International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61

Contents lists available at ScienceDirect

International Journal of Rock Mechanics Mining Sciences

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

Analytical solution for assessing continuum buckling in sedimentary rock

slopes based on the tangent-modulus theory
crossmark
⁎
Sergio Esteban Rosales Garzon
Drummond Ltd., Mina Pribbenow km 31 vía San Roque - Bosconia, Cesar, Colombia, South America

A R T I C L E I N F O

Keywords:
Stratified rock
Surface coal mine
Slope stability
Buckling failure mode

1. Introduction
Slopes in stratified rock masses cut or naturally parallel to bedding
form a non-daylighted slab structure. When these slopes contain a
continuous weak bedding plane or a persisting discontinuity, and the
rock mass lacks any relevant cross-jointing able to favor wedge, planar
or ploughing failure, then the kinematic potential for buckling emerges.
Buckling collapse hazard in a rock mass slope1–6 suppose an absolute
instability where the failure is often rapid and without previous easily
recognizable warnings. When the slope reaches a critical height, any
extra height, pore pressure increase or small external load could cause
it to buckle. Whether in the case of a footwall slope of a surface coal
mine or in a natural slope, that potential risk demands a rigorous
assessment of the slope stability for proper design or mitigation. The
use of the limit equilibrium technique in combination with Euler's
buckling theory has resulted in some closed solutions.7–12 However, as
stated by others1,6,13,14 and from the author’s own experience, these
approaches have proven unable to consistently explain the buckling
failure mode. This article suggests a formula that can be used to
calculate the length of the passive segment and the use of the tangent-
modulus theory to identify the mode of failure and evaluate slope
stability. Nine real buckling cases were analyzed with acceptable
results.
2. Length of the passive segment
In a slope under buckling potential, the overall slope length, Ls, can
be divided into two different regions: the “passive segment”, which is at
the toe of the slope, and the active region, which is the section above
the passive segment and is called the “driving segment” (see Fig. 1).
During failure, the “passive segment” starts bulging upward and
progressively releases the “driving segment” along the dip direction
in a concomitant translational failure mode.
A one span beam-column with a pin-roller support on its left end
and a pin support on its right end is used to model the passive segment
(see Fig. 2). The bedding parallel to the slope face is modeled as an
orthogonal fracture system with a flat overall geometry. Since the
beam-column is assumed to be straight, it is necessary to recreate this
same natural condition in the model by introducing an auxiliary
eccentric compressive force, T, at the level of the pinned supports.
The magnitude of this eccentric force shall be exactly the force needed
to compensate the downward deflection due to self-weight and thus
keep the beam-column straight.
Beam-column cannot deflect downwards due to natural constraints,
nor can they deflect upwards assuming negligible tensile strength (as is
common for major surface stratified rock masses). Therefore, in this
compression-only setting, the critical point from which equilibrium is
lost can be defined as the moment when the beam-column begins to
deflect upward. Before this critical point, the beam-column is still
straight and can develop compressive strength, and after this critical
point, the beam-column fails by losing its compression-only status. The
word “straight” is used here to mean a beam-column in a compression-
only setting; consequently, curved beam-columns while working in a
compression-only setting, in essence, also adhere to this meaning.
Based on this failure mechanism, at the critical point, the length of
the passive segment,ℓ , shall be such that the total midspan deflection
in the beam-column model is zero. In addition, the driving force, P, is
evaluated by isolating the potential driving segment as illustrated in
Fig. 3. By summing the forces along dip, we can obtain the expression
(1) for the resultant driving force P. In this expression, rock mass unit

⁎
Correspondence address: Cra 82A #50A 83 Calasanz, Medellín, Antioquia, Colombia, South América
E-mail addresses: srosales@drummondltd.com, ingsergiorosales@gmail.com.

http://dx.doi.org/10.1016/j.ijrmms.2016.10.002
Received 28 March 2016; Received in revised form 29 August 2016; Accepted 11 October 2016
1365-1609/ © 2016 Elsevier Ltd. All rights reserved.
S.E. Rosales Garzon
Fig. 1. Sketch of buckling failure mode in slopes of stratified rock.
Fig. 2. Model for the passive segment (eccentrically loaded beam-column with self-
weight).
Fig. 3. Equilibrium of forces for the driving segment.
weight, γ, slab thickness, d, friction angle of the failure surface, ɸ,
cohesion of the failure surface, C, slope angle, α, and dip overall slope
length,Ls , are known, and the length of the passive segment along the
dip, ℓ , is unknown. Under certain conditions of slope angle, depth of
the potential failure surface and sliding surface friction and cohesion, it
is possible to obtain a negative driving force. This negative result
means that the buckling failure mode could be not a real concern in
such a setting.
P = (γd sin α − C − γd cos α tan φ)(Ls − ℓ)
The two segments, active and passive, have been evaluated in-
dependently. To integrate both together, we need to ensure the
equilibrium condition. The driving force, P, in the driving segment
acts along dip direction as well as axially, assuming that the center of
gravity matches the geometric center of the slab. As explained before,
the eccentric force and the self-weight in the model for the passive
segment cancel bending moments each other, leaving only the axial
force along dip, T. Consequently, and only at the critical point, it is true
International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61
that the force P shall be equal to the force T to guarantee the overall
equilibrium of the slope. This equilibrium exercise implies rigid blocks;
nevertheless, for the purpose of surface sliding analysis where the
stress is relatively low, it is reasonable to assume that the blocks are
rigid.2
Using deflection formulas derived from the Euler-Bernoulli beam
equation15,16 we can evaluate the midspan deflection. If only the weight
of the beam is considered in the beam-column model (T =0 ), then the
maximum downward elastic deflection (ymax ↓)16 (p. 193) is:
ymax ↓ = 5qℓ 4/384EI (2)
If only the eccentric force is considered in the beam-column model
(q=0 ), then the maximum upward elastic deflection (ymax ↑)16 (p. 194) is:
ymax ↑ = Teℓ 2/8EI (3)
Therefore, to obtain zero midspan total deflection, through the use
of the principle of superposition and setting T equal to P, we find at the
critical point, the downward deflection from Eq. (2) shall be equal to
the upward deflection from Eq. (3):
5qℓ 4/384EI = Peℓ 2/8EI (4)
Substituting P as defined in Eq. (1), inserting the eccentricity, e,
equal to d /2 and then simplifying the results for Eq. (4), we obtain the
quadratic equation for ℓ :
ℓ 2 + [9.6(γd sin α − C − γd cos α tan φ)/ q]ℓ
− [9.6(γd sin α − C − γd cos α tan φ) Ls / q] = 0 (5)
The quadratic Eq. (5) has the following simplified solution for its
positive root; in this equation, the constant C1 and the beam-column
self-weight, q, are known:
ℓ= C12 + 2C1 Ls − C1 (6)
C1 = 4.8(γd sin α − C − γd cos α tan φ)(d /2)/ q
q = dγ cos α
In conclusion, for any given slope, the length of the passive segment
can be calculated using Eq. (6) as a function of the overall slope length
along the dip. When the toe slope contains a nose or a natural
imperfection, the passive segment length is no longer the one identified
above. In this case, it is believed that passive segment length must be
the same as the nose length or controlled by the imperfection itself.2,3,6
3. Compressive strength
The tangent-modulus theory by Shanley for column buckling
concluded that the lateral deflection starts very near the tangent-
modulus load, Pt, that is, the concept of tangential modulus predicts
the maximum load that can be applied to a column before it undergoes
lateral deformation.17 Given that we previously defined the straight
state as the needed requirement for the beam-column model to be in
equilibrium, it is reasonable to adopt the tangent-modulus theory.
To properly investigate the buckling strength of columns loaded in
the inelastic range using tangent-modulus theory, it is necessary to
know the real stress-strain diagram of the rock mass uniaxial state of
stress. Using previous stress-strain diagram, we can deduce the
(1) corresponding tangent-modulus, Et. Finally, by introducing this tan-
gent-modulus into Engesser's formula, we can obtain the buckling
formulas (7) and (8) needed to assess, respectively, the compressive
load and the compressive stress in the inelastic domain.18
Pt = π 2Et I /ℓ 2 (7)
σt = π 2Et /(ℓ/ r )2 (8)
On the other hand, the buckling force of a column in the elastic
domain is obtained from Euler's buckling formulas (9) and (10) which
54
S.E. Rosales Garzon
provide, respectively, the compressive load and the compressive stress
in the elastic domain.
Pe = π 2EI /ℓ 2 (9)
σe = π 2E /(ℓ/ r )2 (10)
Euler's formulas (9) and (10) are valid only when the compressive
stress, σe, is less than the proportional limit, σp. By introducing the
expression for σe from (10) in the inequality σe < σp and solving for
slenderness ratio (λ = ℓ/ r ) we obtain the valid limit (11) of formulas (9)
and (10).
λ > π E / σp (11)
By definition, the ultimate uniaxial compressive stress σu of the
column material defines the upper-bound of the buckling stress range
according to tangent-modulus theory. In other words, compressive
stress shall be always equal to or less than the ultimate compressive
stress,σ ≤ σu . As such, we assumed that the ultimate compressive
strength of the rock column corresponds to the uniaxial compressive
strength of the rock mass, UCS, because, according to Hoek-Brown
failure criterion, failure initiates at the boundary of an excavation when
UCS is exceeded by the stress induced on that boundary.19
Once Et is known, the rest of the variables in formulas (7)–(11) can
also be deduced. The length of the column, ℓ , can be estimated as
explained earlier in Section 1. Because the bedding is parallel to the
driving compressive force, the deformation modulus of the rock mass,
E, can be set equal to of the deformation modulus of the intact rock, Ei.
The radius of gyration r = I / A is given by the geometry of the
problem. The column's effective length factor k = 1.0 is used to account
for the column with pinned ends as modeled in the previous section.
This end condition is very often encountered in practical applications13
and is also the fundamental case from which Euler’s formula was
derived.
4. Critical slope height
The critical slope height Hcr is defined as the slope height for which
the ratio between the critical force of the column Pcr and the driving
force P acting on it (1) is 1.0. Pcr could be defined as either Pt or Pe,
depending on the case.
5. Case study results
Nine buckling failures reported in the literature were selected to
conduct an evaluation of the above approach for buckling assessment.
Three of them describe different buckling failures at the surface coal
mines of Malvern Hills1, Quintette2,20 and Westfield4. Other three
come from physical models (friction table) developed to understand
buckling failures at Grand-Baume coal mine.3,21 To complete the
sample, other three cases were selected to account for buckling in
natural slopes namely in Bawang Mountain5,22 and Lavini di Marco6.
Next, we describe the failure mechanism and the factors that could or
not influence the instability for each case study.
Seale1 described a large slab toe-buckle footwall failure that
occurred overnight on 1 July 2004, when the excavator operator
arrived to extract the three-meter-thick coal seam that was exposed
the day before. Sometime during the night, a two-meter-thick slab had
translated 6.2 m down dip and buckled, developing a broadly cylind-
rical fold structure at the toe of the slab partially covering the main
seam coal, see the photo on the left in Fig. 4. The failure extended 85 m
along the length of the 355 RL bench. To achieve an economical pit
design and minimize waste in the coal-bearing units that are steeply
dipping (45 ± 5°), the footwall batters were cut to parallel the dip of the
bedding. The footwall design has 5 m wide benches cut at 15 m vertical
intervals, creating batter dip slopes of 21 m in length. The failure
occurred immediately following the rapid excavation of the toe (355 RL
55
International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61
bench formation). Both ends of the failure terminate at a joint
perpendicular to bedding that intersects the 370 RL bench. This end
joint offers a side release mechanism that may also contribute to the
character of toe-buckle fold. Jointing perpendicular to bedding occurs
frequently in the pit, with several joints separated by up to 3 cm of
orange gouge. Daily June rainfalls never reached more than 8 mm and
there was no measurable rainfall from 26 June to the time of the toe-
buckle failure, therefore, rainfall may not be a direct contributing
factor. The relatively rapid unloading of the slope during batter
formation was believed to cause both pore pressure effects and
strength-lowering to residual friction angles of the failure surface
material by wetting up.
Scoble4 described that an extensive but thin slab in the pavement of
the Westfield Shale coal (“P” coal) failed catastrophically at 10:30 pm
on 14 January 1974 at the western wall of the mine's Phase II
excavation, see photo on the right in Fig. 4. A bulldozer at the base
of the slope was engulfed and the driver killed. Illumination of the
slope at the time was confined to the toe area but two witnesses in a
tractor at the northern toe area stated that the middle of the face first
bulged and then slid down over the lower part, breaking into fragments
as it fell. No signs of distress were observed in the slope prior to failure.
It had been standing for four years. The possibility exists that an initial
displacement of the lower slab, caused by buckling at its toe, sliding on
a cross-over joint or fracture through intact mudstone may have
occurred, thus allowing the more noticeable and devastating overriding
movement of the upper slab. Groundwater pressures are not directly
indicated to have contributed significantly to failure. No blasting
occurred nearby the pavement during the period immediately preced-
ing the failure.
Quintette Operating Corporation owned an open pit coal mine
located near the town of Tumbler Ridge in northeastern British
Columbia, Canada. At this property, hosted in stratified sedimentary
rock situated on the eastern slope of the Rocky Mountains, Wang et al.2
described a buckling failure event. The footwall with sandstone slabs
potentially subject to sliding was exposed from a crest elevation of
approximately 1650 m to the pit floor at approximately 1455 m. The P2
Pit footwall had been gradually exposed since startup without sig-
nificant problems. In August 1997, when the pit floor reached an
elevation of 1530 m, two thin slabs, each approximately 0.75 m thick,
slid over a wedge-shaped area, see Fig. 5. The failure extended
approximately 90 m above the pit floor and was approximately 150 m
wide. The unloading during the end footwall slope formation was the
only factor believed to have caused this failure event.
Choquet et al.3 and Soukatchoff et al.21 described the analysis of the
strata buckling mechanism at the Grand-Baume coal mine in France.
At this mine, a 250-meter-high footwall with an adverse bedding dip of
40° was planned. The open pit of Grand-Baume (Gard-France) allows
access to the Stephanian coal deposits within; the northwest area of
this open pit comprises an overthrust fold. The coal is extracted in the
reverse side wall of the fold and the slope of the open pit, which has an
angle of approximately 40°. In this slope, planar failures between
benches have been observed, see Fig. 6. These phenomena are
explained by a buckling effect. Choquete et al.3 analyzed the slope
using a combination of physical modeling (friction table) and three-
hinge buckling limit equilibrium analysis. Assuming a passive region of
concave curvature toward the pit extending approximately 60% of the
overall slope length, friction table analysis yielded a critical height of
115 m for a layer 0.5 m thick, a critical height of 170 m for a layer 1 m
thick, and a critical height of 240 m for a layer 1.5 m thick.
Bawang Mountain landslide is a natural slope failing through
buckling described by Yang.5 The Bawang Mountain landslide, with a
slope height of 940 m and a slope angle of 40°, occurred in 1983 (see
schematic graph on the top of the Fig. 7). The slabs of this slope are
composed of Denying limestone strata. The total thickness of the failed
slabs is 10 m, and 3 separate slabs were involved in the failure. The
slabs, approximately parallel with the slope surface, are separated from
S.E. Rosales Garzon International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61

Fig. 4. On the left a photo of the large toe-buckle, footwall failure that occurred on 1 July 2004 at the Malvern Hills surface coal mine, inland Canterbury, New Zealand indicatig bench
levels and the toe-buckle slab dimensions (From Seale1). On the right a photo of pavement slab failure occurred on 14 January 1974 at Westfield surface coal mine in the United
Kingdom (From Scoble4).

one another and from an underlying weak interlayer composed of
broken marlite, formed by relative sliding between interfaces. The
stratum below the weak interlayer consists of massive limestone. Under
the long-term gravitational force the slabs near the toe of the slope
buckled gradually and the slabs close to the slope top were torn away.
Tommasi et al.6 described a buckling instability in a high natural
slope known as the case of Lavini di Marco. For approximately 3 km
south of the town of Rovereto (Trento, Italy) the left flank of the Adige
river valley is formed by well stratified limestones with clayey and
marly interbeds. The slope surface created by the youngest and largest
of these slides (Lavini di Marco) is wrinkled by buckle folds which were
first noticed in 1969 as a sliding of rock layers after their progressive
“heave”. Two nearby but independent buckling areas are reported. The
first, which is referred as “three-hinge buckling” (case 5) is located on
the Lavini slide surface between elevations 700 and 750. In this area
layer dip progressively increases proceeding downhill, giving the slope
surface an overall gentle convexity. Buckling involves a 2.7 m thick
pack of several layers which have a thickness ranging between 0.5 m
and 1 m, and appear subdivided by the subvertical joint sets. The
rotation axis is parallel to the slope strike and normal to the sliding
direction. Sliding of the pack of layers produced a gaping tension crack
which results from the combination of two joint sets. The second,
which is referred as “flexural buckling” (case 6) is observed in an area
located at a lower elevation, downhill from the change of curvature
marking the axis of the “three-hinge buckling”. In this area the lower
part exhibits an apparent concavity. “flexural buckling” involves a thick
pack (5–6 m) of layers that are significantly thinner than those
displaced by the “three-hinge buckling” (thickness ranging between
10 cm and 20 cm). See schematic graph of these two cases on the
bottom of Fig. 7. Deformations develop according to cylindrical
symmetry parallel folds. All the area was monitored between 1992
and 2002 using a Synthetic Aperture Radar Interferometry (INSAR)
resulting in an active downward displacement of average velocity of
about 3 mm/year. During the autumn 2000 a period of intense rainfall
lasting some weeks triggered important landslides in the Trento
province but it was no detectable in the displacement-time curves.
Such behavior suggests that the relationship, if any, between buckling
deformations and hydraulic conditions in not simple. Furthermore,
after numerical analysis the buckling is only explained by introducing
significant uplift pressures; however such water pressure induces

Fig. 5. Photograph showing buckling failure surface occurred on August 1997 at Quintette coal mine in northeastern British Columbia, Canada (From Wang et al.2).

56
S.E. Rosales Garzon International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61

Fig. 6. Left photograph shows the Sandstone layers at elevation 250 m of West Wall in Grand-Baume coal mine in France (note the concave curvature of layers towards inside of pit) and
on the right the corresponding graphical interpretation of the failure mechanism (From Choquet et al.3).

ties for each case. Data constraining the geometric parameters for each
case, including the thickness of the unstable rock slab, d, and its
corresponding slope angle, α, have been compiled from the literature.
Additionally, the rock mass unit weight, γ; the strength properties of
the failure surface, ɸ, C, the uniaxial compressive strength of the intact
rock material, σci, and the modulus of deformation of the rock mass, E,
are also sourced from authors’ reports.

5.1. Ultimate compressive strength UCS

The rock mass uniaxial compressive strength, UCS, is determined

Fig. 7. The top schematic diagram shows the failure process of Bawang Mountain
landslide (From Yang5); a) Prior to deformation; b) During deformation. And the bottom
schematic diagram shows the profile of the left flank of the Adige valley at south of
Rovereto, North-Easters Italy (From Tommasi et al.6).
noticeable displacements normal to the bedding that are not observed
in situ.
Table 1 summarizes the reported parameters and material proper-
on a case-by-case using the Hoek-Brown failure criterion.19 With this,
we need three parameters to derive UCS; uniaxial compressive strength
of the intact rock material, σci, disturbance factor, D, and the geological
strength index, GSI. The value of σci is known, see Table 1. In cases 4, 5
and 6, D is equal to zero to account for the undisturbed rock mass of
natural slopes. The end footwalls at coal mines (cases 1, 2, 3, 7, 8 and
9) are composed of the remaining in-situ material after the recovery of
the last recoverable coal seam; because the coal is brittle and soft,
excavation can be usually carried out by ripping and dozing (mechan-
ical excavation) so the degree of damage to the slope correlates well
with a disturbance factor of D=0.7. To determine GSI, we correlated

Table 1
Physical and mechanical properties of the rock mass and failure surface materials.

# Case Rock mass α (°) Failure surface

Name Kind of slope Type σci (MPa) E (GPa) γ (kN/ Material d (m) Ø (°) C (kPa)
m3)

1 Westfield Coal mine footwall Mudstone 42 11 20.5 35 Thin clay band 3.5 22 0
2 Malvern Hills Mudstone/Siltstone 25 0.61 22 42 Thin coal seam 2 3 0
3 Quintette Sandstone 50 30 27 44 Carbonaceous parting 1.5 25 0
4 Bawang Natural Limestone 80 50 27 40 Marlite layer 10 17 40
5 Lavini di Marco Limestone 55 30 27 22 Clayey-marly interbed 2.7 18 0
6 20 5.5 15
7 Grand-Baume Physical model (friction Sandstone/ 51.5 30 25 40 Planar clean contact 0.5 25 0
8 table) Siltstone 1
9 1.5

57
S.E. Rosales Garzon
Table 2
Selection of the Geological Strength Index, GSI, based on descriptive inputs sourced from the cases’ literature, and the required parameters to apply the Hoek-Brown failure criterion and
thus obtain the uniaxial compressive strength of the rock mass, UCS, for each case.
the rock types and the descriptive inputs on structure and surface
condition with the existing charts for GSI selection.23 Coal mine cases
are heterogeneous rocks, so accordingly, we used the GSI chart
“Geological Strength Index for heterogeneous rocks such as flysch”
provided by Marinos and Hoek,23 while for the homogeneous rock of
cases 4, 5 and 6, we used the “General chart for GSI estimates from
geological observations” provided by Marinos and Hoek23. Table 2
summarizes the corresponding parameters used to apply the Hoek-
Brown failure criterion and the resulting UCS for each case. The
columns “Rock Type” and “Descriptive Input” of Table 2 contain the
descriptive inputs sourced from forensic descriptions, pictures and
field observations, reported in the literature and earlier described in
this section for each case study.
5.2. Tangent-modulus Et
The stress-strain diagrams are not reported for the case studies.
Nevertheless, by using some rock mechanics concepts and the method
of determining the buckling stress as developed by Ylinen 24, we can
estimate a solution. For the approximation of the stress-strain diagram,
Ylinen proposes the expression (12) in terms of the first derivate with
respect to the strain.
∂σ /∂ε = (a − σ )/(b − c′σ )
where σ is the stress, ϵ is the strain and a, b, c′ are three free
parameters, the values of which should be determined so that the
stress-strain function deduced from Eq. (12) by integration suitably
International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61
agrees with the stress-strain diagram.24 After applying some boundary
conditions Ylinen provides these three parameters on the basis of the
modulus of elasticity, E, proportional limit, σp, and compressive
strength, UCS, of the material:
a = UCS
b = UCS / E
c′ = c / E
c = 1 + (Eδp /UCS )/(σp /UCS ) + ln(1 − σp / UCS )
In addition, the parameter c is affected by the quantity δp , which
indicates the allowable deviation from Hooke's law at the proportional
limit. In all study cases, the required δp to calibrate the stress-strain
diagrams and capture the expected behavior for each rock type was
0.0000005; this figure also conforms to the acceptable range of δp since
it is less than the maximum values allowed between 0.00002 and
0.0002. 24
Deere and Miller classified the uniaxial stress-strain curves into six
types.25 We considered the first three types of behavior, given that they
correspond better with the rock types of the nine selected cases. With
this, we defined the values for the proportional limit, in terms of the
ratioσp / UCS , that better describe the type of curve of stress-strain
behavior, after matching and grouping all case studies by rock type.
(12)
Table 3 summarizes previous analyses and shows the resultant para-
meter c and proportional limit for each type of behavior.
By integrating the expression (12) and applying additional bound-
ary conditions, Ylinen demonstrates the expression (13) as the stress-
58
S.E. Rosales Garzon International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61

Table 3
Correlation between the Rock Type for each case study and the types of curve of stress-strain behavior according Deere and Miller25, and Bell26.

Fig. 8. Dimensionless stress-strain diagrams for each case study applying Ylinen's method24 and the types of curve of stress-strain behavior according Deere and Miller25, and Bell26.

strain function.24 By introducing the required parameters for each case
study into this expression, we can develop a stress-strain function for
our cases. To obtain a general idea of the form of the stress-strain
curves in an advantageous comparative form, Fig. 8 shows the
dimensionless diagram of σ / UCS plotted against E ϵ/ UCS . We see that
the more inelastic the rock is, the greater is the deviation of the stress-
strain curve from the Hooke's line.
ε = 1/E [cσ − (1 − c ) UCS⋅ ln(1 − σ / UCS )]
Finally, by introducing the tangent-modulus function (12) into
Engesser's formula (8) and taking into consideration that σ = σt when
inelastic buckling occurs, Ylinen's method provides the following
equation for buckling stress σt as a continuous function of the
slenderness ratio λ valid for the entire domain (elastic and inelastic):
σt = (π 2E + UCSλ2 )/2cλ2 − (π 2E + UCSλ2 )2 − 4π 2EcUCSλ2 /2cλ2 (14)
In Fig. 9, the ratio σt / UCS is plotted against the dimensionless
quantity λ / π σp / E . At great values of the slenderness ratio, when the
buckling stress is low, the value of the tangent-modulus is constant and
(13)
the buckling stress diagrams coincide with Euler's hyperbola. When the
slenderness ratio decreases, the buckling stress increases, the value of
the tangent-modulus decreases and the buckling stress diagrams
deviate from Euler's hyperbola and approach the ultimate compressive

59
S.E. Rosales Garzon International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61

Fig. 9. Buckling stress diagrams for each case study in dimensionless form according to formula (14).

strength of the material, UCS.
In terms of the mode of failure, when λ approaches zero, stress
approaches UCS and failure is brittle. When λ satisfies the limit
stablished by expression (11), the failure is elastic. In between,
buckling occurs after the stress in the column exceeds the proportional
limit of the column material and before the stress reaches the ultimate
strength. This last situation is called inelastic buckling.
5.3. Results
Table 4 summarizes the results; in Fig. 9, the dots labeled with a
number indicate the exact point in which the failure in each case
occurred. It is important to notice that cases 7 and 8 fall in the zone of
pure elastic behavior, case 9 falls in the zone of inelastic behavior close
to the elastic zone, and cases 1, 2, 3, 4, 5 and 6 behave brittlely since
they plot in the top left area of the stress diagrams, where failure is
linked to the ultimate compressive strength of the column material and
is not related to its geometry. This brittle behavior appears when the
stress equals or is very near to the UCS; although there is no defined
limit for this extreme, a sensitivity analysis shows that a ratio of
σt / UCS ≈ 0.98 overestimates the critical height by approximately 20%.
Table 4 also shows that the actual failure height, Hf, for each case is
satisfactorily predicted compared with the predicted critical height
Hcr. The critical slenderness ratio λcr =π E / σp is compared with the
slenderness ratio for each case to define the elastic behavior, as
explained in Section 3. Finally, the critical stress or stress at failure,
σf, is presented to highlight that for the elastic or inelastic behaviors
the stress is always lower than UCS.
6. Discussion
Other methods use empirical ratios of the length of the buckle to the
total length of the slope (ℓ/ Ls ) to determine the length of the buckle.
The determination of these ratios usually depends upon individual
author criteria or upon base friction modeling.27 The ratios appear
remarkably varied in the literature with values from 0.36 to 0.46,1 0.5,9
or even as high as 0.66.21 In contrast, the proposed method is based on
the deterministic calculation of the length of buckle through a formula
that uses the geometry of the slope and the strength of the potential
failure surface. Although, in practice, a direct measurement of this
length is rare due to the immediate collapse of the structure after
failure, the Malvern Hills case reports an estimated buckle segment
length of 12 m, which agrees with the length of 11 m calculated by Eq.
(6) for this case. Additionally, the length of the buckle as calculated
herein has proven valid to the extent that the critical height prediction
for all cases was satisfactory. The mean relative error for the predicted
critical height, Hcr, according to this method, is 2%, with a standard
deviation of 7%. This can be contrasted with the solution applying the
approach developed by Cavers 9, for which we obtained a relative error
for the predicted critical height, Hc, of 211%, with a standard deviation
of 249%. Although Cavers’ approach fails in the overall prediction, it
seems to approach better to the results when the case is closer to the
elastic mode. This is reasonable because this approach is based on
Euler's buckling theory.
Analysis of the results also highlights the relevance of the passive
segment slenderness ratio (ℓ/r) and, to select the proper stability
criteria for each case, the necessity of establishing if the mode of
failure is brittle, elastic buckling or inelastic buckling. When stability is
controlled by elastic buckling, the slope follows the Eulerian behavior
which is only based on the geometry of the rock column and the
modulus of elasticity of the material. On the contrary, when brittle
failure controls, the slope stability appears to depend exclusively on the
rock mass uniaxial compressive strength of the material forming the
slab. In between, under inelastic buckling, the slope stability obeys to a
combination between material strength and buckling. The review of the
cases shows that the buckle geometry sometimes could appear in the

Table 4
Results for the length of the passive segment, ℓ ; critical and at-failure slenderness ratios λcr andλ; compressive strength and at-failure stress UCS and σf ;and critical height predicted by
this method Hcr contrasted with the actual height of failure Hf and the critical height predicted by Caver's approach Hc.9

# Case Rock type σf (kPa) UCS (kPa) ℓ (m) λ λcr Mode of failure Hf (m) Hcr (m) Hc (m)

1 Westfield coal mine Mudstone 612 612 25 24 596 Brittle 92 84 403

2 Malvern Hills coal mine Mudstone/siltstone 197 197 11 19 247 15 17 68
3 Quintette coal mine Sandstone 1076 1076 20 46 587 90 92 225
4 Bawang Mountain landslide Limestone 9958 9959 152 53 235 940 1005 1119
5 Lavini di Marco Limestone 1651 1651 29 38 446 170 171 582
6 796 796 29 18 525 95 88 817
7 Grand-Baume coal mine Sandstone/siltstone 516 1342 109 754 525 Elastic 115 117 127
8 818 172 595 170 184 202
9 1061 223 515 Inelastic 240 239 265

60
S.E. Rosales Garzon
brittle mode, for example in the cases 2, 4, 5 and 6. In these cases, the
buckle form observed on field after collapse could become a confounder
when identifying the specific mode of failure. The most plausible
explanation is a progressive failure related to the non-uniform stress
distribution in the slab; failure initiates at the most crucial point close
to the bottom of the slab and propagates gradually to ultimate failure.
In the process, the rock mass at the bottom of the slab crushes and
distorts relatively faster, resulting in the heaving (in the form of buckle)
of the relatively less affected rock mass at the top of the slab.
Additionally, the results also reveal that buckling can be explained
only by gravity without invoking other extreme factors like significant
water pressures, which usually are difficult to substantiate for near
surface fractured rock masses.
7. Conclusion
A formula to calculate the passive segment length was derived by
treating this passive segment as a beam-column stability problem and
guaranteeing equilibrium in elastic range. Then, by considering the
known extension of the passive segment and the tangent-modulus
theory, the article provides a conceptual understanding of the buckling
stability of stratified slopes, yielding a proper and workable tool for
evaluating whether the mode of failure is brittle, elastic buckling or
inelastic buckling. If the failure mode is brittle, the critical load
depends exclusively on the uniaxial compressive strength of the rock
mass. If the failure mode is through elastic buckling, the critical load
can be calculated applying Euler's theory. Finally, if the failure mode is
through inelastic buckling, the critical load can be assessed using the
tangent-modulus according to Shanley's theory. The stability analysis
on nine reported cases using the suggested method shows good
correlation between the predicted critical height and the actual height
of failure in each case.
Conflict of interest
I wish to confirm that there are no known conflicts of interest
associated with this publication and there has been no financial support
for this work that could have influenced its outcome.
References
1 Seale J. An Engineering Geological Investigation of Footwall Toe-Buckle Instability at
the Malvern Hills Opencast Coal Mine (M.Sc. thesis), Christchurch, New Zealand:
University of Canterbury; 2006.
2 Wang B, Cavers D, Wong B. Surface buckling failure study and support design at the
Quintette coal mine, Canada. In: Taylor & Francis Group, editor. Landslides:
Evaluation and Stabilization/Glissement de Terrain: Evaluation et Stabilisation, Set
of 2 Volumes. London: CRC Press; 2004:475–480.
61
International Journal of Rock Mechanics & Mining Sciences 90 (2016) 53–61
3 Choquet P, Hadjigeorgiou J, Manini P, Mathieu E, Soukatchoff V, Paquette Y.
Analysis of the strata buckling mechanism at the Grand-Baume coal mine, France. Int
J Surf Min – Reclam Environ. 1993;7:29–35.
4 Scoble MJ. Studies of Ground Deformation in British Surface Coal Mines (Ph.D.
thesis), Nottingham: University of Nottingham, Mining Engineering Department;
1981.
5 Yang ZL. Instability behavior for side slope with bedding rock mass. Chin J Geotech
Eng. 2010;32(12):1888.
6 Tommasi P, Verrucci L, Campedel P, Veronese L, Pettinelli E, Ribacchi R. Buckling of
high natural slopes: the case of Lavini di Marco (Trento-Italy). Eng Geol.
2009;109(1):93–108.
7 Goodman RE. Introduction to Rock Mechanics, , New York: Wiley; 1989.
8 Walton G, Coates H. Some footwall failure modes in South Wales opencast workings.
In: Proceedings of the 2nd International Conference on Ground Movements and
Structures. Cardiff; April 1980:435–448.
9 Cavers DS. Simple methods to analyze buckling of rock slopes. Rock Mech.
1981;14(2):87–104.
10 Cavers DS, Baldwin GJ, Hannah T, Singhal RK. Design methods for open pit coal
mine footwalls. In: Proceedings of the International Symposium on Geotechnical
Stability in Surface Mining. Calgary; 6–7 November 1986:79–86.
11 De Renobales TS. Strata buckling in footwall slopes in coal mining. In: Proceedings
of the 6th International Conference on Rock Mechanics. Montreal; 30 August–10
September 1987:527–531.
12 Hu X-Q, Cruden DM. Buckling deformation in the Highwood Pass, Alberta, Canada.
Can Geotech J. 1993;30(2):276–286.
13 Stead D, Eberhardt E. Developments in the analysis of footwall slopes in surface coal
mining. Eng Geol. 1997;46(1):41–61.
14 Dawson RF, Bagnall AS, Barron K. Rock anchor support systems at Smoky River Coal
Limited, Grande Cache, Alberta. CIM Bull. 1995;88(992):60–65.
15 Truesdell C. The rational mechanics of flexible or elastic bodies: 1638–1788. In:
Leonhard Euler. Opera Omnia. Birkhäuser; 1960.
16 Roark RJ, Young WC, Budynas RG. Roark’s Formulas for Stress and Strain, , New
York: McGraw-Hill; 2002.
17 Shanley FR. Inelastic column theory. J Aeronaut Sci. 1947;14:261–267.
18 Engesser F. Ueber die Knickfestigkeit gerader Stäbe. Zeitschr. D Arch - U Ingen Ver
Zu. 1889:455.
19 Hoek E, Carranza-Torres C, Corkum B. Hoek-Brown failure criterion-2002 edition.
In: Proceedings of the North American Rock Mechanics Symposium and Tunneling
Association of Canada Conference. Toronto; 7–10 July 2002:267–273.
20 Bahrani N, Tannant D. Shear strength assessment of a footwall slab using finite
element modeling. CIM Bull. 2009;102(115):48–57.
21 Soukatchoff V, Hantz D, Mathieu E, Paquette Y. Renforcement et contrôle de
parements dans une mine à ciel ouvert de charbon. In: Proceedings of the 7th
International Congress on Rock Mechanics. Aix-la-Chapelle; September 1991:965–
968.
22 Sun GZ. Structure Mechanics of Rock Masses, , Beijing: Science Press; 1988.
23 Marinos P, Marinos V, Hoek E. Geological Strength Index (GSI): a characterization
tool for assessing engineering properties for rock masses. In: Proceedings of the
International Workshop on Rock Mass Classification for Underground Mining.
Pittsburgh: U.S. Department of Health and Human Services; May 2007:87–94.
24 Ylinen A. A method of determining the buckling stress and the required cross-
sectional area for centrally loaded straight columns in elastic and inelastic range.
Mem Assoc Int Ponts Charpentes. 1956;16:529–550.
25 Deere DU, Miller RP. Engineering Classification and Index Properties for Intact
Rocks, Technical Report AFWL-TR-65-116. New Mexico: Air Force Weaponds
Laboratory, Kirtland AFB, and University of Illinois; 1966.
26 Bell FG. Engineering behaviour of rocks. In: Engineering Properties of Soils and
Rocks, Blackwell Science Ltd, editor. Oxford: MPG Books Ltd. Press; 2000:229–232.
27 Goodman RE. Methods of Geological Engineering in Discontinuous Rocks, , St Paul,
MN: West Publishing Co; 1976.