Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Powder Technology 359 (2020) 305–313

Contents lists available at ScienceDirect

Powder Technology

journal homepage: www.elsevier.com/locate/powtec

Wet and dry grinding of coal in a laboratory-scale ball mill:


Particle-size distributions
Xiangning Bu a,⁎, Yuran Chen a, Guangxi Ma a, Yujin Sun b, Chao Ni a,c,⁎⁎, Guangyuan Xie a
a
Key Laboratory of Coal Processing and Efficient Utilization (Ministry of Education), School of Chemical Engineering and Technology, Xuzhou, 221116, China
b
College of Mining Engineering, Taiyuan University of Technology, Taiyuan, 030024, China
c
Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta T6G 1H9, Canada

a r t i c l e i n f o a b s t r a c t

Article history: In this study, experimental data for wet- and dry-ground coal samples under wet and dry grinding are
Received 11 June 2019 characterized by commonly used distribution functions. First, both the R-R and Swrebec functions have
Received in revised form 28 August 2019 superior fitting performances for cumulative particle size curves compared to the other studied func-
Accepted 18 September 2019
tions. On this basis, a time-dependent expression is drawn to describe the cumulative particle size distri-
Available online 6 October 2019
bution. Second, the R-R function produces a significantly superior fit to the relative mass distributions of
Keywords:
the ground products compared to those of the others at a short grinding time. The goodness of fit for all
Particle size distribution distribution functions studied performs marginally worse at approximately 3 min, which can be associ-
Coal ated with a change in the dominant breakage mechanisms from impact to abrasion-chipping. With an in-
Ball milling crease in the grinding time, the G-G-S function is the optimal function for characterizing the particle size
Wet grinding probability mass distributions of wet grinding, whereas the G-M function provides the best fitting perfor-
dry grinding mance when applied to the experimental dry-grinding data. Further, the optimal particle size probability
density functions are associated with the difference in breakage mechanisms between wet and dry
grinding.
© 2019 Elsevier B.V. All rights reserved.

1. Introduction distribution, pulp density, coal rank, the mode of grinding (wet or
dry), material hardness, and dissemination features [7–15]. The mode
Coal has an important role in electricity generation worldwide. of grinding, wet or dry, has significant differences in grinding kinetics
According to data from the World Coal Association, coal-fueled power and the properties of the ground products. Breakage rates (first-order
plants currently provide 37% of the global electricity [1]. In 2017, hypothesis) for wet grinding are faster than those for dry grinding
China's coal-fired power generation accounted for 64.5% of the total when the solid concentration is less than 50 vol %. This is attributable
power generation [2]. Nearly 1.76 billion tons coal produced in China to the interactions between the broken surface bonds and water mole-
is used for generating electricity in power plants [3,4]. Grinding is a nec- cules, the settling of the coarse particles in water, and the reduced cush-
essary operation for the preparation of finely powdered coal (b75 μm) ioning of the coarse particles in water [15–17]. However, the breakage
in power plants [5]. Particle size, in general, also has an influence on rates for wet grinding with higher slurry densities are significantly less
the setting time of cements, flow characteristics of granular materials, compared to those for dry grinding. As observed by Klimpel [18] and
compacting and sintering behavior of metallurgical powders, masking Tangsathitkulchai and Austin [15], the rapid increase in viscosity of
power of paint pigment, and flotation of coal and minerals [6]. Thus, it the slurry for wet grinding is apparent, where a layer of particles begins
is important to control the particle size or particle-size distribution to deposit around the inner mill wall. Thus, the contact efficiency be-
(PSD) in a ground product. tween the balls and particles reduces, resulting in a decrease in the
The efficiency of size reduction is influenced by a number of vari- breakage rate for wet grinding. It was concluded that the variations in
ables including the mill speed, ball size, filling rate, feed size solid concentrations of wet grinding are a determining factor in compar-
ing the breakage rates between wet and dry grinding. Further, the devi-
ation of breakage rates from the first-order hypothesis has been
⁎ Corresponding author. observed by numerous researchers [19–27]. This phenomena is related
⁎⁎ Corresponding author. Key Laboratory of Coal Processing and Efficient Utilization
(Ministry of Education), School of Chemical Engineering and Technology, China
to the energy transfer mechanisms in comminution machines whereby
University of Mining and Technology No1, Daxue Road, Xuzhou, Jiangsu, 221116, PR China. coarser particles could be ground preferentially or be protected by fines,
E-mail addresses: xiangning.bu@foxmail.com (X. Bu), sunnichao@126.com (C. Ni). by energy dissipation through inter-particle friction in compressed bed

https://doi.org/10.1016/j.powtec.2019.09.062
0032-5910/© 2019 Elsevier B.V. All rights reserved.
306 X. Bu et al. / Powder Technology 359 (2020) 305–313

quartz particle size distribution produced by ball milling were


minimal in wet grinding with small balls and were maximal in
dry grinding with small balls. A primary advantage of wet pro-
cesses compared with dry, for example, is in the production of
sub-micron particles, where it is easier to prepare a higher de-
gree of dispersion of fired (sintered) bauxite [34].
The majority of the published literature for wet and dry ball
milling is limited to grinding kinetics and the properties of the
ground products regarding liberation degree and fineness. The
results of PSD analysis can be expressed in terms of cumulative
percent oversize or undersize in relation to the diameters of the
particles (cumulative size distributions), or as a distribution of
the amounts or relative mass present in each of a defined num-
ber of size classes (probability density distribution) [9]. To the
best of our knowledge, the studies mentioned above have
never systematically discussed the PSD (especially for probabil-
ity density distribution) difference of coal particles between
wet and dry conditions. The aim of this study is to determine
the optimal distribution functions to characterize the cumulative
mass distributions of ground products obtained by wet and dry
grinding of mono-sized coal particles. On this basis, a time-
Fig. 1. Schematic of energy consumption, energy efficiency, breakage mechanisms, and
proper cumulative distribution functions at different grinding stages. Raw data sources
dependent form is established to describe the cumulative parti-
(energy consumption and efficiency) from Refs. [39,40]. cle size distribution of the ground product. Furthermore, the op-
timal particle size probability density functions are associated
with the difference in breakage mechanisms between wet and
dry grinding.
comminution, and possibly from heterogeneities produced in the feed
particles [28].
Researchers have also investigated the differences in the prop- 2. Particle size distribution function
erties of ground products. Yamamoto et al. [29] demonstrated
that the initial rates of the DNA (deoxyribonucleic acid) extrac- The selection of the best-fitting model depends highly on
tions by a dry-based method are greater in comparison to those both the data set being analyzed and the comminution devices
of the wet-based method owing to higher collision efficiency [35,36]. In previous research, Grady and Kipp [37] suggested
among the fine beads and fungal spores. Zhang et al. [30] ob- that the R-R (Rosin-Rammler) function is capable of characteriz-
served that the dry crushing method is considered as an advanta- ing the cumulative distribution of fragment fractions for impact
geous method in the crushing of spent lithium-ion batteries for comminution. Taşdemir and Taşdemir [36] compared the fitting
their subsequent recycling process. Experimental results provided performances of PSD functions selected for chromite ores com-
by Leewatchararongjaroen and Anuntagool [31] revealed that minuted by different devices. They found that the R-R function
wet-milled rice flour contains granules with smaller average was best at describing the cumulative particle size distribution
size compared to dry-milled samples. However, the dry-milling of the ground product generated by the hammer crusher and
process caused the destruction of the crystalline structure and ball mill data, whereas the G-G-S (Gates-Gaudin-Schumman)
yielded flour with reduced crystallinity compared to a wet- function provided the best fitting results for jaw and cone
milling process. Feng and Aldrich [32] investigated the effect of crusher. Wang and Forssberg [38] reported that the R-R function
dry and wet grinding on the flotation of complex sulfide ores. was an acceptable estimate of the cumulative PSD of ground
The ground particle surfaces for dry grinding were relatively products from stirred media mills. A schematic of the energy
rough, whereas the ground particle surfaces of wet grinding consumption, energy efficiency, breakage mechanisms, and
were smoother and cleaner. The dry-ground samples exhibited proper cumulative distribution functions at different grinding
more stable, higher loaded froths and faster flotation kinetics stages is summarized in Fig. 1.
owing to a high concentration of microstructural defects as com- From this, it can be concluded that the G-G-S function is preferable
pared to those of samples obtained from wet grinding. Kotake for fitting the cumulative profiles of product size distributions from
et al. [33] demonstrated that the product size and width of a crushers such as jaw and cone. The typical crushing is accomplished

Table 1
Summary of distribution functions.

Distribution Differential distribution function Cumulative distribution function Ref.


 m1 "   # "   #
Rosin-Rammler m d d
m
d
m [52]
f ðdÞ ¼ exp  F ðdÞ ¼ 1−exp −
l l l l
"  m # "  m #
Modified Rosin-Rammler m lnm−1 d ln d ln d [53]
f ðdÞ ¼ m exp − F ðdÞ ¼ 1−exp −
d ln l ln l ln l
 m
Gates-Gaudin-Schumman md
m−1
d [54–56]
f ðdÞ ¼ m
F ðdÞ ¼
l l
 m
Gaudin-Melody mðl−dÞ
m−1
d [57]
f ðdÞ ¼ m
F ðdÞ ¼ 1− 1−
l l
"    # a
Swrebec dmax dmax 1 [47]
f ðdÞ ¼ ln =ln F ðdÞ ¼ "    #a
d d50 dmax dmax
1 þ ln =ln
d d50
X. Bu et al. / Powder Technology 359 (2020) 305–313 307

Fig. 2. Cumulative grain size curves for wet and dry grinding.

Fig. 3. Comparisons of amount of fines produced by wet and dry grinding.

by slow compression of large particles (N1000 μm) against rigid surfaces the involvement of two distinct breakage mechanisms. Allen [48]
[39]. The cumulative size distribution of ground products more closely reviewed the most important two-parameter PSD functions such
follows the R-R function for hammer crushers, tumbling, and stirred as the R-R and G-G-S models. Several three- and four-parameter
mills. These devices combine impact, chipping, and abrasion events models have also been reported for greater accuracy in the charac-
caused by energy transferred from moving media such as hammers, terization of PSDs, such as the Swrebec function [49], modified
balls, rods, or large particles [39]. It should also be noted that the char- Nukiyama-Tanasawa distribution function [50], and Nukiyama-
acteristics of a PSD produced by comminution is also dependent upon Tanasawa equation [51]. However, their wide-spread application
the operating conditions, characteristics of the raw ore, comminution has been limited owing to their greater mathematical complexity
circuits, and other factors. [36,52]. Further, the normal distribution and logarithmic-normal
It is noted that Mandelbrot [40] introduced the fractal geometry functions are of minimal practical use because they typically fail
to describe the dimensional properties of irregular shapes, which
are produced in such a way that they cannot be described by
Euclidean geometry [41]. Although this concept is able to describe
complex natural phenomena by a simple exponential function, ac-
tually it is failed to describe the distribution of the comminution
products. The PSD of a ground product is used to evaluate the per-
formance of grinding circuits. It is well established that the PSD
curve of a ground product is typically similar (parallel) to the
PSD of the feed, which depends primarily on the properties of the
materials being ground [33,42,43]. Several PSD functions have
been developed to characterize fragment generation mathemati-
cally after grinding [44]. Lowrison [45] and Prasher [46] have al-
ready discussed on the cumulative particle size distribution as the
PSD functions (log-normal distribution, Rosin-Rammler-Bennett
equation (R-R), Gates-Gaudin-Schuhmann equation (G-G-S), and
Gaudin-Meloy equation (G-M). Petrakis et al. [47] observed that
the single linear GGS model is in some cases unable to fully de-
scribe the particle size distribution of comminution products.
They divided the G-G-S model into two straight lines, which indi- Fig. 4. Ash content of particles in top size interval (−3.0 + 1.0 mm) at different grinding
cates the presence of two domains of particle sizes and therefore times.
308 X. Bu et al. / Powder Technology 359 (2020) 305–313

Fig. 5. RMSE and Adj. R2 values produced by different cumulative distribution functions for wet grinding (means and error bars). R–R: Rosin-Rammler; MR-R: Modified Rosin-Rammler; G-
G-S: Gates-Gaudin-Schumman; G–M: Gaudin-Melody.

to fit the experimental data [53]. Therefore, five commonly used of −3.0 + 1.0 mm with a 7.58% ash was used for the grinding
distribution functions (see Table 1) were used to fit the differential tests. XRD analysis (Bruker D8 Advance, Karlsruhe, Germany) in-
or cumulative undersized distribution frequencies of the ground dicated that quartz was the major impurity mineral. The detailed
products under wet and dry ball milling. operating process of the XRD measurements is described in the
literature [54,55].
3. Experimental
3.2. Grinding tests
3.1. Materials
Grinding test were performed in an XMQ-Φ240 × 90 laboratory-
The sample used in this study was coking coal, which was col- scale conical ball mill (Wuan Exploring Machinery Factory, Wuhan,
lected from the overflow (clean coal) of the dense medium cy- China). For the wet- and dry-grinding tests, the ball diameter, frac-
clone in the Shanghaimiao Central Coal Preparation Plant tional ball filling, and fractional powder filling were 35 mm,
(Ordos, Inner Mongolia, China). A narrow size fraction 19.31%, and 6.80%. The solid concentration for the wet grinding

Fig. 6. RMSE and Adj. R2 values produced by different cumulative distribution functions for dry grinding (means and error bars).

Fig. 7. Goodness of fit of R-R and Swrebec functions versus the average product particle size.
X. Bu et al. / Powder Technology 359 (2020) 305–313 309

Fig. 8. Relationship between average product particle size and parameters (m and l) of R-R function.

Fig. 9. Relative mass probability density of ground products for wet and dry grinding.

was 70 vol %. The experimental procedure for wet grinding and the the ground products were calculated according to the calculating
detailed information regarding the ball mill are described in our pre- procedures described in the literature [51].
vious work [10]. After each dry-grinding test with a selected grind-
ing time, the mill was held sufficiently long to allow the fines to 4. Results and discussion
settle. Then, the balls were removed and the ground particles were
discharged into a container. The mill and balls were washed with 4.1. Cumulative size distributions
tap water to collect the adsorbed fines on their surface. The ground
Fig. 2 depicts the cumulative particle size curves for the wet and dry grinding
sample was sieved to determine the cumulative mass distributions
under different grinding times. As the grinding time increased from 1 min to
of ground products using a set of 1000, 500, 250, 125, 74, and 45 10 min, the dry grinding indicated a faster reduction of the average particle size
μm standard laboratory sieves. The relative mass distributions of (d 50 ) compared to the wet grinding. This is caused by the difference in breakage

Fig. 10. RMSE and Adj. R2 for different probability distribution functions: wet grinding.
310 X. Bu et al. / Powder Technology 359 (2020) 305–313

Fig. 11. RMSE and Adj. R2 for different probability distribution functions: dry grinding.

energy and mill environment between the wet and dry conditions. It has been dem- age of the top size interval is not greater than approximately 50%. According to the lit-
onstrated that the grinding energy in water (wet grinding) is approximately 20% erature [56–59], when the grinding time is chosen to provide not more than
less than in air (dry grinding) if the elastic strain energy is proportional to the square approximately 30%–40% broken from the topmost size interval, it is considered that
of the strength [16]. Moreover, the rapid increase in the viscosity of the slurry for wet there is minimal or no secondary breakage. In general, wet milling is more effective
grinding is clear, where a layer of particles begins to deposit on the inner mill wall at reducing the particle size of a large variety of materials to submicron scale than
[15,18]. Consequently, the contact efficiency between the balls and particles reduced dry milling [45]. However, with increased grinding time, the amount of fines gener-
considerably, resulting in a decrease in the breakage rate for the wet grinding (see ated from the secondary breakage for dry grinding becomes considerably greater
Fig. 3). It should be mentioned that wet grinding produces a greater amount of than that of the wet grinding, i.e., the main source of fine particles for the dry grinding
fines compared to dry grinding with a short time, and the breakage weight percent- is secondary breakage. The ash content of the wet-ground −3.0 + 1.0 mm coal parti-

Fig. 12. SEM micrographs of ground particles of −74 + 45 μm from wet grinding.

Fig. 13. SEM micrographs of ground particles of −74 + 45 μm from dry grinding.
X. Bu et al. / Powder Technology 359 (2020) 305–313 311

cles was greater than that of the feed. However, the ash content of the dry- products for wet grinding was best represented by the G-G-S distribution func-
ground −3.0 + 1.0 mm coal particles was slightly smaller than that of the feed (see tion. However, the G-M distribution function provided the best fitting perfor-
Fig. 4). This indicated that the higher grinding energy for dry grinding is conductive mance when it was applied to the experimental data for dry grinding. Fig. 12
to the breakage of the stronger mineral (relatively high-ash fraction). displays SEM micrographs of ground particles of −74 + 45 μm from wet grinding
Figs. 5 and 6 display the comparison of fitting performances among different cumula- under different times. The figure indicates that the surfaces of the ground prod-
tive distribution functions using root mean square error (RMSE) and adjusted R-squared ucts from wet grinding when the grinding time was 10 min were smoother than
(Adj. R2) under wet and dry grinding conditions. The means and errors of the cumulative those obtained at 2 min grinding time. The same result was confirmed under
distribution functions for wet and dry grinding were calculated according to the supple- dry conditions (Fig. 13). The smoother surface of the ground products is consid-
mental file. ered to be produced by abrasion mechanisms [64,65]. As provided in Fig. 3, in-
The assessment results indicate that both the R-R and Swrebec functions have a creasingly more fines are generated from the secondary breakage of the
superior fitting performance over the other studied functions. However, the aver- daughter fragments, which further promotes the occurrence of abrasion mecha-
age RMSE for the fitted data by the Swrebec function (0.0218) is marginally less nisms. Thus, it can be concluded that the smooth surfaces of the ground particles
than that of the R-R function (0.0249). Moreover, the average Adj. R 2 obtained correlate significantly with the deviations of the amount of fines and the presence
for the Swrebec function (0.9971) is marginally greater than that of the R-R func- of water, indicating that the breakage mechanisms change to provide more
tion (0.9950). These results of the goodness-of-fit for the compared functions are chipping and abrasion fracture with changes in grinding time and mill environ-
in accordance with the previous studies, i.e., the Swrebec function is marginally ment. In addition, the surfaces of the dry-ground particles were rougher com-
better for predicting the cumulative size distribution of the ground products in a pared with the wet-ground ones. Detailed description on the deference of
ball mill compared to the R-R function [49,60,61]. However, Liu et al. [62,63] re- morphological characterization in dry and wet grinding can be found in our pub-
ported that the R-R function provides a marginally greater determination coeffi- lished article [66].
cient for fitting a data set of copper tailing powder compared to that of the
Swrebec function.
5. Conclusions
To compare the fitting performances of the R-R and Swrebec functions more
carefully, the RMSE and Adj. R 2 values were plotted against the average product
size (d 50 ) of the ground products under wet and grinding. In general, the trend • Both the R-R and Swrebec distribution functions demonstrate su-
lines displayed in Fig. 7 imply that, a marginally better fitting performance for perior fitting performance for the cumulative particle size curves
coarser sizes is obtained using the Swrebec function, and the R-R function fits the over the other studied functions. The R-R function provides mar-
data marginally better at finer sizes. Overall, it was confirmed that both the R-R
and Swrebec demonstrated greater accuracy in describing the cumulative particle
ginally better fitting performance for coarser sizes, whereas the
size curves obtained from wet and dry grinding of coal in a laboratory-scale ball Swrebec function fits the data somewhat better at finer sizes.
mill. The R-R function, therefore, was used to characterize the size distribution of • For a short grinding time, a superior fitting performance to the
the ground products owing to the greater mathematical complexity of the Swrebec experimental particle size probability mass distributions of a
function.
laboratory-scale ball mill is obtained when using the R-R func-
For the R-R function, the parameter m is a measure of the spread of particle
size (size distribution modulus). A smaller m value indicates a wide spread in tion. The particles are nipped against and between the balls in
the particle size [52]. The parameter l represents the mean particle size (μm). m the bed on the bottom of the mill during this period.
and l are adjustable parameter characteristics of the size distribution, which can • When grinding time increases to approximately 3 min, the good-
be evaluated via nonlinear least square methods provided by the MATLAB ness of fit for all distribution functions studied performs margin-
software.
As presented in Fig. 8, the values of m and l are significantly decreased with in-
ally worse. This phenomena can be associated with a change in
creased grinding time. It can also be observed that dry grinding demonstrates supe- the dominant breakage mechanisms from impact to abrasion-
rior advantage in size reduction compared to that of wet grinding, which is in chipping.
acceptable agreement with the findings from Fig. 2. Moreover, the two m-curves • With increasing time, it can be observed that the G-G-S and G-M
for wet and dry grinding coincide with each other at short grinding time (b10 min).
functions are the optimal functions to describe the particle size
The decreasing R-R distribution modulus m indicates a wider distribution of particle
size when the grinding time increases. For a short grinding time, the dry grinding can probability mass distributions for wet and dry grinding, respec-
produce a slightly wider range of particle size distribution compared to that of wet tively. Further, morphology characterization of the ground parti-
grinding. cles confirms the occurrence of abrasion or attrition with a longer
grinding time.
4.2. Relative mass probability density distributions

Fig. 9 illustrates the relative mass probability densities of the daughter fragments
produced by primary breakage under wet and dry conditions. The calculation method Author contributions
of the relative mass probability density was described in Ref. [51]. The relative mass
distributions of the ground products for wet and dry grinding were used to verify the
goodness of fit of these probability density functions (see Table 1). Comparisons of Xianging Bu and Chao Ni: conceptualization, funding acquisition,
the fitting performances among the different probability distribution functions roles/writing - original draft, and writing - review & editing; Guangxi
using RMSE and Adj. R2 for wet and dry grinding are displayed in Figs. 10 and 11, Ma: investigation and methodology; Yuran Chen: data curation and
respectively. formal analysis; Yujin Sun: validation; Guangyuan Xie: resources and
For wet and dry grinding, the R-R probability distribution function produces a
significantly better fit to the experimental particle size probability mass distribu-
visualization.
tions than others when the grinding is limited to a relatively short time (b2 min).
The source of the daughter fragments is mainly from the primary breakage of the Acknowledgements
feed material in the topmost size interval in this period. In a ball mill, the break-
age of particles depends on the impact and cascading motions of the balls. The
The authors gratefully acknowledge financial support from the
particles are nipped against and between balls in the bed on the bottom of the
mill [56]. Project funded by China Postdoctoral Science Foundation (No.
When grinding time increases to approximately 3 min, the goodness of fit for 2019M652024).
all distribution functions studied performs marginally worse. In particular, the
Adj. R 2 values given by these functions range from −0.5167 to 0.1351, which Appendix A. Supplementary data
could be related to a change in the dominant breakage mechanism. It can be ob-
served from Fig. 3 that the amount of fines smaller than 74 μm increases to ap-
proximately 8%, which is conducive to the generation of abrasion or attrition. As Supplementary data to this article can be found online at https://doi.
the grinding proceeded (≥3 min), the relative mass distributions of the ground org/10.1016/j.powtec.2019.09.062.
312 X. Bu et al. / Powder Technology 359 (2020) 305–313

References [32] D. Feng, C. Aldrich, A comparison of the flotation of ore from the Merensky Reef after
wet and dry grinding, Int. J. Miner. Process. 60 (2000) 115–129.
[1] World Coal Association, Coal & electricity, https://www.worldcoal.org/coal/uses- [33] N. Kotake, et al., Influence of dry and wet grinding conditions on fineness and shape
coal 2018, Accessed date: 13 March 2018. of particle size distribution of product in a ball mill, Adv. Powder Technol. 22 (2011)
[2] China Electricity Council, Analysis and forecast report of national power supply 86–92.
and demand situation in 2017-2018, http://www.cec.org.cn/guihuayutongji/ [34] Y.E. Pivinskii, et al., A study and comparison of the properties of bauxite
gongzuodongtai/2018-02-01/177584.html 2018, Accessed date: 1 February treated by dry and wet grinding techniques, Refract. Ind. Ceram. 44 (2003)
2018. 399–404.
[3] W. Xia, et al., Effect of particle shape on bubble-particle attachment [35] R. Zisselmar, H. Kellerwessel, Approximate mathematical description of
angle and flotation behavior of glass beads and fragments, 338 (2018) particle-size distributions – possibilities and limitations as to the assessment
168–172. of comminution and classification processes, Part. Part. Syst. Charact. 2
[4] National Bureau of Statistics of China, China statistical yearbook-2017, http:// (1985) 49–55.
www.stats.gov.cn/tjsj/ndsj/2017/indexeh.htm 2018, Accessed date: 13 October [36] A. Taşdemir, T. Taşdemir, A comparative study on PSD models for chromite
2017. ores comminuted by different devices, Part. Part. Syst. Charact. 26 (2009)
[5] U. Ulusoy, C. Igathinathane, Particle size distribution modeling of milled coals by dy- 69–79.
namic image analysis and mechanical sieving, Fuel Process. Technol. 143 (2016) [37] D.E. Grady, M.E. Kipp, Fracture mechanics of rock, in: B.K. Atkinson (Ed.), Dy-
100–109. namic Rock Fragmentation, London Academic Press, London, UK 1987,
[6] X. Bu, et al., The order of kinetic models in coal fines flotation, Int. J. Coal Prep. Util. pp. 429–475.
37 (2017) 113–123. [38] Y. Wang, E. Forssberg, Product size distribution in stirred media mills, Miner. Eng. 13
[7] Y. Umucu, et al., A new model for comminution behavior of different coals in an im- (2000) 459–465.
pact crusher, Energy Sources Part A 36 (2014) 1406–1413. [39] J.A. Herbst, et al., Size reduction and liberation, in: M.C. Fuerstenau, K.N. Han (Eds.),
[8] V. Deniz, Comparisons of dry grinding kinetics of lignite, bituminous coal, and petro- Principles of Mineral Processing, Society for Mining, Metallurgy, and Exploration,
leum coke, Energy Sources Part A 35 (2013) 913–920. Inc., Littleton, Colorado, USA 2003, pp. 61–118.
[9] V. Deniz, The effects of moisture content and coal mixtures on the grinding behavior [40] B.B. Mandelbrot, The Fractal Geometry of Nature, W.H. Freeman, New York, 1983.
of two different coals, Energy Sources Part A 36 (2013) 292–300. [41] D.L. Turcotte, Fractals and fragmentation, J. Geophys. Res.-Solid Earth 91 (1986)
[10] X. Bu, et al., Grinding kinetics of coal in wet ball-milling using the Taguchi 1921–1926.
method, Int. J. Coal Prep. Util. (2019) https://doi.org/10.1080/19392699.2019. [42] F.C. Bond, Crushing and grinding calculations, Part II, Br. Chem. Eng. 6 (1961)
1603147. 543–548.
[11] M.S. Celik, A comparison of dry and wet fine grinding of coals in a ball mill, Powder [43] F.C. Bond, Crushing and grinding calculations, Br. Chem. Eng. 6 (1961)
Technol. 55 (1988) 1–9. 378–385.
[12] V. Deniz, The effects of ball filling and ball diameter on kinetic breakage parameters [44] A.W. Momber, The fragmentation of standard concrete cylinders under com-
of barite powder, Adv. Powder Technol. 23 (2012) 640–646. pression: the role of secondary fracture debris, Eng. Fract. Mech. 67 (2000)
[13] V. Deniz, The Effect of Mill Speed on Kinetic Breakage Parameters of Clinker and 445–459.
Limestone, vol. 34, 2004 1365–1371. [45] G. Lowrison, Crushing and Grinding: the Size Reduction of Solid Material, CRC Press,
[14] V. Deniz, Relationships between Bond's grindability (G bg) and breakage pa- London, 1974.
rameters of grinding kinetic on limestone, Powder Technol. 139 (2004) [46] P.L. Prasher, Crushing and Grinding Process Handbook, Wiley, New York,
208–213. 1987.
[15] C. Tangsathitkulchai, L.G. Austin, The effect of slurry density on breakage parameters [47] E. Petrakis, et al., Evaluation of the relationship between energy input and particle
of quartz, coal and copper ore in a laboratory ball mill, Powder Technol. 42 (1985) size distribution in comminution with the use of piecewise regression analysis,
287–296. 35 (2017) 479–489.
[16] N. Kotake, et al., A fundamental study of dry and wet grinding in bending tests on [48] T. Allen, Powder Sampling and Particle Size Determination, Elsevier, Amsterdam,
glass—effect of repeated impact on fracture probability, Miner. Eng. 17 (2004) 2003.
1281–1285. [49] F. Ouchterlony, The Swebrec© function: linking fragmentation by blasting and
[17] H. El-Shall, P. Somasundaran, Physico-chemical aspects of grinding: a review of use crushing, Trans. Inst. Min. Metall. A 114 (2005) 29–44.
of additives, Powder Technol. 38 (1984) 275–293. [50] P. González-Tello, et al., A modified Nukiyama–Tanasawa distribution function and a
[18] R. Klimpel, Laboratory studies of the grinding and rheology of coal—water slurries ☆, Rosin–Rammler model for the particle-size-distribution analysis, Powder Technol.
Powder Technol. 32 (1982) 267–277. 186 (2008) 278–281.
[19] L.G. Austin, P. Bagga, An analysis of fine dry grinding in ball mills, Powder Technol. [51] C.A. Dunbar, A.J. Hickey, Evaluation of probability density functions to approximate
28 (1981) 83–90. particle size distributions of representative pharmaceutical aerosols, J. Aerosol Sci.
[20] C. Tangsathitkulchai, Acceleration of particle breakage rates in wet batch ball mill- 31 (2000) 813–831.
ing, Powder Technol. 124 (2002) 67–75. [52] K.M. Djamarani, I.M. Clark, Characterization of particle size based on fine and coarse
[21] L.G. Austin, et al., An analysis of ball-and-race milling: Part I. The Hardgrove mill, fractions, Powder Technol. 93 (1997) 101–108.
Powder Technol. 29 (1981) 263–275. [53] P. González-Tello, et al., Analysis of the mean diameters and particle-size distribu-
[22] L. Tong, et al., Stirred milling kinetics of siliceous goethitic nickel laterite for selective tion in powders, Part. Part. Syst. Charact. 27 (2010) 158–164.
comminution, Miner. Eng. 49 (2013) 109–115. [54] X. Bu, et al., Removal of fine quartz from coal-series kaolin by flotation, Appl. Clay
[23] E. Bilgili, B. Scarlett, Population balance modeling of non-linear effects in milling Sci. 143 (2017) 437–444.
processes, Powder Technol. 153 (2005) 59–71. [55] X. Bu, et al., Multi-stage flotation for the removal of ash from fine graphite using me-
[24] E. Bilgili, et al., Formulation of a non-linear framework for population balance chanical and centrifugal forces, Minerals 8 (2018) 15.
modeling of batch grinding: beyond first-order kinetics, Chem. Eng. Sci. 61 (2006) [56] L.G. Austin, et al., Process Engineering of Size Reduction: Ball Milling, SME/AIME,
33–44. New York, 1984.
[25] S. Samanli, et al., Investigation of breakage behavior of coal in a laboratory-scale [57] A. Ozkan, M. Yekeler, Fine dry grinding of zeolite in a laboratory ball mill, Part. Part.
stirred media mill, Int. J. Coal Prep. Util. 30 (2010) 20–31. Syst. Charact. 20 (2003) 276–282.
[26] G. Danha, et al., Application of basic process modeling in investigating the breakage [58] A. Ozkan, et al., Kinetics of fine wet grinding of zeolite in a steel ball mill in compar-
behavior of UG2 ore in wet milling, Powder Technol. 279 (2015) 42–48. ison to dry grinding, Int. J. Miner. Process. 90 (2009) 67–73.
[27] Z. Ma, et al., Breakage behavior of quartz in a laboratory stirred ball mill, Powder [59] M. Yekeler, et al., Kinetics of fine wet grinding in a laboratory ball mill, Powder
Technol. 100 (1998) 69–73. Technol. 114 (2001) 224–228.
[28] D.W. Fuerstenau, et al., Linear and nonlinear particle breakage processes in commi- [60] J.M. Menéndez-Aguado, et al., Particle size distribution fitting of surface detri-
nution systems, Int. J. Miner. Process. 74 (2004) S317–S327. tal sediment using the Swrebec function, J. Soils Sediments 15 (2015)
[29] N. Yamamoto, et al., Comparison of dry- and wet-based fine bead homogeni- 2004–2011.
zations to extract DNA from fungal spores, J. Biosci. Bioeng. 107 (2009) [61] A.M. Osorio, et al., Fine grinding size distribution analysis using the Swrebec func-
464–470. tion, Powder Technol. 258 (2014) 206–208.
[30] T. Zhang, et al., Characteristics of wet and dry crushing methods in the recycling pro- [62] S. Liu, et al., Effect of grinding time on the particle characteristics of glass powder,
cess of spent lithium-ion batteries, 240 (2013) 766–771. Powder Technol. 295 (2016) 133–141.
[31] J. Leewatchararongjaroen, J. Anuntagool, Effects of dry-milling and wet-milling on [63] S. Liu, et al., Study on the grinding kinetics of copper tailing powder, Powder
chemical, physical and gelatinization properties of rice flour, Rice Sci. 23 (2016) Technol. 330 (2018) 105–113.
274–281.
X. Bu et al. / Powder Technology 359 (2020) 305–313 313

[64] C. Hicyilmaz, et al., Flotation responses to the morphological properties of particles [66] X. Bu, et al., Differences in dry and wet grinding with a high solid concentration of
measured with three-dimensional approach, Int. J. Miner. Process. 75 (2005) coking coal using a laboratory conical ball mill: breakage rate, morphological char-
229–236. acterization, and induction time, Adv. Powder Technol. (2019) https://doi.org/10.
[65] C. Hiçyilmaz, et al., Effects of the shape properties of talc and quartz particles on the 1016/j.apt.2019.08.016.
wettability based separation processes, Appl. Surf. Sci. 233 (2004) 204–212.

You might also like