floatation of the material. This is can be broken down into many categories such as, “process
labor”, “maintenance labor”, “process power, propane and fuel”, “supplies (operating and
maintenance)” and “general and administration.” [13] Figure 9 shows about 25% of the operating
costs are the power to run the machinery and of this, grinding is a substantial amount. This is
why there is a large focus on the Bond Work Index and the energy required to grind a specific
$6,922,054 $3,727,000
$2,288,000 Process Labor
$2,389,000
Maintenance Labor
Figure 9: Annual operating cost summary for Morrison Copper/Gold Project based on processing 30,000
tonnes of feed per day. [13]
By using the calculated kWh/t values from the BWI, the cost of energy used per year can
be calculated for each of the tested mills and materials. Table 13 shows the estimated $/year
spent on power for a 20, 5000 and 80000 tonnes feed per day mills.
23
Table 13: The power operational cost for each of the test runs from the different mills based on 20, 5000,
and 80000 tonnes of feed per day.
Figure 10 shows the direct comparison between the values of the BWI of the BICO and
the Sepor ball mills on a plant scale of producing 20 tonnes per day. On average, the BICO ball
mill had lower kWh/t values for each of the tests run on both Dr. Usman’s and the Australian
material. This leads to lower overall operating costs for the plant to grind the same material.
When comparing these values to the round robin average of test from around the world, both the
BICO and Sepor mills produced lower kWh/t values and hence, lower estimated operating costs.
24
$10,000
Power Operational Cost ($/year)
$9,500
Husni 1
$9,000
Husni 2
$8,500 Aus. 1
Aus. 2
$8,000
$7,500
BICO SEPOR
Figure 10: This graph shows a direct comparison between the different materials run on both the BICO
and Sepor mills and the operating costs estimated for a 20 tonnes of feed per day mill.
Companies often underestimate the CAPEX to compete for financing which leads to
lower bidding from contractors. [14] This explains the massive amounts of budget overruns at
the end of projects. By correctly estimating the CAPEX early in the feasibility studies, there is
A useful tool when predicting CAPEX values is the six-tenths rule. Approximate costs
can be obtained if the cost of a similar item of different size or capacity is known. [15] The
approximate cost of the equipment can be multiplied of a ratio of corresponding sizes (cfm, Hp,
ft2, etc.) and raised to the six-tenths. Six-tenths is an average size exponent for many different
process equipment. [16] For ball mills specifically, the exponent is .65 which was used in the
calculations for CAPEX with tons/hr being the corresponding size used in the ratio.
𝑆𝐵 0.6
𝐶𝐵 = 𝐶𝐴 ( )
𝑆𝐴
(4)
25
Where Cb is the approximate cost ($) of equipment having size Sb (cfm, Hp, ft2, etc.), Ca
is the known cost ($) of equipment having corresponding size Sa (same units as Sb) and Sb/Sa is
The typical size for the Bond Ball Mills is about 0.3 m x 0.3 m. The cost for a BICO ball
mill is about $6900. [17] The Sepor ball mill costs about $9000 but includes an additional
charge, table and electronics. [18] Using the six-tenths rule, it was estimated that the average ball
mill cost for the round robin tests was about $7600.
By knowing the values for Bond Ball Mills, a ball mill for a larger plant scale can be
calculated by again using the six-tenths rule. Table 14 shows the estimated capital costs for a 20,
5000 and 80000 tonnes of feed per day mill with the BICO, Sepor and round robin tests. The
dimensions of the ball mill needed for the capacities were taken from Cost Mine and the costs
were extrapolated. [19] Because there is a dependence of kWh/t when using the six-tenths rule,
the BICO results showed the lowest cost estimate for the ball mills followed by Sepor and then
the round robin. These values differ by about a half million at the largest scale of plant which
would cause either over or under estimates of CAPEX depending on which value was chosen
Table 14: The capital cost estimates of ball mills for various size plant scales using the test results from
the BICO, Sepor and Round Robin.
20 tonnes feed per 5000 tonnes feed per 80000 tonnes feed per
day day day
kWh/mt (diam. X $/ball mill (diam. X $/ball mill (diam. X $/ball mill
length) length) length)
BICO 13.01 0.8x1.4m $249,655 5x9m $3,660,112 7x12m $5,491,460
Sepor 14.71 0.8x1.4m $268,847 5x9m $3,941,485 7x12m $5,913,618
Round 15.39 0.8x1.4m $275,917 5x9m $4,045,138 7x12m $6,069,135
Robin
26
Overall, the results from the BICO ball mill tended to underestimate the values for both
CAPEX and OPEX while the Sepor ball mill and ball mill used during the round robin testing
were similar with overestimating the values. Under or over estimations of these values greatly
affect the overall project budget which is why it is so important to perform very precise
7.0 Conclusion
The data from Dr. Usman’s material suggests that there is no statistical difference
between the BICO and Sepor mills using a 95% confidence interval, but there is a 10%
difference between the average values from each machine. This 10.0% increase from the BICO
to the Sepor will greatly increase the predicted required energy and, therefore, the cost of the
process.
The data from the Australian material suggests that there is a statistical difference
between the two machines. There is an 11.6% increase in the calculated value from the BICO
mill to the Sepor mill. The underestimation of the BICO mill could cost a mining company up to
$5.1 million in a year in power operational costs, assuming 80,000 tonnes of feed per day.
Many mining projects are having cost overruns and getting an accurate Bond Work Index
will help to mitigate this problem. The Sepor mill reports a higher value, which will help in not
This is the desired ball charge analysis, while below are the analyses of the ball charges that were used.
28
1.45 39 8.794
1.17 64 7.16
1 10 0.668
0.75 68 2.436
0.61 91 1.495
Total 272 20.553
.
29
0
2 2.2 2.4 2.6 2.8 3 3.2 3.4
-0.1
A: log(% Passing) = 0.7436(ln(size)) - 2.55
R² = 0.9982
-0.2
B: log(% Passing) = 0.7375(ln(size)) - 2.4621
R² = 0.9979
-0.3
C: log( % Passing) = 0.6605(ln(size)) - 2.2729
R² = 0.9982
Log(% Passing)
-0.4
Test A
-0.5
Test B
-0.6 Test C
-0.7
-0.8
-0.9
-1
Log(Size) (µm)
The Gates-Schumman plot used to determine the F80 of Dr. Usman’s material.
30
0.5
0
Ln(-Ln(1-% Passing/100))
-0.5
Test A
Test B
-1 Test C
A: ln(-ln(1-% Passing/100)) = 0.9287(ln(size)) - 6.7884
R² = 0.9958
-1.5 B: ln(-ln(1-% Passing/100)) = 0.9841(ln(size)) - 6.9037
R² = 0.9932
C: ln(-ln(1-% Passing/100)) = 0.9268(ln(size)) - 6.594
-2 R² = 0.9873
-2.5
Ln(size) (µm)
The Roslin Rambler plot used in determining the F80 for Dr. Usman’s material
31
Test A
Size Weight Percent Cumulative Percent Cumulative Percent Percent Product in
Mesh
(µm) Retained Retained Retained Passing Feed
10 1700 99.2 29.6 29.6 70.4 11.2
16 1000 77.0 23.0 52.6 47.4
28 600 46.3 13.8 66.5 33.5
60 250 52.6 15.7 82.2 17.8
100 149 22.1 6.6 88.8 11.2
Pan 0 37.5 11.2 100.0 0.0
Total 334.7
Initial
336.5
Mass
32
Test B
Mesh Size Weight Percent Cumulative Percent Cumulative Percent Percent Product in
(µm) Retained Retained Retained Passing Feed
10 1700 71.9 18.9 18.9 81.1 13.3
16 1000 90.0 23.6 42.5 57.5
28 600 72.5 19 61.5 38.5
60 250 66.2 17.4 78.8 21.2
100 149 30.0 7.9 86.7 13.3
Pan 0 50.7 13.3 100.0 0.0
Total 381.3
Initial 382
Mass
33
Test C
Mesh Size Weight Percent Cumulative Percent Cumulative Percent Percent Product in
(µm) Retained Retained Retained Passing Feed
8 2380 55.8 11.2 11.2 88.8 13.9
10 1700 79.6 16.0 27.3 72.7
16 1000 107.4 21.6 48.9 51.1
28 600 67.5 13.6 62.5 37.5
60 250 80.3 16.2 78.7 21.3
100 149 36.5 7.4 86.1 13.9
Pan 0 69.0 13.9 100.0 0.0
Total 496.1
Initial 497.6
Mass
34
4.5
3.7
3.5 Test B
Test C
3.3
Test A
3.1
2.9
2.7
2.5
5 5.5 6 6.5 7 7.5 8
Log(Size) (µm)
The Gates-Schumman plot used to determine the F80 of the Australian material.
35
0.5
0
5 5.5 6 6.5 7 7.5 8
-0.5 R² = 0.9891
Test A
-1
Test B
Test C
-2.5
Ln(Size) (µm)
A Roslin Rambler plot used to determine the F80 of the Australian material.
36
80.0
A: % Passing = 0.023(Size) + 12.79
70.0 R² = 0.9932
20.0
10.0
0.0
0 500 1000 1500 2000 2500
Size (µm)
The linear plot used to determine the F80 of the Australian material.
37
Test A
Mesh Size Weight Percent Cumulative Percent Cumulative Percent Percent Product in
(µm) Retained Retained Retained Passing Feed
10 1680 156 49.3 49.3 50.7 14.7
16 1000 45.7 14.4 63.7 36.3
28 595 25.9 8.2 71.9 28.1
60 250 29.5 9.3 81.3 18.7
100 149 12.8 4.0 85.3 14.7
Pan 0 46.5 14.7 100.0 0.0
Total 316.4
Initial 316.9
Mass
38
Test B
Mesh Size Weight Percent Cumulative Percent Cumulative Percent Percent Product in
(µm) Retained Retained Retained Passing Feed
8 2380 84.2 29.8 29.8 70.2 13.4
9 2000 28.6 10.1 40.0 60.0
10 1680 24.7 8.8 48.7 51.3
16 1000 44.6 15.8 64.6 35.4
28 595 25.0 8.9 73.4 26.6
60 250 27.0 9.6 83.0 17.0
100 149 10.3 3.7 86.6 13.4
Pan 0 37.7 13.4 100.0 0.0
Total 282.1
Initial 284.3
Mass
39
Test C
Mesh Size Weight Percent Cumulative Percent Cumulative Percent Percent Product in
(µm) Retained Retained Retained Passing Feed
8 2380 106.7 25.5 25.5 74.5 14.7
9 2000 38.9 9.3 34.7 65.3
10 1680 35.3 8.4 43.2 56.8
16 1000 70.3 16.8 59.9 40.1
28 595 43.2 10.3 70.2 29.8
60 250 45.4 10.8 81.1 18.9
100 149 17.8 4.2 85.3 14.7
Pan 0 61.5 14.7 100.0 0.0
Total 419.1
Initial 421.1
Mass
40
2.05
1.95
Grams/Revolution
1.9
1.85
Usman 1.1
1.8 Usman 1.2
1.75
1.7
1.65
1.6
2 3 4 5 6 7 8 9
Cycle Number
As the test reaches closure, the grams/revolution, when compared to the previous cycle, are close
to each other.
41
Usman 1.1
Cycle New Feed Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Number (g) Feed (g) /Cycle (g) (g) Load (g)
0 1166.8 149.35 N/A N/A 149.4 N/A 333.4
1 270.7 34.65 100 896.1 257.1 2.225 333.4
2 276.7 35.42 134 890.1 264.4 1.705 333.4
3 361.5 46.27 175 805.3 340.3 1.682 333.4
4 377.4 48.31 171 789.4 352.8 1.784 333.4
5 397.5 50.88 160 769.3 379.2 2.055 333.4
6 321.3 41.13 138 845.5 306.9 1.933 333.4
7 337.7 43.23 151 829.1 320.3 1.832 333.4
8 351.5 44.99 158 815.3 337.7 1.848 333.4
42
Usman 1.2
Cycle Number New Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Feed (g) Feed (g) /Cycle (g) (g) Load (g)
0 1167 149.38 N/A N/A 149.4 N/A 333.4
1 209.8 26.85 100 957.2 194.8 1.679 333.4
2 360.9 46.2 183 806.1 349.9 1.664 333.4
3 380.5 48.7 173 786.5 353.9 1.768 333.4
4 378.4 48.44 161 788.6 362.7 1.952 333.4
5 341.2 43.67 146 825.8 331 1.968 333.4
6 341.8 43.75 147 825.2 325.3 1.912 333.4
43
1.65
1.6
Grams/Revolution
1.55
Usman 2.1
Usman 2.2
1.5
1.45
1.4
2 3 4 5 6 7
Cycle Number
Grams/revolution for each cycle, as the test closes the change in grams/revolution is minimal.
44
Usman 2.1
Cycle New Feed Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Number (g) Feed (g) /Cycle (g) (g) Load (g)
0 1166.8 149.35 N/A N/A 149.4 N/A 333.4
1 228.5 29.25 100 938.3 197 1.678 333.4
2 322.8 41.32 181 844 300.2 1.428 333.4
3 396.3 50.73 205 770.5 365.1 1.537 333.4
4 380.1 48.65 184 786.7 346.9 1.622 333.4
5 363.4 46.52 176 803.4 337.8 1.659 333.4
6 350.5 44.86 173 816.3 324.3 1.616 333.4
45
Usman 2.2
Cycle New Feed Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Number (g) Feed (g) /Cycle (g) (g) Load (g)
0 1167.1 149.39 N/A N/A 149.4 N/A 333.4
1 197.8 25.32 100 969.3 180.6 1.553 333.4
2 325.6 41.68 198 841.5 311.1 1.358 333.4
3 382.1 48.91 215 785 362.9 1.462 333.4
4 356.1 45.58 195 811 335.5 1.490 333.4
5 373.5 47.81 193 793.6 350.7 1.568 333.4
6 358.6 45.9 182 808.5 342.8 1.630 333.4
7 348.5 44.61 176 818.6 333.3 1.636 333.4
46
0
3.5 3.7 3.9 4.1 4.3 4.5 4.7 4.9
-0.2
Usman 1.1
-0.4 Usman 1.2
-0.6
-0.8
-1
-1.2
-1.4
ln(size) (µm)
The Roslin Rambler plot used to determine the P80 of Dr. Usman’s material used in the BICO
mill.
47
90.0
60.0
Usman 1.1
Usman 1.2
50.0
40.0
30.0
30 50 70 90 110 130
Size (µm)
The linear plot used to determine the P80 of Dr. Usman’s material from the BICO mill.
48
Usman 1.1
Mesh Size (µm) Weight Retained (g) % Retained Cumulative % Retained Cumulative % Passing
115 125 32.8 14.8 14.8 85.2
150 105 27.3 12.4 27.2 72.8
200 74 44.2 20.0 47.2 52.8
270 53 28.3 12.8 60.0 40.0
400 37 15.4 7.0 67.0 33.0
Pan 0 73.0 33.0 100.0 0.0
Total 221.0
49
Usman 1.2
Mesh Size (µm) Weight Retained (g) % Retained Cumulative % Retained Cumulative % Passing
115 125 41.2 16.4 16.4 83.6
150 105 31.3 12.4 28.8 71.2
200 74 52.0 20.7 49.5 50.5
270 53 34.1 13.6 63.1 36.9
400 37 15.3 6.1 69.1 30.9
Pan 0 77.6 30.9 100.0 0.0
Total 251.5
50
0
3.5 3.7 3.9 4.1 4.3 4.5 4.7 4.9
Usman 2.1
Usman 2.2
-0.5
-1
-1.5
ln(size) (µm)
A Roslin Rambler plot used to determine the P80 of Dr. Usman’s material used in the Sepor mill.
51
100.0
90.0
2.1: % Passing = 0.5468(Size) + 15.09
R² = 0.9967
80.0 2.2: % Passing = 0.683(Size) + 3.468
R² = 0.9957
70.0
% Passing
Usman 2.1
60.0
Usman 2.2
50.0
40.0
30.0
30 50 70 90 110 130
Size (µm)
A linear plot used to determine the P80 of Dr. Usman’s material used in the Sepor mill.
52
Usman 2.1
Mesh Size (µm) Weight Retained (g) % Retained Cumulative % Retained Cumulative % Passing
115 125 38.7 15.9 15.9 84.1
150 105 27.8 11.4 27.4 72.6
200 74 44.6 18.4 45.8 54.2
270 53 26.8 11.0 56.8 43.2
400 37 15.5 6.4 63.2 36.8
Pan 0 89.4 36.8 100.0 0.0
Total 242.8
53
Usman 2.2
Mesh Size (µm) Weight Retained (g) % Retained Cumulative % Retained Cumulative % Passing
115 125 28.0 10.5 10.5 89.5
150 105 37.0 13.9 24.4 75.6
200 74 61.2 23.0 47.4 52.6
270 53 39.0 14.7 62.1 37.9
400 37 18.5 7.0 69.1 30.9
Pan 0 82.3 30.9 100.0 0.0
Total 266.0
54
1.750
1.650
Grams/Revolution
1.550
AUS 1.1
AUS 1.2
1.450
1.350
1.250
2 3 4 5 6 7 8 9
Cycle Number
A plot of the grams/revolution for each cycle, as the test nears the end the grams/revolution
stabilizes.
56
AUS 1.1
Cycle New Feed Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Number (g) Feed (g) /Cycle (g) (g) Load (g)
0 1331.1 189.02 N/A N/A 189 N/A 380.3
1 302.4 42.94 100 1028.7 289.6 2.467 380.3
2 241.3 34.26 137 1089.8 222.7 1.378 380.3
3 410.2 58.25 251 920.9 399.1 1.357 380.3
4 424.1 60.22 237 907 417.6 1.506 380.3
5 405.1 57.52 213 926 396.2 1.594 380.3
6 405.3 57.55 203 925.8 393.7 1.660 380.3
7 396.9 56.36 194 934.2 384.4 1.687 380.3
8 375.8 53.36 192 955.3 361.8 1.606 380.3
57
AUS 1.2
Cycle New Feed Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Number (g) Feed (g) /Cycle (g) (g) Load (g)
0 1331.3 189.04 N/A N/A 189 N/A 380.4
1 262 37.2 100 1069.3 247.1 2.099 380.4
2 253.7 36.03 164 1077.6 242.5 1.263 380.4
3 435.2 61.8 273 896.1 418.2 1.307 380.4
4 465.4 66.09 242 865.9 455.8 1.610 380.4
5 392.9 55.79 195 938.4 380.8 1.665 380.4
6 412 58.5 195 919.3 398.3 1.743 380.4
7 392.2 55.69 185 939.1 384 1.778 380.4
58
1.450
1.400
1.350
Grams/Revolution
1.300
1.250
AUS 2.1
1.200 AUS 2.2
1.150
1.100
1.050
1.000
2 3 4 5 6 7 8
Cycle Number
A plot of the grams/revolution of each cycle, as the test nears its end the change in
grams/revolution is minimal.
59
AUS 2.1
Cycle New Feed Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Number (g) Feed (g) /Cycle (g) (g) Load (g)
0 1329.3 188.76 N/A N/A 188.8 N/A 379.8
1 318.3 45.2 100 1011.0 304.5 2.593 379.8
2 217.9 30.94 129 1111.4 204.5 1.345 379.8
3 354.3 50.31 259 975.0 341.1 1.121 379.8
4 459.1 65.19 294 870.2 444.5 1.291 379.8
5 406.2 57.68 244 923.1 391.1 1.368 379.8
6 406.6 57.74 236 922.7 393.1 1.424 379.8
7 330.2 46.89 192 999.1 318.1 1.413 379.8
60
AUS 2.2
Cycle New Feed Product in Revolutions Oversize Product Grams/Revolution 250% Circulating
Number (g) Feed (g) /Cycle (g) (g) Load (g)
0 1331.6 189.09 N/A N/A 189.1 N/A 380.5
1 277.1 39.35 100 1054.5 264.3 2.250 380.5
2 232.2 32.97 152 1099.4 216.4 1.210 380.5
3 381 54.1 287 950.6 360.1 1.065 380.5
4 461.5 65.53 306 870.1 461.5 1.294 380.5
5 416.4 59.13 243 915.2 403.9 1.419 380.5
6 489.1 69.45 286 842.5 475.3 1.419 380.5
61
75
65 AUS 1.1
AUS 1.2
45
35
30 50 70 90 110 130
Size (µm)
Linear plot used to determine the P80 of the Australian material from the BICO mill.
62
AUS1.1
Mesh Size (µm) Weight Retained % Retained Cumulative % Cumulative %
Retained Passing
115 125 31 11.1 11.1 88.9
150 105 29.7 10.7 21.8 78.2
200 74 52.8 18.9 40.7 59.3
270 53 36.7 13.2 53.9 46.1
400 37 20.5 7.4 61.2 38.8
Pan 0 108.1 38.8 100.0 0.0
Total 278.8
63
AUS 1.2
Mesh Size (µm) Weight Retained % Retained Cumulative % Cumulative %
Retained Passing
115 125 34.5 12.4 12.4 87.6
150 105 29.5 10.6 23.1 76.9
200 74 51.1 18.4 41.5 58.5
270 53 33.9 12.2 53.7 46.3
400 37 21.1 7.6 61.3 38.7
Pan 0 107.3 38.7 100.0 0.0
Total 277.4
64
75
65 AUS 2.1
AUS 2.2
55
45
35
30 50 70 90 110 130
Size (µm)
Linear plot used to determine the P80 of the Australian material from the Sepor mill.
65
AUS 2.1
Mesh Size (µm) Weight Retained % Retained Cumulative % Cumulative %
Retained Passing
115 125 17.6 6.6 6.6 93.4
150 105 30.4 11.3 17.9 82.1
200 74 54.1 20.2 38.1 61.9
270 53 35.5 13.2 51.3 48.7
400 37 24.8 9.2 60.5 39.5
Pan 0 105.9 39.5 100.0 0.0
Total 268.3
66
AUS 2.2
Mesh Size (µm) Weight Retained % Retained Cumulative % Cumulative %
Retained Passing
115 125 20.5 7.8 7.8 92.2
150 105 28.2 10.7 18.5 81.5
200 74 50.9 19.3 37.8 62.2
270 53 34.3 13.0 50.8 49.2
400 37 19.5 7.4 58.2 41.8
Pan 0 110 41.8 100.0 0.0
Total 263.4
Appendix M: Statistical Analysis of Australian
Material
t-Test: Two-Sample Assuming Unequal
Variances
COST SUMMARY
Operating Costs Cost Percentage
Supplies and Materials $ 6.88 81%
Labor $ 0.63 7%
Administration $ 0.22 3%
Sundry Items $ 0.77 9%
Total $ 8.50
Capital Costs Cost Percentage
Equipment $ 100,729,500.00 29%
Installation Labor $ 63,013,100.00 18%
Concrete $ 8,282,800.00 2%
Piping $ 25,031,700.00 7%
Structural Steel $ 9,009,600.00 3%
Instrumentation $ 6,189,400.00 2%
Insulation $ 3,215,700.00 1%
Electrical $ 12,882,900.00 4%
Coatings and Sealants $ 1,063,000.00 0%
Mill Building $ 7,040,300.00 2%
Tailings Embankment $ 61,399,100.00 18%
Engineering/Management $ 27,078,600.00 8%
Working Capital $ 19,736,900.00 6%
Total $ 344,672,600.00
Total $ 344,672,608.50
*Based on 40,000 tons/day
Example pulled from Cost Mine for a mill that has 40,000 tonnes of feed per day. [19]
Power Operational Costs
20 tonnes feed per 5000 tonnes feed per 40000 tonnes feed per day 80000 tonnes feed per day
day day
kWh/mt $/d $/yr $/d $/yr $/d $/yr $/d $/yr
BICO Mill
Usman 1 13.67 $ 24.33 $ 8,881 $ 6,083 $ 2,220,350 $ 48,665 $ 17,762,798 $ 97,330 $ 35,525,596
Usman 2 13.10 $ 23.32 $ 8,511 $ 5,830 $ 2,127,768 $ 46,636 $ 17,022,140 $ 93,272 $ 34,044,280
Aus. 1 13.24 $ 23.57 $ 8,602 $ 5,892 $ 2,150,507 $ 47,134 $ 17,204,056 $ 94,269 $ 34,408,112
Aus. 2 12.77 $ 22.73 $ 8,297 $ 5,683 $ 2,074,167 $ 45,461 $ 16,593,338 $ 90,922 $ 33,186,676
Sepor Mill
Usman 1 15.20 $ 27.06 $ 9,875 $ 6,764 $ 2,468,860 $ 54,112 $ 19,750,880 $ 108,224 $ 39,501,760
Usman 2 14.53 $ 25.86 $ 9,440 $ 6,466 $ 2,360,035 $ 51,727 $ 18,880,282 $ 103,454 $ 37,760,564
Aus. 1 14.67 $ 26.11 $ 9,531 $ 6,528 $ 2,382,775 $ 52,225 $ 19,062,198 $ 104,450 $ 38,124,396
Aus. 2 14.75 $ 26.26 $ 9,583 $ 6,564 $ 2,395,769 $ 52,510 $ 19,166,150 $ 105,020 $ 38,332,300
ROUND ROBIN
Average 15.39 $ 27.39 $ 9,999 $ 6,849 $ 2,499,721 $ 54,788 $ 19,997,766 $ 109,577 $ 39,995,532
71
Plant Scale 20 tonnes feed per day 5,000 tonnes feed per day 40,000 tonnes feed per day 80,000 tonnes feed per
day
kWh/mt (diam. x $/ball (diam. x $/ball mill (diam. x $/ball mill (diam. x $/ball mill
length) mill length) length) length)
BICO
Usman 1 13.67 0.8 x 1.4 m $249,655 5x9m $3,660,112 6 x 10 m $4,412,702 7 x 12 m $5,491,460
Usman 2 13.10 0.8 x 1.4 m $249,655 5x9m $3,660,112 6 x 10 m $4,412,702 7 x 12 m $5,491,460
Aus. 1 13.24 0.8 x 1.4 m $249,655 5x9m $3,660,112 6 x 10 m $4,412,702 7 x 12 m $5,491,460
Aus. 2 12.77 0.8 x 1.4 m $249,655 5x9m $3,660,112 6 x 10 m $4,412,702 7 x 12 m $5,491,460
SEPOR
Usman 1 15.20 0.8 x 1.4 m $267,478 5x9m $3,921,418 6 x 10 m $4,727,738 7 x 12 m $5,883,511
Usman 2 14.53 0.8 x 1.4 m $267,478 5x9m $3,921,418 6 x 10 m $4,727,738 7 x 12 m $5,883,511
Aus. 1 14.67 0.8 x 1.4 m $267,478 5x9m $3,921,418 6 x 10 m $4,727,738 7 x 12 m $5,883,511
Aus. 2 14.75 0.8 x 1.4 m $267,478 5x9m $3,921,418 6 x 10 m $4,727,738 7 x 12 m $5,883,511
ROUND
ROBIN
Average 15.39 0.8 x 1.4 m $269,647 5x9m $3,953,211 6 x 10 m $4,766,067 7 x 12 m $5,931,211
SPECIFICATIONS Capital
Air Swept Ball Diam. x length HP Cost
Mills
air swept ball 1.8 x 2.3 m 150 $583,980
air swept ball 2.4 x 2.9 m 300 $874,800
air swept ball 2.7 x 3.0 m 400 $970,880
References
[1] F.P. Van Der Meer and W. Maphosa, “High pressure grinding moving ahead in copper, iron,
and gold processing,” in Journal of the South African Institute of Mining and Metallurgy. Vol.
http://www.scielo.org.za/scielo.php?script=sci_arttext&pid=S0038-223X2012000700008
[2] “Opportunities to Enhance Capital Productivity,” Ernest & Young, London, UK, 2015.
http://www.ey.com/Publication/vwLUAssets/EY-opportunities-to-enhance-capital-
productivity/$FILE/EY-opportunities-to-enhance-capital-productivity.pdf
[3] “Mining Industry Energy Bandwidth Study,” BCS, Incorporated, Laurel, MD, 2007.
http://energy.gov/sites/prod/files/2013/11/f4/mining_bandwidth.pdf
[4] “Crushing and Grinding Theory A New Industrial Frontier,” Allis-Chalmer, Milwaukee, WI,
1953.
[5] E. Kaya, P. C. Fletcher and P. Thompson, “Reproducibility of Bond Work Index with
Different Standard Ball Mills,” Phelps Dodge, Safford, AZ and Dawson Metallurgical Lab, Inc.
[6] H. Usman, “Measuring the Efficiency of the Tumbling Mill as a Function of Lifter
Configurations and Operating Parameters,” Ph.D. dissertation, Dept. Mining Eng., Colorado
[8] “Standard Bond Ball Mill Grindability Test,” Colorado School of Mines, Golden, CO, May,
1999.
[9] “Bond Ball Mill Grindability Test Procedure,” Phillips Enterprises LLC., Golden, CO, 2016.
75
[10] D.C. Montgomery, “Inferences about Process Quality,” in Introduction to Statistical Quality
Control,” 7th ed. Hoboken, NJ: Wiley, 2013, ch. 4, sec. 4.2, pp. 139-141.
[11] R.L. Bullook, “Accuracy of feasibility study evaluations would improve accountability,”
Jackling Lecture in Mining Engineering, April 2011, Vol. 63, No. 4, SME, pp. 78-85.
[12] “Sensitivity Analysis for Mineral Processing Projects”, PEC Consulting, Saint Louis, MO,
2015.
[13] R. Anguelov, “Operating Cost Estimate (OPEX),” Project Feasibility Study, February 2009,
[14] R. L. Bullock, “Mine Feasibility Studies Need Improving!” Missouri University of Science
[15] R. W. Whitesides, “Process Equipment Cost Estimating by Ratio and Proportion”, PDH
[16] C.G. Anderson, “Engineering Design and Cost Analysis”, Colorado School of Mines,
http://www.bicoinc.com/ballMill.html.
[19] J. B. Leinart, “Mining Cost Service”, Cost Mine, Spokane Valley, WA, 2015.