Professional Documents
Culture Documents
2010-Alehossein - Poulsen - Stress Analysis of Longwall Top Coal Caving
2010-Alehossein - Poulsen - Stress Analysis of Longwall Top Coal Caving
International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms
a r t i c l e in f o a b s t r a c t
Article history: Longwall top coal caving (LTCC) is a relatively new method of mining thick coal seams that is currently
Received 3 November 2008 achieving high productivity and efficiency in application, particularly in China. The technique is similar
Received in revised form to traditional longwall mining in that a cutting head slices coal from the lower section of the coal seam
3 May 2009
onto a conveyor belt installed in front of the hydraulic support near the cutting face. In modern LTCC an
Accepted 17 July 2009
additional rear conveyor belt is located behind the support, to which the flow of the caved coal from the
Available online 8 August 2009
upper part of the seam can be controlled by a moveable flipper attached to the canopy of the support.
Keywords: The mining method relies on the fracturing of the top coal by the front abutment pressure to achieve
LTCC satisfactory caving into the rear conveyor.
Caving
This paper develops a yield and caveability criterion based on in situ conditions in the top coal in
Chinese caving index (CCI)
advance of the mining face (yield) and behind the supports (caveability). Yielding and caving effects are
Plasticity
Mohr–coulomb combined into one single number called caving number (CN), which is the multiplication result of
Numerical caving factor (CF) and yield factor (YF). Analytical derivations are based on in situ stress conditions,
Finite difference Mohr–Coulomb and/or Hoek–Brown rock failure criteria and a non-associated elastoplastic strain
Finite element method (FEM) softening material behaviour. The yield and caveability criteria are in agreement with results from both
Discrete element method numerical studies and mine data.
The caving number is normalised to mining conditions of a reference Chinese mine (LMX mine) and
is used to assess LTCC performance at fourteen other Chinese working longwalls that have had varying
success with the LTCC technology. The caving number is found to be in good agreement with
observations from working LTCC mines. As a predictive model, results of this analytical/numerical study
are useful to assess the potential success of caving in new LTCC operations and in different mining
conditions.
& 2009 Elsevier Ltd. All rights reserved.
1365-1609/$ - see front matter & 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijrmms.2009.07.004
ARTICLE IN PRESS
H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41 31
Fig. 1. A conceptual model of top coal caving method showing roof rock stress variations behind and ahead of face; after [5].
p p
2p = −p +3p
−p
B
B
p A 2p = 3p−p
A 3p
p
=1 K=0 =1 K=1
p p
p p
5p 3p
B B
A p A 3p
2p 2p
2p 2p
p p
p p
9p
B
1.4p
A0 B A 19p
2p 2p
2p 2p
p p
Fig. 2. Influence of vertical and horizontal pre-mining in situ stresses on hoop boundary stress at points A (x-axis where y ¼ 0) and B (y-axis where y ¼ 901) in circular and
elliptical excavations, identified by their two orthogonal diameter ratios, l, for various stress ratios K ¼ horizontal stress/vertical stress) by a 2D isotropic elastic analysis.
ARTICLE IN PRESS
H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41 33
Therefore, there are several hypotheses, which can be made for the that they may be assumed to remain as principal stresses in these
failure behaviour of the front top coal. Some of these possible formulations. Extension of the theory to include principal stress
mechanisms are schematically illustrated in Fig. 3. rotation is possible and does not require much effort.
Finite element (FE) modelling has been used to address
answers to such hypotheses assuming a range of stress ratio,
3. Yield factor and inelastic deformation
K ¼ (horizontal stress)/(vertical stress). Only results for the
unit in situ stress ratio, K ¼ 1, are shown in Figs. 4–6. Two
dimensional, eight-noded, isoparametric Mohr–Coulomb, Rock mass ahead of the support (chock) starts a failure process
elastoplastic (without softening) finite elements have been used when the major principal stress (e.g. the vertical stress) is equal or
in these analyses. Strain softening materials are normally mesh- greater than a failure function of the minor principal stress (e.g.
dependent and require special treatments [15]. the horizontal stress) and rock-mass strength properties (k, m, s,
These elastic and elastoplastic finite element analyses indicate n). With the convention that compressive stresses are positive, the
that all the assumptions and particularly either of the vertically or linear Mohr–Coulomb (MC) failure criterion may be written as
horizontally normal compression deformation modes of the front f ¼ s1 YFðKp s3 þ sc Þ ð1aÞ
top coal, as hypothesised as case (a) or (b) in Fig. 4, are possible
potential modes of rock mass deformation, which led us to when compressive failure is dominant, and/or,
develop a simple practical site independent model. In the f ¼ s3 YFðKa s1 sc Þ ð1bÞ
analytical formulation explained in the next sections we assume
that the horizontal and vertical in situ stresses are principal when tensile failure is dominant. YF is a yield factor which makes
stresses and each can become a major principal stress depending the above yield functions zero, i.e. f ¼ 0, when it is equal to unity.
on the value of the stress ratio K. Furthermore, their slight Therefore, YF must be either equal to or greater than unity for a
rotations, as a result of coal excavation behind it, are ignored so rock to fail, and so YF ¼ 1 is therefore the critical condition.
Premining, original, insitu, front & top coal line Deformed line due to excavation
Front
Front Top
top K<1 K>1
coal
coal block
block
Support Support Support
(Chock) (Chock) (Chock)
Fig. 3. Some of the possible hypotheses for deformation and failure mechanism of front top coal: (a) vertically normal compression, (b) horizontally normal compression,
and (c) normal and shear.
Fig. 4. Elastic finite element (FE) results of element deformation and displacement vectors around the excavation area.
ARTICLE IN PRESS
34 H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41
A rock brittleness coefficient k, which is analogous and The caving model developed in the following sections of this
identical in value to the more familiar soil mechanics terms of paper will not be altered by the method and diversity of
active (Ka) and passive (Kp) earth pressure coefficients (i.e. converting one criterion to another. Alternatively, equivalent MC
material stress limit ratios), now can represent either of the parameters, which can approximate the same H&B failure
factors Ka or Kp, depending on the dominant coal failure criterion, can also be derived from HB parameters by a linearisa-
mechanism in Eq. (1). sc is the uniaxial compressive strength tion process [16].
(UCS), st is the tensile strength (e.g. Brazilian tensile strength; Using direct horizontal and vertical stress notation is most
BTS), s1 and s3 are the major and minor principal stresses, which often more convenient than principal stress notation, especially
are limited by material strength. The active and passive coeffi- when the principal stress directions coincide with the vertical and
cients of rock failure are related to each other via the relation, horizontal axes:
Ka ¼ 1/Kp, where each represents the linear MC envelope slope of
s1 ¼ sh ; s3 ¼ sv ; sh 4sv : K41 ð8aÞ
one principal stress against the other. Due to the material strength
limitations, the magnitude of any negative stress cannot be
greater than the value of the rock uniaxial tensile strength (st) and s1 ¼ sv ; s3 ¼ sh ; sv 4sh : Ko1 ð8bÞ
the magnitude of any positive stress, at any negative confinement
stress, cannot be greater than the value of the rock uniaxial When the value of Eq. (1) or (3) is zero at the critical YF ¼ 1, the
compressive strength (sc), i.e.: rock no longer behaves elastically, and its behaviour is more
s3 min ¼ st for s1 40; s1 max ¼ sc for s3 o0 ð2Þ inelastic and brittle rather than plastic and ductile. However, here
we loosely use the same classical terminology of plasticity in this
context [17]. The potential function corresponding to the yield
In terms of the non-linear Hoek and Brown (HB) criterion, the f criterion (1) and (3) are simply
function in Eq. (1a) becomes n
n s3
s3 g ¼ s1 YF s3 þ sc m þs ð9Þ
f ¼ s1 YF s3 þ sc m þs ð3Þ sc
sc
The power factor n is normally o1. For example, n ¼ 0.5 for most g ¼ s1 YFðKp s3 þ sc Þ ð10Þ
hard rocks, and n ¼ 0.65 for a few Australian coal seams. Notice
that criterion (3) transforms to the same linear MC criterion in Eq. which dictates a general non-associated flow rule of plasticity. The
(1a) when n ¼ 1, s ¼ 1 and m ¼ Kp1. The parameter s depends on parameter m*rm corresponds to the rock dilation angle crf. A
the rock fracturing condition, with a maximum value of unity for flow rule is either associated or non-associated. An associated flow
the intact rock and the parameter m identifies the rock strength rule corresponds to the case where both friction and dilation
for a particular rock type. It is similar to the passive strength angles are equal, i.e. all cases of c ¼ f. A non-associated flow rule
coefficient (Kp), which activates rock frictional behaviour once corresponds to the general cases of caf. Most soils and rocks
cohesive bonds have been broken. In general the following follow the non-associated flow rule of plasticity. Metals, or almost
relations hold true for the two methods: zero friction solids, correspond to a specific case of associated flow
m ¼ sKp ðKp Þðn1Þ=n ð4aÞ rule where c ¼ fE0. The value of the dilation angle is normally
o23 of the friction angle (f) for most rocks. Applying the theory of
scMC ¼ scHB sn ð4bÞ plasticity of cohesive frictional materials [17] we find the
following stress-strain relations for the vertical and horizontal
There is no comparable UCS strength reduction factor s in the MC stresses.
criterion to be applicable to fractured rock masses as well. ! ! !
Ds1 A B e1 A B u1 =L1
Because the rock strength parameter (sc) has two different ¼ ¼ ð11Þ
interpretations in the two criteria, it is better to distinguish the Ds3 C D e3 C D u3 =L3
two strength terms by the subscripts MC and HB, particularly,
when they are meant to be different, as shown in Eq. (4b). The where e1 and e3 are the principal strains and u1 and u3 are the
friction angle in the HB criterion is not constant and depends on corresponding deformations over the principal lengths L1 and L3.
the magnitudes of the principal stresses: The elastoplastic stress-strain coefficients in the rigidity matrix
8" #1 9
< = 2 T 3
1 2 s1 s3 ð1nÞ=n @g @f
fHB ¼ sin 1þ ð5Þ 6 ½De ½De 7
: mn sc ; A B 6 @s @s 7
¼ 6½De T 7 ð12Þ
C D 4 @f @g 5
½De þH
The cohesion intercept is a more complex function of the @s @s
HB parameters. Letting R0 ¼ (s1s3)/2, which is the shear
stress at the point of zero normal stress, the CHB can be calcu- are stress dependent as shown in Eq. (12). The rigidity matrix
lated by depends on the elastic matrix De and the normal vectors (qg/qs
s ðs mR0 =sc Þ and qf/qs) to the plastic yield (f) and potential (g) functions [17].
CHB ¼ (c )! ð6Þ
ð1nÞ=n 1 De is the elastic matrix, i.e. D11e ¼ D22e ¼ l+2G and
1 2 2R0
2 cos sin 1 þ mn sc D12e ¼ D21e ¼ l, in which l ¼ nE/(1+n)/(12n) is Lame constant,
G ¼ E/(1+n)/2 is shear modulus, E is Young’s modulus and n is
Poisson’s ratio. H is the hardening (H40) or softening (Ho0)
The values of the fHB and CHB above are not identical to those
parameter, which is assumed to be equal to E/100, unless
obtained by the MC model expressed in the following equations.
otherwise stated. As mentioned earlier, for softening material
Typically, fMC4fHB and CMCoCHB:
modelling, finite element solutions become mesh dependent and
sc st 1 pffiffiffiffiffiffiffiffiffiffi hence require special regularisation techniques to overcome the
sinðfÞMC ¼ ;C ¼ ss ð7Þ
sc þ st MC 2 c t mesh dependency problem [15].
ARTICLE IN PRESS
H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41 35
4. In situ stress and solutions in Fig. 1, caving at the rear of the support occurs in an LTCC method
if and only if the rock ahead of the face has already cracked, failed or
Excluding any possible surcharge loadings, the underground yielded. When this happens, the stresses in the rock satisfy the yield
vertical stress before any mining is normally assumed to be due to criterion (1) or (3) causing inelastic deformations governed by the
the gravity force. In situ pre-mining horizontal stress is usually stress strain relations (11). In other words, the initial state of stress,
related to vertical stress by a line having a slope K and a constant q0: sh0 and sv0, have been disturbed to new stresses sh and sv as a
sv0 ¼ gz ð13Þ result of the induced excavation stresses Dsh and Dsv. Therefore,
we can write:
Fig. 5. Elastoplastic finite element results of element deformation and displacement vectors around the excavation area.
Fig. 6. Elastoplastic finite element results of displacement vectors around the excavation area.
ARTICLE IN PRESS
36 H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41
In terms of principal stresses, these can be written as and for the HB model these functions are
s1 ¼ s10 þ Ds1 ð17Þ A ðYFÞC B ðYFÞD
f ðu1 ; u3 Þ ¼ s10 ðY:F:Þs30 þ u1 þ u3
L1 L3
s3 ¼ s30 þ Ds3 ð18Þ ðYFÞsc f1n
¼0 ð23Þ
Using Eqs. (15)–(18) in Eqs. (1) and (3), we obtain the following
separate system of equations in terms of the original in situ stresses u1 u3
(13) and (14): gðu1 ; u3 Þ ¼ ð1 þ m ng1n1 Þ þ ¼0 ð24Þ
L1 L3
s1 s3
YF ¼ orYF ¼ ðMCÞ ð19Þ where
Kp s3 þ sc Ka s1 st
s30 þ ðCu1 =L1 Þ þ ðDu3 =L3 Þ
f1 ðu1 ; u3 Þ ¼ m þs ð25Þ
s10 þ Ds1 sc
YF ¼ n ðHBÞ ð20Þ
s þ Ds3
s30 þ Ds3 þ sc m 30 þs s30 þ ðCu1 =L1 Þ þ ðDu3 =L3 Þ
sc g1 ðu1 ; u3 Þ ¼ m þs ð26Þ
sc
For example, a top coal will yield ahead of the face (Fig. 1),
if the following properties and stress conditions are assumed 1.8
for the coal seam, which is possible in practice: n ¼ 0.65, sc ¼ 20
MPa, s ¼ 0.25,st ¼ 1 MPa, sv0 ¼ (150 m)(0.025 MN/m3) ¼ 3.75 MPa,
Dsv ¼ 2sv0 ¼ 7.5 MPa, sh ¼ Kasv ¼ (1 MPa/25 MPa) (3.75 MPa+7.5
1.4
þ ð22Þ
L1 L3 1.5
2
0.5
1.5 0
Relative yield factor
0 1 2 3 4
In-situ stress ratio (K)
0.5
v v v
0 h bγc h
b
0 100 200 300 400 500
Depth from free surface (m)
a = c + h tan()
Fig. 7. Effect of rock cover depth on the coal seam relative vertical displacement at
yield. sc ¼ 25 MPa, s ¼ 1, K ¼ 1, g ¼ 0.025 MN/m3, gc ¼ 0.013 MN/m3, Kp ¼ 15, Fig. 10. Caving condition for a rectangular 2D block sandwiched between two
K*p ¼ 1.7, a ¼ 6 m, b ¼ 8 m. vertical blocks.
ARTICLE IN PRESS
H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41 37
in which
b
x A ðYFÞð1 þ mnf1n1 ÞC
F1 ¼ ð30Þ
L1
d
B ðYFÞð1 þ mnf1n1 ÞD
a F3 ¼ ð31Þ
L3
y
Fig. 11. Caving condition for a prismatic 3D block sandwiched between four 1 þ m ng1n1 m2 nðn 1ÞCg1n2
G1 ¼ þ u1 ð32Þ
vertical blocks. L1 L21 sc
JOB TITLE : Top coal caving - Chock loading analysis - Dpth 200m (*101)
LEGEND
3.000
6-Apr-04 11:09
cycle 24050
time 2.269E+00 sec
2.500
no. zones : total 6650
at yield surface (*) 284
yielded in past (X) 1114
tensile failure (o) 96 2.000
block plot
1.500
1.000
0.500
0.000
CSIRO
E&M
3.850 3.900 3.950 4.000 4.050 4.100 4.150 4.200 4.250
(*102)
Fig. 12. Numerical model of top coal caving at GM mine. Plot shows extent of yield for base case parameters.
ARTICLE IN PRESS
38 H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41
1 m2 nðn 1ÞDg1n2 however, it reduces with increasing coal strength UCS (Fig. 8) and
G3 ¼ þ u1 ð33Þ
L3 L1 L3 sc it is a periodic, non-monotonic function of the in situ stress ratio K
(Fig. 9).
Figs. 7–9 demonstrate the application of the above equations. The
effects of depth of cover (z), coal compressive strength (UCS) and
in situ stress ratio (K) on the yield factor have been calculated. YF 5. Coal caving factor (CF) and caving number (CN)
has been normalised with a standard reference case, i.e. the case
where the relative yield factor is unity. The reference case is a coal Gravity force is the main driver for coal caving both for a small
seam, in which z ¼ 150 m, sc ¼ 25 MPa, s ¼ 1, K ¼ 1, rock mass and large block of coal. For example, consider caving stability of a
unit weight, g ¼ 0.025 MN/m3, coal seam unit weight, 2D rectangular block of a coal shown in Fig. 10. The coal, which is
gc ¼ 0.013 MN/m3, Kp ¼ 15, Kp ¼ 1:7, a ¼ 6 m, b ¼ 8 m. In each sandwiched between two other vertical blocks from the sides, is
figure all parameters are the same as the reference case above, under a top vertical stress (sv) and its own gravity stress (bgc).
except one parameter, the independent variable on the x-axis. In Because of the lateral confinement, there are, first cohesive and
Fig. 7, the variable parameter is the depth of rock, z, in Fig. 8, the then frictional, lateral resistance forces against the gravity forces.
changing parameter is the coal UCS, or sc, and it is the in situ Caving occurs when the ratio of the vertical gravity force exceeds
stress ratio K in Fig. 9. The results are consistent with both the the shear resistance force. This ratio is defined as the caving factor
field caving and the numerical results discussed in Sections 4 and (CF). For this particular simple 2D problem we can write
5. In summary they indicate that, for the same rock properties,
yielding or caving improves with increasing depth z (Fig. 7),
a sv þ bgc
CF ¼ a ð34Þ
b 2c þ 2sh tanðfÞ þ b st
Table 2
Base case parameters in numerical study. Note that a and b can accept a range of values from cleat
dimensions to the whole coal block. We suggest we should check
Depth of mining (m) 200 both small and large dimensions to see which is more dominant.
Horizontal to vertical stress ratio 1 In Eq. (34), a and b are the width and height of the block, c is coal
Coal strength (MPa) 8
cohesion, f is coal friction angle, gc is coal unit weight, st is coal
tensile strength and sv and sh are the vertical and horizontal
JOB TITLE : Top coal caving simulation, Depth = 300m, K = 1.0, Base coal strength (*101)
1.750
LEGEND
25-Sep-01 9:03
cycle 2248171
time 4.121E+02 sec
1.250
density
1.330E+03
2.545E+03
2.550E+03 0.750
2.555E+03
2.560E+03
2.565E+03
block plot
Support Element Locations 0.250
principal stresses
minimum = -2.052E+07
maximum = 3.766E+06
-0.250
0 1E 8
-0.750
CSIRO
E&M
3.025 3.075 3.125 3.175 3.225 3.275 3.325
(*102)
stresses applied to the block. The equilibrium equation (34) can be Yield in the top coal was observed to a greater or lesser extent
easily extended to Eq. (35) for general three-dimensional blocks in all the numerical analyses and extended from the cave free
shown in Fig. 11: surface, over the support (chock) to a distance in advance of the
0 1 face. The measure used to compare the analyses is the extent of
ad B
B sv þ bgc C
C yield detected on a horizontal line at a 2 m distance above the
CF ¼ B C chock. Along this horizontal line, the distance to which yield
b @ cx cy ad A
2d þ 2a 2 þ 2dshx tanðfx Þ þ 2ashy tanðfy Þ þ st
2
sin ðaÞ sin ðbÞ b extends in advance of the face is measured and normalised to the
ð35Þ base case as defined in Table 2.
where a and b are the inclination angles towards the x and y axes,
as shown in Fig. 11. Several parallelograms (instead of rectangles)
Yield in Top Coal vs Depth
have formed as a result of the inclined position of the block. In Eq.
1.4
(35) the x and y indices refer to the projected x and y planes with
these axes as their normal lines. 1.3 yield
We now combine the effects of the caving and yield factors into
one factor called caving number (CN): 1.2
CN ¼ 100ðCFÞðYFÞ ð36Þ 1.1
Yield
0.9
0.8
6. Numerical analysis—distinct/discrete element method
0.7
Numerical analysis provides the opportunity to verify the
calculations for the yield function for the particular geometric 0.6
layout of a top coal caving mine in addition to considering
0.5
additional variables that may influence the yield of top coal. For
this purpose, a representative top coal mining layout has been 0.4
assessed with the numerical code UDEC by the Itasca Consulting 100 150 200 250 300
group [18], Fig. 12. The depth of cover, thickness of top coal, angle Depth (m)
of break etc are based on conditions at the GM mine and
presented in [19] who undertook a physical model of the LTCC Fig. 15. Variation in yield in the top coal with increasing depth, normalised to
200 m.
process.
Fig. 14. PFC2D discrete model of LTCC. A model showing the extent of vertical cracking and the relaxation of the roof behind the face line—after [7].
ARTICLE IN PRESS
40 H. Alehossein, B.A. Poulsen / International Journal of Rock Mechanics & Mining Sciences 47 (2010) 30–41
Yield in Top Coal vs UCS of Coal Mixed distinct–discrete element modelling allows for rock to
1.6 be modelled after it has broken as the modelled rock mass is made
Yield up of smaller individual elements bonded together. Fig. 13 shows a
1.4 UDEC model illustrating the formation of fractures ahead of the
mining face and coal breakage over the top of the supports
1.2 followed by caving at the rear.
The model in Fig. 14 shows a different discrete element code
using spherical elements and is able to accurately simulate the
1
flow of the fractured coal in the goaf due to LTCC. The model
automatically extracts the coal that falls within 2 m of the rear of
Yield
0.8
the support to simulate the rear face conveyor. (The LTCC support
is not shown in this model). Depth of mining, horizontal to
0.6 vertical stress ratio and UCS strength of the top coal were varied
from the base values with the results presented in Figs. 15–17. The
0.4 patterns of the resulting functions are in good agreement with the
analytical results presented in Figs. 7–9. The numerical results of
0.2 Figs. 15–17 generally confirm the change patterns of the analytical
predictions of the yield function presented in Figs. 7–9. In
0 summary the numerical modelling results indicate that yield
3 4 5 6 7 8 9 10 11 12 13 14 15 increases with increasing depth (Fig. 15), decreases with
UCS (MPa) increasing coal strength (Fig. 16) and is non-monotonic with
variation in K (Fig. 17), as picked up by the analytical model of Fig.
Fig. 16. Variation in yield in the top coal with increasing coal strength, normalised 9. The results for K effects in Fig. 17, however, needs more careful
to strength of 8 MPa.
considerations, as it also represent non-uniform and complex
responses from a collection of elements forming the top coal block
in the numerical model, in which each element can be under any
active, passive, compression, tensile, shear failure mode. In other
words, the yield function here is a strong function of not only the
Yield of Top Coal vs Horizontal to Vertical Stress Ratio
stress ratio K, but also the element size and the type of the
1.3
dominant failure mechanism explained by Eqs. (1).
Yield
1.2
Table 3
Qualitative calculations of the factors CF, YF and CN, with respect to those for the LMX mine.
Seam and coal mine Coal UCS (MPa) Cover depth (m) CF YF CN Cavability classification result (CCI)