You are on page 1of 15

Computers and Geotechnics 36 (2009) 1142–1156

Contents lists available at ScienceDirect

Computers and Geotechnics


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

A numerical modeling approach for assessment of progressive caving of strata


and performance of hydraulic powered support in longwall workings
G.S.P. Singh a,*, U.K. Singh b
a
Department of Mining Engineering, Institute of Technology, Banaras Hindu University, Varanasi 221 005, India
b
Department of Mining Engineering, ISM University, Dhanbad 826 004, India

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

Article history: A numerical modeling approach is proposed for predicting the progressive caving behaviour of strata and
Received 21 January 2009 performance of powered roof support in a given geo-mining and strata condition. The study suggests that
Received in revised form 17 April 2009 face convergence slope of 75 mm/m of face advance may be considered as the maximum permissible
Accepted 1 May 2009
value for selection of optimal capacity support for safe working in a given strata condition. The study also
Available online 23 May 2009
reveals that en masse caving of strata is responsible for dynamic loading leading to collapse of supports if
they are not designed to discharge the fluid at the desired rate.
Keywords:
Ó 2009 Elsevier Ltd. All rights reserved.
Longwall
Progressive caving
Support load
Abutment stress
Peak face convergence slope
Numerical modeling

1. Introduction For estimation of support capacity, theoretical models [18,19]


based on soil mechanics approach have been suggested. A number
Strata control in longwall mining has been a grey area of re- of empirical models [2,7,11,15,16,20–38] have also been proposed
search since its introduction in underground coal mining industry for this purpose. A ground response curve devised on the basis of
worldwide. A reliable prediction of the caving behaviour of strata data obtained from automatic data acquisition system for leg pres-
and its interaction with the roof support helps in selection of sus- sure monitoring, leg stiffness test and routine underground obser-
tainable mining parameters and rational capacity of supports. It is vations has also been proposed [39] for assessment of support
pre-requisite for developing a reliable support selection tool essen- performance at a longwall face. The curve has been subsequently
tial for successful planning of longwall working in a given geo-min- used for projecting the support requirement under a different
ing condition. A number of approaches have been developed to geo-mining condition at the same mine. Although application of
address this important subject of applied rock mechanics. numerical modeling technique for strata control in longwall work-
Theoretical models for prediction of main fall and periodic cav- ings is not a new topic of research, the study has not been done to
ing span are based on plate–beam theory [1] and bending moment assess the strata–support interaction with progressive face ad-
approach [2]. A number of empirical models have been developed vance in most of the cases. Most of these studies have been done
on the basis of either certain concept or some field experience to using elastic analysis where simulation of face advance bears no
assess the caving behaviour of strata. Some of these approaches importance. Some of the non-linear modeling studies [40–44] have
suggest roof classifications for qualitative assessment of caving been done with some different objectives and hence, follow a dif-
behaviour [3–7]. Some other models propose quantitative relation ferent modeling scheme. The discrete element model analysis re-
to predict the span of main fall [8–13]. Similar relations have been ported in [45] does not consider the progressive face advance
proposed by various researchers to estimate the span of periodic and the model has been studied using arbitrarily selected block
caving [11,14–16]. A few models give both the options of qualita- size of strata.
tive assessment of roof caving and the quantitative assessment of Experience shows that although the existing approaches offer
caving span [15–17]. important contributions towards understanding strata–support
interactions, but do not provide effective means of support specifi-
cation [46]. Their applicability in Indian geo-mining condition is
* Corresponding author. Tel.: +91 9451828337; fax: +91 5422369434. limited and suffers from considerable error resulting in inconsis-
E-mail address: gspsingh@yahoo.com (G.S.P. Singh). tent results [15,47,48]. There are numerous examples where

0266-352X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2009.05.001
G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156 1143

improper planning has resulted in wrong site selection and under- computed as a direct function of thickness of the rock beds in
rated capacity of supports giving rise to a lot of problems like the layer. The modulus of the layer is computed as
threats of face instability, frequent face stoppages, irregular face t t1 t2 t3
advance, poor face productivity and collapse of workings in the ex- ¼ þ þ ð1Þ
E E1 E2 E3
treme cases. In view of the above, a fresh attempt is required to de-
velop a new approach by which gap of uncertainty could be where t1, t2, t3, . . . are the thickness of individual rock beds 1, 2, 3,
reduced. If the problem is predicted by suitable pre-investigation, . . .; E1, E2, E3, ... the Young’s modulus of individual rock beds 1, 2,
it can be dealt in a well planned and confident manner. 3, . . .; t the thickness of the layer (t = t1 + t2 + t3 +   ); and E is the
This paper describes a numerical modeling approach to assess equivalent Young’s modulus of the layer comprising rock beds 1,
the progressive caving of strata and performance of powered roof 2, 3, . . . .
support of a given capacity deployed in a given geo-mining condi- In absence of laboratory measured values of elastic moduli for
tion. The various features of the numerical model are described. different rock types from various longwall panels studied in this
The performance of the numerical model is compared with field work, a uniform criterion for estimation of Young’s moduli has
observations of 23 case studies taken from different coal fields in been adopted. For coal, a single value of 2 GPa has been assumed
India to assess the acceptability of the model for predicting the for all cases, irrespective of its compressive strength. The modulus
caving behaviour of strata and support performance in a given stra- of other types of coal measure rocks are estimated using [51].
ta condition.
E ¼ 0:31rc ð2Þ
2. Estimation of rock parameters where E is the Young’s modulus, GPa; and rc is the uni-axial com-
pressive strength, MPa.
Estimation of rock parameters consists of identification of dif- The phrase ‘immediate roof’ has been used in this work for a
ferent layers in the overlying strata and their properties, estima- layer just above the coal seam to be extracted. It is amenable to
tion of rock mass properties and in situ stress field. cave after advance of the support if it is weak and thin or suffi-
ciently laminated. If it is strong and difficult to cave, it overhangs
2.1. Identification of layers in roof and their properties to a considerable length in the goaf and plays a vital role on loading
over the support at the face. Here, we consider that the ‘immediate
The major mechanism of caving in sedimentary rock formation roof’ and the ‘immediate caving roof’ are two different entities.
is controlled by dominant planes of parting in the strata. The ap- Immediate caving roof may be same as the immediate roof in the
proach of Roof Separation Index (RSI) [48,50] is used to identify first case, i.e. when the roof stratum is weak, thin and laminated.
the prominent layers of roof separation in the caving zone. In order However, in the latter case, when the immediate strata is thick
to identify the different layers in roof, different rock beds are iden- and competent, the immediate caving roof may be only a portion
tified as per lithology of the rock formation above the coal seam. of the immediate roof which is caving in the goaf due to local spall-
The RQD, average core length and strength values are assigned to ing during the progressive face advance. Hence, in the difficult cav-
these rock beds on the basis of detailed logging results of the core ing condition, the thickness of immediate caving roof may even be
samples and laboratory test. zero. In some cases, where a thick coal seam is mined leaving a
Once the roof separation layers in the overlying strata are iden- portion of coal in the roof, the thickness of coal left in the roof
tified, the mechanical properties of different layers in roof existing may act as ‘immediate caving roof’ due to its easy caving character-
within the zone of roof caving for a given geo-mining condition are istics. The ‘main roof’ lies just above the immediate roof. The fail-
assessed. The values of uni-axial compressive strength (rc), tensile ure of this layer in large span, conceptually, causes loading of
strength (rt), RQD and Young’s modulus for the layers are support at the face resulting in face convergence as well as high
computed. The weighted strength and RQD values of a layer are abutment stress during the periods of main fall and periodic caving

Fig. 1. Roof zones above a longwall face illustrating idealized block distributions after caving [11]: he is extraction height, h1 is immediate roof thickness, hc is caving height
and d is deflection of main roof (1).
1144 G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156

of the strata. The main roof may consist of either a single layer or a field measurement using any established method. Unfortunately,
series of layers which may separate from each other during pro- in situ stress measured data for Indian coalfields are very limited.
gressive caving. The ‘overburden’ strata lie above the main roof. Therefore, some theoretical approach is required for a proper esti-
It may consist of a single layer or a number of layers. However, mation of the reasonable values of the in situ stresses. The vertical
we do not simulate parting of layers in the overburden when it lies stress is generally estimated using
at a height more than 15 times height of extraction. It is considered
rv ¼ cz ð3Þ
that such strata should not play a direct role on loading of the sup-
port at the face. The layers in the overburden strata gradually settle where rv is the vertical stress, MPa; c the unit weight of the overly-
over the caved goaf. The schematic diagram of various layers in ing rock, MPa/m; and z is the depth below surface, m.
roof, as discussed above is depicted in Fig. 1. The average horizontal stress (rh) estimated by the elasto-static
thermal stress model given by Eq. (4) [75] which is reported to fit
2.2. Estimation of rock mass properties reasonably with the in situ stress measurement data obtained in
many countries worldwide [76] has been used to generate the hor-
Proper estimation of rock mass strength and modulus values is izontal stress field in the model
very important for a meaningful numerical modeling study of cav- m bEG
ing behaviour and support requirements. It deserves an extensive rh ¼ cH þ ðH þ 1000Þ ð4Þ
1m 1m
literature review and personal experience before proposing a final
approach to address the subject properly. The review of literatures where b is the thermal expansion coefficient of rock: 8  106/°C for
for conversion of intact rock properties to rock mass properties sandstone, 30  106/°C for coal; G the average value of geothermal
shows a widely varying number of approaches. In most of the gradient for Indian coal fields: 0.03 °C/m; H the depth from surface,
cases, Hoek–Brown failure criteria [52] and Geological Strength m; E the Young’s modulus of rock, MPa; c the unit weight of rock,
Index have been used to estimate the rock mass strength [53,54]. MN/m3; and m is the Poisson’s ratio of rock.
A few others have used rock mass classification approach [52,55–
59]. In India, the empirical failure criteria [60] are mostly used in 3. Numerical simulation
combination with Bieniawski RMR [61]. A proper use of these ap-
proaches can only be ensured if the detailed information about A two dimensional plain strain plasticity model has been devel-
condition of joints is either available on record or the core samples oped to assess the progressive caving behaviour of strata and sup-
are available for generating necessary information. In cases, where port performance in longwall workings. Such a model has been
such geotechnical data are not available, most of the estimations opted mainly for two reasons: (i) it is valid for simulating the con-
are done on the basis of approximations of joint roughness, aper- dition at centre of a longwall face till the ratio of face length to the
ture, etc., based on personal judgment. These approaches involve face advance is 2, however for practical purposes; it can simulate
a lot of assumptions which increase the possibility of error in the condition up to a face advance till the ratio is 1.5 without
end results. A number of other researchers have used a direct ap- any considerable error in the end modeling results and (ii) it en-
proach based on scale effect, joint frequency, RQD, etc., to estimate ables simulation of a field representative progressive face advance
the in situ rock mass strength from intact rock strength (Table 1). due to lesser run time requirement enabling a more realistic and
Based on the common observations of literature review as cited minute study of progressive failure and caving of strata as well
above and personal judgment, a scale effect of 50% combined with as the loading behaviour of powered roof support deployed at
the influence of RQD has been adopted to estimate the rock mass the face as the face advances through critical mining cycles involv-
strength. The values of bulk moduli, shear moduli and cohesion ing periods of roof caving including local fall, main fall and a few
are calculated using the standard relations [73]. In absence of large periodic caving. Finite difference based numerical modeling soft-
scale tested value of friction angle for sandstone, coal and other ware FLAC Version 4.0 [49] has been used for this work.
coal measure rocks, friction angle of 25° is taken for coal and 40° Necessary FISH modules (user defined functions) have been
for other rock types in all cases. The angle of dilation for coal is developed and implemented successfully to simulate the progres-
taken as 2° and that for rock strata is considered as 5°, uniformly sive failure and caving of strata, and assessment of strata and sup-
in all cases [48]. These values are also in accordance with port behaviour parameters in most useful terms like load on the
suggestion given by [74]. support, face convergence, front abutment stress and vertical
strain in coal at the face with progressive face advance. The mod-
2.3. Estimation of in situ stress field eling scheme comprises schemes for model formulation, stable
solution for every stage of progressive face advance, simulation
In situ stress field plays a very important role in caving behav- of mining sequence, i.e. progressive face and support advance,
iour of strata. The best way to estimate in situ stress is to conduct and simulation of failure and caving of strata with the progressive
face advance.
Table 1
Value of reduction factor used by various researchers for conversion of laboratory 3.1. Modeling of powered roof support
strength to rock mass strength.

Sl. no. Authors reference Reduction factor to convert Remarks A realistic simulation of the complex loading behaviour of pow-
laboratory strength ered roof support is very important. The powered roof support
to in situ strength modeled in this two dimensional plain strain model (Fig. 2) com-
(i) [62] 0.25 prises of a canopy, a base and four support members interacting
(ii) [63] 0.14–1 According to RQD between the canopy and base against roof and floor at the coal face.
(iii) [64] 0.4
These four support members form two groups simulating the front
(iv) [65] 0.05–0.5 According to rock type
(v) [66] 0.2 and rear legs. The two support members of each group act at an an-
(vi) [67] 0.25 gle of 121° and 59°, respectively from the horizontal position so
(vii) [68] 0.2 Scale effect for coal that the modeled support is capable of providing a maximum lat-
(viii) [69–71] 0.1–0.5 General scale effect
eral load resistance of 30% of its yield load capacity in the worst
(ix) [72] 0.35
loading condition of strata. In case of purely vertical loading, the
G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156 1145

lateral load provided by the modeled support remains zero, as the based on the 0.3 coefficient of friction between the canopy and the
lateral load components are counterbalanced by the two members coal measure roof strata [77].
of each group. This modeling capability of lateral loading up to 30% The stiffness of powered roof support is estimated using Eq. (5)
of the yield load capacity has been introduced in the present work considering the bulk modulus of water–mineral oil mixture as

Fig. 2. A window showing the schematic diagram of modeled powered roof support simulating the vertical and lateral loading of support.

Fig. 3. General grid density in different zones of the model. Mining in coal seam is simulated along X direction.
1146 G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156

1.94 GPa. The fluid column height in the hydraulic leg for a given main roof layer 1 and main roof layer 2, as explained above. These
extraction height is estimated using the approach suggested by layers comprise the caving zone of the overlying strata. The over-
[78]. The theoretically calculated value of support stiffness has also burden rock up to the surface lies above the top layer. The side
been verified by laboratory tests for hydraulic legs of different boundary of the model has been selected in such a way that the
capacities at Rock Mechanics and Strata Control Laboratory in the boundary stress of the model remains unaffected due to extraction
Department of Mining Engineering, ISM University Dhanbad of the complete mining zone. Accordingly, for a planned extraction
of length ‘a’ (mining zone) in the panel, the left and right bound-
KwA
ks ¼ n ð5Þ aries are at ‘2.5a’ distance from the centre of mining zone. The ele-
h
ment size along X direction in the mining zone is 1 m. In Y
where ks is the support stiffness, N/m; n the number of legs working direction, the size of elements in the coal face has been kept equal,
in parallel in between canopy and base of the support; A the area of while the element size of rock layers has been increased gradually
cross section of the leg piston, m2; h the fluid column height in the by 10%. Sufficient care has been taken so that the size of the ele-
hydraulic leg, m; and Kw is the bulk modulus of the water–oil mix- ments do not vary significantly from one model to another as the
ture, N/m2. results have been found to be dependent on element size. A sche-
The structural element ‘prop’ has been used to model the ‘sup- matic layout of one of the model showing the grid density in its dif-
port’ members, which interact between canopy and base set ferent zones is shown in Fig. 3. Origin (0, 0) has been taken at the
against roof and floor at the coal face. The yielding characteristics coal roof and at the extraction starting point in all the models. A
of the modeled system of support (the effect of pressure relief typical input parameter for Panel A4 at Moonidih mine in Jharia
valve) has been modeled by incorporating a load closure table and Panel 21 at PVK 5 mine in Singereni coalfields in India are gi-
where the further development in load from its setting load upon ven in Tables 2 and 3, respectively.
installation of the support at the face is governed by the linear rela- The different layers in roof are liable to slip and separate from
tion governed by its stiffness till it achieves the yield load value. each other due to presence of weak parting planes. These major
Once the support achieves the yield load value, the stiffness be- parting planes have been simulated by interface elements imple-
comes zero and the load experienced by the support does not in- mented in FLAC. These elements are characterized by Coulomb
crease further but it continues to yield at the constant yield load sliding and separation in tension. It facilitates separation and slip-
value till the system becomes stable. ping of the nodes joining the adjacent grid-points, which forms the
interface to simulate the load-deformation behaviour of disconti-
3.2. Model formulation nuities. An interface is assigned pre-defined normal and shear stiff-
ness, joint friction, joint cohesion and joint tension to simulate the
Formulation of the two dimensional plain strain plasticity mod- representative behaviour of the parting plane. The boundary condi-
el consists of construction of model geometry, defining of the con- tions have been applied in such a way that they represent the ac-
stitutive relation and material properties for rock mass and parting tual field conditions. The floor boundary has been fixed in both X
planes, in situ stress initialization and assignment of boundary con- and Y directions. The sides are fixed in X direction only (roller
ditions of the model. In addition to this, some monitoring parame- boundary) till the virgin model gets converged after initialization
ters like history of unbalanced force for a given point have also of in situ stress. The side boundaries are later on locked after initial-
been introduced to check the convergence requirement of the vir- izing the X and Y displacements and the velocities to zero, simulat-
gin model after its solution. ing the clamped side boundary in the virgin model. These
The model geometry of a longwall panel consists of 50 m floor boundary conditions allow vertical and shear displacements in
rock overlain by a coal seam. The coal seam is overlain by roof con- the model without affecting its external geometry. The X and Y
sisting of two to three layers. They are named as immediate roof, displacement and velocity values in the virgin model are initialized

Table 2
Numerical model input data for Moonidih A4 panel.

Thickness Density Young’s mod. Shear mod. Bulk mod. rcm rtm Cohesion Friction Dilation
(m) (kg/m3) (GPa) (GPa) (GPa) (MPa) (MPa) (MPa) (°) (°)
Overburden 370.35 2144 12.77 5.11 8.51 13.18 1.03 3.08 40 5
Main roof
Layer 2 10.17 2003 6.04 2.42 4.03 9.52 0.86 2.22 40 5
Layer 1 5.34 1935 6.21 2.48 4.14 10.78 1.03 2.52 40 5
Immediate roof 5.64 1879 5.07 2.03 3.38 8.15 0.76 1.90 40 5
Coal 3.65 1406 2.00 0.80 1.33 2.04 0.04 0.65 25 2
Floor 50.00 1935 6.21 2.48 4.14 10.78 1.03 2.52 40 5

Note: rcm = rock mass compressive strength, rtm = rock mass tensile strength.

Table 3
Numerical model input data for panel 21 of PVK 5 mine.

Thickness (m) Density Young’s mod. Shear mod. Bulk mod. rcm rtm Cohesion Friction Dilation
(kg/m3) (GPa) (GPa) (GPa) (MPa) (MPa) (MPa) (°) (°)
Overburden 166.34 2050 3.86 1.55 2.58 4.76 0.41 1.11 40 5
Main roof 15.94 2050 3.86 1.55 2.58 4.10 0.35 0.96 40 5
Immediate roof 30.02 2045 3.01 1.20 2.00 4.01 0.39 0.94 40 5
Coal 3.00 1400 2.00 0.80 1.33 4.63 0.31 1.48 25 2
Floor 50.00 2300 3.86 1.55 2.58 4.10 0.35 0.96 40 5
G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156 1147

to zero after achieving the equilibrium condition to simulate the


virgin ground condition.

3.3. Solution scheme

This model solution scheme is based on the basic concept of


ground response curve (Fig. 4) showing the relationship between
roof convergence and the support pressure. The initial roof conver-
gence (AB) after excavation of the rock is due to the elastic defor-
mation of the strata. The support pressure required to prevent any
convergence in this condition is very high and is of the order of the
primary stress level at point A. As the roof begins to deform, the
support resistance required to prevent further convergence re-
duces, as arching and self-supporting capacity of the ground is in-
duced at point B. The roof, then reaches a point where failure
begins to develop at point C. The required support resistance, then,
begins to increase as self-supporting capacity is lost and a support
of higher capacity is required for failed ground at point D.
In the numerical model, for obtaining reasonable load on sup-
port at the face, we obtain roof strata convergence curve with re-
spect to time-stepping for a typical numerical model. A general
nature of the curve is shown in Fig. 5. In this curve, we find that
the rate of roof convergence is very steep and elastic in nature Fig. 5. Numerical model obtained ground response vs. time stepping curve for roof
for time-steps from A1 to A2. Installing a support to prevent this strata.

elastic roof convergence is not beneficial rather it only leads to


undue load on the support. Further, the support resistance
required to prevent this section of roof convergence is very high flag is less than 0.06. During this process, the value of ‘d2’ is up-
which is neither required nor desirable. Installing the support dated by the value of ‘d3’at the end of each loop and a new value
any time between the roof convergence corresponding to time- of ‘d3’ is obtained. This normalization criterion of strata conver-
steps A2 and A3 is more useful for reducing the rate of post elastic gence is helpful to find the point of time-steps, when the elastic
roof convergence so that the roof condition at the face remains limit of roof convergence just passes off.
stable. Therefore, the simulated support is installed in the numer- Once this point is achieved, the support is installed at the new
ical model at every stage of face advance after a time-step location of the face and the model is solved for a stable condition
corresponding to convergence at A2 so as to obviate the effect of of the strata convergence. This is achieved by activating another
elastic roof convergence on the support. FISH module. In this module, the vertical displacement of the roof
A FISH module has been developed for the purpose of checking strata is monitored at the same location, i.e. 3 m behind the face
the point A2. It monitors the vertical displacement of the roof stra- and one grid above the face roof, after every 100 time-steps for ini-
ta 3 m behind the face and one grid above the face roof after every tial 200 time-steps. Here, the displacement obtained after first 100
100 time-steps for initial 300 time-steps. The displacement ob- steps is defined as ‘dx’ and that after the next 100 steps as ‘dy’. A flag
tained after first 100 steps is defined as ‘d1’, that after the next is defined as the absolute value of ‘(dy  dx)’. Now the time-step-
100 steps as ‘d2’ and that after the last segment as ‘d3’. The absolute ping scheme enters into a loop of 500 time-steps in five cycles,
value of ‘(d2  d1)’ is defined as ‘dsl’. A flag is defined as the ratio of each of 100 steps till the value of flag is less than 106. During this
absolute value of ‘(d3  d2) to ‘dsl’. Now the stepping scheme enters process, the value of ‘dx’ is updated by the value of ‘dy’ at the end of
into a loop to make 100 time-steps in each cycle till the value of each loop and a new value of ‘dy’ is obtained. When the time step-
ping comes out of the loop after satisfying the flag condition, the
roof convergence with respect to time-stepping becomes stable
and thus a stable solution is obtained in terms of a practical strata
monitoring parameter for every face position. Here, it may be
noted that this time stepping procedure used by the model does
not reflect the real time. Rather it is an iteration procedure used
for solving the model from previous state of equilibrium to the
new state of equilibrium following a physical solution.

3.4. Simulation of failure and caving of strata

Mohr–Coulomb plasticity model with tension cut off has been


used to simulate the roof failure, which assumes that rock will fail
under shear. This is the simplest failure model with minimum
number of assumptions and gives accurate representation of rock
mass behaviour under high confinement stress on a large scale
[79]. The yield function of this model is given by
pffiffiffi
fs ¼ r1  ðr2 q þ 2c qÞ ð6Þ
ft ¼ r2 þ rt ð7Þ
where fs is the Mohr–Coulomb shear yield function; ft the tension
Fig. 4. Rock–support interaction diagram. yield function; r1 the major principal stress, MPa; r2 the minor
1148 G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156

principal stress, MPa; q ¼ ð1þsin /Þ


ð1sin /Þ
; u the friction angle, degree; c the These FISH modules make use of the sub-modules to simulate
cohesive strength, MPa; and rt is the tensile strength, MPa. the mining sequence and progressive face advance.
Shear yielding of strata is detected in the model when fs < 0 for
Eq. (6) and tensile failure is detected when ft < 0 for Eq. (7). Con- 3.6. Modeling of goaf material in longwall caving
ventionally, all compressive stresses are positive while tensile
stresses are negative. The review of literature also suggested that modeling of goaf is
In the present work, it has been considered that caving of roof necessary where a number of periodic caving with a very long ad-
strata is different from failure or yielding of the strata. Yielding vance of face (the whole panel length has been simulated in some
of the strata is a result of induced stress exceeding the strength cases) have been considered [40,41,45,80,81]. The goaf compaction
limit following the Mohr–Coulomb failure criteria with tension has also been considered for the study of subsidence or the design
cut off. This original failure criterion has been used with a slight of yield pillars [82–85]. However, a number of authors [42,86] did
modification in the present work to consider the reduction in cohe- not simulate the goaf material while simulating the effect of
sive strength of strata which has failed in tension. In the normal change in characteristics of strata with progress of a longwall face
failure criteria, there is no such consideration, whereas here, the for a limited face advance. The compaction of the goaf and its resul-
cohesive strength of the element failing in tension is reduced to tant resistance do not occur during local falls and main fall with
zero, which seems to be valid consideration as observed from lab- progressive advance of the longwall face. It occurs far behind the
oratory test observations. This modification simulates the strain face only during periodic caving. Here, in the present work, numer-
softening behaviour of the strata without much of assumption. A ical simulation has been done within the limits of two dimensional
FISH module is used to monitor the tension and cohesion status configuration of face length and face advance, i.e. the simulated
of strata elements. If any element of the strata is detected for fail- face advance is restricted to only 0.67 times the face length. There-
ure in tension, its cohesion value is set to zero. fore, we have simulated only a few periodic caving which falls
Although, failure of the strata is the preliminary requirement within this limit of face advance. In fact, the probability of goaf
for its caving, whether a failed strata caves or not, it is guided resistance arises only when the overlying strata is deformed con-
by the shear strain or vertical displacement that the failed strata siderably over the fully pack goaf material. Such a condition is pos-
has undergone. The literature survey conducted in this work did sible only much behind the face. Hence, the probability of goaf
not disclose any such work done previously, which have consid- compaction and resultant resistance in the vicinity of the face,
ered these criteria for modeling the caving of strata. The present where the support is deployed, do not arise. Further, the strata,
wok has used these criteria intuitively where the limit of maxi- which play role on loading of the support at the face, are not sup-
mum allowable shear strain is set to 0.25 and the maximum ver- ported by the filled and compacted goaf. In view of this, modeling
tical displacement is set to 1 m (1 m in downward direction) for of goaf has not been done in the numerical model without any
caving of the failed strata to occur in the model. The models were appreciable effect on the result.
run automatically without much problem of bad geometry with
this limit of maximum shear strain and vertical displacement
while solving the models in large strain mode. These functions 4. Result extraction and model verification
are performed using two FISH modules. The first FISH module is
used to monitor shear strain in the roof strata. If the shear strain The results of the model are obtained in terms of front abut-
in any element exceeds the maximum limit (set as ‘0.25’), it is ment pressure, load on support, face convergence slope and verti-
deactivated to represent caving of the strata under excessive shear cal strain in coal at the face using separate FISH modules for all the
strain. Similarly, the second module monitors the vertical dis- parameters. A FISH module searches for the maximum value of
placement in the roof strata. If the vertical displacement in any vertical stress at the middle level of the coal face and prints its va-
element exceeds the maximum limit (set as ‘1 m’), it is deacti- lue and the location with respect to the face for every stage of face
vated to represent caving of the strata under excessive vertical advance. Similarly, the load on support for every stage of face ad-
displacement. vance is obtained using another FISH module.
The convergence at a point is calculated as the difference of ver-
3.5. Simulation of mining sequence tical displacement of the node in the roof level and that of the cor-
responding node in the floor level. The face convergence data so
The progressive face advance is simulated in stages of 1 m till obtained are retrieved by a FISH module for nodes located at 0,
the main fall and first two periodic caving are observed within 1, 2, 3 and 4 m distance from the face at every stage of the face ad-
the two dimensional modeling limitations. Once the face advances vance and the convergence slope for 1, 2, 3 and 4 m location at the
by 6 m, a powered roof support of given capacity and load defor- face is computed with respect to the convergence at 0 m location
mation characteristics is installed at the face simulating the se- (i.e. at the face point). The average face convergence slope is ob-
quence of longwall face advance. For simulation of mining tained by averaging the convergence slope of these four points.
sequence and support advance, a separate file is activated. This file The vertical strain in coal at the face is computed using a separate
uses various FISH modules for support installation and advance FISH module, which computes the percentage of vertical closure of
commands. It also saves the model status after every 1 m of face the face with respect to the height of extraction. Only the post-
advance. The FISH modules used in the file perform following elastic stage of roof–floor convergence or strain is considered for
functions: necessary studies.
A separate log file is used to store the data of specific parameter
(a) Simulates initial face advance of 5 m till the formation of for all the simulated face positions and results of all the parameters
installation chamber is simulated, are stored separately in specific log files. The data from these log
(b) Simulates the face advance and delay the support installa- files are fed into Microsoft Excel files using various macros for edit-
tion at the new face location till the elastic range of roof con- ing the format of stored data. These macros are very convenient for
vergence occurs, obtaining the final results containing large volume of data from the
(c) Installs the support at the new face location, models. The formatted outputs of the data are finally presented in
(d) Iterates to converge the solution after installation of the sup- graphical forms with respect to face advance. A movie file is also
port in the advanced condition of the face. created to visualize the progressive failure and caving pattern of
G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156 1149

Main fall Periodic caving


3.50 Yield load = 3.19 MN

3.00

2.50

Load (MN) 2.00


Numerical modeling result
Field observation
1.50

1.00

0.50

0.00
5 10 15 20 25 30 35 40 45 50 55 60
Face advance (m)
Fig. 6. Model simulated vs. field observed load on 4  3.19 MN support deployed in Panel A4 at Moonidih mine in Jharia coal field.

7.44 Yield load = 7.46 MN


Periodic caving
Main fall
6.51
Field observation
5.58

4.65
Load (MN)

3.72
Numerical modeling observation
2.79

1.86

0.93

0.00
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95
Face advance, m

Fig. 7. Model simulated vs. field observed load on 4  7.46 MN support deployed in Panel 21 at PVK 5 mine in Singereni coalfield.

strata with face advance. The outputs are finally analyzed to obtain to be periodically stopped for recovery operation at the face during
the response of the model for simulating the features of local fall, periods of severe roof caving, and collapsed longwall workings
main fall and periodic roof caving. where the face could not be recovered after occurrence of severe
In the progressive failure plots, for example Figs. 8a–f and 9a roof degradation during uncontrolled caving of the strata and the
and b, the following conventions have been adopted to indicate working of the face was permanently stopped.
the various mechanism of roof failure. The sign  1(red in colour) The numerical models assess the caving behaviour of strata and
indicates yield in shear, o (pink in colour) indicates yield in tension performance of the field deployed powered roof support. A movie
and x (green in colour) indicates elastic zone in current state, but of progressive face advance has also been prepared for each model
yield in past. The zones showing ‘elastic in current state, but yield to visualize the failure and caving of strata with progressive face
in past’ should ideally occur due to relaxation of stress after failure advance. The comparative study of the model and the field ob-
of the strata. served load on the support with progressive face advance for two
typical case studies are shown in Figs. 6 and 7. The model input
4.1. Model verification data for these case studies are given in Tables 2 and 3, respectively.
The plots clearly show that numerical modeling results are very
Numerical simulation of strata behaviour and performance of effective in estimating the peak loading conditions of the support.
the supports deployed in 23 different longwall panels, operated It may be noted here, that as the field observation is manual and
in various coal seams of India have been studied. These panels in- not continuous, it has not been possible to record all the peak load-
clude successfully operated panels where no significant strata con- ing conditions during the progressive face advance from the field
trol difficulty was experienced during extraction of the panel, observation data. Figs. 8a–f and 9a and b show some of the simu-
panels extracted with considerable difficulty where the face had lated plots of progressive roof failure, parting of layers and the cav-
ing of strata for these cases. The numerical modeling results of
main fall and periodic caving spans have been compared with
1
For interpretation of color in Figs. 8 and 9, the reader is referred to the web those of the field values. The comparative plots of numerical mod-
version of this article. eling results of main fall and periodic caving spans with their field
1150 G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156

Fig. 8. Progressive caving of roof at Moonidih A4 panel (X and Y scale  100 m). (a) 11 m Face advance, progressive failure and caving of sandstone immediate roof. (b) 25 m
Face advance, failure of main roof layer 1 during main fall. (c) 26 m Face advance, caving due to main fall. (d) 35 m Face advance, failure of main roof layer 2 during first major
periodic caving. (e) 36 m Face advance, caving of upper strata during first major periodic caving. (f) 54 m Face advance, regular caving of strata.

observed values (Figs. 10 and 11) also show that numerical simu- numerical modeling results of the various case studies. The obser-
lation findings corroborate with the field observations. vations have been used to classify the performance of different
A compilation of the qualitative comparison of the face condi- longwall faces in these three groups on the basis of field observa-
tion observed in the field with that of the numerical model obser- tions and their numerical modeling observations. Group 1 covers
vation (Table 4) also shows that the modeling results are in good the faces, where extraction was completed successfully without
agreement with the field observations. However, in some of the any considerable strata control problem. The faces under Group 2
models (case studies under sl. nos. 11–17 and 21), where en masse faced frequent face stoppages for face recovery operations neces-
caving of strata is observed, dynamic loading of support was expe- sary to overcome roof deterioration due to excessive convergence
rienced in the actual operating condition in field resulting in dam- at the face during the period of main fall or periodic caving of stra-
age of supports and collapse of the face. The model observed load ta. The plot shows that for all the faces under Group 1, the peak
on the supports and the face convergence for these cases, do not face convergence slope obtained from numerical model is less than
reflect the actual field value. It is because; the present static model 75 mm/m of face advance. It is P75 mm/m of face advance in case
does not consider the effect of dynamic loading on the support. of faces in Group 2, which could be extracted at the cost of periodic
However, the model shows the en masse caving of strata in all such face stoppages. Here, the roof at the face deteriorated due to high
cases which cause dynamic loading. Such caving condition of strata face convergence, which could not be controlled by the support
demands special considerations such as use of hydraulic powered provided at the face. The face stoppage was necessary for its recov-
support with suitable rapid yield valve, resorting to roof softening, ery. In some of the faces, induced blasting from underground was
etc. also carried out to limit the overhang of immediate roof. In view of
Fig. 12 shows the three group classification of different longwall these observations, a safe convergence slope (PFCS) of less than
panels based on peak face convergence slope obtained from 75 mm/m may be considered as a design criterion for optimal
G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156 1151

Fig. 8 (continued)

support capacity selection for all static loading conditions, where quence of strata lying within 15 times extraction height above
the failure and subsequent caving of rock is regular. Accordingly, the coal seam. The major planes of parting among the rock beds
any support giving a face convergence slope P75 mm/m obtained are identified using an index called Roof Separation Index (RSI).
from numerical simulation may be considered as under-rated The mechanical properties of different layers in roof existing with-
support for working the face. However, the modeling observation in the caving zone for a given geo-mining condition are assessed
for dynamic loading (Group 3 faces) cases can not be solely relied after identification of the parting planes in the overlying strata.
upon the face convergence value observed in the model because it The maximum height of caving zone in the roof is considered as
is not representative of the actual loading condition observed in 15 times height of extraction assuming the bulking factor as low
the field. Dynamic loading results in much heavy loading on the as 1.07. The rock mass strength for various layers in roof and the
support in actual field condition compared to the value estimated floor rock is estimated using scale effect and thickness weighted
by the present static model which does not consider effect of average RQD of the layer. The in situ stress in the virgin ground
dynamic loading. An approach for accounting dynamic loading on condition is estimated using a theoretical approach in absence of
powered support has been developed in Ph.D. work [87]. field-measured values. Parting planes are simulated using interface
elements.
5. Discussions Modeling of a hydraulic powered support has been done consid-
ering its ability to resist the maximum lateral load limited to 30% of
The numerical modeling approach reported in this paper is use- its yield load capacity, while in case of pure vertical loading the lat-
ful for simulating the progressive caving of strata and assessing the eral load provided by the modeled support remains zero. The stiff-
performance of a given hydraulic powered roof support in a given ness of the supports of various capacities has been determined
geo-mining and strata conditions. It makes use of borehole lithol- using stiffness of hydraulic legs and verified in the laboratory.
ogy and laboratory test data to identify the rock beds in the se- The stable solution scheme, which has been used in numerical
1152 G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156

Fig. 9. Progressive caving of strata at panel 21 of PVK 5 mine (X and Y scale  100 m). (a) 44 m Face advance, progressive parting and failure of immediate roof. (b) 48 m Face
advance, failure zone extends to main roof, shear failure of immediate roof.

160
140 Field observed Model estimated

120
Main fall span (m)

100
80
60
40
20
0
Moonidih A4

Jhanjra W 1

Jhanjra W 2

Jhanjra AW1

Jhanjra AW 2

Jhanjra AE 1

Jhanjra AW 3

Jhanjra AW 4

Jhanjra AW 8

Jhanjra AW 9

Khottadih P1

Khottadih P 2

Balrampur P 1

Balrampur B7

Kumda K 5

Rajendra P 2

Churcha

PVK P 21

VK 7 P 4

JK 5 P 3

GDK 9 P 3T

GDK 10A P 11

GDK 10A P 12

Longwall panel

Fig. 10. Field observed vs. numerical model estimated span of main fall in various longwall panels.

modeling, is based on the concept of Ground Response Curve. It en- sively advancing face. The ground response curve is obtained auto-
ables a reasonable simulation of the interaction between strata and matically using a FISH module. The support is installed after
the yielding support of given capacity deployed at the progres- completion of elastic deformation leading to face advance. The
G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156 1153

35

Periodic caving span (m )


30 Field observed Model estimated
25
20
15
10
5
0
Moonidih A4

Jhanjra W 1

Jhanjra W 2

Jhanjra AW1

Jhanjra AW 2

Jhanjra AE 1

Jhanjra AW 3

Jhanjra AW 4

Jhanjra AW 8

Jhanjra AW 9

Khottadih P1

Khottadih P 2

Balrampur P 1

Balrampur B7

Kumda K 5

Rajendra P 2

Churcha

PVK P 21

VK 7 P 4

JK 5 P 3

GDK 9 P 3T

GDK 10A P 11

GDK 10A P 12
Longwall Panel

Fig. 11. Field observed vs. numerical model estimated span of periodic caving in various longwall panels.

Table 4
Qualitative comparison of field observed condition with numerical model observation.

Sl. no. Longwall panel Field observation Numerical modeling observation


Face condition Load on support Face condition Load on support
1. Moonidih A4 Stable, no significant problem Achieved yield load during main Stable, regular caving in goaf Achieved yield load during main
fall and first periodic caving fall and first periodic caving
2. Jhanjra W 1 Stable, no significant problem Did not achieve yield load during Stable, regular caving in goaf Achieved yield load during
weight at the face periodic caving
3. Jhanjra W 2 Stable, no significant problem Did not achieve yield load during Stable, regular caving in goaf Did not achieve yield load
weight at the face
4. Jhanjra AW1 Stable roof till 420 m face Achieved yield load during Stable, regular caving in goaf Achieved yield load during
advance weight at the face periodic caving
5. Jhanjra AW 2 Deteriorated during weights Achieved yield load during Increased face convergence Achieved yield load during main
weight at the face during weights fall and periodic caving
6. Jhanjra AE 1 Stable, no significant problem Achieved yield load during Stable, regular caving in goaf Achieved yield load during main
weight at the face fall and periodic caving
7. Jhanjra AW 3 Deteriorated during weights Achieved yield load during Increased face convergence Achieved yield load during main
weight at the face during weights fall and periodic caving
8. Jhanjra AW 4 Stable, no significant problem Achieved yield load during Stable, regular caving in goaf Achieved yield load during
weight at the face periodic caving
9. Jhanjra AW 8 Deteriorated during weights Achieved yield load during Very high face convergence Achieved yield load during main
weight at the face during weights fall and periodic caving
10. Jhanjra AW 9 Stable, no significant problem Achieved yield load during Stable, regular caving in goaf Achieved yield load during main
weight at the face fall and periodic caving
11. Khottadih P1 Sudden collapse of strata, Achieved yield load during En masse caving Achieved yield load during
damage of support at several weight at the face periodic caving
instances
12. Khottadih P2 Sudden collapse of strata, Achieved yield load during En masse caving Achieved yield load during main
damage of support at several weight at the face fall and periodic caving
instances
13. Balrampur P1 Damage of support during main Achieved yield load during En masse caving Peak load less than support
fall weight at the face capacity
14. Balrampur B7 Extraction done with induced Achieved yield load during En masse caving Peak load less than support
caving by blasting from surface weight at the face capacity
15. Kumda K5 Extraction done with induced Achieved yield load during En masse caving Peak load less than support
caving by blasting from surface weight at the face capacity
16. Rajendra P2 Extraction done with induced Achieved yield load during En masse caving Achieved yield load during main
caving by blasting from surface weight at the face fall
17. Churcha P1 Caving of strata accompanied by Achieved yield load during En masse caving Most frequently loaded to yield
air blast, face collapsed during weight at the face load
periodic weight
18. PVK 5 panel 21 Stable, no significant problem Achieved yield load during Stable, regular caving in goaf Achieved yield load during some
weight at the face of the periodic caving
19. VK 7 panel 4 Stable, no significant problem Achieved yield load during Stable, regular caving in goaf Achieved yield load during main
weight at the face fall
20. JK 5 panel 3 High face convergence, severe Achieved yield load during Significant strata failure in Achieved yield load from a
strata problem weight at the face advance of the face, high face prolonged period during main
convergence fall and periodic caving
21. GDK 9 panel 3T Sudden collapse of strata, Achieved yield load during En masse caving Achieved yield load during
damage of support weight at the face periodic caving
22. GDK 10A panel 11 High face convergence and roof Achieved yield load during Significant strata failure in Achieved yield load for a
degradation when face under weight at the face advance of the face, high face prolonged period during main
weight stopped due to convergence fall and periodic caving
breakdown
23. GDK 10A panel 12 Stable, no significant problem Achieved yield load during Stable, regular caving in goaf Achieved yield load during main
weight at the face fall and periodic caving
1154 G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156

Longwall Panel

New Kumda, K 5

Khottadih, 1& 2

GDK 10A, P 12

GDK 10A, P 11
Balrampur, B 7

Balrampur, P 1

Jhanjra, AW 3
Jhanjra, AW 4

Jhanjra, AW 9

Jhanjra, AW 2

Jhanjra, AW 8
Moonidih, A4

Jhanjra, AW1
Rajendra, P 2

GDK 9, P 3T

Jhanjra, AE 1
Churcha, P 1

Jhanjra, W 2

Jhanjra, W 1
PVK, P 21

VK 7, P 4

JK 5, P 3
0
Peak face convergence slope, mm/m of face advance

2 3 6 9 14 17
24 26
50 35 37 36
45 46 48
56 55 55
75
100 90
102
114
3 1
150
2

200

250
257

300

Fig. 12. Three group classifications of numerical modeling observed peak face convergence slope based on general field experience: longwall faces in Group 1 were extracted
successfully without any considerable strata problem; Group 2 longwall faces experienced excessive face convergence and frequent face stoppages and Group 3 faces
experienced dynamic loading of support.

state of completion of elastic deformation is automatically de- as a post-failure strain softening behaviour of strata failed in
tected by the FISH module on the basis of history of convergence tension.
of roof at the face. Another FISH module has been developed to The numerically simulated modeling results of caving span for
simulate the post-failure strain softening behaviour of Mohr–Cou- main fall and periodic caving corroborated with their field ob-
lomb rock material, where cohesion of the element is reduced or served values. The field observations also supported the caving
set to null in the case of failure of an element in tension. The caving behaviour, failure characteristics, load on the support, abutment
of strata is simulated using another FISH module which monitors stress and vertical strain in coal at the face as observed in numer-
the shear strain and the vertical displacement of the failed rock ele- ical simulation. The three group classification of longwall panels as
ments. This approach differentiates between the failure and the per the caving behaviour and face convergence obtained in numer-
caving of strata. Only those failed elements, where the shear strain ical modeling study is reasonable and representative of the field
exceeds 0.25 or the vertical downward displacement exceeds 1 m, experience.
are assigned caved.
The model observations shows that weight at the face is not al- 6. Conclusions
ways observed due to failure of main roof especially when the
immediate roof is thicker. In this case, the immediate roof itself A numerical modeling approach is proposed for simulation of
acts as the main roof and the role of immediate roof gets different progressive caving behaviour and support performance in longwall
from the conceptual considerations. The numerical modeling tool mining. The model uses Mohr–Coulomb plasticity model with ten-
obviates such conceptual considerations and hence is more effec- sion cut off and incorporates parting planes, in situ stress field, rock
tive in understanding the mechanism of failure and caving of strata mass strength and failure criteria, and proper modeling of support
in a given geo-mining condition. The model observed load on sup- setting and yielding behaviour as well as field representative se-
port are quite representative of the field observations as depicted quence of face advance. The model performance has been tested
from the comparative plots of two cases. for 23 different strata conditions from different longwall panels
The modeling results show that the panels were extracted in the and verified with their field observations. The model observations
field without any severe strata problem, where regular caving of are in lien with the field observations in varying geo-mining condi-
strata with peak face convergence slope of less than 75 mm/m of tions. The modeling study combined with the field experience of
face advance was observed in the numerical model. However, in- panels extracted in different conditions suggest that a safe conver-
creased convergence at the face was observed in the numerical gence slope of less than 75 mm/m of face advance may be consid-
model of other longwall panels where the longwall face had to ered as a design criterion for optimal support capacity selection for
be stopped periodically to recover the deteriorating roof condition all static loading conditions, where the failure and subsequent cav-
due to excessive convergence at the face during periods of major ing of rock is regular. Additional study needs to be made for esti-
roof caving. Instances of en masse caving were also observed in mation of raid yield valve requirement for safe working of
some of the longwall panels. A common modeling observation of supports in en masse caving leading to dynamic loading conditions.
these panels is that a thick layer of strata provides a very poor ten-
dency of bending and prevents the tensile crack from progressing Acknowledgements
significantly with progressive face advance, leading to en masse
failure and caving of strata in such condition. Dynamic loading The authors are thankful to the Head, Department of Mining
on the support was experienced while working these faces in the Engineering, ISM University for providing necessary laboratory
field. Tensile failure was observed as the predominant failure facility for conducting the numerical modeling work. The field data
mechanism in most of the cases. Shear failure of immediate roof were compiled from BCCL and SCCL mines. The authors are thank-
in advance of the face was also observed in some of the cases ful to the management of various longwall mines for their co-oper-
where the depth of working is more. Shear failure was also noticed ation and encouragement for this work. The work reported in this
G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156 1155

paper is part of the Ph.D. Thesis submitted to ISM University by the [30] Porter I, Aziz N I. Longwall facelines: geology, convergence and powered
support rating. Min Sci Technol 1988;7:243–52.
first author. The views and findings expressed in this paper are
[31] Jiang YM, Peng SS, Chen JS. DEPOWS – a powered support selection model. In:
opinion of the authors and not necessarily of the organization they Proceedings of the rock mechanics as a guide for efficient utilization of natural
serve. resources. Rotterdam: AA Balkema; 1989. p. 141–8.
[32] Jackson DJ, Newson SR. The design and application of longwall face support
systems. In: Proceedings of the international strata control conference,
References Düsseldorf; May 1989. p. 296–315.
[33] Peng S S. Ground control models for coal mine design. J Mines Metals Fuels
[1] Obert L, Duvall WI. Rock mechanics and the design of structures in rock. New 1992:169–78.
York: John Wiley and Sons, Inc.; 1967. [34] Das SK. Optimized selection of powered supports in mechanized coal mines in
[2] Majumdar S. The support requirement at a longwall face – a bending moment India. Ph.D. Thesis, Indian School of Mines, Dhanbad; 1994.
approach. In: Proceedings of the rock mechanics: key to energy production: [35] Das SK. Determination of optimum capacity of powered roof supports in the
27th US symposium on rock mechanics, The University of Alabama, longwall face in medium thick coal seam. J Mines Metals Fuels 1995:99–105.
Tuscaloosa, Alabama; June 1986. p. 325–32. [36] Das SK. Indian coal measure rocks and their realistic classification for the
[3] Zamarski B. Control of roof in longwall faces of OKR. VVUU; 1970. selection of roof support. Minetech 1996;17(4):55–62.
[4] Arioglu E, Yuksel A. Design curves for hydraulic face supports. J Mines Metals [37] Das SK. Standardization of powered roof supports for longwall faces. Min
Fuels 1984:173–8. Technol 1997;79(908):95–107.
[5] Hongzhu Z. A study of strata behaviour and support resistance of a fully [38] Das SK. Behaviour of the fractured roof strata over a longwall panel and
mechanized longwall face. In: Proceedings of the international symposium on determination of capacity of powered roof supports. In: Proceedings
mining technology and science, China Institute of Mining and Technology, of the international conference on mining, New Delhi; November 1999. p.
Xezhou; September 1985. p. 67–73. 563–89.
[6] Peng SS, Wu J, Li HC, Chen SL. How to determine yield load of longwall roof [39] Medhurst TP, Kevin R. Ground response curves for longwall support
supports. Coal Min 1986;23(10):40–3. assessment. Min Technol (Trans Inst Min Metall, Section A) 2005;114:A81–8.
[7] Peng SS, Zhu DR, Jiang YM. Roof classification and determination of the support [40] Kwasniewski M, Wang J. Computer simulation of longwall coal mining with
capacity for the fully mechanized longwall faces. J Mines Metals Fuels, roof caving III. Effect of mining rate on strata behavior. Zeszyty Naukowe
Longwall Min Developments – Special Number 1989:289–96. Politechniki Slaskiej, Seria:Gornictwo Z. 221 Nr kol. 1256; 1994.
[8] Pawlowicz K. Classification of rock capability of coal measure strata in Upper [41] Kwasniewski M, Wang J. Computer simulation of longwall coal mining
Silesia coalfield, Prace GIG, Komunikat, No. 429; 1967 [in Polish]. with roof caving I. Displacement field and fracture zones in the rock mass.
[9] Bilinski A, Konopko W. Criteria for choice and use of powered supports. In: Zeszyty Naukowe Politechniki Slaskiej, Seria:Gornictwo Z. 221 Nr kol. 1256;
Proceedings of the symposium on protection against roof falls, Katowice, 1973, 1994.
paper no. IV-1; 1973. [42] Poulsen BA, Guo H. Numerical modeling of longwall coal mining through a
[10] Unrug K, Szwilski. An influence of strata control parameters on longwall weak to strong transition in strength of immediate roof strata. In: Proceedings
mining design. In: Proceedings of the 21st US symposium on rock mechanics, of the FLAC and numerical modeling in geo-mechanics; 2001. p. 287–93.
Rolla, Morgantown; May 1980. p. 720–8. [43] Yasitli NE, Unver B. 3-D numerical modeling of stresses around a longwall
[11] Peng SS, Chiang HS. Longwall mining. John Wiley and Sons; 1984. panel with top coal caving. J South African Inst Min Metall
[12] Singh TN, Singh B. Design of support system in caved longwall faces. In: 2005;105(5):287–300.
Proceedings of the colloquium on longwall face supports, Dhanbad; 1979. p. [44] Unver B, Yashitli NE. Modeling of strata movement with a special reference to
79–85. caving mechanism in thick seam coal mining. Int J Coal Geol 2006;3(4):
[13] Singh TN, Singh B. Design criteria of face supports. In: Proceedings of the 227–52.
symposium on state of the art of ground control in longwall mining and [45] Richardson AM, Gilbride LJ. The influence of massive sandstones in the main
mining subsidence, Organized by Society of Mining Engineers, New York, roof on longwall support loading. In: Proceedings of the 15th international
1982. p. 145–50. . conference on ground control in mining, Golden, Colorado, August 1996. p. 95–
[14] Kuznetsov ST, Pekarskii DG, Korovkin VT. Determining the normal stresses in a 109.
uniform bent beam cantilever. Sov Min Sci 1973;9(5):478–82. [46] Trueman R, Lyman G, Callan M, Robertson B. Assessing longwall support–roof
[15] Sarkar SK, Dhar BB. Strata control failures at caved longwall faces in interaction from shield leg pressure data. Min Technol (Trans Inst Min Metall,
India – experience from Rana to Churcha (1964 to 1990). In: Proceedings of Section A) 2005;114(September):A176–84.
the fourth Asian mining, Organized by MGMI at Calcutta; November 1993. p. [47] Sarkar S K, Singh B. The investigation into strata control failures at caved
361–80. longwall faces in India and a new approach for support planning to avoid such
[16] Sarkar S K. Mechanised longwall mining – the indian experiences. Oxford & occurrences. J Mines Metals Fuels 1979;May:127–34.
IBH Publishing Company Private Limited; 1998. [48] Singh GSP. Development of a model for cavability assessment in longwall
[17] Ghose AK, Dutta D. A rock mass classification model for caving roofs. Int J Min panels in India. M.Tech Thesis, Department of Mining Engineering, Indian
Geol Eng 1987;5:257–71. School of Mines, Dhanbad; 2004.
[18] Terzaghi K. Theoretical Soil Mechanics. 3rd ed. New York: John Wiley; 1965. [49] Itasca. FLAC version 4.0 user’s manual. Minneapolis (Minnesota, USA): Itasca
[19] Evans I. Face support requirements: a problem in arching. NCB MRDE report Consulting Group, Inc.; 2000.
no. 64; 1975. [50] Singh GSP, Singh UK, Banerjee G. Cavability assessment model for longwall
[20] Barry AJ, Nair OB, Miller JS. Specifications for selected hydraulic powered roof working in India. In: Proceedings of the third Asian rock mechanics
supports with a method to estimate support requirement for longwalls. US symposium, Organized by ISRM, Kyoto, Japan; November 2004. p. 295–300.
Department of Interior, Bureau of Mines, IC 8424; 1969. [51] Wilson AH. The stability of underground workings in the soft rocks of the coal
[21] Ashwin DP. Some fundamental aspects of face powered support design. Min measures. Ph.D. Thesis, University of Nottingham, UK; 1980.
Eng 1975;119:659–75. [52] Hoek E, Brown E T. Practical estimates of rock mass strength. Int J Rock
[22] Wade LV. Longwall support load predictions from geological information. In: Mechanics Min Sci 1997;34(8):1165–86.
Proceedings of the SME–AIME fall meeting & exhibition, Denver, Colorado, [53] Tsiambaos G, Sabatakakis N. Considerations on strength of intact sedimentary
September 1976, No. 76-I-308; 1976. rocks. Eng Geol 2004;72:261–73.
[23] Josien JP, Gouilloux C. Present and future roof control and support in longwall [54] Sonmez H, Gokceoglu C, Nefeslioglu HA, Kayabasi A. Estimation of rock
faces in French coal mines. Colliery Guard Int 1978:49–58. modulus: for intact rocks with an artificial neural network and for rock masses
[24] Ming-Gao C. A study of the behaviour of overlying strata in longwall mining with a new empirical equation. Int J Rock Mech Min Sci 2006;43:224–35.
and its application to strata control. In: Proceedings of the symposium on [55] Robert H. Rock mass strength by rock mass classification. In: Proceedings of
strata mechanics, Newcastle upon Tyne; April 1982. p. 13–17. the South African rock engineering congress, Johannesburg; 1997. p. 346–56.
[25] Yuanwei S. Possibility and approaches for developing more compact and light [56] Aydan O, Dalgiç S. Prediction of deformation behaviour of 3-lanes Bolu tunnels
weight powered support. In: Proceedings of the international symposium on through squeezing rocks of North Anatolian fault zone (NAFZ). In: Proceedings
mining technology and science, China Institute of Mining and Technology, of the regional symposium on sedimentary rock engineering, Taipei; 1998. p.
Xezhou; September 1985. p. 55–66. 228–33.
[26] Wilson AH. The problems of strong roof beds and water bearing strata in the [57] Yudhbir LW, Prinzl F. An empirical failure criterion for rock masses. In:
control of longwall faces. In: Proceedings of the symposium on ground Proceedings of the fifth international congress of rock mechanics (ISRM), vol.
movement and control related to coal mining, Organized by The Australian 1; 1983. p. B1–B8.
Institute of Mining and Metallurgy, Illawara Branch, University of [58] Ramamurthy T. Stability of rock masses. Indian Geomech J 1986;16(1):1–74.
Wollongong; August 1986. p. 1–8. [59] Barton N. TBM tunneling in jointed and faulted rock. Rotterdam: Balkema;
[27] Budirsky S, Martinec P. Influence of support resistance on roof control in single 2000.
pass thick seam mining (4.5 m) – a case history. Min Sci Technol [60] Sheorey PR. Empirical Rock Failure Criteria. Rotterdam (Netherlands): A.A.
1986;4:59–67. Balkema; 1997.
[28] Bigby DN. Strata deformation measurements on longwall coalfaces and their [61] Bieniawski ZT. Rock mass classification in rock engineering. Proceedings of the
implications for design of powered support systems. Int J Min Geol Eng symposium on exploration for rock engineering. Cape Town: AA Balkema;
1987;5:59–73. 1976. p. 97–106.
[29] Peng SS, Shen LS, Wu J. How to select the proper type of powered support. [62] Peng SS, Su WH. The causes of cyclic excessive convergence at the longwall tail
Colliery Guard 1987;235(2):74–7. entry. Int J Min Eng 1983;1:27–41.
1156 G.S.P. Singh, U.K. Singh / Computers and Geotechnics 36 (2009) 1142–1156

[63] Smart BGD, Metcalfe K, McGraw WM. A new approach to the specification of [78] Smart BGD, Aziz NI. The influence of caving in the Hirst and Bulli seams on
powered supports. In: Proceedings of the international strata control powered support rating. In: Proceedings of the symposium on ground
conference, Düsseldorf; May 1989. p. 279–95. movement and control related to coal mining, Organized by The Australian
[64] Vujec S, Dunda S, Hrzenjak P, Kujundzic T. Excavation of natural stone in Institute of Mining and Metallurgy, Illawara Branch, University of
Croatia. In: Proceedings of the 19th world mining congress, mining in the 21st Wollongong; August 1986. p. 1–12.
century-quo vadis, New Delhi; November 2003. p. 609–19. [79] SIMRAC. Numerical modeling of mine workings. 2nd ed. GAP 728, 2000. <http:
[65] Beckett LA. Longwall mining: a finite element study of strata behaviour and //www.simrac.co.za/report/Reports/thrust2/gap728/gap728.htm>; December
expected subsidence. USBM progress report 10054, Denver Research Centre; 2006.
1984. [80] Banerjee G, Kushwaha A, Ray AK, Yadava KP, Sheorey PR. Application of
[66] Hsiung SM, Peng SS. Chain pillar design for US longwall panels. Min Sci numerical modeling to study the behaviour of caved goaf at longwall
Technol 1985;2:297–305. workings. In: Proceedings of the ROCKSITE-99: international conference on
[67] Tang DHY, Peng SS. Causes and mechanisms of surface fractures in a Central rock engineering technique for site characterization, Bangalore, India;
West Virginia coal mine. Min Sci Technol 1986;4(1):41–8. December 1999. p. 381–8.
[68] Evans I. The tensile strength of coal. Colliery Eng 1961;38:428–34. [81] Deb D, Verma AK. Numerical analysis of support and rock strata interaction at
[69] Holland CT. The strength of coal in mine pillars. In: Proceedings of the sixth longwall faces. In: Proceedings of the NSAGE – 2004 national symposium on
symposium on rock mechanics, University of Missouri; 1964. p. 450–66. advances in geotechnical engineering, Karnataka Geotechnical Centre of Indian
[70] Bieniawski ZT. An analysis of results from underground tests aimed at Geotechnical Society & Department of Civil Engineering, IIS, Bangalore; 2004.
determining the in situ strength of coal pillars. CSIR report MEG 569, Pretoria, p. 223–8.
South Africa; July 1967. [82] Badr SA, Mendoza R, Kieffer S, Salamon MDG, Ozbay MU. Numerical modeling
[71] Evans I, Pomeroy CD. The strength, fracture and workability of coal. NCB of longwalls in deep coal mine. In: Proceedings of the 22nd conference on
MRDE, Burton-on-Trent; 1973. p. 32–65. ground control in mining, West Virginia University, WV, USA; 2003. p. 37–43.
[72] Rajmeny PK, Singh UK, Rathore SS. A new model to estimate rock mass [83] Yavuz H, Fowell R J. A physical and numerical modeling investigation of the
strength accounting for the scale effect. Int J Rock Mechanics Min Sci roadway stability in longwall mining with and without narrow pillar protection.
2004;41:1013–21. Min Technol (Trans Inst Min Metall, Section A) 2004;113:A59–72.
[73] Jaeger JC, Cook NGW. Fundamentals of Rock Mechanics. Chapman and Hall; [84] Badr AE. Numerical analysis of coal yield pillars at deep longwall mines. Ph.D.
1979. Thesis, Colorado School of Mines, 2006. <http://www.mines.edu/research/
[74] <www.rocscience.com/downloads/phase2/webhelp/phase2_model/strength_ wmrc/Webpage/RP-2_pillars_files/NumericalMModel_pdf>; December 2006.
parameters.htm>; December 2006. [85] Esterhuizen GS, Karacan CO. Development of numerical models to investigate
[75] Sheorey PR. A theory for in situ stresses in isotropic and transversely isotropic permeability changes and gas emission around longwall mining panel, 2006.
rock. Int J Rock Mechanics Min Sci Geomech Abstracts 1994;31:23–34. <http://www.cdc.gov/niosh/mining/pubs/pdfs/donmt.pdf>; February 2006.
[76] Shoerey PR, Mohan GM, Sinha A. Influence of elastic constants on the [86] Mitra R, Westman, EC, Mrugala, MJ. Use of numerical modeling in longwall
horizontal in situ stress. Int J Rock Mechanics Min Sci Geomech Abstracts roof performance analysis. In: Proceedings of the sixth North America rock
2001;38:1211–6. mechanics symposium (NARMS): rock mechanics across borders and
[77] Barczak TM. State of the art testing and analysis of mine roof support systems. discipline, Houston, Texas; June 2004.
Eastern coal mine geo-mechanics: In: Proceedings of the bureau of mines [87] Singh GSP. Cavability assessment and support load estimation for longwall
technology transfer seminar, Pittsburgh, PA; November 1986, US Department workings in India. Ph.D. Thesis, Department of Mining Engineering, Indian
of Interior, Bureau of Mines, IC 9137; 1986. School of Mines University, Dhanbad; 2007.

You might also like