Download as pdf or txt
Download as pdf or txt
You are on page 1of 29

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/337401569

Grindability and abrasive behavior of coal blends: analysis and prediction

Article  in  International Journal of Coal Preparation and Utilization · November 2019


DOI: 10.1080/19392699.2019.1694009

CITATIONS READS

0 63

6 authors, including:

Alamin Idris Mohamad azmi Bustam


Universiti Teknologi PETRONAS Universiti Teknologi PETRONAS
22 PUBLICATIONS   53 CITATIONS    263 PUBLICATIONS   2,721 CITATIONS   

SEE PROFILE SEE PROFILE

Nurul Ekmi Rabat Fahim Uddin


Universiti Teknologi PETRONAS Universiti Teknologi PETRONAS
18 PUBLICATIONS   62 CITATIONS    22 PUBLICATIONS   31 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Ionic Liquids Toxicity on Fresh Water Microalgae, Scenedesmus quadriqauda, Chlorella vulgaris & Botryococcus braunii ; Selection Criterion for Use in a Two-Phase
Partitioning Bioreactor (TPPBR) View project

Bio-MEMS Sensor Device for Non-Invasive Screening of Diabetes via Exhaled Breath Analysis View project

All content following this page was uploaded by Alamin Idris on 28 November 2019.

The user has requested enhancement of the downloaded file.


International Journal of Coal Preparation and Utilization

ISSN: 1939-2699 (Print) 1939-2702 (Online) Journal homepage: https://www.tandfonline.com/loi/gcop20

Grindability and abrasive behavior of coal blends:


analysis and prediction

Alamin Idris, Zakaria Man, Azmi Bustam, Nurul Ekmi Rabat, Fahim Uddin &
Hafiz Abdul Mannan

To cite this article: Alamin Idris, Zakaria Man, Azmi Bustam, Nurul Ekmi Rabat, Fahim
Uddin & Hafiz Abdul Mannan (2019): Grindability and abrasive behavior of coal blends:
analysis and prediction, International Journal of Coal Preparation and Utilization, DOI:
10.1080/19392699.2019.1694009

To link to this article: https://doi.org/10.1080/19392699.2019.1694009

Published online: 20 Nov 2019.

Submit your article to this journal

View related articles

View Crossmark data

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


https://www.tandfonline.com/action/journalInformation?journalCode=gcop20
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION
https://doi.org/10.1080/19392699.2019.1694009

ARTICLE

Grindability and abrasive behavior of coal blends: analysis and


prediction
Alamin Idris a, Zakaria Mana, Azmi Bustamb, Nurul Ekmi Rabata, Fahim Uddinc,
and Hafiz Abdul Mannanc
a
Chemical Engineering Department, Center of Contaminant Control & Utilization, Universiti Teknologi
PETRONAS, Bandar Sri Iskandar, Perak, Malaysia; bChemical Engineering Department, Institiute of
Contaminant Management, Universiti Teknologi PETRONAS, Bandar Sri Iskandar, Perak, Malaysia; cChemical
Engineering Department, Universiti Teknologi PETRONAS, Bandar Sri Iskandar, Perak, Malaysia

ABSTRACT ARTICLE HISTORY


Low-grade coals are blended with high-quality coals to meet eco- Received 27 June 2019
nomic, environmental, and quality specifications. Hence, the grind- Accepted 13 November 2019
ability and abrasiveness of coal blends are crucial economic and KEYWORDS
operational parameters. This work evaluates, analyzes, and predicts Abrasiveness index; coal
the grindability and abrasive behavior of coal blends. Three binary blends; Hardgrove
coal blends with common low-grade coal were first prepared at grindability index; low-grade
various ratios. Blends 1 and 2 were composed of identical and similar coal; predictive model
ranks, whereas Blend 3 was composed of different ranks. The blends
were analyzed using proximate, ultimate analyzers, and a Bomb
calorimeter. The grindability and abrasive behavior of the blends
were measured using Hardgrove grindability index (HGI) and
Yancey, Geer, and Price methods, respectively. Further, the coarser
(+75μm) and finer (−75μm) fractions of HGI experiment were char-
acterized using proximate, ultimate and heating value analyses. The
additivity of HGI values was observed for Blend 1 and Blend 2,
whereas, the non-additive behavior was observed in Blend 3.
Further, the blends’ mineral matter contents and abrasiveness index
were found to be additive. Several existing models were found to be
inaccurate for HGI predictions. Therefore, a new cross-validated
model using multi-linear regression was proposed. The model exhib-
ited better HGI predictions of coal blends with a coefficient of deter-
mination R2 = 0.9416.

Introduction
Today, coal blending is a common practice in coal-fired power plants to enhance eco-
nomic performance, meet boiler requirements, and comply with environmental regula-
tions (Shen et al. 2019). Although the cost and availability of coal are of prime
consideration, the reasons for coal blending vary depending on the challenges faced at
the coal-fired power plants. Generally, coal blending is performed to meet stringent
environmental regulations such as SO2 and NOx emissions, minimize operational pro-
blems, i.e., slugging, fouling and deposit formation, extend fuel flexibility, and enhance
combustion or boiler performance. Low-grade coal can be used in the blend to reduce the

CONTACT Alamin Idris alamino97@yahoo.com Institute of Contaminant Management, Chemical Engineering


Department, Universiti Teknologi PETRONAS, 32610 Bandar Sri Iskandar, Perak, Malaysia
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/gcop.
© 2019 Taylor & Francis Group, LLC
2 A. IDRIS ET AL.

cost of feed coal. Coals with high sulfur, nitrogen, and ash contents can be blended with
high-quality coals to overcome the emission issues and operational difficulties (Mohanty
et al. 2019). Indigenous coals are blended with commercially available coals to secure the
supply line for extended periods. In most cases, coals are blended to enhance the
performance of boilers in terms of energy production, cost, and operational safety
(Dobó and Fry 2018).
Grindability of coal or coal blend is a vital quality parameter in its mining, beneficia-
tion, and utilization, which determines the economics, capacity and efficiency in pulver-
ized-fuel fired power plants (Wang et al. 2018). For instance, in most of the coal-fired
power plants, about 5% of the energy produced is utilized in size-reduction operations
(Doroodchi, Zulfiqar, and Moghtaderi 2013). The performance of the mill (i.e., capacity
and efficiency) in a power plant is directly affected by the grindability of the coal
(Özbayoğlu, Özbayoğlu, and Özbayoğlu 2008; Radic et al. 2011; Tymoszuk et al. 2019).
The behavior is a reflection of interrelated characteristic properties such as hardness,
fracture, and tenacity (Rubiera et al. 1999), which are affected by the rank, moisture
content, organic composition and texture, brittleness, and composition of the mineral, and
maceral contents (Trimble and Hower 2003; Ural and Akyıldız 2004). Several authors have
even suggested the options of coal grindability enhancement using dispersant, microwave
treatment, ultrasonic wave irradiation methods, etc. (Atesok et al. 2005; Kingman 2006;
Lester, Kingman, and Dodds 2005). The grindability parameter can be measured with
a standard laboratory apparatus known as Hardgrove Grindability Index (HGI) machine,
which offers a measure of relative grindability or ease of pulverization of coal compared to
standard coals. Most of the pulverization operations are designed so that 70% of the
output coal passes through a 75-µm sieve (Mesh 200)(Steer et al. 2015). Thus, the HGI
apparatus is standardized similarly to grind coal sample in 60 revolutions, and the final
products are sieved through a 75-µm for the determination of HGI value (Sahoo et al.,
2011). Coals with higher HGI value are softer to grind, and thus, based on the relative
HGI values, coals are often classified as very hard (HGI<40), hard (40< HGI<60), med-
ium-hard (60< HGI<80), soft (80< HGI<100), very soft (100< HGI<120), and extremely
soft (HGI>120) (Kanwal, Sana, and Shafqat Nawaz 2017; Tichánek 2008).
The grindability index (HGI) value is related to the amount of energy required by the
mill to grind coal into the desired fineness (Tichánek 2008; Xie et al. 2015, 2018). The
work index (Wi) is commonly used to describe the amount of energy required in kWh/ton
of coal. Several authors have reported approximate correlations of Wi and HGI values, as
summarized chronologically by Hower (1998). The first expression correlating the work
index with the grindability index in wet grinding condition was proposed by Bond, as
follows (1961):

Wi ¼ 435 ðHGIÞ0:91 (1)

Later, the correlation was modified to cover a wide range of brittle materials, limestone,
subbituminous and bituminous coal, and materials with Wi > 8.5 kWh/ton. The modified
correlation between Wi and HGI is expressed as follows (Williams et al. 2015):

Wi ¼ 1622 ðHGIÞ1:08 (2)
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 3

If the HGI value is lower, the coal is harder to grind to the desired size and hence, requires
higher energy/power consumption. Higher mill capacity can be achieved with higher HGI
value coal.
The demand for energy conservation in the pulverization plant is attributed to the
grindability of the coal blend and the operating cost of the mill. Moreover, the main-
tenance cost of the milling parts is also of great concern. Coal ash contains minerals such
as quartz and pyrite which are known to be harder than mild/carbon steel. When these
mineral constituents are contacted during the pulverization process, they tend to abrade
the steel of the milling part, which, in turn, affects the performance of the mill (Nahvi,
Shipway, and McCartney 2009; Wells et al. 2004). Thus, besides the grindability index, the
abrasive behavior of the coal blends must also be identified for utilization considerations.
The abrasiveness index (AI) of coal or coal blends is commonly determined to measure
the abrasive behavior of coal using a standard method known as Yancey, Geer, and Price
(YGP) method (Wells et al. 2005). The method quantifies the amount of steel metal that
would be abraded away if contacted with a unit weight of coal. The abrasiveness of coals is
influenced by coal constituents such as moisture, ash content, minerals (such as quartz,
pyrite, and clays), and microlithotypes and macerals. Hardness and grindability of coals
may also influence the abrasive behavior.
The test procedures for the grindability and abrasiveness indices (the HGI and YGP
methods) are tedious, time-consuming, and require special equipment (Deniz and Umucu
2013). Thus, the need for predictive models to estimate these important parameters (i.e.,
the grindability and the abrasiveness indices) always remained in demand. Several authors
developed models based on the proximate analysis, ultimate analysis, and petrographic
analysis. For instance, Sengupta (2002) correlated the proximate analysis and grindability
referred to as Statistical Grindability Index (SGI), with a high coefficient of determination
(R2 = 0.93). Peisheng et al. (2005) reported the prediction of HGI using Generalized
Regression Neural Network (GRNN) based on the proximate analysis (R2 < 0.95).
A support vector regression (SVR) model based on proximate analysis was also proposed
with a high coefficient of determination (R2 = 0.93) by Venkoba Rao and Gopalakrishna
(2009). Khoshjavan and Rezai (2013) predicted HGI with R2 = 0.96 using Artificial Neural
Network (ANN) based on proximate, ultimate, and heating value data. Another ANN
prediction was also reported based on petrographic analysis with a maximum R2 of 0.92
(Bagherieh et al. 2008). Chelgani et al. (2008) reported three prediction models of HGI
using the least-square regression method based on the proximate, ultimate, and petro-
graphic analysis. Further, the authors improved the predictions using Feed-forward
Artificial Neural Networks (FANN) with high coefficients of determination R2 = 0.89,
0.89, and 0.95, respectively. Matin et al. (2016) developed a random forest (RF) model
(R2 = 0.90) to predict HGI values using data obtained from proximate, ultimate, petro-
graphic, and ash analysis. The results in these studies seem to predict HGI values with
acceptable accuracies; however, they are not cross-validated against other coal samples.
This allows the possibility of overfitting, which is one of the reasons why a model with
high accuracy for its training data can fail to predict HGI value for other coal samples or
blends.
This work aims to evaluate, analyze, and model the grindability and abrasiveness
behavior of coal blends. The HGI and AI measurement of coal blends were carried out
using the standard HGI and YGP methods, respectively. Parent coals with different ranks
4 A. IDRIS ET AL.

and fuel quality were blended at various proportions. The parent coals and blends were
first characterized at particle size −212 μm using standard methods for the proximate,
ultimate, and heating value analyses. Later, the grindability index of coal and coal blends
was determined using the standard HGI machine, whereas the abrasiveness index of coal
blends was determined using the YGP method. To identify the impact of coal constituents
on its grindability, the coarser (+75 μm) and finer (−75 μm) fractions of the HGI
experiment were further analyzed for the proximate, ultimate, and heating value analyses.
To avoid particle size differences on the analysis results, the coarser fraction was first
crushed to pass −75 μm sieve prior to the analysis. The results of fuel quality parameters,
experimental HGI values, work index, and abrasiveness index of the blends were inves-
tigated as a function of wt.% of the low-grade coal. The existing models’ HGI predictions
were compared with experimental observations, and their drawbacks were identified.
Accordingly, a new cross-validated model based on quality parameter data using a multi-
linear regression technique has been proposed to predict the HGI value of coal blends.

Experimentals
Coal and Blend Sample Preparation
Four parent coals with varying ranks originating from different geographical locations
were selected for the binary coal blend preparation. Among the coals, one is low-grade
/high-ash coal. The remaining parent coals were chosen to minimize the negative impacts
of low-grade coal. The sample names and rank of the parent coals are shown in Table 1.
For the grindability test, the parent coal samples were first crushed to pass a sieve of
4.75 mm aperture and dried for 24 h in open-air separately. The grindability test samples
were prepared according to ASTM D 409 02 (ASTM International 2002a), where the
parent coals were sieved/screened in a series of sieves with 1.18 mm and 0.6 mm aperture
sizes (i.e., Mesh No. 16 and Mesh No. 30). The HGI sample is thus the fraction retained
between the sieves (1.18 mm-0.6 mm). The two fractions (i.e., > 1.18 mm and < 0.6mm)
are discarded. Fifteen coal blends were prepared for HGI measurement with particle sizes
of 1.18 mm-0.6 mm on weight basis with blending ratios described in Table 2, maintain-
ing the low-rank coal in all the binary coal blends.

Characterization of Coal and Coal Blends


The proximate analysis of the parent coal samples with particle size −212 μm was carried
out using TGA 701, (LECO, Michigan, USA) according to the ASTM D5142-02a test
method (ASTM International 2002b). The moisture (MC), volatile matter (VM), fixed
carbon (FC), and the ash (AC) contents in wt.% were determined. The ultimate analysis

Table 1. The rank of the sample parent coals.


Sample Name Coal Rank
Coal 1 Sub-bituminous coal A (SubA)
Coal 2 Sub-bituminous coal A (SubA)
Coal 3 Sub-bituminous coal B (SubB)
Coal 4 High volatile C bituminous coal (hvCb)
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 5

Table 2. Coal blends sample preparation with different blending


ratios.
Blend 1 Blend 2 Blend 3
S.No. Coal 1:Coal 2 Coal 1:Coal 3 Coal 1:Coal 4
1 95:05 95:05 95:05
2 80:20 80:20 80:20
3 70:30 70:30 70:30
4 50:50 50:50 50:50
5 30:70 30:70 30:70

was also carried out using CHN628 (LECO, Michigan USA) according to ASTM D3176-
09 (ASTM International 2009), where the carbon (C), hydrogen (H), and nitrogen (N)
contents were determined, whereas the sulfur (S) content (wt.%) was determined using
TruSpec S (LECO, Michigan USA) according to ASTM D 4239–04 (ASTM International
2004). Oxygen (O) content was calculated by difference. The mineral matter (MM%)
content of the sample was calculated using Parr formula
MM% ¼ 1:08AC þ 0:55S (3)

The heating value, i.e., gross calorific value (GCV) of the parent coals, were determined
following ASTM D 5865-11a (ASTM International 2011) using automatic Bomb calori-
meter AC 500(LECO, Michigan, USA). The calorific value of the parent coals was also
used to confirm the coal rank, according to ASTM D 388–99 (ASTM International 2000).
Similarly, the proximate, ultimate, and heating value analyses were also conducted for 15
coal blend samples blended at −212 μm particle sizes.

Hardgrove Grindability Index Measurements


For the determination of the HGI value of the parent coal and coal blends ASTM
D 409–02 (ASTM International 2002a) test procedure was adopted, where the HGI
machine was first calibrated against standard coal samples with known HGI values (i.e.,
40, 60, 80, 100). 50.00 ± 0.01 g of the parent coal or coal blends sample (0.6 mm-1.18 mm)
was placed in the bowl of the machine and ground for 60 ± 0.25 revolutions at speed of 20
rev/min. The ground sample was then sieved with a 75-µm screen using an electromecha-
nical vibrator for 5 min. HGI values were calculated from the weights of the fraction
retained on 75 µm screen or the fraction that passed through 75 µm screen using
Equations (4,5), respectively.
HGI ¼ 13 þ 6:93ðYÞ (4)

HGI ¼ 359:5  6:93ðXÞ (5)

where, X and Y are the weights of −75 µm and +75µm fractions (g), respectively.
Experiments were repeated to ensure the repeatability of HGI values lay within ±2 units
(Özbayoğlu, Özbayoğlu, and Özbayoğlu 2008). To account for the weight loss during
sample transfer and sieving as dust or by sticking on walls of the bowl, the HGI values
calculated from both fractions were averaged and reported.To compare the energy/power
6 A. IDRIS ET AL.

requirements for grinding the blends of subbituminous and bituminous coals, the mod-
ified Bond’s correlation shown by Equation (2) was used to calculate the work index in
kWh per ton of the material.

HGI Fractions Analysis


Fractions of the HGI measurement were analyzed to assess the impact of coal or blend
characteristics on grindability behavior. The proximate and ultimate analyses were used to
characterize both (−75 µm and +75 µm) fractions. The +75 µm fractions were first crushed
to pass the 75 µm sieve using a vibratory mill. This was made to avoid the effects of the
particle size difference of the fractions on the proximate and ultimate analyses results.

Abrasiveness of Coal Blends


The abrasiveness index values of coal or blends were determined using the YGP test rig
according to British standard BS 1016:19 (British Standard 1980), which consisted of four
mild steel blades positioned at the center of the vessel. About 2 kg of coal or blend batch
sample was placed in the YPG test setup, and the steel blades were rotated mechanically
with a velocity of 12,000 ± 6 revolutions at a speed of 1470 rev/min. The weight of the
blades was measured before and after the test. Abrasiveness index (AI) was calculated by
the total weight loss from the four steel blades in milligrams per kilogram of coal or coal
blend tested (Wells et al. 2004). AI is described as follows:
mg of metal lost
AI ¼ (6)
kg of coal or coal blend

Prediction of Grindability
The predictions for the HGI values of blends were assessed using various existing
predictive models, i.e., the weighted average linear additive model, statistical grindability
index by Sengupta (2002), and HGI prediction models by Chelgani et al. (2008) and
Peisheng et al. (2005). The results are compared with experimental HGI values. The linear
additive model is expressed as:
M ¼ x1 M1 þ x2 M2 (7)

where M is the predicted coal blend HGI value, whereas, M1, x1, and M2, x2 are HGI
values, and weight fractions of the parent coal 1 and 2 in the binary coal blend,
respectively. The percentage deviation from the experimental values (Me) is calculated as:
D% ¼ ½ðMe  M Þ=M   100 (8)

The statistical grindability Index (SGI) proposed by Sengupta is determined using the
proximate composition analysis (Sengupta 2002). This SGI is claimed to cover all coal-
fields in India with 300 samples, having 0.7-19% moisture, 3-65% ash, 16-45% volatile
matter and 15-66% fixed carbon. Due to the high correlation coefficient (R2 = 0.77),
similar samples origin, and large ranges of proximate analysis variables, the correlation is
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 7

expected to provide accurate predictions of the coal samples used in this study.
Mathematically, (Sengupta 2002):
SGI ¼ 93:25 þ MCð0:256 þ 0:196MCÞ þ 3ACð1:097  0:009ACÞ
(9)
 3VM ð1:165  0:029VM Þ  5FCð1:103  0:00166FCÞ

The HGIs proposed by Chelgani et al. (2008) were determined for a wide range of Kentucky
coal samples using the proximate and ultimate composition analyses, respectively. These
HGI determination models were developed using 600 samples, having 0.8–13.2% moisture,
0.64–59.8% ash, 16.6–46.4% volatile matter, 0.30–7.60% total sulfur, 28.9–83.5% carbon,
2.48–6.05% hydrogen, 0.09–2.34% nitrogen and 0–20.6% oxygen. These ranges are large and
should be able to predict the HGI for other coal samples if the data is not over-fitted.
For proximate analysis (R2 = 0.77),
HGI ¼ 102:69 þ 4:227Stotal  1:634VM  0:569AC  0:237MC (10)

For ultimate analysis (R2 = 0.75),


HGI ¼ 77:162 þ 3:994lnðStotal Þ  10:920H þ 1:904MC  0:424AC
 
OþN
 11:765ln (11)
C

Peisheng et al. (2005) also proposed two correlations for the estimation of grindability
index of Chinese coals. Proximate analysis parameters were utilized for the prediction
using GRNN. The first correlation (R2 = 0.66) was linear, whereas the second correlation
was quadratic with a higher coefficient of determination (R2 = 0.76).
The linear correlation was defined as,
HGI ¼ 95:11266  1:869MC  0:9145VM þ 0:46AC (12)

Whereas the quadratic correlation was defined as,


HGI ¼ 167:5  10:1MC þ 4:933AC þ 2:17VM þ 2:55FC
(13)
þ 1:99MC2  0:0356AC2  0:015VM 2  0:0038FC2

Results and Discussion


Parent Coals
The parent coal quality parameters obtained from proximate, ultimate, and heating
value analyses are shown in Table 3. The moisture content of the parent coal samples
has the range (3.48 − 10.2 wt.%). It can also be observed that coal 1 contains high ash
content (51.7 wt.%), low fixed carbon content (FC = 19.4 wt.%), and gross calorific
value (GCV = 2420 kcal/kg) as compared to the other parent coals. Moreover, the
sulfur content of the coal 1 is 1.09 wt.% which could be objectionable by environ-
mental regulatory authorities. Coal 1 contains the highest amount of incombustible
mineral matter (56.44 wt.%), which lowered its energy content, affecting the overall
quality of the coal. Therefore, coal 1 can be conveniently referred to as low-grade coal,
8 A. IDRIS ET AL.

Table 3. Proximate, ultimate analysis, heating value, grindability and abrasiveness indices
of parent coals.
Parent Coal Coal 1 Coal 2 Coal 3 Coal 4
Proximate Analysis wt.%
MC 5.52 7.97 10.2 3.48
VM 23.3 43.20 55.1 27.79
FC 19.4 39.46 29.0 53.41
AC 51.7 9.37 5.52 15.32
Ultimate Analysis wt.%
C 31.2 62.0 59.7 67.7
H 2.47 4.62 4.35 3.74
N 0.38 0.62 0.26 1.42
S 1.09 0.62 0.26 0.78
O 7.61 14.8 19.4 7.60
Heating Value, kcal/kg
GCV 2420 4970 4770 5350
Mineral Matter wt.%
MM% 56.44 10.46 6.10 16.97
Grindability Analysis
HGI 94 59 72 64
Abrasiveness Index, mg/kg
AI 80 26 19 16

which is required to be upgraded for its ultimate utilization. Hence, blending the low-
grade with coals of better quality is the best strategy to meet the quality parameters
needed for the efficient operation of the combustion chamber in PF boilers. Thus,
coals with low ash contents (5.52–15.32 wt.%) or lower mineral matter contents and
higher fixed carbon contents (29.0–53.41 wt.%) were blended with low-grade coal 1.
The ranks of the parent coals were first reconfirmed using their heating values, i.e., GCV,
according to the method ASTM D 388–99(ASTM International 2000), where coals are
classified into ranks according to their respective gross calorific values on a moist, mineral
matter free (mmf) basis. Thus, accordingly, coal 1 was identified as Subbituminous A coal
(SubA), whereas coal 2, 3, and 4 were identified as Subbituminous A coal (SubA),
Subbituminous B coal (SubB), and high volatile C bituminous coal (hvCb), respectively.
Therefore, according to the ranks of the parent coals, the blends were made among similar
and different coal ranks, i.e., blend 1 (SubA-SubA), blend 2 (SubA-SubB), and blend 3 (SubA-
hvCb). It is clear that blend 1 (coal 1-coal 2) is of the same ranks, i.e., subbituminous A coal,
but the parent coals in blend 1 are different in terms of quality parameter contents. Further,
coal 3 is also of similar rank with least ash content (5.52 wt.%) as compared to coal 1. However,
coal 4 is of different rank with the highest fixed carbon content (53.41 wt.%) and energy
content 5350 kcal/kg. The blend of coal 1 (low volatile coal) and coal 4 (hvCb) (i.e., Blend 3)
presents an apparent rank.
The grindability indices of the parent coals were determined using both Equations (4),
(5). The experiments were repeated three times, and the HGI values of both fractions were
averaged and considered to determine the parent coal grindability with less than ±2 units
tolerance, as shown in Table 4.
It is surprising to observe that the softer coal (Coal 1) with HGI value of 94 has a severe
abrasive behavior on the surfaces of mild steel parts of the mill, as shown in Table 3. It
must be noted that coal 1 has the highest mineral matter content as compared to other
coals studied. Despite being a low-grade coal based on the quality parameters, Coal 1 is
observed to be significantly abrasive with an abrasive index of 80 mg of mild steel lost/kg
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 9

Table 4. Measured HGI values from the weight of finer and coarser fractions.
Sample HGI (−75 fraction) HGI (+75 fraction) Average HGI Hardness
Coal 1 93.56 94.60 94 Soft
Coal 2 59.30 57.79 59 Hard
Coal 3 71.54 73.06 72 Medium hard
Coal 4 63.53 64.87 64 Medium hard

of coal. The least abrasive coal is the most preferable for the safe operation of pulveriza-
tion mills. The increasing order of the abrasiveness index among the parent coals is as:
Coal 4 < Coal 3 < Coal 2< Coal 1. The energy/power requirement to crush these coals to
a pulverized form (<75 μm particle sizes) for its utilization in PF boiler combustion
chamber in terms of Bond’s work index (Wi) were calculated from HGI values according
to Equation (2). Thus, the order of work index of parent coals was obtained as Coal 1
(12.00 kWh/ton) < Coal 3 (16.00 kWh/ton) < Coal 4 (18.17 kWh/ton) < Coal 2 (19.84
kWh/ton).
According to the grindability results of these parent coals, higher ash or higher mineral
matter content coal exhibited higher HGI value, implying that coals with higher ash or
mineral matter content are softer to grind, thus, the low-grade coal (Coal 1: SubA) with
the highest mineral matter content (56.44 wt.%) is softer to grind with HGI value of 94.
However, coal 2 (SubA), though of the same rank as coal 1, the grindability index was
measured to be 59, attributed to its lower mineral matter content (10.46 wt.%). The
comparison of grindability of coal as a function of its ash or mineral matter content
may not be a good indicator if the ranks of the coals compared are different. In contrast, it
is the type of minerals present in the coal matrix that determines the grindability behavior
(Hansen and Hower 2014). For instance, coal 3 (SubB) has the lowest mineral matter
content (6.10 wt.%), but it is softer to grind than coal 2 (SubA) and coal 4 (hvCb) with
mineral matter contents of 10.46 and 16.97 wt.%, respectively.
It is evident that in addition to the ash/mineral matter contents, there are also other
components within the coal matrix that can affect the grindability of coal. To identify
these components and their impact on the coal grindability of different ranks, the fractions
produced in HGI measurements, i.e., the coarser (+75 μm) and finer (+75 μm) fractions
were further analyzed using proximate, ultimate and heating value analyses. The prox-
imate analysis results of these fractions for the parent coals are shown in Fig. 1.
According to the proximate analysis of HGI fractions, it is clear that the ash content of
coal 1 is significantly higher in the finer fraction. This implies that the coal’s ash or
mineral matter contents are relatively softer to grind among the other constituents,
whereas the quantities of the fixed carbon and volatile matter contents are observed to
be higher in the coarser fraction. These do not contribute to hardness of the coal; rather
they are measures of coal ranks. Similar observations can be made for coals of similar
ranks, i.e., coal 2 and 3. Moreover, the fractional difference in volatile matter contents for
different rank coal, i.e., coal 3, appeared to be insignificant, where almost similar amounts
were determined with both fractions. Moreover, the moisture contents appeared to be
similar with both fractions. Furthermore, these moisture contents of fractions display
a lower amount when compared with the untested samples, as shown in Table 3; the
10 A. IDRIS ET AL.

Figure 1. Parent coals HGI fractions proximate analysis of a) finer (−75 μm) and b) coarser (+75 μm)
fractions.
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 11

Figure 2. Parent coals HGI fractions ultimate analysis a) Carbon, b) Hydrogen, c) Nitrogen and d) Sulfur
contents.

reduced amount corresponds to the amount of moisture lost due to grinding the coal for
60 revolutions in the HGI machine.
The ultimate analysis results for the HGI fraction of parent coals are presented in Fig. 2.
Complimentary to the proximate analysis for fixed carbon contents shown in Fig. 1,
among the organic elemental composition of coal, it is observed that elemental carbon
is found higher in the coarser fraction. A similar trend in hydrogen content values was
also observed. However, for the nitrogen and sulfur contents, the results obtained are
inconclusive, with negative results for coal 2 (SubA) parent coal. Furthermore, the values
of energy content (GCV) were observed to be higher in the coarser fraction, as shown in
Fig. 3. The results are in good agreement to the proximate and ultimate analyses results,
which revealed an increased fixed carbon and elemental carbon contents in the coarser
fractions, respectively.

Coal Blends
Parent coals with a particle size range of −212 μm, were blended according to the ratios
presented in Table 2. The three sets of binary blends with 5 different blending ratios were
characterized using proximate, ultimate, and heating value analyses. The HGI and AI
indices were measured in typical conditions as that of the parent coals. The results of the
12 A. IDRIS ET AL.

Figure 3. Parent coals HGI fractions heating values (GCV).

aforementioned analyses and measurements for 15 blends are presented in Table 5.


According to the results obtained, the hardness of the blend to grindability has increased
with a decrease in HGI values. In all cases of the blends, it is harder to grind the composite
coal blend than the low-grade high ash coal. The other parent coals in the blends are hard
and medium-hard coals based on their grindability index values as shown in Table 4.
Thus, the decrease in the blending ratio of the low-grade coal in all blends with such
medium-hard and hard coals resulted in reduced HGI values or decreased grindability
behavior of the blend, which would require higher energy requirement to grind to
pulverized coal particle size ranges.
The effects of moisture content on HGI grindability have been studied for a wide
range of moisture contents by several authors (Kanwal, Sana, and Shafqat Nawaz 2017;
Shahzad et al. 2014; Vuthaluru et al. 2003). Besides the fact that increased moisture in
blend reduces the energy content of the blend, as such, in most of the coal utility plants,
such as power plants boilers, cement kiln, steel industries, etc. prefer the optimum
moisture content range of 4–8 wt.% (Pankaj et al. 2017). Fortunately, the moisture
contents of the blends studied are mostly within the acceptable ranges for direct utiliza-
tion. Thus, the investigation of moisture effects on HGI value at these ranges is not
significant.
Comparing the mineral matter contents of the coal blends with the HGI grindability
index, it clear that some proportionality relationships can be established despite the
variations caused due to blending coal of different rank or maturity. More elaborative
studies of HGI fractions on combined maceral and microlithotype composition of high-
volatile bituminous coal blends of progressive grinding were reported by Trimble and
Hower (2000). They concluded that the comparisons of pulverized coal with the
laboratory experiments were imprecise due to probable shifts in the blend composition.
Generally, mineral matter/ash contents of the coal blend are known to cause the
slugging and fouling problems; their behavior in the boiler is unpredictable in most of
Table 5. Proximate, ultimate analysis, heating value, grindability and abrasiveness indices of coal blends (−212 μm).
Blends Blend 1 Blend 2 Blend 3
Ratio 95/05 80/20 70/30 50/50 30/70 95/05 80/20 70/30 50/50 30/70 95/05 80/20 70/30 50/50 30/70
Proximate Analysis, (wt.%)
MC 5.91 6.02 6.33 6.62 6.75 5.54 6.04 6.48 7.91 8.66 5.33 5.19 4.65 4.37 3.99
VM 25.97 29.79 31.23 35.17 38.58 26.59 29.44 32.43 35.55 40.18 24.80 25.96 25.84 27.00 27.79
FC 22.09 25.82 26.52 30.85 33.21 21.60 24.00 25.99 27.78 31.73 22.69 28.04 30.34 37.31 44.05
AC 46.03 38.37 35.91 27.36 21.46 46.27 40.52 35.10 28.76 19.43 47.18 40.81 39.17 31.32 24.17
Ultimate Analysis, (wt.%)
C 34.01 38.88 40.59 47.23 51.51 33.65 37.14 40.43 43.74 48.59 33.80 39.82 41.62 49.54 57.11
H 2.81 3.16 3.26 3.58 3.86 2.69 2.97 3.13 3.27 3.51 2.59 2.82 2.81 2.91 3.41
N 0.57 0.63 0.65 0.73 0.76 0.53 0.54 0.55 0.52 0.52 0.74 0.91 0.95 1.17 1.38
S 0.94 0.99 1.02 0.91 0.84 1.10 0.92 0.92 0.88 0.67 1.13 1.05 1.00 0.97 0.87
O 9.73 11.94 12.24 13.57 14.82 10.22 11.87 13.39 14.92 18.62 9.24 9.40 9.80 9.72 9.07
Heating Value, (kcal/kg)
GCV 3270 3700 3850 4470 4870 3180 3500 3790 4100 4480 3190 3720 3820 4470 5220
Mineral Matter (wt.%)
MM 50.23 41.99 39.34 30.05 23.64 50.58 44.27 38.41 31.54 21.35 51.57 44.65 42.85 34.36 26.58
Grindability Index
HGI 94 91 82 79 73 93 92 92 89 81 76 72 68 72 70
Abrasiveness Index (mg/kg)
AI 76 70 62 55 44 79 69 60 51 36 75 69 62 47 34
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION
13
14 A. IDRIS ET AL.

the cases. The ash or mineral matter in coal blends interacts at high temperatures to yield
the slugging and fouling issues on the walls of the boiler.
In most coal blending activities, the power plant operators strive to

(1) meet the boiler requirements by defining the maximum ash or mineral matter
contents
(2) enhance the combustion performance by increasing the heating value and reducing
the fuel ratio (FC/VM),
(3) meet environmental restrictions by using coals of reduced pollutant contents in the
blend, and
(4) perform economic energy production by using a cheaper/low-grade, a softer to
grind, and less abrasive coal in the blend.

It is desirable that the composite coal blend has higher heating value and lower fuel
ratio for better combustibility and lower mineral matter/ash, sulfur, and nitrogen contents
to meet the emission regulation. Furthermore, for economic and safety reasons, it is also
equally desirable to have coal blends with enhanced grindability (as an energy-saving
incentive) and reduced abrasiveness behavior of blends (operational equipment safety).
These parameters have a strong impact on the performance (i.e., efficiency and capacity)
of the coal-fired PF boilers.
To demonstrate the important parameters in coal blends, plots of heating value GCV,
mineral matter contents, work index, as well as the abrasive index as a function of wt.% of low-
grade coal in the coal blends are depicted in Fig. 4. It can be observed that with an increase in
the low-grade coal content in the blends, the heating value (GCV) of the blends have decreased
almost linearly. The calorific value is an additive coal property, which can be estimated as the
weighted average of the parent coals’ heating values. However, the mineral matter contents
displayed a linear increase with the increase in low-grade coal content in the blends, similar
observation of linear increase can also be seen with the abrasiveness index of the coal. These
imply that the mineral matter contents, as well as the abrasiveness index, are additive proper-
ties. The extent of deviation may differ based on the rank of the parent coal in the blends, but
in all cases, the deviations are within the acceptable range. Besides the abrasive behavior of the
coal blends, the work index, a measure of the energy required to grind the coal blends, is also
an important parameter in the pulverization plant. It’s assessment prior to utilization of the
coal blends is crucial for the economic and capacity considerations. The parameter is estimated
from the HGI grindability value of the blend using nonlinear correlation; thus, obviously, it
appears to have deviated from the weighted average of the parent coals. Moreover, the severity
of the nonlinearity depends on the rank of the parent coals used in the blend.
Therefore, the operators of the power plants must assess and evaluate the coal blend
properties before utilization. The criterion of the assessment could be more detailed or
extended to include other crucial parameters such as fuel (FC/VM) ratio, O/C ratio, sulfur,
or nitrogen contents depending on the challenges encountered at the plant.

Prediction of HGI Values of Coal Blends


The experimental HGI values and the weighted average (linear additive model prediction)
values are compared in Fig. 5 and Fig. 6 for blend 1, blend 2, and blend 3, respectively. For
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 15

Figure 4. Effects of low-grade coal blending ratio on a) heating value GCV, b) mineral matter content,
c) work index, and d) abrasiveness index of coal blends.

Figure 5. Experimental and predicted HGI of coal blends a) Blend 1, and b) Blend 2.
comparison purposes, the parent coals’ HGI values can be observed at 0% and 100% low-
grade coal. According to the results, it can be observed that the linearity of HGI values
depends on the parent coal ranks present in the blend. Based on the comparison with
16 A. IDRIS ET AL.

Figure 6. Experimental and predicted HGI of coal blend (Blend 3).

weighted average, blend 1 is additive with close linearity of R2 = 0.969, however, as the
rank of one of the parent coal in the blend is changed, the deviation increases as observed
in Fig. 7 for reduced correlation values with R2 = 0.897 and 0.414, for blend 2 and blend 3,
respectively.
It can be noted that HGI values for all coal blends lie within the parent coals HGI
values. The measured HGI values of the coal blends, i.e., Blend 1 and Blend 2, are
observed to be higher than the predicted values, which are in agreement with blends
reported by Rubiera et al. (1999), and Hower (1988). However, for the case of blend 3 as
shown in Fig. 6, the measured HGI exhibits a different trend when compared to the
weighted average predictions with severe deviation departing from the linearity

Figure 7. Correlation coefficient of predicted and measured HGI values for coal blends.
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 17

relationship. The reason for such deviation is that the Blend 3 is composed of coals of
different ranks (i.e., SubA and hvCb). If the ranks of parent coals are similar, the HGI
value of the blend can be predicted as the weighted average of the individual coals. In
other words, the HGI value is additive property for similar coal ranks (see Fig. 7).
However, determining the linearity of the coal blend based on the percentage difference
in HGI values of the parent coals may not be valid for all cases. For instance,Ananda Babu,
Lawrence, and Sivashanmugam (2018) set the limit for the HGI value percentage differ-
ence of the parent coals to 25% for the HGI value of the blend to be additive. However, the
percentages differences in the HGI value of the blends studied here are above 25%, i.e.,
blend 1 (45.75%), blend 2 (26.50%), and blend 3 (37.97%). In contrast to the 25%
criterion, blends with similar ranks regardless of the percentage difference in the HGI
values of parent coal are still additive (e.g., Blend 1, and Blend 2). However, if the coal
ranks of the parent coals are dissimilar (e.g. Blend 3), the 25% percentage difference
criterion is still valid. Therefore, it can be concluded that the 25% percentage difference of
HGI values criterion can only be applied to the coal blends of different ranks.
The statistical grindability index (SGI) proposed by Sengupta is a predictive correction
based on the proximate analysis results (Sengupta 2002). By using the proximate analysis
data for the coal blend, the SGI values were calculated. It should be noted that the
proximate analysis results of the coal samples used in this study are within the ranges
defined by the Sengupta. The comparison of SGI values with the measured HGI grind-
ability for all blends is plotted as shown in Fig. 8 (a). According to the correlation
coefficient obtained (i.e., R2 = 0.1288), no significant relationship is observed between
the predicted and the measured values, so the SGI model cannot be used for the blends
studied. Only a few values are close to the experimental observation, which does not
justify the usage of this correlation for HGI predictions. Some of the magnitude of errors
are as high as 50% of the observed HGI value. Therefore, the SGI expression by Sengupta
does not apply to the coal blends in this study.
HGI model proposed by Chelgani et al. (2008) predicts the HGI based on the prox-
imate analysis results, as shown in Equation (10). HGI values were calculated using the
proximate analysis data for the coal blends. The proximate analysis results of the coal
samples used in this study lie within limits defined by Chelgani et al. (2008). The
comparison of HGI values with the measured HGI grindability for all blends is plotted
as shown in Fig. 8 b). The magnitude of errors are high, starting from 13 and the
correlation coefficient is very low (R2 = 0.009). The correlation grossly underestimates
the HGI values of the coal blends in this study. Some HGI values are even predicted to be
less than ten, which is unrealistic.
Chelgani et al. (2008) also proposed an HGI model based on the ultimate analysis of
coal, as presented in Equation (11). The correlation also claims to predict HGI values for
a wide range of coal properties using ultimate analysis. However, this correlation also fails
to predict the HGI values for the coal samples in this study, as shown in Fig. 8 b).
Interestingly, the prediction error of relatively harder coals is less than the softer coals.
The magnitude of errors are high, starting from 11%, and the correlation coefficient is very
low (R2 = 0.009). The correlation also underestimates the HGI values of the coal blends in
this study; therefore, it cannot be used to predict the HGI of the coal samples in this study.
Peisheng et al. (2005) also proposed two models for the prediction of HGI for coal
samples from China using proximate analysis. One model is linear, whereas the other is
18 A. IDRIS ET AL.

Figure 8. Comparison measured HGI values of coal blends and predicted HGI values using a) Sengupta
(Sengupta 2002), b) Chelgani et al,(Chelgani et al. 2008) c) Peisheng et al.(Peisheng et al. 2005) linear,
and d) quadratic model.

quadratic. These correlations are also not suitable for HGI predictions for the selected
blends, as shown in Fig. 8 c) and Fig. 8 d). The predicted values of these correlations are
only able to achieve weak coefficients of determination R2 = 0.14, and 0.24, respectively.
The R2 value for the quadratic expression, however, seems to be better as compared to the
other correlations selected for HGI predictions. Therefore, it can be concluded that none
of the correlations can predict the HGI values for the coal samples (i.e., the parent coals
and coal blends) in this study.

Proposed HGI Prediction Model


It can be noted that the current models are unable to provide reliable estimates for HGI
values of different coal blends. The primary reason for this discrepancy can be attributed
to the absence of inverse terms. All the terms of the current correlations are proposing
either a linear or a quadratic correlation of some or all of the related variables, which is
questionable in practice. Furthermore, it is not clear if the correlations developed by the
researchers were cross-validated. Here, cross-validation of the proposed correlation is
introduced to avoid any possibility of over-fitting. Hence, the use of all four parameters
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 19

in Sengupta’s SGI correlation based on proximate analysis is not justified, as the system is
closed, i.e., the four parameters add up to 100% (Chelgani et al. 2008).
Keeping these points under consideration, the following measures have been taken to
model the HGI value based on the proximate and ultimate analyses’ results:

(1) The data of parent coals as well as the blends before crushing is used (~212µm).
(2) Out of the 19 points available for analysis, only 15 are used for modeling, and 4 are
kept for cross-validation. This eliminates the possibility of getting an over-fitted
correlation, which cannot be used for other datasets.
(3) Linear, quadratic, inverse linear, and inverse quadratic values of the variables are
used to explore the correlation between various variables and the corresponding
HGI values.

Normalized data is used for determining the coefficients of the correlation so that relative
change in the magnitude of inputs and outputs is taken into consideration.
The best correlation (with cross-validation) was found between the fixed carbon (linear
and inverse quadratic), hydrogen, total sulfur, oxygen (linear and inverse quadratic),
nitrogen (inverse quadratic), and the product of ash content and oxygen. Regression
based on least-squares determined the coefficients for the following expression:

346:70 61853:84
HGI ¼ 1:85FC  35:94H  46:05S  4:91O þ 0:05AC  O  
FC2 N2
26726:08
þ þ 268:42 (14)
O2

which can be rearranged as:


 
5445:73
HGI ¼ 268:42  35:94H  46:05S  4:91 O  0:01AC 
O2
 
187:169 61853:84
þ 1:85 FC  2
 (15)
FC N2

The performance of the proposed correlation is shown in Fig. 9. Since the cross-
validated points are near the 45° line, the prediction of correlation is satisfactory with
a coefficient of determination R2 = 0.9416. According to the proposed correlation
Equation (15), the HGI values are expected to increase linearly with the decrease in
hydrogen and sulfur content. This means that coals with lower hydrogen and sulfur
content will be softer to grind. Similarly, the coals with high nitrogen content are expected
to be softer, and the correlation predicts that such coal will have relatively higher values of
HGI. Higher ash content is also expected to make the coal softer, as verified by the
equation. Although the relation between oxygen content and fixed carbon is not clear, the
correlation predicts that higher oxygen content and lower fixed carbon content will yield
relatively harder coals, despite its actual relevenace. For example, 5% FC content is
predicted to increase the HGI value by 15.06, whereas a 25% FC content is predicted to
increase the HGI value by 45.75. On the other hand, 15% oxygen content is predicted to
increase the HGI value by 64.22, whereas 25% oxygen content is predicted to decrease the
20 A. IDRIS ET AL.

Figure 9. Proposed model validation.

Table 6. Parameter ranges of coal blends (as determined).


Parameters (wt.%) Min Max Mean Std
FC 19.4 53.41 30.17 8.25
AC 5.52 51.7 31.78 12.85
H 2.47 4.62 3.26 0.56
N 0.26 1.42 0.73 0.31
S 0.26 1.125 0.89 0.20
O 7.6 19.4 12.00 3.30

HGI value by 48.17 if the ash content is fixed at 25%. The ranges of parameters for this
correlation are presented in Table 6.

HGI Fractions Analysis


Similar to the parent coals, the HGI fractions of the coal blends were characterized using
proximate, ultimate, and heating value analysis, mainly to identify the relative contribu-
tions of the blend quality parameters on the grindability of the blend. Further, these
fractions analyses results can be utilized to construct a characteristic curve for a particular
blend describing the changes in the parameter of interest as a function of the blending
ratio. The characteristic curves may be utilized by the plant operators to decide the
appropriate blend ratio for utilization (Ananda Babu, Lawrence, and Sivashanmugam
2018).
It is worth mentioning that the samples fractionated at HGI measurement were tested
with an equivalent particle size of −75µm to avoid the impact of particle size difference on
the analysis. The proximate and ultimate analyses results, heating values GCV, and HGI
Table 7. Proximate, ultimate analysis, heating value, grindability index of finer (−75µm) HGI fraction.
Blends Blend 1 Blend 2 Blend 3
Blend Ratio 95/05 80/20 70/30 50/50 30/70 95/05 80/20 70/30 50/50 30/70 95/05 80/20 70/30 50/50 30/70
Proximate Analysis, (wt.%)
MC 5.61 5.73 6.14 6.76 6.92 5.05 5.31 5.61 5.97 7.72 5.47 5.27 5.45 5.08 4.89
VM 20.04 22.46 24.25 29.59 34.41 20.34 23.23 25.99 30.74 38.97 22.54 23.44 25.52 25.97 25.95
FC 16.62 18.06 19.18 22.52 26.62 16.15 16.85 18.98 21.71 27.6 22.08 26.96 31.79 36.74 42.07
AC 57.73 53.75 50.43 41.13 32.05 58.46 54.61 49.42 41.58 25.71 49.91 44.33 37.24 32.21 27.09
Ultimate Analysis, (wt.%)
C 25.82 28.77 30.89 37.57 44.12 25.8 27.67 30.89 35.48 45.07 32.11 37.81 43.55 48.62 54.13
H 2.61 2.88 3 3.53 3.77 2.57 2.73 3.01 3.31 4.19 2.98 3.14 3.41 3.43 3.51
N 0.4 0.46 0.46 0.56 0.74 0.38 0.35 0.37 0.38 0.44 0.52 0.64 0.78 0.95 1.08
S 0.72 0.71 0.72 0.7 0.66 0.75 0.78 0.83 0.81 0.6 0.79 0.78 0.79 0.78 0.76
O 7.11 7.7 8.36 9.74 11.8 6.99 8.55 9.88 12.5 16.3 8.22 8.02 8.79 8.93 8.54
Heating Value, (kcal/kg)
GCV 2360 2910 3080 3780 4110 2600 2780 3080 3490 4430 3120 3710 4020 4470 4930
Mineral Matter (wt.%)
MM 62.74 58.44 54.86 44.81 34.97 63.55 59.41 53.83 45.35 28.1 54.34 48.31 40.65 35.21 29.67
Grindability Index
HGI 93.25 90.26 81.04 78.1 72.36 92.63 91.61 91.3 87.73 80.41 75.34 70.74 67.55 70.53 69.27
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION
21
22
A. IDRIS ET AL.

Table 8. Proximate, ultimate analysis, heating value, grindability index of coarser (+75µm) HGI fraction.
Blends Blend 1 Blend 2 Blend 3
Blend Ratio 95/05 80/20 70/30 50/50 30/70 95/05 80/20 70/30 50/50 30/70 95/05 80/20 70/30 50/50 30/70
Proximate Analysis (wt.%)
MC 6.03 5.73 6.52 7.23 7.2 5.13 5.33 5.66 6.13 7.62 5.71 5.44 5.39 5.08 4.76
VM 23.89 27.38 27.91 32.82 36.62 23.69 25.61 28.85 33.75 40.43 28.37 28.34 29.79 28.91 28.33
FC 21.3 24.11 24.56 29.05 35.37 20.05 20.78 23.36 26.46 32.23 26.53 31.46 35.58 40.67 45.2
AC 48.78 42.78 41.01 30.9 20.81 51.13 48.28 42.13 33.66 19.72 39.39 34.76 29.24 25.34 21.71
Ultimate Analysis (wt.%)
C 31.18 35.61 37.12 44.3 45.79 30.35 31.5 35.87 40.9 49.62 39.18 44.39 49.73 54.66 58.81
H 3.13 3.44 3.49 3.97 3.93 2.88 2.99 3.34 3.7 4.39 3.55 3.69 3.89 3.85 3.83
N 0.58 0.63 0.66 0.74 0.67 0.55 0.52 0.57 0.57 0.68 0.81 0.93 1.1 1.21 1.36
S 0.99 0.94 0.95 0.78 0.73 1.03 0.99 0.83 0.79 0.65 1.25 1.07 0.91 0.95 0.97
O 9.31 10.9 10.3 12.1 20.9 8.94 10.4 11.6 14.3 17.3 10.1 9.73 9.73 8.91 8.57
Heating Value, (kcal/kg)
GCV 3050 3580 3710 4400 4290 3030 3110 3540 4010 4820 3910 4410 4650 5030 5390
Mineral Matter, (wt.%)
MM 53.23 46.72 44.81 33.8 22.88 55.78 52.69 45.69 36.78 21.65 43.23 38.13 32.08 27.89 23.98
Grindability Index
HGI 94.59 91.78 82.39 79.97 73.41 93.59 92.79 92.73 89.33 82.42 77.37 72.71 68.93 72.6 71.05
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 23

values are presented in Table 7 and Table 8 for the finer (−75 μm), and coarser (+75)
fractions, respectively. The mineral matter contents presented for each fraction were
calculated using Equation (3). According to the results, the following comparative obser-
vations can be made.

● Although coal blends were tested at air-dry condition during the HGI measurement,
the moisture content was found higher at the coarser fraction for Blend 1 and Blend
2. However, the trend of moisture content for Blend 3 was observed overlapping.
● The quantities of the volatile matter VM were found to be in higher proportions at
the coarser fraction, and the corresponding elemental hydrogen, nitrogen, and oxy-
gen contents were also found to be supportive for the volatile matter contents in both
fractions.
● The mineral matter/ash contents are the softest solid constituents of the coal; these
incombustible inorganic components were found to be significantly in higher pro-
portions in the finer (−75 μm) fractions.
● Another important parameter is the proximate fixed carbon and the ultimate ele-
mental carbon contents. It can be seen that the weight percentages of these para-
meters are higher with the coarser (+75 μm) fraction than that of the finer (−75 μm)
fraction. Furthermore, the corresponding heating values GCVs are also in good
agreement with the carbon contents.
● The sulfur content is found among the harder coal constituent to grind. The higher
proportions of the sulfur content in coarser fraction indicate that harder mineral
forms of a sulfur compound such as pyrites and others exist in the coal matrix. For
instance, pyrite FeS is known to be harder than mild steel, which is believed to be one
of the coal mineral components responsible for the abrasive effects on mill parts
(Wells et al. 2005).

Based on these fractions analyses, characteristic curves as a function of the blend ratio
can be constructed, forming a system of the upper and lower limit curve, which is helpful
for the plant operators to make quick decisions on what blend ratio may be utilized.
Ananda Babu, Lawrence, and Sivashanmugam (2018) constructed the characteristic ash
contents as a function of the blending ratio of the blends which can potentially assist
operators in adjusting the blending ratio to get the target ash contents. However, for
a blend to be eligible for utilization, there are several factors need to be considered, the
characteristic ash contents curve may be utilized if the objective is to control the slugging
and fouling behaviors of the blend, considering the combustibility and efficiency as an
overall objective, then characteristic curves of fuel ratio and heating values become crucial,
but if the operational safety of the boiler and environment are the targets, then the
construction of characteristic curves corresponding to the mineral matter content and
nitrogen/sulfur is important to monitor the blending ratio. Other blend-specific charac-
teristic curves can also be constructed depending on the objectives and challenges
encountered. Fig.10 displays coal blends characteristic curves for parameters such as
mineral matter content, heating value, nitrogen content, and fuel ratio for each blend
studied. It can be observed that different blends exhibited different curves. The theoretical
average curve is plotted for comparison reasons. For every blend ratio or low-grade
24 A. IDRIS ET AL.

Figure 10. Coal blends characteristic curves in terms in mineral matter, heating value nitrogen and fuel
ratio.

contents in the blends, the corresponding parameter may vary within the coarser and finer
fraction curve. Accordingly, the adjustments can be made to achieve the target value.

Conclusion
Binary coal blends were prepared using low-grade coal and coals of different ranks. The
fuel quality parameters, HGI grindability values, and the abrasiveness index (AI) of the
coal blends were determined using standard methods. Results revealed that the HGI value
of the coal blend is an additive property if the ranks of the parent coal are identical or
similar; otherwise, it is a non-additive parameter. Whereas, the AI value is found to be an
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 25

additive property for all the coal blends tested. The effects of the blending ratio of low-
grade coal as a function of mineral matter, heating value, Bond’s work index, and
abrasiveness index were discussed. It is concluded that the blending of low-grade coal
with high-quality coal has an inverse effect on heating value and Bond’s work index of the
coal blends. However, a linear relation was observed for mineral matter and abrasiveness
index. To elucidate more on the effects of coal constituents on grindability values, the
fractions of HGI samples were analyzed. Observations on the analyses results indicate that
the ash contents/mineral matter contents were found to be softer components. The
elemental nitrogen, hydrogen, and oxygen were found to be supportive of this observation.
Besides, the data of HGI fractions analyses were used to construct a characteristic curves
for specific blends to assist power plant operators. Several relevant existing models such as
Sengupta SGI, Chelgani, and Peisheng models, were found to be unable to predict the HGI
value of coal blends prepared in this study. Therefore, a new cross-validated model has
been proposed for the prediction of HGI value, which shows a good agreement between
the experimental and calculated HGI values.

Acknowledgments
Authors gratefully acknowledge the Department of Chemical Engineering, Universiti Teknologi
PETRONAS for the facility and support given for this research work.

ORCID
Alamin Idris http://orcid.org/0000-0001-6697-9248

References
Ananda Babu, K., A. Lawrence, and P. Sivashanmugam. 2018. Grindability studies on blended coals
of high-ash Indian coals with low-ash imported coals. International Journal of Coal Preparation
and Utilization 38:433–42. doi:10.1080/19392699.2017.1281254.
ASTM International. 2000. Standard classification of coals by rank, in: ASTM D 388 – 99, ASTM
international. United States, 1–7.
ASTM International. 2002a. Standard test method for grindability of coal by the hardgrove-machine
method, in: ASTM D 409-02, ASTM international. United States, 1–9.
ASTM International. 2002b. Standard test methods for proximate analysis of the analysis sample of
coal and coke by instrumental procedures, in: D 5142 – 02a, ASTM international. United States,
1–5.
ASTM International. 2004. Standard test methods for sulfur in the analysis sample of coal and coke
using hightemperature tube furnace combustion methods, in: ASTM D 4239 – 04, ASTM
international, United States, 1–8.
ASTM International. 2009. Standard practice for ultimate analysis of coal and coke, in: D 3176 – 09.
United States, 1–3.
ASTM International. 2011. Standard test method for gross calorific value of coal and coke, in:
ASTM D 5865-11a ASTM international. United States, 1–12.
Atesok, G., M. Ozer, F. Boylu, and H. Dıncer. 2005. The effect of anionic dispersants on grindability
of different rank coals. International Journal of Mineral Processing 77:199–207. doi:10.1016/j.
minpro.2005.06.004.
26 A. IDRIS ET AL.

Bagherieh, A. H., J. C. Hower, A. R. Bagherieh, and E. Jorjani. 2008. Studies of the relationship
between petrography and grindability for Kentucky coals using artificial neural network.
International Journal of Coal Geology 73:130–38. doi:10.1016/j.coal.2007.04.002.
Bond, F. C. 1961. Crushing and grinding calculations. British Chemical Engineering 6:378–91.
British Standard. 1980. Methods for analysis and testing of coal and coke part 19: Determination
of the index of abrasion of coal in: BSI - BS 1016–19 British standard institution. United
Kingdom.
Chelgani, S. C., J. C. Hower, E. Jorjani, S. Mesroghli, and A. H. Bagherieh. 2008. Prediction of coal
grindability based on petrography, proximate and ultimate analysis using multiple regression and
artificial neural network models. Fuel Processing Technology 89:13–20. doi:10.1016/j.
fuproc.2007.06.004.
Deniz, V., and Y. Umucu. 2013. Interrelationships between the bond grindability with physicome-
chanical and chemical properties of coals. Energy Sources, Part A: Recovery, Utilization, and
Environmental Effects 35:144–51. doi:10.1080/15567036.2010.504942.
Dobó, Z., and A. Fry. 2018. Investigation of co-milling Utah bituminous coal with prepared woody
biomass materials in a Raymond bowl mill. Fuel 222:343–49. doi:10.1016/j.fuel.2018.02.181.
Doroodchi, E., H. Zulfiqar, and B. Moghtaderi. 2013. A combined experimental and theoretical
study on laboratory-scale comminution of coal and biomass blends. Powder Technology
235:412–21. doi:10.1016/j.powtec.2012.10.054.
Hansen, A. E., and J. C. Hower. 2014. Notes on the relationship between microlithotype composi-
tion and Hardgrove grindability index for rank suites of Eastern Kentucky (Central Appalachian)
coals. International Journal of Coal Geology 131:109–12. doi:10.1016/j.coal.2014.06.010.
Hower, J. C. 1988. Additivity of hardgrove grindability: A case study. Journal of Coal Quality
7:68–70.
Hower, J. C. 1998. Interrelationship of coal grinding properties and coal petrology. Mining,
Metallurgy & Exploration 15:1–16. doi:10.1007/BF03403218.
Kanwal, S., H. Sana, and S. Shafqat Nawaz. 2017. Munir, influence of relative humidity and
temperature on grindability of Pakistani and Afghani Coals. The Journal of Energy and
Environmental Science. Photon 132:663–68.
Khoshjavan, R. K. S., and B. Rezai. 2013. Evaluation of the effect of coal chemical properties on the
Hardgrove Grindability Index (HGI) of coal using artificial neural networks. The Journal of the
Southern African Institute of Mining and Metallurgy 113:505–10.
Kingman, S. W. 2006. Recent developments in microwave processing of minerals. International
Materials Reviews 51:1–12. doi:10.1179/174328006X79472.
Lester, E., S. Kingman, and C. Dodds. 2005. Increased coal grindability as a result of microwave
pretreatment at economic energy inputs. Fuel 84:423–27. doi:10.1016/j.fuel.2004.09.019.
Matin, S. S., J. C. Hower, L. Farahzadi, and S. C. Chelgani. 2016. Explaining relationships among
various coal analyses with coal grindability index by Random Forest. International Journal of
Mineral Processing 155:140–46. doi:10.1016/j.minpro.2016.08.015.
Mohanty, A., S. Chakladar, S. Mallick, and S. Chakravarty. 2019. Structural characterization of
coking component of an Indian coking coal. Fuel 249:411–17. doi:10.1016/j.fuel.2019.03.108.
Nahvi, S. M., P. H. Shipway, and D. G. McCartney. 2009. Effects of particle crushing in abrasion
testing of steels with ash from biomass-fired powerplants. Wear 267:34–42. doi:10.1016/j.
wear.2009.01.054.
Özbayoğlu, G., A. M. Özbayoğlu, and M. E. Özbayoğlu. 2008. Estimation of Hardgrove grindability
index of Turkish coals by neural networks. International Journal of Mineral Processing 85:93–100.
doi:10.1016/j.minpro.2007.08.003.
Pankaj, P. K., S. K. Kushwaha, K. K. Manjhi, T. K. Das, and A. Gupta. 2017. Effect of coal-blend
crushing on coke quality. International Journal of Coal Preparation and Utilization 37:124–30.
doi:10.1080/19392699.2016.1140651.
Peisheng, L., X. Youhui, Y. Dunxi, and S. Xuexin. 2005. Prediction of grindability with multivariable
regression and neural network in Chinese coal. Fuel 84:2384–88. doi:10.1016/j.fuel.2005.04.016.
Radic, D., M. Obradovic, M. Stanojevic, A. Jovovic, and D. Stojiljkovic. 2011. A study on the
grindability of Serbian coals. Thermal Science 15:267–74. doi:10.2298/TSCI1101267R.
INTERNATIONAL JOURNAL OF COAL PREPARATION AND UTILIZATION 27

Rubiera, A. A. F., E. Fuente, N. Miles, and J. J. Pis. 1999. Effect of the grinding behaviour of coal
blends on coal utilisation for combustion. Powder Technology 105:351–56. doi:10.1016/S0032-
5910(99)00158-8.
Sahoo, B. K., S. De, and B. C. Meikap. 2011. Improvement of grinding characteristics of Indian coal
by microwave pre-treatment. Fuel Processing Technology 92 (10):1920–1928. doi:10.1016/j.
fuproc.2011.05.012.
Sengupta, A. N. 2002. An assessment of grindability index of coal. Fuel Processing Technology
76:1–10. doi:10.1016/S0378-3820(01)00236-3.
Shahzad, K., S. Kanwal, S. Nawaz, N. Sheikh, and S. Munir. 2014. Effects of moisture and coal
blending on the Hardgrove grindability index of Pakistani Coals. International Journal of Coal
Preparation and Utilization 34:1–9. doi:10.1080/19392699.2013.776961.
Shen, Y., M. Wang, Y. Hu, J. Kong, J. Wang, L. Chang, and W. Bao. 2019. Transformation and
regulation of sulfur during pyrolysis of coal blend with high organic-sulfur fat coal. Fuel
249:427–33. doi:10.1016/j.fuel.2019.03.066.
Steer, J. M., R. Marsh, D. Morgan, and M. Greenslade. 2015. The effects of particle grinding on the
burnout and surface chemistry of coals in a drop tube furnace. Fuel 160:413–23. doi:10.1016/j.
fuel.2015.07.094.
Tichánek, F. 2008. Contribution to determination of coal grindability using hardgrove method.
GeoScience Engineering, LIV. 27–32.
Trimble, A. S., and J. C. Hower. 2000. Maceral/microlithotype analysis of the progressive grinding
of a Central Appalachian high-volatile bituminous coal blend. Mining, Metallurgy & Exploration
17:234–43. doi:10.1007/BF03403240.
Trimble, A. S., and J. C. Hower. 2003. Studies of the relationship between coal petrology and
grinding properties. International Journal of Coal Geology 54:253–60. doi:10.1016/S0166-5162(03)
00039-9.
Tymoszuk, M., K. Mroczek, S. Kalisz, and H. Kubiczek. 2019. An investigation of biomass
grindability. Energy. doi:10.1016/j.energy.2019.05.167.
Ural, S., and M. Akyıldız. 2004. Studies of the relationship between mineral matter and grinding
properties for low-rank coals. International Journal of Coal Geology 60:81–84. doi:10.1016/j.
coal.2004.07.001.
Venkoba Rao, B., and S. J. Gopalakrishna. 2009. Hardgrove grindability index prediction using
support vector regression. International Journal of Mineral Processing 91:55–59. doi:10.1016/j.
minpro.2008.12.003.
Vuthaluru, H. B., R. J. Brooke, D. K. Zhang, and H. M. Yan. 2003. Effects of moisture and coal
blending on Hardgrove grindability index of Western Australian coal. Fuel Processing Technology
81:67–76. doi:10.1016/S0378-3820(03)00044-4.
Wang, G., X. Bai, C. Wu, W. Li, K. Liu, and A. Kiani. 2018. Recent advances in the beneficiation of
ultrafine coal particles. Fuel Processing Technology 178:104–25. doi:10.1016/j.fuproc.2018.04.035.
Wells, J. J., F. Wigley, D. J. Foster, W. H. Gibb, and J. Williamson. 2004. The relationship between
excluded mineral matter and the abrasion index of a coal. Fuel 83:359–64. doi:10.1016/S0016-
2361(03)00262-X.
Wells, J. J., F. Wigley, D. J. Foster, W. R. Livingston, W. H. Gibb, and J. Williamson. 2005. The
nature of mineral matter in a coal and the effects on erosive and abrasive behaviour. Fuel
Processing Technology 86:535–50. doi:10.1016/j.fuproc.2004.04.002.
Williams, O., C. Eastwick, S. Kingman, D. Giddings, S. Lormor, and E. Lester. 2015. Investigation
into the applicability of Bond Work Index (BWI) and Hardgrove Grindability Index (HGI) tests
for several biomasses compared to Colombian La Loma coal. Fuel 158:379–87. doi:10.1016/j.
fuel.2015.05.027.
Xie, W., Y. He, C. Luo, X. Zhang, H. Li, H. Wang, and F. Shi. 2015. Energy-size reduction of coals
in the Hardgrove machine. International Journal of Coal Preparation and Utilization 35:51–62.
doi:10.1080/19392699.2014.960079.
Xie, W., Y. He, L. Qu, X. Sun, and X. Zhu. 2018. Effect of particle properties on the energy-size
reduction of coal in the ball-and-race mill. Powder Technology 333:404–09. doi:10.1016/j.
powtec.2018.04.018.

View publication stats

You might also like