Spatial context in the calculation of gas emissions for underground coal mines

Abstract

The prediction of gas emissions arising from underground coal mining has been the subject of extensive research for several decades, however calculation techniques remain empirically based and are hence limited to the origin of calculation in both application and resolution. Quantification and management of risk associated with sudden gas release during mining (outbursts) and accumulation of noxious or combustible gases within the mining environment is reliant on such predictions, and unexplained variation correctly requires conservative management practices in response to risk. Over 2500 gas core samples from two southern Sydney basin mines producing metallurgical coal from the Bulli seam have been analysed in various geospatial context including relationships to hydrological features and geological structures. The results suggest variability and limitations associated with the present traditional approaches to gas emission prediction and design of gas management practices may be addressed using predictions derived from improved spatial datasets, and analysis techniques incorporating fundamental physical and energy related principles.

Full text

Spatial context in the calculation of gas emissions for underground coal
mines
Patrick Booth
⇑
, Heidi Brown, Jan Nemcik, Ren Ting
School of Civil, Mining and Environmental Engineering, School of Earth and Environmental Sciences, University of Wollongong, Wollongong 2500, Australia
article info
Article history:
Received 15 January 2017
Received in revised form 5 March 2017
Accepted 12 April 2017
Available online xxxx
Keywords:
Gas emission prediction
Spatial analysis
Underground coal mining
Risk management
Greenhouse gas
Climate
abstract
The prediction of gas emissions arising from underground coal mining has been the subject of extensive
research for several decades, however calculation techniques remain empirically based and are hence
limited to the origin of calculation in both application and resolution. Quantification and management
of risk associated with sudden gas release during mining (outbursts) and accumulation of noxious or
combustible gases within the mining environment is reliant on such predictions, and unexplained vari-
ation correctly requires conservative management practices in response to risk. Over 2500 gas core sam-
ples from two southern Sydney basin mines producing metallurgical coal from the Bulli seam have been
analysed in various geospatial context including relationships to hydrological features and geological
structures. The results suggest variability and limitations associated with the present traditional
approaches to gas emission prediction and design of gas management practices may be addressed using
predictions derived from improved spatial datasets, and analysis techniques incorporating fundamental
physical and energy related principles.
Ó2017 Published by Elsevier B.V. on behalf of China University of Mining & Technology. This is an open
access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Underground mining methods account for approximately 20%
of total black coal and a proportionally higher amount of metallur-
gical coal production in Australia [1]. In NSW, hard metallurgical
coal is exclusively mined from the Illawarra coal measures in the
southern region of the Sydney Basin. Co-located with these coal
reserves are significant quantities of methane (CH
4
) and carbon
dioxide (CO
2
) gas [2].
Fugitive emissions of gas from mining via ventilation air not
only contribute towards greenhouse gas (GHG) inventory, but in
the case of methane, also represent a lost opportunity for energy
recovery. Gas reserves are not limited to economically recoverable
coal seams, but also include coal measures and other porous
stratigraphy both above and below the working seam [3].
Emission predictions are essential for the quantification and
management of risk associated with sudden gas release during
mining (outbursts), and accumulation of noxious or combustible
gases within the mining environment. Unexplained variation in
gas character rightly requires conservative mining practices to
manage such risks [4].
In many cases, risks are identified later in the mining cycle
where remedial action is typically more expensive and is more
likely to incur production delay or loss.
Over 2500 gas core samples from three southern Sydney Basin
mines producing from the Bulli seam have been analysed in vari-
ous geospatial context including relationships to hydrological fea-
tures and geological structures.
Improved spatial datasets, particularly those containing a verti-
cal dimension and derivatives thereof, may be applied to predic-
tion and management of gas emission using fundamental
principles. The application of the physical and spatial techniques
described enhance the potential future use of high volume and
high resolution real time measurement data for proactive manage-
ment of gas emission risk much earlier in both the gas and mining
life cycle.
The improved resolution and definition in the prediction of site
specific transient gas emission character, in terms of source loca-
tion, quantity, composition, flow path and timing is acknowledged
by several authors as critical for maintaining current production
rates in higher gas content environments [3,5,6].
Gas emissions will increase well beyond the practical manage-
ment capacity of ventilation and current pre and post drainage sys-
tems at several Australian underground coal mines [4]. Hence the
traditional approach of increasing ventilation quantity is unlikely
http://dx.doi.org/10.1016/j.ijmst.2017.07.007
2095-2686/Ó2017 Published by Elsevier B.V. on behalf of China University of Mining & Technology.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
⇑
Corresponding author.
E-mail address: peb987@uowmail.edu.au (P. Booth).
International Journal of Mining Science and Technology xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
International Journal of Mining Science and Technology
journal homepage: www.elsevier.com/locate/ijmst
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007

to be sustainable due to practical constraints such as roadway area
and maximum air velocity therein.
Only a step change improvement in gas drainage, capture and
utilisation practices will allow coal to remain a sustainable source
of energy in a low emission world [3]. The identification and use of
gas management controls which are fundamentally based and
incorporate improved spatial and time resolution will not only
make mining safer, delivery of this outcome will reduce interrup-
tions for reasons of safety management and lift both coal and over-
all energy productivity.
2. Historical gas emission prediction
The prediction of methane emissions arising from underground
coal mining has been the subject of extensive research for several
decades and techniques range from simple geometric models to
modern finite element models [7–12]. Despite improvement in
computation processing power and speed over this time period,
calculation techniques remain empirically based and are hence
limited to the origin of information in both application and
resolution.
Gas emissions due to mining extraction are transient and a
complex function of the in-situ resource character, the space where
in-situ character and gas equilibrium is affected by extraction, the
degree to which character and equilibrium is affected, and the sys-
tem response [9].
In order to simplify the calculation process of most current pre-
diction techniques, key inputs for gas content, material properties
and spatial attributes are generally either (1) provided as input
variables at low resolution, (2) held constant, or (3) neglected
altogether.
Of the many prediction techniques available, the Flügge tech-
nique continues to be used for the purpose of total specific gas
emission calculation at many Australian mines [13,14]. However,
limitations in describing spatial and time based gas emission char-
acter with any resolution renders this technique ineffective for
design of gas drainage programs. Evidence provided through finite
element analysis and micro seismic observations suggest the trian-
gular prism representation is only valid in specific geological con-
ditions and does not cater well for changes in either geology or
operational practices [15].
Research by Lama in the 1990s led to the significant reduction
of risk associated with gas outburst through the development of
composition dependent gas content threshold levels for the Bulli
seam [16]. These thresholds or derivations thereof largely remain
in place in the Australian coal industry to the present day due to
the principles based methodology used. Further research during
the latter part of the decade also focussed on developing an under-
standing of fundamental mechanisms driving gas emission beha-
viour from coal and surrounding strata [16–18]. The importance
of cleat and joint geometry and net effective stress in the control
of fluid movement was highlighted.
A detailed description of the process for measurement of gas
content and its’ contributing components may be found in Aus-
tralian Standard AS 3980 [19]. Limitations of some of the measure-
ment techniques used, specifically including assumptions of the
timing of initial desorption and the lost gas component Q
1
, are dis-
cussed further by Saghafi [20].
Other factors considered in emission prediction include differ-
ential sorption properties of coal under the effect of a shear struc-
ture, and gas pressure measurements which change as a result of
changes in the permeability of the structure. Significantly, the frac-
ture density and sorption properties may change up to 20 m away
from the shear structure, but gas pressure changes can occur up to
100 m away from the structure.
The GeoGAS Longwall ‘‘pore pressure” model described by
Ashelford took account of many gas reservoir and geological
parameters of coal seams and allowed variation of mining opera-
tions in arriving at a gas emission value [11]. The model relies upon
measured gas reservoir properties for the determination of gas
release such as; measured gas content (Q
m
), gas desorption rate,
gas composition, gas sorption capacity, seam thickness and min-
eral matter above and below the working section, pore pressure
and coal and sandstone porosity. The model parameters and how
they are measured are described by Williams et al. [21].
The advantage of this model over prior techniques was its’ abil-
ity to accurately predict the magnitude of gas emission from the
floor seams below the Bulli seam in the southern Sydney Basin.
This was due to the significant deformation and order of magni-
tude changes in horizontal and vertical stress in the floor strata
recognised and displayed by finite element software. Whilst the
pore pressure model remains the most adaptive and fundamen-
tally based calculation of gas emission for longwall operations,
the input assumptions limit the application of this technique to
the increasing spatial and time resolution required for design of
gas drainage programs.
The availability of increasing computational processing capabil-
ity has enabled the management of the increased size and com-
plexity of the data available for gas emission analysis in recent
years. Studies including those by Karacan used statistical, principle
component analysis (PCA) and artificial neural network (ANN)
based approaches to predict the ventilation methane emission
rates of U.S. longwall mines [10,22,23].
Critically, all techniques which involve the use of large histori-
cal data sets for gas emission prediction by analysis using statisti-
cal, PCA or ANN approaches rely on a fundamental assumption that
input conditions will not materially change. Model outputs are
based in fundamental scientific principles however the model
design and structure limits the ability for its use in locations where
input conditions change rapidly.
Comparison of the output of various prediction models is diffi-
cult due to lack of a common gas, material and spatial datum ref-
erence and also for the reasons discussed in Jensen et al. [24].
3. Relevant gas fundamentals used
3.1. Gas generation
Coalbed or coal seam gas are general terms used to describe
gases contained within coal measures that are generated as part
of coalification and other geological and hydrogeological processes
[25]. Similar to the creation of coal itself, coal bed gas generation
pathways are also dependent on fundamental physical and chem-
ical character and changes in both level and form of energy within
the environment. Coal bed methane can be classified as either bio-
genic or thermogenic in origin [26].
Biogenic methane is generated at low temperature by anaerobic
microbes (methanogens) when coal beds are exposed to ground-
water recharge after basin deformation. The dominant biological
processes involved in the generation of biogenic methane include
carbon dioxide reduction and acetate reduction or fermentation
which are described in chemical Eqs. (1)–(3). Two significant fac-
tors must be carefully considered in the characterisation of the ori-
gin of biogenic gas. Firstly, for carbon dioxide reduction to
methane, hydrogen must be present. Secondly, in addition to the
methane, the two-part acetate fermentation process also produces
CO
2
[27].
CO
2
þ4H
2
!CH
4
þ2H
2
Oð1Þ
CH
3
COO

þH
þ
!H
4
þCO
2
ð2Þ
2P. Booth et al. / International Journal of Mining Science and Technology xxx (2017) xxx–xxx
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007

CH
3
COO

þH
2
O!CH
4
þHCO
3
ð3Þ
Availability of hydrogen ions is increased via groundwater flow
and recharge in subterranean aquifers. Such aquifers may include
seawater sources, noting that sea water is under saturated with
respect to its salts, except for calcium carbonate which may occur
in saturated or near-saturated state. The flow pathway of water is
therefore an important factor in characterising gas reservoir condi-
tions. The relative rate of change of coal seam gradient and orien-
tation hence provide information on available potential energy
under the influence of gravity. The effect of gravity on hydrogeo-
logical and material deposition character has remained constant
over geological time.
Thermogenic gas is generated at high temperature during late
stage coalification and generally contains heavier carbon isotopes
than biogenic gas. The results described by Moore indicate that
the first gas generated via thermogenic processes is CO
2
at approx-
imately 50 °C[26]. Above this temperature, increasing amounts of
hydrocarbons (methane, ethane and higher) and nitrogen are pro-
duced at maximum volume at approximately 150 °C. At higher
temperature, gas generation reduces, producing a parabolic maxi-
mum gas volume trend with temperature and/or rank. Such para-
bolic gas content trends have been reported from a number of
Australian Basins including the southern Sydney Basin which is
the subject location of this research [28].
3.2. Gas storage
Over 90% of gas storage in coal occurs by physical adsorption to
the surface of the coal matrix, including the surfaces of all internal
pores and cleats or fractures [25]. The remaining is free gas, which
may also reside within internal pores depending on pore geometry,
and also within cleats or fractures. It is the physical adsorption pro-
cess which differentiates coal bed reservoirs from conventional gas
reservoirs. Conventional gas reservoirs may contain only one-sixth
to one-seventh of the equivalent coal bed reservoir by rock volume,
as the gas is free within porous spaces and not held to surfaces via
adsorption.
Adsorption concepts between gas and a solid surface are usually
described in terms of isotherms, where the amount of adsorbate on
adsorbent is shown as a function of pressure at constant tempera-
ture as depicted for three gases (CO
2
,CH
4
and N
2
) at one of the
study sites in Fig. 1.Fig. 1 also demonstrates the range of potential
variation in sorption capacity from upper, middle and lower sec-
tions within the 2–3 m thickness of the Bulli seam.
Langmuir suggested that adsorption takes place through a sin-
gle layer equilibrium mechanism [29]. At lower pressures, a molec-
ularly denser state allows greater volumes to be stored by sorption
than is possible by compression, and higher pressures increase the
potential that a particular molecule will find an adsorption site.
Gas composition is a fundamental controlling variable in determin-
ing total possible sorption capacity due to the relative size, struc-
ture and energy levels of relevant gas molecules, particularly CH
4
relative to CO
2
[30–32]. However, the availability of adsorption
sites or coal internal surface area remains the key limiting
parameter.
The coal structure hence sets the adsorption and desorption
character, which also changes with gas type, but this does not nec-
essarily mean that a coal of certain properties and whose sorption
capacity is described by a particular isotherm actually contains
that amount of gas for a given volume [14]. The ratio between
actual measured gas content and the theoretical sorption capacity
is known as the degree of saturation and is expressed as a percent-
age. Lower in-situ degree of saturation is an indicator that other
mechanisms, such as lowering of hydrostatic pressure through
fluid movement, have potentially allowed gas to migrate or other-
wise be released from the coal after initial gas generation.
Gas storage capacity is hence defined by the combination of gas
composition and coal properties and structure. Using fundamental
physical, chemical, energy and geometric relationships, it is postu-
lated that for the purpose of gas emission prediction, dynamic
response of the gas reservoir to mining extraction can be reliably
predicted using higher resolution input spatial parameters and
measured coal property data which is largely available through
proximate characterisation parameters such as rank, carbon con-
tent, macerals, and moisture content. The range of possible varia-
tion in coal properties for one of the study sites is demonstrated
in Table 1.
Coal structural properties are also unlikely to remain constant
over a given mining horizon, but rather be significantly influenced
by the landscape at time of deposition. Hence these properties are
influenced by spatial factors which may be either measured
directly or reliably interpolated using fundamental spatial relation-
ships. Mineralisation also influences internal structure, geometry,
pore availability to gas adsorption, ability of gas to flow, shrinking
and swelling, gas content, gas recoverability, and potential for
enhanced gas recovery [33].
Experimental evidence closely correlates increasing coal rank
with higher proportions of micropores [31]. An increase in micro-
pore distribution per unit of coal volume also increases surface
area available for gas adsorption, hence explaining observed exper-
imental increase in gas storage capacity with coal rank, and
increase in rate of change of volumetric capacity per unit pressure
change as described by Kim in the review by Moore [26].
3.3. Gas flow
The movement of gas molecules through either other gases, flu-
ids or solids are described by Fick’s Laws [20]. The key point for the
diffusive behaviour of gas described by these laws is that the
energy driving the diffusion process is atomic energy, and molecu-
lar vibration motion in response to this energy [34]. The gas con-
centration gradient is a proxy term for a molecular energy
density per unit volume gradient across three-dimensional space.
This is relatively small in total available energy terms, in the
absence of other forces (e.g. pressure gradients). The critical point
being that on reducing spatial component of the denominator of
both of Fick’s equations (dx), it becomes more probable that mole-
cules will be subject to much larger external energy forces (e.g.
pressure gradients) in shorter time frames.
Darcy’s law is an expression of conservation of momentum and
describes a proportional relationship between the instantaneous
discharge rate through a porous medium, viscosity of the fluid
and pressure drop over a given distance. This equation can also
be solved for permeability, allowing for relative permeability to
be calculated. In practice, this measurement is difficult and expen-
Fig. 1. Range of calculated sorption capacities for Bulli seam at Mine A with typical
seam reservoir pressure range (shaded).
P. Booth et al./ International Journal of Mining Science and Technology xxx (2017) xxx–xxx 3
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007

sive to complete in-situ, but is the only method of obtaining a true
permeability result which reflects the reservoir conditions [17].
In case of coal, permeability is a complex, multi-dimensional
function of several influences such as width, length, height, aper-
ture spacing, frequency or density, and connectivity of cleats or
fractures [25]. Many of these influence functions are non-linear,
however, have components that can be either readily measured
directly or indirectly or otherwise grouped without affecting mate-
rially affecting calculation results. Changes in permeability in coal
may be summarised into two main components; the effective
stress effects, and the shrink and swell strain effects on the coal
matrix with desorption or adsorption which may increase or
decrease relative permeability [35,36].
In case of coal, permeability is a complex, multi-dimensional
function of several influences such as width, length, height, aper-
ture spacing, frequency or density, and connectivity of cleats or
fractures [25]. Many of these influence functions are non-linear,
however, have components that can be either readily measured
directly or indirectly or otherwise grouped without affecting mate-
rially affecting calculation results. Coal composition hence controls
a broad range of gas reservoir properties including gas adsorption
capacity, gas content, porosity, permeability and gas transport.
4. Calculation of relevant spatial properties
The fundamental nature of the physical and chemical interac-
tions between the principal components of coal, coal seam gases
and other substances found in the mining environment have
remained constant over geological time. These interactions are sig-
nificantly influenced by the various forms of energy applied over
time, however the potential energy involved in sedimentary depo-
sition, gas generation and flow is of particular relevance to analysis
of gas emission at higher resolution. An overview of the process for
calculation of the relevant spatial properties is depicted in Fig. 2.In
the absence of vertical dimension data specific to the location of
gas core samples, alternate sources of vertical information or inter-
polation can be used to inform predictive modelling of gas
emissions.
4.1. Derivation of the elevation surface
The goal of the first stage of model development is to obtain the
best possible representation of the relative level (RL) of the floor of
the coal seam in the subject area using a common datum. To
achieve this, original sources of data included but were not limited
to, manual survey, drilling records and seismic interpretation. Ide-
ally, all input data is provided in the form of three-dimensional
points. However, this is not always available.
4.1.1. Location (point) data sources
Care must be taken to ensure the use of a common reference
datum for all available location and level measurements entered
as input data. The Map Grid of Australia (MGA) Zone 56 and Aus-
tralian Height Datum (AHD) were selected as the common refer-
ence datum for all location and level information used, and
several data sources required conversion to this datum initially.
The dimensional convention used throughout this study is X
related to longitudinal co-ordinates, Yrelated to latitudinal co-
ordinates, and Zrelated to height or RL co-ordinates.
Although sources of RL data may include contours, these are
generally previous interpolations of X,Y, and Zpoint data sources.
To allow for future model prediction, development and improve-
ment in real time location measurement technology, all RL input
data sources were converted to X,Yand Zpoint data before pro-
ceeding to the next stage.
4.1.2. Interpolation technique selection and output
Selection of the interpolation technique suitable for creation of
the elevation surface used in this study considered a range of selec-
tion criteria. Input data and processing constraints and future use
of the interpolation outputs in later model processes received
higher weighting in the assessment process. Of the many tech-
niques described and compared in the literature, the spline with
barriers interpolation was selected due to the use of exact mea-
sured point data as input data, the ability of the technique to man-
age known abrupt changes in level (e.g. geological faults), the
maintainability of the interpolation process with updated input
data, and the ability to balance competing requirements for pro-
cessing time and output resolution.
The result of the interpolation is a raster surface of configurable
grid cell size, using barriers, from measured points using a mini-
mum curvature spline technique. The barriers are entered as poly-
line features, and the resulting smooth surface is constrained by
the input barrier features. Input datasets may also have several
points with the same xand ycoordinates. An important feature
of this technique is that if the values of the points at the common
Table 1
Range of coal properties within the Bulli seam–Mine A.
Sample Proximate analysis (%, by weight) Coal grain density (kg/m
3
) Intraparticle porosity (%) Surface area (m
2
/g)
Moisture Ash Volatiles Fixed Carbon
Top 1.35 12.53 17.98 68.14 1466 4.37 0.333
Middle 1.00 10.53 23.55 64.92 1579 1.92 0.144
Bottom 1.15 25.88 15.06 57.91 1311 3.83 0.292
Note: all proximate analysis on percentage by weight air dried basis; and surface area data is calculated based on the pores whose diameter above 30 nm.
Fig. 2. Process for calculation of spatial properties.
4P. Booth et al. / International Journal of Mining Science and Technology xxx (2017) xxx–xxx
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007

location are the same, they are considered duplicates and have no
effect on the output.
Several tests were undertaken of total surface calculation area
versus raster resolution (individual cell size) and processing time.
The final selected digital elevation model (DEM) configuration for
the two study sites covered an area of approximately 200 km
2
each, contained multiple barrier features of both two and three-
dimensional data types, at a 1 m 1 m resolution and processed
in approximately 10 min.
This configuration was deemed acceptable for future use and
maintainability of the modelling process, and considering eventual
development to multi-stratum environments.
Fig. 3 provides an example three-dimensional DEM representa-
tion of the Illawarra coal measures viewed from the north looking
south. The Bulli seam is the uppermost seam in the sequence dis-
played in light grey, conformably overlying up to 5 other seams.
Surface elevations appear in green, generally above the AHD zero
reference shown in blue. The study area is shown in light tan with
relevant geological structures shown in red.
4.2. Spatial parameters derived from elevation surface
Spatial parameters deemed essential for the eventual calcula-
tion of gas emission character across the mining environment
and calculated at the required higher resolution (i.e. each
1m1 m cell) are the vertical dimension Z in metres (AHD), the
maximum slope in degrees (0–90°), the aspect in degrees (0–
360°), and the curvature in metres per metre, overall and then sep-
arately in plan and in profile.
Calculation of the above italicised terms all involve the evalua-
tion of the cell in consideration against each of up to eight of its
neighbouring cells in the horizontal X-Yplane. The slope is a repre-
sentation of the maximum rate of change of elevation (Z) with
respect to both Xand Y. The aspect represents the orientation of
the slope, where values near 0°and 360°indicate a north facing
area, 90°an easterly facing, 180°a south facing, and 270°a west
facing area.
The calculation of curvature is the derivative of the slope (i.e.
d
2
z/dxy
2
) using a similar process to the slope calculation, but using
the slope value for each cell as the input to the curvature calcula-
tion. By definition, the curvature at a point with zero slope (flat)
will also be zero. This phenomenon may be used to determine
areas likely to retain fluid, also known as sinks. Curvature may
be further defined into profile and plan curvatures, which are use-
ful for describing the acceleration or deceleration of flow paths in
the case of the profile curvature, or convergence or divergence of
flows in the case of plan.
5. Application of spatial parameters to gas samples
Over 2500 gas core sample locations from two mine sites were
initially provided in the form of AutoCAD drawing files, complete
with two-dimensional Xand Yco-ordinates. Drilling trajectories
to obtain the core samples were also provided in most cases. Lab-
oratory sample analysis results containing a range of gas properties
were provided in the form of MS Excel spreadsheets. Gas parame-
ters included gas content, gas composition and concentration, and
desorption characteristics with a unique reference to a sample or
core identification number. An overview of the collation process
and gas property information contained in the dataset is depicted
in Fig. 4.
5.1. Gas sample location referencing
The first stage of assessment of spatial parameters involved ref-
erencing AutoCAD two-dimensional location information for each
unique sample to the laboratory analysis results. Significantly, this
process revealed a 5–10% mismatch error rate, which was identi-
fied using standard database tools.
Errors appeared to be caused by either incorrect data entry into
spreadsheet or incorrect placement of sample location or sample
number within the AutoCAD drawing. Such errors were resolved
by manually reconfirming sample results and locations with dril-
ling records.
5.2. Allocation of spatial parameter data to gas samples
Once gas core sample locations were confirmed as two-
dimensional X,Ypoints, the next stage of assessment involved
the allocation of all previously calculated values for elevation,
slope, aspect, and curvature to each of the gas samples. The output
of these previously calculated spatial parameters was, in each case,
a raster surface of 1 m 1 m resolution. As each of the gas sample
locations were specified as unique points, the process of extracting
the relevant spatial parameters from the raster surfaces and allo-
cating the value to the gas sample point was completed in less than
5 min.
The final stage of spatial data allocation involved the calculation
of the distance and direction to the nearest geological structure.
Structures may include faults, dykes or other anomalies which
may be represented as either a two or three dimensional features.
For initial assessment, a simple two-dimensional planar distance
and direction was selected, although the software is capable of full
three-dimensional calculation.
6. Initial observations and results
A range of two and three-dimensional representations of the
dataset were prepared for preliminary interpretation and visual
trend observation. As existing mining threshold limits are deter-
mined primarily by measured gas content and gas composition,
these dependent variables were considered initially.
Assessment of the full dataset’s gas content result using simple
scatter distribution analysis by X,Yand Zlocation did not reveal
any significant first order linear trend.
Whilst the magnitude of observations in Fig. 5 might suggest
linear trends with respect to the vertical Zdimension, this is a
function of the greater sampling density within the mining hori-
zons of the subject mines.
Assessment of gas composition using scatter distribution analy-
sis of CO
2
concentration by X,Yand Zdimensions revealed a loca-
lised trend with respect to the Zdimension at each individual mine
as shown in Fig. 6.
Analysing the dataset collectively, a distinct layering of gas
composition and concentration is observed with respect to the ver-
tical dimension as shown in Fig. 7.
Localised trends appeared at each mine with increasing CO
2
concentrations being observed downslope of higher CH
4
concen-
trations and geological features. Increasing observed gas content
Fig. 3. Example three-dimensional DEM of study area.
P. Booth et al. / International Journal of Mining Science and Technology xxx (2017) xxx–xxx 5
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007

with CO
2
concentration shown in Fig. 8 is primarily accounted for
by comparison to the experimentally determined isotherms for the
mine (Fig. 1), recognising that coal structural properties and hence
sorption capacity is also likely to vary relative to many spatial
parameters. Difference in localised seam hydrostatic pressure
may also account for such observations, however in the absence
of in-situ pressure measurement, this could not be confirmed.
Of significance in Fig. 8 is the apparent single outlier having a
CO
2
concentration of approximately 10% and measured gas content
of 20 m
3
/t. On further investigation, it was found that this core
sample was actually taken from a cross-measure drill hole from
the Bulli seam to the underlying Wongawilli seam and had a CH
4
concentration of approximately 85%. Such display demonstrates
the potential ability of the spatial techniques used for easy anom-
aly detection within large datasets.
The introduction of further independent spatial variables for
slope, aspect, curvature and geological structures visually sug-
gested a strong dependence between higher gas content and areas
where localised fluid accumulation or flow restriction was likely to
occur. The number of core samples taken in these areas over an
extended time period, combined with the number of gas drainage
holes drilled in the immediate area, suggests that these areas were
also difficult to drain.
An example of these areas within Mine A is depicted in Fig. 9.
The gas composition of this particular area was greater than ninety
percent CO
2
, however the dependence between areas of likely fluid
accumulation and higher gas content appeared to be independent
of gas composition. Other areas of Mine A with higher CH
4
compo-
sition also demonstrated a similar relationship. The seam reservoir
gas pressure for this area was estimated to be in the order of
3 MPa.
Although not finalised, datasets collected from each mine
include attributes which will allow calculation of gas drainage
quantities and timing. Due to the configuration of the database,
spatial relationships between various attributes may be assessed
using the same process as described in Section 4.2.
Data from Mine B also suggests a similar strong relationship
between spatial characteristics and higher gas content. At this
mine, such spatial relationships also appeared to be independent
of gas composition. Areas dominated by higher CO
2
concentration
were laterally separated from areas of higher CH
4
concentration by
over 2000 m. However, a similar localised trend of higher CO
2
con-
centrations downslope of geological features and higher CH
4
con-
centrations was observed.
In summary, each observed high gas content sample location
exhibited one or more the following spatial characteristics;
(1) Immediately adjacent to and upslope from structures form-
ing flow barriers or restrictions to the general trend of
within seam flow,
(2) An adjacent high rate of change of slope (curvature) tending
to localised minima where both slope and curvature tends to
zero,
Fig. 4. Gas database collation process.
Fig. 5. Three-dimensional gas content distribution.
Fig. 6. Three-dimensional gas composition distribution.
Fig. 7. Multivariate scatter analysis of gas content and RL (m) by gas composition
using CO
2
concentration as color.
Fig. 8. Multivariate scatter analysis of gas content and gas composition using gas
content as legend colour.
6P. Booth et al. / International Journal of Mining Science and Technology xxx (2017) xxx–xxx
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007

(3) A coincident or immediately adjacent change in aspect from
the general aspect trend.
In general, very localised areas featuring all of the above charac-
teristics tended to exhibit higher gas content towards the upper
extreme of the sample range. As these gas content observations
also approached the sorption capacities displayed on the experi-
mentally derived gas isotherm, it is suggested these areas are at
or near saturation for the given seam reservoir pressure.
7. Preliminary multivariate statistical analysis of spatial
parameters
A preliminary ordinary least squares linear regression test was
undertaken, using gas content as the dependent variable, in order
to statistically confirm the observations made visually. The statis-
tical significance of the input candidate variables, with the con-
tributing direction of the relationship is shown in Table 2.
As expected, due to the form of the input candidate variables for
aspect, slope and near angle in degrees, the adjusted R
2
of any pre-
liminary linear predictive model tested was poor. However, the
results demonstrate that predictive models incorporating deriva-
tives of the candidate spatial variables are worthy of further inves-
tigation. Furthermore, due to the use of two-dimensional planar
proximity calculations for preliminary derivation of the proximity
candidate variable, it is expected that a three-dimensional proxim-
ity assessment yielding full three-dimensional magnitude and
bearing values to near structures will significantly improve the
predictive model fit and eventual results.
8. Conclusions and future directions
Over 2500 gas core samples from two southern Sydney basin
mines producing metallurgical coal from the Bulli seam have been
analysed in various geospatial context. A robust foundation for the
process to obtain, prepare and load the relevant spatial input data-
sets into a predictive model has been described.
Spatial relationships between measured gas content, gas com-
position, and spatial parameters such as RL, slope, aspect and cur-
vature have been determined. The relevance and importance of
determining these relationships at a localised or site-specific,
rather than regional level have been demonstrated.
Statistically significant determining factors, including those
influenced by hydrological features and geological structures have
been identified and will be investigated further.
Further development of the predictive model to include time
and material property dimensions, full three-dimensional assess-
ment of proximity to adjacent structures and gas drainage holes,
techniques to normalise input datasets to improve calculation
speed, and incorporation of hydrological assessment tools, will sig-
nificantly improve model outcomes. This will allow further appli-
cation of the model to site specific and more complex geology
including multi-seam mining environments.
The results suggest variability and limitations associated with
the present traditional approaches to gas emission prediction and
design of gas management practices may be addressed using pre-
dictions derived from improved spatial datasets, and analysis tech-
niques incorporating fundamental physical and energy related
principles. This foundation will allow increasingly complex factors,
such as strata material properties, and stress directions and magni-
tude to be incorporated into predictive models.
The application of the physical and spatial techniques described
enhances the potential for use of high volume and high resolution
real time measurement data in management of gas emission risk.
By proactively addressing such risks earlier in both the gas and
mining life cycle, material reduction of costs and improvement
Fig. 9. Representation of gas content distribution analysis at Mine A using aspect, slope and curvature parameters.
Table 2
Summary of variable statistical significance.
Variable Significant (%) Negative (%) Positive (%)
X(longitude) 100 100 0
Y(latitude) 96.93 100 0
CO
2
(%) 87.12 0 100
Aspect (°) 60.12 66.87 33.13
Near angle (°) 53.37 100 0
Z (RL metres) 49.69 44.79 55.21
Proximity (m) 25.77 80.37 19.63
Slope (°) 1.84 83.44 16.56
P. Booth et al. / International Journal of Mining Science and Technology xxx (2017) xxx–xxx 7
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007

production and environmental outcomes are more likely to be
obtained.
Acknowledgments
This research has been conducted with the support of the Aus-
tralian Government Research Training Program Scholarship. The
authors also gratefully acknowledge the direct financial support
of MeCee Solutions Pty Ltd. The technical assistance and provision
of data from the respective anonymous study sites is appreciated.
The support of fellow researchers at the University of Wollongong
in the provision of coal sample analysis is acknowledged, as well as
comments from anonymous reviewers.
References
[1] Department of Industry, Innovation and Science. Resources and Energy
Quarterly-December Quarter; 2015.
[2] Geoscience Australia. Coal Seam Gas Fact Sheet Canberra: Australia; 2015.
[3] Karacan CÖ, Ruiz FA, Cotè M, Phipps S. Coal mine methane: a review of capture
and utilization practices with benefits to mining safety and to greenhouse gas
reduction. Int J Coal Geol 2011;86(2–3):121–56.
[4] Rao Balusu, Srinivasa Yarlagadda, Ting R, Shi S, Roy Moreby. Strategic review of
gas management options for reduced ghg emissions. ACARP; 2010 1/5/2010.
Report No.: C17057.
[5] Packham R, Cinar Y, Moreby R. Simulation of an enhanced gas recovery field
trial for coal mine gas management. Int J Coal Geol 2011;85(3–4):247–56.
[6] Wang F, Ren TX, Hungerford F, Tu S, Aziz N. Advanced directional drilling
technology for gas drainage and exploration in Australian coal mines. In:
Proceedings of the 1st International Symposium on Mine Safety Science and
Engineering, ISMSSE. Beijing; 2011. p. 25–36.
[7] Curl SJ. Methane prediction in coal mines. IEA Coal Research - Technical
Information Service; 1978.
[8] Creedy DP. Methane emissions from coal related sources in Britain:
development of a methodology. Chemosphere 1993;26(1–4):419–39.
[9] Lunarzewski LW. Gas emission prediction and recovery in underground coal
mines. Int J Coal Geol 1998;35(1–4):117–45.
[10] Karacan CÖ. Modeling and prediction of ventilation methane emissions of U.S.
longwall mines using supervised artificial neural networks. Int J Coal Geol
2008;73(3–4):371–87.
[11] Longwall Ashelford D. ‘‘pore pressure” gas emission model. In: Proceedings of
the AusIMM Coal Conference. University of Wollongong, New South Wales,
Australia; 2003.
[12] Guo H, Yuan L, Shen B, Qu Q, Xue J. Mining-induced strata stress changes,
fractures and gas flow dynamics in multi-seam longwall mining. Int J Rock
Mech Min Sci 2012;54(3):129–39.
[13] Meyer T. Surface goaf hole drainage trials at illawarra coal. In: Proceedings of
the AusIMM Coal Conference. University of Wollongong, New South Wales,
Australia; 2006.
[14] Black DJ. Factors affecting the drainage of gas from coal and methods to
improve drainage effectiveness. School of Civil, Mining and Environmental
Engineering. New South Wales: University of Wollongong; 2011.
[15] Kelly M, Gale W, Hatherly P, Balusu R, Luo X. Combining modern assessment
methods to improve understanding of longwall geomechanics. In: Proceedings
of Coal Conference. University of Wollongong, New South Wales, Australia;
1998.
[16] Lama RD, Bodziony J. Management of outburst in underground coal mines. Int J
Coal Geol 1998;35(1–4):83–115.
[17] Gray I. Reservoir engineering in coal seams: part 2—observations of gas
movement in coal. SPE Reservoir Eng 1987;2(1):35–40.
[18] Barker-Read GR, Radchenko SA. Gas emission from coal and associated strata:
interpretation of quantity sorption-kinetic characteristics. Min Sci Technol
1989;8(3):263–84.
[19] Australian Standard. AS3980-1999 Reconfirmed 2013-Guide to the
determination of gas content of coal—Direct desorption
method. Sydney: Standards Australia; 2013.
[20] Saghafi A. Determination of the gas content of coal. In: Proceedings of 16th
Coal Operators’ Conference. University of Wollongong, New South Wales,
Australia; 2016.
[21] Williams RJ, Yurakov E, Ashelford DJ. Gas emission modelling of gate road
development. Australia: The AusIMM; 2001. p. 45–51.
[22] Karacan CÖ, Goodman GVR. A CART technique to adjust production from
longwall coal operations under ventilation constraints. Safety Sci 2012;50
(3):510–22.
[23] Karacan CÖ, Olea RA. Inference of strata separation and gas emission paths in
longwall overburden using continuous wavelet transform of well logs and
geostatistical simulation. J Appl Geophys 2014;105:147–58.
[24] Jensen B, Gillies ADS, Anderson JM, Jones N. Review of Methane emission and
prediction in longwall coal mines. In: Proceedings of Aus Inst Min Met; 1992.
p. 11–17.
[25] Flores RM. Coalbed methane: from hazard to resource. Int J Coal Geol 1998;35
(1–4):3–26.
[26] Moore TA. Coalbed methane: a review. Int J Coal Geol 2012;101:36–81.
[27] Burra A, Esterle JS, Golding SD. Coal seam gas distribution and hydrodynamics
of the Sydney Basin, NSW, Australia. Austral J Earth Sci 2014;61(3):427–51.
[28] Faiz M, Saghafi A, Sherwood N, Wang I. The influence of petrological properties
and burial history on coal seam methane reservoir characterisation, Sydney
Basin, Australia. Int J Coal Geol 2007;70(1–3):193–208.
[29] Langmuir I. The adsorption of gases on plane surfaces of glass, mica and
platinum. J Am Chem Soc 1918;40(9):1361–403.
[30] Booth P, Nemcik J, Ren T. A critical review and new approach for determination
of transient gas emission behaviour in underground coal mines. In:
Proceedings of the 16th Coal Operators’ Conference. University of
Wollongong, New South Wales, Australia; 2016. p. 367–79.
[31] Mosher K, He J, Liu Y, Rupp E, Wilcox J. Molecular simulation of methane
adsorption in micro- and mesoporous carbons with applications to coal and
gas shale systems. Int J Coal Geol 2013;109–110(2):36–44.
[32] Mohanty MM, Pal BK. Sorption behavior of coal for implication in coal bed
methane an overview. Int J Min Sci Technol 2017;27(2):307–14.
[33] Flores RM. Coal and coalbed gas: fueling the future. Burlington: Elsevier
Science; 2013.
[34] Pan Z, Connell LD. Modelling permeability for coal reservoirs: a review of
analytical models and testing data. Int J Coal Geol 2012;92:1–44.
[35] Cai Y, Liu D, Mathews JP, Pan Z, Elsworth D, Yao Y, et al. Permeability evolution
in fractured coal—combining triaxial confinement with X-ray computed
tomography, acoustic emission and ultrasonic techniques. Int J Coal Geol
2014;122(1):91–104.
[36] Li J, Heap AD. A review of spatial interpolation methods for environmental
scientists. Canberra, Australia: Department of Resources Energy and Tourism;
2008.
8P. Booth et al. / International Journal of Mining Science and Technology xxx (2017) xxx–xxx
Please cite this article in press as: Booth P et al. Spatial context in the calculation of gas emissions for underground coal mines. Int J Min Sci Technol (2017),
http://dx.doi.org/10.1016/j.ijmst.2017.07.007