BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science

9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

Araştırma Makalesi - Research Article

A New Chemical Method to Predict Grindability Index

(HGI) for Limestones
Kireçtaşları İçin Yeni Bir Kimyasal Öğütme İndeksi
Candan Bilen1*
Geliş / Received: 18/09/2021 Revize / Revised: 20/02/2022 Kabul / Accepted: 08/03/2022

ABSTRACT

In this study, limestone samples (a total of 58 sample) were investigated in terms of their grindability and chemical
composition. Grindability tests were carried out on standard HGI mill. Limestone samples were collected from
two different limestone quarry and they were characterized in terms of their chemical composition. In the order of
technological utilization, grindability nature of limestones is as significant as their chemical composition.
Chemical composition of the limestone samples from the quarries under investigation differs and so does the
grinding index, i.e. HGI (Hardgrove Grinding Index). In the context of this study, chemical composition data of
limestone samples were correlated with the results of the grinding tests (HGI values). In addition, abovementioned
correlations were provided as graphical demonstrations in this context. After these abovementioned graphical
demonstration of the relationships between HGI values and chemical composition data, the role of the each
chemical composition item in terms of grindability was understood. Based on this understanding, an empirical
formula employing the chemical composition data was proposed to predict HGI.

Keywords- Limestone Quarry, Chemical Composition, HGI, Modeling, Limestone Grinding

ÖZ

Bu çalışmada, kireçtaşı numuneleri (toplam 58 numune) öğütülebilirlik ve kimyasal bileşim açısından

incelenmiştir. Öğütülebilirlik testleri standart HGI değirmeninde gerçekleştirilmiştir. Kireçtaşı örnekleri iki farklı
kireçtaşı ocağından alınmış ve kimyasal bileşimleri açısından karakterize edilmiştir. Teknolojik kullanım sırasına
göre, kireçtaşının kimyasal bileşimi kadar öğütülebilirliği de önemlidir. İncelenen ocaklardan alınan kireçtaşı
numunelerinin kimyasal bileşimleri ve öğütme indeks (Hardgro ve Öğütülebilirlik Indeksi) değerleri farklıdır. Bu
çalışma kapsamında, kireçtaşı numunelerinin kimyasal bileşim verileri ile öğütülebilirlik test sonuçları (HGI
değerleri) ile ilişkilendirilmiştir. Bunun yanında, belirtilen bu ilişkilendirmeler grafiksel gösterim olarak bu
kapsamda verilmiştir. HGI değerleri ile kimyasal bileşim verileri arasındaki ilişkilerin grafiksel gösteriminden
sonra, her bir kimyasal bileşim öğesinin öğütülebilirlik açısından rolü anlaşılmıştır. Bu çalışma kapsamında HGI'yi
tahmin etmek için kimyasal bileşim verilerini kullanan ampirik bir formül önerilmiştir.

Anahtar Kelimeler- Kireçtaşı Ocağı, Kimyasal Bileşim, HGI, Modelleme, Kireçtaşı Öğütme

1*Corresponding author contact: [email protected] (https://orcid.org/0000-0001-9209-0872)

Department of Geological Engineering, Engineering Faculty, Zonguldak Bülent Ecevit University, Zonguldak, Turkey
124
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

I. INTRODUCTION
Limestone which has calcite as dominant mineral is being utilized as a raw material by human beings for
a long time. Limestone has wide field of utilization and it is mainly demanded by the construction industry. In
addition to utilization in terms of construction purposes, it has also potential usage in metallurgy, agriculture. Flue
gas desulfurization is also another field of industry which limestones are widely utilized as raw materials. In many
industries abovementioned however, in addition to its specific chemical composition limestone should also be
ground and have the form of powder.
Limestones are formed mainly as sedimentary rocks and the main formation conditions include climate
and absence of clay or sandy material [1]. Limestones have mostly calcium carbonate or magnesium carbonate
compounds and some impurities like iron, aluminum, silica, and sulfur presents at different amounts (depending
on the type and formation) in their structures. Limestones are mostly classified in terms of their origin, geological
formation, mineralogical structure, crystal form, chemical composition, color and hardness. Limestones are also
classified depending on their carbonate amounts. In terms of carbonate classification, Folk [2,3] and Dunham [4]
classification systems are the most common ones.
Grindability of a material is characterized by the amount of the work/energy required for a unit
weight/volume of material to be ground to a specific size distribution [5]. Ore grindability is represented by the
Bond work index value [6] and Mucsi et al. [5] have summarized the most widely known and utilized grindability
tests as Bond, Hardgrove and Zeisel methods [7-9]. Many researchers [10-14] have employed BWI to address the
limestone grindability. Musci et al. [5] have investigated the grindabilities of andesite, basalt, clinker, limestone
and quartz. Referring back and forth to the study of Mucsi et al. [5], although it has not been addressed and
emphasized, chemical composition along with the HGI values of abovementioned materials (andesite, basalt,
clinker, limestone and quartz) were provided. As it is obvious and expected, change in chemical composition
results in a change in grindability values, i.e. HGI in this case. According to the Seo et al. [15], difference in
grinding efficiency depends on the chemical composition and crystal structure. Same authors [15] have employed
HGI to understand the grindabilities of limestones and they have claimed the fact that HGI measurement is rather
easier and not as much time consuming as compared to BWI (Bond Work Index) measurement. HGI method,
although some disadvantages [16-18] and repeatability issues [19, 20] are associated with, is widely known and
mostly addressed in terms of grindability specifications (especially for coal in terms of international trade) and
mostly considered for design and optimization of grinding circuits [16].
In the study of Mendis et al. [21], chemical composition, moisture content and particle size are considered
as the effective factors of raw material in terms of grindability and authors have claimed the fact that grindability
is mainly affected by the chemical composition. According to Kural and Ozsoy [22], chemical and physical factors
of raw material are the main reasons in terms of low grindability. Mendis et al. [21] have addressed the percentages
of SiO2, Al2O3, Fe2O3, CaO, MgO, Cl, SO3, Na2O, and K2O in limestones in order to assess relationships between
grindability and chemical composition. In the study of same authors [21], it was shown that increase in CaO (%)
resulted in an increase in grindability. Although some chemical composition data is interrelated to grindability,
authors [21] have employed no.212 sieve residue rather than any grindability index values. Although employment
of the method abovementioned to address the relationship between grindability and chemical composition is
acceptable to some extent, assessment in terms of grindability index (HGI or BWI) could have been more
reasonable.
Researchers [23] have investigated the effect of chemical composition on Portland cement clinker
grindability. In addition, Ürünveren et al. [24] have tried to predict HGI values of Afşin-Elbistan (Turkey) Low-
grade coals based on proximate analysis and ash chemical composition. Although prediction of HGI (for coal
mostly) is more of concern by many researchers [25-27], some other researchers [28] have investigated the
grindability behavior of clinker and colemanite. Although no recent research have investigated the relationship
between chemical composition and grindability index values (HGI) as in the same context of this study, still some
studies [23,29] have been conducted in terms of clinker composition and grindability.
Grindability is most of the time significant issue in terms of energy requirements of milling, and
pulverized limestones are generally desired for the further utilization of limestones. In this regard, not only
chemical composition data of limestones are mostly referred but also the grindability is. While having chemical
composition data on hand, if further utilization needs milling, field engineers should be aware of the grindability
values of the corresponding limestones. So, in this context, chemical composition data can be interrelated with
grindability index values, i.e. HGI, which is not only time consuming experimental procedure but also needs
125
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

expertise and significant amount of laboring. In this study, two different limestone quarries (Quarry-1 and Quarry-
2) were taken into consideration and a total of 58 limestone samples, i.e. 22 samples from Quarry-1 and 36 from
Quarry-2, were collected. Collected samples were analyzed in terms of their chemical composition and grindability
tests (HGI) were performed. These analyzes abovementioned resulted in an understanding of the relationship
between grindability values and chemical composition data for limestones. Having evaluated the relationships
between each chemical component and HGI values, a new chemical grinding index for limestones was proposed.
II. MATERIAL AND METHOD
A. Samples and Sample Preparation
Limestone quarries under investigation are located in Gebze/Kocaeli, Turkey. Number of limestone
quarries are being operated in the region where this study conducted. Location map of the quarries where samples
are collected from is provided in Figure 1. Geological map of the study area (adapted from the study of Gedik et
al. [30]) was provided in Figure 2. General view of limestone quarries studied is provided in Figure 3. Laboratory
work along with sampling in the quarry is schematized in Figure 4.

Figure 1. The location map of the study area

126
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

Figure 2. Geological map of the study area (adapted from the study of Gedik et al. [30])

(a-1) (a-2)

(b-1) (b-2)

Figure 3. General view of the limestone quarries under investigation, refer to (a-1) and (a-2) for Quarry-1 & (b-1) and (b-2) for Quarry-2.

Figure 4. Schematical representation of the laboratory work and sampling in the quarry.

B. Grinding Tests
Grinding tests were carried out with the standard HGI mill employment. All of the 58 samples were
characterized in terms of their HGI values. HGI tests was carried out as in the standard described in ASTMD409-
71[8]. In order perform HGI tests, samples were crushed step by step to the size fraction of -1.18+0.6 mm, as
standard implies for the feed size of HGI mill. After the grinding with HGI mill, final amount(s) ground under
75 µm were noted. This final amount (ground under 75 µm with HGI mill) was placed in the equation (1) and HGI
value of that specific sample is determined accordingly.

127
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

HGI=13.6+6.93.w (1)

where w is the weight of the test sample passing through 75 µm sieve.

HGI tests were realized and corresponding HGI values for each limestone samples (a total of 58) were
recorded. Representation of HGI testing environment is provided in Figure 5.

(a) (b)
Figure 5. Representation of HGI testing environment, (a) Grinding elements of Hardgrove Machine (adapted from the study of Tichanek [16]),
(b) HGI mill employed in this study.

C. Chemical Analyzes
Representative samples of limestone (a total of 58) were collected from the quarries (Quarry-1 and
Quarry-2). Collected samples were initially prepared for chemical analysis. This preparation includes size
reduction and grinding. In terms of chemical analysis, standard method (ASTM C1271-99 [31]) was taken into
consideration and the analysis was carried out with X-ray Fluorescence (XRF). A laboratory view of the XRF
equipment (Philips PW-2404) employed for the chemical analyses is provided in Figure 6.

Figure 6. Laboratory view of the XRF equipment.

III. RESULTS AND DISCUSSION

Limestone samples collected from the quarries were investigated in terms of their chemical composition
and their grindability behavior. In order to understand the grindability behavior, HGI tests were performed. The
chemical composition data and the corresponding HGI value of each sample were tabulated in Table 1 and Table
2, respectively for Quarry-1 and Quarry-2. As it is previously explained, a total of 58 limestone samples (22 of
which is collected from Quarry-1 and 36 of which is collected from Quarry-2) were analyzed and corresponding
data was tabulated.

128
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

Table 1. Chemical composition data and HGI values of the limestone samples from Quarry-1.

SAMPLE HGI MgO (%) SiO2 (%) CaO (%) Fe2O3 (%) Al2O3 (%) LOI (%) TOTAL (%)

Q1-1 75.82 2.83 8.11 43.68 1.96 3.64 38.43 98.65

Q1-2 74.68 1.62 1.49 52.83 0.83 0.00 42.93 99.70

Q1-3 63.64 4.46 6.86 40.91 2.82 3.52 39.81 98.39

Q1-4 78.34 1.76 1.29 51.68 1.25 0.51 43.21 99.69

Q1-5 65.18 5.80 8.47 40.00 3.59 1.98 39.72 99.55

Q1-6 72.66 2.81 2.71 49.13 1.80 0.00 43.03 99.49

Q1-7 75.82 3.10 3.52 47.80 1.68 1.44 41.80 99.33

Q1-8 79.15 2.12 1.24 51.52 0.57 0.47 43.78 99.69

Q1-9 84.36 0.76 0.97 53.99 0.24 0.32 43.30 99.58

Q1-10 88.16 0.63 1.24 54.42 0.24 0.00 43.14 99.68

Q1-11 79.79 0.87 1.05 53.55 0.43 0.45 42.99 99.34

Q1-12 77.79 12.22 6.74 34.80 1.33 0.00 44.23 99.32

Q1-13 78.48 12.52 2.51 37.09 1.39 0.00 46.06 99.57

Q1-14 78.18 12.11 6.33 35.56 1.39 0.00 44.03 99.41

Q1-15 75.29 11.48 9.39 33.92 1.54 0.00 42.62 98.96

Q1-16 47.71 3.87 5.38 40.48 7.21 2.21 38.83 97.99

Q1-17 69.42 7.99 3.35 41.73 2.08 0.00 44.31 99.46

Q1-18 73.15 12.96 2.34 37.03 1.44 0.00 45.83 99.60

Q1-19 78.89 13.11 1.24 38.08 1.07 0.00 46.12 99.62

Q1-20 66.78 12.78 2.67 37.37 1.03 0.00 45.71 99.56

Q1-21 74.40 12.52 2.32 37.63 1.15 0.00 45.88 99.49

Q1-22 62.20 12.82 2.56 37.26 1.35 0.00 45.52 99.51

Table 2. Chemical composition data and HGI values of the limestone samples from Quarry-2.

SAMPLE HGI MgO (%) SiO2 (%) CaO (%) Fe2O3 (%) Al2O3 (%) LOI (%) TOTAL (%)

Q2-1 96.69 0.83 4.71 54.24 0.73 1.97 36.71 99.20

Q2-2 95.94 0.46 4.23 54.47 0.68 1.67 37.93 99.44

Q2-3 78.18 0.46 5.23 53.75 1.23 1.86 36.85 99.38

Q2-4 86.75 0.80 11.67 46.59 1.11 2.73 36.19 99.09

Q2-5 66.30 1.76 30.42 27.95 3.39 12.86 19.30 95.68

Q2-6 83.07 0.54 3.81 54.36 0.61 1.33 38.90 99.55

Q2-7 72.14 1.03 14.30 46.58 2.20 6.65 27.18 97.95

Q2-8 86.30 0.53 2.83 54.60 0.85 1.26 39.50 99.57

Q2-9 79.49 0.40 4.70 53.95 0.97 1.45 38.04 99.50

Q2-10 76.91 1.79 6.57 49.87 1.42 2.51 36.56 98.72

129
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

Table 2. Continues

Q2-11 82.78 0.94 4.35 53.79 0.78 1.48 38.05 99.38

Q2-12 72.78 10.98 12.64 35.21 1.98 5.43 32.84 99.07

Q2-13 70.01 4.29 47.33 9.94 2.02 22.28 10.06 95.93

Q2-14 69.55 9.51 2.07 42.57 1.98 1.22 42.40 99.75

Q2-15 80.34 1.41 3.65 53.48 0.65 1.26 39.22 99.67

Q2-16 92.04 0.59 3.28 54.43 0.34 0.59 40.54 99.77

Q2-17 77.91 0.81 7.10 52.63 1.30 3.54 33.46 98.84

Q2-18 93.84 0.45 2.85 54.78 0.28 0.56 40.83 99.75

Q2-19 78.46 0.90 11.09 47.69 1.88 3.71 33.52 98.79

Q2-20 64.84 4.03 24.81 30.77 3.46 8.04 25.95 97.06

Q2-21 78.97 0.90 4.87 53.12 0.83 1.73 37.86 99.31

Q2-22 78.46 1.67 9.55 50.04 1.56 4.03 31.66 98.50

Q2-23 92.33 0.60 2.55 55.37 0.62 0.98 39.56 99.68

Q2-24 96.65 0.60 3.84 54.59 0.72 1.46 38.27 99.47

Q2-25 88.39 0.84 3.81 54.14 0.70 1.64 38.26 99.40

Q2-26 82.59 1.25 4.13 53.01 0.82 1.73 38.21 99.15

Q2-27 88.13 1.07 3.31 54.95 0.61 1.44 38.07 99.46

Q2-28 79.87 1.80 6.59 50.02 1.05 2.41 37.08 98.95

Q2-29 99.23 1.04 2.46 55.25 0.56 1.01 39.26 99.58

Q2-30 75.19 1.16 5.43 52.66 1.51 2.22 36.18 99.15

Q2-31 86.97 1.02 1.78 55.94 0.55 0.64 39.51 99.44

Q2-32 85.88 1.28 4.73 53.34 0.71 1.87 37.39 99.30

Q2-33 78.85 1.66 10.95 47.21 1.65 3.86 33.23 98.55

Q2-34 80.23 0.99 2.27 54.04 0.64 0.77 40.91 99.63

Q2-35 71.17 2.62 9.81 45.89 1.50 2.48 36.64 98.94

Q2-36 80.81 2.30 11.81 45.30 1.59 4.28 33.01 98.30

As it is presented in Table 1 and Table 2, chemical composition data for both quarries includes major
oxide percentages (MgO, SiO2, CaO, Fe2O3, Al2O3) and LOI (%). Although chemical composition data for both
quarry is available for other element oxides like K2O, TiO2, P2O5, Na2O, BaO, MnO and etc, tabulated data only
includes the abovementioned major oxides, i.e. MgO, SiO2, CaO, Fe2O3, Al2O3. This is because of the fact that
total percentages of these major oxides and LOI (%) is approximately 100 (See Table 1 and Table 2) for all samples
and so the rest of the element oxides can be neglected to count in this case. In order to evaluate the change in
grindability (HGI) with respect to the chemical composition data, graphical demonstrations of the relationships
were provided between Figure 7 and Figure 12 [Figure 7(a)-12(a) for Quarry-1 and Figure 7(b)-12(b) for Quarry
-2].

130
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

100,00
120,00
80,00 100,00
60,00 80,00
HGI

HGI
40,00 60,00
y = -0,2758x + 75,526 40,00
20,00 y = -1,9744x + 85,364
R² = 0,0263
20,00 R² = 0,258
0,00
0,00
0,00 5,00 10,00 15,00
0,00 5,00 10,00 15,00
MgO (%)
MgO (%)

(a) (b)
Figure 7. Relationship between HGI values and MgO (%) content of limestone samples (a) from Quarry-1, (b) from Quarry-2.

100,00 120,00
80,00 100,00
80,00
60,00
HGI

HGI

60,00
40,00
y = -1,1402x + 77,87 40,00 y = -0,5724x + 86,589
20,00 R² = 0,1296 20,00 R² = 0,3454
0,00 0,00
0,00 2,00 4,00 6,00 8,00 10,00 0,00 10,00 20,00 30,00 40,00 50,00
SiO2 (%) SiO2 (%)

(a) (b)
Figure 8. Relationship between HGI values and SiO2 (%) content of limestone samples (a) from Quarry-1, (b) from Quarry-2.

100,00 120,00
80,00 100,00
80,00
60,00
HGI

HGI

60,00
40,00
40,00 y = 0,6179x + 51,572
20,00 y = 0,4947x + 52,259
R² = 0,1698 20,00 R² = 0,4424
0,00 0,00
30,00 40,00 50,00 60,00 0,00 20,00 40,00 60,00
CaO (%) CaO (%)

(a) (b)
Figure 9. Relationship between HGI values and CaO (%) content of limestone samples (a) from Quarry-1, (b) from Quarry-2.

131
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

100,00 120,00
80,00 100,00
80,00
60,00
HGI

HGI
60,00
40,00
y = -4,9744x + 81,86 40,00 y = -9,5969x + 93,469
20,00 R² = 0,7172 20,00 R² = 0,6701
0,00 0,00
0,00 2,00 4,00 6,00 8,00 0,00 1,00 2,00 3,00 4,00
Fe2O3 (%) Fe2O3 (%)

(a) (b)
Figure 10. Relationship between HGI values and Fe2O3 (%) content of limestone samples (a) from Quarry-1, (b) from Quarry-2.

100,00 120,00
80,00 100,00
80,00
60,00
HGI

HGI

60,00
40,00
40,00 y = -1,2027x + 85,729
y = -3,1379x + 75,706
20,00 20,00 R² = 0,307
R² = 0,1749
0,00 0,00
0,00 1,00 2,00 3,00 4,00 0,00 5,00 10,00 15,00 20,00 25,00
Al2O3 (%) Al2O3 (%)

(a) (b)
Figure 11. Relationship between HGI values and Al2O3 (%) content of limestone samples (a) from Quarry-1, (b) from Quarry-2.

100,00 120,00
80,00 100,00
80,00
60,00
HGI

HGI

60,00
40,00 y = 1,353x + 15,126 40,00 y = 0,805x + 53,286
20,00 R² = 0,1334
20,00 R² = 0,3308
0,00 0,00
35,00 40,00 45,00 50,00 0,00 10,00 20,00 30,00 40,00 50,00
LOI (%) LOI (%)

(a) (b)
Figure 12. Relationship between HGI values and LOI (%) content of limestone samples (a) from Quarry-1, (b) from Quarry-2.

Initial evaluations of the relationships (Figure 7 - Figure 12) are as following: i. Fe2O3 content (%) is the
most meaningful chemical composition parameter in terms of HGI for both quarries, ii. relationships obtained for
the samples from Quarry-2 resulted in higher R2 values (coefficient of determination) except Fe2O3(%). In terms
of these initial understandings abovementioned, one can claim the fact that any empirical equation to be proposed
for the prediction of HGI should include Fe2O3 (%) content.
Limestone grindability is a significant issue since limestone as raw material has a wide range of
utilization, mostly requires powder form. In order to improve the efficiency of grinding circuits of limestones, iron
content should be taken into consideration in the beginning. Graphical demonstrations of the relationship of HGI
dependency on Fe2O3 [Refer to Figure 10 (a) and (b)] simply includes the understanding as following: higher the

132
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

Fe2O3 results in lower the HGI value and so harder (more difficult) is the grinding operation and vice versa,
respectively. This actually makes sense due to the fact that easier milling with any pre-treatment (microwave)
would end up less wear of the mill, mill liner, and milling medium, summarized by Kumar et al. [32]. Here, in
order to have higher grinding efficiency for limestone grinding, pre-treatment including magnetic separation could
be employed.
In addition to the abovementioned interrelations, a new empirical equation for the prediction of HGI in
terms of chemical composition data was proposed in the context of this study. In order to have less number of
parameters contributing to HGI and to obtain the highest R2, several attempts were carried out on a statistical
software, i.e. XLSTAT. In this context, “Gamma Regression” in the body of “Log-Linear Regression” of that
abovementioned software was taken into consideration. The model proposed (See (2) and (3)) has only 4
parameters employing and it has the corresponding R2 of 0.74.

𝑃𝑟𝑒𝑑𝑖𝑐𝑡𝑖𝑜𝑛 𝑀𝑒𝑡ℎ𝑜𝑑 𝑜𝑓 𝐻𝐺𝐼 = 𝑒 (4.19+0.004𝑥Si𝑂2+0.0056𝑥𝐶𝑎𝑂−0.084𝑥𝐹𝑒2 𝑂3 −0.0022𝑥𝑟𝑎𝑡𝑖𝑜) (2)

Here, “ratio” is defined as (3):

𝐹𝑒2 𝑂3 + 𝐴𝑙2 𝑂3 + 𝑀𝑔𝑂 (3)

𝑟𝑎𝑡𝑖𝑜 =
Si𝑂2

Note that all chemical composition items, i.e. SiO2, CaO, MgO, Fe2O3, Al2O3 are in percentages. Trial
error processing of linear and nonlinear regressions have only resulted as either too many number of parameters
involving or less R2 obtained at the end. So this equation presented in (2) is regarded as the best equation in terms
of easier evaluation and better prediction. In this context, graphical representation of the comparison between
experimentally obtained HGI values and the predicted (Eqn.2 is employed) HGI values (See Table 3 and See
Figure 13).

133
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

Table 3. Predicted (Eqn.2 is employed) and experimentally obtained HGI values.

Sample Experimentally Predicted HGI Sample Experimentally Predicted HGI

Obtained HGI values Obtained HGI values
values values
Q1-1 75.82 73.71 Q2-1 96.69 85.57
Q1-2 74.68 82.95 Q2-2 95.94 85.94
Q1-3 63.64 67.11 Q2-3 78.18 82.06
Q1-4 78.34 79.34 Q2-4 86.75 81.74
Q1-5 65.18 63.03 Q2-5 66.30 65.50
Q1-6 72.66 75.25 Q2-6 83.07 86.24
Q1-7 75.82 75.69 Q2-7 72.14 75.32
Q1-8 79.15 83.95 Q2-8 86.30 84.27
Q1-9 84.36 87.63 Q2-9 79.49 83.79
Q1-10 88.16 88.03 Q2-10 76.91 79.36
Q1-11 79.79 86.00 Q2-11 82.78 84.93
Q1-12 77.79 73.41 Q2-12 72.78 71.42
Q1-13 78.48 72.15 Q2-13 70.01 71.12
Q1-14 78.18 73.17 Q2-14 69.55 70.59
Q1-15 75.29 72.59 Q2-15 80.34 85.43
Q1-16 47.71 45.92 Q2-16 92.04 88.10
Q1-17 69.42 70.51 Q2-17 77.91 81.63
Q1-18 73.15 71.67 Q2-18 93.84 88.57
Q1-19 78.89 73.19 Q2-19 78.46 76.89
Q1-20 66.78 74.62 Q2-20 64.84 64.71
Q1-21 74.40 73.76 Q2-21 78.97 84.41
Q1-22 62.20 72.48 Q2-22 78.46 79.52
Q2-23 92.33 86.18
Q2-24 96.65 85.55
Q2-25 88.39 85.41
Q2-26 82.59 84.15
Q2-27 88.13 86.27
Q2-28 79.87 82.02
Q2-29 99.23 86.46
Q2-30 75.19 79.69
Q2-31 86.97 86.59
Q2-32 85.88 85.33
Q2-33 78.85 78.13
Q2-34 80.23 85.23
Q2-35 71.17 78.14
Q2-36 80.81 77.96

Based on the data provided in Table 3, it can be easily noticed that experimentally obtained or predicted
HGI values are higher for the samples collected from Quarry 2. This abovementioned difference between each
quarry can be associated with the quarry location, quarry altitude difference, heterogeneous structure of the
samples, alteration differences on each quarry, proximity to underground water supply & fault zone,
meteorological differences, mineralogical and crystallographic differences and effects of freezing & thawing and
etc.

134
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

100
HGI values (pred.)
80

60

40
y = 0,7243x + 21,583
20 R² = 0,7417

0
40,00 50,00 60,00 70,00 80,00 90,00 100,00 110,00
HGI values (exp.)

Figure 13. Relationship between HGI values (exp.) and HGI values (pred.), “exp.” stands for experimentally obtained and “pred.” stands for
predicted.

Referring to Figure 13, the correlation between experimentally obtained and predicted HGI values is
significant, which can be interpreted as the achievement of the model proposed. In terms of linear regression
models, proposed empirical equation (2) and (3) is the most significant, which is obtained with number of
iterations. Prediction of HGI based on the chemical composition data has not been considered for limestones,
which means it can be improved. Still, the proposed model represents a quite significant understanding of the
parameters contributions and it can be employed as a guide for the initial estimates of HGI for limestones.
Accordingly, one can refer to the coefficients of the each parameter of the proposed model (Eqn.2). In this context,
it can be claimed that, increase in percentage of CaO results in an increase in HGI, which means easier grindability
and increase in “ratio” results in a decrease in HGI, i.e. more difficult is the grinding. The contribution by SiO2
and Fe2O3 is kind of complicated in this case, since “ratio” does also depend on these parameters. Increase in Fe2O3
content results in a increase in “ratio” and it therefore results in a duplicated decrease in HGI, which means harder
grinding operation likely to be occurring. Increase in SiO2 content results in an increase in HGI in some limits
only. In addition to this abovementioned parametric evaluation, one can easily have initial estimates of HGI based
on chemical composition data of limestones by avoiding the time consuming and laboring intense downside of the
HGI method itself.
IV. CONCLUSION
In this study, a total of 58 limestone samples were studied in terms of their chemical composition and
grindability value (HGI). Limestone samples were collected from two different quarry, i.e. Quarry-1 and Quarry-
2. Analysis results of chemical composition were interrelated to the results of HGI tests for the corresponding
samples. Further evaluations were made in this context with the referrals back and forth to the graphical
representations of the relations (relationships between chemical composition data and HGI values). Major oxides
(%) and LOI percentages for all samples from each quarry were tabulated. Tabulated data of chemical composition
of the limestone samples was restricted to major oxides (SiO2, CaO, MgO, Fe2O3, Al2O3) since XRF analysis
resulted as negligible for some oxides like Na2O, K2O and etc. In this context, graphical representations of the
each chemical composition data (included) and the HGI values resulted with the understanding of the Fe2O3
percentage significance in terms of HGI. Initial evaluations are summarized as following: i. Fe2O3 content (%) is
the most meaningful chemical composition parameter in terms of HGI for both quarries, ii. relationships obtained
for the samples from Quarry-2 resulted in higher R2 values (coefficient of determination) except Fe2O3(%). Latter
in the context of this study, a new chemical grinding index (an empirical equation) was proposed to predict HGI.
This model proposed includes 4 parameters (5 major oxides) and it has R 2 of 0.74 between HGI exp. and pred.
values. The equation proposed includes a parameter stated as “ratio” which takes SiO2, MgO, Al2O3 and Fe2O3
into account. A correct evaluation in terms of the HGI values dependency on chemical composition data is as
following: “i. increase in percentage of CaO results in an increase in HGI, which means easier grindability, ii.
increase in Fe2O3 content results in a increase in “ratio” and it therefore results in a duplicated decrease in HGI,
which means harder grinding operation likely to be occurring, iii. increase in SiO2 content results in an increase in
HGI in some limits only”. By this method presented in the scope of this study, grindability index values for

135
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

limestones can be easily predicted for initial estimates without employing the time consuming and laboring intense
HGI testing procedure.
ACKNOWLEDGEMENTS
Authors of this study would like to thank to Limestone Quarry Companies’ owners and the operators for
providing the samples.
REFERENCES
[1] Šiler, P., Kolarova, I., Bednarek, J., Jana, M., Musil, P., & Opravil, T. (2018). The possibilities of analysis
of limestone chemical composition. IOP Conf. Series: Materials Science and Engineering.
[2] Folk, R. L. (1959). Practical petrographical classification of limestones. Amer. Ass. Petrol. Geol. Bull., 43(1),
1–38.
[3] Folk, R. L. (1962). Spectral subdivision of limestone types. Amer. Ass. Petrol., Geol. Mem., 1, 62–84.
[4] Dunham, R. J. (1962). Classification of carbonate rocks according to depositional texture. Mem. Amer. Ass.
Petrol. Geol., 1, 108–121.
[5] Mucsi, G., Rácz Á., Mag G., Antal G., & Csőke B. (2019). Volume based closed-cycle Hardgrovegrindability
method. The Mining-Geology-Petroleum Engineering Bulletin, 34(4), 9-17.
[6] Todorovic, D., Trumic, M., Andric, L., Milosevic, V., & Trumic, M. (2017). A quick method for bond work
index approximate value determination. Physicochem. Probl. Miner. Process., 53(1), 321−332.
[7] Bond, F.C., & Maxson, W. L. (1943). Standard grindability tests and calculations. Trans. Soc. Min. Eng.
AIME, 153, 362–372.
[8] ASTM D409 (1991). Standard Test Method for Grindability of Coal by the Hardgrove Machine Method.
[9] Zeisel, H. G. (1953). Schriftenreihe der Zementindustrie. 14, Verein Deutcher Zementwerke, Düsseldorf, 51.
[10] Hoşten, Ç. & Gülsün, M. (2004). Reactivity of limestones from different sources in Turkey. Minerals
Engineering, 17(1), 97-99.
[11] Yin, L. & Guo J. (2011). Study on Wet FGD Limestone Quality, Third International Conference on
Measuring Technology and Mechatronics Automation, Shangshai, 605-608.
[12] Deniz, V. (2014). Relationships between Bond's grindability (Gbg) and breakage parameters of grinding
kinetic on limestone. Powder Technology, 139(3), 208-213.
[13] Altun, N.E. (2014). Assessment of marble waste utilization as an alternative sorbent to limestone for SO2
control. Fuel Processing Technology, 128, 461-470.
[14] Ozkahraman, H.T. (2005). A meaningful expression between bond work index, grindability index and
friability value. Minerals Engineering, 18(10), 1057-1059.
[15] Seo, J.H., Baek, C. S., Cho, J.S., Ahn, Y.J., Ahn, J.W., & Cho, K.H. (2019). Evaluation and application of
grinding index of domestic desulfurization limestone. Journal of Energy Engineering, 28(1), 1-9.
[16] Tichanek, F. (2008). Contribution to determination of coal grindability using Hardgrove Method. GeoSci.
Eng., LIV (1).
[17] Chelgani, S. C., Hower, J. C., Jorjani, E., Mesroghli, Sh., & Bagherieh, A. H. (2008). Prediction of coal
grindability based on petrography, proximate and ultimate analysis using multiple regression and artificial
neural network models. Fuel Proc. Technol, 89, 13–20.
[18] Sahoo, R.K. (2006). Review: An investigation of single particle breakage tests for coal handling system of
the gladstone port, Powder Technology, 161(2), 158-167.
[19] Ozbayoglu, G., Ozbayoglu, A. M., & Ozbayoglu, M. E. (2008). Estimation of Hard grove grindability index
of Turkish coals by neural networks. Int. J. Miner. Proc., 85, 93–100.
[20] Sengupta, A.N. (2002). An assessment of grindability index of coal, Fuel Processing Technology, 76(1) 1-
10.

136
 BŞEÜ Fen Bilimleri Dergisi BSEU Journal of Science
9(1), 124-137, 2022 https://doi.org/10.35193/bseufbd.997319

e-ISSN:2458-7575 (https://dergipark.org.tr/tr/pub/bseufbd)

[21] Mendis, B.S.M., Jayathunga, T.H.G.S., Madurapperuma, H.H., Rohitha, L.P.S., Dharmarathna, P.G.R., &
Hemalal, P.V.A. (2017). Determination of existing relationship amonggrindability, chemical composition
and particle size of raw material mix at Aruwakkalu Limestone.
[22] Kural, A., & Ozsoy, C. (2004). Identification and control of the raw material bending process in cement
industry. International Journal of Adaptation Control Signal Processing, 18(5), 427-442.
[23] Frigione, G., Zenone, F. & Esposito, M. V. (1983). The effect of chemical composition on Portland cement
clinker grindability. Cement and Concrete Research, 13(4), 483-492.
[24] Ürünveren, A., Altıner, M., Kuvvetli, Y., Ural, O. B., & Ural, S. (2020). Prediction of Hard grove grindability
index of Afsin-Elbistan (Turkey) Low-grade Coals based on proximate analysis and ash chemical
composition by neural networks. International Journal of Coal Preparation and Utilization, 40(10), 701-
711.
[25] Bilen, M., Kizgut, S., Cuhadaroglu, A., Yilmaz, S., & Toroglu, İ. (2017). Coal grindability and breakage
parameters. International Journal of Coal Preparation and Utilization, 37(5), 279-284.
[26] Bilen, M., Kızgut, S., Yilmaz, S., Baris, K., & Cuhadaroglu, D. (2018). Grindability of coal changing with
burial depth. International Journal of Coal Preparation and Utilization, 38(2), 75-87.
[27] Yilmaz, S., & Bilen, M. (2016). Empirical Relationships of HGI in terms of proximate analysis of coal. In
XVIII International Coal Preparation Congress, 953-957.
[28] Cuhadaroglu, A. D., Kizgut, S., Yilmaz, S., & Zorer, Y. (2013). Characterization of the grinding behavior of
binary mixtures of clinker and colemanite. Particulate Science and Technology, 31(6), 596-602.
[29] Gouda, G. R. (1979). Effect of clinker composition on grindability. Cement and Concrete Research, 9(2),
209-218.
[30] Gedik, İ., Duru, M., Pehlivan, Ş., & Timur, E., (2005). 1/50.000 ölçekli Türkiye Jeoloji haritaları, İstanbul-
F22c Paftası, MTA Raporu, No:11, Ankara.
[31] ASTM C1271-99. (2012). Standard Test Method for X-ray Spectrometric Analysis of Lime and Limestone.
[32] Kumar, P., Sahoo, B.K., De, S., Kar, D.D., Chakraborty, S., & Meikap, B.C. (2010). Iron ore grind ability
improvement by microwave pre-treatment. Journal of Industrial and Engineering Chemistry, 16(5), 805-812.

137