You are on page 1of 8

Salamon, M.D.G., Pillar failure in deep coal seams: Numerical Simulation.

ISRM 2003–Technology roadmap for rock mechanics, South African Institute of Mining and Metallurgy, 2003.

Pillar Failure in Deep Coal Seams: Numerical Simulation


La Rupture des Piliers dans les mines souterraines de
charbon: Une Simulation Numérique
Bruchverhalten von Pfeilern in tiefen Kohlenflözen:
Numerische Simulation
M.D.G. Salamon, S. Badr, R. Mendoza, M.U. Ozbay
Colorado School of Mines
Golden - CO - 80401 - USA

ABSTRACT: The use of coal pillars at depth is fraught with problems due roof control difficulties
and the possible occurrence of coal or pillar bumps. In the USA the largest and most important
application of pillars at depth is in longwalling, where chain pillars are the customary supports of
the multiple entry developments. At first, the chain pillars were designed using conventional
approaches, where the principle is that the strength of the pillar should exceed the load imposed on
it. Such approach to design led to larger and larger pillars at great depths, but instead of
eliminating, it tended to aggravate the roof control problems. Eventually the idea was born that
small sized yielding pillars, which have proved to be successful elsewhere, might provide the
answer to this pressing problem as well.
The aim of this paper is to study the modes of pillar failure at great depths, using two solution
methods. The first of these comprises some elementary solutions obtained with the aid of a
simplified conceptualization of the rock mass. This model consists of a strain-softening coal seam
sandwiched between two stratified masses. The second approach utilizes FLAC3D, which is a
commercially available three-dimensional numerical modeling software package. To gain a better
insight, both the strain-softening and ideally plastic Mohr-Coulomb constitutive laws embodied in
FLAC3D were exploited.
As expected, the width to height ratio of the pillar has proved to be the primary factor that
determined the behavior of the pillars. Narrow pillars fail in a stable manner without the
occurrence of instabilities. Intermediate w/h ratio pillars under the appropriate loading conditions,
may fail suddenly and may even collapse totally. Squat pillars do show the potential for sudden,
coal bump like side failure, but because the remaining large width of the inner core, the ruble
around the pillar provides sufficient confinement so as to reestablish stability. Finally, the study
revealed that the traditional Mohr-Coulomb law does not provide a useful insight into the failure
of coal pillars.

RESUME: L’utilisation des piliers de charbon dans les mines souterraines de grande profondeur
occasionne des problèmes due aux difficultés de control de plafond et possibles fracture ou
explosions au niveau des piliers. Aux Etats-Unies d’Amérique, l’utilisation des piliers dans les
mines souterraines de charbon comme moyen de support est considèrees comme étant la plus
grande et la plus importantes des méthodes utilisée dans le ‘’ longwalling method’’ ou une série
des piliers sont disposé de manière à supporter les multiple entrées d’exploitation de charbon. En
premier lieu, cette série des piliers étaient construit de manière conventionnelle avec un principe
basé sur le fait que le poids exerce sur les piliers doit être inférieur a la force que ces piliers exerce
sur le plafond. Ce principe a conduit a ce que le volume des piliers augmente au fur et à mesure
que la profondeur au de la mine augmente. Malheureusement, au lieu de résoudre le problème de
support, ce principe n’a fait qu’aggraver le problème de contrôle de plafond. Eventuellement une
idée s’est dégagé allant du fait que les piliers endommagés de taille moyenne qui se sont avérés
être efficaces ailleurs, pourraient fournir un indice à ce particulier problème.
Le but de cette recherche est celui d’étudier les différent types de ruptures qu’un pilier peut
occasionner lorsqu’il est utilisé a grande profondeur, tout en utilisant deux possible méthodes. La
première d’entre elle comprend quelques solutions élémentaires obtenues avec l’aide d’une
méthode simplifié et conceptuelle de la masse de la roche. Ce modèle consiste à prendre en
sandwich la bande de minerais de charbon entre deux masses stratigraphiques sans pour autant
endommager le plafond. La seconde méthode nécessite l’utilisation du ‘’ FLAC3D’’, qui est un
logiciel commercialisé disposant d’un modèle numérique en 3 dimensions. Dans le but d’obtenir
1
un meilleur résultat, les lois de ‘’Strain-softening’’et idéalement celle constitutive de Mohr-
Coulomb compris dans le logiciel ‘’FLAC3D’’ont été utilise.

Comme prévu, la proportion entre la largeur et la hauteur des piliers a démontré être un facteur
majeur qui détermine le comportement essentiel de la manière dont les piliers réagissent. Les
piliers de taille étroite se fracturent de manière stable sans pour autant occasionner d’autres
instabilités. Par ailleurs, un pilier de proportion moyenne (entre la largeur et la hauteur) sous des
conditions appropriée de poids ou de masse peut occasionne des ruptures soudaine ou encore
s’effondrer totalement. Les piliers qui sont petit et épais représentent un potentiel danger pour une
soudaine explosion au alentour du pilier de charbon. Toutefois due au fait que la plus grande partie
de la largeur du pilier demeure intact, les roubles de charbon autour du pilier procurent assez de
confinement à tel point qu’ils rétablissent une stabilité.
Finalement, l’étude a révèlee que la traditionnelle loi de ‘’Mohr-Coulomb’’ ne fourni pas assez
d’information qui mène a la rupture des piliers de charbon.

ZUSAMMENFASSUNG: Die Anwendung von Kohlenpfeilern in großen Teufen ist behaftet mit
Problemen aufgrund Schwierigkeiten bei der Beherrschung der Firste und Gebirgs- und
Flözschlägen. In den USA werden Pfeiler hauptsächlich im Strebbau angewandt und dienen in
Reihen zur Sicherung der mehrfachen Eingänge von Abbaustrecken. Ursprünglich wurden solche
Pfeilerreihen mit konventionellen Methoden dimensioniert, wobei die Festigkeit eines Pfeilers
größer sein sollte als die Last auf dem Pfeiler. Dies führte zu immer größeren Pfeilerausmaßen mit
zunehmender Teufe und die Schwierigkeiten bei der Standsicherheit der Firste nahmen eher zu als
ab. So wurde die Idee geboren, dass klein dimensionierte Bruchpfeiler, die sich schon in anderen
Bereichen bewährt hatten eine Lösung zu diesem dringenden Problem darstellen könnten.
Diese Arbeit hat zum Ziel das Bruchverhalten von Pfeilern in großen Teufen unter Awendung von
zwei Lösungsansätzen zu untersuchen. Der erste Ansatz beinhaltet einige grundlegend Lösungen,
die mit einem vereinfachten Modell des Gebirgsverbandes erlangt wurden. Dieses Modell besteht
aus einem Kohlenflöz mit mechanischer Entfestigung, der zwischen zwei geschichteten Körpern
eingespannt ist. Im zweiten Ansatz wird das frei erhältliche Computerprogramm Flac3D zur
numerischen Simulation genutzt. Zum besseren Verständniss wurden sowohl die Modelle mit
mechanischer Entfestigung, als Mohr Coulombs ideal plastisches Fesigkeitsmodell, die in Flac3D
enthalten sind, untersucht.
Wie erwartet hat sich herausgestellt, dass das Verhältnis von Pfeilerbreite zu Pfeilerhöhe den
größten Einfluss auf das Verhalten der Pfeiler hat. Schmale Pfeiler brechen in einer stabilen Weise
ohne das Auftreten von Instabilitäten. Pfeiler mit einem mittleren Breiten/ Höhen Verhältniss
können unter den entsprechenden Belastungszuständen plötzlich versagen und sogar ganz
zussammenbrechen. Breite Pfeiler haben ein Potenzial für plötliche, flözschlagartige Seitenbrüche,
doch aufgrund der großen Restbreite des Kerns bietet das Geröll um den Pfeiler herum genügend
Einschluß um die Standsicherheit wieder herzustellen. Letztlich wurde festgestellt, dass das Gesetz
von Mohr- Coulomb keine nützlichen Hinweise zu dem Bruchverhalten von Kohlenpfeilern bietet.

idea was born that in relatively deep seams yielding pillars


Introduction might prove to be more successful. Several field
experiments appear to substantiate this supposition. As
The obvious application of coal pillars at depth is related to yielding pillars are of relatively low width to height ratio,
longwall mining. In many instances the economic success clearly there is a sharp discord between these two
of underground mining of coal in the United States hinges approaches to chain pillar design. To resolve this conflict
on the application of the longwall method. The longwalls and to ensure that yield pillars do not create inherently
in the country usually retreat and they are developed using dangerous conditions, it is necessary to investigate and
some multi-entry system. The rows of pillars separating the eventually to understand the mechanisms involved in the
entries in this system, the chain pillars, play an essential loading and failure of coal pillars. Previous incidents have
role in the operation. The correct design of these pillars is a demonstrated that pillar failures can be gradual or sudden
key factor to the safety of longwall extraction. and even violent. The aim of this paper is to throw some
Initially conventional design of pillars was employed by the light on the design of chain pillars with special reference to
operators. This approach to design has proved to be their stability.
reasonably effective in relatively shallow depths, but it has
led to pillars of increasingly larger width to height ratios as Basic Principles
the depth of mining increased. Experience has shown that
this approach is often ineffective if the depth of the seams It is obvious to experienced miners that the sudden or
exceeds about 400 m. Such deep seams are frequently violent events represent the most serious hazard, because
encountered in the western United States; consequently, they occur unexpectedly, without warning. Clearly, the
several trials have been conducted in this region by mine investigation of the behavior of yielding pillars would not
operators to find a solution to this problem. Eventually the be complete without an examination of the modes of their
failure.
2 ISRM 2003 – TECHNOLOGY ROADMAP FOR ROCK MECHANICS
It was discovered in the mid 1960’s (Cook 1965:389; Deist
1965:502) that rocks can sustain load even when they are Elementary Studies in Two Dimensions
deformed beyond the strain corresponding to their
maximum resistance, or strength. The stress-strain curve of The simplified model has the property that the stress and
rocks has two branches: (a) the ascending branch where the displacement components (vertical, horizontal and shear
rock’s resistance increases with increasing straining and (b) stresses, the roof to floor convergence and the horizontal
the descending or strain-softening branch, where the displacement) induced by partial extraction of the coal can
resistance diminishes with increasing strain. Further be defined in the seam in terms of a system of ordinary
studies have revealed (Starfield and Fairhurst 1968:78; differential equations. In the case of elastic and Deist’s
Salamon 1970:613) that the mode of the rock failure strain softening behaviors, since both of these are
strongly depends on the stiffness of the loading mechanism essentially linear, this system of equations can be solved
and the slope of the descending branch of the rock’s stress- analytically, so the stress and displacement components can
strain curve. This slope, or the rate of a rock’s strain- be expressed in terms of analytical formulae. This feature
softening, is a measure of the substance’s brittleness. If the permits the strict analysis of the solutions and the detection
loading system is stiff relative to the rock’s rate of strain- of unusual or unexpected features arising in them. Since
softening, the failure of the rock will be gradual or stable. the method of solution is not dynamic, the instabilities
Conversely, if the loading system is soft relative to the manifest themselves as singularities in the quasi-static
softening of the rock, the failure will be violent or unstable. solutions. Due to space limitations the analytical solutions
Thus, rocks in their descending or strain softening branch are not presented here, but they can be found in an earlier
are only conditionally stable. Since the discovery of these publication (Salamon 1992: 94).
basic principles some progress has been made, but the Empirical field trials have suggested that narrow chain
understanding of the phenomenon is far from perfect. A pillars (width/height < 4) tend to be associated with
significant fraction of the available progresses are related to relatively low stresses and their deformation is gradual and
the global stability of pillar workings enunciated by free of sudden changes. As the width to height ratio is
Starfield and Fairhurst (1968:78) and Salamon (1970:613). increased, the behavior changes and, due to the higher load
These analyses are global in a sense that individual pillars bearing capacity of the squat pillars, the magnitude of
are treated as single units without regard to the distribution stresses increase and instabilities start to emerge.
of stress and strain in them. This approach provides an Four solutions obtained with the aid of the simple model
explanation of catastrophic collapses involving many, are discussed. The first three cases involve a longwall
sometimes even hundreds of pillars, but cannot clarify the development with two entries, which are separated by a
mechanisms involved in localized failures, such as the long strip pillar. The simplicity of this geometry is a
bumping of coal from the side of a pillar or roadway. consequence of the 2D solution. In all cases the physical
Thus, if progress is to be made in rational yield pillar properties and geometrical parameters (the numerical
design, it is necessary to evolve methods of analysing the values of these variables are summarized in the Appendix)
behaviour of yielding pillars, up to the moment of are identical, with one exception. The pillar width, W, is
instability, taking into account the detailed distribution of different in the three cases (W = 6 m, W = 10 m and W =
stress and strain in and around them. 20 m). Since the seam thickness is taken to be 2 m, these
This is a formidable task, since only dynamic analysis that widths give the following width to height ratios: 3, 5 and
accounts for inertia forces can tackle satisfactorily the 10. The seam is assumed to be at a depth of 700 m, which
stable and unstable behavior of structures. In the case of by coal mining standards is deep level extraction. At this
complex structures, which can be depicted adequately only depth the virgin vertical stress is 17.5 MPa. No attempt is
in three dimensions and involve longwall(s), chain pillars made in these elementary solutions to simulate the effects
and entries, dynamic analysis is not practical at present, of the extraction of the longwall panels. The pillars are
hence a less than perfect substitute method needs to be loaded by simply widening the entries. While this approach
found and applied. does not faithfully reproduce the mining geometry, it does
Two methods will be presented to illustrate how a non- offer the opportunity of analysing the behavior of pillars of
dynamic approach can be used to gain insight into the various widths to height ratios under differing loads.
problem. First, using a simplified model, a few two The fourth elementary solution involves a single panel
dimensional example problems are tackled. This model which corresponds to one of the entries in a longwall
consists of an upper mass (or roof) and a lower mass (or development. The ribsides delimiting the span of the panel
floor) which are both stratified and elastic. Between the can be viewed as a pillar of infinite width, so in this sense
two masses a coal seam is sandwiched, which exhibits this geometry can be regarded as the limiting case of the
elastic and strain softening properties. The behaviour of the development with two entries when the width/height ratio
seam follows the strain softening concepts proposed by of the chain pillar is increased without bounds.
Deist (1965). This construction of the model was proposed Narrow pillar: width to height ratio is 3. The model
by Salamon (1991:9 and 1992:94). The other approach reveals that in the specified circumstances the strength of
involves extensive numerical studies using FLAC3D (Itasca this pillar is 19.25 MPa. This value is only slightly higher
2002). This approach allows the treatment of realistic than the virgin stress, so it can be expected that the pillar
geometries. Both strain-softening and Mohr-Coulomb edges will start to yield immediately as the entries are
constitutive laws are explored. developed. This means that the pillar initially will have an

3
elastic core, which is surrounded by yielding coal. In 3.8
Figure 1 the depth of yielding is illustrated as a function of
3.6
the width of the entries (i.e., loading). It is noteworthy that

Depth of yielding into pillar sides (m)


narrow, 5 m wide entries already create yield zones that are 3.4
Max. yielding depth
2.29 m in width. If the width of the entries is increased to
6.45 m, a change of only 1.45 m, the two yield zones meet 3.2
at the center of the pillar, so the coal throughout the full
3
width of the pillar will be in the descending branch of its
stress-strain curve. At this moment the load bearing 2.8
capacity of the pillar is reduced to 14.7 MPa. A further
widening of the entries will induce additional deformation 2.6 Max. entry
width
and additional loss in the load bearing capacity of this
2.4
relatively narrow pillar. Thus, this narrow pillar deforms
gradually and does not exhibit any sign of sudden or violent 2.2
deformation. Also, the stresses imposed on the roof and
floor by the resistance of the pillar are modest. These are 2
important and favorable features of narrow pillars. The 7 8 9 10 11 12
main disadvantage of these support elements is their Width of entry (m)
relatively low load bearing capacity after yielding. This
feature suggests that narrow yield pillars should be Figure 2. Depth of yielding versus entry width for a w/h = 5
restricted to developments involving one row of chain pillar.
pillars with the associated two entries.
Intermediate pillar width: width to height ratio is 5. The dary between the yielding and still elastic coal inside the
strength of this pillar is 28.6 MPa, which is reached when pillar becomes unstable and consequently, the yielded pillar
yielding penetrates into the sides of the pillar to a depth of edges are likely to disintegrate. Due to the instability this
2.95 m. This level of loading is achieved when the width of process is likely to be sudden and results in an event that
the entries is 10.6 m. Further widening of the entries may be described as a ‘coal bump’. The disintegration of
induces a growth in the depth of yielding with concomitant the edge of the pillar results in a loss of pillar core
reduction of the load bearing capacity of the pillar. In confinement. This, in turn, is likely to initiate rapid inward
Figure 2 the depth of yield as a function of the width of the progression coal failure. The rubble produced by the
entries is depicted. The relationship reveals an accelerating process of disintegration rests on the floor and through
yield penetration and shows an unexpected feature. At an friction attempts to provide confinement to the remaining
entry width of 11.2 m the rate of yield penetration runs part of the pillar. The final fate of the pillar depends on the
away, that is, the slope of the curve in Figure 2 becomes width of the elastic core at the time when the instability
vertical; the solution has a singularity. This is an entirely occurs. If the core is of sufficient width the pillar will
new attribute of pillar behavior. stabilize. If, however, the core width is insufficient, the
No similar behavior was observed in the case of the narrow whole pillar will collapse. In the present example the width
pillar. In practical terms this result indicates that the boun- of the core is 10 - 2×3.45 = 3.10 m, which appears
insufficient to allow the stabilization of the pillar, so it is
3.2 likely that the pillar will collapse. Interestingly, this
analysis reveals that the widening of the pillar leads to an
unexpected result. The squatter pillars provide higher
Depth of yielding into pillar sides (m)

3
resistance to ground movement which in certain
circumstances, especially if the roof and/or floor are
2.8 intensely laminated, could induce a serious deterioration in
the strata conditions. Furthermore, under certain
2.6 circumstances, the edge of a squatter pillar may fail
suddenly in a manner that seems to resemble a coal bump.
With intermediate width pillars, such sudden pillar edge
2.4
failure may in turn causes the collapse of the whole pillar.
Squat pillar: width to height ratio is 10. The strength of
2.2 this pillar, according to the model employed here, is 33.6
MPa and this strength is attained when the pillar sides have
2 yielded to a depth of 3.67 m. The behavior of this pillar is
5 5.25 5.5 5.75 6 6.25 6.5 similar to that of the previous one, because, as the entries
Entry width (m) are widened and the depth of yield increases, eventually
here again a situation is reached when the boundary inside
Figure 1. Depth of yielding versus entry width for a w/h = 3 the pillar between the elastic and yielded part of the seam
pillar. becomes unstable. This situation is illustrated in Figure 3.

4 ISRM 2003 – TECHNOLOGY ROADMAP FOR ROCK MECHANICS


4.2 4.4

4 4 Max. yielding depth


Max. yielding depth
Depth of yielding into pillar sides (m)

3.8 3.6

Depth of yielding in ribsides (m)


3.6 3.2
3.4 2.8
3.2
2.4
3
2
Max. entry
2.8
width 1.6
Max. panel
2.6
1.2 width
2.4
0.8
2.2
0.4
2
12 13 14 15 16 17 18 19 20 21 0
0 5 10 15 20 25
Width of entry(m)
Span of panel (m)

Figure 3. Depth of yielding versus entry width for a w/h =


10 pillar. Figure 4. Depth of yielding versus panel width for the
ribside (infinite w/h ratio pillar).
The critical depth of yield in this instance is 3.87 m and the
corresponding load bearing capacity of the pillar is 33.46 instance is 23.4 m, which is only 14.5% greater than the
MPa. This loading is achieved at an entry width of 20.05 critical entry width in the development supported by the
m. Note that the instances of maximum load bearing and squat pillar that has a width to height ratio of ten.
instability are very close to each other. It is noteworthy that If the critical span of a single entry is known it is possible
the tributary load with this geometry is 35.04 MPa. to estimate, at least in principle, the strength of squat
There is likely to be a significant difference between the pillars. However, because the so obtained estimate is an
behavior of the intermediate and the squat pillars. In the upper bound, this method is of little practical significance.
latter case the width of the elastic central core at the
moment of instability is 20 - 2×3.87 = 12.26 m, which Numerical Studies with FLAC3D
appears to be sufficient large to ensure that the rubble at the While the simplified model of the previous section does
edges can stabilize the pillar. So in the case of large width provide an insight into the possible behavior of yielding
to height ratios coal bump type of pillar edge failure may chain pillars, it is unlikely to provide results that can
occur, but the pillars themselves are unlikely to collapse. withstand comparisons with the behavior of the prototype
Conditions around a single panel. After the consideration in the field. Thus, to get closer to a quantitative method of
of wide pillars, it seems reasonable to suppose that the design, it is necessary to employ a sophisticated stress
edges of such a pillar must behave similarly to the ribsides analysis package to perform the necessary numerical
of a single panel that has a span comparable to the width of studies. For the present case, FLAC3D was chosen as the
the entries in the longwall development. Clearly these numerical code because of its flexibility in employing a
ribsides can be viewed conceptually as pillars of infinite variety of constitutive laws and failure criteria, and also due
width. Next, this notion is tested using the model to pillars being three-dimensional structures.
employed up to now. Assuming that all parameters remain To keep the analysis relatively simple, it was assumed that
unchanged, the conditions around a single panel of span S non-linear behavior is restricted to the coal bed. It was
are examined. decided to compare results obtained using two models as
Postulate that the panel is first developed with a narrow alternatives to describe the post failure behavior of the coal.
span and then gradually widened in a manner similar to the The alternatives used were the Mohr-Coulomb and strain-
widening of the entries in the previous examples. First, the softening models. Both of these are readily available in
rearrangement of stresses and strains around the panel FLAC3D.
accompanies readily the increasing span. Notably the depth Results relating to pillars in the central region of an
of the yield zones into the ribsides increases smoothly as extensive room and pillar workings are reported on in this
seen in Figure 4. This illustration depicts a curve that is paper. As the central region is postulated to be remote from
comparable to the curves already seen in Figures 2 and 3. the boundaries of extraction, here each pillar is subjected to
Observe that there is a critical span, Sc, at which the yield identical loading conditions. Both strain-softening and
penetration curve has a singularity and this span is the Mohr-Coulomb type models have been investigated. In the
upper bound of the panel width that can be supported by the present paper there is opportunity to discuss only a small
ribsides without collapse. part of the obtained results. As in the previous section, the
This collapse is associated with instability so it could focus remains on the modes of pillar failure. Incorporation
manifest itself as a coal bump. The critical span in this of results into full longwall mine layouts is given elsewhere
(Badr 2003).

5
In this example, a 3D domain was used to simulate pillars
of square cross-section with width/height ratios of one to
eight. There are several planes of symmetry in the problem
Figure 5. Three of these were taken advantage of when
constructing the model. Two of these are obvious and a
third is not so obvious. The obvious symmetries occur at
the two perpendicular axes of the imprint of the pillar on
the floor. The third plane of symmetry is the horizontal
median plane of the pillar. Strictly speaking this is not a
symmetry plane, because the presence of stress free ground
surface but, because of the large depth, this discrepancy is
negligible. In the presence of these planes it is sufficient to
model only one eighth of the pillar. The interaction of the
modelled pillar with the surrounding room and pillar
system can be simulated through the specification of
appropriate boundary conditions. The aim of the exercise is
to determine the behaviour of the pillar both before and
after the point when its maximum resistance, that is its
strength, is overcome. Consequently the pillar must be
loaded by displacement, that is, by the uniform lowering of Figure 6. Average pillar stress-deformation relationship
the mid plane of the pillar. The room width at 6.5 m and using strain softening criterion.
the height of the pillar, which was taken to be 1 m,
remained fixed throughout the runs and the various widths curves corresponding to ratios four and higher show a
to height ratios were achieved by changing the widths of pronounced drop in resistance, which is followed by
the pillars. oscillations.
This behavior appears to suggest that these pillars show
instability similar to that exhibited by the intermediate
Loading
width and squat pillars examined in the previous section.
Unfortunately it is not possible to be absolutely certain
Entry 3.25 m whether the same phenomenon is being observed. The
0.5 m difficulty arises from the numerical nature of the solution.
Coal While it can be expected that at the moment of instability
FLAC3D would not converge to a stable solution, however,
Cross-cut
the possibility that a stable solution is found at some point
5m beyond the occurrence of a local instability cannot be
excluded.
To get a better appreciation of the behavior of these pillars
a closer examination was performed next. The stress-
deformation relations of zones along the diagonal of the
pillar were examined for different width to height ratios. In
Figures 7 and 8 the behaviors of the zones are shown for
width to height ratios of two and six, respectively. The first
of these illustrations (Figure 7) indicates that the
deformation of all zones in the narrow pillar (w/h=2) is
Figure 5. The room and pillar model. smooth; the curves corresponding to all zones are smooth
and well behaved. Thus, the behavior of this pillar is
Strain-Softening behavior similar to that observed when analyzing the narrow pillar in
The load-deformation curves of pillars with varying the previous section employing the simplified model.
width/height ratios are plotted in (Figure 6). Here the mean However, the curves depicting the deformation of some
stress values are plotted as the function of the mean value diagonal zones are no longer smooth in the wider pillar
of pillar compression. The curves are depicted for integer (w/h=6) (Figure 8). Over zones numbered between 10 and
values of the width/height ratio from one to eight. This 15 (see Figure 9) for zone numbering sudden drop in
family of curves reveals quite strong trends. All curves resistance occurs. The curves relating to the other zones
show, as expected, the ascending and the descending remain regular and smooth. The suddenness of the stress
branches. The peak values that separate the branches drops in the easily recognizable region suggests the
represent the pillar strengths. The strength increases presence of an underlying instability; although
markedly with increasing width to height ratio. The strain unfortunately this supposition cannot be confirmed with
softening or the descending branches deform smoothly up certainty.
to about width/height ratio of three. All descending
.

6 ISRM 2003 – TECHNOLOGY ROADMAP FOR ROCK MECHANICS


30
29
28
27
1 26
0.9 Numerals refer 25
to distance 24
23
from pillar rib in 22
Zone vertical stress (MPa x 10)

21
meters. 20
19
18
0.7 17
16
15
14
13
12
11
0.5 10
9
8
Sloughed zones 7
6
Stress-drop zones 5
0.3 Elastic zones 4
3
1 Zone number 2
0.1 1

Elastic core size changes with w/h ≅ 0.5m ≅ 1m

Zone vertical deformation (m x 10-2)


Figure 9. Behavior of zones mapped at one quarter of w/h =
5 pillar using strain softening criterion.
Figure 7. Behavior of zones across the diagonal for w/h = 2
pillar.
so striking that it is reasonable to postulate that instabilities
occur in both models.
Numerals refer to
distance from pillar Mohr-Coulomb representation of pillars
Zone vertical stress (MPa x 102)

rib in meters.
For the sake of comparison, the same problems have been
recalculated by assuming that the failure of the coal is
controlled by the traditional Mohr-Coulomb criterion. A
comparison of Figures 6 and 10 reveals that the
fundamental differences which are known to exist between
the strain-softening and Mohr-Coulomb approaches
1.1 manifest themselves in the behavior of pillars as well. A
1 1.3 1.4
1.2
load deformation curve of a pillar obtained by employing
the strain-softening model defines a unique value for the
strength of the pillar. This strength is given by the pillar’s
peak resistance, which is the value that separates the
ascending and descending parts of the resistance curve. In
Zone vertical deformation (m x 10-2) contrast, the load deformation curves obtained using the
Figure 8. Behavior of zones across the diagonal for w/h = 6 Mohr-Coulomb model does not exhibit peak values,
pillar. because the curves do not have descending branches.
Furthermore, the load-deformation curves in Figure 10 are
The repetition of calculations for a series of width/height all well behaved; no sign of any type of singularity can be
ratios reveals a repeatable pattern. Pillars of width to height detected. This behavior is consistent with the properties of
ratios 5, 6, 7, and 8 all show similar stress drops over the the ideally plastic Mohr-Coulomb model. It would appear
same zones (Figure 9). therefore, that the approach to modelling which is based on
Geometrically this means that the region of stress drop is the Mohr-Coulomb concept has little or no value when the
essentially in a fixed position relative to an edge or a corner goal of the investigation is the clarification of pillar failure.
of the pillar. For the sake of comparison, recall now the Also, this approach cannot be of any utility if the goal is to
depths of yield penetration from Section 3. The critical investigate the prospect of coal bumps.
depths of yield penetration are 3.45 m, 3.87 m and 3.89 m
for three-meters height pillars of 5, 10 and infinite width to Conclusion
height ratios (note that the pillar height in Figure 9 is one
meter). While these values are not identical, they are In summary, a coal seam sandwiched between stratified and
undoubtedly close to each other and largely independent of elastic roof and floor masses, reveals that three different
the pillar’s width to height ratio. Thus, it can be said that chain pillar behaviors can be distinguished in longwall
the instability in the simplified model also occurs in a development. The single most important factor that
location that, for practical purposes, remains unaltered by influences behavior is the width to height ratio.
pillar geometry. Hence, the similarity between FLAC3D At low width to height ratios the pillar deformation is
results and the behavior observed in the simplified model is controlled, the yielding zones progress towards the center

7
Acknowledgment
This publication was supported by Cooperative Agreement
number U60/CCU816929-02 from the Department of
Health and Human Services, the Center for Disease Control
and Prevention (CDC). Its contents are solely the
responsibility of the authors and do not necessarily
represent the official views of the Department of Health and
Human Services, CDC. Support provided by Department
of Health and Human Services, CDC, is greatly
acknowledged. The work presented is part of the Health
and Safety research activities currently carried out at
Western Mining Resource Center (WMRC) at the Colorado
School of Mines.

References
Badr, S. A. (2003): Numerical Analysis of coal yield pillars
. at deep longwall mines. Ph .D. Thesis in preparation.
Figure 10. Average pillar stress-deformation relationship Department of Mining Engineering, Colorado School of
using Mohr-Coulomb failure crieterion. Mines, Golden, Colorado.
Cook N.G.W. (1965): The failure of rock. Int. J. Rock
of the pillar smoothly and a pillar that is yielding Mech. Min. Sci. Vol. 2, pp. 389-403.
throughout its width can be created readily. If the depth of Deist F.H. (1965): A nonlinear continuum approach to the
mining is great, this fully yielding state can be reached problem of fractured zones and rockbursts. Jl. S. Afr. Inst.
during the development of the entries. Hence, these narrow Min. Metall. Vol. 65, pp. 502-522.
units are ideal choices as yield pillars. Their only FLAC3D (2002): Itasca Inc. Minnesota, USA.
shortcoming is that their load bearing capacity is low, so Salamon M.D.G. (1970): Stability, instability and design of
they should be considered only in development involving pillar workings. Int. J. Rock Mech. Min. Sci. Vol. 7, pp.
two entries. Essentially the same picture emerges, at least 613-631.
qualitatively, regardless whether the results of the Salamon M.D.G. (1991): Deformation of stratified rock
simplified or FLAC3D modelling are examined. masses: A laminated model. Int. J. Rock Mech. Min. Sci.
Pillars with intermediate width to height ratios have a Vol. 91, pp. 9-25.
number of weaknesses. Firstly, it would appear that these Salamon M.D.G. (1992): Strength and stability of coal
pillars cannot be brought into a fully yielding state, because pillars. Proceedings of the Workshop on Coal Pillar
their failure process becomes unstable at a certain depth of Mechanics and Design. Santa Fe, 1992. US Bureau of
yield penetration. This critical moment is associated with a Mines, Information Circular/1992, IC 9315, pp. 94-121.
stage of loading which may be attained in the course of Starfield A.M. and Fairhurst C. (1968): How high-speed
development of the entries or only later when the load computers can advance design of practical mine pillar
transferred from a longwall panel becomes sufficiently systems. Engineering/Mining Journal. Vol. 169, pp. 78-84.
high. This instability may induce a coal bump like event
and even a total collapse of the pillar, with the concomitant Appendix
hazards of such eventualities. Ideally pillars with
Summary of the data used for calculations in the two
intermediate width to height ratios should be employed in
dimensional studies, are as follows:
situations where the depth of yield can be kept below the
Modulus of the rock mass (E): 20 GPa; Modulus of the coal
critical value. This suggests that their utilization should be
seam (Es): 3 GPa; Poison ratio of coal seam (νs): 0.2;
restricted to coal seams under relatively shallow cover.
While squat pillars are unlikely to suffer total collapse, Specific weight of the rocks (γ): 25 kPa/m; Horizontal to
they are not immune to sudden failure of their sides. For vertical virgin stress ratio (k): 0.7; Working height: 2 m;
this reason and also, because they adversely influence coal depth below surface: 700 m; Internal angle of friction (ρ):
recovery, they are not ideal as chain pillars in deep longwall 30o; Uniaxial compressive strength (σc): 5 MPa;
mining.
Coefficient of friction (between seam and roof/floor) (µ):
The discussion avoided the quantification of the terms
‘narrow’, ‘intermediate’ and ‘squat’. This is a deliberate tan (20o); Normalized measure of strain softening (δ): 0.05.
omission, which unfortunately is unavoidable at this stage The negative slope is given by:
of development. It can be stated, however, observations in
the field do support the conclusions, at least in a qualitative − 2δE s
sense. E neg = = −239 MPa
(1 − 2ν s + sin 30)(1 + ν s )(1 − δ)

8 ISRM 2003 – TECHNOLOGY ROADMAP FOR ROCK MECHANICS

You might also like