International Journal of Coal Geology 106 (2013) 39–47

Contents lists available at SciVerse ScienceDirect

International Journal of Coal Geology

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

A numerical study of stress changes in barrier pillars and a border area

in a longwall coal mine
Mahdi Shabanimashcool ⁎, Charlie C. Li
Department of Geology and Mineral Resource Engineering, Norwegian University of Science and Technology (NTNU), Trondheim, Norway

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

Article history: In longwall mining, stope voids are filled by caved-in rock materials. Cave-in and fracturing of roof strata
Received 3 September 2012 cause severe disturbance to and stress changes in the host rocks. The study site is the Svea Nord coal mine.
Received in revised form 20 December 2012 Longwall mining method was implemented to extract coal at the study site. It is intended that the coal in
Accepted 21 December 2012
the border area on one side of the longwall panels will be mined after completion of the longwall mining.
Available online 30 December 2012
There is a concern about how the longwall mining affects the stress state in the border area and how the
Keywords:
stress changes would affect future mining in the border area. 3D numerical modelling was conducted to con-
Longwall mining sider the above concerns. Two models were constructed in the study. The first one is a local model that in-
Numerical modelling cludes only two panels to study how the stresses in the barrier pillars are changed during panel mining.
Barrier pillars The second one is a mine-scale global model that includes all the panels with the aim of studying the distur-
Large-scale numerical modelling bance to the border area by the longwall mining. The simulations show that the stresses in the barrier pillars
Stress changes fluctuate up and down during mining because of periodic cave-in events behind the longwall face. A failure
zone of about 12 m exists in the wall of the barrier pillar. A large portion of the barrier pillar is still intact and
is thus capable of protecting the border area. A disturbed zone of about 20 m is found to have developed in
the wall of the main tunnels on the side of the border area after all the longwall panels have been mined out.
The stress state in the remaining portion of the border area remains unchanged. Therefore, by considering a
pillar with width of 25–35 m into the border area along the access tunnels, it will be possible to mine the rest
of the border area with room-and-pillar method in the future.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction
In longwall mining, stope voids are filled by caved-in rock mate-
rials. Cave-in of the roof strata results in stress changes in the rock
mass surrounding the longwall panels. The stress redistribution is
mainly associated with the in-situ stress state, the layout of the
panels, and the mining sequence. The vertical stress in panel walls
would increase with an enlargement of the mined-out region. The
horizontal stresses in the roof and floor are also changed by the ad-
vance of the longwall face as well as the enlargement of the
mined-out voids.
Whittaker (1974) developed an empirical approach to calculate
the vertical stress ahead of the longwall face for rock conditions
prevailing in the UK. In his approach, a yield zone exists ahead of
the face. The vertical stress is zero both at the face and at the rib
side. It increases quickly with the distance from the face. Wilson
(1983) developed an analytical–empirical method for the same
purpose, whose results agree with those of Whittaker (1974).
⁎ Corresponding author at: Sem Sælands veg 1, N-7491, Trondheim, Norway. Tel.: +47
73594831; fax: +47 73594814.
E-mail address: mahdi.shabanimashcool@ntnu.no (M. Shabanimashcool).
Numerical analyses conducted by, such as, Park and Gall (1989)
agree well with the results of Whittaker and Wilson. All of these stud-
ies showed that the maximum abutment stress of four to six times the
original vertical stress happens at a close distance from the stope face.
Recently, however, numerical studies and field measurements in
Australia showed that the maximum abutment stress of about twice
the original one occurs at a great distance from the stope face. Kelly
et al. (1996, 2002) argued that those diversities are owed to the
high horizontal stress acting along the panel. High horizontal stress
causes shearing of the intact rock and bedding planes which results
in abutment stress concentration far ahead of the stope.
The horizontal stress redistribution as a result of longwall mining
has been studied through field measurements and numerical modelling
as well. Such cases are, however, limited and there is no internationally
accepted approach to assess the horizontal stress redistribution due to
longwall mining. Gale and Blackwood (1987) carried out a numerical
investigation on the influences of the horizontal stress orientation on
the stability of the gates at an Australian colliery. They devised a
graph, depicting the stability of the gate roof in relation to the orienta-
tion of the principal in-situ stress with alignment of the gates. Recently
Coggen et al. (2012) carried out two and three dimensional numerical
simulations in order to study influences of the magnitude and orienta-
tion of the in-situ horizontal stresses on the stability of the gates. They

0166-5162/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.coal.2012.12.008
40 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47

concluded that gates driven at an angle to the in-situ stress field suffer 4200 m
asymmetric deformation with pronounced stress effect.
Mark et al. (2007) investigated horizontal stress changes in the
tailgate of a coal mine in the USA using detailed field measurements.
They showed that horizontal stress in the direction of the perpendic-
ular to the longwall panels dramatically increases owing to mining
and the principal stresses reorient owing to roof strata deformation
and yielding. They also claimed that the required length of the rock
bolt for reinforcing of the roof strata is dependent on the magnitude Global
of the horizontal stress perpendicular to the gates. model
The stress redistribution is associated with the geological condi-
tions, the mechanical properties of the rocks, the state of the in-situ
stresses, the layout of the panels, the mining sequence, and so on.
Neither analytical nor empirical approaches can take into account
too many factors in their estimations, so their applications have
limitations (Brady and Brown, 2002). Oversimplifications would
bring about uncertainties in the results. Numerical modelling, howev-
er, provides us with an opportunity to include a good number of

82300 m
relevant factors in the analysis so that the results may be more realis-
tic than those of analytical and empirical approaches. Furthermore,
the effects of single factors can be easily studied numerically, which
1000 m
is helpful to identify the most important factors for use in practice.
Numerical modelling together with field measurements is an appro-
priate way of studying stress redistributions in longwall coal mines.
Previous studies of this kind have proven the validity of the method,
e.g. Gale and Blackwood (1987), Singh et al. (2001), Mark et al.
(2007) and Jiang et al. (2012).
Svea Nord is a coal mine in the archipelago of Svalbard in the
North Sea. The longwall mining method is used at present, but it is
intended that the coal in the border area of the longwall panels will
be mined by room-and-pillar method in the future. There are
concerns about how the longwall mining affects the stress state in
the mining region, particularly in the border area, and how the stress
changes would affect future mining in the border area. This study Y
aims to investigate the above concerns through 3D numerical model-
ling. An algorithm developed by Shabanimashcool and Li (2011,
2012) was used for simulations. Two types of modelling were
X
conducted in this study. The first one is a local model that includes
only two panels to study how the stresses in the barrier pillars are
changed during mining in a panel. The second one is a mine-scale Fig. 1. Mine map and regions chosen for numerical modelling.

global model which includes all the panels. The aim of the global
model is to study the disturbance to the border area caused by the
longwall mining. The significance of the study is that it proves that
the failure zone in the country rock surrounding an underground
void of large volume is limited to a relatively small extent as long as
the void is backfilled by either caved-in materials or other types of
loose materials.
2. Study site
The Svea Nord mine produces about 3 million tonnes of coal annual-
ly. The longwall panels are approximately 250 m wide and 2500 m
long. The overburden varies from 50 to 400 m, and most of the ground
surface above the panels is covered by glaciers of up to 250 m in thick-
ness. Fig. 1 is a map showing the layout of the longwall panels.
In-situ rock stresses were measured in six locations in the mine by
the overcoring relief method. The principal stresses in the mine area
are vertical and horizontal. The vertical stress σv is equal to the
weight of the overburden and the horizontal stresses in the central
mining area are σH = 10 MPa and σh = 8 MPa. The horizontal stresses
σH and σh are approximately parallel with and perpendicular to the
panel length (a deviation of 15 to 20°), respectively. The ratios of
the horizontal principal stresses to the vertical stress are approxi-
mately kmax = σH / σv = 2 and kmin = σh / σv = 1.7 in the mining region
with the lowest overburden.
There are a number of thrust faults in the mining region, which
strike approximately perpendicularly to the panels and have limited
persistence.
Fig. 2 is a column map showing the sequence of the laminated
sedimentary coal/rock layers below and above the coal seam being
mined. The panel roof is marked by 0 on the scale ruler. The roof strata
are composed of siltstone and finely grained sandstone interblended
with thin coal and bentonite layers. A bentonite layer lies about 2–3 m
above the coal seam. The roof strata may be divided into three units:
unit 1 ranges from 0 to 2.5 m above the coal seam and is composed of
siltstone with thin coal interlayers; unit 2 is from 2.5 to 10 m and is
composed of sandstone and siltstone with coal interlayers; and unit 3
refers to the rock strata from 10 m upward and is mainly composed of
relatively massive sandstone and siltstone. Unit 1 is the immediate roof
of the panels. A bentonite layer marks the upper border of this unit.
The rock mass is in general dry in the mining region but wet in the
cave-in area. The water comes from the glacier above the cave-in
area. Water is not a big issue in the mine.
3. Numerical simulation
3.1. Simulation logarithm
In longwall mining, roof cave-in is a complex dynamic process in-
volving rock fracturing, disintegration, movement, stress redistribution,
 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47
(m)
100
30
Unit 3 90
20
80
10 Unit 3
70
Unit 2
Bentonite
Unit 1
0
60
-10
50
-20 40
-30
Sandstone
Siltstone
Coal
Fig. 2. Geological column of the mine.
and so on. The materials that have caved into the mined-out stopes are
compressed under the pressure of the overlying rock and become con-
solidated after a certain period. The consolidation of the caved-in mate-
rials would reduce the pressure on the chain pillars between panels
because a portion of the vertical load would be absorbed by the consol-
idated materials (Brady and Brown, 2002). The roof strata above the
panels can be divided into three zones in accordance with the extent
of disturbance induced by the mining underneath (Duplanic and
Brady, 1999; Peng and Chiange, 1984):
1. The cave-in zone, which is characterised by disintegrated and
downward-moving rock blocks. The stope voids are filled by the
disintegrated blocks.
2. The fracture zone, which is characterised by separation of bedding
planes and vertical fractures caused by the cave-in underneath.
The rock blocks are still confined and cannot move downward
freely under gravity.
3. The elastic continuum zone, which refers to the portion of the rock
mass above the fracture zone. No fractures are created due to the
mining underneath and the rock only reacts to the mining by elas-
tic deformations.
Shabanimashcool and Li (2012) developed an algorithm for
longwall mining simulations using FLAC 3D. The algorithm was devel-
oped based on a scenario for the caving process proposed by Peng and
Chiange (1984). The scenario states that caving starts when the
longwall face advances to a critical length lp, at which the cave-in ma-
terials fully fill the void underneath. After that, subsequent cave-in
events occur periodically after every face advance of ls until the entire
panel is mined out. In the algorithm shown in Fig. 3, the upper bound-
ary of the cave-in zone is defined by a critical value of the maximum
principal strain, ε t1. The rock caves in when the strain ε t1 is larger
than a critical value. It is found, through back analyses of subsidence
data measured on the ground surface, that the critical value of ε t1
for caving at the study site is 0.05 (Shabanimashcool and Li, 2012).
According to Shabanimashcool and Li (2012) the maximum principal
strain of the rock, ε t1, is larger than or equal to 0.05 in the cave-in
zone; that is,
t
ε1 ≥0:05:
41
FLAC3D Model
Excavation advance:
(1) 5 m, or
(2) l s if available
Run 100 cyc
Yes Unbal. force ratio <1e-5
No
Yes
If dzmax<hc
No
(1) Outline cave-in zone, ε1t >0.05
(2) Null elements - Create void
Yes
BF≥1.43
No
No Yes
l p and b available?
Determine l p and b
Yes l s available?
(1) Change cave-in zone to double- No
yield material (Backfill)
Determine l s
(2) Reduce E & v in fracture zone
Fig. 3. Flow chart illustrating the algorithm for numerical modelling of longwall mining
(Shabanimashcool and Li, 2012). hc is the height of the longwall panel and dzmax is the
maximum roof–floor convergence of the open stope.
In the numerical modelling, the cave-in process is simulated
through the following steps. The longwall face starts with an advance
of 5 m every time at the beginning of the simulation. The boundary of
the cave-in zone is outlined using the above criterion. The mesh
elements of the model in the cave-in zone are nulled; that is, zero
values are assigned to the relevant mechanical properties in the
mesh elements when the bulking factor, BF, of the caved-in materials
is larger than 1.43. BF is a parameter describing the increase in
volume of a material due to cave-in with respect to its original vol-
ume. The BF of the caved-in materials is defined as the ratio of the
volume of the cave-in zone to the volume of the rock strata that
caves in. BF is infinite before caving occurs and gradually declines
with the upward extension of the cave-in zone. It has been reported
that the void underneath is fully filled by caved-in materials when
BF is equal to or less than 1.43 (Brown, 2007). In the numerical
modelling, calculation iterations are carried out and BF is checked.
Excavation advances until BF is less than 1.43, which indicates that
the caved-in materials have fully filled the mined-out void under-
neath, and the cave-in process stops. From now on the materials
will be consolidated under the pressure of the strata above. A
double-yield material model is used for the consolidation of the
caved-in materials. The elastic properties, that is, the Young's modu-
lus and Poisson's ratio, of the rock mass in the fracture zone are
then reduced by a certain amount from their original values. At this
moment, the critical caving length lp and the thickness, b, of the
caving strata are determined. After that, a similar excavation process
is performed to find the critical advance length ls at which subsequent
ð1Þ cave-in events occur. Finally, the longwall face advances a length of ls
42 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47
every time and calculation iterations are carried out to update the
boundary of the cave-in zone and the BF in the non-filled voids of
the stope. This process continues until the entire panel has been
mined out.
The mechanical properties of the cave-in and fracture zones are
presented in Table 1. The values were obtained through back calcula-
tions of the surface subsidence measured on the ground surface. The
algorithm was calibrated in previous work (Shabanimashcool and Li,
2011, 2012). The simulation results of the modelling are in reasonable
agreement with field measurements. The calibrated model was used
to study the stress changes in the border area and in barrier pillars
in this study.
The sizes of the cave-in zone and fracture zone above the longwall
panels were determined using the algorithm described above. Fig. 4
shows the shapes of the cave-in and fracture zones in the portion
extending to the starting edge of the panel under three representative
overburdens of the mine. The panel height is 4 m in the model. After a
face advance of about 36 m the cave-in zone reaches its maximum
height of about 16 m. After that, the upper boundary of the cave-in
zone remains unchanged with further face advances regardless of
the thickness of the overburden. The upper boundary of the fracture
zone has a convex shape in the roof of the panel portion extending
to the abutment side. The height of the fracture zone remains at its
maximum in the portion of the mined-out panel which is at a dis-
tance of more than about 300 m from the starting edge of the panel.
The height of the fracture zone increases slightly with the overbur-
den. These results were used in the global model to investigate stress
changes in the border area.
Two types of model were used in this study. The first one is a local
model including only two panels to study how the stresses in the bar-
rier pillars are influenced by the mining activity in panels. The second
one is a mine-scale global model and includes all the panels. The aim
of the global model is to study the disturbance to the border area
caused by the longwall mining. The simulation results of the two
models are provided below.
3.2. Local model
It is intended that the coal in the border area of Svea Nord mine
will be mined out in the future. The barrier pillars beside the main
tunnels play a significant role in protecting the border area from dis-
turbance by the longwall mining. A local 3D numerical model was
established to study the stress changes in the pillars induced by
longwall mining activities. The local model includes only two panels
in order to simulate the influence of the mining process on the pillars
in detail.
The local model is composed of two panels, C4 and C6 (shown in
Fig. 1). Taking into account the symmetry of the displacement and
stresses in the cross section of a panel, the model is bordered along
the centre lines of the two panels in the horizontal plane. The size
Table 1
Mechanical properties of the cave-in and fracture zones.
Material
Constitutive
Properties
model
Cap pressure E0
εm
v (%)
(MPa) (MPa)
Cave-in
material
Double-Yield
80εv
p= 80 0.3
1−εv / εmv
E (MPa)
Fractured zone − failure zones are in the outer portion of the pillar. The maximum
150
εv is the volumetric strain; εmv is the maximum volumetric strain, P is the cap pressure,
E0 is the initial tangential modulus, C is the cohesion, E is the Young's modulus and v is
the Poisson's ratio.
160
110 m Rock + 250 m Ice
300 m Rock + 100 m Ice
Height above the panel floor (m)
400 m Rock
140
Upper boundary
120
of fracture zone
100
80
60
40
Upper boundary
20
of cave-in zone
0
0 50 100 150 200 250 300 350
Distance to the panel starting-end (m)
Fig. 4. The upper borders of the cave-in and fracture zone for three different overbur-
dens (Shabanimashcool and Li, 2012).
of the model is 1500 × 300 × 400 m (length × width × height; Fig. 5).
The overburden is assumed to be composed of 300 m of rock and
100 m of ice. The rock part of the overburden is included in the
model, while the ice part is replaced by a constant vertical stress on
the upper boundary of the model (Fig. 5b). The in-situ stresses includ-
ed in the model represent the field measurements. The rock mass is
assumed to be an isotropic material obeying a strain-softening law.
The mechanical properties of the rock materials are given in Table 2.
Bedding planes and sandwiched layers of coal and bentonite are in-
cluded in the models as discontinuities. These discontinuities obey
the Mohr–Coulomb law. The material parameters of the discontinu-
ities are given in Table 3. History and monitoring points within the
model are placed at the mid-height of the panel and along the centre
line of the barrier pillar (Fig. 5a).
The algorithm presented in Fig. 3 is implemented in the model.
Panel C4 is mined out first, followed by C6. It should be noted that a
local co-ordinate system is used in the local model with the horizon-
tal y-axis in the cross section of the panels, the horizontal x-axis along
the panel length, and the z-axis in the vertical direction (Fig. 5). Fig. 6
shows the stress changes along the centre line of the pillar at different
stages of mining. Mining in panel C4 causes negligible stress changes
in the barrier pillar. Noticeable stress changes occur only when the
distance from the longwall face to the pillar edge is less than 150 m.
Taking the vertical stress as an example, the vertical stress σzz in
the centre line of the barrier pillar starts to increase noticeably
when the longwall face advances to a position about 150 m from
the pillar edge (Fig. 6a). The stress gradually increases with shorten-
ing of the distance until the face is about 20 m from the pillar. After
that, rock failure starts to occur in the wall of the pillar, resulting in
a reduction in the vertical stress because of rock failure. The failure
zone starts at the pillar wall and extends to a depth of about 12 m
in the wall when the longwall face reaches the position of the pillar
edge. The vertical stress reaches its maximum, about 20 MPa, at the
end of the failure zone. The vertical stress drops to 14 MPa after
cave-in occurs in panel C6. The horizontal stresses change in a similar
Friction
angle.
way to the vertical stress until the longwall face reaches the pillar
C (MPa)
edge. Differently from the vertical stress, the horizontal stresses
(Deg.)
increase slightly after cave-in occurs in the mined-out area (Fig. 6b
and c).
0 30
Fig. 7 shows the modelled failure zones in pillar walls, marked by
green and red colours, after completion of the mining in panel C6. The
v
0.3
depth of the failure zone is about 12 m in the side wall of the barrier
pillar facing the mined-out panel. The huge inner portion of the pillar
is still intact. Thus, the barrier pillar can provide the border area with
good protection from the disturbance of longwall mining.
 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47 43

a Monitoring points

y
C6
Barrier pillar

300 m
x

Barrier pillar

b 300 m

Chain pillar
C6 C4
100 m

Fig. 5. The 3D local model: a) horizontal view; b) cross section of the model and the boundary conditions.

3.3. Global model of the overburden. The stress on the upper boundary is unevenly
distributed because of the variation in the elevation of the ground
The stress changes in the border area are affected by all the surface. It is assumed that the two in-situ horizontal stresses have
longwall panels. A mine-scale 3D global model was constructed to the same directions and magnitudes with the field measurements.
simulate the stress state in the border area after completion of mining The properties of the rock masses and discontinuities are identical
in all the longwall panels. The global model includes all the panels to those used in the local model (see Tables 1, 2, and 3). Three lines
within it. Small volumes of excavations in the panels have little in the border area (lines 1, 2, and 3) at the middle of the coal seam
impact on the border area. Therefore short excavation advances are chosen to monitor the stress changes in the border area. The
were omitted in the large-scale simulations. Instead the entire panel monitoring lines are marked in Fig. 1.
is mined out in every mining stage in order to save time. A global The model is first run to equilibrium before excavation. Mining in
co-ordinate system is used in the global model with the horizontal a panel is simulated by changing the mechanical properties and
X-axis lying approximately in the eastward direction, the horizontal constitutive law of the rock materials in the cave-in and fracture
Y-axis approximately in the northward direction, and the Z-axis still zones whose borders are outlined in the local model. This procedure
in the vertical direction (Figs. 1 and 8). starts from panel B4 and continues until panel D10. A similar simula-
The global model is 8230 m long, 4200 m wide, and 200 m thick tion approach was also employed by Tajdus (2009) and Whittles et al.
(see Fig. 1 and also Fig. 8). The vertical stress applied on the upper (2006).
boundary of the model is equal to ρgz, where ρ is the density of the The principal stresses along the three monitoring lines are
rock mass, g is the gravitational acceleration, and z is the thickness presented in Fig. 9. The major principal stress σ1 is in the vertical
direction in the near-field of the main tunnel, but it becomes horizon-
tal at a long distance from the tunnel. The stresses are zero on the
tunnel wall and increase to their maximum values at a depth of
Table 2
Mechanical and physical parameters of the rock mass. about 4 m. That implies that the first 4 m of the wall fails. The stresses
drop to the in-situ stress levels when the distance is more than 25 m
Rock mass Coal Sandstone Siltstone from the wall, which indicates that the stress state in the area at least
K (GPa) 2.33 6.8 5.3 25 m from the main tunnels is not influenced by the longwall mining.
G (GPa) 0.5 5.1 3.2
c (MPa) 1.06 2.8 1.9
ϕ (deg.) 29 49 37
σt (MPa) 0.033 0.28 0.42 Table 3
cr (MPa) 0.0106 0.28 0.19 Mechanical properties of the discontinuities used in the numerical models.
ϕr (deg.) 29 30 30 Discontinuity Cohesion (MPa) Friction (deg.) Tensile strength (MPa)
ep % 0.5 0.05 0.05
Bedding 0.3 30 0.03
K is the bulk modulus, G is the shear modulus, c is the cohesion, ϕ is the internal fric- Thin coal interlayer 0.3 30 0.1
tion, σt is the tensile strength, cr is the residual cohesion, ϕr is the residual friction and Bentonite interlayer 0.5 25 0.4
ep is the plastic strain parameter at the residual strength.
44 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47

12 m

Cave-in zone
C6
Barrier Pillar

C4
Barrier
Pillar Cave-in zone

Fig. 7. Failure zones in the barrier pillar.

barrier pillars is associated with the size of the fracture zone above
the panel stopes and in turn with the thickness of the overburden.
Three different overburdens were examined in the model. They are
Case I: 110 m rock + 250 m ice; Case II: 300 m rock + 100 m ice;
and Case III: 400 m rock. The vertical stress along the centre line of
the pillars is shown in Fig. 12 for the three cases of overburdens.
The vertical stress is zero at the side wall surface of the pillar and
increases to its maximum value at a certain depth in the wall. The

Fig. 6. Stress changes along the centre line of the barrier pillar in different stages of
mining: a) the vertical stress σZZ, b) the horizontal stress σXX parallel to the panel,
Y Z
and c) the horizontal stress σYY perpendicular to the panel.
X
82
30
m

Fig. 10 shows the horizontal stresses (σXX and σYY) in a horizontal

plane 2 m above the panel floor in a part of the barrier pillar region. 200 m
The local model has shown previously that stress concentrations Coal seam
occur in the barrier pillars. The global model shows that the stresses
in the border area on the western side of the main tunnels are not
4200 m
noticeably changed by the longwall mining.
Three panels, B4, B6, and C2, were mined out in sequence in the
model in order to examine how mining in panels B6 and C2 influences b
the stress state in the barrier pillar of B4. The horizontal stress
changes along the centreline of the barrier pillar of panel B4 during
mining in panels B4, B6 and C2 are presented in Fig. 11. The results
show that mining in panels B6 and C2 does not affect the stresses in
the pillar.

Coal seam
4. Discussion
100 m

4200 m
4.1. Load in the pillar Z

X
The barrier pillars protect the border area and the main tunnels of
the mine from disturbance by the longwall mining. The load in the Fig. 8. The global model: a) 3D view; b) cross section in the X–Z plane.
 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47 45

D10

D8
XX

D6
Stress
concentration

D4

D10

D8
YY

Stress
D6
concentration

D4
Fig. 9. Principal stresses in the border area after excavation of panels C4 to D10, along:
a) line 1, b) line 2, and c) line 3.

rock from the wall surface to the position of the maximum stress is in
the post-failure stage, resulting in a reduced stress in that region. In
the region beyond the position of the maximum stress, the rock/coal
has not failed and the stress decreases exponentially with distance
from the wall. It can be seen in Fig. 12 that the position of the maximum
stress moves toward the inside of the pillar with an increase in the rock
cover, which indicates that the failure zone becomes larger with a larger
overburden.
4.2. Stress changes in the pillars
With the conventional tributary area design method, the vertical
pillar load is equal to the weight of the rock prism overlying above
the pillar (Mark et al., 1999). Mining-induced cave-in results in stress
relaxation in the fracture zone above the mined-out panels. A
pressure arch is formed in the dome-shaped boundary of the fracture
zone unless the fracture zone has reached the ground surface. The
caved-in materials and the fracture zone are at least partially
Fig. 10. Horizontal stresses at the level of the coal seam after completion of the
longwall panel mining: a) σXX; b) σYY.
de-stressed. In this case, however, numerical modelling would be
able to provide more reasonable results on stress mitigation than
the tributary area method. For the sake of comparison, the maximum
vertical pillar stresses in the three cases of overburdens are presented
in Fig. 13. The x-axis of the diagram is the relative height of the
fracture zone compared to the overburden, hf / H, where hf is the
height of the fracture zone and H the thickness of the overburden.
In the case of the overburden composed of 110 m rock + 250 m ice
(case 1), the fracture zone reaches the ground surface. The maximum
vertical stress in the pillar is about 3.2 times the in-situ vertical stress.
For cases 2 and 3, which have more rock cover in the overburden, the
vertical pillar stress is about 2.3 times the in-situ stress. In these two
cases the absolute pillar stresses are a little higher than those in case
1, but the ratio to the in-situ stress is smaller.
46 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47

3.5 Case I

2.5 Case II
Case III

σ max
zz
1.5

σ in − situ 1

zz 0.5

0
0.2 0.4 0.6 0.8 1
hf
Relative height of the fracture zone
H
Fig. 13. The maximum vertical stress in the barrier pillar when the entire panel is
mined out for three cases of overburdens.

Fig. 11. Changes in horizontal stresses σXX and σYY in the barrier pillar after mining-out
of panels B4, B6, and C2.
The stresses in the barrier pillars fluctuate up and down following
periodic cave-in events behind the longwall face during mining. The
stress change is related to the critical advance length ls. Fig. 14
shows changes of the vertical and horizontal stresses in the pillars
for the three cases of overburdens. The vertical stress change is a re-
duction of 5–6.5 MPa depending on the critical advance length ls
and equivalently on the overburden. The horizontal stress change is
an increase of a small amount almost irrespective of the advance
length ls and the overburden.
5. Conclusion
The 3D local model was constructed for the purpose of examining
the stress changes in the barrier pillars during longwall mining in
Svea Nord coal mine. The model includes two panels and simulates
the mining process in a manner similar to reality. The simulations
show that the stresses in the barrier pillars and abutments fluctuate
up and down during mining because of periodic cave-in events occur-
ring immediately behind the longwall face. The vertical stress in the
pillars decreases by 5 MPa under an overburden of 400 m when the
mine-out stopes are filled by cave-in rock blocks. A failure zone of
about 12 m is found to have developed in the side wall of a barrier
pillar after completion of a panel mining at the study site. A signifi-
cantly large portion of the barrier pillar is still intact and is able to
provide protection to the border area at the western side of the
main access tunnels of the mine.
A mine-scale 3D global model was constructed to examine the
influence of the longwall mining on the stress state in the border
area. The model includes all the longwall panels. The simulation
results show that a disturbed zone of about 20 m has developed in
the wall of the main tunnels on the side of the border area after all
the longwall panels have been mined out. Beyond the first 20 m
from the main tunnels, the stress state in the border area remains
unchanged after completion of the longwall mining. With considering
barrier pillars with a width of 25 to 35 m within the border area along
the access tunnels of the mine, the rest of the border area will be
mineable in the future with room-and-pillar method.
The general significance of the study is that it proves that the failure
zone in the country rock surrounding an underground void of large vol-
ume is limited to a relatively small extent as long as the void is
backfilled by either caved materials or other types of loose materials.

Periodic caving distance- ls (m)

7 8 9 10 11 12 13 14 15
6
Stress reduction (MPa)

4
σzz

σxx
2 σyy

0 Case III Case II Case I

-2

Fig. 14. Stress changes in the barrier pillar after cave-in in the mined-out panel. De-
creases and increases in the compressive stress are shown with positive and negative
Fig. 12. Vertical stresses in the barrier pillar under three different overburdens. magnitudes, respectively.
 M. Shabanimashcool, C.C. Li / International Journal of Coal Geology 106 (2013) 39–47
References
Brady, B.H.G., Brown, E.T., 2002. Rock Mechanics: For Underground Mining, 2nd edn.
Springer, Netherlands.
Brown, E.T., 2007. Block caving geomechanics. Indooroopilly, Qld.: Julius Kruttschnitt
Mineral Research Centre.The University of Queensland, Brisbane.
Coggen, J., Gao, F., Stead, D., Elmo, D., 2012. Numerical modelling of the effects of weak
immediate roof lithology on coal mine roadway stability. International Journal of
Coal Geology 90–91, 100–109.
Duplanic, P., Brady, B.H., 1999. Characterization of caving mechanism by analysis of
seismicity and rock stress. Proceeding 9th International Congress Rock Mechanics,
Paris, France, pp. 1049–1053.
Gale, W.J., Blackwood, R.L., 1987. Stress distributions and rock failure around coal mine
roadways. International Journal of Rock Mechanics and Mining Science and
Geomechanics Abstracts 24 (3), 165–173.
Jiang, Y., Wang, H., Xue, S., Zhao, Y., Zhu, J., Pang, X., 2012. Assessment of mitigation of
coal bump risk during extraction of an island longwall panel. International Journal
of Coal Geology 95, 20–33.
Kelly, M., Gale, W., Hatherly, P., Balusu, R., Luo, X., 1996. New understanding of longwall
caving processes: results of microseismic monitoring and initial computational
modelling at Gordonstone mine. Proceeding of Geology in Longwall Mining Sympo-
sium, Coalfield Geology Council of New South Wales, Sydney, pp. 153–161.
Kelly, M., Luo, X., Craig, S., 2002. Integrating tools for longwall geomechanics. Interna-
tional Journal of Rock Mechanics and Mining Science 39 (5), 661–676.
Mark, C., Russel, F., Mark, C., 1999. Analysis of Longwall Tailgate Serviceability (ALTS):
A Chain Pillar Design Methodology for Australian Conditions: Proceeding of 2nd In-
ternational Workshop Coal Pillar Mechanics and Design1066-5552, pp. 33–49.
47
Mark, C., Gale, W., Oyler, D., Chen, J., 2007. Case history of the response of a longwall
entry subjected to concentrated horizontal stress. International Journal of Rock
Mechanics and Mining Science 44, 210–221.
Park, D.W., Gall, V., 1989. Supercomputer assisted three-dimensional finite element
analysis of a longwall panel. Proceeding of 30 th Rock Mechanics Symposium,
Rock Mechanics as a Guide for Efficient Utilization of Natural Resources, Morgan-
town. ISBN: 9061918715, pp. 133–140.
Peng, S.S., Chiange, H.S., 1984. Longwall Mining. John Wiley, New York.
Shabanimashcool, M., Li, C.C., 2011. Numerical study on disturbance of longwall mining
to stress state in the rock mass. Proceeding 12th International Congress in Rock
Mechanics, Beijing, pp. 1579–1585.
Shabanimashcool, M., Li, C.C., 2012. Numerical modelling of the longwall mining and
stability study of the gates in the Svea Nord coalmine. International Journal of
Rock Mechanics and Mining Science 51, 24–34.
Singh, R.N., Porter, I., Hematin, J., 2001. Finite element analysis of three-way roadway
junctions in longwall mining. International Journal of Coal Geology 45, 115–125.
Tajdus, K., 2009. New method for determining the elastic parameters of rock mass
layers in the region of underground mining influence. International Journal of
Rock Mechanics and Mining Science 46, 1296–1305.
Whittaker, B.N., 1974. An appraisal of strata control practice. Mining Engineering 134,
9–24.
Whittles, D.N., Lowndes, I.S., Kingman, S.W., Yates, C., Jobling, S., 2006. Influence of geo-
technical factors on gas flow experienced in a UK longwall coal mine panel. Inter-
national Journal of Rock Mechanics and Mining Science 43, 369–387.
Wilson, A.H., 1983. Stability of underground working in the soft rocks of the coal mea-
sures. International Journal of Mining Engineering 1 (2), 91–187.