17th International Mining Congress and Exhibition of Turkey- MCET 2001, ©2001, ISBN 975-395-417-4

Yielding Pillar Concept and its Design

H.Yavuz
Department of Mining Engineering, Süleyman Demire! University, İsparta, Turkey

ABSTRACT: Yielding pillar theory and the comparative performance of yield-pillar-protected roadways to
critical and stable pillar systems are described. The applicability of empirical methods, the mine stiffness
concept, enhanced confined core concept and numerical modelling in designing a yield pillar during the
development stage of mining in two-entry systems was investigated and the drawbacks of each method are
outlined. These methods were compared by evaluating published data from field measurements. Finite
difference models were arranged for estimation of strata stiffness at the pillar location for comparison with the
post-failure slopes of pillars of 5, 7.5 and 10 m İn width, which are the intended preliminary design for a UK
coal mine. The estimated pillar width ranges of the methods for this example were compared. The findings
are verified by those from previous investigations.

1 INTRODUCTION such a pillar with a view to safer supply gate

location (Figure 1) is sought by evaluating various
The safety and productivity of the longwall mining design methods and previous measurements of
method depend on the maintainence of ground different field site applications. A finite-difference-
control in the gate entries. The stability of a method-based two-dimensional code, FLAC, was
development roadway serving a new panel is also used to determine the local mine stiffness by
ensured by leaving a stable pillar of such a width applying force to the roof for various pillar width
that the stresses induced by the nearby excavation do configurations. Numerical analysis was also
not significantly influence the level of original virgin performed for the same configurations in order to
vertical stress over the working region of a new investigate the magnitude of vertical stresses over
panel. As longwall mining progresses to greater the pillar after yielding.
depths, conventional stable pillar designs require
greater and greater pillar width. Widths in excess of
60 m are necessary to provide ground control under
600 m depth of cover. The narrow pillar, designed to
yield during longwall mining operations, has been
popular especially in the US and recently in the UK.
It is used to increase productivity and to minimise
problems associated with mining at greater depths.
The pillar between the gate entries is designed such
that the width/height ratio provides a destressed zone
over the supply gate of the new panel by gradual
yielding and transfer of the potentially dangerous
stress concentrations to the adjacent pillar.
This work arose from the necessity of leaving a
pillar between the gate entries of extraction and
development panels in Bilsthorpe coal mine. This
was the case after an extensive fall of the supply
gate of a development panel driven immediately
adjacent to the loader gate of an excavation panel
had occurred. First, the theoretical basis for yielding
Figure 1. The layout of workings in Bilsthorpe colliery for
pillars is explained. Then, a preliminary design for preliminary design of a yield pillar
397
2 YIELDING PILLAR THEORY sizes of 12.2 m and 10.6 m were found to be
yielding pillars (Koehler et al., 1996). Studies İn
The yielding pillar concept was first developed for British coal mines have shown that narrow pillars
room and pillar mining to create a destressed zone in with widths in the range of 10-30 m result in a high
the working area. The size of the pillars is degree of gate roadway closure, producing a 70%
determined so that yielding of the pillars is ensured change in cross-sectional area (Whittaker & Singh,
and loads are transferred from the working area to 1979). Figure 3 shows that when very large
the adjacent barrier pillars. The yield pillar system conventional pillars were used, better conditions
has proven more stable and more economic than were common. However, yield pillars exhibit better
abutment pillars at mining depths greater than 450 performance than critical-size pillars.
m. However, there have been documented cases of
yield pillar successes at mining depths of less then
450 m (Kripakov et al., 1994). A conceptualised
relationship between yield, critical and stable pillars
in terms of gate road performance is illustrated in
Figure 2. The horizontal axis represents the
minimum performance standard differentiating
stable gate road configurations from unstable
configurations. A pillar design whose performance
falls above the horizontal axis is considered
successful, while a design whose performance falls
below the horizontal axis is considered unsuccessful.
The deterioration of ground conditions is generally
more gradual for abutment pillars as pillar size
decreases. Changes in performance are witnessed by
the onset of minor floor heave, an increase in Figure 3 Statistical analysis of roadway closure for various
audible coal popping and an increase in the pillar widths (Whittaker & Singh, 1979)
frequency of roof-related problems (Koehler et al.,
1996).
Holland (1973) studied the pressure arch concept
for board and pillar mining and suggested that a
pressure arch develops between the two barrier
pillars in a panel when the pillars are designed to
yield. İn this approach, the yield pillars are expected
to redirect the overburden stresses to the solid
abutments, thereby allowing greater extraction ratios
within the panels. In the development stage of
longwall mining, the pillar will be loaded by the
tributary area theory. Where yield pillars are left
under an intact pressure arch, the effective upper
boundary of the tributary area is the bed separation
limit. The pillar can be designed to yield either at the
development stage or at the longwalhng stage.
Figure 4a shows that the pillar is too wide to yield
and yielding occurs at the ribs; the load carried by
Figure 2. Conceptualisation of the yield pillar concept
the yielded section is probably transferred to the
inner pillar elastic core and to the abutment sides. In
Incorrect sizing of yield pillars can result in the second case (Figure 4b), the pillar is not
worsened entry conditions. Mark (1990) stated that proportioned so that it will retain sufficient
between yield and abutment pillar sizes, there are flexibility and not pick up the full overburden load.
intermediate pillar sizes that are too stiff to yield but The pillar load increases when die size of the entries
also too small to redistribute stresses effectively is increased. However, tensile failure probability at
within themselves. Such pillars, called "critical the mid-span of the entry restricts the size of the
pillars", maximise disturbance of the surrounding entry. Failure of the pillar should occur İn a non­
ground. Field observation and stress measurements violent manner and it must maintain enough residual
in a US coal mine showed that a 16.8-m pillar width strength to support the weight of the rock within the
at a depth of 800 m is a critical pillar design. Pillar pressure arch.

398
 adjusted so that yielding is ensured and that the
width-height ratio should be adjusted so that it has
high residual strength, maintaining the stability of
the roadway. The failure and load-carrying capacity
of a pillar depends on the intact and broken strength
properties of the coal, the width-height ratio of the
pillar, the rock properties and the properties of the
rock-coal interfaces for a site-specific investigation.
Current methods do not consider all of these factors.
However, they can be applied to specific cases with
the necessary experience in die field.

3.1 Empirical methods

In a two-entry system, the design of a yield pillar
may be accomplished using empirical formulae and
by determining a mine-specific safety factor through
field experience. All empirically derived equations
can be written in two types of expression.
Figure 4 Red İ sin button of ground stresses for different pillar
widths due to entry development (1)

(2)
In the longwalling stage, cantilevered beds above
the caved zone load the rib of the coal seam. These
loads can cause excessive damage to the roadway where Sp is die pillar strength in MPa, Si is the in-
depending on pillar behaviour. A theory was situ strength of coal in MPa, wp is the width of the
developed in the UK which states that a destressed pillar in m, hp is the pillar height m m and A, B, a
zone develops inside an ellipsoid, while outside the
ellipsoid the stress is high (Alder et al., 1951). The and b are constants expressing the shape effect
width of the ellipsoid 'b' shown in Figure 5 depends (Table 1).
upon the cover depth. The vertical width of the
ellipsoid is twice the horizontal width.
Table I. Details of pillar strength equations.
Author A B a b

Obert& Duval! (1967) 0.778 0.222

Holland (1973) - - 0.5 05
Bieniawski (1984) 0 64 0.36
Salamon & Munro ( 1967) - - 0.46 0 66

The empirically derived equations given above

can predict the overall strength of squat pillars. The
strength of infinitely long and rectangular pillars can
be significantly greater than that of square pillars
due to the greater confinement generated within
them. The strength of large rectangular specimens
may be expected to be the same as that of large
square specimens with a side length equal to the
effective width of rectangular specimens. Wagner
(1974) suggested the following equation for
estimating the effective width of long pillars:
Figure 5 Pressure arch concept for longwalling stage of mining.

(3)
3 DESIGN OF A YIELDING PILLAR

A pillar design method should simply take into where Arp is the area of the pillar and Upp is the pillar
consideration that the width of a pillar should be circumference. According to equation 3, the

399
effective width of an infinitely long barrier pillar is seen clearly. Wagner's tests proved conclusively that
twice the actual width. a pillar has a significant load-bearing capacity even
The load coming over the pillar İs calculated when its maximum resistance, which is traditionally
using the tributary area method, which is commonly regarded as the strength of the pillar, has been
used in room and pillar mining. This load is called overcome. This is the main area of interest in the
the development load and İs calculated as follows: design of yield pillars.

(4)

where wp is the pillar width (m), we is the entry

width (m), H is the-depth (m) and y is the unit
3
weight of the overburden (MN/m ).
Empirical methods can be successful provided
that correct selection of the safety factor, which is
the ratio of pillar strength to stress imposed on the
pillar, is made through field experience for each
specific formula. A comparison was made for a
depth of 600 m and a 2-m-thick seam. The strength
equations give varying safety factors for the same
pillar width as illustrated İn Figure 6. This difference
İs more significant when the width of the pillar
increases. Carr (1992) reported that safety factors in Figure 7. In-situ complete stress deformation curve and stress
the range of between 0.52 and 0.73 were successful. distribution of pillar with width/height ratio of I (Wagner,
The width of the pillar to yield ranges between 6 and 1974).
11.5 m. This range of widths İs in agreement with
successful field applications, as shown below.
Although the empirical design of pillars gives an The deformation characteristics of coal pillars
idea of the width-height ratio of the pillar where were investigated by means of underground tests on
yielding is expected, it does not estimate the failure large coal specimens which were carried out by Van
behaviour of a pillar that exhibits either gradual or Heerden (1975) in South African coal mines. The
sudden failure. results of these tests indicated that the deformation
characteristics, in particular the post-peak behaviour,
of coal pillars are not only a function of die coal
itself, but most importantly of the pillar geometry
(Figure 8). These tests can be used for the estimation
of stiffness estimation in pillars with width/height
ratios of up to 3.5.

Figure 6. Yield pillar width ranges for 2-m pillar height.

3.2 Mine stiffness concept

In-situ tests performed by Wagner (1974) clearly Figure 8. Effect of width/height ratio on the post-tailure slope
showed that the failure of a pillar is a gradual of coal pillars (Van Heerden, 1975).
process as shown in Figure 7. The distribution of the
load across the cross-section of the pillar varies as
the loading proceeds. Yield at the pillar edges and an Bearing in mind the yielding mechanism of
increase in the load bome by the pillar core can be pillars, Salamon (1970) showed that equilibrium

400
between the loading of a pillar and post peak pillar strata given in Figure 11 and the post-failure slope
resistance is stable, regardless of the convergence values are put into equation 5, an unstable condition
experienced by the pillar if; will be expected for 5 and 7.5-m. pillar widths. This
finding is questionable in terms of the predicted
(5) post-failure slopes for pillars with width/height
ratios of 3.75 and 5 due to the unavailability of data.
where K^ is the stiffness of the loading strata and
KP is the minimum slope of the post-peak load-
deformation relation for the pillar (both K^ and
KP have negative values). These cases are
illustrated in Figure 9, showing that the relation
between the stiffness of the strata and post-peak load
deformation determines the stability condition of the
pillar.

Figure 10. Modei conditions for mine stiffness estimation

Figure 9 Stable and unstable cases for a pillar depending on

stiffness of loading strata.

In assessing the stability of a mine structure, the

required information consists of the post-peak
stiffness of the pillar and die mine local stiffness at
various pillar positions. The stiffness of the strata at
an individual pillar location, local stiffness, is
described as the load deformation between the Figure 11 Stress & strain relation from roof to floor for 5, 7 5
hanging wall and footwall. The finite difference and 10-m pillar widths
method was utilised to find the stiffness of strata at a
pillar location at Bilsthorpe coal mine. The model
conditions are illustrated in Figure 10. The average 3.3 Enhanced confined core method
depth is 600 m. The deformation modulus of the The friction between the loading platens and the
rock mass is 14 GPa. The detailed input data for the specimen significantly affects the strength of the
Mohr-Coulomb plasticity model was described by specimen under uniaxial loading conditions This
Yavuz (1999). will be a primary factor in the generation of
The pillar was replaced by stresses for pillar confining pressure and increase in strength. The
widths of 5, 7.5 and 10 m as illustrated in Figure 10. Wilson confined core concept considers confinement
The strains between the hanging wall and footwall at of the pillar; how ever, this method İs based on an
the pillar location are plotted against the stress assumed horizont.il restraint and does not consider
applied on the roof (Figure 11). The post-failure the level of confinement for varying properties of
slopes of the 5, 7.5 and 10-m pillars were found to the coal-rock interfaces. It is a known fact that if the
be 0.73, 0.42 and 0.28 respectively from the friction angle is reduced due to filling material, the
equation given in Figure 8. If the slope values of the strength of the coal is reduced.

401
 Salamon (1992) developed an enhanced confined
core concept which takes the coal-rock interface
properties into account Failure at pillar edges (10)
depends upon the following criterion:

(6)
(11)
where k= , is the friction
angle of the coal, p İs the pillar side restraint, S, is The load on the pillar is estimated by using the
the in-situ compressive strength of the coal, and tributary area method in the following way:
is the maximum stress at the edge of the fully elastic
pillar, found by the following equation:
(12)

If the load q m exceeds the load-bearing capacity of

(7)
die pillar , then the pillar fails. If it does not
exceed the load-bearing capacity, then the pillar
where q is the vertical virgin stress, we = we / 2, we edges will yield, but the pillars will be able to
sustain the load.
is the entry width, wp = w p / 2 , wp is the pillar
The changes in the width of die pillar for failure
width (Figure 12), h is the seam thickness, and the were investigated for a depth of 600 ra, wheic S, is
constant 6 İs given by the following relation: 5 MPa,
3
is 0.025 MN/m , E is 10 GPa, Es is 2.5
GPa, is 0.3, is 7, h is 2 m and w e is 5 m. The
width of the pillar was calculated. This width is
given in Table 2 for different friction angles of coal
and interface and for different side restraints.

Table 2. The width of the pillar for complete yield for various
Figure 12 Geometry and notations for yield pillar design. friction angles for coal bedding friction a n g l e s a n d
edge restraints (p).
(deereet (degree) p(MPa) Wp(m)
35 25 0.1 8.4
30 25 0.) 10.9
(8) 35 20 0.1 9.9
35 25 0.02 10.9

where E, andt), are the Young's modulus and

Poisson's ratio of the seam. E is the Young's 3.4 Numerical method
modulus of the rock. X is a constant related to the As mentioned above, a yielding pillar provides a
deformation of the surrounding strata found from destressed zone around the entries by transferring
X= H / 2(û2, (0 is a constant related to surface the stresses from over the pillar to over abutment
subsidence. Salamon suggested values for to of 5-7 sides. The numerical modelling technique, when
for coal mining. If equation 6 satisfies the condition, compared to the other design methods, is a powerful
the pillar edges will yield and the load-bearing method for demonstrating the yielding situation of a
capacity of the pillar is found as follows: pillar and stress state over the working area provided
that enough in-situ data are available to construct the
models. Although most design methods ignore the
(9) stress distribution within the pillar and interaction
between the roof, pillar and floor, these data can be
taken into consideration in numerical models.
Parametric studies and field applications showed the
where Rp is the width/height ratio of the pillar, p. is
importance of these factors in the design of yielding
the interface coefficient of friction(ti = tan(p), <p is pillars. Numerical models using a two-dimensional
the interface friction angle and finite difference code, FLAC, were arranged for
investigation of the stress magnitudes over the pillar

402
under the influence of depth stress and front stresses and the stress at the centre of the pillar
abutment stresses. The boundary conditions of the decreased from 25 MPa to 19 MPa. The 10-m pillar
models are the same as those given in Figure 10. A width can be regarded a critical width for the
strain-softening model constituted the post-failure properties assigned to the model.
behaviour with widths of 6, 7.5 and 10 m for the
model pillars. By applying back analysis to the data
available from the large-scale tests of Van Heerden
(1975), the friction angle and cohesion of the yielded
coal were found to be 23° and 0.35 MPa. The in-situ
compressive strength and deformation modulus of
the coal mass are 4.78 MPa and 2.5 GPa. The rock
mass surrounding the working area is quite strong.
Bedding planes, illustrated in Figure 10, were also
modelled and the friction angle between the coal-
rock interface is 27 degrees.
The distribution of vertical stresses immediately
above the pillar, gate entry and abutment are
illustrated in Figure 13. In this model, it is assumed
that the Iongwall face is far enough for building up
front abutment stress over the modelled section. The
stresses are illustrated for just a half side of the pillar
due to its symmetry. The findings from these models Figure 14. The stress state over the working region and pillar at
600 m depth with front abutment stress
are that a 6-m pillar yields and retains a small
amount of stress, about 10 MPa, at the centre of the
pillar, while 7.5 and 10-m pillar rib sides yield. Parametric studies showed that the properties of
However, the pillar core retains a significant the pillar-rock interface, as well as the softening
quantity of stresses. properties of die coal material, significantly
influence the yielding state of a pillar and stresses
over the pillar and working area.

3.5 Field experience and stress measurements

The difficulty in stress measurement is the reliability
of the obtained values. However, the evaluation of
stress measurements and observation of roadway
stability is a practical method of site-specific design,
especially for yield pillars. The width of the pillar to
yield during the development loading or the side
abutment loading stage ensures improved ground
conditions for pillars between 6 and 10 m wide,
depending on the coal mass strength, seam height,
roof and floor constraint and mining depth. The
application of these sizes is generally successful in
deep mines (Table 3). However, a 9-m-wide pillar
Figure 13. The stress state over the wotting region and pillar at failed to yield properly under 365 m of cover load,
a depth of 600 m with no front abutment stress. and at some mines, the changing of a proven
successful design to slightly larger 12-15-m yield
pillars under deeper cover seems to have resulted in
The vertical stress was gradually increased over renewed ground control problems, both in the roof
the vertical upper boundary of the models in order to and floor (Demarco et al., 1988).
represent the front abutment stresses over the
modelled section. The stress distribution over the
modelled section illustrated in Figure 14 is just for a 4 CONCLUSIONS
25-MPa boundary stress condition. In this case, no
increment occurred over the 6-m pillar; excess For a pillar to be called a yielding pillar, it must
stresses were transferred to the abutment side. It is display gradual yielding, not sudden failure, and
interesting to note that the 7.5-m pillar failed maintain enough residual strength to support the
gradually with the increment of front abutment weight of the rock within the pressure arch during

403
 Table 3. Yield and critically sized pillar widths at various the main factors into account in the yielding process
depths in the field. of the pillar. However, the prediction and reliability
Pillar Depth Yield Pillar core Face Reference
width (m) condition stress position
of results in mis method depend on the availability
(m) (MPa) (m) of in-situ data. For a pillar height of 2 m, the
16.8 795 critical 65 0 Ko*leraaL1996 predicted width of the yielding pillar for the
12.2 580 yield 20 -24 KœhkyetaU996 properties assigned to the models is 6 m, without
10.6 855 yield - - KœhfcxetaU996 considering the front abutment stress. The 7.5-m
9 460 yield 12 0 Demacoaall988 pillar can also be regarded as a yield pillar. The 10-
9 460 yield 14 0 Danacoetall988 m pillar, which was determined for this example
6.1 610 yield 7 76 Newiml989 mine as a critical width, could be successfully
10 840 yield 10.5 25 HadnetaL1997 applied in other mines, depending on the shear
*(0 longwall race is approaching measurement point strength properties of rock-coal interfaces as well as
the softening properties of the coal material.

the development and side abutment loading stages of

mining. The current design methods introduced in REFERENCES
this study are not unique themselves to explain the
yielding mechanism of these pillars. All the methods Alder, H., Potts, E. L. J. & Walker, A. 1951. Research on strata
used for the preliminary estimation of long yielding control in the northern coalfield of Great Britain. A8,
International Conference About Rock Pressure and Support
pillar width ranges for Bilsthorpe Colliery estimated in the Workings. : 106-134.
a reasonable width range when compared to the Bieniawski, Z. T. 1984. Rock Mechanics Design in Mining and
previous applications. The main conclusions of this Tunnelling. A. A. Balkema.
study can be summarised as follows: Carr, F. 1992. Ten years' experience of the Wilson/Carr pillar
sizing method at Jim Walter Resources, inc. Workshop on
1. Empirical equations with regard to the tributary Coal Pillar Mechanics and Design. : 166-179.
area method predicted a width range for a yielding Demarco, M. J., Koehler J. R. & Lu, P. H. 1988.
pillar of between 6 and 11.5 metres for Bilsthorpe Characterisation of chain pillar stability in a deep Western
colliery by considering safety factors between 0.5 Coal Mine - a case study. Min. Eng. 40(12), 1115-1119.
and 0.73. The ultimate strength concept, of course, Hendon, G., Carr, F., Lewis, A. C. & Cassie, J. 1997. A co­
operative study of gate entry designs - Welbeck Colliery
does not consider the residual strength of the pillar, (UK) and Jim Walter Resources (USA). Mining
interaction between die roof, pillar and floor, and the Technology. 79(909), 115-121.
stress distribution within the pillar. However, diese Holland, C. T. 1973. Mine pillar design. SME Mining
factors can be taken into account by determining a Engineering Handbook. Soc. Min. Eng. AIME.: 1: 13/96-
site-specific safety factor. This means that generally 13/118.
Koehler, J. R., Demarco, M. J. & Wuest, W. J. 1996. Critical
the predicted width ranges of the empirical equations
pillar concept in yield pillar based longwall gate road
are acceptable based on field experience. design. Min. Eng. 48( 8), 73-78.
2. The mine stiffness concept has two weaknesses Kripakov, N. P., Sun, M. C. & Donato, D. A. 1994.
Automation of a progressive failure procedure for analysis
in predicting the stability of a yielding pillar. Firstly, of underground mine pillar designs. Proc. of IS'* Int. Conf.
there are not enough data available for the post- on Ground Control in Mining., 59-68.
failure slopes of large-scale pillars. Secondly, it does Mark, C. 1990. Pillar design methods for longwall mining.
not take the pillar-roof interface properties into BuMines IC 9247.
consideration. The predicted post-failure slopes and Newman, D. A. 1989. In-situ yield behaviour of a coal pillar.
calculated stiffness values from the numerical Int. J. of Mining and Geological Engineering. 7,163-170.
Obert, L & Duvall, W. 1967. Rock Mechanics and Design of
models suggested that a 10-m pillar could be Structures in Rock. Wilev. New York.
regarded as a yielding pillar, while the 5 and 7.5-m Salamon, M. D. G. & Munro, A. H. 1967. A study of the
pillar widths were found to be unstable in a two- strength of coal pillars. Journal of the South African
entry system. This finding is true for the 10-m pillar; Institute of Mining and Metallurgy. 55-67.
however, the 7.5-m pillar width in the two-entry Salamon, M. D. G. 1970. Stability, instability, and design of
pillar workings. Int. J. Rock Mech. Min. Sei. & Geomech.
system is a yielding pillar in some mines. Abstr.l, 613-631.
3. The confined core concept is a valuable Salamon, M. G. D. 1992. Strength and stability of coal pillars.
analytical method in comparison to the other Workshop on Coal Pillar Mechanics and Design. 94-121.
Van Heerden, W. L. 1975. In-situ determination of complete
analytical methods in which no account is paid to the stress-strain characteristics of large coal specimens. J. S.
properties of coal-rock interfaces. However, pillar Afr. Inst. Min. Metall. 75(8), 207-217.
side restraint is based on an assumed value which Wagner, H. 1974. Determination of the complete load-
strongly affects prediction of the yielding width of a deformation characteristics of coal pillars. Proc. 3rd Int.
pillar. There İs no attention paid İn this method to Cong. Rock. Mech. 1076-1081.
Whittaker, B. N. & Singh, R. N. 1979. Design and stability of
the post-failure properties of the pillar. pillars in longwall mining. The Mining Engineer. 59-71.
4. Numerical modelling has certain advantages Yavuz, H. 1999. Physical and Numerical Modelling of Pillar
when compared to the other methods since it takes Protected Mine Roadways. PhD. Thesis, Leeds University.

404