English
Italiano
Forgot your password?
New user?
Document Actions
ijege-13_bs-tecca-et-alii.pdf
wide. The relevance of this process as a natural hazard
stresses the need for a more complete knowledge of
both the triggering mechanism and the prediction of
life span of rockslide dams.
the discontinuities (roughness, wall strength and per-
sistence) and on the weathering action on the intact
rock and discontinuities.
and bedding planes. In general, because the presence
or absence of discontinuities has a great influence on
the stability of rock slopes, their behavior plays a criti-
cal role in a stability evaluation.
reaching implications for the safe development of both
inhabited alpine valleys and engineering projects.
to model the evolution of a rock slope tought to have
failed in response to a seismic shaking. The numeri-
cal simulation is aimed, then, to explain whether the
their development, so that consequences and possible
prevention and mitigation actions can be envisaged.
kinematics of a rock slide/avalanche in the north-east-
ern Italian Alps. The “La Marogna” rock avalanche, in
the Vicenza Province (Venetian Pre-Alps, North-East-
ern Italy), with a volume of about 17
morphological investigations highlight that the whole
rock avalanche mass is formed by two distinct overlap-
ping bodies and that apparent poor stability conditions
characterize the slope above the present main scarp.
neering geological model has been built and analyses of
the triggering conditions have been performed using the
bi-dimensional continuum (FLAC) and discontinuum
(UDEC) codes UDEC on the re-constructed original
slope profile. Different situations have been simulated
for gaining a better understanding of the effect of static
and dynamic loads on the modeled rock slope.
crease results in the instability of a rock mass limited
at its bottom by both bedding and a pre-existing dis-
continuity.
mass movement.
ley, with a local NW-SE trend. The slope is composed of
the Mesozoic formation of “Dolomia Principale” char-
acterized by well-bedded carbonates. This slope belongs
to the northern limb of an anticline fold so that the bed-
ding dip, NNW trending, increases progressively from
20°-25° at the slope top to about 50° at the bottom (Dal
Prà, 1992). The fold limb is affected in the higher part of
the slope by a normal fault, dipping 70° ENE, by vertical
persistent tectonic lineaments and by some trenches not
directly correlated to the instability phenomenon (Z
m long and 160-180 m high.
scarp, where the beds dip 30°-35° as a mean (Fig. 2).
possible to note the presence of a NNW dipping thrust
(“b” in figure 2) with a “stair case” geometry: steep
reverse fault tracts in stiff layers connect flat tracts
sub-parallel to the bedding.
the upper thicker layers of the “Dolomia Principale”,
avalanche (“La Marogna” rock avalanche), was trig-
gered by the earthquake occurred on 1117.01.03 (I0
IX MCS, M 7.0) (M
glaciation most-unstable situation.
has been analyzed both in static and dynamic condi-
tions to investigate the trigger mechanism and provid-
ing the possibility to evaluate the stability conditions
of the present slope. The geological and structural set-
ting of the area has been investigated by Z
ing strength parameters.
discontinuity-controlled failures. Discontinuum
modeling approach explicitly simulates the geo-
logical structure treating the problem domain as an
assemblage of independent units, corresponding to
blocks formed by the intersection of joints (l
based methods is that contacts between blocks are
continuously changing with the deformation process.
A more realistic response can be, then, modeled and
the specific failure mechanism, controlled by pre-
defined discontinuities, may be captured .
to rock slope analysis.
values of the properties of rocks and discontinui-
ties. Uniaxial compressive tests were conducted in
laboratory, while joint parameters JRC0 and JCS0
were estimated by field Barton roughness profiles
and Schmidt hammer tests in 8 and 140 kPa respec-
tively. Values of the rock mass strength and defor-
mation parameters were obtained using the RocLab
program (H
shown in Tab. 1.
planes are prevailingly undulating and less rough. The
residual friction angle (φ
discontinuities have been back-calculated from data on
deformability of both intact rock and rock mass (i
been derived using the corrections proposed by b
moduli were calculated on the basis of rock mass defor-
mation modulus and a Poisson coefficient of 0.23.
ible just down-hill (Fig. 2) (Z
indicated with “c” in Fig. 2, is present. Three main sub-
vertical joint systems cross the rock mass, trending ap-
proximately N310°-320°, N230° and N260°-270°.
from the base of the main scarp to the valley bottom. It
is covered by a fan-shaped body rising up the opposite
slope for about 40 m, due to a later process with the
features of a rock/debris flow. The deposits dammed the
Astico River, and the resulting lake should have drained
in a very short time, since no lacustrine sediments have
been found upstream. The deposit is formed by sands,
gravels and pebbles with blocks from 1 to several tens
m
sistent with the 1117.01.03 earthquake (b
5.5x10
collapsed rock volume is about 13
compute the interaction between different materials,
site geometry and wave propagation in case of seismic
inputs (b
continuous material is replaced by a hypothetical con-
tinuous material using a homogenization technique.
When coupled with a discontinuum modeling, two-
dimensional continuum modeling may be used to pre-
liminarily examine stress distribution and evolution,
possible phases of stress-induced progressive failure,
and plastic yielding within the rock mass. As such, a
preliminary set of 2-D models were run using the con-
are listed in Tables 2, 3 and 4.
original slope has been re-drawn on the basis of the
surrounding morphology and the calculated land-
slide volume; 2) the stresses have been initialized
considering the existence of a 800 m high glacier
progressively reducing its height; 3) in continuum
modeling, ubiquitous-joint and strain-hardening/
softening ubiquitous-joint models have been consid-
ered; 4) in discontinuum modeling, Mohr-Coulomb,
Coulomb slip and continuously yielding models
have been considered; 5) seismic loading has been
with a seismic event, continuum analysis have been
carried out in dynamic conditions.
estimated in VIII-IX MCS, that is a magnitude of
M=6-7, and the corresponding PGA (Peak Ground
Acceleration) considered equal at least to 0.32g with
frequencies from 2 to 5 Hz. However, the Italian Risk
Map (INGV, 2007) shows, in the same area and for
an excedence probability of 1% in 50 years, a PGA
of 0.25g. The analyses have been, then, performed
considering two values of the maximum acceleration
amax : 0.25g and 0.32g. Using existing empirical cor-
relations, the duration of the seismic loading has been
evaluated in a minimum of 8 seconds.
to propagate upwards and, for simplicity, isotropic
conditions have been assumed. The input variables
for dynamic analysis have been calculated from the
expressions (b
for viscous boundaries. The sinusoidal shear wave has
been applied for different periods (8, 12, 24 s) repre-
senting cycles of motion from 16 to 120. Values of
input seismic parameters are summarized in Tab. 5.
uum model) or a sinusoidal velocity wave (discon-
tinuum model), whose frequencies and intensities
have been selected on the basis of the Italian Seismic
Risk Maps (INGV, 2007).
continuum numerical methods have been preferred to
preliminarily simulate the mass structure behavior sub-
jected to quasi-static or dynamic loading.
pendent failure mechanisms. The model incorporated
837 quadrilateral elements and the grid has been drawn
to fit the existing anticlinalic structure. An elasto-plas-
tic constitutive criterion has been assigned to the slope
materials assuming a Mohr-Coulomb yield criterion;
stresses have been initialized assuming a gravity load-
ing and an elastic homogeneous isotropic rock mass.
Before the removal of the glacier, an ubiquitous-joint
model and a strain-hardening/softening ubiquitous-
joint model have been assigned respectively to the base
rock mass and to the rock collapsing mass, to take ac-
count of both material anisotropy and the continuously
changing dip of bedding (Fig. 3).
radation of rock mass strength with stress variation and
time has been simulated by gradually decreasing the
values of rock properties up to reaching the calculated
properties values of weathered rock mass (Tab. 2).
steeper parts of the reconstructed slope (Fig. 3). In
case of the presence of a groundwater with a maxi-
mum height of 10 m over the indicated shear sur-
face, the rock slope is still stable but, at the base and
at the top of the potentially collapsing mass, finite
slips along the ubiquitous joints develop (Fig. 4). In
conclusion, the examination of the 2-D finite-element
results in static conditions shows that a condition of
instability cannot be reached, even considering the
presence of a perched water table in the potentially
collapsing mass.
maximum horizontal displacements, greater than 6.0
m in the steeper part of the slope, propagate towards
the higher part of the slope, being maximum at the
toe and approximately null at a distance of 550 m;
ii) slips along ubiquitous joints are distributed in all
the collapsing mass; iii) the upper end of the sliding
mass is limited by zones at failure in tension. The in-
stability condition is shown by the horizontal veloci-
ties progressively increasing with time, monitored at
selected points on the shear surface (Fig. 6).
curred in 1117 A.D. might have triggered the "La
Marogna" rock avalanche but, with continuum mod-
eling, the shape of the collapsed mass does not com-
pletely reflect the present morphology of the slope: the
slip surface follows the local maximum rock strata dip,
but the simulated sliding mass extends far beyond the
location of the present landslide scarp. In any case, the
examined rock slope seems to be more sensitive to the
seismic frequency than to the seismic acceleration, at
least considering a dynamic time of 8 seconds.
analyses, in both static and dynamic conditions, were
carried out using the two-dimensional distinct element
code UDEC (Z
ing model for bedding and tectonic discontinuities.
brated under the gravity force with the obtained rock
parameters values (Tab. 3); ii) the load due to ice sheets
has been applied to the model and later removed in suc-
cessive stages; iii) static and dynamic analyses have been
performed with rock and discontinuities properties and
models shown in Tables 3 and 4 respectively.
ity (“b” in Fig. 2) and the shorter one (“c” in Fig. 2):
computed horizontal displacement rate induced by ice
sheets removal is null. The results diverge only as re-
gards the distribution of the calculated displacements:
in the former case they develop over the persistent tec-
tonic discontinuity up to the top of the modeled slope,
while in the last case, they are limited upslope in cor-
respondence of the present location of the landslide
crown (Fig. 8), representing in a better way the future
collapsing mass.
amplitude of 0.08, 0.1, 0.13, 0.16 m/s (acceleration of
0.25 g and 0.32 g) at 3 and 5 Hz for 24 and 40 cycles
(duration of dynamic input of 8 seconds). Frequency,
amplitude and accelerations values have been selected
as those used in the continuum modeling dynamic
analysis. Free-field boundaries are invoked along
the left and right boundaries to absorb energy and no
displacement is allowed in the x-direction along the
lateral sides of the model. The Rayleigh damping ra-
tio, which reproduces the energy losses in the natural
system when subjected to dynamic loading, has been
assumed to equal 2%.
seismic parameters for the dynamic analysis are dis-
played in Tab. 5.
onds after the end of the dynamic input (test 2), shows
that the slope is entirely unstable. However, the col-
lapsing mass does not match the field observation:
significant displacements involve also the upper part of
the slope, beyond the present location of the landslide
crown (Fig. 9).
bedding planes and cross fractures, are represented
in the model.
In dynamic conditions, in fact, the mesh size is con-
trolled by the shortest wavelength of input fluctuation
(i
of seismic wave λ. The highest frequency of the input
wave f
that the maximum input frequency of considered dy-
namic loads is 5 Hz.
gular elements. The base is assumed to be flexible but
the boundary is fixed in the y-direction. Figure 7 dis-
plays the model of the jointed slope and the ice sheets.
presence of a persistent tectonic discontinuity (“b” in
Fig. 2) and a shorter and steeper discontinuity (“c” in
Fig. 2); iii) material properties deriving from the con-
tinuum modeling analyses; iv) glaciation and deglacia-
tion processes have been modeled after the application
of the rock mass properties values (Tab. 4, Static analy-
sis), in order to consider the modification of stresses
in the transition from glacial to non-glacial conditions
(I
vertical joints and Coulomb Slip model for the tectonic
discontinuities; vi) in dynamic analysis, Coulomb Slip
of normal load and frictional resistance activation on
the sliding surface and at the blocks contacts.
based on geological observations, field data, and labora-
tory tests. In order to analyze the influence of geological
and seismic factors on slope failure, numerical simula-
tions were performed using both continuum (FLAC)
and discontinuum (UDEC) bi-dimensional approaches.
isting anticlinalic fold and, apparently, partly with a long
thrust cutting the upper portion of the same anticline.
chanical characteristics on slope stability was made un-
der static conditions. Both continuum and discontinuum
modeling revealed the stability of the slope with the
strength parameters values obtained by field and labora-
tory tests and realistic water pressures on sliding surface.
uphill extension of the collapsed rock mass may be
simulated only considering the existing secondary
tectonic discontinuity, shorter and more tilted than the
more evident and long one. The discontinuum approach
simulates effectively this discontinuity, showing that
the slope instability but, in this case, the collapsing mass
matches rather well the present slope geometry. The dis-
placement dynamics, highlighted also by the amplifica-
tion of blocks deformation, shows that the collapsing
mass is split in two main parts as indicated by the sub-
vertical joint progressive aperture, the most upward part
being characterized by a delayed dynamics.
(monitored points PT 2 and PT 3, Fig. 11). It is also
worthwhile to note the pulsating trend of the horizontal
circumstance is confirmed by the different values and
trend of the velocities registered at the surface ground
monitored points (Fig. 11).
as a rock avalanche and deteriorated the rear part that,
however, remained on site. Only later this rock mass
was destabilized probably by an increase of pore pres-
sure due to heavy rain, so accounting for the fan shape
of its deposit. This hypothesis seems to be validated
also by the observation that the first phenomenon is
smaller than the later one and this results also by the
numerical simulation.
the different nature of the two parts constituting the
sliding surface: the tectonic discontinuity in the upper
part and the bedding planes in the lower one; iii) the
non uniform distribution of the shear resistance on the
sliding surface, due to the different characteristics of
these two parts and to the presence of rock bridges at
the transition between them; iv) the presence of a real-
istic groundwater level in the rock mass.
represents the major risk factor for the valley bottom
villages: type of earthquake source and site conditions
need to be, then, better understood both from field evi-
dence and modeling.
scarp and, so, stressing the necessity for accurate and
detailed structural geology surveys in order to repro-
duce the real landslide formation mechanism.
that the shear resistance of the controlling slip surface
may be exceeded by the shaking-induced inertial forces
due to, at least, a medium intensity earthquake, that is
lower than that (M=7.0) of the 1117 A.D. earthquake.
The duration of the seismic loading seems to be not so
relevant as collapse may be modelled with a seismic
shaking duration of only 8 s. In general terms, the slope
collapse occurs for source frequencies in the range of
those that may be attributed to this earthquake and large
continuous post-seismic displacements develop with
increasing velocities. Maximum horizontal velocities at
the surface of the landslide body reach 0.6 m /s after 60
s from the end of the seismic loading (Fig.10) increas-
ing, then, abruptly so that “La Marogna” rockslide may
be categorized as a high-speed landslide.
and input frequency is included between 0.13 and
0.26, that is generally considered critical for seismic
amplification (d
be considered more reliable, but it requires more accu-
rate and detailed field surveys that not always can be
easily and completely obtained. Continuum approach
is, on the other hand, useful in order to preliminar-
ily simulate the mass structure behavior subjected to
quasi-static or dynamic loading and analyze the cor-
responding stresses distribution.
is the main object of coming researches. As a matter
of facts, the field observation indicates the presence of
two different landslide deposits attributable to differ-
ent processes, while the numerical simulation shows a
unique sliding phenomenon. It should be highlighted,
nevertheless, that the sequence of displacements fol-
lowing the seismic loading (Fig. 10) displays a rock
mass that, before the general collapse (Fig. 10 e) is bro-
ken up in two parts with different dynamics: the small-
er front part, where the reconstructed slope is steeper,
moves earlier, while the larger rear one moves more
ITASCA. UDEC (2011) - Universal distinct element code, version 5. Minneapolis: Itasca Consulting Group, Inc.
i
The issue 16 volume 01 has been published