Impact of ROM PSD on the Crushing and Grinding Circuit Throughput

Fernando Fernandez, Split Engineering Chile

María Rocha, Rocha Blast Engineers
John Kemeny, Split Engineering, University of Arizona
Tom BoBo, Split Engineering
Cristian Rodriguez, Split Engineering Chile
Rodrigo Fuentealba, Split Engineering Chile

Abstract

Mathematical modeling in conjunction with mineral characterization is widely used as a method for design
and optimization of comminution circuits, to simulate the processes of crushing, grinding and
classification. An important input parameter to the simulation models is the Run-of-Mine Particle Size
Distribution (ROM PSD). The run-of-mine size distributions due to blasting have a great influence on the
performance of a SAG mill. In this paper, a mine case study was conducted where there are four primary
geologic units and the material properties between the four units vary considerably. Also, the mixture of
the four unit that will be mined at any given time also vary considerably. This presents a challenge in
terms of optimizing the blasting for mill production throughout the mine life. As part of the case study,
field rock mass characterization, laboratory rock mechanics tests, and laboratory breakage tests were
conducted. The field characterization and rock mechanics tests provide material properties used to
simulate the effect of different blast designs on run of mine particle size distributions (ROM PSD). The
JKSimBlast software was used for these simulations. The rock breakage tests provide material properties
used to simulate crushing and grinding performance. The JKSimMet software was used for these
simulations.

The goal of the mine is a mill throughput of 95,000 tonnes (104,720 tons) per day (TPD). Based on this
goal, run-of-mine fragmentation analysis was conducted for four blasting designs resulting in powder
factors of 0.44, 0.54, 0.64, and 0.94 kg/t. The powder factor of 0.44 represents a standard blasting design,
while the 0.54 and 0.64 kg/t designs are high-energy blasts specifically to assist with mill production. The
0.94 kg/t design was investigated for the one geologic unit with very high strength. The results of the run-
of-mine fragmentation analysis was then input into a crushing and grinding simulation model to predict
the mill throughput for the time periods of 1-5 years and 6-10 years. The results indicate that certain blast
designs are able to achieve the 95,000 tonnes TPD (104,720 tons TPD) goal, while other blast designs
were not. A standard blast design with a 0.44 kg/t powder factor, for example, was not able to achieve
the throughput goal, while a design with a 0.64 kg/t powder factor was able to achieve the goal in both the
1-5 and 5-10 year time periods.
1.0 Introduction
In recent years, mathematical modeling in conjunction with mineral characterization has been widely used
as a method for the design and optimization of comminution circuits. The Run-of-Mine Particle Size
Distribution (ROM PSD) is a function of rock properties and design parameters of the blast and has a
significant impact on the crushing and grinding treatment circuit. Traditionally, design of the grinding
treatment is estimated using only the F80 value instead of the full distribution of sizes in the feed, but the
finer sizes have great influence on the treatment of a grinder, particularly in the SAG mill.

In this paper, a mine case study was conducted where there are four primary geologic units and the material
properties between the four units vary considerably. Also, the mixture of the four unit that will be mined
at any given time also vary considerably. This presents a challenge in terms of optimizing the blasting for
mill production throughout the mine life. As part of the case study, field rock mass characterization,
laboratory rock mechanics tests, and laboratory breakage tests were conducted. The field characterization
and rock mechanics tests provide material properties used to simulate the effect of different blast designs
on run of mine particle size distributions (ROM PSD). The JKSimBlast software was used for these
simulations, using the Kuz-Ram fragmentation model along with a JKMRC fines correction (JKMRC,
2014). The rock breakage tests provide material properties used to simulate crushing and grinding
performance, including the JK DropWeight test and the SAG Mill Comminution test (SMC Test), which
is an abbreviated, less expensive version of the JK Drop-weight test. The JKSimMet software was used
for these simulations (JKMRC, 2014). The objective is to find a nominal ROM PSD Curve for the project
to ensure the throughput of 95,000 tonnes (104,720 tons) per day according to the current Mining Plan.

2.0 Mineral Characteristics

In order to simulate crushing and grinding, the material properties of the ore materials must be determined.
Even though standard rock mechanics properties such as the uniaxial compressive strength and the tensile
strength are useful in understanding the hardness of different ore materials, traditionally crushing and
grinding simulation models have used specific laboratory breakage tests such as the JK DropWeight.
These tests break the rock in a similar manner to what occurs in actual crushing and grinding equipment.
Details on these tests are given in Napier-Munn et al. (1996).

The geology of the mine consists of altered sedimentary and porphyry rocks. Some alterations render the
rock hard, while other alterations render the rock soft. Overall, the rock is divided into four geologic units,
referred to as SED-H, SED-S, POR-H, and POR-S. SED refers to sedimentary and POR refers to
porphyry, and H refers to a hard alteration and S refers to a soft alteration. In addition to field rock mass
characterization and standard rock mechanics laboratory tests, laboratory rock breakage tests were
conducted to provide the information needed for crushing and grinding simulation. Table 1 shows the
rock breakage properties for the four geologic units. Table 1 shows mean values for each of the rock
properties. Statistical variations in these properties were also determined but are not shown here. In this
study, the 50th percentile (mean) and the 80th percentile (stronger fraction) for the breakage parameters are
considered, and also the values associated with two time periods in the mine life, 1 to 5 years and 6 to 10
years. These periods will be studied first to ensure the throughput of 95,000 tonnes (104,720 tons) per day
for the first ten years of mine life.
 Table 1. Example of Mineral Properties for 1 to 5 years period
Geological
SG SPI A b A*b Ta BMWi (KWh/t)
Units
SED H 2.62 85.4 46.9 0.82 38.7 0.38 14.3
SED S 2.62 68.2 46.9 0.95 44.4 0.44 13
POR H 2.62 51.6 46.9 1.13 52.8 0.52 12.6
POR S 2.62 50 46.9 1.15 53.9 0.53 11.6
SG = Specific Gravity
SPI = SAG Power Index
A, b = Impact breakage parameters, are determined using a high energy impact breakage device called JK Drop Weight test
A*b = Hardness of the ore parameter
Ta = Abrasion breakage parameter
BMWi = Ball Mill Work Index

3. Comminution Models and Simulation Results

Primary Crusher Simulation

The crushing process may be regarded as a flow chart describing a series of simple and repetitive events.
The new feed entering the first crusher is 'classified'. Very fine material breakage skips and reports directly
to a product. The very large material reports directly to breakage. The material of intermediate size reports
to break with a probability that increases with the size. After the break, the classification process is
repeated, with fine material reporting to product, and material that is still very large reporting to breakage,
and so on. This process is illustrated in Figure 1.

Figure 1. Concept of classification and breakage in a crusher (from Napier-Munn et al., 1996)

The gyratory crusher breakage model described in Napier-Munn et al. (1996) is used in this paper. This
model uses four parameters, K1, K2, K3 and t10, as described below:

K1 = Crusher Closed Side Setting (CSS)

K2 = Crusher CSS + Eccentric Throw (ET)
K3 = Constant (usually 2.3)
t10 = percentage passing one tenth (1/10) of the average initial particle size (from drop weight test).

The parameter K3 is normally a constant with a value of 2.3, and for the other parameters a CSS value of
150mm, an ET value of 45mm, and a t10 value of 5% were used in this study.

An example of the simulation of primary crushing using the SED-H rock type with a blasting powder
factor of 0.44 kg/t is shown in Figure 2. Figure 2b shows the particle size distribution before and after
primary crushing. As expected, most of the size reduction occurs in particles above about 70 mm.

Figure 3. Simulation of primary crushing using SED-H rock type with powder factor of 0.44 kg/t.

The primary crusher model described above was used to simulate the crushing of the four geologic units
under different blast designs. When considering different blast designs for the mine, certain variables will
be fixed. This includes the drill diameter of 270 mm, the bench height of 15 meters, and the drill hole
length of 17 meters. The explosive used is also fixed, a Fortran Mex65 emulsion. The properties of the
emulsion include a density of 1.3 kg/m3, a VOD of 5200 m/s (17,060 ft/s)95, an energy of 3.55 MJ/kg,
and RWS and RBS values of 1.16 and 1.97, respectively. The primary variables in blast design, therefore,
are the burden and spacing. By varying the burden from 6.5 m (21.33 ft) to 4.5 m (14.76 ft) and the
spacing from 7.5 m (24.62 ft) to 5.2 m (17.06 ft), four case studies have been conducted with powder
factors of 0.44, 0.54, 0.64 and 0.94 kg/t. The case with a burden of 6.5 m, (21.33 ft), spacing of 7.5 m
(24.62 ft), and powder factor of 0.44 kg/t, is considered to be a normal blast design, while the others have
elevated energies to assist with mine-to-mill optimization. Also, the 0.94 kg/t powder factor is only
considered for the hardest rock unit, POR-H. For each of the four geologic units and for the four powder
factors, the run-of-mine particle size distributions (ROM PSD) were determined using the JKSimBlast
software (JKMRC, 2014). These ROM size distributions were then fed into the crusher model to give
primary crusher product size distribution curves. These curves for the four geologic units and the four
powder factors are shown in Figure 2. Both the mean size distributions and the 80th percentile size
distributions were calculated, as shown in Figure 2. The 80th percentile curves represent worst-case
conditions for courser fragmentation that could occur.
 Figure 2. Primary crusher product size distributions for the four geologic units (SED-H, SED-S,
POR-H, POR-S) and for four blast designs (0.44, 0.54, 0.65, 0.94 powder factors). The 0.94
powder factor only used for POR-H.

Grinding Circuit Simulation

Mill models are based on the assumption that the contents of the mill are thoroughly mixed. The
following equation describes the steady state conditions for material balance for each size fraction.
Feed + Appearance = Breakage to finer particles + Product Equation 1
Appearance is the distribution of fragment sizes in an AG / SAG mill for a given material. It is
determined experimentally by using at least three size fractions and it is defined in terms of three
parameters of abrasion and impact breakage (A, b and t10). Appearance functions are dependent on the
size, power and energy levels prevailing in the mill estimated from the load and size of the mill. Details
of the appearance function can be found in Napier-Munn et al. (1996).
The amount of breakage or rupture rate, t10, is related to the specific comminution energy as shown in
equation 2:
t10  A[1  e( b.Ecs ) ] Equation 2
Where,
t10 = percentage passing one tenth (1/10) of the average initial particle size.
Ecs = specific grinding energy (kWh/t)
A, b = parameters from impact breakage of the ore.

For the hydrocyclone and other classification models such as trommels or screens, the concept of an
efficiency curve is used, which defines the behavior of the classifier. The hydrocyclone model used in this
paper is based on this concept, using equations which correlate with certain equipment dimensions such
as the diameter of the cyclone, apex angle and taper angle (Napier-Munn et al., 1996).

The grinding circuit model used in this work considers a SAG mill with a 40' x 26' (12.2 x 7.9m) Effective
Grinding Length (EGL), in combination with two ball mills with a 27' x 45' (8.2 x 13.7m) EGL. The SAG
mill has an installed capacity of 26 MW and the ball mills have a capacity of 37.5MW. Also, the ball mills
are in closed circuit with gMax33 hydrocyclones. Finally, the product limit, P80, for flotation is assumed
to be 150 microns (0.0059 inches). An example of the results of this grinding circuit simulation for the
SED H geologic unit with a blasting powder factor of 0.44 kg/t is shown in Figure 4.

Figure 4. Example Grinding Circuit and Classification for the SED H geologic unit.

The output from the grinding circuit simulation is tonnes per day (TPD). The mining plan is based on a
throughput of 95,000 tonnes (104,720 tons) per day. In order to determine a blasting pattern that will
ensure that a throughput of 95,000 tonnes (104,720 tons) per day can be achieved, four blast designs
resulting in powder factors of 0.44 (standard), 0.54 (optimized), 0.64 (high energy) were considered. A
very high energy powder factor of 0.94 kg/t was also considered for the hardest unit, POR-H. Figure 5
shows the results of grinding throughput simulations for each of the four geologic units. Results are
shown for both the mean size distributions and the 80th percentile distributions with represent a worst-
case coarse distribution. Results are shown as tonnes per day (TPD) vs. powder factor (kg/t), and the
throughput goal of 95,000 tonnes (104,720 tons) per day is also shown on the figures. The results in
Figure 5 clearly show that the harder alterations (SED-H and POR-H) require higher powder factors to
achieve the throughput goal of 95,000 tonnes (104,720 tons) per day. The results show that in general
the high-energy powder factor of 0.65 kg/t can achieve 95,000 tonnes (104,720 tons) per day in all units
except POR-H.

Figure 5. Tonnes per day vs. blast design for the four geologic units

4. Study of production by time period and blast design

The results shown in Figure 5 above do not take into account the distribution of the four geologic units
that will be mined in the first ten years of the mine life. Table 2 shows the distribution of mining the four
geologic units for years 1-5 and years 6-10. In years 1-5 for example, 78% of the mined ore will be of the
soft alterations and only 22% will be of the hard alterations. In years 6-10 this changes to 70% soft and
30% hard. In this section, these distributions are considered to determine the expected throughput.

Table 2. Distribution of the four geological units that will be mined in the first 10 years
1-5 6-10
Period
(Years) (%) (%)
All 100% 100%
SED-H 11% 23%
SED-S 31% 33%
POR-H 11% 7%
POR-S 47% 37%
First of all, we consider the case where the same blast design will be applied to all rock types in a given
time period, but the blast designs for different time periods can be modified. The primary blast design
variables, as discussed previously, are the burden and spacing, and four cases have been considered
resulting in powder factors of 0.44, 0.54, 0.64, and 0.94 kg/t. Figure 6 shows the results of grinding
throughput simulations for years 1-5 and years 6-10. Results are shown as tonnes per day (TPD) vs.
powder factor (kg/t), and the throughput goal of 95,000 tonnes (104,720 tons) per day is also shown on
the figures. These results take into account the distributions of the different geologic units as presented
in Table 2. The results show that a powder factor of 0.6 kg/t is sufficient to meet the 95,000 TPD (104,720
tons) throughput requirement for both the 1-5 year and 6-10 year time periods.

0.44 0.54 0.64 0.94

TPD 1-5
83,524 91,598 99,921 102,913
years
TPD 6-11
79,289 89,807 98,466 100,370
years

Figure 6. Comparison between Powder Factor vs. Tonnage (TPD) for each period

Now we consider the case where the blast design for a given time period does not have to be the same in
all the geologic units. For example, it might be advantageous to have a higher powder factor in the hard
alterations compared with the soft alterations. Table 3 shows four mixtures of powder factors that were
considered. The first one for example (labeled 0..54 + 0.64), considers a powder factor of 0.54 in the
sedimentary units and 0.64 in the porphyry units. The last one (labeled 0.64 + 0.94) considers 0.64 in all
the units except POR-H, where 0.94 is used.

Table 3. Summary of blasting mixtures for each geological unit and powder factor

0.54+0.64 0.44 0.54 0.64 0.94 0.54+0.64+0.96 0.44 0.54 0.64 0.94
SED H X SED H X
SED S X SED S X
POR H X POR H X
POR S X POR S X

0.64 0.44 0.54 0.64 0.94 0.64+0.94 0.44 0.54 0.64 0.94
SED H X SED H X
SED S X SED S X
POR H X POR H X
POR S X POR S X
Table 4 gives the grinding circuit throughput results (TPD) for the powder factor mixtures given in Table
3. It shows, for example, that the 0.54+0.64 mixture is able to meet the throughput requirements in the
first 5 years but not in years 6-10. It also shows that the 0.64 mixture is able to meet the requirement in
both the 1-5 year and 6-10 year periods. These results are also shown graphically in Figure 7. Figure 8
shows the primary crusher product size distributions for each of the mixtures, for both the 1-5 and 6-10
year time frames.

Table 4. Numerical data of TPD for each blasting mixture

0.54+0.64 0.54+0.64+0.94 0.64 0.64+0.94

TPD 1-5 years 94969 97961 99922 102914
TPD 6-11 years 92418 94322 98467 100371

Figure 7. Tonnes per day versus the powder factor mixtures shown in Table 3.

Figure 8. SAG Feed Curve Optimized (1-5 and 6-10 years)

5.0 Conclusions

The run-of-mine size distributions due to blasting have a great influence on the performance of a SAG
mill. In this paper, a mine case study was conducted where there are four primary geologic units and the
material properties between the four units vary considerably. Also, the mixture of the four unit that will
be mined at any given time also vary considerably. This presents a challenge in terms of optimizing the
blasting for mill production throughout the mine life. As part of the case study, field rock mass
characterization, laboratory rock mechanics tests, and laboratory breakage tests were conducted. The field
characterization and rock mechanics tests provide material properties used to simulate the effect of
different blast designs on run of mine particle size distributions (ROM PSD). The JKSimBlast software
was used for these simulations. The rock breakage tests provide material properties used to simulate
crushing and grinding performance. The JKSimMet software was used for these simulations.

The goal of the mine is a mill throughput of 95,000 tonnes (104,720 tons) per day (TPD). Based on this
goal, run-of-mine fragmentation analysis was conducted for four blasting designs resulting in powder
factors of 0.44, 0.54, 0.64, and 0.94 kg/t. The powder factor of 0.44 represents a standard blasting design,
while the 0.54 and 0.64 kg/t designs are high-energy blasts specifically to assist with mill production. The
0.94 kg/t design was investigated for the one geologic unit with very high strength. The results of the run-
of-mine fragmentation analysis was then input into a crushing and grinding simulation model to predict
the mill throughput for the time periods of 1-5 years and 6-10 years. The results indicate that certain blast
designs are able to achieve the 95,000 TPD (104,720 tons) goal, while other blast designs were not. A
standard blast design with a 0.44 kg/t powder factor, for example, was not able to achieve the throughput
goal, while a design with a 0.64 kg/t powder factor was able to achieve the goal in both the 1-5 and 5-10
year time periods.

References

Bailey, C, Lane, G, Morrell, S and Staples, P.(2009), What can go wrong in comminution circuit design?,
in Proceedings Tenth Mill Operators’ Conference, pp 143–149.
Brent G.F., Hawke S.J., Gómez R (2012), Ultra-High intensity blasting for improved ore comminution,
Fragblast10, Nueva Delhi, Noviembre.
JKMRC. 2014. http://www.jkmrc.uq.edu.au
Napier-Munn, T J, Morrell, S, Morrison, R D and Kojovic, T. (1996),Mineral Comminution Circuits, their
Operation and Optimization, JKRMC Monograph Series in Mining and Mineral Processing 2, Ed.
Rocha M., Kemeny J., BoBo T., Rodriguez C. (2014),Calibration of a Fragmentation Model for a New
Mining Operation, 40th Symposium ISEE, Denver, February.
Stark, S, Perkins, T and Napier-Munn, T J.(2008), JK drop weight parameters – A statistical analysis of
their accuracy and precision and the effect on SAG mill comminution circuit design, in Proceedings
Metallurgical Plant Design and Operating Strategies 2008, pp 147-156.