Available online at www.sciencedirect.

com

Fuel 87 (2008) 125–130

www.fuelfirst.com

Effect of mineral matter on coal self-heating rate

B. Basil Beamish a, Ahmet Arisoy b,*

a
School of Engineering, The University of Queensland, Brisbane Queensland 4072, Australia
b
Faculty of Mechanical Engineering, Istanbul Technical University, Gümüsßsuyu, 34437 Istanbul, Turkey

Received 27 April 2006; received in revised form 27 March 2007; accepted 28 March 2007
Available online 21 May 2007

Abstract

Adiabatic self-heating tests have been conducted on subbituminous coal cores from the same seam profile, which cover a mineral mat-
ter content range of 11.2–71.1%. In all cases the heat release rate does not conform to an Arrhenius kinetic model, but can best be
described by a third order polynomial. Assessment of the theoretical heat sink effect of the mineral matter in each of the tests reveals
that the coal is less reactive than predicted using a simple energy conservation equation. There is an additional effect of the mineral mat-
ter in these cases that cannot be explained by heat sink alone. The disseminated mineral matter in the coal is therefore inhibiting the
oxidation reaction due to physicochemical effects.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Coal; Spontaneous combustion; Mineral matter; Non-Arrhenius kinetics

1. Introduction thermal runaway at temperatures up to 160 °C and the

Coal self-heating, leading to spontaneous ignition, con-
tinues to be a hazard that must be managed by coal mines
to create a safe work environment [1]. Spontaneous com-
bustion management plans (SCMPs) should contain data
that can be used to assess the inherent propensity of the
coal to self-heat. This usually involves obtaining results
from small-scale laboratory tests. One such test that is fre-
quently used by the Australian coal industry is the adia-
batic oven R70 self-heating rate test, which can also be
used to study and quantify the factors that affect coal
inherent self-heating propensity [2–6]. The test measures
the average self-heating rate between 40 °C and 70 °C,
known as the R70 self-heating rate index [7]. The higher
the R70 value, the higher the propensity of coal for self-
heating. Tests of this genre have been criticized in the past
on the grounds that they disregard different thermal diffu-
sivities of the coals tested. These tests also provide a full
temperature history of the self-heating process through to
*
Corresponding author. Tel.: +90 212 252 9587; fax: +90 212 245 0795.
E-mail address: arisoyah@itu.edu.tr (A. Arisoy).
data can be used to define the reactivity of coal.
One important factor that affects coal reactivity and
hence self-heating propensity is the mineral matter content
of the coal. However, little is known about mineral matter
effects on coal self-heating, other than brief investigations
by Humphreys et al. [7], Smith et al. [8] and Beamish and
Blazak [4]. Smith et al. [8] looked at the heat sink effect
of adding different minerals and inorganic constituents to
coal as a possible solution to inhibiting coal self-heating.
Humphreys et al. [7] added differing amounts of coal ash
to the same coal sample to test a mineral matter free correc-
tion equation for the R70 self-heating rate. Beamish and
Blazak [4] and Beamish et al. [9] showed a strong negative
correlation exists between ash content and R70 self-heating
rate of both low and high rank coals. Such a relationship is
not accounted for in most models of coal spontaneous
combustion.
This paper presents a more detailed kinetic analysis of
the self-heating rate data obtained by Beamish and Blazak
[4] from subbituminous coal cores. The effects of mineral
matter in the coal are clearly defined by means of mathe-
matical modelling of the heat sink effect over the entire

0016-2361/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2007.03.049
126 B.B. Beamish, A. Arisoy / Fuel 87 (2008) 125–130

temperature range of the adiabatic oven test from 40 to The gas selection switch was turned to oxygen with a con-
160 °C. stant flow rate of 50 mL min1. The temperature change of
the coal with time was recorded by a data logging system
2. Experimental for later analysis. It was assumed sensible heat removal
by the gas is insufficient to affect the combustion phenom-
2.1. Coal samples enology. The oven limit switch was set at 160 °C to cut off
the power to the oven heating element and stop the oxygen
The coal samples used in this study were obtained from a flowing when the sample reached this temperature.
geotechnical drilling program in the Callide Basin, Queens-
land. The cores selected for testing came from the one bore- 2.3. Evaluation of adiabatic oven tests
hole and were thus of the same rank suite of samples. To
maintain sample integrity, all cores were firstly wrapped in In an adiabatic oven, it is assumed that an ideal adia-
plastic cling wrap then aluminium foil and finally an outer batic condition for the sample is created. It means that
layer of masking tape, before being chilled on ice at the drill all the heat produced by coal oxidation is used for heating
site. These samples were then transported to the laboratory of the coal sample. A simplified energy conservation equa-
in an insulated container full of ice and transferred immedi- tion for the coal sample in the adiabatic oven can be writ-
ately to a freezer for storage until adiabatic testing was per- ten in this case. This ignores finite convective and diffusive
formed. Four samples from the original suite tested by heat transfer rates and heat given to the incoming gas flow
Beamish and Blazak [4] are considered in this study. These and also ignores reactant depletion. Dry coal is considered
are TG1, TG2, TG5 and TG6. Coal quality data for the to consist of two constituents, namely mineral matter free
samples are given in Table 1. The coal is classified as subbi- coal (mmfc) and mineral matter (mm). Mineral matter free
tuminous A in rank, with the low volatile matter contents coal reacts with oxygen and releases heat of reaction,
being a reflection of the high inertinite content of the Callide meanwhile mineral matter does not react (unless it is finely
coal [10]. X-ray powder diffraction analysis of the samples disseminated pyrite [8]) and has no contribution to the oxi-
identified kaolinite as the predominant mineral present in dation reaction between carbon and oxygen. The energy
the coal, with subordinate amounts of quartz and siderite. conservation equation for coal in the reaction flask can
be written as follows:
2.2. Adiabatic oven tests
ðmmm C p;mm þ mmmfc  C p;mmfc ÞdT =dt ¼ V mmfc  q0ox;mmfc ð1Þ
Full details of the adiabatic oven used for the self-heat- where q0ox;mmfc ¼ heat release rate of reaction in oxygen
ing experiments are given in Beamish et al. [2]. Each sample atmosphere during adiabatic oven testing [W m3],
was crushed to <212 mm just before testing in the adiabatic Vmmfc = volume of mineral matter free coal in the sample
oven to minimise the effects of oxidation on fresh surfaces [m3], mmm and mmmfc = mass of mineral matter and min-
created by the grinding of the coal. A 150 g sample of the eral matter free coal respectively [kg], Cp,mm and Cp,mmfc =
crushed coal was dried under nitrogen at 110 °C for specific heat of mineral matter and mineral matter free coal
16 h to ensure complete drying of the sample. Once dry- respectively [J kg1 K1].
ing was complete, the coal was allowed to cool down to
40 °C before being placed into the reaction vessel and mmm ¼ mcoal  Rmm
transferred to the adiabatic oven where it was allowed to mmmfc ¼ mcoal  ð1  Rmm Þ
equilibrate at a temperature of 40 °C under nitrogen. When V mmfc ¼ mmmfc =qmmfc ¼ mcoal  ð1  Rmm Þ=qmmfc
the sample temperature had stabilised, the oven was
switched to remote monitoring mode. This enabled the where Rmm is the ratio of mineral matter in dry coal. Insert-
oven to track the coal temperature rise due to oxidation. ing these equations into Eq. (1) results in the heat release
rate of coal being defined as follows:
Table 1
Coal quality and parameters used for adiabatic testing
q0ox; mmfc ¼ ½qmmfc  ðRmm  C p;mm þ ð1  Rmm Þ
Sample TG5 TG6 TG1 TG2  C p;mmfc Þ=ð1  Rmm ÞdT =dt ð2aÞ
Total test mass (g) 150.3 150.1 150.2 150.2
or
Moisture removed (g) 21.4 17.9 18.4 13.5
Dried reactant mass (g) 128.9 132.2 131.8 136.7 q0ox;mmfc ¼ qmmfc  C p  dT =dt ð2bÞ
Mineral matter mass (g) 14.4 41.6 59.0 97.2
Mineral matter free coal mass (g) 114.5 90.6 72.8 39.5 where Cp = (RmmÆCp,mm+ (1Rmm) Æ Cp,mmfc)/(1Rmm).
Moisture (%, ar) 14.2 11.9 12.3 9.0
Mineral matter (%, db) 11.2 31.5 44.8 71.1
Ash (%, db) 9.8 28.4 39.1 62.2 2.4. Non-Arrhenius rate data from adiabatic oven tests
MM/ash 1.14 1.11 1.15 1.14
Volatile matter (%, dmmf) 27.2 27.7 27.8 28.0 Data files from the adiabatic oven tests contain temper-
Fixed carbon (%, dmmf) 72.8 72.3 72.2 72.0
ature values of the coal sample from 40 °C up to approxi-
 B.B. Beamish, A. Arisoy / Fuel 87 (2008) 125–130 127

-1
y = 13626260439.781x3 - 104390885.900x2 + 255969.558x - 201.764
-2

ln (dT/dt) -3

-4

-5

-6

-7
0.0026 0.0027 0.0028 0.0029 0.0030 0.0031 0.0032
1/T

Fig. 1. Data points for coal sample TG5 and fitted third order polynomial.

mately 160 °C. Coal temperatures are recorded every min- 2

ute. From this recorded time–temperature data, tempera- 0
ture increase rate versus temperature can be derived. A

ln (dT/dt)
plot of ln(dT/dt) versus 1/T plots for coal sample TG5 is -2
shown in Fig. 1. Rate data can be derived from this plot -4
by considering Eq. (1). This type of plot is also the basis
of the ‘heat release method’ by Jones et al. [11]. Generally, -6

it has been assumed that the coal oxidation reaction rate -8

follows Arrhenius kinetics. However, recent work by Jones 0.0022 0.0024 0.0026 0.0028 0.0030 0.0032
and Newman [12] has questioned this assumption. It can be 1/T
seen in Fig. 1 that the straight line relationship predicted by TG5 (11.2%) TG6 (31.5%) TG1 (44.8%) TG2 (71.1%)
the Arrhenius equation does not appear to be valid for the
entire temperature range of the coal self-heating. Although Fig. 2. Reactivity curves of different mineral matter containing coals. The
upper curve indicates a more reactive coal than the lower one.
the results in Jones and Newman [12] were obtained in a
microcalorimeter where conditions are highly non-adia-
batic; Jones and Newman [12] suggested the rate data is 3. Calculation of heat sink effect on self-heating rate of coal
more appropriately described by a third order polynomial.
Accordingly, a third order polynomial has been fitted to Under the assumption of mineral matter in coal has only
the data in Fig. 1. The equation obtained is in the following a heat sink effect on self-heating rate, the self-heating rate
form: of mineral matter free coal can be calculated from the data
lnðdT =dtÞ ¼ a  ð1=T Þ3 þ b  ð1=T Þ2 þ c  ð1=T Þ þ d ð3aÞ of a mineral matter containing coal. If Eq. (3c) is inserted
into Eq. (2b), the heat release rate expression for mineral
or matter free coal is obtained as follows:
3 2
dT =dt ¼ exp½a  ð1=T Þ þ b  ð1=T Þ þ c  ð1=T Þ þ d ð3bÞ q0ox;mmfc ¼ qmmfc  C p  exp½f ðT Þ ð4Þ

or If we carry out an adiabatic oven test for a known coal and

dT =dt ¼ exp½f ðT Þ
where a, b, c and d are constants of the polynomial which
are obtained from the fitted function. Values of these con-
stants for the TG5 coal sample are a = 13626260439.781,
b = 104390885.900, c = 255969.558 and d = 201.764.
Using the above defined procedure for plotting dT/dt ver-
sus 1/T, third order polynomials are fitted for each coal of
the data set. The derived third order polynomials for each
coal are shown in Fig. 2. As can be seen from Fig. 2, reac-
tivity of the coal increases with decreasing mineral matter
content. This relationship is also consistent with the mea-
sured R70 values of the samples.
ð3cÞ derive the constants for the heat release polynomial as well
as derive the mineral matter content and specific heat val-
ues, we can express heat release rate of mineral matter free
coal. This is true under the assumption that a heat sink ef-
fect from mineral matter is the only parameter acting on
heat release rate of dry coal. Similarly, without doing any
further experiments we can also calculate a polynomial
function for another coal of different mineral matter con-
tent using the expression for mineral matter free coal heat
release rate. If we define the known polynomial function as
f1(T) and the unknown polynomial function as f2(T), an
expression can be written relating these two functions as
follows:
128 B.B. Beamish, A. Arisoy / Fuel 87 (2008) 125–130

exp½f2 ðT Þ ¼ ðC p;1 =C p;2 Þ  exp½f1 ðT Þ Table 4

Heat capacities of mineral matter and mineral matter free coal at 55 °C
From this expression f2(T) can be written as follows:
Sample TG5 TG6 TG1 TG2
f2 ðT Þ ¼ lnðC p;1 =C p;2 Þ þ f1 ðT Þ ð5Þ Cp,mm (J kg K ) 1 1
924 914 982 971
Cp,mmfc (J kg1 K1) 1128 1135 1136 1139

4. Results and discussion

Data for each of the coal samples are contained in Table 2

1 and the measured mineral matter compositions are given
0
in Table 2.

The mineral matter value in Table 1 has been calculated
using the equation:
mm ¼ A=ð0:86  K þ Q þ 0:62  SÞ
where A = ash content, K = kaolinite fraction, Q = quartz
fraction, and S = siderite fraction. This equation allows for
the loss of hydration from the kaolinite and the carbonate
loss from the siderite during the ash determination.
Heat capacity of the coal sample mineral matter Cp,mm
can be calculated in terms of its constituents as follows:
C p;mm ¼ %Kaolinite  C p;Kaolinite þ %Quartz  C p;Quartz
þ %Siderite  C p;Siderite
where, the heat capacities of the mineral constituents are
calculated from the temperature dependent equations in
Table 3.
Heat capacity of the mineral matter free coal is calcu-
lated by the following empirical formula [14]:
C p;mmfc ¼ f%FC  ½0:165 þ ð3:4 T  104 Þ  ð1:4 T 2  107 Þ
þ %VM  0:8  ð0:395 þ 8:1 T  104 Þ þ %VM  0:2
 ð0:71 þ 6:1 T  104 Þg  4186:8 ðJ kg1 K1 Þ
An example of the calculated Cp,mm and Cp,mmfc values of
the four coals at 55 °C is given in Table 4.
The same curves shown in Fig. 2 have been calculated by
using Eq. (5) with sample TG5 (mineral matter content
11.2%) as the reference coal and the results are given in
Fig. 3. The reactivity curve for mineral matter free coal
Table 2
Mineral matter compositions of Callide samples
Sample TG5 TG6 TG1
Kaolinite (%) 66.4 60.8 91
Quartz (%) 24.9 35.8 9 13.9
Siderite (%) 8.7 3.4 0 0
ln (dT/dt)
-2
-4
-6
-8
0.0022 0.0024 0.0026 0.0028 0.0030 0.0032
1/T
MMFC (0.0%) TG5 (11.2%) TG6 (31.5%)
TG1 (44.8%) TG2 (71.1%)
Fig. 3. Calculated polynomials for mineral matter free coal, TG1, TG2
and TG6 based on TG5 test results.
has also been added to Fig. 3. At low temperatures, the cal-
culated curves in Fig. 3 are similar to the experimental
curves in Fig. 2. However, as temperature increases there
is substantial deviation between the calculated and experi-
mental curves, with the calculated curves indicating the
coal should be more reactive than it actually is. Therefore,
assuming a simple heat sink effect of the mineral matter in
the coal does not fully explain the observed self-heating
behaviour.
This point can be examined further by a slightly different
evaluation of the test results of Humphreys et al. [7] and
Beamish and Blazak [4]. In the first group of tests mineral
matter (coal ash) was added to the coal externally. In this
case the coal and mineral matter were mixed with each
other and only the heat sink effect of the mineral matter
was expected to play a role. However, in the tests of Beam-
ish and Blazak [4] the mineral matter was already present in
the coal as a natural constituent that was finely dissemi-
nated throughout the coal.
Humphreys et al. [7] calculated R70 values of mineral
matter free coal basing on measured R70 values of mineral
TG2 matter containing coals. Using the correction factors given
86.1 in the paper, R70 values can be predicted for each of the
mineral matter cases. Both measured and predicted R70 val-
ues are listed in Table 5 and the results are graphically illus-
trated in Fig. 4. Measured and calculated points are very

Table 3 Table 5
Heat capacities of mineral matter constituents as a function of temper- Measured R70 values with correction factors given by Humphreys et al. [7]
ature (T) in degrees Celsius (based on values from Knacke et al. [13]) and calculated R70 values considering only the heat sink effect
Mineral Formula Heat capacity equation (J kg1 K1) Mineral matter content (%) 0 9.9 29.9 49.9
Siderite FeCO3 0.97T + 683.81 Measured R70 values (°C/h) 1.375 1.098 0.767
Quartz SiO2 1.41T + 704.93 Correction factor for R70 1.00 1.085 1.328 1.766
Kaolinite Al2O3 Æ 2SiO2 Æ 2H2O 1.71T + 907.83 Calculated R70 values (°C/h) 1.492 1.375 1.123 0.845
 B.B. Beamish, A. Arisoy / Fuel 87 (2008) 125–130 129

R70 self-heating rate ( oC/h, db) 1.6 5. Conclusion

1.4
1.2 Oxidation of a coal particle in atmospheric conditions is
1.0 a complex process. There are many competing chemical
0.8 and physical processes occurring simultaneously at low
0.6 temperatures. One of the important parameters affecting
0.4 Calculated the inherent self-heating propensity of coal is the mineral
0.2 Measured matter content. A definite negative correlation exists
0.0 between mineral matter content and the R70 self-heating
0 10 20 30 40 50 60 rate of coal. Similarly, the heat release rate (reactivity) of
Mineral matter (%, db) dry coal under adiabatic test conditions from 40 to160 °C
Fig. 4. Relationship between average self-heating rate and mineral matter also displays a negative correlation with mineral matter
content for the results obtained by Humphreys et al. [7]. content.
This relationship is partly due to the mineral matter in
the coal acting as a heat sink. If coal and ash are simply
admixed with each other, then only the heat sink effect of
Table 6 the mineral matter is observed. However, if the mineral
Measured R70 values for Callide samples and calculated R70 values matter is naturally disseminated in the coal then the heat
considering only the heat sink effect
sink effect does not account for the total observed decrease
Mineral matter content (%) 0 11.2 31.5 44.8 71.1 in the coal self-heating rate. In this case the mineral matter
Measured R70 values (°C/h) 16.22 9.31 5.98 1.69 inhibits oxygen reaching carbon surfaces and probably acts
Calculated R70 values (°C/h) 17.95 16.22 13.02 10.48 5.74 on the chemical reaction mechanism.

Acknowledgements
20
R70 self-heating rate (oC/h, db)

18
16
14
12
10
8
6 nical University enabled Dr Beamish to spend time at ITU
Measured
4 to write this paper.
Calculated
2
0
0 20 40
Mineral matter (%, db)
Fig. 5. Relationship between average self-heating rate and mineral matter
content using test results obtained by Beamish and Blazak [4].
close to each other, indicating that only the heat sink effect
of the mineral matter is apparent for this situation.
A similar comparison can be made between measured
and calculated R70 values for the Callide samples. The
R70 value can be calculated from each of the reactivity
curves in Fig. 3 by taking the exponential of the average
ln(dT/dt) from 40 to 70 °C and multiplying by 3600 to con-
vert to °C/h. These values are listed in Table 6 and graph-
ically illustrated in Fig. 5. There are considerable
differences between measured and calculated points in this
case. The calculated R70 values consider only the heat sink
effect. In this case predictions are not as successful as in the
case of admixing ash (mineral matter) and coal. Therefore,
when the mineral matter is in the structure of coal (dissem-
inated), it has additional effects on the self-heating process.
These additional effects could be on chemical processes and
on transport properties.
Anglo Coal Callide Mine and the Australian Coal Asso-
ciation Research Program are thanked for their continued
support of spontaneous combustion testing research at The
University of Queensland. The University of Queensland
Travel Grant and financial assistance from Istanbul Tech-
60 80 100
References
[1] Moreby R. Management of seam gas emission and spontaneous
combustion in a highly gassy, thick and multi seam coal mine – a
learning experience. In: Proceedings of the eighth international mine
ventilation congress, 2005. p. 221–30.
[2] Beamish BB, Barakat MA, St. George JD. Adiabatic testing
procedures for determining the self-heating propensity of coal and
sample ageing effects. Thermochim Acta 2000;362:79–87.
[3] Beamish BB, Barakat MA, St. George JD. Spontaneous-combustion
propensity of New Zealand coals under adiabatic conditions. Int J
Coal Geol 2001;45:217–24.
[4] Beamish BB, Blazak DG. Relationship between ash content and R70
self-heating rate of Callide coal. Int J Coal Geol 2005;64:126–32.
[5] Beamish BB, Hamilton GR. Effect of moisture content on the R70
self-heating rate of Callide coal. Int J Coal Geol 2005;64:133–8.
[6] Beamish BB. Comparison of the R70 self-heating rate of New Zealand
and Australian coals to suggate rank parameter. Int J Coal Geol
2005;64:139–44.
[7] Humphreys D, Rowlands D, Cudmore JF. Spontaneous combustion
of some Queensland coals. In: Proceedings of the ignitions, explosions
and fires in coal mines symposium; 1981. p. 5-1–19.
[8] Smith AC, Mıron Y, Lazzara CP. Inhibition of spontaneous
combustion of coal. US Bureau of Mines Report of Investigations
RI 9196, 1988.
[9] Beamish BB, Blazak DG, Hogarth LCS, Jabouri I. R70 relationships
and their interpretation at a mine site. In: Proceedings of the
COAL2005 conference, 2005. p. 183–5.
130 B.B. Beamish, A. Arisoy / Fuel 87 (2008) 125–130

[10] Biggs MS, Burgess AW, Patrıck RB. Callide Basin. Geology of [13] Knacke O, Kubaschewski O, Hesselmann K. Thermochemical
Australian coal basins, vol. 1. Geological Society of Australia; 1995. properties of inorganic substances. 2nd ed. Berlin: Springer-Verlag;
p. 471–88. Special Publication. 1991.
[11] Jones JC, Henderson KP, Littlefair J, Rennie S. Kinetic parameters of [14] Jones JC. Environmental and safety aspects of combustion technol-
oxidation of coals by heat-release measurement and their relevance of ogy. Caithness: Whittles Publishing; 1997.
self-heating tests. Fuel 1998;77:19–22.
[12] Jones JC, Newman SC. Non-Arrhenius behaviour in the oxidation of
two carbonaceous substrates. J Loss Prevent Process Ind 2003;16:
223–5.