You are on page 1of 53

Particulate Science and Technology

An International Journal

ISSN: 0272-6351 (Print) 1548-0046 (Online) Journal homepage: http://www.tandfonline.com/loi/upst20

Evaluation of the Relationship Between


Energy Input and Particle Size Distribution
in Comminution with the Use of Piecewise
Regression Analysis

E. Petrakis, E. Stamboliadis & K. Komnitsas

To cite this article: E. Petrakis, E. Stamboliadis & K. Komnitsas (2016): Evaluation of


the Relationship Between Energy Input and Particle Size Distribution in Comminution
with the Use of Piecewise Regression Analysis, Particulate Science and Technology, DOI:
10.1080/02726351.2016.1168894

To link to this article: http://dx.doi.org/10.1080/02726351.2016.1168894

Accepted author version posted online: 30


Mar 2016.

Submit your article to this journal

Article views: 100

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=upst20

Download by: [Orta Dogu Teknik Universitesi] Date: 10 May 2016, At: 03:04
Evaluation of the Relationship Between Energy Input and

Particle Size Distribution in Comminution with the Use of

Piecewise Regression Analysis

E. Petrakis*

Technical University of Crete, School of Mineral Resources Engineering, Chania, Greece


Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

E. Stamboliadis

Technical University of Crete, School of Mineral Resources Engineering, Chania, Greece

K. Komnitsas

Technical University of Crete, School of Mineral Resources Engineering, Chania, Greece

Abstract

It is known that the single linear Gates-Gaudin-Schuhmann (GGS) model is in some cases unable

to fully describe the particle size distribution of comminution products. In order to overcome this

shortcoming, piecewise regression analysis was used to predict the size distributions derived,

after grinding for various periods four mono-sized fractions of quartz and marble in a laboratory

ball mill. The single line was divided into two straight lines, which indicate the presence of two

domains of particle sizes and therefore the involvement of two distinct breakage mechanisms.

The obtained GGS model parameters were used to determine the evolution of particle size

distribution as a function of the energy input.

1
In addition, the existing relationships between energy input and particle size distribution were

improved, by taking into account the effects of the feed size and material type. The new

relationships obtained can be used for a more accurate estimation of the required energy for

breakage.

Keywords: comminution, Gates-Gaudin-Schuhmann, piecewise regression analysis, specific

energy
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

INTRODUCTION

Size reduction is one of the most fundamental and energy intensive operations in the

mineral processing industry, consuming 34% of the electrical energy generated worldwide

(Deniz 2013a). Estimations on energy consumption in mining operations in the USA indicate

that 39% is used for beneficiation and processing operations, 75% of which is consumed in

comminution (BCS 2002). In addition, a large share of this energy is absorbed by the device and

only a small share, given that the efficiency of Semi-Autogenous Grinding (SAG) mill is about

15%, is used for size reduction (Djordjevic 2010). Therefore, continuous research efforts are

required to reduce energy consumption and obtain products with the desired size distribution.

The main and well known theories, that first described the relationship between specific

energy requirements and size reduction in comminution, are those of Rittinger (1867), Kick

(1885) and Bond (1952). Walker and Shaw (1954) proposed the differential Equation (1), which

considers the theories of Rittinger, Kick and Bond as partial cases and states that the energy

required to make a small change in the size of an object is proportional to the size change and

inversely proportional to the object size raised to the power m.

2
dx
d C (1)
xm

where: d is the infinitesimal specific energy required to reduce by dx the size of a

particle with size x, while C is a constant related to material type and m is a constant indicating

the order of the process. If m in Equation (1) equals to 1, 1.5 or 2, then the theories of Kick,

Bond and Rittinger apply, respectively.

Since these theories do not incorporate the concept of particle size distribution (PSD),
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

new approaches were developed over the years. Charles (1957) first proposed Equation (2)

assuming that both the initial feed and the final product size distribution were described by the

Gates-Gaudin-Schuhmann (GGS) model.

C
x 1m (2)
m 1 1 m 0

where and x0 are the distribution and size modulus, respectively, as derived by the GGS

model.

However, for the derivation of Equation (2) Charles used Equation (1), that has no

physical equivalent since it assumes that the particle size diminishes continuously, without

breakage into smaller daughter particles.

Stamboliadis (1996, 2000) stated that comminution is based on the following

assumptions: (i) when sufficient force is applied upon a particle, this breaks into smaller ones

that all together have the same initial mass, (ii) the shape of size distribution, which follows the

GGS model, for the produced particles is similar for all energy levels, and (iii) the energy

required to produce a single particle depends on the surface area of the new particle to the power

n. His approach can be expressed by Equations (3) and (4)

3

A x02n3 , for 2 n 3 0. (3)
2 n 3

lnx0
, for 2 n 3 0. (4)
x0

If the exponent n takes the value 1.5, 1.25 or 1, these equations describe the theories of

Kick, Bond and Rittinger, respectively.

Many other specific energy equations have been proposed in the scientific literature for
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

the design of milling circuits (Turner 1982; Morrell 1996), but the results were not always

accurate. In general, these equations do not consider explicitly the effect of the feed size on the

consumption of specific energy in mills. Some factors, such as the feed size or the material type,

may have decisive influence on the specific energy consumption and therefore on the milling

efficiency. A noteworthy exception is the Morrells (2008) equation, which relates specific

energy consumption with the mill feed size (F80) and the product size (P80). However, the

assumption of representing particle size distributions with one point, (F80) or (P80), does not seem

appropriate, since different size distributions may occur at this point. Furthermore, Kapur (1972)

showed that well-developed self-similar size distributions should be used to develop energy-size

reduction equations. However, it was revealed that self-similarity does not exist in the presence

of non-first-order breakage kinetics (Bilgili 2007).

During the last decades, considerable work has been done, using a phenomenological

grinding kinetics mathematical model derived from population balance considerations, to

optimize energy consumption in grinding circuits. This model uses two sets of parameters,

namely the specific breakage rate and the breakage function for various size fractions (Austin,

Klimpel, and Luckie 1984). This function provides the fundamental size-mass balance equation

for fully mixed batch grinding systems and its analytical solution has been successfully applied

4
to metallic, intermetallic and ceramic systems (Matijai and Kurajica 2010; Ian, Bailon, and

L'Esprance 2011). Many researchers have mentioned the advantages of this equation (Ipek and

Gktepe 2011; Wang et al. 2011; Deniz 2013a; Gupta and Sharma 2014), while the scale-up of

laboratory data to full-scale mills has also been discussed (Deniz 2013b).

The variation of the kinetic model parameters, when different mill operating conditions

are used, has been investigated in several recent studies. These parameters depend on factors

such as ball size (Katubilwa 2008; Katubilwa and Moys 2009; Deniz 2012; Olejnik 2013), media
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

shape (Qian, Kong, and Zhang 2013; Umucu, Altnine, and Deniz 2014), mill speed (Carvalho

and Tavares 2013; Gupta and Sharma 2014) and powder loading (Shin et al. 2013). However,

other researchers have criticized the fact that the population balance model has no intrinsic

capability to simulate the process under different operating conditions (Carvalho and Tavares

2013). Therefore, the relationship between energy input and particle size distribution, which is a

fundamental aspect of comminution, still requires further elucidation.

Other authors focused their efforts on the advantages of the Attainable Region (AR)

approach. Although this approach has been initially proposed for the analysis of chemical

engineering processes, it can also be used for the design of comminution circuits. The AR

approach has been used successfully to optimize batch milling of various materials (Khumalo et

al. 2008; Katubilwa et al. 2011; Danha et al. 2015).

Particle size is probably the most important physical characteristic of a particulate

material and the knowledge of its distribution is essential for many industrial processes since it

affects product quality and behavior (Ramakrishnan 2000; Tademir, Ozdag, and Onal 2011;

Zhang et al. 2012). In mineral processing operations, the determination of particle size

distribution is the key to improve process efficiency (Xia, Yang, and Zhau 2012; Xia et al. 2012).

5
Several mathematical models and expressions have been proposed for the description of the

particle size distribution of a particulate material (Allen 2003).

It is known that the GGS model is suitable for the determination of particle size

distribution in comminution (Wills and Napier-Munn 2006). As already mentioned, energy input

affects particle size distribution (Charles 1957; Stamboliadis 2000, 2002, 2003), providing that

the comminution products follow the GGS model and the shape of size distribution curve is kept

constant for all energy levels. However, the single linear GGS model is unable to fully describe
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

the entire data and this is why only the lower part of the particle size distributions was used in

these studies. In addition, no explanation was provided for the method used for fitting the dataset

and thus the validity of results is under question. Several years later, comminution tests on

chromite ores using different devices showed that two different lines, indicating the presence of

two different domains of particle sizes, could be used in the GGS model plots (Tademir and

Tademir 2009).

Furthermore, the exponents (1 m) and (2n 3) can be calculated from Equations (2)

and (3) using the loglog plot of the energy consumed versus the size modulus x0. Charles (1957)

and Stamboliadis (2000, 2002) using different experimental results, confirmed that a relationship

between these exponents and the distribution modulus exists, and also that the terms

(1 m + ) and (2n 3 + ) for most minerals have values very close to zero. Also, they found

out that these terms are greater than zero for minerals that exhibit cleavage in their structure,

such as galena and salt (NaCl). However, it is not known whether these terms will be affected by

the amount of energy consumed for the reduction of the materials size. Therefore, several issues

related to comminution of particulate materials still remain unclear, and there is an urgent need

for the development and validation of new approaches.

6
The present study aims to elucidate the evolution of PSD in relation to energy input and

improve the established energy-size reduction relationships, using quartz and marble as test

materials. For this purpose, piecewise regression analysis was applied to the GGS model data, by

dividing the single line into two straight lines. The parameters derived were used to elucidate the

effect of material type and feed size on energy demand during comminution in a ball mill.

MODELS USED FOR THE SIMULATION OF PARTICLE SIZE


Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

DISTRIBUTION

Many mathematical models have been developed to determine the size distribution of

particles. The most important two-parameter models are the well-known normal, log-normal,

Gates-Gaudin-Schuhmann (GGS) and Rosin-Rammler (RR), as described by Allen (2003). King

(2001) has described truncated distributions e.g. logarithmic distribution, truncated RR and log-

normal distribution, while the fractal size distribution is another approach that can used to

describe particulate materials (Ahmed and Drzymala 2005; Tademir 2009). The GGS model

that was used in the present study is expressed as


x
P 100 (5)
x0

where: x is the screen size, P is the cumulative weight % finer than size x, x0 is the

maximum particle size (size modulus) and is the distribution modulus. Theoretically, lower

values of suggest that more fines, more large particles and fewer particles in the middle range

will be produced (Lu et al. 2003). Equation (5) denotes that as the value of increases the

distribution is getting narrower.

Equation (5) can be rewritten as logarithmic Equation (6):

7
logP logx 2 logx0 (6)

If logx and logP are plotted this equation produces a straight line. The slope of the line is

the distribution modulus , while the size modulus x0 can be calculated from the intersection

point.

However, single linear models may not prove adequate for the description of the entire

dataset. Piecewise linear regression is a type of regression that allows multiple linear models to

fit the data, while the lines intersect at a point called breakpoint. Thus, the single power-law of
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

Equation (5) would yield the following two power-law equations

f
x
P 100 , x x b (7)
xf

c
x
P 100 , x x b (8)
xc

where: xf or xc is the maximum particle size (size modulus), is the distribution modulus,

f and c are indices which correspond to the finer and coarser domain of products, respectively

and xb is the breakpoint, which is defined as the transition point between two domains.

Therefore, using the piecewise regression model, Equation (5) can be rewritten as

logP f logx b f logx logxb c logx bc (logx logxb ) (9)

where f and c are the slopes and bf and bc are the intercepts of the straight lines. Each

term in parenthesis represents a logical operation i.e., if the term is true it will get the value 1,

whereas if it is false it will get the value 0. Therefore, Equation (9) actually represents two

models. The value of logxb can be considered as the upper limit of the first domain (fine

products) and the lower limit of the second domain (coarse products).

8
MATERIALS AND METHODS

The materials used in the present experimental study are quartz obtained from Assiros,

near Thessaloniki, Greece and marble obtained from west Crete, Chania area, Greece. Both

materials were obtained from quarries of the respective areas. 100 kg of each material was

sampled from several locations to obtain representative samples. Quartz has a microcrystalline

structure, while marble which is of metamorphic origin consists of carbonates. The microscopic
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

examination of both materials in polished and thin sections showed that they are compact and

homogeneous (Figures 1 and 2). No open cracks, discontinuities or alteration phenomena were

noticed. Marble is a relatively soft material with a hardness varying between 3 and 4 on Mohs

scale, while quartz is a much harder material (7 on Mohs scale).

The experiments were performed in a laboratory ball mill (Table 1) at a constant speed of

= 66 rpm (1.1 Hz), corresponding to 70% of its critical speed. The mill charge consisted of

25.4 mm (1 in.) stainless steel balls with density b = 7.85 g/cm3. The parameters that describe

the mass of grinding media J and material fc, respectively, are calculated from Equations (10)

and (11). J is the ball filling ratio, which is the fraction of the mill filled by the media bed, while

fc is the filling ratio of the material fed in the mill.

Volumeof solid balls 1


J 1 (10)
mill volume

Volumeof solid material 1


fc 1 (11)
mill volume

where is the bed porosity of balls and material (assumed to be 40% for both quartz and

marble) (Austin, Klimpel, and Luckie 1984).

9
The ball filling ratio was 0.2, while the material filling ratio was 0.04 corresponding to

345.0 g and 351.5 g of quartz and marble, respectively. The fraction of the spaces between the

balls at rest, U, that is filled with material bed (interstitial filling) was 0.5 and can be calculated

by Equation (12)

fc
U (12)
0.4 J

The samples were homogenized by the cone and quarter method and about 6 kg of each
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

material was used for the tests. Both materials were crushed in a jaw crusher to 4 mm, wet

screened at 150 m and the +150 m fraction dried and screened in a series of screens with an

aperture ratio 2 to obtain the feed fractions. Four mono-size fractions of each material

(3.35 + 2.36 mm), (1.7 + 1.18 mm), (0.850 + 0.600 mm) and (0.425 + 0.300 mm) were

obtained. The grinding tests were performed at various grinding times t (0.5, 1, 2, 4 or 8 min),

while the products obtained at each time t were wet sieved in a series of screens with a ratio of

2 for the determination of particle size distribution. All tests were carried out in duplicate and

average values are given in the results section.

It should be also noted that the following issues were taken into consideration:

i. A fresh feed charge was used in each test.

ii. The feed charge was sampled and screen analyzed for size distribution in the same way

as the mill product.

iii. The mill was layer loaded with balls and feed, following the standard practice (Gupta,

Zouit, and Hodouin 1985)

The mill power W was calculated by the following formula (Stamboliadis,

Emmanouilidis, and Petrakis 2011)

10
W 9.9 M N D (13)

where: is the total mass (kg) of the feed material and balls, is the rotational speed

(Hz) and D is the internal diameter of the mill (m).

The energy consumption in the mill is proportional to time t and can be calculated by the

formula:

E W t (14)
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

where: W is the net mill power (W), t is the grinding time (s) and E is the net energy

consumption (J). If Mp is the mass of the feed material (kg), then the specific energy required

for grinding (J/kg) is calculated by Equation (15)

W t
(15)
Mp Mp

The specific energy in kWh/t is calculated by dividing Equation (15) by 3,600.

RESULTS AND DISCUSSION

Product Size Distribution

As mentioned earlier, the GGS model was selected for the description of particle size

distribution of the mill products obtained after different grinding times and corresponding to

different energy input. The linear best fit of the experimental distributions enables the calculation

of the single GGS model parameters and x0. Tables 2 and 3 show the values of the model

parameters obtained for each feed size using Equation (6), for quartz and marble respectively.

The results show that the type of the material, the amount of energy required per ton of

material (specific energy, ), as well as the feed size, affect the particle size distribution. The

11
accuracy of the GGS model was assessed using the correlation coefficient R2. t is observed that

the model cannot describe with high accuracy the data across the entire range of the produced

particle sizes. This is also shown in Figure 3 which presents data derived by the use of Equation

(6). It is shown that the size distribution of quartz grinding products is not linear and deviates

from the straight line for coarser particles. Figure 3, which is used as example, presents data for

the 3.35 + 2.36 mm feed fraction of quartz after 0.5 min of grinding (equals to 0.29 kWh/t). The

same trend is observed after longer grinding times, as illustrated in Figure 4. The GGS plots for
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

marble show similar behavior. Based on these, data it is deduced that the particle size

distributions do not belong to one population of particle sizes.

Application of Piecewise Linear Regression

Figures 3 and 4 show the presence of two different domains of particle sizes. Since the

data points deviate from linearity, the predicted straight lines are unable to fully describe the

particle size distribution of the comminution products. These domains can be clearly identified

by applying piecewise regression analysis, using Equation (9).

The GGS model parameters for quartz and marble using piecewise regression analysis by

the Quasi-Newton non-linear estimation method are presented in Tables 4 and 5. In these Tables

the breakpoints with the highest correlation coefficient R2 were also estimated for each dataset.

Obviously, in all cases the values of R2 are higher compared to those shown in Tables 2 and 3.

Piecewise regression analysis was also found to improve the accuracy of the GGS model, during

comminution of chromite ores in different devices (Tademir and Tademir 2009).

Figure 5 shows the GGS plot of particle size distribution for quartz after 0.5 min of

grinding using piecewise regression analysis. The evolution of the particle size distribution in

12
relation to the grinding time, which is proportional to the energy input, is also shown in Figure

6. It is seen from these figures that two different linear regions exist, indicating the action of two

distinct breakage mechanisms. The first one is associated with the finer, while the second one to

the coarser fraction of products. As seen in Figure 6, the first domain (fine fraction) shifted

almost parallel with increasing grinding time, while for the second domain (coarse fraction) a

change in the slope of the straight lines is observed. It is known that breakage of particles in

tumbling mills is achieved mainly by impact, compression or attrition. When a single particle is
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

broken by compression the products fall into two distinct size ranges, namely coarse particles

resulting from the induced tensile failure, and fines from compressive failure near the points of

loading, or by shear at projections (Wills and Napier-Munn 2006). In impact crushing, a particle

experiences higher average stress, that is necessary to enable simple fracture, and thus tends to

break apart rapidly, mainly by tensile failure. Attrition causes steady erosion of comparatively

smooth surfaces of lumps, due to friction of surfaces being in relative motion. The attrition

process generates a significant number of particles that are much finer than the parent size (King

2001). During milling of a mono-sized feed fraction, breakage is accomplished by a combination

of mechanisms and results in the production of two domains of particle sizes. As milling goes on

more fines are produced. This is also shown by plotting the breakpoint xb, connecting the two

domains of particle sizes, with the specific energy required to reduce each fraction (Figures 7

and 8). It is obvious that the breakpoint xb is reduced when specific energy is increased and

reaches a limit value. xb also depends on material type since for a given specific energy, e.g.

1 kWh/t, the values of xb, are higher for quartz than those of marble, for each feed fraction. The

latter indicates that marble is ground more easily that quartz. It is also pointed out that abnormal

13
grinding of quartz is observed, since for the coarse feed fraction, 3.35 + 2.36 mm, the xb values

are slightly smaller than the ones for the 1.7 + 1.18 mm fraction.

Figures 9 and 10 show the xb values obtained for each feed fraction as a function of size

modulus xf. It is revealed that the two variables are mainly very well and in some cases

moderately correlated. This means that the domain of fine particle sizes is closely related to the

coarse one. These figures also indicate that grinding of the coarse feed fraction of quartz is

abnormal.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

The Relationship Between Energy and Feed Size

It is seen from Tables 4 and 5 that piecewise regression analysis is more suitable for

describing the particle size distribution for quartz and marble comminution products. It is also

revealed that, for the same material and feed size, the f value remains constant or almost

constant for different specific energies, whilst the c value decreases with increasing energy

input. On the other hand, it is observed that for each energy input, as the size of feed decreases

the values of f and c become higher. Regarding the effect of material type, their respective

values are lower in the case of marble. As already mentioned, lower values of indicate the

production of more fines, more large particles and fewer particles in the middle range, while the

higher the value of the narrower is the size distribution.

Using Equations (7) and (8) the size moduli xf or xc can be calculated for the two different

domains of particles sizes. Size modulus denotes the theoretical maximum size of particles in the

assembly and corresponds to the size at which each straight line intercepts the 100% passing

level. The value of xc is supposed to remain constant for the same feed size and is independent of

the energy input, thus only the determination of xf requires investigation. Variations of xf values

14
for each feed size and at different energy inputs are presented in Tables 6 and 7, for quartz and

marble respectively. For the same material and the same feed size the value of xf decreases with

increasing energy input, suggesting obviously that particle size is reduced as grinding proceeds.

By comparing the two materials tested, it can be said that the size distribution of marble

comminution products is shifted to smaller sizes for all energy levels.

Figures 11 and 12 show the specific energy values versus the size modulus xf for quartz

and marble, respectively. These figures indicate the specific energy required to reduce each
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

initial size fraction and produce an assembly of particles with a size modulus xf, where the index

f refers to the finer domain of products. It is observed from the trendlines, that the relationship

between specific energy and size modulus can be described by Equation (16)

C x f b (16)

where C' and b are functions that depend on the material, as well as the feed size.

Generally, it is shown that for the same material the C' value decreases with decreasing

feed size, while the b value increases with decreasing feed size. The units for C' are the same

with the units of specific energy (kWh/t), while (b) is the slope of the straight line and is equal

to the exponents proposed by Charles and Stamboliadis (Equations (2) and (3)). C' and b values

are described with high accuracy as a function of the feed size by fitted logarithmic trendline

equations, with R2 values greater than 0.97. Thus, the general form of Equation (16) can be

expressed as a function of feed size xi, as follows

For quartz 1.05 lnxi 1.33 x f 0.54lnxi 1.36 (17)

For marble 0.50 lnxi 0.88 x f 0.33lnxi 1.04 (18)

15
It is noted that the feed size is supposed to be a mono-sized fraction and xi (mm)

corresponds to the upper size of the size fraction i.

Equations (17) and (18) are alternative expressions of the energy-size reduction

relationship, which derive from the general form of Equation (16), while (b) depends on the

material type and the feed size. These equations consider the presence of two different domains

of particle sizes and are applicable to sizes finer than the breakpoint xb. Despite this latter

inability, this study shows that the establishment of the relationship between size modulus xf and
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

breakpoint xb, as shown in Figures 9 and 10, enables the full definition of the relationship

between energy input and particle size distribution of the milling products.

Table 8 shows that the existing theories apply only to a specific feed size. It is observed

that for larger feed sizes (around 12 and 23 mm for quartz and marble, respectively) the energy

required for breakage is proportional to the size or volume (x3) of the material, as Kick suggests.

On the other hand, for smaller fractions (around 5 mm) the energy is proportional to the size

(x2.5), as Bond proposes, while for finer fractions (around 12 mm) the energy is proportional to

the new surface area (x2), as Rittingers theory indicates. Figure 13 shows that with the use of

Equations (17) and (18) the relationship between feed size xi and exponent n can be determined.

It is revealed that for the feed sizes examined in the present study, the effect of the material type

on energy consumption is not significant. It is expected that the slope of the curve of Figure 13

will be higher as the material becomes softer, while n will take a higher than 1.25 value.

Interrelation Between f and (1 m) or (2n 3)

Tables 9 and 10 present the interrelation between the size modulus f and the exponent of

Equation (3) for each feed size and various specific energies, for the two materials tested. The

16
exponent of Equation (2) is equal to the one of Equation (3) and for this reason the exponent m is

not included in the Tables. Furthermore, it has been shown that the exponents are equal to the

feed depended (b) slope. As a result, n can be determined for each feed size and various

specific energies. In addition, when piecewise regression analysis is used, the slope f can be also

determined from the GGS size distribution plots. Hence, the value of the term (2n 3 + f) is

well determined for the materials tested. Results can be summarized as follows:

i. Equation (3) is valid if > (2n 3) and thus the term (2n 3 + f) is supposed to be
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

positive. This was not shown in this study except for the case of quartz, when coarser

feed size was used. No theoretical explanation can be provided at present, but it can be

said that this term was calculated indirectly as the difference between the measured

terms, (2n 3) and f, and thus the experimental error bears on both terms. In this study,

the use of piecewise regression analysis indicates that the calculation of the slope f was

accurate.

ii. Despite the fact that for the same feed size the value of f changes as grinding proceeds, it

can be considered as constant because of its little variation. Hence, the average values of

f can be determined for each feed size. Figures 14 and 15 show the specific energy

versus size modulus xf for quartz and marble, respectively, thus enabling the investigation

of the variation of f. It is shown that the energy required for the production of an

assembly of particles with the same size modulus xf increases as f decreases and

consequently, the energy demand for breakage becomes higher. Also, Tables 9 and 10

show that there is an inversely proportional relationship between f and the term

(2n 3 + f). Based on these findings, it is deduced that the energy demand increases

when the term (2n 3 + f) decreases. Stamboliadis (2002) concluded that the term

17
(2n 3 + f) is affected by the type of the material; for example, in materials with weak

structure, which require less energy demand for breakage, this term obtains a relatively

large value. However, in this study, it was revealed that even for the same material the

energy required for size reduction influences this term.

iii. The same conclusions can be drawn for all feed size fractions and energy inputs.

Obviously, as the specific energy increases the term (2n 3 + f) decreases, thus

indicating the increasing difficulty for breakage as the material becomes finer.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

iv. Quartz is much harder than marble and this is the main the reason for its higher energy

demand during grinding. Thus, one would expect that the term (2n 3 + f) would be

smaller in the case of quartz. However, this applies only in the case of the two finer feed

sizes examined (0.850 + 0.600 mm and 0.425 + 0.300 mm). To address this paradox,

its size reduction was investigated in this study using approaches based on kinetic

models. Kinetic models use the concept of specific breakage rate, which is the rate with

which a material of a defined size fraction breaks out to finer fractions. Variations of

specific breakage rates of the materials tested at various feed size fractions showed that

there is an optimum size (around 1.7 mm) at which this rate is maximized. Above this

value breakage becomes abnormal, because the particles are too large to be nipped and

fractured properly by the grinding media used. This is the reason why the two coarser

feed size fractions used behave abnormally during grinding.

CONCLUSIONS

Batch grinding experiments using quartz and marble were performed in a ball mill to determine

the relationship between energy input and size reduction. The results obtained show the existence

of two different domains of particle sizes, thus a single linear model, namely the GGS model,

18
cannot describe with adequate accuracy the particle size distribution across the entire range of

particle sizes. On the other hand, this study proved that piecewise regression analysis is an

accurate approach for describing the particle size distribution of the comminution products. With

the use of this approach the GGS model parameters, namely f and xf, were determined. It is

mentioned that in general, the f value remains constant or almost constant for all energy demand

levels. Also, it is observed that for each energy input, the value of f increases as the feed size

decreases. As regards the effect of the material type, the f value for quartz is higher, indicating
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

that the size distribution of the comminution products of harder materials is more narrow.

By plotting the specific energy versus size modulus xf, it is shown that a power-law can

describe the relationship between energy input and particle size. The resulting curves depend

both on material type and feed size. This power-law is an improved energy particle size

relationship and its main novelty is the consideration of two different domains of particle sizes,

as a result of the involvement of two distinct breakage mechanisms. Despite the fact that this law

applies to sizes finer than the breakpoint xb, this study shows that the establishment of the

relationship between size modulus xf and breakpoint xb, enables the full definition of the

relationship between energy input and particle size distribution of the milling products.

he present study, with the use of piecewise regression analysis, has proven that after

comminution two populations of grinding products are obtained and also made clear that the

produced particle size distributions do not follow the single linear GGS model. This finding is in

contrast with the existing theories, which assume that a particle of size x is continuously reduced

by dx, without breaking into smaller daughter particles.

The proposed power-law also shows that the existing theories of Rittinger, Kick and

Bond are valid for a specific feed size. Moreover, it is revealed that for the size range for which

19
the theories of Rittinger and Bond apply (n equals 1 or 1.25, respectively), the effect of the

material type on energy consumption is considered minor.

Furthermore, this study revealed that, even for the same material, the energy required for

size reduction influences the term (2n 3 + f), as opposed to earlier studies which mention that

this term is only affected by the material type. Finally, grinding kinetics shows that when the

feed size is above nearly 1.7 mm breakage becomes abnormal and the effect of the material type

on energy demand cannot be accurately assessed.


Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

References

Ahmed, H. A. M., and J. Drzymala. 2005. Two-dimensional fractal linearization of distribution


curves. Physicochemical Problems of Mineral Processing 39:12939.
Allen, T. 2003. Powder sampling and particle size determination. Amsterdam: Elsevier.
Austin, L. G., R. R. Klimpel, and P. T. Luckie. 1984. Process engineering of size reduction: Ball
milling. New York: AIME, SME.
BCS. 2002. Energy and environmental profile of the U.S. mining industry. Mining industry of
the future. Technical report, BCS Inc. for the U.S. Department of Energy, Office of
Energy Efficiency and Renewable Energy. December 2002. Available at
http://www1.eere.energy.gov/manufacturing/resources/mining/pdfs/overview.pdf
Bilgili, E. 2007. On the consequences of non-first-order breakage kinetics in comminution
processes: Absence of self-similar size spectra. Particle & Particle Systems
Characterization 24:1217. doi:10.1002/ppsc.200601043
Bond, F. C. 1952. The third theory of comminution. Transactions of the AIME 193:48494.
Carvalho, R. M., and M. Tavares. 2013. Predicting the effect of operating and design variables
on breakage rates using the mechanistic ball mill model. Minerals Engineering 43
44:91101. doi:10.1016/j.mineng.2012.09.008
Charles, R. J. 1957. Energy-size reduction relationships in comminution. Transactions of the
AIME 208:8088.
Danha, G., D. Hildebrandt, D. Glasser, and C. Bhondayi. 2015. A laboratory scale application of
the attainable region technique on a platinum ore. Powder Technology 274:1419.
doi:10.1016/j.powtec.2014.12.048
Deniz, V. 2012. The effects of ball filling and ball diameter on kinetic breakage parameters of
barite powder. Advanced Powder Technology 23:64046. doi:10.1016/j.apt.2011.07.006
Deniz, V. 2013a. Comparisons of dry grinding kinetics of lignite, bituminous coal and petroleum
coke. Energy Sources, Part A: Recovery, Utilization and Environmental Effects 35:913
20. doi:10.1080/15567036.2010.514591

20
Deniz, V. 2013b. Effects of mill speed on kinetic breakage parameters of four different
particulate pumices. Particulate Science and Technology 31:10108.
doi:10.1080/02726351.2012.658903
Djordjevic, N. 2010. Improvement of energy efficiency of rock comminution through reduction
of thermal losses. Minerals Engineering 23:123744. doi:10.1016/j.mineng.2010.08.019
Gupta, V. K., and S. Sharma. 2014. Analysis of ball mill grinding operation using mill power
specific kinetic parameters. Advanced Powder Technology 25:62534.
doi:10.1016/j.apt.2013.10.003
Gupta, V. K., H. Zouit, and D. Hodouin. 1985. The effect of ball and mill diameters on grinding
rate parameters in dry grinding operation. Powder Technology 42:199208.
doi:10.1016/0032-5910(85)80054-1
Ian, B. -P., J. -P. Bailon, and L. G. L'Esprance. 2011. Ball mill grinding kinetics of master
alloys for steel powder metallurgy applications. Powder Technology 210:26772.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

doi:10.1016/j.powtec.2011.03.028
Ipek, H., and F. Gktepe. 2011. Determination of grindability characteristics of zeolite.
Physicochemical Problems of Mineral Processing 47:18392.
Kapur, P. C. 1972. Self-preserving size spectra of comminuted particles. Chemical Engineering
Science 27:42531. doi:10.1016/0009-2509(72)85079-6
Katubilwa, F. M. 2008. Effect of ball size distribution on milling parameters. Master of science
diss., University of the Witwatersrand, Johannesburg.
Katubilwa, F. M., and M. H. Moys. 2009. Effect of ball size distribution on milling rate.
Minerals Engineering 22:128388. doi:10.1016/j.mineng.2009.07.008
Katubilwa, F. M., M. Moys, H. Michael, D. Glasser, and D. Hildebrandt. 2011. An attainable
region analysis of the effect of ball size on milling. Powder Technology 210:3646.
doi:10.1016/j.powtec.2011.02.009
Khumalo, N., D. Glasser, D. Hildebrandt, and B. Hausberger. 2008. Improving comminution
efficiency using classification: an attainable region approach. Powder Technology
187:25259. doi:10.1016/j.powtec.2008.03.001
Kick, F. 1885. Das Gesetz der proportionalen Widerstnde und seine Anwendungen. Leipzig:
Felix.
King, R. P. 2001. Modeling and simulation of mineral processing systems. Oxford: Butterworth-
Heinemann.
Lu, P., I. F. Jefferson, M. S. Rosenbaum, and I. J. Smalley. 2003. Fractal characteristics of loess
formation: Evidence from laboratory experiments. Engineering Geology 69:28793.
doi:10.1016/s0013-7952(02)00287-9
Matijai, G., and S. Kurajica. 2010. Grinding kinetics of amorphous powder obtained by solgel
process. Powder Technology 197:16569. doi:10.1016/j.powtec.2009.09.010
Morrell, S. 1996. Power draw of wet tumbling mills and its relationship to charge dynamics
Part 2: An empirical approach to modelling of mill power draw. Transactions of the
Institution of Mining and Metallurgy (Sect C) 105:C5462.
Morrell, S. 2008. A method for predicting the specific energy requirement of comminution
circuits and assessing their energy utilisation efficiency. Minerals Engineering 21
(3):22433. doi:10.1016/j.mineng.2007.10.001
Olejnik, T. P. 2013. Selected mineral materials grinding rate and its effect on product
granulometric composition. Physicochemical Problems of Mineral Processing
49:40718.

21
Qian, H. Y., Q. G. Kong, and B. L. Zhang. 2013. The effects of grinding media shapes on the
grinding kinetics of cement clinker in ball mill. Powder Technology 235:42225.
doi:10.1016/j.powtec.2012.10.057
Ramakrishnan, K. N. 2000. Fractal nature of particle size distribution. Journal of Materials
Science Letters 19:107780.
Rittinger, P. R. 1867. Lehrbuch der Aufbereitungskunde. Berlin: Ernst and Korn.
Shin, H., S. W. Lee, H. S. Jung, and J. B. Kim. 2013. Effect of ball size and powder loading on
the milling efficiency of a laboratory-scale wet ball mill. Ceramics International
39:896368. doi:10.1016/j.ceramint.2013.04.093
Stamboliadis, E. 1996. The relationship of energy and particle size in comminution. Min. Metall.
Ann. 6:921 (in Greek).
Stamboliadis, E. 2000. The relationship of energy and particle size in comminution. In 3rd
conference for mineral wealth, Technical Chamber of Greece, Athens, 25161.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

Stamboliadis, E. 2002. A contribution to the relationship of energy and particle size in the
comminution of brittle particulate materials. Minerals Engineering 15:70713.
doi:10.1016/s0892-6875(02)00185-1
Stamboliadis, E. 2003. Impact crushing approach to the relationship of energy and particle size
in comminution. European Journal of Mineral Processing & Environmental Protection
3:16066.
Stamboliadis, E., S. Emmanouilidis, and E. Petrakis. 2011. A new approach to the calculation of
work index and the potential energy of a particulate material. Geomaterials 1:2832.
doi:10.4236/gm.2011.12005
Tademir, A. 2009. Fractal evaluation of particle size distributions of chromites in different
comminution environments. Minerals Engineering 22:15667.
doi:10.1016/j.mineng.2008.06.002
Tademir, A., H. Ozdag, and G. Onal. 2011. Image analysis of narrow size fractions obtained by
sieve analysis. An evaluation by log-normal distribution and shape factors.
Physicochemical Problems of Mineral Processing 46:95106.
Tademir, A., and T. Tademir. 2009. A comparative study on PSD models for chromite ores
comminuted by different devices. Particle & Particle Systems Characterization 26:69
79. doi:10.1002/ppsc.200800035
Turner, R. R. 1982. Selection and sizing of primary autogenous and semi-autogenous grinding
mills. In Design and installation of comminution circuits, ed. A. Mular and G. Jergensen,
47490. AIME.
Umucu, Y., M. Y. Altnine, and V. Deniz. 2014. The effects of ball types on breakage
parameters of barite. Journal of the Polish Mineral Engineering Society (Inzynieria
Mineralna) 15:11317.
Xia, W., J. Yang, Y. Zhao, B. Zhu, and Y. Wang. 2012. Improving floatability of taixi anthracite
coal of mild oxidation by grinding. Physicochemical Problems of Mineral Processing
48:393401.
Xia, W., J. Yang, and B. Zhau. 2012. Flotation of oxidized coal dry-ground with collector.
Powder Technology 228:32426. doi:10.1016/j.powtec.2012.05.043
Walker, D. R., and M. C. Shaw. 1954. A physical explanation of the empirical laws of
comminution. Transactions of the AIME 199:31320.

22
Wang, X., W. Gui, C. Yang, and Y. Wang. 2011. Wet grindability of an industrial ore and its
breakage parameters estimation using population balances. International Journal of
Mineral Processing 98:11317. doi:10.1016/j.minpro.2010.11.008
Wills, B. A., and T. Napier-Munn. 2006. Wills mineral processing technology. An introduction
to the practical aspects of ore treatment and mineral recovery, 7th ed. London: Elsevier
Science & Technology Books.
Zhang, Z. L., J. G. Yang, L. H. Ding, and M. Zhaoy. 2012. An improved estimation of coal
particle mass using image analysis, Powder Technology 229:17884.
doi:10.1016/j.powtec.2012.06.027
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

23
Table 1. Grinding conditions

Mill Diameter, D (cm) 20.4

Length, L (cm) 16.6

Volume, V (cm3) 5,423

Operational speed, (rpm) 66

Critical speed, c (rpm) 93.7


Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

Balls Diameter, d (mm) 25.4

Number 77

Weight (g) 5,149

Specific gravity (g/cm3) 7.85

Porosity (%) 40

Ball filling volume, J (%) 20

Material Specific gravity (g/cm3) Quartz (2.65)

Marble (2.7)

Material filling volume, fc (%) 4

Interstitial filling, U 0.5

24
Table 2. GGS model parameters for quartz

Tim 3.35 + 2.36 mm 1.7 + 1.18 mm 0.850 + 0.600 m 0.425 + 0.300 m

e (kWh/t m m

(min )
x0 R2 x0 R2 x0 R2 x0 R2
)
(m) (m) (m) (m

)
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

0.5 0.29 1.0 8,88 0.94 1.1 3,14 0.96 1.3 1,58 0.94 1.6 640 0.94

6 8 5 9 7 5 1 3 1 8 0

1 0.58 0.9 6,04 0.97 1.0 2,30 0.99 1.1 1,20 0.97 1.4 581 0.96

2 0 8 4 4 0 5 7 7 8 3

2 1.16 0.7 3,80 0.99 0.8 1,59 0.99 0.9 905 0.99 1.2 505 0.98

8 8 6 9 7 8 9 6 6 8

4 2.31 0.6 2,24 0.97 0.7 1,09 0.97 0.7 669 0.97 1.0 427 1.00

3 3 3 1 2 0 8 8 2 0

8 4.63 0.4 1,37 0.86 0.5 937 0.93

4 3 3 7 7

25
Table 3. GGS model parameters for marble

Tim 3.35 + 2.36 mm 1.7 + 1.18 mm 0.850 + 0.600 m 0.425 + 0.300 m

e (kWh/t m m

(min )
x0 R2 x0 R2 x0 R2 x0 R2
)
(m) (m) (m) (m

)
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

0.5 0.28 0.7 7,09 0.96 0.8 2,72 0.97 0.9 1,49 0.94 1.2 700 0.91

4 8 0 3 2 8 8 2 8 6 8

1 0.57 0.6 4,13 0.99 0.7 1,81 0.99 0.8 1,07 0.98 1.0 552 0.97

3 9 2 3 1 8 5 9 9 6 6

2 1.14 0.5 2,22 0.97 0.5 1,20 0.98 0.6 770 0.99 0.8 458 0.99

0 4 7 8 1 1 9 7 7 8

4 2.27 0.4 946 0.93 0.4 623 0.96 0.4 587 0.96 0.6 387 0.99

1 0 6 4 9 5 4 6

8 4.54 0.3 400 0.93

6 0

26
Table 4. GGS model parameters using piecewise regression analysis for quartz

Ti 3.35 + 2.36 mm 1.7 + 1.18 mm 0.850 + 0.600 mm 0.425 + 0.300 mm

m (k
f lo xb R f lo xb R f lo xb R f lo xb R
e W 2 2 2 2
c g( m c g( m c g( m c g( m
( h/t
xb) m xb) m xb) m xb) m
m )

in
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

0. 0.2 0 2 3. 1. 0. 1 3 3. 1. 0. 1 4 2. 0. 0. 1 4 2. 0. 0.

5 9 . . 19 5 9 . . 21 6 9 . . 80 6 9 . . 41 2 9

8 9 4 6 9 0 2 3 3 9 1 1 2 3 9 2 1 8 6 9

6 2 4 8 3 1 5 9 2 2 4 8 8 7 2 9

1 0.5 0 1 3. 1. 0. 0 1 3. 1. 0. 1 2 2. 0. 0. 1 3 2. 0. 1.

8 . . 18 5 9 . . 13 3 9 . . 82 6 9 . . 41 2 0

8 9 7 3 9 9 8 0 5 9 0 6 6 7 9 2 1 1 5 0

1 1 8 9 6 2 0 9 5 8 0 6 0 2 8 0

2 1.1 0 0 2. 0. 0. 0 0 2. 0. 0. 0 0 2. 0. 0. 1 1 2. 0. 0.

6 . . 90 8 9 . . 95 9 9 . . 65 4 9 . . 49 3 9

7 7 3 0 9 8 8 4 0 9 9 9 3 5 9 1 9 1 1 9

8 8 0 9 9 9 0 8 9 9 0 6 7 9 0 8

4 2.3 0 0 2. 0. 0. 0 0 2. 0. 0. 1 0 2. 0. 0. 1 1 2. 0. 1.

. . 52 3 9 . . 80 6 9 . . 47 3 9 . . 41 2 0

27
1 8 4 9 3 9 8 1 5 3 9 1 6 7 0 9 0 0 5 6 0

4 3 8 9 3 8 8 6 1 2 0 5 2 2 0 0

8 4.6 0 0 2. 0. 0. 0 0 2. 0. 0.

3 . . 55 3 9 . . 58 3 9

7 0 0 5 9 7 1 2 8 9

8 9 5 6 9 7 2 7
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

28
Table 5. GGS model parameters using piecewise regression analysis for marble

Ti 3.35 + 2.36 mm 1.7 + 1.18 mm 0.850 + 0.600 mm 0.425 + 0.300 mm

m (k
f lo xb R f lo xb R f lo xb R f lo xb R
e W 2 2 2 2
c g( m c g( m c g( m c g( m
( h/t
xb) m xb) m xb) m xb) m
m )

in
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

0. 0.2 0 1 3. 1. 0. 0 1 2. 0. 0. 0 2 2. 0. 0. 0 3 2. 0. 1.

5 8 . . 14 3 9 . . 90 7 9 . . 72 5 9 . . 44 2 0

6 6 2 8 9 7 5 0 9 9 8 8 4 3 9 9 7 4 7 0

1 7 7 9 2 6 4 8 0 3 0 9 3 8 8 0

1 0.5 0 0 3. 1. 0. 0 0 2. 0. 0. 0 1 2. 0. 0. 0 2 2. 0. 1.

7 . . 07 1 9 . . 81 6 9 . . 61 4 9 . . 43 2 0

5 8 2 8 9 7 7 3 5 9 7 2 1 0 9 9 1 9 7 0

9 3 0 7 3 3 0 8 7 8 8 9 1 5 5 0

2 1.1 0 0 2. 0. 0. 0 0 2. 0. 0. 0 0 2. 0. 0. 0 0 2. 0. 1.

4 . . 87 7 9 . . 68 4 9 . . 47 3 9 . . 39 2 0

6 3 5 5 9 6 3 9 8 9 6 6 7 0 9 8 9 8 5 0

5 7 0 7 7 2 9 7 9 9 0 7 4 2 0 0

4 2.2 0 0 2. 0. 0. 0 0 2. 0. 0. 0 0 2. 0. 0. 0 0 2. 0. 0.

. . 55 3 9 . . 53 3 9 . . 40 2 9 . . 37 2 9

29
7 5 0 7 6 9 5 1 7 4 9 6 2 3 5 9 6 4 9 3 9

9 9 0 3 7 0 4 4 3 4 3 7 8 5 9 9

8 4.5 0 0 2. 0. 0.

4 . . 37 2 9

4 0 8 3 9

9 2 9 5
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

30
Table 6. GGS size modulus xf for quartz

Time xf (mm)

(min (kWh/t
3.35 + 2.36 m 1.7 + 1.18 m 0.850 + 0.425 m 0.425 + 0.300 m
) )
m m m m

0.5 0.29 23.3 5.4 2.6 1.3

1 0.58 10.0 2.9 1.5 1.0


Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

2 1.16 3.8 1.6 0.9 0.6

4 2.31 1.0 0.8 0.6 0.4

8 4.63 0.4 0.5

31
Table 7. GGS size modulus xf for marble

Time xf (mm)

(min (kWh/t
3.35 + 2.36 m 1.7 + 1.18 m 0.850 + 0.425 m 0.425 + 0.300 m
) )
m m m m

0.5 0.28 16.3 4.4 2.8 1.6

1 0.57 5.1 1.8 1.4 0.8


Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

2 1.14 1.1 0.9 0.8 0.5

4 2.27 0.5 0.4 0.4 0.3

8 4.54 0.3

32
Table 8. Upper feed size for which the existing theories apply

Existing theories Exponent Upper feed size, mm

m n Quartz Marble

Rittinger 2 1 1.95 1.13

Bond 1.5 1.25 4.92 5.14

Kick 1 1.5 12.41 23.37


Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

33
Table 9. Interrelation between f and (2n 3) for each feed size fraction and at various specific
energies for quartz

3.35 + 2.36 mm 1.7 + 1.18 mm 0.850 + 0.600 mm 0.425 + 0.300 mm

(kW
2n n f 2n 2n n f 2n 2n n f 2n 2n n f 2n
h/t)
3 3 + f 3 3 + f 3 3 + f 3 3 + f

0.29 0. 1. 0. 0.19 1. 0. 1. 0.11 1. 0. 1. 0.28 1. 0. 1. 0.55

7 1 8 1 9 0 4 8 1 8 5 2
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

7 6 3 3 0 2 9 8

0.58 0. 1. 0. 0.14 1. 0. 0. 0.18 1. 0. 1. 0.35 1. 0. 1. 0.63

7 1 8 1 9 9 4 8 0 8 5 2

7 1 3 6 0 5 9 0

1.16 0. 1. 0. 0.11 1. 0. 0. 0.26 1. 0. 0. 0.41 1. 0. 1. 0.66

7 1 7 1 9 8 4 8 9 8 5 1

7 8 3 9 0 9 9 7

2.31 0. 1. 0. 0.17 1. 0. 0. 0.31 1. 0. 0. 0.43 1. 0. 1. 0.81

7 1 8 1 9 8 4 8 9 8 5 0

7 4 3 3 0 7 9 2

4.63 0. 1. 0. 0.11 1. 0. 0. 0.35

7 1 7 1 9 7

7 8 3 9

Ave 0. 0. 1. 1.

34
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

rage

1
8

0
9

35
3
0

7
1
Table 10. Interrelation between f and (1 m) for each feed size fraction and at various specific
energies for marble

3.35 + 2.36 mm 1.7 + 1.18 mm 0.850 + 0.600 mm 0.425 + 0.300 mm

(kW
2n n f 2n 2n n f 2n 2n n f 2n 2n n f 2n
h/t)
3 3 + f 3 3 + f 3 3 + f 3 3 + f

0.28 0. 1. 0. 0.02 0. 1. 0. 0.19 1. 0. 0. 0.25 1. 0. 0. 0.41

6 1 6 9 0 7 1 9 8 3 8 9
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

9 1 5 2 8 0 3 3

0.57 0. 1. 0. 0.04 0. 1. 0. 0.18 1. 0. 0. 0.28 1. 0. 0. 0.43

6 1 5 9 0 7 1 9 7 3 8 9

9 9 5 3 8 7 3 1

1.14 0. 1. 0. 0.02 0. 1. 0. 0.24 1. 0. 0. 0.36 1. 0. 0. 0.50

6 1 6 9 0 6 1 9 6 3 8 8

9 5 5 7 8 9 3 4

2.27 0. 1. 0. 0.04 0. 1. 0. 0.34 1. 0. 0. 0.42 1. 0. 0. 0.66

6 1 5 9 0 5 1 9 6 3 8 6

9 9 5 7 8 3 3 8

4.54 0. 1. 0. 0.14

6 1 4

9 9

Ave 0. 0. 0. 0.

36
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

rage

9
5

7
6

37
2
7

4
8
Figure 1. Compact mass of quartz. Sample VP30, reflected light, // Nicols.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

38
Figure 2. Coarse grains of calcite with polysynthetic twinning. Sample VP15, transmitted light,
X Nicols.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

39
Figure 3. GGS plot of particle size distribution for quartz after 0.5 min of grinding.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

40
Figure 4. GGS plots of particle size distributions for quartz for different grinding times.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

41
Figure 5. GGS plot of particle size distribution for quartz after 0.5 min of grinding, using
piecewise regression analysis.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

42
Figure 6. GGS plots of particle size distributions for quartz after different grinding times, using
piecewise regression analysis.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

43
Figure 7. Breakpoint xb versus specific energy at each feed size for quartz.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

44
Figure 8. Breakpoint xb versus specific energy at each feed size for marble.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

45
Figure 9. Breakpoint xb versus size modulus xf at each feed size for quartz.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

46
Figure 10. Breakpoint xb versus size modulus xf at each feed size for marble.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

47
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

48
Figure 11. Specific energy versus size modulus xf for quartz.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

49
Figure 12. Specific energy versus size modulus xf for marble.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

50
Figure 13. Relationship between n and upper feed size fraction.
Figure 14. Specific energy versus size modulus xf at different f values, for quartz.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

51
Figure 15. Specific energy versus size modulus xf at different f values, for marble.
Downloaded by [Orta Dogu Teknik Universitesi] at 03:04 10 May 2016

52