You are on page 1of 20

International Journal of Mining Science and Technology 25 (2015) 687–706

Contents lists available at ScienceDirect

International Journal of Mining Science and Technology


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

A holistic examination of the load rating design of longwall shields after


more than half a century of mechanised longwall mining
Frith Russell C. ⇑
Mine Advice Pty Ltd., Beresfield, NSW 2322, Australia

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

Article history: It is a commonly asked question: how big should the longwall shields be? The answer is a key aspect of a
Received 15 January 2015 longwall mining feasibility study when the consequences of inadequately rated shields are considered.
Received in revised form 11 March 2015 This paper addresses this question based on the measured nature of the loading environment in which
Accepted 16 June 2015
shields are required to operate, the various geological and geometrical controls of that environment
Available online 24 August 2015
and the various links between their load rating, a range of other relevant shield design factors and the
loss event they are required to prevent a major roof collapse on the longwall face. The paper concludes
Keywords:
that despite the tremendous advances that have been made in shield design and load rating over the past
Longwall geomechanics
Longwall face instability
50 years, the same drivers that caused longwall miners of the past to seek improved roof control on the
Longwall shield design longwall face via the use of ever-higher rated shields, are still as relevant today. However at the current
Periodic weighting of massive strata time, the limits of the largest available longwall shields have yet to be tested, therefore industry focus for
Longwall panel geometry the foreseeable future should possibly be in achieving the maximum level of roof control on the face via
their optimum operational use rather than considering further shield rating increases and incurring the
inevitable downsides in terms of capital cost and shield weight.
Ó 2015 Published by Elsevier B.V. on behalf of China University of Mining & Technology.

1. Introduction (1) Longwall shields.


(2) Manner by which the longwall face is managed during
Few in the coal industry would argue that most longwall faces operations.
were not adequately stable for the majority of their operating time.
It is demonstrably the exception rather than the rule that longwall Good longwall operational practice (e.g. keeping the face
faces become unstable with either large cavities hindering produc- straight, setting of shields, equipment maintenance, etc.) has been
tion requiring both remediation and recovery or the shields the subject of significant industry interest via the regular column
becoming ‘‘iron-bound’’ due to excessive roof to floor convergence. ‘‘Wills on Walls’’ in the ‘‘Australian Longwalls’’ journal, whereby
That is not to say that some longwall faces are not more prone to one of the Australian industry authorities on the subject, Nick Wills
such difficulties than others (the specific of which will be discussed provides an informative and at times, amusing commentary on the
in more detail in the paper) and certainly, when major events do subject.
occur the operational difficulties and costs involved in the recovery However, the author contends that unless longwall shields are
are substantial, with very severe events being considered as ‘‘catas- well-designed and adequately rated in the first place, even
trophic’’ to the business. Therefore understanding the primary best-practice operational controls may be insufficient to prevent
causes and controls of such events, even if only at a conceptual major unplanned loss events occurring. For example, if the shield
level, should be of great interest to every mining company that legs are leaking significantly or minor amounts of roof loss (which
either operates longwall faces or is considering doing so. can always occur) has resulted in the excavation height being
During longwall extraction, mine operators have only two sig- greater than the operating range of the shields, the potential for
nificant pro-active controls at their disposal to maintain adequate major instability on the face may be substantially increased above
strata stability on the longwall face: that which can be controlled by operational controls such as ‘‘keep-
ing going’’.
The focus of this paper is to consider one key aspect of longwall
shield design, namely defining a suitable load rating, in a more
⇑ Tel.: +61 409 056 514. holistic manner in order to assist mining companies make
E-mail address: russellfrith@mineadvice.com.au

http://dx.doi.org/10.1016/j.ijmst.2015.07.001
2095-2686/Ó 2015 Published by Elsevier B.V. on behalf of China University of Mining & Technology.
688 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

informed decisions when procuring face equipment, thereby inevi-


tably maximizing the likelihood that best-practice operational con-
trols will be effective during longwall extraction.

1.1. Defining a suitable design process for longwall shield rating

Engineering design is at the heart of the engineering method


and it always requires certain definitions to be in place before it
can be undertaken on a credible basis. These are: (1) what is the
specific problem being designed against? (2) how does the prob-
lem manifest? (3) what, if any, are the external drivers that act Fig. 2. Example of large stone ‘‘lumps’’ developed during a major longwall roof fall.
to influence whether the outcome will be satisfactory or unsatis-
factory? (4) how do specific element(s) of the problem being con-
sidered act to promote a satisfactory outcome? (5) is there a
credible engineering model directly linking the design of elements
of the problem with the required outcome? (6) how is an accept-
able level of design risk determined and quantified? 1 Spalling of coal face and
increasing tip to face
The load-rating design of longwall shields will be considered by
2 Guttering type failure of
reference to these 6 fundamental elements of engineering design, immediate roof
including making comments as to the usefulness of some of the 3 Delineation of large intact blocks
via the propogation of weighting 3
previously published design methodologies. induced sub-vertical tensile
ractures
2

1.2. What problem is being designed against? 1

For the purpose of this paper, the problem being designed


against by the load rating of the longwall shields is the occurrence Fig. 3. Schematic illustration of primary roof instability mechanisms on the
of major roof collapses ahead of the shields during longwall extrac- longwall face.
tion (see Figs. 1 and 2). Such events can stop longwall production
for weeks or even months in extreme cases-particularly if the
AFC becomes over-loaded and needs to be unloaded by hand (see 1.3. How does the problem manifest?
Fig. 1), and remedial measures such as cavity fill and strata
re-consolidation ahead of the face using some form of injected Ignoring the impact of major geological structures (e.g. faults
medium such as polyurethane or phenolic resins are very and dykes) for the moment, which simply due to the presence of
expensive. structurally disturbed strata can cause major longwall roof falls,
Equipment damage can also be significant such as increased the author has only ever been able to identify two fundamental
wear on the AFC due to the conveying of stone or damage to roof instability mechanisms on the longwall face (Fig. 3 which also
hydraulic lines and electro-hydraulic equipment should be includes the influence of face spall increasing the effective tip to
required to break-up large pieces of fallen stone from the roof face distance which is therefore common to both):
(see Fig. 2).
Therefore the problem being designed against by determining a  Horizontal stress-driven guttering ahead of the shields (akin to
suitable load rating for the longwall shields is as follows: ‘‘the cutter roof, albeit with a different horizontal stress driver as will
occurrence of major roof cavities ahead of the shields with the be explained later)-termed ‘‘guttering’’ from here on.
potential to stop the longwall face for an extended period and  Delineation of large blocks due to a proliferation of
require significant resources to return the face to normal produc- mining-induced sub-vertical fracturing-termed ‘‘block delin-
tion conditions’’. eation’’ from hereon (see Figs. 2 and 4).
This definition underpins the contents of this paper and should
be referred to when considering any of the comments made herein. These two basic mechanisms are illustrated schematically in
Fig. 3 with Fig. 2 showing the types of large blocks that can form
and their potential impact upon the longwall face should they fall
out of the roof ahead of the shields. With two distinctly different
roof instability mechanisms ahead of the longwall shields being
defined, the geo-mechanisms by which they both manifest need
to be identified and explained.

1.3.1. Conceptual models for defined longwall roof instability


mechanisms
In any engineering discipline it is useful if not vital to refer to a
conceptual mechanistic model when discussing various aspects of
an engineering problem, such a model at least allowing basic
‘‘cause and effect’’ relationships to be identified and potentially
proven experimentally. The availability of such a model is also a
necessary pre-cursor to attempting to analyze any engineering
Fig. 1. Example of over-loaded AFC due to large amounts of fallen stone. problem numerically.
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 689

Cantilever driven
bending moment

Fulcrum point W

Wshield Horizontally
bedded roof
σh x l strata

l
du/dt
Fig. 4. Longwall face roof cavity containing a proliferation of sub-vertical mining t P
induced fractures.

Vertically cleated coal


R
An effective mechanistic model should ideally reduce the engi- KEY
neering problem to its key elements so that (1) it is as simple as W - abutment loading
W shield - abutment loading on shields
possible but captures all of the key elements, (2) but it is readily t - extraction height
understandable, at least at a conceptual level, by all. This type of l - tip to face distance prior to spall
x - increase in tip to face distance due to spall
simplification in science and engineering is termed ‘‘reductionism’’ σ h - horizontal stress changes due to roof strata bending
and should be at the heart of the engineering method-as stated by du/dt - rate of shield/roof strata convergence at the shield
P - shield loading
Antoine de Saint Exupéry ‘‘a designer knows he has attained per- R - rate of face retreat
fection not when there is nothing left to add, but when there is
nothing left to take away’’. A good example of reductionism in Fig. 5. Simplified mechanistic model of a longwall face.
practice are Isaac Newton’s Three Laws of Motion whereby a very
complicated first principles problem was simplified to three simple
statements and mathematical equations that even high school stu- moment being generated in the roof at or close to the coal face.
dents could understand and work with perfection indeed. This bending moment is at a maximum at the fulcrum point and
acts to generate both compressive and tensile horizontal stres-
1.3.2. Guttering type roof instability ses in the immediate roof strata as a direct result of roof strata
Fig. 5 shows a conceptual mechanistic model first developed bending.
and proposed by the author in 2005, which attempts to define  Horizontal stress induced for a short distance into the immedi-
the key parameters involved in either the absence or conversely, ate roof is inevitably compressive due to the sense of bending of
the development of guttering type roof instability on the longwall the roof strata, which in conjunction with the structural compe-
face [1]. It will be regularly referred to during this paper and can be tence of the immediate roof (in particular its bedded nature)
summarized as follows: and the effective tip to face distance (to be discussed later),
can result in a buckling type failure of the roof ahead of the
 The model contains a vertically cleated coal face of a given shields and the onset of guttering (cutter roof) type cavity
height t and an immediate roof strata against which the shields formation.
are set against being horizontally bedded.  The higher the level of absolute shield closure in any given
 Longwall face area and shields are directly affected on a shear shear cycle, the higher the resultant bending moment and
by shear basis by a varying vertical load, termed W. The relative therefore, the higher the induced horizontal stresses in the roof
proportions of W which directly load the coal face as compared at the face line.
to the longwall shields are indeterminate.
 Roof strata is directly supported by a longwall shield of a given The main points of the model in Fig. 5 are that it contains a sim-
load rating P with a tip to face distance from the tip comprising ple but rational explanation for the commonly observed guttering
both a face geometry (l) and face spall (x) component. The or cutter roof type roof instability ahead of the longwall shields.
model assumes that the tip is in contact with the roof strata, Specific detail such as what factors control the magnitude of W
although as will be discussed later in the paper, there are sev- and Wshield on a shear by shear basis and how the shields act
eral reasons why this does not always occur in practice, the against the formation of gutter type roof cavities will be discussed
result being a further increase in effective tip to face distance. during the paper.
 The result of the component of the vertical load W acting on the One critical aspect of guttering type roof cavity development is
longwall shield (Wshield) commonly being higher than the shield that whilst on its own it is rarely sufficient to stop a longwall face,
load rating P (as evidenced from shield convergence rate data – it has the ability to: (a) significantly reduce the overall effective-
see later) is that the longwall shields close-up vertically at a cer- ness of the longwall shields in preventing a larger cavity develop-
tain rate over time (du/dt) which in conjunction with the rate of ing (by preventing canopy tip contact with the roof and often the
face retreat R (or shear cycle time) determines the absolute need to turn-off the posi-set system to maintain canopy attitude)
level of roof convergence or shield closure in any given cutting and (b) can at times result in the roof to floor height along the face
cycle of the longwall face. being greater than the maximum working height of the shields,
 Relatively high strength and stiffness of the coal face as com- this again significantly reducing their effectiveness.
pared to that of the longwall shields results in the roof strata Nonetheless, a horizontal stress-driven guttering type roof
above the longwall face behaving as a form of cantilever with instability mechanism does not offer a credible explanation for
the fixed end or fulcrum being located some distance head of the formation of sub-vertical fractures in the roof (Fig. 4) and large
the as-cut longwall face. Cantilever action of the immediate roof ‘‘block delineation’’ (Fig. 2). Therefore a different driving mecha-
strata from the face to out over the shields results in a bending nism is required.
690 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

1.3.3. Large intact block delineation Cantilever driven


bending moment
The easiest method to demonstrate how large block delineation
occurs above the longwall face in general terms is by reference to
high level periodic weighting due to near-seam massive strata Fulcrum point
(Fig. 4 was taken on a longwall face that was beset by high level
periodic weighting roof cavities which were always observed to
contain such sub-vertical fractures in the roof ahead of the shields).
Observation of sub-vertical roof fractures on many longwall σh
faces firstly confirms that they are not directly related to each con-
secutive shear of coal being taken. If they would be regularly Gutter
u
spaced on around 0.9–1 m intervals (unless a half-web cutting sys-
tem was being used) and this is rarely if ever the case, as can be
seen in Fig. 4. Therefore, another geotechnical driver other than Fig. 7. Conceptual model for guttering of roof ahead of the powered supports.
the simple act of taking individual shears of coal must be at work.
It is also the case that the strike of such features only rarely
aligns with a known geological jointing direction in the same area to form) and it is judged from the strata noises commonly heard
of the mine and they are almost always absent in the adjacent gate before they appear on the face, that they incrementally work their
roads. Therefore it is logical to conclude that they are some form of way down through the overlying strata from below the massive
Mining-Induced Fracturing (MIF) related to longwall extraction. weighting unit toward the face horizon (see Figs. 3 and 6). Only
A strong clue as to the mechanistic driver for these fractures when they intersect the roof of the longwall do they pose a signif-
emanated from work undertaken by Frith and Creech, which was icant production threat and it is at this point that large blocks of
subsequently completed by Frith and McKavanagh, into the phe- intact rock material are able to fall out onto the face (see Fig. 2).
nomenon of periodic weighting of near-seam massive strata units In the experience of the author, the development of closely
[2,3]. Various longwall mines in the Newcastle Coalfield of NSW spaced sub-vertical tensile fractures above a longwall face is not uni-
occasionally contain thick (up to 50 m) massive conglomerate versally linked to the occurrence of high level periodic weighting
channel units in close proximity (<5 m in places) to the longwall due to thick near-seam massive strata units, although it is inevitably
extraction horizon. Such a lithological setting is relatively rare (for- more prolific under such circumstances. Therefore as a general phe-
tunately) but when longwall extraction was attempted under such nomenon, it is hypothesized that any significant rotation of the over-
conditions, it afforded a unique opportunity to investigate the burden about the longwall face, however caused, will generally act
associated geomechanics and various threats to longwall to drive their development, the faster the rotation, the more rapid
production. their formation below the rotating unit or section of overburden.
The evaluation of periodic weighting intensity will be discussed Similarly, the closer the rotating unit is to the extraction horizon,
in more detail later in the paper, suffice to state at this stage that the sooner they will appear in the roof ahead of the shields.
the driving mechanism for intense sub-vertical MIF above the
longwall face was inferred to be as one uncontrolled rotation of
1.4. External drivers for roof instability on the longwall face
massive strata units about the face, rather than simply the high
abutment load W that was also generated.
The preceding section of the paper has defined two distinctly
Fig. 6 illustrates this schematically, the main point being that
different roof instability mechanisms (namely guttering and block
once the massive strata unit becomes unstable, in addition to caus-
delineation) both of which are potentially aggravated by spalling of
ing intensive vertical loading W on the coal face/shields and
the coal face acting to increase the effective tip to face distance.
thereby producing additional face spall, if its center of gravity is
Guttering cavity development is driven by bending of the roof
located well behind the longwall shields (which with high level
strata about the face and associated stress-driven buckling within
periodic weighting will inevitably be the case), it will naturally
the tip to face distance (see Fig. 7), with large block delineation
and uncontrollably rotate as a single intact block about the long-
being directly related to uncontrolled overburden rotation about
wall face until it comes to rest on any underlying caved or goafed
the face (Fig. 6). This section of the paper attempts to determine
material.
and quantify the various primary drivers or controls for both
These fractures are generally observed to be very clean (i.e. they
behaviors.
contain little or no rock flouring), there is commonly no evidence
that they existed prior to longwall extraction, they naturally open
up in tension as they develop (thus allowing the large intact blocks 1.4.1. Guttering cavity development
Referring to Fig. 5 and considering the basic mechanics of both a
cantilever and end-loaded slender column beam behavior (includ-
Overburden surcharge
ing Euler Buckling) as described by Colwell and Frith, the following
statements are taken to be generally true [4]:
Tensile fracture of
cantilever
propagates ahead Centre of gravity (1) The higher the bending moment is, the higher the horizontal
of face
Multiple sub-vertical stress changes within a bending beam.
fractures caused by (2) The higher the maximum deflection of a cantilever, the
rotation of cantilever
higher the associated bending moment at the fulcrum point.
Possible
(3) The thinner an end-loaded slender beam, the more prone it
roof cavity is to unstable buckling.
Goafed material (4) The longer an end-loaded slender beam, the more prone it is
to unstable buckling.
(5) Any end-loaded beam or column of a geometry consistent
Fig. 6. Schematic illustration of periodic weighting and associated fracturing with Euler Buckling, will be naturally unstable should it
mechanism of immediate roof. become overloaded.
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 691

 Roof convergence above the shields (function of roof conver-


W/H 1 (sub-critical) gence rate du/dt in Fig. 5 and elapsed time) and the resultant
Spanning upper strata
bending moment above the longwall face driving localized roof
Depth of cover H buckling and consequent guttering within the tip to face dis-
tance (Fig. 7).
 Rate of longwall retreat R (i.e. the time for each individual
shear).

Less than full depth of cover These three influences can be analyzed within the conceptual
loading impacts face
model shown in Fig. 5 by applying Euler Buckling concepts (which
Face width W represent inevitable beam or column instability), in particular that
of critical buckling stress (for both the coal face in a vertical sense
Fig. 8. Sub-critical overburden conditions and likely abutment load distribution
and the roof horizontally within the tip to face distance) so that a
between the face and surrounding coal pillars.
relative measure of roof fall potential (RFP) via stress-driven gut-
tering can be gained according to both face width and extraction
height.
14
Fig. 5 shows the conceptual model for both the coal face and the
roof above the longwall face and as the model is only being used
Average convergence rate (mm/h)

12
for relative comparison purposes, there is no need to factor in such
10
issues as the nature of the coal, nature of the roof or shield rating
8 and location, etc. These will all be considered as constants for the
6
purposes of this illustration but considered individually later.
In terms of coal face spalling (i.e. causing an increase in tip to
4 face distance or x as shown in Fig. 5), the exceeding of the critical
2 buckling stress for a well-cleated coal in a vertical sense is a func-
tion of both rv (i.e. the vertical stress acting in the coal at any given
0 100 200 300 400 500 600 time) and t (the height of extraction) squared. As such it can be sta-
Depth of working (m) ted that:

Fig. 9. Plot of average shield convergence rate against depth of cover [8]. x ¼ f ðrv ; t 2 Þ ð1Þ
For isolated sub-critical panel conditions, panel width w is
These statements allow the three main controls for guttering taken to be the overriding determining factor for rv so that the fol-
type roof cavity development to be identified (other than the long- lowing is also true:
wall shields themselves), namely (a) amount of roof convergence
x ¼ f ðw; t 2 Þ ð2Þ
above the longwall shields, (b) effective tip to face distance and
(c) nature of the immediate roof strata and its susceptibility to It is noted that mechanistically the use of coal buckling in a ver-
undergo horizontal stress driven buckling (i.e. guttering/cutter roof tical column as the conceptual model for face spall is fully consis-
development) when end-loaded. tent with the primary coal rib failure mechanism for roadways
Frith proposed a basic model for the relative influence of differ- reported by Colwell [7].
ent longwall mining geometry (i.e. face width, extraction height However, face spall on its own is not a direct reason for a long-
and cover depth) on the general face loading environment W and wall face to be unable to produce coal and is therefore only an
therefore as a direct consequence, the general propensity for gut- input into the potential for guttering type roof instability ahead
tering type roof cavity development due to roof convergence above of the shields in the tip to face distance as will now be discussed.
the shield and associated strata bending about the face [1]. This In terms of the roof material just ahead of the shields, exceeding
was termed Roof Fall Potential (RFP) at that time remembering that the critical buckling stress for the bedded roof and so becoming
the development of small guttering type cavities can be the critically unstable is a direct function of both the horizontal stress
pre-cursor to larger collapses due to the potential corrupting of rh acting and the tip to face distance (l + x) squared, so that:
shield effectiveness, as previously described.
2
For demonstration purposes herein, the potential for guttering Roof fall potential ðgutteringÞ ¼ f ðrh ; ½l þ x Þ ð3Þ
type roof instability along a single longwall face in sub-critical over-
For a compressive horizontal stress that is generated in the
burden conditions (i.e. typically the higher depths of cover) with
immediate roof as a result of cantilever bending about the face, it
stable surrounding chain pillars (see Fig. 8) will be analyzed in detail.
can also be shown that rh is a direct function of the vertical dis-
As a general concept for sub-critical overburden conditions as
placement of the free end of the cantilever (u), which in turn is a
defined herein, the maximum possible abutment load affecting
direct function of both the rate of roof displacement above the
the center of the longwall face itself will be assumed to be defined
shields (du/dt) and the rate of longwall retreat during production
by a section of overburden equivalent in height to the panel width.
(R).
Note that this assumption also infers an equivalent caving angle of
The rate of shield convergence (du/dt) was generally linked to
around 26° which is at the upper end of measured abutment angles
depth of cover by Frith and Stewart (see Fig. 9 whereby all deep
in both Australia and the USA [5,6].
monitoring sites contained yielding chain pillars under double goaf
In terms of the driving potential for a guttering type roof cavity
loading), but it logically is more a function of panel width w in iso-
ahead of the shields, the following three factors are logically of
lated sub-critical longwall extraction environments as being con-
some direct relevance:
sidered herein [8].
Similarly the rate of longwall retreat R is also to some degree a
 Face spall and the associated increase in the tip to face distance
function of panel width w (assuming similar cutting and flitting
(x in Fig. 5).
speeds for the shearer) so that Eq. (3) can be re-written as:
692 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

2
Roof fall potential ðgutteringÞ ¼ f ðw2 ; ½l þ x Þ ð4Þ Composite super-critical subsidence trough across
both sub-critical longwall panels and chain pillar
2
When it is also remembered that x = f(w, t ) as given in Eqs. (2)
and (4) can be re-written as: Fully collapsing
upper overburden
2
Roof fall potential ðgutteringÞ ¼ f ðw2 ; ½l þ f ðw; t2 Þ Þ ð5Þ
In order to use Eq. (5) for comparative purposes, the absolute Face loading
term l (i.e. the geometrical tip to face distance without spall) needs Highly exposed to full
First compressed depth of cover
to be removed and this is justified on the basis of the overwhelm- longwall chain pillar
Second longwall
ing experience of the author being that major roof instability
almost always occurs in conjunction with significant face spall sig- Fig. 11. Super-critical overburden conditions across two sub-critical longwall
nificantly greater than the geometrical minimum tip to face dis- panels and a yielding chain pillar.

tance. It is not dimensionally correct to do this, but is judged to


be a reasonable approximation to re-write Eq. (5) as:
coal being a weak but more importantly, relatively thickly bedded
2 2 2 4 4 material as compared to many laminated stone types.
Roof fall potential ðgutteringÞ ¼ f ðw ; ½w; t  Þ ¼ f ðw ; t Þ ð6Þ
In other words, the driving potential for a guttering type roof 1.4.2. Large block delineation and overburden weighting/rotation
fall on the longwall face in an isolated sub-critical environment mechanisms
only is a direct function of both the panel width w and extraction It goes without saying that when large intact blocks of overbur-
height t to the power 4. It is noted that depth of cover on its own is den material form above the longwall face and develop periodic
not a controlling factor in the isolated sub-critical environment. weighting, the magnitude of shield loading Wshield and the associ-
Without going through a detailed derivation, in a super-critical ated level of shield convergence rate (du/dt) inevitably increase as
environment (see Fig. 10), total depth of cover has a far greater a direct consequence (see Fig. 12 taken from Frith who shows vary-
control on face loading conditions so that Eq. (6) becomes: ing rates of leg pressure rise after set (surrogate for rate of leg con-
vergence) during known severe periodic weighting on LW5 at
Roof fall potential ðgutteringÞ ¼ f ðH3 ; w; t4 Þ ð7Þ
Newstan Colliery) [9]. Therefore an increase in guttering type cav-
Eq. (7) is also considered to be generally applicable in those sit- ity development is inevitable under such conditions, as was always
uations whereby chain pillars between longwall panels undergo observed to be the case on LW5 at Newstan before the full roof col-
yield and significant compression with goaf on both sides (see lapse occurred via large intact blocks falling out onto the AFC
Fig. 11), so that adjacent longwall panels are not isolated from each (Fig. 2). This section of the paper simply considers a number of
other in terms of overburden abutment loading. potential causes for significant overburden rotation above and near
In a similar manner to which periodic weighting RFP due to the longwall face.
massive strata in the near-seam overburden (which will be dis- Three different possible scenarios will be described by reference
cussed in the next section of the paper) is very sensitive to changes to either published research studies or case examples known to the
in the thickness of the massive strata unit causing the periodic author, namely (a) periodic weighting, (b) overburden fracturing
weighting [i.e. RFP = f(thickness3)], so it appears that the propen- and weighting adjacent to chain pillar cut-throughs and (c) retreat-
sity for guttering type cavity development on the longwall simply ing out from under large vertical surface cliffs at low cover depth.
due to the mining geometry, is highly sensitive to variations in
panel width, cover depth, chain pillar stability and extraction 1.4.2.1. (1) Periodic weighting. The general mechanistic concept of
height (all other factors being equal). periodic weighting of near-seam thick massive strata units has
In the same way that roof guttering (cutter roof) in a mine road- already been described in general detail, as shown in Fig. 6.
way is at least in part related to the structural competence of the Research work reported by Frith and McKavanagh presented an
roof strata, it is self-evident that the same should be true for gut- empirical classification (see Fig. 13) for periodic weighting severity
tering type cavity development ahead of the longwall shields. in terms of the extent to which the resultant face conditions were
The experience of the author dictates that the more laminated an manageable or not when trying to prevent a major roof cavity
immediate roof type, the greater the propensity for uncontrolled forming ahead of the longwall shields [3]. The classification at that
buckling and associated guttering cavity development (all other time used both the known thickness of the massive strata unit and
factors being equal). It is further suggested that this is why many the width of the longwall panel, these being two critical controls.
longwall operations have found that leaving a relatively thin The weighting classification is based primarily upon longwall
immediate coal roof in place tends to prevent guttering cavity extraction experiences in the Newcastle Coalfield of NSW, but
development in an otherwise unstable friable overlying stone unit, was supplemented by more wide-scale case histories from other
parts of Australia to give it a more general applicability (see Ref.
[3] for details).
W/H 1 (suber-critical)
It is noted that the massive strata thickness used is based on
what is termed as an ‘‘Equivalent Conglomerate Thickness’’ or
Depth of cover H

‘‘ECT’’, massive sandstone bodies having been shown to have lower


periodic weighting potential than the same thickness of conglom-
erate. A general conversion factor, from a massive sandstone thick-
ness to an ECT, of 0.7 m was found from the back-analysis of
Full depth of cover loading various mining outcomes, meaning that a 30 m thick massive
impacts face
sandstone has the same periodic weighting potential as a 21 m
thick massive conglomerate (all other factors being equal).
Face width W
The key elements of periodic weighting and the rotational
Fig. 10. Super-critical overburden conditions and likely load distribution to the behavior of massive strata units in close proximity to the extrac-
longwall face. tion horizon are assessed to be as follows:
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 693

120 1800-2000
110 1600-1800
100 1400-1600
90 1200-1400
1000-1200

Chock #
80
70 800-1000
600-800
60
400-600
50 200-400
40 0-200
30 -200-0
251 253 255 257 259 261 263 265 267 269 271 273 275 -400--200
Shear #

Fig. 12. Isopachs of measured leg pressure rises in 5 min period after set (in psi)-longwall 5 at Newstan [9].

300
Rapid
convergence
Uncontrollable
250 longwall Caving
Readily Operationally weighting shortwall
controllable controllable
longwall
Panel width (m)

longwall
200 weighting weighting
?
Bridging
shortwall
150

Potential for windblasts


100 according to proximity
* Massive strata unit located within of bridging channel to
40 m of top of extraction horizon seam horizon

50
0 5 10 15 20 25 30 35 40 45 50
Equivalent conglomerate unit thickness ECT (m)

Fig. 13. Near-Seam, massive strata weighting classification-data points omitted for the sake of clarity [3].

Protection Protection
 In direct contrast, if a massive unit cannot span the panel at the
zone zone face, it will naturally become unstable and so cantilever from a
position ahead of the face, this setting up the potential for as
one block rotation about the face (see Fig. 6). This effect will
Overburden weighting zone
be generally confined to the central portion of the face, with
the two face ends being protected due to the presence of stable
Goaf
flanking coal pillars (see Fig. 14).
 The periodic weighting cycle length appears to increase in
(a) Hypothetical near-overburden fracturing arc- direct proportion to the thickness of the massive unit (which
longwall geomechanics (Frith and Stewart 1993) makes perfect sense). Similarly increasing roof fall propensity
50 can also be anecdotally directly linked to an increasing thick-
45 ness of massive strata as was clearly evident during the extrac-
40 tion of LW5 at Newstan Colliery in the mid 1990s. This can be
Leg pressure (MPa)

35 seen in Fig. 15 which shows the roof fall plan for LW5, the vari-
30
ations in roof fall frequency being almost entirely linked to dif-
25
20
ferences in conglomerate channel thickness above the
15 extraction horizon-this being known from detailed up-hole dril-
10 ling studies in the adjacent gate roads and surface borehole
Av. Back press.
5 Av. Front press. information.
0 20 40 60 80 100 120 140
 Mechanistically it can be argued that the potential for a block
Support # delineation type roof fall on the longwall face is a function of
(b) Measured powered support loading distribution the cube of the thickness of massive strata (i.e. RFP = f[thick-
across a longwall face (Frith and Stewart 1993) ness3]). The associated logic behind this statement is that the
Fig. 14. Typical support load and overburden fracturing distribution across a
rate of block rotation increases as function of thickness2 (both
longwall face. the moment arm and load acting through the center of gravity
are a direct function of thickness), with the distance that the
face has to retreat in order to progress beyond the extent of
the weighting cycle which also increases in relation to thickness
 If a massive unit is sufficiently thick to span the extraction panel
[9]. Combining these independent factors allows the thickness3
at the face, even temporarily, the longwall face is not exposed to
relationship to be derived.
block rotation effects as instability of the massive unit is initi-
ated behind rather than ahead of the face (termed ‘‘shortwall’’
Without doubt, longwall roof control experiences that relate
overburden fracturing).
back to the periodic weighting of thick, near-seam massive strata
694 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

index peaks were around 100 m apart which effectively elimi-


nated any form of periodic weighting of near-seam massive

e
strata as just described.

ra e bl
m enc ure

)
ct
80 ld es
 Rises and falls in energy index were plotted on the panel layout

tra
(1 ubs st m
a ir
as shown in Fig. 16, it was apparent that each rise corresponded
F
with the longwall face approaching a maingate (MG)
cut-through (C/T). Similarly the finish of the falling energy
index corresponded with the face crossing the line of a tailgate
Zone 1
(13) (TG) C/T (the MG and TG C/T’s are only slightly offset as can be
seen in Fig. 16).
A

 The majority of the reported abnormally large goaf falls corre-


B
C
D
sponded with the face crossing the zone between
E

Zone 2 semi-adjacent MG and TG C/T’s.


(7)
F

Zone 3
This result suggested that both the anecdotal mining evidence
(79) (i.e. the reported large goaf falls) and the measured
micro-seismic overburden behavior were somehow linked and
Zone 4 the presence of chain pillars C/T’s might be the trigger for such
(29)
behavior. Whilst this one example did not conclusively makes
the link between chain pillar cut-throughs and the measured over-
Zone 5
Roof fall location (29) burden behavior, it resulted in a search for other evidence in sup-
(29) Days of lost production per port of the theory, as will now be detailed.
zone total=157 or 22.5 weeks
CSIRO/SCT reported the outcomes of micro-seismic monitoring
Fig. 15. Major roof falls and other weighting related events, LW5, Newstan [9]. during part of LW28A extraction at Appin Colliery which was
aimed at investigating overburden geomechanics associated with
gas liberation from beneath the Bulli Seam [10]. Fig. 17 is copied
from that report and the following comments are made:
units shows quite clearly that RFP increases disproportionately
with increasing strata unit thickness. Similarly though, highly  Luo and Hatherly first reported this Appin Colliery data and
aggressive periodic weighting effects can be largely eliminated made the comments to the effect that surges in micro-seismic
through relatively minor changes in longwall panel width so as activity were directly associated with increased gas makes from
to induce strata bridging or spanning at the face line (shortwall the floor measures [11].
geomechanics).  They also stated that ‘‘the fractures occur in zones which are
spaced at approximately 100 m intervals’’.
1.4.2.2. (2) Overburden weighting in proximity to chain pillar cut-  It is understood that the longwall chain pillars were in fact
throughs. During research studies into both periodic weighting and 100 m in length and it is clearly evident in Fig. 17 that each of
windblasts and longwall extraction, a significant amount of evi- these ‘‘fracture zones’’ or ‘‘clusters’’ are situated almost exactly
dence was unexpectedly gathered in support of the view that chain above or in close proximity to TG chain pillar cut-throughs.
pillar cut-throughs have a major influence on overburden fractur-
ing, and consequently, face loading condition [3]. This was not a From this evidence it is difficult to arrive at any conclusions
focus of the research work at the outset, and to the best of the other than that the C/T’s in the TG chain pillar are having a funda-
authors’ knowledge had never even been postulated as a hypothe- mental controlling influence on the occurrence of surges in over-
sis. However, over time the data set became such that the existence
of a real and significant overburden weighting mechanism could
no longer be ignored or dismissed. A brief summary of some of
the supporting data is given herein for reference purposes.
The outbye end of LW19 at West Wallsend Colliery was overlain Solid
by a thick conglomerate channel that was thought to be
non-windblast prone but would nonetheless result in large goaf
falls. As such the mine decided to install a full micro-seismic mon-
itoring system and track overburden behavior during extraction
through this area. As expected the area proved to be
non-windblast prone, but the results of the micro-seismic data
analysis combined with reported large goaf falls showed a result
that was totally unexpected. Solid
Fig. 16 shows the overall outcome of the micro-seismic data
analysis plotted on the mine plan in conjunction with reported
large or abnormal goaf falls. The following comments are made:
Rising enargy index
Falling enargy index
 Micro-seismic data indicated that the ‘‘energy index’’ of over- Locelion of reported
burden fracturing exhibited periodic rises and falls in a manner abnormal goaf falls
that suggested the occurrence of periodic weighting of some
form. However, surface subsidence indicated that the
near-seam conglomerate channel in its entirety had not in fact
caved and was almost certainly still spanning the longwall
Fig. 16. Results of micro-seismic monitoring linking large goaf falls with enhanced
panel following extraction. It was also the case that the energy overburden fracturing and chain pillar cut-throughs [3].
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 695

1400
Drainage holes
Borehole
1300 Station

27
Longwall face

LW
1200

1100

28
LW
1000
Northing (m)

900 Depth: 80 m
SS thick: 14 m
800
Longwall face position
700 Sep6-13 Monltoring
Longwall face position test zone
600 Sep 21-Oct 1

500
Longwall face position
Approximately
Oct 23-Nov 14 uniform increase in
400
cover depth and
200 300 400 500 600 700 800 900 1000 1100 1200 sondstone thickness
Easting (m)

Fig. 17. Micro-seismic monitoring outcomes from LW28 at app in colliery showing N
event clusters on 100 m cycles and TG cut-through lines [3].

Depth: 160 m

l3
SS thick: 36 m

l2
burden fracturing, which in turn are linked to periods of increased

Pane

l1
Pane
l6

Pane
gas make from the floor measures. Anecdotally, later communica-

Pane

l4
l5
tions with mine site staff indicated that during the extraction of

Pane
Pane
subsequent longwall panels it was very common for gas ‘‘blowers’’
to occur along the face as it was crossing a C/T. In between C/T’s Fig. 18. South Bulga mine-location of face weightings from deputies reports [3].
such occurrences were far less frequently.
Early longwall faces at South Bulga were overlain by massive
sandstone (SS) units with various thicknesses. When deputies  As the longwall face retreats, abutment loading of the
reports of discernible face weightings were plotted onto a mine super-incumbent strata is ‘‘thrown’’ onto the flanking chain pil-
plan (see Fig. 18), it was immediately obvious that all of those lars incrementally shear by shear.
weightings that were designated as being [M] ‘‘major events’’  However, each C/T and the extreme ends of the chain pillar rep-
(i.e. those that stopped face production for an extended period) resent an area that cannot readily accept abutment load (as
aligned almost exactly with MG and TG cut-throughs (at 100 m there is either no coal there or the coal is of significantly
spacing) being at each end of the longwall face. No major weight- reduced load bearing capacity).
ings were ever experienced outside of a few meters from a chain  Therefore, as the face approaches a C/T, vertical abutment load
pillar cut-through. must logically concentrate inbye of the C/T.
Furthermore later longwall panels in the same area of the mine  If the chain pillar is highly loaded just inbye the C/T, but imme-
that were formed up using MG cut-throughs spacing of only 50 m diately outbye the C/T, the next pillar has a significantly lower
(for development productivity reasons) as opposed to the previ- load acting upon it, vertical shear movement must occur within
ously used 100 m, never experienced similar major face weight- the overburden above the longwall face due to what may be ter-
ings, despite the overburden lithology, depths of cover and panel med as ‘‘differential chain pillar compression’’ either side of the
geometry being almost identical. C/T.
The South Bulga data set provides a further credible ‘‘cause and  It is hypothesized that this ‘‘differential chain pillar compres-
effect’’ relationship between chain pillar cut-through layout and sion’’ acts to form a vertical discontinuity in the overburden
the occurrence of major weightings on the longwall face with the aligned sub-parallel with the longwall face (this then explains
ability to stop face production for a period. the measured micro-seismic behavior discussed previously),
It is finally noted that the same basic link with chain pillar hence the observed surges in vertical abutment stress acting
cut-throughs can also be found in such phenomena as surface sub- in and around the longwall face itself.
sidence cracking (including the varying location of transverse
cracks parallel with the face) and the occurrence of windblasts dur-
ing extraction. They are not described in detail herein due to space
restrictions, but are discussed in detail by Frith and McKavanagh C/T's cannot accept overburden loading
[3].
From the small body of evidence presented herein, the presence Pillar loading area
of chain pillar C/T’s demonstrably has a significant impact upon High pillar Low pillar
overburden behavior, which in turn is influencing longwall face loading loading
conditions in a number of different ways. Other examples of this Vertical fracturing
link have been found as part of consulting studies which are not Retreat due to differential
direction pillar compression
discussed; herein it is suffice to say that the same basic relation- across the C/T

ship between chain pillars C/T’s, observed overburden behavior


and production loss scenarios can be fully made.
The hypothetical model used to offer some rational explanation
for this behavior is shown schematically in Fig. 19 and the C/T’s are
likened to discontinuities running across the longwall panel. In this Fig. 19. Conceptual model linking cut-throughs location to increased overburden
regard the following comments are made as follows: fracturing.
696 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

inability to advance the face in an attempt to reduce the tip to face


distance and (c) significantly increased volume of unstable roof
material that will slump onto the AFC and needs to be removed
before remediation measures can be applied.
The subject of the effects of geological discontinuities such as
faults and dykes on longwall face stability is a major subject in
itself and deserving of a separate technical paper. However, for
the purpose of this paper it can be summarized as simply a local
geological anomaly with the ability to significantly increase the
potential and severity of major roof instability on the face as com-
pared to it not being present. Relevant aspects include increasing
levels of face spall, increasing face loading conditions W, signifi-
cantly altered immediate roof conditions and out of seam cutting
Fig. 20. Schematic illustration of the stability threat when extracting out from cliff thus slowing down the rate of longwall retreat R.
lines at shallow cover depth.

1.6. Summary

 Further apart chain pillar C/T’s are, the higher the surge in abut- The various conceptual and mechanistic models described
ment stress as the face crosses each C/T line appear to be (the herein in combination could possibly be viewed as a ‘‘unified the-
formation of significantly more chain pillar cut-throughs in US ory’’ linking a whole range of geometrical and geological parame-
longwall panels as compared to Australian counterparts is ters to the potential for significant roof instability on the
intriguing in this regard). longwall face via two distinctly different fall mechanisms. When
 Other evidence also strongly suggests that staggering C/T’s on combined with the uncertainties inherent in any rock mass, they
the MG and TG side increases the frequency of weighting events provide a credible explanation for many major loss events on long-
and so reduces their severity. wall faces as well as an insight into the characteristics of measured
variations in either shield convergence rate or leg pressure rise rate
It is not clear as to why a measured or inferred overburden on a shear by shear basis, an example of which was previously
weighting does not occur at every C/T line. However, based on both shown in Fig. 12 with two further examples being provided in
measured data and anecdotal outcomes, the effect is demonstrably Figs. 22 and 23 [8].
real and often significant in terms of the occurrence of unwanted However, the term ‘‘unified theory’’ cannot in reality be applied
events that detract from face production. It is also the case that as to the best of the authors’ knowledge, there is no realistic way of
even though the phenomenon was first identified through research bringing them all together into a single methodology whereby
studies relating to longwall extraction under massive immediate shear by shear behavior in terms of key roof stability controls
roof conditions, it has been observed and identified under a much issues, such as level of face spall, shield convergence rate, overbur-
wider range of overburden conditions including lithologies that den rotation effects and rate of longwall retreat can even be esti-
contain no discernible massive strata units. mated, let alone credibly predicted. Therefore they stand as
individual explanatory conceptual empirical models that may
1.5. Retreating out from under high vertical cliffs at low cover depth never be combined into an all-encompassing predictive
methodology.
Although never formally published, the author is aware of at The best that can be achieved with the various models is to
least one shallow longwall mine whereby face line roof control dif- allow longwall mining environments and longwall layout to be
ficulties commonly occurred when the face was retreating out evaluated for both mining geometry and geological/geotechnical
from under a high steep cliff line at surface (see Fig. 20), whereas risk, this then potentially allowing:
such difficulties never occurred when the face retreated under sim-
ilar cliff lines from the shallow depth side. (a) Objective comparisons to be made between different long-
The likely technical reasons behind this observation are rela- wall mines-which to the knowledge of the author has never
tively obvious having described in detail both general face loading been done on an industry-wide scale similar to other strata
and overburden rotation mechanisms, as the presence of such a control aspects such as coal pillar stability and roadway
significant vertical discontinuity as a cliff line would inevitably ground control, and more relevantly to this paper.
both (a) increase vertical face loading beneath the cliff line (see (b) Providing input into the selection of shield load-rating based
Fig. 20) and (b) result in a natural rotation of the significant over- on an informed opinion on the likely propensity for major
burden ‘‘block’’ immediately behind the cliff line. roof instability at any given longwall mine or area of a long-
wall mine.

1.5.1. Impact of major geological discontinuities This then represents the authors’ opinion as to the current
The previous two sections of the paper have described in detail ‘‘state of the art’’ in terms of predicting longwall face conditions
some of the more general geometrical and geotechnical drivers for and instability risk according to defined characteristics of the geo-
known longwall roof instability mechanisms in close proximity to logical environment and mining geometries. It therefore becomes a
the longwall face. The final issue that needs to be briefly addressed key factor or indeed limitation when considering how the load rat-
before considering the load-rating design of longwall shields (i.e. ing of longwall shields is assessed and determined.
the main subject of the paper) is the potential influence of signifi-
cant localized geological discontinuities such as major faults. 2. General shield design comments
Fig. 21 shows the actual impact of a mid-angled fault plane hid-
ing outbye of a longwall face in terms of the exacerbation of roof Having provided a detailed description of both roof instability
instability ahead of the shields. Of particular note are (a) significant mechanisms on the longwall face and the various geometrical,
increase in tip to face distance due to the hade of the fault, (b) geotechnical and geological drivers involved, the reader will hope-
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 697

(a) Mining towards fault (b) Face fails on fault plane (c) Cavity develops (d) Large cavity remains (e) Rocsil is used
leaving rubble sitting on further while clearing once chain is cleared to fill the cavity
chain and large tip to chain more material falls
face/exposed roof as chain is cleared

Fig. 21. Schematic illustration of the substantial roof instability mechanism caused by a mid-angled fault hiding outbye the face.

35 (6) Leg hydraulics are not impaired by fluid leaks that would
Convergence rate (mm/h)

30 allow leg loads and stiffness’s to drop during the mining


25 cycle (self-evident requirement and considered no further
20 herein).
15
10
However, when reviewing the requirements listed above, it is
5
0 immediately obvious that some lead to compromises with other
170 180 190 200 210 220 230 240 250 longwall design aspects.
Shear # For example:
Fig. 22. Example of measured shield convergence rate on a shear by shear basis [8].
 Location of the shield legs with respect to the longwall face is
limited by the presence of the AFC and a forward walkway, both
fully now have a good appreciation of the difficulties and complex- having width requirements if they are to meet other longwall
ities involved in being able to define the potential for a major roof design outcomes (e.g. AFC conveying capacity, front walkway
fall at any given point within a longwall panel. Essentially, there is ergonomics, etc.).
no comprehensive predictive model available that adequately  Even if the canopy tip could be set against the roof at the face,
accounts for all of the various drivers, only part models that deal risks associated with the shearer drums colliding with the
with different aspects of the problem. How this can be used as part canopy preclude this as a practical proposition.
of longwall shield design will be detailed later on in the paper, the  Yield loads are limited by such issues as capital cost and shield
focus now changing to considering how others have attempted to weight in terms of transporting those underground using
determine suitable load-ratings for longwall shields in the past. rubber-tyred vehicles.
As a starting point, it is hypothesized by the author that long-
wall shields offer the highest or optimum level of roof control Recognizing that such compromises exist, the practical or real-
when they meet the following six operational requirements: istic design of longwall shields should seek to determine the best
overall balance such that none of the listed operational require-
(1) They are able to be set to the roof along all or most of their ments are unduly neglected. Therefore it is immediately evident
canopy length. that compromising on the load rating of the shields is possibly
(2) Leg loads are applied to the roof as close to the face as pos- desirable under certain circumstances, the logical question being
sible, albeit within practical mining considerations. how to determine such circumstances and defining the magnitude
(3) Tip to face distance is as small as possible, noting that the tip of the compromise on a prudent commercial investment basis.
to face distance is defined by the closest position to the face
that the canopy is first in contact with the roof and the
amount of face spall after the cut is taken, not simply by 3. Determining the load rating of longwall shields
the location of the canopy tip and as-cut face (i.e. ‘‘l’’ in
Fig. 5). Determining the load rating of a longwall shield appears to be
(4) Yield rating of the shield P is fit for purpose to the inherent dictated by two distinct and seemingly opposing schools of
face conditions (to be discussed in more detail later). thought:
(5) Set: yield ratio is as high as possible, this leading to
improved levels of direct or active reinforcement to the (a) Shield rating is matched to the expected geotechnical condi-
immediate roof of the longwall face (again to be discussed tions and mining geometry.
in detail later). (b) The bigger, the better.

The fundamentals of these two approaches to shield load rating


20
design are the key to understanding whether either should be
Convergence rate (mm/h)

15 applied or some other approach developed, remembering that


the overall objective of a longwall shield is to prevent major roof
10 collapses ahead of the shields so as to assist the longwall mine to
5
produce large volumes of low cost coal on a semi-continuous basis
and provide an acceptable return on capital to the shareholders or
owners of the mine.
0 10 20 30 40 50 60 70 80
Shear #
In other words, the argument is not really one of technical
approach or the elegance of the analyses, but being founded in
Fig. 23. Example of measured shield convergence rate on a shear by shear basis reality and contributing to minimize the investment risk associ-
showing a distinct periodic weighting characteristic [8]. ated with each new longwall mine or longwall face procurement.
698 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

R
r Support E
loading
B D
C Support
D E lowering
W Support
h1

Shield load
W B C setting
F A
Gob Support
F advance Adjacent shield
A advance GRC
curve
Face P P Shearer cut
GRC curve
After wilson Condition
at setting Adjacent shield
advance GRC curve
Fig. 24. Detached block model of longwall roof behavior and shield rating design
[15]. Convergence

Fig. 27. Cyclic loading pattern on ground reaction curve for individual or a small
group of shields [13].

Main roof loading F

being included in the design methodology, it is straightforward


Conver gence D
to argue that the method should not be used, albeit accepting that
Roof structure each method contributed to the developing state of knowledge at
the time it was published.
K Coal K Shieid K Gob
3.1.1. Ground reaction curve concept
Fig. 26 is directly re-produced from Medhurst and illustrates
Floor structure
how the ground reaction curve concept can seemingly be used
for longwall shield design [12]. Fig. 27 is taken from Barczak and
Fig. 25. Concept for evaluating roof shield response based on the stiffness of the Tadolini and the same basic elements are also present [13].
shield, coal and gob foundation [16].
Reviewing these two figures and technical papers that demon-
strate the proposed use of the ground reaction curve in longwall
Contact shield design, several characteristics are immediately apparent:
load

Theoretical
Effect of hydraulic
leakage can result in
(a) Ground and shield curves are preferably brought together at
set-to-yield lower set pressures and a common point so that presumably a state of ‘‘equilibrium’’
reduced support stiffness
800 t is achieved, this making logical sense when designing
ground support for civil tunnels which was the original
A Working
640 t range intent of the method.
Degradati
B on of roof (b) Effect of time, which as any longwall operator knows can be
Support setting line with time
Setting supports later critical to strata stability via the face being stood for an
C Roof failure
increases convergence
and results in poorer
extended period, is not included.
roof conditions (c) As more face convergence occurs, the necessary shield load
to reach equilibrium seemingly reduces, at least initially.
Convergence
Initial roof Set-to-yield (d) Neither paper provides any guidance or assistance to how
convergence leg closure
the necessary ground reaction curves are developed on a
Fig. 26. Strata-support interaction diagram for longwall support [12]. site-specific basis.
(e) Neither technical paper considers any direct link between
3.1. Matching the shield rating to the expected geotechnical conditions the analyses undertaken and the potential for major roof
and mining geometry falls to occur on the longwall face, either in general or
site-specific terms.
The most recent and commonly cited approaches to longwall
shield rating design relate to either: Frith and Stewart undertook shield monitoring, including shield
convergence monitoring, on various longwall faces at some 12 dif-
(1) Use of the ‘‘ground reaction curve’’ concept which was orig- ferent Australian longwall mines, this being thought to be the most
inally developed for use as part of the ‘‘New Austrian Tun- comprehensive longwall shield convergence study ever under-
nelling Method’’ in the early 1960s [12,13]. taken and reported [18]. They noted that in almost all cases long-
(2) Numerical modeling [14]. wall shields never came to equilibrium during the shear cycle or
even over 2 day longwall stoppages (as was common practice dur-
However before these two recent design methods are critiqued, ing weekends in the early 1990s) and that in instances of very high
the historical shield design methods of Wilson and Smart will be initial shield convergence rates after shield set, only minimal slow-
briefly commented upon [15,16]. ing down of the convergence rate over time. (see Fig. 28
The basic arrangements used by Wilson and Smart are provided re-produced from Frith and Stewart) [18].
in Figs. 24 and 25 respectively. Furthermore, Figs. 22 and 23 show measured examples of aver-
Without even considering details of the analysis conducted, age convergence rate on a shear by shear basis, this indicating the
both methods can be eliminated from current use, as neither caters degree of variation in face loading that can occur as a longwall
for the loss event that the shield should be designed to either pre- retreats in shear cycle increments. Whilst the lowest convergence
vent or reduce the potential for, namely the occurrence of major rate measured in these two examples is zero, typical low-end val-
roof cavities ahead of the shields. With the required outcome not ues are less than 5 mm/h with occasional high values in excess of
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 699

15 mm/h peaking at 35 mm/h (i.e. at least 7 times the typical (3) The tip to face distance to maximize roof integrity and limit
low-end values). This reality of measured shear by shear shield dilation should be less than approximately 0.6 m. The smal-
convergence rate on two different longwall faces during retreat ler the distance, the better the result is.
distances of less than 70 m is not effectively captured in the speci-
fic discussions or examples given regarding the use of the ground If these key requirements are met, then the ground about the
reaction curve concept for longwall shield rating design. face and above the supports should be controlled under most geo-
The ground reaction curve was originally developed for and logical circumstances.
applied to a ground stability problem whereby long-term stability In other words, it appears that an almost universally applicable
(i.e. equilibrium) was a necessary outcome, a primary focus being longwall shield design is being presented as one of 100–
in determining a suitable timing for the emplacement of a perma- 110 ton/m2 at yield, a set-yield ratio of 80%, a canopy ratio of
nent concrete tunnel lining which must not be fractured and <2.4 and preferably less than 2 and a tip to face distance of
cracked by excessive ground movements. In complete contrast, <0.6 m. The following comments are made in response to this
monitoring data indicates that longwall faces are more typically research finding:
in a state of non-equilibrium during the vast majority of
short-term shear cycles, such that achieving equilibrium between (1) Numerical model used is presumably only two-dimensional
the strata and the shields is neither necessary or indeed practical although this is never stated in the report, whereas the prob-
given the obvious disparity between the magnitude of possible lem is one of at least three dimensions.
strata loads acting on the shields (Wshield) and available shield load (2) In the opinion of the author, the term ‘‘weak’’ strata is too
ratings (P), which even in the extreme are limited to no more than generic and the research does not include consideration of
approximately 120 ton/m2 or 1.2 MPa of vertical pressure. varying extraction height, periodic weighting, cover depth,
The above disparities between measured shield convergence panel width, chain pillar stability, etc. as design variables
behavior with time, the non-inclusion of elapsed time in the anal- that are either known via other industry research publica-
ysis, the non-inclusion of a major roof fall within the assessment tions that are not considered or referenced in the project
process and the lack of any useable guidance about how a ground report or can be logically concluded as having some level
reaction curve may be developed for any given longwall panel or of influence over the face loading environment during long-
area of a given longwall panel, means that despite the apparent wall extraction, weak strata or otherwise.
link to what is undoubtedly a seminal concept in civil tunneling, (3) Whilst the comments on canopy balance are sensible as they
it is judged that the ground reaction curve concept offers little to can be engineered into the shield at the design stage, the
the problem of determining suitable shield ratings in varying long- suggestion that tip to face distance should be less than some
wall settings on a prudent investment risk basis. pre-determined amount for the shields to be effective is
problematical. Firstly significant face spall can occur (which
3.1.2. Numerical modeling approach is controlled by other factors not least of which is extraction
Others have utilized a numerical modeling approach in longwall height) to significantly increase the effective tip to face dis-
shield design (e.g. Gale who addresses shield design in ‘‘weak’’ tance and secondly, variations in the as-cut roof horizon and
ground or strata) and it is worth briefly reviewing the outcomes of also small scale gutters in the roof ahead of the shields will
this most recent industry funded research into the shield design change the location along the canopy where it first touches
topic [14]. the roof, again increasing the effective tip to face distance.
The research project’s aims are to: (a) better understand frac- Therefore defining an optimum tip to face distance, credible
ture mechanics and geotechnical requirements for face support or otherwise, has no practical value.
in ‘‘weak’’ strata, (b) investigate factors which optimize face sup- (4) The findings from Gale are simply a confirmation of what was
port and (c) review operational factors to optimize support perfor- almost certainly the state of the art in longwall shield ratings
mance within a weak strata section. before the advent of the Moranbah North Mine 2 leg, 1750
The stated findings of the research project are as follows: tons replacement shields in 2009, with two supporting state-
ments to improve their overall effectiveness, one being able to
(1) Yield capacity to control the caving line and provide confine- implemented in the design of the shields with the other out-
ment to the fractured material is recommended to be in the side the control of the shield designer [14,17].
100–110 ton/m2 range with a set of approximately (5) Research confirmed that the specified longwall shields will be
80 ton/m2. effective under ‘‘most geological circumstances’’. The author
(2) The canopy balance is recommended to be less than 2.4 has no major disagreement that this outcome is generally cor-
preferably less than 2. In most instances this relates to a rect. As stated in the first sentence of the paper ‘‘few in the coal
tip to leg distance in the range 3–3.5 m. industry would argue that most longwall faces were not ade-
quately stable for the majority of their operating time’’.
Gale effectively confirms the industry status quo as of the end of
A
2009 whereby longwall shields at that time were effective in the
Roof convergence (mm)

majority of geological conditions [14]. Therefore on the assump-


A - Independent of time tion that most longwall instability events are therefore linked to
B - Decoy of rate with time the more extreme geological and mining geometry circumstances,
B the relevant shield design question post-2009 is seemingly
whether there is tangible benefit in increasing shield ratings
beyond the levels stated by Gale and if so, where to apply them?
This question will be addressed by reference to two issues:

Time (h) (1) Whether the bigger is better approach has any future merit,
including the post-2009 advent of the 2 leg 1750 tons
Fig. 28. Various forms of time-dependent shield convergence behavior during an
individual shear [18]. shields at Moranbah North?
700 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

1400 (3) As a geotechnical profession, the site specific design of long-


Avg support capacity
1300
Max support capacity
wall shields clearly remains elusive, with the science follow-
1200 ing rather than driving the increases in shield ratings as
1100
demanded by industry.
Shield capacity

1000
(ton)

900
The design and procurement of the 1750 tons MNC shields was
800
demonstrably driven by both factors (1) and (2) above, hence the
700
drivers for ‘‘bigger is better’’ were alive and well as recently as just
600
500
prior to 2009. Furthermore, ACARP commissioned a concept study
1984 1988 1992 1996 2000 2004 2008 to examine the feasibility of producing 15MTPA from a single long-
Year wall face, one of the major findings being that further research was
required in specifying shield capacity, this concern emanating from
Fig. 29. Historical overview showing trend of increasing shield capacities from
1985 to 2006 [13]. the recognition that the higher the required face production, the
less tolerable are major strata control events and the associated
losses [19].
Two published anecdotes will be used to justify point (3),
(2) Application of the models presented earlier in regards to namely that geotechnical engineering has been following rather
geotechnical and mining layout drivers for longwall instabil- than leading the development of ever-higher shield ratings.
ity to provide guidance on which longwall environments are McDonald states that ‘‘ACIRL suggested that Westcliff should
most prone to face instability? Is bigger better? seek hydraulic supports with a capacity of 800 tons after conduct-
ing exhaustive tests on physical scale models of the strata above
3.2. Bigger is better? the Westcliff/Bulli Seam area [18]. Westcliff increased this to
900 tons and chose Gullick Dobson 4 leg chock shields from the
The bigger is better approach to shield design has been several manufacturers who were able to meet the specification.
on-going in industry since the advent of mechanized longwall min- During the presentation of Klenowksi, it was stated that as a
ing over 50 years ago. result of the numerical modeling study to assess the powered sup-
McDonald made the following statement in regards to the port requirements at Central Colliery, it was concluded that a rat-
development of shield load ratings in Australia following the intro- ing of around 650 tons would suffice [20]. However, the final
duction of longwall mining in 1963: ‘‘every Australian longwall has support rating of 800 tons was selected for Central by a partner
been subject to home-made modifications. These modifications in the project which was more comfortable using a higher rated
were undertaken because of improved roof conditions allowed support.
greater face cutting rates. Improved roof conditions were achieved With the recent procurement of 2 leg 1750 tons longwall
using heavier supports; the area of greatest concern for early long- shields at Moranbah North and the coal industry asking the ques-
wall miners’’ [18]. tion as to the various barriers to producing 15MTPA from a single
The same basic trend of ever-increasing shield ratings over time longwall, it is hard to conclude that the ‘‘bigger is better’’ approach
is also evident in the USA, and Fig. 29 is provided by Barczak and is necessarily at an end at the current time.
Tadolini [13]. However, 2015 is not 1986 and as stated previously; most long-
Finally, the procurement in 2009 by Moranbah North Mine in walls remain adequately stable for most of the time. Therefore,
Australia of the highest rated longwall shields ever manufactured accepting that there may still be a demand for ever-increasing
(2 leg 1750 tons shields) was driven solely by the need for improved shield load ratings in extreme mining circumstances or in conjunc-
roof control on the longwall face, the cost and operational disadvan- tion with abnormally high production expectations, perhaps what
tages of shields weighing more than 60 ton each being set-aside for is really needed is some form of rationale, even if only judgment
improved roof control and longwall production rates [17]. based, as to what shield rating may be appropriate for any given
Therefore, at face value it would seem that the mining industry set of circumstances. This will be considered in the final section
has been informally applying the ‘‘bigger is better’’ approach to long- of the paper after a number of other key shield design issues relat-
wall shield rating design for some time now, in fact since the first ing to strata control are briefly mentioned.
mechanized longwall supports were first used. This raises one sim-
ple question: why has this continued to be necessary, to as recently
as 2009, in spite of the claimed advances in geotechnical modeling 4. Other shield design aspects relating to strata control
and longwall shield rating design by the geotechnical profession?
There are three distinct answers to this question which provides 4.1. Shield width
full context and may pre-empt that the increasing of longwall
shield load ratings is perhaps not at an end, even only a few years Shield widths have increased during the past 20 years from 1.5
after the ‘‘quantum leap’’ associated with the 1750 tons Moranbah to 1.75 m, and currently 2 m wide shields are becoming more com-
North shields: mon. This evolution has been driven largely by operational effi-
ciency considerations, such as reducing the face production
(1) Industry is still seemingly driven by the statement of constraint due to shield lower-advance-reset (LAR) cycle times,
McDonald, the only change being that acceptable levels of reducing the number of shields and AFC pans to be moved during
face production in the past were much lower than they are face relocations (the move to wider longwall faces has simply exac-
now, hence the need for even fewer roof control problems erbated this requirement), and less components on the longwall
during longwall extraction and therefore higher rated face requiring maintenance or that can fail.
shields [18]. However, it is also true to say that as individual shields get
(2) Longwall faces are wider, higher, deeper and in more aggres- wider, higher load capacities can be achieved (without increasing
sive strata conditions than was commonly the case leg pressures markedly) as increased leg diameters that are
pre-1986. This has undoubtedly made faces less stable over- directly proportional to the increased shield width, act to increase
all and so driven the need for higher rated shields. the shield load rating by the leg diameter squared. The load rating
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 701

of the current 2 m wide Moranbah North longwall shields 4.3. Canopy ratio
(875 ton/m of face) could not have been achieved within a
1.75 m wide shield without significantly increased leg pressures. Canopy ratio is basically the ratio between the canopy length
ahead of the legs and the canopy length behind the legs (A/B in
4.2. Tip to face distance Fig. 32). The importance of the canopy ratio is to provide a bal-
anced canopy whereby its center of mass is reasonably close to
Tip to face distance is undoubtedly the simplest concept in rela- the location of the legs. This ensures that the Canopy tip does
tion to roof stability on the longwall face, and it is being directly not have a significant tendency to drop away from the roof during
analogous to roadway width. It represents the unsupported span shield lower, advance and re-setting (LAR), which will either tend
between the face and the tip of the shields where roof instability to reduce tip loading of the shield if it stays in contact with the roof
emanates from. Few longwall operators would disagree that main- or increase the tip to face distance if it does not, neither of these
taining the tip to face distance as low as possible, which has a pos- outcomes are beneficial in terms of roof control.
itive impact on roof stability on the longwall face. Whilst the author is not aware of any rigid design values for
The actual tip to face distance at any points within the mining canopy ratio that is backed by fundamental shield design research,
process will be governed by a number of independent factors: optimum values commonly used are between 2 and 2.5, with val-
ues closer to 2 being preferred. Obviously, a better canopy balance
(1) Minimum tip to face distance as determined by the face could be achieved by making the length of the canopy behind the
equipment and method of working (see ‘‘l’’ in Fig. 5) which legs (B) longer, although this simply adds weight and hence cost to
is commonly found to be in the order of 500 mm. the shields. It also tends to reduce the shield Support Load Density
(2) An increased tip to face distance due to face spall effects (see (SLD) (ton/m2), which whilst being a commonly used parameter is
‘‘x’’ in Fig. 5). not a particularly useful means of assessing shield design. This is
(3) An increased tip to face distance caused by the tip of the because maximizing SLD (by reducing the distance B behind the
shields not being set to the roof. This can have a number legs) will almost certainly lead to the canopy becoming less bal-
of different causes, such as the cutting of steps in the roof anced via an increased canopy ratio, which is counter-productive
(Fig. 30), small gutter type roof cavity development ahead in terms of roof control ahead of the canopy tip.
of the shields (Fig. 7) or whether two or four leg shields The other obvious outcome when considering canopy ratio is
are being used (Fig. 31). that lengthening the canopy to reduce tip to face distance for strata
control reasons is also potentially counter-productive, as it will
It is usual for the tip to face distance to be analyzed and opti- tend to increase the canopy ratio thus making the canopy less bal-
mized by reference to the position of the shields and face before anced and so pulling the tip away from the roof.
the cut is taken. The tip location in relation to the face relates to Certainly a combination of reducing tip to face distance by mak-
the shields being back from the AFC and the cut yet to be taken. A ing the canopy longer ahead of the legs (A) and also increasing the
theoretical tip to face of approximately 500 mm is consistent with shield SLD by reducing the canopy length behind the legs (B) is a
current well-designed longwall shields operating in the Australian recipe for disaster in terms of canopy ratio and shield effectiveness.
coal industry and cannot be reduced further, primarily due to It is noted that the author is aware of at least one set of relatively
shearer collision concerns. In other words, operational rather than recently procured shields in the Australian coal industry that have
geotechnical considerations are the limiting factors in this regard. seemingly been designed with exactly this concept in mind.
In reality, the geometrical tip to face distance is a minimum
value and will only fleetingly if ever be as low as 500 mm. The influ-
ence of face spall can be very significant at times (e.g. under peak 4.4. Planned operating height range in relation to the maximum
face loading conditions or in proximity to a major geological struc- operating height
ture) and the author has observed face spall of several meters in
conjunction with major instability on the face, rendering any argu- It is important that when a shield is working at the upper end of
ments about whether the geometrical tip to face distance should be its designated operating height range, and there is still an amount
450 or 500 mm for roof control reasons as essentially meaningless. of leg travel to allow the shield to be correctly set to the roof
Even though the method of face working can be changed to should either the cut inadvertently be taken too high or a small
reduce the tip to face figure under extreme face loading conditions roof cavity forms ahead of the shields. Having the ability to set
by an amount determined by the web depth (e.g. double chocking the shield to the roof for a distance above the designated planned
when working Uni Di IFS which can only be used when significant maximum working height is a prudent shield design consideration,
face spall is occurring anyway), tip to face distances of 1–2 m with as once the shield cannot be correctly set to the roof, the threat of
the shields behind the AFC waiting for the AFC to be advanced (so major roof instability occurring increases significantly.
that double-chocking can again take place) are an inevitable out- Again there are no industry accepted design guidelines in this
come during mining under high face loading conditions, particu- regard, the amount of extra operating height above the planned
larly in thick seam environments. maximum working height being a risk mitigation measure to be
The designed tip to face distance, as is shown on OEM face decided upon by the mine itself. However, as a general rule, the
cross-sections, is no more than a practical minimum value rather author is of the view that two criteria should be applied:
than what is likely to be the case during mining. Accepting that
the tip to face distance prior to the cut should be minimized, it (1) The higher the general likelihood of roof instability occur-
should not be done at the expense of front walkway utility, AFC ring on the longwall face for geotechnical reasons, the
width or canopy ratio (see next section) due to the potentially greater the additional operating height, over and above the
overriding influence of face spall. It is assessed to be of little overall planned maximum working height, should be; and
value to minimize the tip to face before the cut by compromising (2) The higher the maximum planned working height, the
the overall shield geometry, only to eventually realize that it makes greater the additional operating height should be. This
little or no difference to the maximum tip to face values that inevi- relates to the argument that the severity of major roof
tably occur during mining when the potential for roof instability is alls on the face increases in line with the face operating
logically also at a maximum. height.
702 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

Fig. 30. Illustration of the impact of ‘‘steps’’ in the roof on effective tip to face distance.

Tip to face Load centre


Load centre Tip to face

(a) Two leg supports (b) Four leg supports

Fig. 31. Illustration of different setting characteristics of two and four leg shields on tip to face distance and load center.

nation angles of substantially less than 20°. Unexpected and seem-


ingly unexplainable face roof control difficulties have commonly
A B
occurred under this scenario.
Maintaining the inclination of the top caving shield at >30°
throughout the operating height range increases in difficulty as
the operating height range is increased. The higher the operating
range, the longer the top caving shield needs to be in order to pro-
vide enough shield reach, therefore the lower its inclination at the
bottom of the operating height range will be (all other factors
being equal).
The conclusion reached is that whilst providing for a large oper-
ating height range may have theoretical benefits, particularly when
Fig. 32. Shield profile showing the definition of canopy ratio (A/B). the seam thickness in a given mine varies significantly, there are
potential major roof control downsides when operating such long-
wall shields at the bottom of their operating range due to the top
Building in increased operating height into the shields results in caving shield having a very shallow inclination angle. The author
additional capital cost and will also make the shields slightly heav- has observed this on several occasions as being a major controlling
ier. However, when this is compared against the consequences of factor in on-going strata control difficulties on the longwall face.
production or reserve losses associated with unplanned roof falls
on the face or having to reduce the face working height to ensure 4.6. Floor bearing pressures
that a sufficient amount of leg travel is available to accommodate
small cavities and/or cut roof horizon variations, such additional Whilst there are no publications discussing this issue in detail
costs and shield weight are surely prudent compromises to make. with respect to longwall shields, there are known examples (albeit
not published) of major longwall collapses occurring either fully or
4.5. Inclination of the top caving shield within the planned operating partly due to the presence of very soft floor beneath the shields
height range causing a dramatic reduction in shield loading to the roof. It is a
fundamental design requirement that any form of standing sup-
It is a fundamental requirement of shield design that the incli- port (including longwall shields) are able to be set against both
nation of the top caving shield is never less than about 30° to the the roof and floor without either undergoing compressive failure.
horizontal. The reason is that broken goaf material should always Whether a longwall shield will fail the floor beneath and so
slide off the top caving shield so that when the shield is being reset effectively ‘‘sink’’ into it, is essentially a bearing failure type mech-
to the roof, the available leg load is applied to the roof above the anism. Therefore it is appropriate to consider it in this manner.
canopy (where it assists roof stability) rather than through the Based on information available to the author, the peak vertical
top caving shield where it has no useful purpose other than to raise stress acting through shield pontoons and into the floor is typically
broken goaf material during the shield setting process. These two no more than around 8 MPa, this being applied at or very close to
scenarios are illustrated schematically in Fig. 33a and b. the toe of the base.
Frith and Stewart undertook shear tests between steel and rock The bearing capacity of any strata unit beneath the shields is a
and found that the limiting angle of friction was in the order of 26°– function of width of the footing (in this case the shield base) and
27° [8]. Therefore a top caving shield inclination of no less than 30° strength and thickness of the floor material. There are many differ-
would logically have a very high likelihood of the resulting broken ent methods of analysis that could be used in this regard. Seeds-
goaf material sliding off and not directly loading the shield. man provides the following equation for use when considering
Whilst not published, the author is aware of several examples in the stability of the floor beneath a coal pillar [21]:
Australia and overseas of longwall shields being worked very close
Bearing capacity ¼ UCS=2ð4:14 þ W=2=hÞ
to the bottom of their operating range with top caving shield incli-
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 703

(a) Goaf load slides off the top caving shield-leg (b) Goaf load direct loads top caving shield-
load applied to roof leg load applied to goaf

Fig. 33. Schematic illustrations of how the inclination of the top caving shield influences leg load distribution.

where W is the pillar width; and h the unit thickness with an immediately above the canopy, this directly acts to reduce the
unconfined compressive strength (UCS). effective tip to face distance whereby unsupported roof strata
Using this equation with a peak applied stress of say 8 MPa, a may have the ability to buckle and so form a gutter type cavity.
UCS can be determined that would represent the limiting equilib- Trueman made the suggestion that under certain face loading
rium condition for the first 1 m of floor beneath the shields in conditions, it may be preferable to set the shields to only 60%
terms of whether a bearing failure would occur or not. On the rather than say 90% of their yield loading. The following quote is
assumption that a pontoon is in the order of 700 mm wide (this taken from that paper: ‘‘Where periodic weighting is high enough
representing the minimum footing dimension expected for a to result in periodic shield overload, it may be better for set pres-
1.75 m shield), it is found that for the first 1 m of floor beneath sures to be nearer 60% of yield than 90% (with shields of support
the base, the limiting equilibrium UCS is 3.6 MPa. For values lower density of 100 ton/m2 or greater before the cut). This relates to
than this floor failure is likely, for values above this, floor failure the effect of time. If the support is set to 60% of yield then it will
becomes less likely. take much longer to the first yield event and for the same cycle
The above equation can be used to consider floor stability for time, there will be fewer yields. Fewer yields will result in less con-
different sections of strata beneath the shields, different maximum vergence and subsequent roof degradation and it will be easier to
floor pressures and different pontoon widths. However, the analy- mine through the periods of support overload. If a shield periodi-
sis present indicates that the typical UCS of the first 1 m of floor cally has an inadequate capacity for the conditions, the authors
needs to be below 4 MPa before the issue of floor failure compro- have seen no evidence that very high setting loads will stabilize
mising shield effectiveness becomes a significant possibility. the roof. The belief that very high set pressures are beneficial
It is noted that this analysis has not considered the extent by may have arisen when support capacities were less and set pres-
which the shields may tend to ‘‘plough’’ the floor material ahead sures close to the yield value were necessary for the set pressure
of the pontoons as they are advanced. This relates to such consid- to be adequate’’ [22].
erations as the effectiveness of base lifting rams to lift the toe of The underlying logic behind this statement seems to be that leg
the base clear of the floor and also whether the shields are being yield events allow increased rates of roof convergence, in which
advanced up-dip or down-dip, the specifics of which are outside case the statement potentially makes sense. However in all of the
the scope of this report and the expertise of the author. shield convergence monitoring studies across the coal industry
reported by Frith and Stewart, not once was the rate of shield clo-
4.7. Set: yield ratio sure seen to increase due to the legs going into yield [8]. The rate
may have remained constant after going into yield (which makes
The history of shield development over the past 50–60 years, all sense) but certainly it never accelerated. Therefore the idea that
points to the concept that as well as the yield load rating being ‘‘the reducing the set pressures (assuming it could be done on an
bigger the better’’, the set: yield ratio should be ‘‘the higher the as-needs basis), thus allowing greater shield convergence to simply
better’’. reduce the number of subsequent yield cycles, may not be
It has been argued that the external driver for guttering type well-founded and certainly goes against the entire history of the
failure of the immediate roof ahead of the shields is bending of development of longwall shields.
the immediate roof strata around the face (see Figs. 5 and 7). Fortunately, Trueman made the following statement: ‘‘opera-
Whilst the shield is rarely ever able to prevent such bending due tional controls can nevertheless be effective in minimizing roof
to the inevitable common disparity between Wshield and P, it can control issues in the presence of high level periodic weighting lead-
surely act to minimize it. Therefore it makes no sense to simply ing to support overload. Specific attention to achieving the highest
allow such bending of the roof to compress the shield and so set pressure practicable without compromising the attitude of the
develop load in the legs. It is far better to apply such load actively support canopy can reduce the extent of cavities and associated
back to the roof as early as possible and so limit the overall level of delays in many instances’’ [23].
shield closure and hence roof bending about the face, this then The author fully agrees with this later statement such that it
contributing positively to roof stability ahead of the shields. It appears that there is no obvious dissention to the universal con-
may also be the case that by setting the shields to the highest pos- cept of maximizing the set: yield ratio of shields in order to
sible or practical level and so reducing roof to floor closure, this increase their supporting effectiveness
acts to reduce face spall, which is again a positive aspect in terms
of roof stability ahead of the shields. 4.8. Flipper arrangement
The other justification for a high set: yield ratio is that the long-
wall shield can also be viewed as a reinforcing device for the The flipper at the tip of a shield serves two distinct purposes:
immediate roof of the longwall face and in the same way that roof
bolts and tendons reinforce the roof of a mine roadway, it acts to (1) It allows some measure of direct support to the upper coal
‘‘clamp together’’ what is typically a stratified strata sequence face which has both roof stability and also safety benefits,
immediately above the canopy. By reinforcing the roof strata and
704 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

(2) In the event that significant amounts of face spall occur, it The author has the view that the active horizontal force concept
can be used to offer confinement to the exposed roof ahead has a minimal impact on shield effectiveness as a result of the
of the tips. following:

Both are legitimate functions although as discussed by Payne, in (1) Active horizontal force applied to the roof strata is limited by
order to function as a roof support ahead of the tips, the flipper friction between the canopy and roof (which is why Frith
needs to be double-articulated [24]. Otherwise, the situation and Stewart undertook shear tests between steel and rock
shown in Fig. 34 will eventuate whereby the flipper can only act as mentioned earlier in the paper). A friction angle of 26°
to control the face and not the roof when the face has spalled. results in a coefficient of friction is only 0.5 [8].
The available load capacity of a flipper arrangement is generally (2) Accepting that the potential for roof instability on a longwall
small, however in terms of acting to control a buckling/guttering face increases as the face height increases, an increasing face
type failure mechanism in the tip to face area, roadway roof sup- height also results in the legs becoming more vertical. There-
port knowledge shows that relatively small vertical loads can pre- fore any active horizontal force that may be generated
vent buckling under much higher levels of horizontal stress [25]. decreases as a direct consequence. In other words, when it
Therefore despite a low load rating, a flipper that can be positively would be most required it is least able to be generated.
set to the roof offers potential roof control advantages. (3) Model shown in Fig. 5 has the immediate roof ahead of the
However, the limitations of a flipper are two-fold, namely: shields initially failing due to horizontal stress driven buck-
ling and guttering, the horizontal stress being generated due
(1) That like a shield, at some point it needs to be retracted from to bending of the roof strata. Therefore adding an additional
the roof during which time if it were preventing roof buck- active horizontal force into this area due to the design of the
ling, such failure could then occur resulting in a small roof shields cannot be beneficial to roof stability in the tip to face
cavity which immediately then acts to reduce shield sup- distance.
porting effectiveness.
(2) Its reach is limited ahead of the tip, therefore as the amount As shown in Fig. 31, an alternative logic as to why 2 leg shields
of face spall increases, its overall effectiveness in terms of demonstrably result in improved roof control than 4 leg shields
roof control decreases. (which was the context behind the development of the active hor-
izontal force concept) is that they are more amenable to set the
Overall, a double-articulated flipper is a useful addition to a canopy tip to the roof and keep the total leg load as close to the
longwall shield, particularly in high seams with friable roof. How- face as possible.
ever it only assists roof control and will not usually prevent large The primary concern with leg inclination is assessed to be to
scale roof collapses that are driven by excessive face loading and ensure that it does not result in a significant reduction in vertical
associated face spall. It is certainly not a substitute for a force being applied to the roof. On the basis that the vertical force
well-designed and proportioned longwall panel layout and highly applied is equivalent to the leg load multiplied by the cosine of the
rated and well-designed longwall shield. leg inclination angle (to the vertical), a leg inclination of 20° only
results in a 6% reduction in the vertical component as compared
4.9. Inclination of legs toward the face to the absolute leg loading. When it is remembered that the max-
imum leg inclination occurs with the shield at the lower end of its
The inclination of the shield legs has been a general shield operating range, such a reduction is judged to be insignificant in
design consideration since Peng first published the concept of an terms of the overall roof instability threat.
‘‘active horizontal force’’ assisting roof stability ahead of the shield As a general shield design concept, it is suggested that leg incli-
tips (see Fig. 35) [26]. nation should be maintained at no more than 20° to the vertical
throughout the defined operating height range.

5. Future design of longwall shield ratings?

Having discussed in detail numerous factors that to varying


degrees influence both the potential for a roof fall to occur ahead
of the longwall shields and the severity of the roof fall, as well as
provided comment on both the history of shield load rating
increases over time and the role of geotechnical based design dur-
H

Fahf

Fig. 34. Inability of single articulated flippers to secure the face or roof due to
previous face spall [24]. Fig. 35. Active horizontal force (Fahf) concept [26].
R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706 705

(10) Even the highest rated longwall shields can have their effec-
tiveness significantly reduced by overlooking a whole range
of other shield design factors or less than adequate opera-
tional practices.

Based on the contents of this paper, the following statements are


made for the consideration of industry as to where to from here:
(1) Industry would benefit from characterizing roof fall poten-
tial (RFP) for longwall mines based on longwall geometries
and other established geological and geotechnical threats
to face stability. The paper has presented a series of concep-
tual mechanistic models backed by empirical data that in
combination allow this to be done on a meaningful basis.
Fig. 36. Example of longwall face instability at shallow cover depth with weathered Such a characterization would be a foundation for future
roof strata.
risk-based discussions on the need or otherwise for
increased longwall shield ratings, whether any given long-
ing that history, the final question addressed is where this may or wall mine had the potential to optimize shield design or
should lead to in the future? indeed, whether longwall extraction should be attempted
In addressing this question, it is first necessary to make a num- at all in given circumstances.
ber of what are considered to be factual statements, albeit that the (2) In the short to medium term, the highest rated shield tech-
relative magnitudes involved may still be the subject of debate: nology available (2 leg 1750 tons 2 m wide shields) has only
been used at one mine site worldwide. Given the quantum
(1) There are demonstrably geotechnical conditions whereby leap that these shields represent, industry focus should per-
longwall extraction using wide panels should not be haps be (a) in using this technology more broadly and (b)
attempted irrespective of shield rating (e.g. very high level making sure that all of the other shield design and opera-
‘‘unmanageable’’ periodic weighting effects or very shallow tional factors that can reduce shield effectiveness, are elim-
cover depth in close proximity to the depth of weathering inated from industry.
(see Fig. 36). In other words, an increased shield rating can- (3) Focusing geotechnical research work at better understand-
not address every level of threat to longwall face stability. ing and identifying (ahead of mining) those extreme geo-
(2) Shield load ratings have increased over time due to mine metrical and geotechnical conditions whereby current
operators (a) wanting to reduce the occurrences of major shield ratings are only likely to be effective in conjunction
roof collapses so as to improve mining economics and (b) with very specific additional operational controls (e.g. hori-
generally not being prepared to procure lower shield ratings zon control through faults, pre-consolidation of strata ahead
as compared to industry best practice due to risk-based con- of mining, avoiding extraction through areas of demonstra-
cerns or the belief that it may represent false economics. ble extremely high risk, etc.). This starts with the availability
(3) The increasing of longwall shield ratings over time has to all of combined industry experience in terms of major
demonstrably been successful in increasing the effectiveness strata control loss events on longwall faces and the inherent
of shields in reducing the occurrence of major roof collapses geotechnical drivers. Unlike coal pillar instability and any
(all other factors being equal). number of roadway strata instability and ground support
(4) The geotechnical design of shields has tended to follow scenarios; this has never been addressed as an
rather than lead the increase in shield ratings over time industry-wide topic, presumably because unlike coal pillar
and even now, the availability of a proven design methodol- and roadway failures, the safety of mine workers is far less
ogy to match shield ratings to a credible evaluation of roof compromised by roof fall events on the longwall face.
instability risk is missing despite many of the claims made
by the geotechnical fraternity. The final comment made is in direct response to a question
(5) Current day longwall shields are sufficient to maintain ade- posed by Barczak and Tadolini at the very end of their paper,
quately stable face conditions for the majority of the time in namely ‘‘do you really think shield capacities can increase forever’’
most geological circumstances. However major roof col- [13]? This question was posed only 2 years before the 2 leg 1750
lapses can and do still occur with some longwall mines being tons shields were procured at Moranbah North Mine, this being a
far more prone than others. quantum leap in shield rating design. Therefore in response to
(6) Longwall production expectations are increasing over time the question the author provides the following: ‘‘probably not,
and longwall mines are inevitably facing more aggressive face but for the foreseeable future it would be unwise for the industry
loading and strata conditions than was the case in the past. to close its mind to the possibility of further substantial shield
(7) Designing longwall faces to operate in ever thicker seams in capacity increases’’.
a single pass is associated with poor strata stability such that
longwall top coal caving (LTCC) may be a better option [27].
(8) Unlike mine roadways where ground support can be incre- 6. Conclusions
mentally increased during the mining cycle to account for
varying conditions, a longwall face is limited by the load rat- In addressing state-of-the-art in longwall shield load-rating
ing of the shields which represents the maximum level of design, the loading environment in which they operate needs
strata support that can ever be used during mining detailed consideration in terms of its manifestation at the longwall
operations. face and shields on a shear by shear basis and the key parameters
(9) Increasing the load rating of longwall shields invokes cost which control this manifestation. It is self-evident that representa-
and weight penalties that individual mines may be unable tively predicting face and shield loading on a shear by shear basis is
to justify or accommodate within their mining and business essentially impossible at the current time given the nature of the
plans. problem and our state of knowledge; therefore there is no credible
706 R.C. Frith / International Journal of Mining Science and Technology 25 (2015) 687–706

basis for longwall shield load rating design using strata loading as a [2] Frith R, Creech M. Face width optimisation in both longwall and shortwall
caving environments. Final Report, ACARP Project C5015; 1997.
design input.
[3] Frith R, McKavanagh B. Optimisation of Longwall mining layouts under
Another key aspect is the recognition from numerous shield massive strata conditions and management of the associated safety and
monitoring studies that they rarely, if ever, reach static equilibrium ground control problems. End of Grant Report, ACARP Project C7019; 2000.
during mining, the point being that the design problem is one of [4] Colwell M, Frith R. A roof support design methodology for longwall installation
roadways. Final Project Report, ACARP Project C19008; 2012.
non-equilibrium and the associated increasing of roof fall propen- [5] Mark C. Pillar design methods for longwall mining. United States Bureau of
sity with time during each shear cycle. Again, this does not obvi- Mines Information Circular 9247; 1990.
ously lend itself to a simplistic analysis to determine the [6] Colwell M. Chain pillar design-calibration of ALPS. End of Grant Report, ACARP
Project C6036; 1998.
necessary load-rating of longwall shields. [7] Colwell M. Rib support design methodology for Australian collieries. Final
The fundamental barrier in terms of attempting to understand Report, ACARP Project C11027; 2004.
and numerically analyze the link between longwall shield rating [8] Frith RC, Stewart AM. Optimisation of powered support performance in
relation to strata loading and engineering criteria. End of Grant Report,
and roof fall propensity ahead of the shields, is that the problem NERDDC Project 1445; 1993.
contains too many indefinable variables. Furthermore, the rating [9] Frith R. Development and demonstration of a longwall monitoring system for
of the longwall shield does not directly control roof fall propensity, operational decision-making. End of Grant Report, ACARP Project C4017; 1996.
[10] CSIRO/SCT. Ground behaviour about longwall faces and its effect on mining.
as they are separated by several other controls, including overbur- End of Grant Report, ACARP Project C5017; 1999.
den weighting, the nature of the immediate roof strata, the amount [11] Luo X, Hatherly P. Pre-cursors and new understanding of high gas emissions at
of face spall and any number of operational variables which may Appin colliery through micro-seismic monitoring. End of Grant Report, ACARP
Project C6025; 1998.
not be consistent on a shear by shear basis. With this being the
[12] Medhurst TP. Practical considerations in longwall support behaviour and
case, the inevitable conclusion reached is that credible and mean- ground response. In: Proceedings of coal. australasian institute for mining and
ingful load-rating design for longwall shields still remains beyond metallurgy. Melbourne; 2005.
the reach of the geotechnical fraternity. [13] Barczak T, Tadolini S. Longwall shield and standing gateroad support designs-
is bigger better? NIOSH Publication; 2007.
The paper has concluded that the same basic operational drivers [14] Gale WG. Matching longwall support design to industry requirements in weak
for increasing shield ratings in the past are still as relevant today, ground. End of Grant Report, ACARP Project C14030; 2009.
albeit that the trend may plateau for a period following the quan- [15] Wilson A. Support load requirements on longwall faces. Min Eng
1975;134:479–91.
tum leap associated with the development of the 2 leg 1750 tons [16] Smart B. The evaluation of powered support performance from geological and
shields for Moranbah North Mine which have yet to be tested at mining practice information. In: Proceedings of 27th US rock mechanics
any other mine. With this being the case, the author contends that symposium. Alabama; 1986.
[17] Martin K, Kizil M, Canbulat I. Analysing the effectiveness of the 1750 tonne
the geotechnical focus for the foreseeable future should be in fur- shields at Moranbah North Mine. In: Proceedings of the coal operators
ther defining and proving the key geological and geometrical con- conference. Wollongong: University of Wollongong; 2012.
trols for longwall face stability and therefore longwall production [18] McDonald A. Longwall and shortwall. Australian Coal Mining Practice; 1986. p.
328–63.
risk as well as ensuring that the management and operation of [19] ACARP. High Capacity underground coal mining: scoping study-defining the
longwall faces makes the best possible use of currently available system needs for a 15 MTPA longwall operation. End of Grant Report, ACARP
longwall shields. Only when the highest rated shields that are Project C15008; 2007.
[20] Klenowski G, Ward B, McNabb K, Dyer G. Prediction of longwall support
available are demonstrably being used to their absolute best per-
loading at southern colliery Queensland. In: Proceedings of 11th international
formance limit, should industry consider whether a further signif- conference on ground control in mining. Wollongong; 1992. p. 155–9.
icant increase in shield load rating is required. Industry is far from [21] Seedsman R. The strength of the pillar–floor system. In: Proceedings of the coal
this point at the current time. operators conference. Wollongong: University of Wollongong; 2012. p. 24–31.
[22] Trueman R, Lyman G, Cocker A. Managing roof control problems on a longwall
The final comment made is one of an historical perspective and face. In: Proceedings of the coal operators conference. Wollongong: University
how it may guide our current day thinking as to the future. In 1986 of Wollongong; 2008. p. 10–21.
when McDonald provided their historical perspective on shield [23] Trueman R, Thomas R, Hoyer D. Understanding the causes of roof control
problems on a longwall face from shield monitoring data – a case study. In:
load rating development over time, it would have been inconceiv- Proceedings of the coal operators conference. Wollongong: University of
able to them that in 2015 a single longwall face would have the Wollongong; 2011. p. 40–7.
ability to produce in the order of 10 MTPA and 2 leg 1750 tons [24] Payne D. Crinum mine, 15 years, 40 million tonnes, 45 roof falls-what did we
learn? In: Proceedings of coal. University of Wollongong, Melbourne,
shields would be available for use [18]. This was less than 30 years Wollongong: Australasian Institute for Mining and Metallurgy; 2008. p. 22–43.
ago, therefore it would perhaps be unwise for any of us to close our [25] Colwell M, Frith R. AMCMRR – an analytical model for coal mine roof
minds as to what new technologies in the future may allow in reinforcement. In: Proceedings of the coal operators conference. Wollongong:
University of Wollongong; 2010. p. 73–83.
terms of longwall shield design, If our predecessors had done this [26] Peng SS. Design of active horizontal force for shield supports for controlling
in 1986, the mining industry today would be little different from roof falls. Min Eng 1990;149(345):457–61.
that time. [27] Hebblewhite B. Status and prospects of underground thick coal seam mining
methods. In: Proceedings of 19th international mining congress and fair of
As Orville Wright stated shortly after the first powered flight
turkey. IMCET; 2005. p. 169–78.
‘‘no flying machine will ever fly from New York to Paris. . .’’ And
then a guy called Lindbergh came along.

References

[1] Frith RC. Half a career trying to understand why the roof along the longwall
face falls in from time to time? In: Proceeding of 24th international conference
on ground control in mining. West Virginia; 2005.

You might also like