Réf. Gunzburger & al. 2005 - A

Référence bibliographique complète

GUNZBURGER, Y., MERRIEN-SOUKATCHOFF, V., GUGLIELMI, Y. 2005. Influence of daily surface temperature fluctuations on rock slope stability: case study of the Rochers de Valabres slope (France). International Journal of Rock Mechanics & Mining Sciences, Vol. 42, 331–349.

Abstract: The present paper describes a rockfall that has affected the Rochers de Valabres slope (France's Southern Alps region) and discusses one possible mechanism for the occurrence of this rockfall, along with the potential for future ones. In the absence of an obvious explanatory trigger factor, we set out to examine whether natural daily surface temperature changes could have played a role in this event. In particular, it is suspected that these slight, yet repeated, perturbations may be a preparatory factor for rockfalls, with a day-to-day cumulative effect. A numerical model strengthens this hypothesis by showing that thermally induced deformations can be sufficient to cause the gradual downward creep of a rock block located in an awkward position. To investigate this notion more thoroughly, a currently vulnerable part of the Rochers de Valabres slope has been instrumented with a high-precision geodetic monitoring system (total station). It is believed that this device is able to capture thermally induced movements if specific precautions are taken. The instrumented rock volume is used as a test site in the aim of better understanding the consequences of surface temperature changes on slope stability. Measured data (with a precision level never before achieved on rock slopes) are compared herein with numerical modelling results. The initial conclusions of the long-term study we are conducting indicate that surface temperature changes play an important, while not easily quantifiable, role in preparing rockfalls.

Thermomechanics - Slope stability - Rockfalls - Geodetic monitoring - Numerical modelling

Organismes / Contact

• LAEGO-INERIS-INPL, Ecole des Mines de Nancy, Parc de Saurupt, 54042 Nancy, Cédex, France - E-mail address: yann.gunzburger@mines.inpl-nancy.fr
• Géosciences Azur, CNRS-UNSA, 250 rue Albert Einstein, 06560 Valbonne, France

This work has been performed thanks to the financial support offered by the INERIS National Institute for Industrial Environment and Risks, within the scope of a study conducted for the Ministry of Ecology and Sustainable Development (the French Ministry responsible for legislation related to environmental issues).

(1) - Paramètre(s) atmosphérique(s) modifié(s)
(2) - Elément(s) du milieu impacté(s)
(3) - Type(s) d'aléa impacté(s)
(3) - Sous-type(s) d'aléa

Pays / Zone
Massif / Secteur
Site(s) d'étude
Période(s) d'observation
French Southern Alps Tinée Valley (Alpes-Maritimes) Rochers de Valabres      

(1) - Modifications des paramètres atmosphériques








Informations complémentaires (données utilisées, méthode, scénarios, etc.)


(2) - Effets du changement climatique sur le milieu naturel








Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)



(3) - Effets du changement climatique sur l'aléa




On the basis of the Rochers de Valabres test site, the authors have begun to test a methodology for investigating the causes of rock slope movements and particularly of rockfalls. Their approach herein is based on a mechanical decomposition of these causes into predisposition factors, preparatory factors and trigger factors as well as on arguments pertaining to both computation and in situ measurements. In this paper, they have focused on the role of the fracture network and surface temperature fluctuations. The main results can be summarised as follows:

• The limit-equilibrium calculation (Section 3.2) demonstrates that the geometrical configuration of the primary fractures with regard to topography can be considered as a predisposition factor; since it is not sufficient for explaining the occurrence of rockfalls, the potential contribution of surface temperature changes were examined.

• A simple 2D model (the ‘‘triangular-block model’’, see Section 2.3) confirms that daily surface temperature changes are able to induce daily plastic deformations of the joints and thus contribute to the preparation of rockfalls. The results from modelling computations are in good accordance with those derived from analytical calculations (temperature profile and normal surface displacement) and can be used as a reference for further investigations.  The numerical calculation whose geometry lies in better agreement with reality (Section 4.1) helps to analyse the movements induced by temperature changes and a study of such calculations proves relevant. In particular, it is shown that daily temperature fluctuations may be responsible for generating irreversible displacements on some fractures.

• The TS technique (Section 4.2) has proven capable of working under natural conditions. Even though the level of confidence we can attribute to its results is still questionable, this technique does represent the first time that such precise monitoring has been achieved. Field monitoring contributions may be noticeably improved with the help of long-term measurements and ground-based measurement devices; in particular, surface tiltmeters have been recently installed on the site.

• In situ measurements are in accordance, in terms of order of magnitude, with numerical modelling results.

• The measurements, as well as the computation, are not sufficient to discern any permanent displacement increase from day to day. For this reason, it will be necessary to examine the influence of seasonal temperature changes by means of measurements and computations.

• Improvements in both measurements and the computational model are possible, with the latter always being easier to achieve: (i) improved thermal boundary conditions may be tested and, (ii) even though the use of 2D modelling has been justified [in Section 4.1.1], greater precision could be obtained by means of a 3D computation.

The limited magnitude of preparatory factors and their cumulative nature represent an intrinsic difficulty with respect to the characterisation of related phenomena. Moreover, in practice, defining the amount of a given preparatory factor necessary for inducing a rockfall is almost impossible. For this reason, the monitoring of preparatory factors in order to predict slope collapse is not sufficient. On the other hand, using trigger factors does not yield any better results since such factors sometimes arise suddenly and just prior to the rockfall, so that a very long period of monitoring is needed before their detection.

One solution may stem from other techniques that do not directly concern rockfall causes, but rather associated phenomena, such as the initiation, propagation or reactivation of cracks within the rock mass. These may be recorded by means of acoustic emission devices. The microseismic method has already been successfully tested to detect cracks in deep coal mines [50], abandoned mines [51], hot dry rock geothermal projects [52] and, more recently, for predicting coastal chalk-cliff collapse [53]. Such a device has been installed on the test site slope; initial results are encouraging, yet still need to be confirmed (see [54] for a description of both the device and these initial results).





Paramètre de l'aléa
Sensibilité des paramètres de l'aléa à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)




(4) - Remarques générales


France’s Southern Alps region is especially prone to slope movement hazards, ranging in magnitude from light rockfalls to extensive gravitational deformations involving several tens of millions of cubic metres of rock [1], such as that at La Clapière. This high rate of slope activity is currently explained by various factors, including sharp topographic contrasts, intensive previous tectonic deformations leading to a high degree of fracturing, seismic activity, and a precipitation profile with rainfall concentration over time (see [2] for an instructive qualitative review).

The Tinée Valley is an economically vital means of communication in this region (Fig. 1); it links the French Riviera with the Mercantour National Park, the large ski resorts Auron and Isola 2000 and the Ubaye Valley through the BonnetteRestefond pass. Due to quaternary glacial erosion, the upper valley is relatively wide, with abundant alluvial fill. In contrast, between the villages of Isola and Saint-Sauveur-sur-Tinée, the Tinée River flows through a deep defile known as the Gorges de Valabres, cut into hard gneissic rock, with steep slopes of several hundred metres on both banks.

In May 2000, the river’s right bank, known as the Rochers de Valabres slope, underwent a large plane failure. The ensuing rockfall, involving about 2000 m3 of material, seriously damaged one pile of a bridge over the Tinée River. This incident caused total isolation of the upper valley for a few weeks due to road traffic interruption, along with major economic and social impacts.

This rockfall did not coincide with any particular rainfall event, seismic activity, freeze-thaw period nor with any of the immediate causes currently cited in the literature. It is therefore difficult to pinpoint an easily intelligible explanation for this event. For this reason, the authors explored another potential rockfall cause: a repetition of smaller events (such as the gradual weathering of fractures), whereby each individual event on its own would almost be imperceptible and insufficient to induce rupture, but when accumulated over time could prove capable of acting as the cause. This paper discusses in greater detail the possible implication of surface temperature changes in rockfalls; while such relationships have been evoked at times in the literature, they have heretofore almost never been quantitatively studied or measured. The primary aim of this research herein is to shed light on the contribution of surface temperature fluctuations to preparing rockfalls, in addition to presenting plausible means to tackle with it.

(2.) Surface temperature changes: a preparatory factor for rockfalls?

(2.1.) Predisposition, preparatory and trigger factors

The distinction between immediate causes (like a strong earthquake) and less-immediate ones (like gradual weathering) stems from the fact that the former acts directly, whereas the latter exerts a slow cumulative effect and therefore requires a long period of time to become effective and ultimately induce a major consequence. This difference may easily be seen in the landslide-sequencing model proposed by Finlayson and Statham [3] and later adapted by Julian and Anthony [2]. In this schematic model (see Fig. 2), progressive failure of a slope (and subsequent rockfall) occurs when the disturbing forces acting on the rock mass have exceeded its resistance. Points 1–5 correspond to such instances where disturbing forces and resistance are equal. From a physical standpoint, various stages may be identified as follows:

• Over the long term (between approximately 104 and 106 years), resistance gradually decreases due to geochemical rock weathering or progressive damage processes, e.g. until the failure threshold has been reached (Point 1).
• Over the medium to long term (typically 102–104 years), the rise in slope angle resulting from erosion may induce slow stress increases until failure (Point 3).
• Short-term strength and stress oscillations (1–102-year period), e.g. due to seasonal changes, are then superimposed upon these effects and can lead to failure (Point 2).
• Lastly, brief and violent phenomena, such as a heavy rainfall event or a strong earthquake, may be directly responsible for either lowering the level of resistance (Point 4) or increasing the magnitude of disturbing forces (Point 5) and lead to failure.

The causes of medium- to long-term changes in resistance or disturbing forces are generally known as preparatory factors: in order to be efficient, their small (and almost imperceptible) effects must be cumulated up until rupture. Causes related to changes over the short-to-medium term are often referred to as triggers (or trigger factors) and constitute the most direct causes of failure. This distinction must not obviously be construed as a dichotomy, due to the existence of a continuous transition between preparatory and trigger factors. The key focus inherent in this model is to essentially convey the notion that rockfalls result from numerous, complex and interacting causes that may act over widely varying time frames; hence, it should never be considered that rockfalls must have been generated by just the most recent and apparent changes.

Moreover, preparatory and trigger factors do not act identically on all slopes, given that initial conditions are not the same. Some slopes are in fact more conducive than others to rockfall activity, due to background factors such as topography (height and steepness of valley walls), vegetation, lithological parameters, and fracture geometry and density. These elements will be referred to as predisposition factors. Unlike preparatory and trigger factors that are defined by their action time frame, predisposition factors do not exhibit any evolution over time; instead, they serve to define the general framework of the slope that incites, to varying degrees, the onset of instabilities.

The most frequently documented factors have been summarised in Table 1. Distinction has been made by their physical nature: mechanical, hydrological, thermal or geochemical phenomena. In this paper, the authors focus on determining whether surface temperature oscillations may serve as a possible preparatory factor for rockfalls. In order to investigate the effects of such a phenomenon, this research was shaped around comparing two means by which these actions can be elucidated: numerical modelling and high-precision geodetic monitoring. These techniques have both been applied to the Rochers de Valabres slope, which is submitted to severe temperature changes due to its southeastward orientation and the dark colour of its component rocks. Moreover, as explained further below, the site’s geometry is favourable to rockfall initiation under repeated thermal loadings. Before proceeding with the site description, let us provide a brief overview of the role of thermal and thermomechanical (TM) phenomena in the field of engineering geology and demonstrate how a simple numerical computation can already lend an idea of the magnitude of TM effects on rock slopes.


Predisposition factors

Preparatory factors

Trigger factors


Steep-sided valley

Well-developed fracture network

Neotectonic stresses

Rise in slope steepness due to valley incision

Regular seismic activity

Damage process

High-magnitude earthquake

Freezing and thawing of water in fractures

Hydrological and meteorological

Climate with a high precipitation rate

Regular rainfall regime

Heavy rainfall episode
Rapid snowmelt


Climate with sharp temperature contrasts

Daily a seasonal surface temperature oscillations



Mineralogical content of the rocks prone to weathering

Progressive weathering


Table 1: Classification of the most widespread rockfall causes (and, more generally, slope movement causes).
[This paper focuses on the influence of surface temperature changes.]

(2.2.) Lack of quantitative studies on the thermomechanical effects occurring on rock slopes

(2.2.1.) Rock thermomechanics and its standard applications to underground phenomena

The measurement and modelling of coupled TM phenomena in porous and fractured rocks has benefited since the early 1980s from both technical and numerical advances in rock mechanics and rock engineering [4], in conjunction with studies conducted on design and performance assessments of heat-producing drift placement in underground waste repositories, in rock salt [5], granite [6,7] or clay formations [8,9]; circulation of hot or cold fluids for geothermal purposes [10–12]; injection of fluids for oil/gas recovery [13]; and storage of cryogenic fluids, such as liquefied gas, in caverns [14].
In most of these problem situations, fluid circulation is also involved, such that the hydromechanical (HM) coupling is then superimposed on the TM coupling. This paper has taken the perspective of the latter.

Whenever the field of thermohydromechanics is involved, rock mass behaviour cannot be correctly predicted if processes are considered independent from each other: one process affects the initiation and progress of another. Consequently, computational methods prove relevant in seeking to understand and reproduce the naturally complex physical interactions (for a review, see [4]). The inherent difficulty arises from the fact that the HM coupling is a total coupling (since both the H and M processes exert reciprocal influences on one another).

When dealing with ‘‘dry rock problems’’ (without any H processes), the situation becomes slightly easier to comprehend, given that the thermal effects of mechanical changes (changes in thermal properties due to deformation and heat production in connection with mechanical energy dissipation) are conventionally assumed to be small and therefore neglected under usual conditions. As a consequence, with respect exclusively to TM process studies, the main phenomenon to be analysed proves to be the mechanical effect of temperature changes; these changes induce both thermal expansion and thermal stresses, which may lead to joint shear movements [6,14] or cause rupture in the rock mass or in joints [12] and hence result in global mechanical deterioration of the rock mass [15]. Let us note that the mechanical characteristics of rocks also depend upon temperature, but only to a significant extent if the temperature changes are large: such is not the case in our study.

Various computational codes have been developed in order to reproduce TM phenomena measurements. Millard et al. [7] offered a comprehensive review of such codes within the scope of the DECOVALEX project. Due to the small variation range in the thermal diffusivity coefficient (in comparison with other parameters, such as hydraulic permeability) between the various rocks [16], evolution in the temperature field due to conduction behaves rather predictably and is reproduced quite reliably, even within fractured heterogeneous rocks. Its mechanical effects also tend to be well understood, provided that the mechanical parameters of both the rock matrix and fractures are known [5,6].

(2.2.2.) Study of surface phenomena induced by temperature changes

It may be noted that the aforementioned studies all relate to underground phenomena. Quantitative studies of phenomena induced near the earth’s surface by means of surface temperature changes are less prevalent. In this field, numerous papers have been published concerning the study of temperature profiles in the ground in relation, e.g. to the passive cooling of buildings using the ground’s thermal inertia [17,18] or to back-calculations of earth surface temperature evolution inferred from deep boreholes for the purpose of reconstituting past climate changes [19]. The mechanical consequences of surface temperature changes however are almost never determined. Within the specific area of slope movements, studies focusing on surface phenomena induced by temperature changes can be divided into three categories: (i) qualitative geomorphological studies, (ii) freeze-thaw phenomenon studies, and (iii) TM phenomenon quantitative studies. Let us now examine these three types one after the other. The notion that rock surface temperature changes (due to air temperature changes, insulation variations, etc.) may cause rockfalls is not altogether new; more specifically, climate changes tied to global warming and the greenhouse effect have been identified by geomorphologists as being responsible for slope instabilities [20,21]. Nonetheless, the phenomenon involved is much more hydro-chemo-mechanical in nature (due to changes in precipitation rate, geochemical weathering) than TM. Moreover, these considerations are generally qualitative and poorly illustrated by in situ measurements.

The fact that freezing and thawing of the water filling rock mass pores and discontinuities may provoke ruptures was also established a long time ago (see e.g. [22]). This phenomenon, known as ‘‘frost shattering’’ or ‘‘cryogenic weathering’’, has been well described by Zaruba and Mencl [23]: ‘‘Water freezing in rock fissures increases in volume and thus tends to widen them; rocks penetrated by fissures consequently show reduced cohesion’’ (Chapter 2, entitled ‘‘Factors causing mass movements’’). Study of this process has led to a large number of papers presenting rockwall temperature measurements in mountainous regions that attempt to correlate temperature with rockwall retreat (see [24,25] for field observations and [26] for laboratory tests and numerical modelling). As a rockfall cause, the freeze-thaw phenomenon is quite limited however to specific climatic environments (featuring moisture and sub-zero temperatures) and hence cannot alone explain the Rochers de Valabres rockfall that occurred during a period with entirely positive temperatures. Furthermore, this phenomenon is probably much more heavily linked to the mechanical processes resulting from water volume increases than to the actual TM processes on which this paper has been focused.

The consequences on slope stability from more frequently encountered daily and seasonal temperature changes are hardly ever considered from a quantitative point of view. Very recently, Vargas et al. [27] observed that rock slope failures in Rio de Janeiro (Brazil) occur predominantly during the dry winter period when almost no rainfall is recorded. They also discussed the idea that within a context of progressive fracture weathering, fluctuations in surface temperature create thermal stresses, which may ultimately reach magnitudes capable of destroying existing rock-bridges and induce rockfalls (equivalent to Point 2 on Fig. 2). This hypothesis has been confirmed by a finite element calculation of the evolution of the socalled ‘‘stress intensity factor’’ during thermal loadings. Vargas et al. also considered that cyclic temperature changes could be responsible for thermal fatigue. This phenomenon has already been cited as regards the superficial microscopic TM weathering of rocks under repeated natural thermal loads (fatigue failure essentially due to the differential thermal expansion of minerals) in the case of historical monument deterioration [28,29].

The phenomena of fatigue and crack growth are related to microscopic processes; they have not been considered in this study, which focuses on the macroscopic TM effects of daily surface temperature fluctuations (e.g. global displacements), as well as on their impact on slope stability. These macroscopic TM effects are also currently being studied at Checkerboard Creek (British Columbia, Canada), where some rather intriguing annual slope displacement cycles have been identified owing to accurate and prolonged extensometric monitoring [30]. In this case, the controlling mechanism has not been completely resolved; in particular, it is not absolutely clear whether displacement periodicity can be explained by annual changes in temperature or precipitation.

(2.2.3.) Quantitative effects of thermally induced deformations on a rock slope: the ‘‘triangular-block model

A slope surface is affected by both daily and seasonal temperature fluctuations caused by air temperature changes, wind cooling and solar radiation. These cyclic fluctuations are partially transmitted to the interior of the rock mass by means of conduction, in accordance with Fourier’s law (see [31] for the associated theoretical considerations), and induce mechanical changes relative to thermal expansion. Even if the temperature changes rapidly decrease with depth, they may still have significant mechanical consequences (in terms of displacements or stress changes far from the ground surface), especially if the thermal expansion of the superficial zone is constrained.
In order to obtain an initial quantitative assessment of the possible impacts of such phenomena on the stability of a rock-block assembly, let us first consider the simple case of a mobile triangular rock block lying on another block, which is fixed and much larger in size (the corresponding conceptual model is presented in Fig. 3). The rock mass is assumed to behave thermoelastically. Contact between the two blocks consists of a 451 dipping joint. The thermoelastic properties of both the rock mass and joint are given in Table 2.
The calculation was performed under plane strain conditions using the distinct element code UDEC [36], considered to be well adapted for such problems [7]. The code was first run in order to achieve stress equilibrium under gravitational loading and with a uniform temperature field T0, such that the blocks remain immobile in the absence of any subsequent changes. This state is referred to as the initial state (time t = 0). The lateral boundary EF of the upper block is then submitted to imposed temperature changes with time T(t) using a 24-h time period (ω = 2π/24 h) and an amplitude of A = 8°C:
T(t) = T0 + A sin ωt. (1)
Temperature changes are propagated within the rock mass by conduction and induce deformations; this step is followed by analysing the displacements of block corners A, E and F as well as the middle of the block’s frontal boundary (Point M) with respect to the initial position.

[See details in the study: '2.3.1. Case of elastic joint behaviour'; '2.3.2. Case of joint plastic behaviour']

(3.) The ‘‘Rochers de Valabres’’ slope: a test site influenced by typical rockfall causes

(3.1.) Site description

Following the May 2000 rockfall on the Rochers de Valabres slope (briefly described in the introduction), works were undertaken by the CETE Méditerranée Public Works Office to secure the area. During this campaign, the presence of other potentially unstable rock blocks was uncovered; such blocks could neither be blasted nor anchored due to their sizeable volumes. It also proved technically difficult to protect the road from potential ensuing rockfalls. As a consequence, the road was reopened to traffic with a residual risk.

The threatening configuration of certain rock blocks on the Rochers de Valabres slope (with widely opened and valley-dipping discontinuities at their base) led us to assume that other rockfalls might occur. We were thus convinced that this site could provide an attractive setting for conducting and then comparing various investigation techniques, as regards their ability to help comprehend the causes behind the rockfalls presented in Table 1. Under such conditions, when it is impossible to completely eliminate risk, the only (and obviously most widespread) approach to ensuring safe traffic conditions consists of foreseeing the evolution in slope with the help of monitoring devices or numerical modelling techniques; such an approach requires extensive knowledge of the mechanisms involved in slope movements.

The presence of both an abandoned road that provides easy access to the most vulnerably positioned blocks (see Fig. 1) and a hydroelectric power plant gallery, which enables performing investigations inside the rock mass, has served to simplify observations, measurements and instrumentation. Scientific investigations were initiated during the summer of 2002 and consist of: field observations and measurements; laboratory testing; numerical modelling; and geodetic, clinometric and acoustic-emission monitoring.

(3.2.) Rock mass geometry as a predisposition factor: the limit-equilibrium calculation

On a yet highly fractured rock mass, such as the Rochers de Valabres slope, thermal effects are more easily likely to cause movements on the existing (and well developed) discontinuities than induce new ones. For this reason, we will first examine how the fracture network layout (with regard to slope topography) delimits those blocks prone to falling under thermal loads. In this aim, the 3D geometry of the Rochers de Valabres slope was plotted using topographic measurements and field compass surveys of fractures along scan lines both on the side and inside the gallery. In spite of measurement scattering due to irregularities within the discontinuities, it was possible to identify eight ‘‘elementary’’ families (F1–F8 on Fig. 7), to be grouped into three main fracture sets with respect to their orientation. The first two sets are generally parallel to the side: one dips towards the inside of the rock mass (families F4 and F5), while the other dips towards the outside on the down slope (F6–F8); dip angles vary from 401 to 601. The third family is composed of sub-vertical fractures with highly variable directions (F1–F3), to such an extent that they determine vertical wedges in the side. These scan-line surveys have been completed by photographic analyses performed on the basis of both land and aerial photographs of the slope. Results have shown that the discontinuities tend to display the same orientations at the various scales and that the previously described discontinuity network heavily influences site morphology (orientation of the gullies, overhangs and downwardly dipping slabs). A study of the specific zone involved in the May 2000 rockfall has revealed that the fallen blocks slid on a preexisting, valley-dipping discontinuity belonging to the F4–F5 fracture set. It must then be considered whether this configuration acts to further propagate instabilities elsewhere on the slope. We undertook a 3D stochastic limit-equilibrium calculation using the RESOBLOK code [38], considering the rock mass as an assembly of undeformable blocks separated by discontinuities. Slope topography is assumed to be plane and the model geometry stochastically reproduces rock mass geometry: the discontinuity network aligns with orientations measured in the field. The stability calculation is performed iteratively: at each iteration, a safety factor is calculated for those blocks exhibiting a free boundary and geometrical mobility. ‘‘Unstable’’ blocks (geometrically and mechanically removable) are then removed and recorded, along with specifying the type of mechanism involved (free-fall, plane slide, wedge slide, etc.). Upon completion of the calculation, once the remaining blocks are all considered ‘‘stable’’ and the iterative process has stopped, it becomes possible to analyse the distribution of fallen blocks within the various categories. In our case, we performed computations by applying zero cohesion to all fractures in order to investigate the primary failure mechanisms, with no intention of calibration in terms of fallen volume. The majority of potential unstable blocks thus result from plane slides on downward-dipping discontinuities (the F4–F5 fracture set). This finding has confirmed that such fractures play a major role in rockfalls occurring on the Rochers de Valabres slope: slope geometry will therefore be considered as a predisposition factor for rockfalls (as presented in Section 2.1). This finding does not suggest that geometry alone can explain rockfalls; rather, it sets the backdrop for failures, yet remains insufficient to instigate one since certain blocks in the previously described configuration are indeed still standing; consequently, the authors have decided to concentrate our research on thermal effects.

(4.) Exploring thermally induced movements in situ

Before presenting the measurement results and in continuity with Section 2.3, the authors carry on with the computational investigation of thermal effect with a geometry much closer to that of the Rochers de Valabres slope. [See sections '4.1. Numerical modelling' + '4.2. High-precision geodetic monitoring procedure' pp. 339-345 of the paper]

4.3. Preliminary results and discussion

(4.3.1.) Thermally induced displacement measured with our monitoring device

It would be impossible to provide the details of all measurements and their interpretation. For this reason, we will concentrate on the distance measurements of four prisms (referred to as Prisms 109, 200, 201 and 202) positioned on the cut presented in Fig. 13. These measurements effectively illustrate the difficulties involved in conducting high-precision remote-sensing monitoring, along with the type of data potentially derived. Distance changes for the prisms considered are shown on Fig. 14, after correction using the a coefficient and incorporation of possible temperature measurement errors.
The distance measurements are quite precise (the uncertainty of 0.75 104 on the data is presented in Fig. 14 as an error bar), but since distance variations are small, they lead to raise the question of accuracy and consistency of determinate error corrections. The reference prism employed herein in fact merely yields an indication of the angular measurement accuracy. As far as distance measurements are concerned, no means are currently available to determine whether our measurements are accurate and whether their accuracy can vary over the course of a day due to changes in meteorological conditions.
In supposing that the variations observed in Fig. 14 are significant, we can note that Prism 109 exhibits the most sizeable distance variation, with an amplitude of about 1 mm. This variation is correlated with air temperature changes and, as a consequence, with temperature of the rock mass surface. A TM origin can therefore be assumed. A delay of approximately 3 h may also be noted; for the time being, this delay remains unexplained since surface displacements were expected to be nearly in phase with temperature changes (see Section 4.1.2). Moreover, it should be pointed out that no significant permanent displacement has been recorded.
These preliminary results reveal that, even if our monitoring device is able to shed light on thermally induced displacements, it needs to be enhanced. Surface tiltmeters recently installed near the prisms will provide additional information about TM movements in the future. High-precision geodetic monitoring should not be considered as a standalone approach, but rather as part of a more complete monitoring device.

(4.3.2.) Comparison between modelling and monitoring

Measurement results may be compared with the computations derived from numerical modelling. For this purpose, we used the same numerical model as that presented in Section 4.1 and applied the air temperature history recorded in situ at its boundary. Numerical modelling results are presented in Fig. 14.
The computed distance changes are similar in shape to those obtained by monitoring (except for Prism 201, which features an inverse evolution path). In particular, the changes are essentially concentrated during the period of high-temperature change rate. Amplitudes are slightly lower in the numerical model than in reality, yet the order of magnitude is indeed respected. The relative deviation between computed and measured distance changes could have been decreased by calibration, e.g. by changing the Young’s modulus or Poisson’s ratio values.
Nonetheless, we prefer improving both the measurement device and numerical modelling before undertaking such a calibration. The effect of boundary conditions must be investigated: we have used thermal boundary conditions with imposed temperature histories, in assuming that both the rock surface temperature and air temperature are equal. As stated earlier, this approach is probably not completely representative of reality; radiation or convective boundary conditions could therefore be better adapted. Jenkins and Smith [49] provided a good illustration of the complexity of rock surface temperature evolution under natural conditions. From their example, it may be concluded that just about the only way to apply real thermal boundary conditions to the model would be to rely upon surface temperature measurements conducted in situ.
The fact that major movements have been measured for Prism 109 and not for the others may be explained by either geometrical or mechanical heterogeneities in rock mass behaviour. Spatial variability of the solar heat influxes (and thus in the thermal loads) may also be cited. This last hypothesis must be explored by calculating the total calorific energy received from the sun at the various points, thereby taking into account both the surface orientations with respect to the sun and the presence or absence of shadow zones.
Another way to enhance modelling could be to acquire better knowledge of the physical parameters of the rock mass. While thermal parameters exhibit a small range of variation and do not need to be measured, the variability in mechanical parameters (such as joint normal and shear stiffness, plastic limit) is of major importance for the quantitative description of the response to a given perturbation. The mechanical laboratory tests now underway on natural joint samples will probably lend a better idea of the joint mechanical properties. Their results however will not be easy to use for numerical modelling, due of course to the scale effect.
The last point to be considered in the model is the influence of seasonal temperature changes on the initial state used for calculation purposes. In our model, results depend strictly on daily temperature changes and are not modified by the initial temperature within the rock mass. This is only true if we consider the case of linear elasticity.

(4.3.3.) Temperature oscillations as a preparatory factor for rockfalls?

With their device, the authors are able to detect a portion of the thermally induced movements. As for now, no strict evidence has been found of permanent displacements induced by thermal effects, neither in numerical modelling nor in surface movement monitoring. As a consequence, efforts should be focused on long-term monitoring and computation since permanent displacements could be of such a small magnitude that they remain almost imperceptible over a short-term monitoring period, such as the one presented herein.
It should be recalled that temperature is also responsible for the slow mechanical weathering of the rock mass. Moreover, being able to quantify TM effects in a rock slope can be useful when interpreting any displacement or deformation measurement: it provides elements to determine the amount of displacement or deformation that may be attributed to thermally induced effects.

(5) - Syntèses et préconisations


Références citées :

[1] Follacci J-P, Guardia P, Ivaldi J-P. Le glissement de terrain de la Clapière (Alpes-Maritimes, France) dans son cadre géodynamique. Proceedings of the Fifth International Symposium on Landslides, Lausanne. Balkema: Bonnard; 1988.

[2] Julian M, Anthony E. Aspects of landslide activity in the Mercantour Massif and the French Riviera, southeastern France. Geomorphology 1996;15:275–89.

[3] Finlayson B, Statham I. Hillslope analysis. London: Heineman; 1980.

[4] Jing L. A review of techniques, advances and outstanding issues in numerical modelling for rock mechanics and rock engineering. Int J Rock Mech Min Sci 2003;40:283–353.

[5] Pudewills A, Droste J. Numerical modeling of the thermomechanical behaviour of a large-scale undeground experiment. Comput Struct 2003;81:911–8.

[6] Rejeb A, Vouille G, Derlich S. Modélisation du comportement thermomécanique d’un massif granitique–application à la simulation de l’expérience THM de Fanay-Augères. Rev Fr Géotech 1990;53:21–31.

[7] Millard A, et al. Discrete and continuum approaches to simulate the thermo-hydro-mechanical couplings in a large, fractured rock mass. Int J Rock Mech Min Sci 1995;32(5):409–34.

[8] Giraud A, Rousset G. Thermoelastic and thermoplastic response of a porous space submitted to a decaying heat source. Int J Rock Mech Min Sci 1995;19:475–95.

[9] Rutqvist J, et al. Coupled thermo-hydro-mechanical analysis of a heater test in fractured rock and bentonite at Kamaishi Mine— comparison of filed results to predictions of four finite element codes. Int J Rock Mech Min Sci 2001;38:129–42.

[10] Bruel D. Modelling heat extraction from forced fluid flow through stimulated fractured rock masses: evaluation of the Soultz-sousforets site potential. Geothermics 1995;24(3):439–50.

[11] Kolditz O. Modelling flow and heat transfer in fractured rocks: conceptual model of a 3-D deterministic fracture network. Geothermics 1995;24(3):451–70.

[12] Bérard T, Cornet FH. Evidence of thermally induced borehole elongation: a case study at Soultz, France. Int J Rock Mech Min Sci 2003;40:1121–40.

[13] Lauwerier HA. The transport of heat in an oil layer caused by the injection of hot fluid. Appl Sci Res 1995;A5:145–50.

[14] Monsen K, Barton N. A numerical study of cryogenic storage in underground excavations with emphasis on the rock joint response. Int J Rock Mech Min Sci 2001;38:1035–45.

[15] David C, Menendez B, Darot M. Influence of stress-induced and thermal cracking on physical properties and microstructure of La Peyratte granite. Int J Rock Mech Min Sci 1999;36:433–48.

[16] Yow JL, Hunt JR. Coupled processes in rock mass performance with emphasis on nuclear waste isolation. Int J Rock Mech Min Sci 2002;39:143–50.

[17] Jacovides CP, Mihalakakou G, Santamouris M, Lewis JO. On the ground temperature profile for passive cooling applications in buildings. Sol Energy 1996;57(3):165–75.

[18] Mihalakakou G. On estimating soil surface temperature profiles. Energy Buildings 2002;34:251–9.

[19] Demezhko DY, Shchapov VA. 80,000 years ground surface temperature history inferred from the temperature-depth log measured in the superdeep hole SG-4 (the Urals, Russia). Global Planet Change 2001;29:219–30.

[20] Schmidt J, Dikau R. Modeling historical climate variability and slope stability. Geomorphology 2004;60:433–47.

[21] Dehn M, Bu¨ rger G, Buma J, Gasparetto P. Impact of climate change on slope stability using expanded downscaling. Eng Geol 2000;55:193–204.

[22] Battle WRB. Temperature observation in bergschrunds and their relationship to frost shattering. Norwegian Cirque Glaciers, Royal Geographical Society series, vol. 4. Boca Raton: Lewis; 1960. p. 83–95.

[23] Zaruba Q, Mencl V. Landslides and their control. Developments in geotechnical engineering, vol. 31. Amsterdam: Elsevier; 1969.

[24] Matsuoka N, Sakai H. Rockfall activity from an alpine cliff during thawing periods. Geomorphology 1999;28:309–28.

[25] Coutard J-P, Francou B. Rock temperature measurements in two alpine environments: implications for frost shattering. Artic Alpine Res 1989;21(4):399–416.

[26] Neaupane KM, Yamabe T, Yoshinaka R. Simulation of fully coupled thermo-hydro-mechanical system in freezing and thawing rock. Int J Rock Mech Min Sci 1999;36:563–80.

[27] Vargas Jr. E, Castro JT, Amaral C, Figueiredo RP. On mechanisms for failure of some rock slopes in Rio de Janeiro, Brazil: thermal fatigue? In: Lacerda, et al., editors. Landslides evaluation and stabilization, Proceedings of the Ninth International Symposium on Landslides. London: Taylor & Francis Group; 2004.

[28] Galan E, Guerrero MA, Vazquez MA, Zezza F. Progressive deterioration of marble columns by thermal changes in relation to their state of superficial decay. In: Delgado Rodrigues, et al., editors. Proceedings of the Seventh International Congress on Deterioration and Conservation of Stone, Lisbon, Portugal, June; 1992.

[29] Royer-Carfagni GF. On the thermal degradation of marble. Int J Rock Mech Min Sci 1999;36:119–26.

[30] Watson AD, Moore DP, Stewart TW. Temperature influence on rock slope movements at Checkerboard Creek. In: Lacerda, et al., editors. Landslides: evaluation and stabilization, Proceedings of the Ninth International Symposium on Landslides. London: Taylor & Francis Group; 2004.

[31] Holman JP. Heat transfer, 9th ed. New York: McGraw-Hill; 2002.

[32] Bieniawski ZT. Engineering rock mass classifications. London: Wiley; 1989.

[33] Gunzburger Y, Merrien-Soukatchoff V. Caractérisation mécanique d’un grand versant rocheux instable au moyen du système RMR—Cas de la Clapière (Alpes françaises). In: PARAM 2002, Proceedings of the International Symposium on Identification and Determination of Soil and Rock Parameters for Geotechnical Design, Paris, 2–3 September 2002. Paris: Presses de l’ENPC; 2002.

[34] Berest P, Weber P. La Thermomécanique Des Roches, vol. 16. Orléans, France: Editions du BRGM; 1988.

[35] Rachez X. Les fondations au rocher de grands viaducs: l’apport de la méthode des éléments distincts. PhD, ENPC, Marne-la-Vallée, France, 1997.

[36] Itasca Consulting Group, Inc. UDEC (Universal Distinct Element Code), version 3.3, Minneapolis, 2003.

[37] Carslaw HS, Jaeger JC. Conduction of heat in solids. 2nd ed. Oxford: Clarendon Press; 1959.

[38] Heliot D. Generating a blocky rock mass. Int J Rock Mech Min Sci 1988;25(3):127–38.

[39] Kadiri I, Merrien-Soukatchoff V, Guglielmi Y. Interpretation and modelling of a hydromechanical in situ experiment within a fractured calcareous rock mass. ISRM 2003 Symposium— Technology Roadmap for Rock Mechanics, 8–12th September 2003, Johannesburg, South Africa, p. 603–608.

[40] Malet J-P, Hartig S, Calais E, Maquaire O. Apport du GPS au suivi continu des mouvements de terrain. Application au glissement-coulée de Spuer-Sauze (Alpes-de-Haute-Provence, France). C-R Géosci, Acad Sci Paris, Earth Planet. Sci. 2000;331:175–82.

[41] Malet J-P, Maquaire O, Calais E. The use of global positioning system techniques for the continuous monitoring of landslides: application to the Super-Sauze earthflow (Alpes-de-Haute-Provence, France). Geomorphology 2002;43:33–54.

[42] Gili JA, Corominas J, Rius J. Using global positionning system techniques in landslide monitoring. Geomorphology 2000;55: 167–92.

[43] Casson B, Delacourt C, Baratoux D, Allemand P. Seventeen years of the «La Clapière» landslide evolution analysed from orthorectified photographs. Eng. Geol. 2003;68:123–39.

[44] Tarchi D, et al. Landslide monitoring by using ground-based SAR interferometry: an example of application to the Tessina landslide in Italy. Eng Geol 2003;68:15–30.

[45] Fruneau B, Achache J, Delacourt C. Observation and modelling of the Saint-Etienne-de-Tinée landslide using SAR interferometry. Tectononohysics 1996;265:181–90.

[46] Rizzo V, Leggeri M. Slope instability and sagging reactivation at Maratea (Potenza, Basilicata, Italy). Eng Geol 2004;71: 181–98.

[47] Bogaard TA, Antoine P, Desvarreux P, Giraud A, van Asch ThWJ. The slope movements within the Mondorès graben (Drôme, France); the interaction between geology, hydrology and typology. Eng Geol 2000;55:297–312.

[48] Milles S, Lagofun M. Topographie et Topométrie Modernes. France: Eyrolles; 1999.

[49] Jenkins KA, Smith BJ. Daytime rock surface temperature variability and its implications for mechanical rock weathering: Tenerife, Canary Islands. Catena 1990;17:449–59.

[50] Mendecki AJ. Real-time quantitative seismology in mines. Rockburst of seismicity in mines. Rotterdam: Balkema; 1993.

[51] Senfaute G, Abdul-Wahed M, Piguet J-P, Josien J-P. Qualification of the microseismic monitoring technique applied to the risk of collapse in iron ore mines. EUROCK Symposium, Aachen 2000:597–602.

[52] Niitsuma H, Chubachi N, Takanohashi M. Acoustic analyses of a geothermal reservoir and its application to reservoir control. Geothermics 1987;16:47–60.

[53] Senfaute G, Merrien-Soukatchoff V, Morel J, Gourry J-C. Microseismic monitoring applied to prediction of chalk sides collapses and contribution of numerical modelling. In: Picarelli, L., editor. Proceedings of the International Conference on Fast Slope Movements, Naples, 11–13 May 2003. Bologna: Pa` tron.

[54] Gunzburger Y, Merrien-Soukatchoff V, Senfaute G, Guglielmi Y, Piguet J-P. Field investigations, monitoring and modeling in the identification of rock fall causes. Proceedings of the Ninth International Symposium on Landslides (ISL), Rio de Janeiro (Brazil), June 28–July 2, 2004.