D46 Implementation of EQ-LS-HY coupling model on type landslide in Alps – Géosciences Azur, ACRI
1. Objectives
Calibrate and validate a theoretical model for landslide motion by hydrological triggers (heavy precipitations and repeated seasonal precipitations).
2. WP Protocole
Building a conceptual hydrogeological/hydrological model for mountainous rock slopes
Numerical tests of infiltration coupled hydromechanical effects on failure propagation within rock slopes.
3. Conceptual
hydrogeological model
3.1 The “La Clapière” case study
Taking the La Clapière slope example, 3 nested discontinuous fractured reservoirs characterize slope hydrogeology. Water flows into fractures whose openings depend on the depth and on the gravitational structures of the slope. Gneisses can be considered as impervious. The actual landslide can be taken as a highly permeable fractured reservoir because the displacements induce the formation of large pores inside opened fractures, breccias and blocks. The landslide is drained at its foot by a group of perennial springs (springs 14, 15, 16, 20 on Fig. 1.1) with a total discharge comprised between 0.95 and 2.35 l.s-1. The springs rise at the bottom of a major N010°E trending major fault zone that cuts the middle part of the main landslide. A perennial spring (spring 1) rises at the foot of the north-eastern compartment at elevation 1550 m, and has a discharge comprised between 0.3 and 0.9 l.s-1. After long precipitation periods, some temporary springs (springs T1 to T3) rise around 1650 m elevation along faults or at the bottom of major tension cracks filled with colluvial deposits (spring T4). All the streams that originate from the springs located in the upper part of the landslide are interrupted a few hundred meters downstream. This means that all the waters re-infiltrate in the main landslide. Outside the landslide, the slope can be divided into a decompression toppled zone and a low uncompressed zone in depth. The decompression toppled zone is a highly fractured area where tension cracks create linear drains with estimated permeabilities ranging between 10-2 and 10-3 m.s-1 (permeability estimations are done from springs yield variations analytic interpretations using Goodman ‘s formula, [8]) . Many of the cracks are filled with colluvial deposits which constitute small reservoirs with an interstitial porosity. These tiny reservoirs are interconnected via the tension crack network. It is possible to look in detail at colluvium filled cracks because some of them are cut by the present-day La Clapière main scarp (for location see black rectangle near the spring 10 on Fig. 1.1). Typically, the filling has a 4 to 20 m wide triangular geometry (Fig. 1.2 and Fig. 1.3). It consists in blocks of various sizes and which arrangement defines a rough bedding. The bedding is warped, showing that sedimentation occurred while gravitational movement was active. The deepest part of the filling often consists of very thin deposits of a buried soil that collapsed when the tensile crack was formed. Blocks and sands that can be found in the upper part of the deposits come from the fractured edges of the crack and from glacial deposits that previously covered the slope. The crack extends inside the slope because it is the superficial reactivation of a tectonic fault. Depending on the places, the colluvial fillings can be completely dry or can be drained by a perennial spring (springs 3, 4 and 5 on Fig. 1.1). In the first case, water infiltrating in the colluvial deposits is drained deeper in the slope through the underlying tectonic fault. In the second case, water is trapped in the filling because the basal buried soil is locally impervious. The interconnection of the fillings creates a perched perennial saturated zone that could explain the presence of springs rising in the upper part of the slope between 1650 and 1400 m in the Tinée valley slope (springs 11 and 12 with a discharge ranging between 0.1 and 0.4 l.s-1) and in the Rabuons valley (springs 4, 5, and 6 with a discharge ranging between 0.1 and 0.6 l.s-1). The low uncompressed part of the slope is outcropping on the Rabuons, the Tenibres and the Tinée river banks below the elevation 1400 m. This zone is fractured by major tectonic joints and can be considered as a relatively low-permeability fissured reservoir (10-8 to 10-9 m.s-1 after data from neighbouring tunnels inflows interpretation, [13]). There is a continuity between the joints and the tensile cracks mapped in the uncompressed toppled zone. No springs were mapped coming from this zone.
Fig. 1: 1) Geostructural, hydrological and landslide context - (2) Hydrogeological cross section – (3) Field observation of a tensile crack filled with superficial deposits.
In order to characterize long term coupling between hydrology and stability of the slope, we compared historic Tinée river flood (French Ministry of Agriculture database) to landslide annual velocities (French Ministry of Equipment database) since 1920 (Fig. 2A). The activation of the La Clapière current movement begins around the years 1950-1955. From 1951 to 1987 there is a constant non linear velocity increase up to a 6 m.y-1 peak. After 1987, there is a small decrease of velocities that show annual variations ranging between 4 and 2 m.y-1 values. During the 1920-1999 period, there are 7 Tinée major flood events corresponding to major precipitation events that caused numerous damages to the valley landscape. Clearly, La Clapière movement triggering fit with 1951-1957 major floods. However 1922 and 1926 flood events did not cause any slope destabilization and 1987 velocity peak does not correspond with any major flood event. For the 1987 to actual period, speeds fluctuations roughly fit with annual precipitation fluctuations [7]. At the year scale and for recent years (since 1998), a reconstitution of infiltration yields were performed using hydrogeochemistry of springs waters [5]. There are two main infiltration peaks that correlate with long duration moderate precipitation amounts (for ex. 426 mm/30 days during 3/99 period) or with short duration high precipitation amounts (for ex. 122 mm/2 days during 18-22/10/99 period). For a 0.6 km2 infiltration area, such amounts correspond to precipitation yields respectively ranging between 0.7 and 2.8 l.s-1. Landslide velocities curves show accelerations that range between 0.02 and 0.25 m.day-1 (Fig. 2B) synchronous to the infiltration peaks periods. Accelerations curves have an asymmetric shape with a rapid rise synchronous with the increasing part of the infiltration yield curve (main groundwater flood infiltration) and a slow decrease synchronous to the decreasing and the drying up part of the infiltration yield curve (slope drying up). Duration of acceleration periods is about the same as infiltration periods.
Fig. 2: (A) Long-term comparison between La Clapière landslide velocity and Tinee river major flood events - (B) Correlation between infiltration and velocity variations at the year scale.
3.2 Towards a general theoretical hydrological-hydrogeological model
for mountainous rock slopes
La Clapière slope hydrogeology was compared to the other mountainous alpine slopes of Séchilienne (French Alps, Grenoble) and Rosone (Italian Alps, Torino)]. In all cases, the same hydrological model can be built with the following characteristics:
The upper part of the slope is highly altered and it is cut by pluri-hectometric tensile cracks (sacking part of the slope). A perched aquifer is nested in this part of the slope. Tensile crack are filled with local material of the slope and they constitute interconnected high permeable drains. Between the tensile cracks, altered blocks of gneiss constitute less permeable porous zones with a high storativity.
If there is a landslide at the foot of the slope, waters from the perched aquifer are drained through the landslide which is made of highly fractured permeable rocks. The infiltration water flow induces acceleration of the landslide. If there is no landslide, perched aquifer drainage takes place in the upper parts of brooks that cut the slope. Then, water flow to the river which is located in the main valley.
The remaining of the slope (deeper parts) is made of low permeable fractured rocks that can be taken as “impervious” compared to the rocks from the two previous zones.
Coupling between pore waters and slope failure propagation is not simple. It appears clearly that heavy precipitation events do not correlate with landslide triggering in the case of rock slopes like La Clapière. This can be due to the high permeability and the high storativity of the rocks under a shallow stress loading. Two cases can be defined depending on the alteration of the slope:
In low fractured zones with small pore sizes poorly interconnected, low permeabilities of rocks induce large pressure variations that cause large deformations. Because water is only located within voids in some fractures, there is no effect on the density variation of the rock mass. This case was first tested considering that it is the initial hydro-mechanical state of the slope.
In very fractured shallow zones where pores sizes exceed the centimetric scale and are well interconnected, the high permeability of the rocks induces low interstitial pressure variations that only cause low deformations. The high porosity and the capability of pores to be easily saturated with water cause a sensitive rock density change and an important loading that induce rock mass deformation. This case was secondly tested considering this is a more advanced case, compared to the previous one, that describes a more weathered rock slope.
4. Numerical models
4.1 First case: Modeling Pressure/deformation coupling effect within
a low weathered rock slope
We performed two parametric simulations with UDEC code and taking the example of a conceptual slope which is of the dimensions of the la Clapière slope. In the first test, only pre-existing fractures were taken into account (Fig. 3A). In the second test, a 28° dipping failure surface was set at the foot of the slope (Fig. 3B). UDEC code allows large finite displacements/deformations of a fractured rock mass under pressure loading. We considered a vertical cross-section extending from the slope crest (2600 m a.s.l) to the valley (1100 m a.s.l). We considered 9 discrete penetrative vertical fractures. So as to hydraulically connect faults between them and to approximate foliation planes geometry, horizontal joints were included in the model. This cross-section is constrained by no bottom vertical and no lateral displacements boundary conditions and, by impervious hydraulic boundary conditions. Rock matrix mechanical behavior is taken as linearly elastic and isotropic. Faults are assumed to behave according to an elasto-plastic law with the Mohr-Coulomb failure criterion (Tab. 1). Fault and matrix mechanical and hydraulical parameters are deduced from laboratory and field measurements.
|
Rock matrix |
Gneiss |
Alluvium |
|
K (Pa) |
5.3 e10 |
2.9 e9 |
|
G (Pa) |
2.5 e10 |
1.3 e9 |
|
d (kg.m-3) |
2200 |
1500 |
|
Joints |
Vertical faults |
Horizontal joint |
|
Jkn (Pa.m-1) |
1.8e10 |
1.8e10 |
|
Jks = Jkn/10
(Pa.m-1) |
1.8e9 |
1.8e9 |
|
Jfric (°) |
30 |
30 |
|
Jperm (Pa-1.s-1) |
83.3 |
83.3 |
|
azero/
ares (m) |
1e-3/1e-4 |
1e-3/1e-4 |
K: Bulk modulus – G: Shear
modulus – d: Matrix density – Jkn: Fracture normal stiffness – Jks: Fracture shear stiffness – Jfric: Fracture friction angle – Jperm: Fracture permeability factor – azero/ares: Range
of variation of fracture aperture.
Table 1: Hydromechanical properties of rock matrix and joints.
Perched saturated zone is simulated affecting a local zero permeability at faults segments corresponding to the basal boundary of this zone (dashed line between 1500 m and 2000 m elevation - Fig. 3). Flow in the faults was set to be compressible. Gravity acceleration was applied. We performed a static hydromechanical calculation with steady-state flow. The cross-section is first consolidated to gravity until stress and displacements are numerically stabilized. Then, initial groundwater conditions were simulated in the basal saturated zone. No interstitial pressure was set in the perched saturated zone. A 0.75 l.s-1 effective infiltration is simulated in the slope at 1900 m elevation (Fig. 3). On the cross-section, we plot maximum displacements induced by the hydraulic loading (Fig. 3).
Fig. 3: Results of coupled hydromechanical modeling.
In the initial slope case (Fig. 3A), pressures increase from 0 to 1 MPa in the perched aquifer. In the basal aquifer, a 0.5 MPa piezometric bump extends from 300 to 1500 m along the x-axis. The maximum calculated values of displacement vector are located between the foot of the slope at 1100 m and the middle part of the slope at 1900 m. This strain zone extends from 50 m to 400 m inside the slope. Displacements values vary between 0.1 m and 1.3 m in that zone. Water pressures are situated in two distinct zones which hydraulically communicate with each other: a basal 500 m thick saturated zone with interstitial pressures ranging between 0 MPa and 5 MPa, and a perched 200 m thick saturated zone with interstitial pressure ranging between 0 MPa and 2 MPa. In the middle part of the slope there is swelling with vectors’ dip towards the top linked with mechanical opening of fractures under pressure elevation in the perched saturated zone. In the upper part of the slope there is a lowering (sackung) with vertical vectors’dip.
Same results are observed in the actual slope case (Fig. 3B). Pressures values are less important and the saturated zones are less extended inside the slope. This means that the slope is better drained after the failure has occured. As a matter of fact, displacement values are less important. Pressures along the failure plane range between 0 MPa at 1500 m and 0.05 MPa at 1150 m. Displacements maximum values of 0.4 to 0.6 m concentrate along the failure plane where vectors are parallel to the plane. However, a 0.1m displacement zone extends further in the stable part of the slope in relation with the basal piezometric bump hydromechanical effect. All the remaining of the slope is also affected by displacements values ranging between 0 and 0.05 m. These upper displacements belong to a deep displacement field generated by the slope foot sliding (mass loss).
4.2 Second case: Modeling
water induced density change effects on failure propagation within a highly
weathered rock slope.
In that second set of numerical tests, same slope topography was considered.
The slope is cut by nine penetrative faults and a highly fractured zone was added in the upper part of the slope in order to model the weathered zone. In that zone, the slope is divided into 10m x 10m diamond shaped blocks. Same mechanical parameters as in the 4.1 case are given to joints and blocks.
Main difference in the 4.2 case relies on the fact that only a water induced density change of the rock is considered. No flow calculation is performed. The cross-section is first consolidated to gravity until stress and displacements are numerically stabilized. Then, the density of a 50m thick layer is changed at the basal part of the weathered zone in order to set initial conditions for a perched saturated zone.
The saturated rocks density of this layer is calculated as follows:
where:
is the density of dry
rock (2200 kg m-3)
n is the rock porosity (0,5)
s is the saturation index (equal to 1)
is the density of
water (1000 kg m-3)
Model is then run until stabilization. Third stage consists in the model loading. A 5m increase of this layer thickness was modelled in order to simulate seasonal water recharge of the perched aquifer.
Joints segments with no stress (which are parts of joints were failure has occurred) are plotted in red on the major principal stress contour map of the slope (Fig. 4). First, it appears that failure is spreading everywhere within the weathered zone. Second, there is a concentration of segments in failure mainly in two parts of the slope:
along the boundary between the weathered and the intact zones;
along the upper part of the penetrative fault above the 2550 y-value.
Principal stresses depend mainly on those two features. It can clearly be seen that the weathered zone basal boundary is a low major principal stress boundary. This low stress boundary is bending towards the foot of the slope within the slope compartment located between the topographic surface and the fault. Close to the intersection between the fault and the basal boundary of the weathered zone (coordinates of that zone are x = 3500 and y =2425 on Fig. 4), there is a stress concentration that is just below the zone where the fault segments have not failed.
8-9 2-3 4-5 6-7 12-13 14-15 28-29
Fig. 4: Major principal stress contour map induced by the density loading of the rock slope. Stresses values are in Pa. Coordinates on the cross section are in meters.
Depending on the points on
the cross section, z-displacements range between -0,4 and -1,4 10-2
m and x-displacements range between -0,5 and -5 10-2 m (
Fig. 5 and Fig. 6). Displacement vectors are directed towards the left side of the slope with a dip angle of 25° for points located below the 2500 y-value and with a dip angle of 40-45° for points located above 2500 y-value. This is explained by a block movement guided by the fault and the basal part of the weathered zone.
Fig. 5: Horizontal displacements variations (m) versus time (s) at points located on Fig. 4.
Fig. 6: Vertical displacements variations (m) versus time (s) at points located on Fig. 4.
5. Conclusion
Hydro-mechanical effects of water flow within fractures are predominant effects that induce strength decrease of rock slopes and progressive failure propagation.
We show in this study that hydrostatic pressures can be generated and concentrated in tensile features of the upper part of the slope with a moderate infiltration yield (mean inter-annual value for example). This can cause sufficient hydrostatic pressures elevation in the cracks to increase slope deformation. At this stage, coupling effects take place within the fractures. A pressure increase induces a fracture deformation and a fracture deformation induces a modification of flow paths and thus a pressure modification. Tilting at the slope surface and progressive failure that propagate deep in the slope generate a thick weathered layer (up to several hundreds meters thick) from the theoretical bottom of the perched aquifer down to the slope foot. At La Clapière such a failure through tilt could have worked for at least 3000 years until 1987 when it is taken that a general failure surface was created.
When the strength of the weathered layer is low enough, water induced effects change. Because rock porosity is very high and permeability became very high, interstitial pressure variations are low (a few meters). To the contrary, the saturated volume of rock is very high and seasonal precipitation recharge of the perched aquifer induces large density variations. Numerical tests clearly show that the state of principal stresses is very low within the weathered zone which all is in tensile stress. Pre-existing penetrative faults still play a role as mechanical heterogeneities within the weathered zone. In that case, even a moderate recharge induce failure propagation that concentrate along pre-existing surfaces like the boundary between the weathered and the intact zones and the penetrative major faults. Model shows that rock remains unfailed only in sparse volumes that are located at the intersections between major discontinuous surfaces.
Publication:
GUGLIELMI Y., CAPPA F. and BINET S. – 2005 – Coupling between hydrogeology and deformation of mountainous rock slopes : insight from La Clapière area (Southern Alps, France), C.R.Geosciences, in press.