Вы находитесь на странице: 1из 14
INTRODUCTION REVIEW CONSTITUTIVE MODELING AND ANALYSIS OF CREEPING SLOPES By Chandra S, Desai,’ Fellow, ASCE, Naresh C. Samtani, ‘Associate Member, ASCE, and Laurent Vulliet,’ Member, ASCE ‘Asstnact: A constitutive model for soils and interfaces involved in ereeping natural slope is proposed, It is based on the hierarchical single-surface plasticity and wscoplastcity approaches and allows for factors such a clastic, plastic and creep strains, normal stress, and stresspath effects. The model isc ‘rom a series of laboratory triaxial (ests for soils obtained from the field site at Villarbeney Landslide, Switzerland, simple shear tests for interfaces, and simple shear creep tests for both. Its verified with respect 10 the laboratory tests used for finding the parameters and independent tests not used in finding the parameters. The model is implemented in a two-dimensional finite-element procedure, whichis then used to back-prediet observed field behavior at two locations At the Villarbeney Landslide. Typical comparisons between finite-clement predictions fd field data show thatthe proposed modeling procedure provides highly satisfactory correlations forthe field situations considered herein. Many natural slopes exhibit slow and continuous movements under gravity load that are affected by creep or viscous response of geologic materials in the slopes. Such movements can lead to landslides with detrimental effects on structures such as roads and buildings that are built on or near the slopes, and can result in the loss of human life. Hence, it is necessary to develop advanced and realistic procedures forthe prediction of the behavior of creeping natural slopes, ‘One of the important factors that influence the behavior of creeping slopes is appropriate characterization of the response of geologic materials and interfaces; in the case of creeping Slopes, the latter can occur at the junction of the creeping mass and the essentially stationary (rock) mass below it “Although a number of investigators have proposed mechanistic idealizations and constitutive models for creeping natural slopes (see review below), none appears to allow consideration of factors such as elastic, plastic, and creep deformations, volume changes, and stress paths in the constitutive models in @ unified manner. The objectives of the present paper are: (1) To present such a unified constitutive model for geologic materials and interfaces; (2) to calibrate the model land determine its parameters from a set of laboratory tests; (3) t0 validate the model with respect to stres-strain-time behavior observed in the laboratory tests; and (4) to illustrate the capability of the model by introducing it in 4 nonlinear finite-element procedure, and by back- predicting observed field behavior of a typical natural slope at Villarbeney Landslide, Swit- 2erland, ‘The literature on analysis of creeping natural slopes and landslides is wide in scope, and itis not intended to provide a comprehensive review herein. Inthe following, a few relevant studies, ‘are briefly reviewed. Various models with different assumptions regarding constitutive behavior have been pro- posed: (Ter-Stepanian 1963; Saito 1965; Yen 1969, Brown and Shew 1975; Savage and Chleborad 1982; Dysii and Recordon 1984; Kawamura 1985; Iverson 1986a,b; Lu and Wang 1988; Vulliet and Hutter 1988a,b; and Davis et al. 1993). Most of these studies are based on simplified Constitutive models such as linear elasticity, classical plasticity (based on von Mises and Drucker- Prager failure criteria), and creep equations that are one-dimensional and often expressed in terms of stress invariants. The latter are often based on Newton, Bingham, and Norton laws (Vulliet and Hutter 1988). However, as indicated previously, these constitutive models are Considered to be restrictive as they do not allow fully forthe foregoing factors under Introduction ‘and multidimensional effects. ‘Regents Prof, Dep of Gi, Engra and Engrg. Mesh, Univ of Arizona, Tucson, AZ NST2I Sr Gcoteh Engh. Fase Bol in | enn Fz New York, NYT, ory Grad Sdn at Dept of CW. Engr. and Engr Mech Unis of Arizona Tueson, AZ "Pro, Dep. of Civ. Engrg. Swiss Fe Inst of Technol, Lausanne Switzerland Note, Discusion open untl June 1, 18S, To extend te cloning dat one month a written request must he fled withthe ASCE Manager of Journals, The manuscript Tor this paper was submitted fr review and pose publication on Februny 7.1982. This paper is pat ofthe Journal af Geotechnical Engineering. Vol. 121. NO. 1. Finusry. 1095. BASCE. ISSN 0730-34109S0RN 0043-00562 00 18.25 per page. Foper No. M21 JOURNAL OF GEOTECHNICAL ENGINEERING 43, ‘The present study toward the analysis of creeping natural slopes is a continuation of the pre vious works by Vulliet and Hutter (1988a,b). They had proposed solution procedures for the slow motion of a sliding fluid-solid mixture based on equilibrium equations derived from balance (of mass and momentum. These equations were solved approximately at speitie points in the sliding mass, which was divided into a grid. The ereep behavior was described by using 3 number of functions, eg. a sliding dominant function and a creep dominant function, which were expressed in terms of the Drucker-Prager failure eriterion, In their work. the main purpose was to simulate slowly moving landslides (or landslides in progress). without considering the starting phase of the movement. und without investigating a possible ereep failure (large acceleration (of movement, influence of inertia, and so forth). On the other hand, they considered three- dimensional effects and hydromechanical coupling. The constitutive law wsed was purely viscous: that is, it related the state of stress to the strain rates. Their model did not consider clastic effects and primary creep. Another important point in the Vulliet and Hutter (1988a,b) approach ‘was that the three-dimensional model was constructed by assuming that the moving mass Was thin, which may be true in general only for large natural slopes. As a consequence. it was pos- sible, using asymptotic developments, to decouple the calculation of the free-surface displace: ‘ment from the determination of the state of stress, Stresses were calculated along a line normal to the Tree surface, taking into account only the longitudinal and transverse slopes of the free surface and of the sliding base. A drawback with this assumption is that the model does not ‘work well at locations where sucden changes (in slope, in thickness. or in material parameters) ‘occur. This is particularly true for the front part (the tongue) of the landslide or at the location of retaining structures (retaining walls, piles, anchors, nails). Therefore. their model cannot be used without modifications for Tactors such as realistic geometry and material properties, ‘As a result, the third writer of the present paper. Vliet, realized a need for more general representation through a unified constitutive model and the finite-element method for more realistic prediction of the behavior of creeping slopes. Toward this end, the researcher visited the University of Arizona, Tucson. fora period of about two years under a grant from the Swiss [National Seience Foundation. As a part of the study, soil from the Villarbency site was brought to Tucson; however. as the amount of soil available was limited, only triaxial and interface tests under most important conditions required for the ealibration of the model were conducted. CONSTITUTIVE MODELS A version in the hierarchical single surface (HIS) family of the constitutive models is for- ‘mulated here for characterizing the elastoviscoplastic behavior of soils and interfaces, Various versions, allowing for associative, nonassociative, anisotropic hardening, disturbance or damage, and fluid flow pressure characteristics and advantages of the HISS approach, have been presented elsewhere (Desai et al. 1986; Desai 1990). The elastoviscoplastic version proposed herein and referred to as the 8,, model, allows for elastic, plastic, and viscous or ereep deformations, continuous yielding Gr hardening, volume changes, and stress-path effects. Descriptions ofthe ‘model for “solid (continuum) geologic materials and interfaces are given herein Tt is noted that the basic framework of the model for solids and interfaces is essentially the same; as described subsequently, the model for interfaces can be derived as a special case from that for the solids. This isa dstinet advantage, particularly since in the finite-element procedure both solids and interfaces have consistent models with the same framework, Solid Geologic Materials “ ‘The proposed model is bused on the theory of elastoviscoplastcity by Perzyna (1966). ‘This theory has often been used to characterize creep or viscous behavior of geologic materials with different plasticity models such as von Mises and Drucker Prager failure criteria (Zienkiewicz ‘and Cormeau 1974; Katona and Mulert 1984). Development towards use of the HISS associative land nonassociative versions, with Perzyna’s theory, have been reported by Desai and Zhang (1987), Valliet and Desai (1989), and Sarntani and Desai (1991). Development of a methodology for finding the parameters for the viscoplastic models, and of extension of the model for vis- coplastic response of interfaces, are among the new contributions in the present paper. ‘Assuming stall strains. the total strain rate tensor. é,, is decomposed into elastic, viscoplastic. €7, parts as and (1a) where uy) and é ae) JOURNAL OF GEOTECHNICAL ENGINEERING where overdot = time rate; oy = effective stress tensor; Ci = elastic constitutive tensor: & "a scalar flow function of the yield function FF, = a normalizing constant having same units as FQ = viscoplastic potential function; and = a material (Musty) parameter. The angle bracket () has the meaning of switch-on-switch-ff operator: when FIF, > 0,(O(E'E,)) = &(F! F). and when FIF, = 0, ((FIF,)) = 0. A number of alternatives are possible for defining & iz and Cormeau 19742 Katona and Mulert 1986; Desai and Zhang 1987). Here, based the following expression was found suitable: o()-() ® where a matefal parameter Based on the laboratory test behavior (described later. its found appropriate to characterize the soil mass using associative plasticity, that is, Q = F. The yield function. Fis given by (Desai et al. 1986) Fa dy ~ (aly + UM ~ pS) = 0 where S, = (VI) oaz!* = stress ratio; J, = fist invariant ofthe effective stress tensor 0,3 nd Jay andJjp are the second and third invariants ofthe devitorie stress tensor. respectively. F and the invariants are nondimensionalized with respect to the atmospheric pressure, P, Parameter n is related tothe state at which transition from volume contraction to dilation occurs, scribes continuous yielding, and a is the hardening or growth function that id is expressed & where &, = f (de? deit)'? is the length of viscoplastic volumetric strains trajectory, and a, and, ae the hardening parameters. Details of the determination of the parameters in the ‘model and their determination are given later and in Appendix 1. The surface given by F plots as continuous and closed in various stress spaces (Desai etal. 1986) o ‘The deformation mechanism at interfaces between two contacting bodies is different than that for an element of the “solid soil material. Relative translational motions can occur at an interface between adjacent points on the two bodies leading to slippage of one body past the other. There are other modes of relative motion possible at the interface such as separation oF Gebonding, rebonding, and interpenetration (Desai et al. 1984); however, in the case of a creeping natural slope. the translational motion is considered to be predominant, In contrast to a relatively distinct interface that ean be assigned finite equivalent smeared thickness, say in the case of interfuee between a structural and a geologic material (Desai et al 1984), in the case of the creeping slope, the interface representation assumed herein involves. a finite “interface zone.” In this zone, near the junction of the moving mass or solid body and the underlying stationary or parent mass, Fig. 1(c), the variation of translational displacements, are much more severe than those in the mass above the interface zone. This is based on the ‘observed pattem of movements in ereeping slopes, Fig. (a) (Gould 1960); a schematic of these patterns is shown in Fig. 1(b). The location O in Fig. I(b) where the movements ate negligible can be assumed as the stationary, or fixed, base, The depth corresponding to OC is the overall depth of the slope that experiences movenients, while the depth corresponding to OB involves high levels of relative motions compared to those in the portion CB. The extent of OB ean be about 25% of the depth of the slope, OC. Zone OA, near the base of the slope, experiences 4 pattern of relative shear displacements that can be considered to approximate simple shear strain conditions. In the present study. this zone OA is treated as the interface; its extent is about 25% of the zone OB. Thus, the thickness of the interface zone is about 6.0% of the thickness of the moving mass. The interface model proposed herein can allow for the variation of stresses and strains in the interface zone. Thus, the possibility of deformations in the entire creeping slope is included in the proposed procedure, by characterizing the mass in AC as solid elastoviscoplastic material, and by char- acterizing the portion OA as an “intesface zone.” with a special model as described later. In contrast to previous models (e.g. Davis etal. 1993), in which the mass AC is assumed to be rigid.” the proposed model allows for the deformation in the moving mass as well asin the interface zone, and hence. is considered to be more general Consider a two-dimensional planar idealization of the interface. The constitutive model for the interface is developed by specializing the three-dimensional model for the solid body by adopting the normal, 6, and shear +, components of the stress at the interface plane, corre: sponding to the stress invariants J, and VJay,. The specialization is obtained by considering the JOURNAL OF GEOTECHNICAL ENGINEERING «5 Horizontal Displacement byrne Sold Body 76.21 Depth below grea sraco o Berehoistor sr.gp| Leeson 1 pes ste) Ver 5.08 356 (Observed Honzntal Movernnt (em) ia) FIG. 1. Landslide and Measured Behavior: (a) Incinometric Profiles of Natural Slope along Coast of ‘Southern California (Gould 1960) (b) and (e ldealtzations of Creeping Slope orm for soi 0 Firs marian, “Analogous tor fr itrtac @ Normal es Votumetre sean, Relative normal displacement Dewstore tain, Relative she (ansatonal placement, Length of viscoplastic strain wajectory. 6, | Length of viscoplastic relative displacement trajectory. é “tur ‘analogies in Table 1. With the analogies, the yield function F, (3) specializes to yield function F for interface as (Desai and Fishman 1991) R= P+ qe — 503 = 0 5) nnondimensionalized with respect to p,:the term (1 ~ 8S,)-""Sin (3) is assumed (0): Bis related to the shape of the yield surface in the principal stress space: plots as continuous function (Fig. 2); m, = parameter related to the state at which the normal displicement transits from compression to dilation; ¥, is related to the Slope of the ultimate envelope for F, when a, = 0 (Fig. 2): and a, is the hardening function given by a 6 where a and b = hardening parameters for the interface, £p.and fy, = lengths of viscoplastic Volumetric and deviatoric strain trajectories, respectively. This subdivision permits specialization for materials for which hardening may be defined based only on the volumetric component. Based on the laboratory test behavior (described later). it was found appropriate to charac terize the behavior of interfaces using nonassociative plasticity. Hence, the plastic potential function, Q,, was defined as follows: Q= 8 salon = 4 o were ag was exprese asthe funtion of a for atoiatve plasty a8 = ‘) “ 46. JOURNAL OF GEOTECHNICAL ENGINEERING 50 2. g 2! 28 = Ukimate & State Change = ° 150 178 200 aera) FIG. 2. Yield Surtaces for interface where a = value of a at the beginning of shear loading: a,, = value of ay at the peak shear Sites; and x — nonastocitive parameter. "The viscoplastic ate of relative incremental strains fr the inertace i given by 20, = oy 0. We Ee = 28 Ad ep, = Wt = normal an shear vicoplestic strains, respective hickitess of the interface (Desai etal. 1984), Based on laboratory tests the following form of flow Function, was found to he appropriate ((B))= (8) where I, and N, = interface viscous parameters MATERIAL PARAMETERS AND LABORATORY TESTS. Soll Material parameters for soils and interfaces are obtained from laboratory triaxial and simple shear tests, respectively. Table 2 presents the values of the parameters for soil and interface materials. The methods for finding the parameters for the elastoplastic part of the model ate given elsewhere (Desai et al. 1986; Desai 1990), while the procedure for finding the viscous parameters is given in Appendix |. Brief descriptions of the laboratory tests pertormed under the present study. as well as those performed previously by Vullit (1986) used for finding the parameters, are given later herein, The constitutive equations are expressed in terms of effective Stresses; fence, for the purpose of finding the parameters, eflective stresses are used. “The soils at the locations £, and Es considered herein (und identified later) have similar physical properties with average water content of 20%. plastic limit of 20%. Higuid limit of 40%. Ary unit weight of 18.5 kNim’, clay fraction of 38%, and specific gravity of solids of 2.66; they are classified as CL. Laboratory experiments (Laboratoire de la Mecanique des Sols 80) showed that the sliding mass is normally consolidated, Also. the material in the landslide originates from the flow of material from an eroding cliff of Oxfordian siltstone, and not from general surface erosion or removal of glacier pressure, which can be 4 souree of overconsolidation, AAs indicated before, only 2 limited amount of the soil was available, Hence. triaxial tests only under limited stress paths were performed: the selected tests are considered to be appro- priate as the elastic and plastic parameters for the proposed model can he found mainly #rom JOURNAL OF GEOTECHNICAL ENGINEERING «7 TABLE 2. Parameters tor Soll and Intertace Peano] Smear] vane | ~Poarwer | Smet] Vane wr i a oe |r ‘a Ee Tans igh [Roms] Ea re Boe in | ae [Srmainees | [Noite ry him 7 ae Yo a tha i na B e erat ‘l 38h | pti ; zu ioe 2 ae ee : de teat q time [fae ; a one te Ronee ott me sr Vises aay Tima oy 7 Tea tn i oa [eae u on na ay 2] a0 {/ | a / en | i) | al \ A L iQ Diasase reo wunamM 9% 16 ow a0 00 3 Axial (Vertical) strain % Jy kPa io io FIG. 3. Resuts for CTC 15 Sol Test: (a) ctahedral Shear Stress and Pore Pressure versus Axial Strain; ‘and (b) Stress Path three compression and one extension tests. Tests performed were: one K,~ consolidation; one isotropic oF hydrostatic compression (HC) (@, = a; = a4) with maximum stress = 480 kPa; yo conventional compression (CTC) (a, > 6; = 0) and one reduced triaxial extension (RTE) (reduce 0 and hold, and er constant): here, ¢,. 62, are the prineipal stresses. The two CTC tests were performed with the initial confining stress, ¢. = 0 ~ 108 and 345 kPa (15 ‘and 50 psi), while an intermediate test at = 200 kPa (29 psi) performed previously by Vulliet, (1986) was used. Typical test results for o = 103 kPa (15 psi) for octahedral shear sttess tn. versus axial strain e,, pore-water pressure t versus e, and stress paths in J, versus VJay space are shown in Fig. 3 “The viscous parameters P’and N were found from the undrained creep tests with measurement of pore water pressure performed by Vulliet (1986) by using a simple shear strain GEONOR HIP apparatus, with normal stress @, of 210) and 400 kPa, and constant shear stress such that the stress ratio 1/a, was 0.6 and 0.8, respectively. A subsequent figure shows typical test results in terms of sheat strain versus time for the stress ratio = 0.6. In these tests, it was assumed that the horizontal stress was approximately equal t0 the vertical stress, which was verified in fone case with measurement of circumferential stress in the reinforced membrane using the reinforcement liner as electrical resistance (Valet 1986). Interface A series of simple shear tests were performed under the present study by using the cyclic rmulti~degree-of-freedom shear device (Desai 198!) for interfaces between simulated base ma- terial and sliding soils (Samtani and Desai 1991). The upper part of the shear (box) device ‘contained the stiffer base, while lower part of the box included the soil contained in a stack of Circular aluminum rings that allowed simple shear deformations. The internal diameter of the ig was 16.5 cm (6.5 in.), with the total thickness of the soil specimen to be 3.6 em (1-4 in.) ‘The normal stresses 103, 207, and 345 kPa (¢, — 15, 30, and 50 psi) were first applied to simulate a range of overburden stresses. After the samples were consolidated under normal stresses, the specimens were sheated under drained conditions with different amplitudes of shear 42 JOURNAL OF GEOTECHNICAL ENGINEERING 0.005 zo 3 coms Fan F001s 8 om F oms of | ____—_| eines hex as ase? 0 0] 03 03 04 03 08 07 tar ipl eten Sher Dipasemen cm ia ‘o FIG. 4. Results for Intertace Test at ¢, = 207 kPa, ur = 0.68 em ‘Experiment > Model cas ‘© 100 200 300 400 500 «00 70 Axial (Vera) strain, % Si kPa a) ‘b) FIG. 5. Comparison between Predictions and Observations for RTE 30 Soll Test © ——____ : ” 2 0m . § I Experiment 0 . | fon ma gu Eepeiment fous & Model i Dor io -oons ° | Fal = 003 04 os 08 1 12 0 02 as 06 08 1 12 14 Relative shear dplacement, em Relative Shear Displacement, em ‘ah ‘61 FIG. 6. Comparison between Predictions and Observations for Interface Test v, = 103 KPa displacements, uz ~ 0.19, 0.64, and 1.27 em (0.075, 0.25, and 0.5 in.). Fig. 4 shows typical results for o,, = 207 KPa (30 psi) and « = 0.64 em (0.25 in.) ‘The viscous parameters for the interface were found by using the simple shear creep tests performed previously by Vulliet (1986). As explained in Appendix I, these parameters for the Sil and interface are difterent because the soils characterized as associative, while the interface is characterized as nonassociative, and because the yield and hardening functions are different. VALIDATION OF CONSTITUTIVE MODELS. The constitutive models for the sol and interface were validated by backpredicting tests used for finding the parameters and independent tests not used to find the parameters; details are sven by Samtani and Desai (1991). Typical results are described in the following. Fig. 5 shows comparisons between backprediction and test data for the soil under the RTE {o, = 207 kPa (30 psi)] stress path. Fig, 6 shows comparisons for the interface simple shear tests for a, = 103 kPa (15 psi); these results were obtained by using average parameters from tests under the various values of «,. Fig. 7 shows comparisons between backpredictions and test data for the simple shear creep test for two samples preconsolidated at 200 kPa (29 psi) and 400 kPa (58 psi), and tested under the same stress ratio ~ 0.6, These results show that the JOURNAL OF GEOTECHNICAL ENGINEERING 49 giz en Hos| _ . Bos ff Experiment ie Model BE | 0 to sm 0 i900 100 o Preconsolidation Strese = 200 KPa; and (o) Preconsolidation Stress = 400 KF proposed model provides satisfactory characterization ofthe elastoplastic and elastoviscoplastic behavior of the soil and interface FINITE-ELEMENT ANALYSIS ‘The viscoplastic models for the solid body and interfaces were implemented in a finite-element procedure. An existing two-dimensional finite-element program, SSTIN (soil structure inter- faction), developed by Desai and coworkers for elastoplastc analysis was modified for the elas- toviscoplastic analysis. Eight-naded isoparametric elements were used t0 represent the Solid body, while six-noded isoparametric elements Were used to represent interfaces, A'2 * 2 rhumerical integration scheme was used. A brief description of the finite-element viscoplastic algorithm is given next (Owen and Hinton 1980; Desai and Zhang 1987) The viscoplastic strain rate at a given time step n is evaluated from (1c). The viscoplasti strain rate at step n + Lean be expressed by using the Taylor's series, and ignoring the higher- order term as ta) = Had + (Ba) to = te) + tote ay where (dor) = vector of sess increment; and [G}* denotes gradient matrix at stop 1. The ‘iscoplistic strain increment {Bey} during atime interval Arg ows ban be writen as (Begt = aula ~ ones) + ote) «a where 6 can take different values, ¢.g. @ = 0 gives the explicit time-integration scheme. "The incremental stress vector (Aa) can be written as tau) = [e-K(BHAg} ~ (6541,) w, where [B] = strain-displacement transformation matrix; {4g} isthe vector of incremental nodal displacements and [Or] = Ai + (eAoan forte] (4) is referred to as the elastoviscoplastic constitutive matrix. In (14), [C*] = elastic matrix, while [7] = unit matris, "The equations of equilibrium to be satisfied at any instant of time f,, can be written in the incremental form as follows: J terriao av = (ao as) where (4Q"} = incremental vector of applied nodal loads. Use of (15) leads to [arKagrt = (80%) (16) where fe eneiaa c isthe tangential sttness matrix 180% = [ larderteian + or as) 50 JOURNAL OF GEOTECHNICAL ENGINEERING [tercovenan av = [ tris av » represents the body load accumulated during each time step. Once the increments of stresses ‘and strains are computed. the total quantities at time f, can be found as for = for) + Aor aD = teh + gr: fer} = (eh + (eh C0a-0) This process is repeated at each time step until stresses at all Gauss points satisfy F = 0. Since the process is incremental, residual out-of-balance forces are obtained from each time step that fare then added to the applied force increment at the next time step. The time step was selected subject to following empirical criteria (Zienkiewiez and Cormeau 1974) lent S ) a= 0 (Edt, = Os, Qi where J; and 1° = second invariants of the viscoplastic strain and strain-rate tensors, respec- tively; and 02 and @ = specified time-step control constants, The first riterion selects a variable size such that the change of strain occurring during the next time interval isa fraction of the total strain accumulated before. The second criterion imposes a restriction on the variable step size between successive intervals calculated by the first criterion to prevent oscillations in solutions as steady-state conditions are approached. VALIDATION FOR FIELD BEHAVIOR Loading Fig, 8(a) shows the site plan for the Villarbeney Landslide, Switzerland (Laboratoire 1980) ‘The locations for inclinometers , E,, and E;, are also shown on the cross section, Fig. 8(b) To monitor the ground-water level and fluctuations thereof, piezometric cells were installed in borings £,, Es, and &,, Fig. 8(a). The depths of inclinometer were 42.75, 54.35, and 37.5 m for E,, E, and Ey, respectively, It was reported that the inclinometer at E, was destroyed after only one week (Laboratoire 1980), and only those at E, and E provided useful information for the time-depth-displacement data. No significant motions were observed beyond depths of about 17 and 7.5 m for locations E, and E,., respectively. Fig. (a and b) shows recorded measurements from the inclinometer data taken along the steepest descent of the slope. Gravity For the purpose of introducing gravity load, the total unit weight of soil was found approx imately as (Laboratoire 1980) Wo #2 (2) where +, = total unit weight; yp = total unit weight at the surface: z = depth below ground surface; and { = constant. For the boreholes at £y and E, the values of yy and { were found to be 21.85 and 21.6 KNim?, and 0.024 and 0.089 kNim*, respectively. The greater value of {for E; indicates that the creeping mass near E; is older and more consolidated compared to that near F, which explains partially why movements near E, ate greater than those near E, Fie, 9 Seepage Piezometric measurements inthe boreholes showed (Laboratoire 1980) that the ground-water level was at about 2 m below the ground surface, Fig. 8(b), did not vary significantly for the time period of about one year considered herein. This is based on the data from Glotzel cells, ‘which showed essentially a constant water head. Hence, the influence of steady state seepage forces for depths below 2 m were introduced by assuming hydrostatic conditions. Inthe fnite- clement analysis, the seepage forces from a flow net analysis were superimposed on those due to the weight of the soil End Tractions In addition to the weight and Seepage forces, surface tractions on the toe of the chosen sections (presented later) were introduced to simulate resistance to the movement of the soil, Fig. 10 As suggested by Ter-Stepanian and Ter-Stepanian (1988). the tractions were assumed corre= JOURNAL OF GEOTECHNICAL ENGINEERING St Finite-Element Mesh @ Vv mE eed 1! 0” ao0 40060 e860 To00 1200 i800 Horizontal Distance, m FIG. 8. Vilarbeney Natural Slope: (a) Site Plan; () Cross Section of Siope sponding to the natural earth pressure equal to 75% of the limiting Coulomb passive earth pressure Finite-clement analyses were performed separately for both locations, Ey and Es, because of different geometries such as the base slope angle of 14° for Ey and 17° for F, and to reduce the extent of finite-element meshes, Since the predictions from both provided similar compar- isons, results for only E, are presented here. “The direction of flow at the surface is indicated by the arrows on Fig. 8(a). Even if lateral spreading exists it does not have a significant influence at section E1-E2 considered here, which is sufficiently far from the edges of the landslide, Hence, the plane strain condition assumed herein can be considered as a teasonable approximation along this section. Fig. 10 shows the finite-element mesh for location £,. The total length discretized was 150 1m, Fig, 10, Both vertical and horizontal displacements were restrained at the nodes on the base. The left-hand end boundary was free to move, while end tractions, as described earlier, were 52 JOURNAL OF GEOTECHNICAL ENGINEERING Results Depth below Ground Surtace, m Depth below Ground Surface, m ri 18 ala ‘0 102030408060708080 (0 10.20 30 40 60 60 70 Displacement, mm Displacement, mm FIG. 9. Observed Inciinometer Profiles at Boreholes: (a) E,; and (b) Es sont Elenents (B-node) Location of Borehole El Elements: (6-node) FIG. 10. Finte-Element Mesh for Borehole E, applied at the right-hand end boundary, as shown in Fig. 10. The initial conditions involved introduction of the in situ state of stress assuming normal consolidation including effect of hydrostatic pressure, zero strains, and zero strain rates, “The mesh in Fig. 10 consists of 60 eight-noded elements with 202 nodes including 12 interface elements. The thickness of the interface element was adopted to be 1.0 m, which is about 6% of the thickness ofthe slope of about 17 m. Since higher-order (quadratic) finite elements were used for soil and interface elements, it was possible to adopt a relatively coarse mesh. Note that With the quadratic element, the stresses and strains vary (linearly) within the elements. Also, it is possible to include more than one element to represent the interface zone. It was found that further refinement of the mesh did not yield significant improvements in the computed results, The vatiable time-stepping scheme was used for the viscoplastic solution with values of 2 = 0.02 and © = 1.2, (21). The viscoplastic solution was continued until the viscoplastic strain rate reached a steady condition at the integration points in the elements: this was id when the viscoplastic strain rate was small (= 10-"/min), Fig. 11 shows typical finite elements results after a period of 354 days in comparison with the field observations for inclinometers E,. Here, the velocity was obtained by dividing the difference in displacements by the corresponding time increment when the steady viscoplastic strain rate was reached. It can be seen that the finite-element predictions compare very well with the observed velocity and displacement profiles, JOURNAL OF GEOTECHNICAL ENGINEERING 53 2 aiete a ron e $ hs gi g 26 manatee a a Ez 8 z 5 3 3 3 6 10 o FS - & a) /f z Sal 7 3" ff g é 8 ey 13 | 0.0001 0.0002 5E-05 0.00015 0 15 30 45 60 75 90 Velocity, mm/min i) Fie. 11 CONCLUSIONS ‘Comparisons of Predictions and Observations at Borehole E,: Displacement, mm % ) Velocity; (b) Displacement A general formulation for modeling creeping natural slopes is proposed, It integrates appro- priate constitutive modeling of geological materials and interfaces, gravity, steady-state seepage, fend tractions, laboratory sind field tests, and computational m jods. New procedures t0 de- termine viscous constitutive parameters and for interface modeling, 1e developed. ‘Al his time, the constitutive and the finite-element models have certain limtations: (1) The tive model can allow for primary and secondary creep and does not allow for tertiary (2) the effect of fluid pressures is superimposed on the mechanical response, and the ful ‘coupling between the fluid and hanical effects is not included, (3) it does not allow for -aastrophic landslides” with precursory creep movements; (4) the two-dimensional idealization is valid essentially forthe central region, andi three-dimensional effects such as late are included, & three-dimensional procedure would be warranted; and (5) field v possible for only two locations for Which reliable data were available, and it woul to perform additional verifications. It may, however be noted that reliable field fare not readily available in most locations, ral spreading, idations were bbe desirable observations Notwithstanding the preceding limitations, because of the improvements provided by the constitutive model by inclusion of some of the icant factors and by satisfactory field val dation, it may be concluded thatthe proposed procedure ean provide an alternativ ind improved ‘means for prediction of velocities and displacements in creeping slopes that can be idealized as ‘wordimensional. It can also allow for inclusion of factors such as inbomoges ‘geometries, and existence of geotechnical structures. ‘ACKNOWLEDGMENTS, APPENDIX |. ies, different Pats of the research herein were supported hy grant No. MSM SSIRBOL/S fom the U.S. National Science Foundation, Washngion, D.C snd Grant No, 8240.0 87 from the Ss National Science Foundation daring the stay of L, Valet tthe Uniserty of Arona The provision of cay wom Villarbeney by E ‘Revordom, Som Mechanics Laburitory, Suis Federal Insitute of Technology. Lausanne, Swilzeriand ts appredated. Useful Sissons wih KG Sharma of fodian litte of Technology. Now Delhi, India. are gratefully PARAMETERS FOR CONSTITUTIVE MODELS This appendix and Table 2 include values of various parameters in the constitut {or the soil and interface, Methods for fin behavior ate given elsewhere (Desai etal ng the parameters related to the elas 1986; Desai 1990; Desai and Fishman I JOURNAL OF GEOTECHNICAL ENGINEERING models ic and plastic 1991). A brief FF inlFF) 9) FIG. 12. Viscous Parameters: ) FI, versus x; (b) InP) vorsus ix) description ofthe method for finding viscous parameters, which was developed under the present study, is given below. Determination of Viscous Parameters I’ and N In the application of Perzyna's viscoplastcity concept, the viscous parameters [and N are often adopted arbitrarily (Zienkiewicz and Cormeau 1974), One of the contributions of the present study is to develop a procedure for finding these parameters based on laboratory creep tests, which is described in the following, ‘Based on the expression forthe viscoplastic strain rate increments, (1c). the following expres- sion is obtained! 1 epep 122 20 pares Or ee @ where i i the second invariant of #. Then Tp re=x= few 4 2 80, 30 Hence, substitution for () from (2) leads to (fe o which on taking the logarithm gives vin (E wr evn (E) = tnx 06) For the soil (Q = F), and the values of FIR, and x are computed from laboratory tests at different points, and a plot of In(F/F,) versus In(x) is obtained; a schematic is shown in Fig. 12 Here, F,/p, = 1, where p, = 100 kPa, is used. The intercept along. ln(x), say ¢. gives the value ‘of F (="exp c) and the slope of the average straight line gives the value of N. ‘The parameters F and NV, are found similarly by using (9a) and (95), APPENDIX Il, REFERENCES Brown, C.B., and Shew, M. S. (1975). “Eifet of deforestation om slopes." J. Geotch., Engrg. Di. ASCE, 012), 147168, Davis, R. O., Desai, CS. and Smith, N, R, (1993). “Stability motions of tansational ses." J. of Geotech ‘Bngrg, ASCE, 119(3) 21-482 Desa C$ (198i). "Behavior of interfaces between sructural and goologic radia” Int, Conf. Recent Adv. in “Geowech. Earthquake Engr. and Soll Dy. St. Lous, Mo. DDesat,C (199). "Modcing ad testing: implementation of nmerial modes and their aplication in practice.” ‘Numerial methods and consaive modeling in peomechones, C8. Desi and. Goda, ed, Springer Vera, ‘Viena, Austria, Desi C'S. and Fishman, KL. (19), “Plesticny-based constitutive model with associated testing for joins Tt 1 Rock Mech. and ining Set 28(), 15-26, DDesai.C.$. Somasundaram, 5. and Frantaskons,G. (1986). “A hierarchical approach for constitutive moving ‘of geological materia. dt. J. Num Analycal Meth. in Gzomech 10, 28-287 Dest C. 8. Zaman, M. M., Lightner, ©. and Sirtvardane, HJ. (988), "Thin-layer element for interfaces ‘and joins" Ines. Num. Avafyacal Meh in Geomech {1 19-88. Desii,C_ 8. and Zhang, D. (1987) "Viscopastie model for geologic materials with generalize flow rule,” Int “1 Num. dnalycal Meth. in Geomech 1 03-620. JOURNAL OF GEOTECHNICAL ENGINEERING 55 Dy. M..and Resondon, E. (1968) “Landis analyzed ax vicous flow." Proc, le nt. Symp. om Landslides, "Toronto. Canada, 05~410, Goa P1960). "A study of shear flue in certain tertiary matine semen.” Proc, ASCE Res. Conf ‘on Shear Srongih of Cohesive So ASCE, New York, NY. 618-6 Iverson, RM (9864). "Unsteady nonuniform landslide motion: I. ther Slate" J. of Geolog)- Te I= 18, Iverson, &. M. (1986), "Unsteady, nonuniform lantide motion: 1, tincaized theory ad the Kinematics of Warsent response". of Geology 8 1. M9364 Katona M. G-rud Mulert M.A. (1984) "A viscoplastic cap model or soils and tock." Mecham of engineering ‘materials, CS. Des ind R. 11 Gallagher eg Jon Wiley and Sons. 835-31, awamura, K-(1ORS). "Methodology for lands pedition.* Pro. th Dat Com om Sol Mech and Faun ‘Ener. Wo, 3, International Sosely of Sit Mechanics und Foundation Engineering, San Frandsco, Ca. Tse i1S8 [LMS (Lat do, Mec, de, Sol). (1980), “The Villarbeneylnstide —2nintemte report.” Rep. No. GXSY. PEL, Lausanne. Statzerland Lu, K-Mand Wang. LS. (988). “On the sidng-bending model of rock mass deformation and alae of ‘iope.” Pro. Sh nz. Symp. on Landslides, Lawns, Switactand, 219-223 ‘owen D. Ra Hinton, E (19M) Fite Element Pasty: Theory and Practice, Pieridpe Press, Swunse. vk erayns, P. (966), “Fundomental problems in viscopncty.” Adv. in Appl. Mech... 248-377. Sato. Ni. (1968). “Forecusting the time of occurence of 2 slope flr” Proc. 6 fn. Conon Soil Mec ‘nd Found Engr. Inertial Society of Sol Mechaties and Foundation Engineering. $37-Sth Samlan, N.C. and Desa. C.8. (1991). “Contittive madeline nd nite element analy of sky moving Tandsldes using the Bcrvchialiscopiastc materi mode.” Rep. to Nationa Science Foundation. Dept a Chil Engrg and Engrg. Mech . University of Atzona. Tucson, Avia ‘Savage, W. Zand Chleborad, A. F-(382),""A mod for erecping Bow in landslides” Bl, Asso. Enere ‘Grologvs. XIN 4, 333-338 ‘Ter Stepanian, G {I843). "On the longterm stability of pes.” Publ No. $2. Norwegian Gootech. tat. . Oso ‘Norway. ‘TerStopinian. G., nd Ter Stepan, H, (ISKH). “Back analyses for determining the lanesile press on ‘midge abutinen” Pro. Sh lar. Sip. on Landldes, Lavsanne-Suitstand, 81 33h ‘Valet L- (1985) "Modkisation des pentes naturelles en mouvement” Thise No. 35. Eoole Polytechnique dae de Lausanne, Lansanne, Swzertan in Breach), ‘Vatlet Land Desai. C. S.( 1989). "Viscoplanticty ad inte elements fr lanstie analysis.” Pro. 0 tt Conf on So Mech and Found. Engrg Intrnaional Soity of Soll Mcchmiy and Foundation Enc Riv de hincio, rz Valet, Ld Tater K, (198), “Set of constittve models for slid under slow’ movements." J. of Gre Engrg” ASCE. 1189} 1022~ 1 ‘Voller son Futcr.K. (1988), “Continuum modes for natura slopes in slow movements.” Catechnignr Lomo, UK. SRO) 1217 ‘Yen. B.C. (1963), "Stability of slopes undergoing ercep deformation." J. Soil Mech. and Found. Di. ASCE. 9S(4), 181086 Zienkicwicr. 0. C.and Cormeau. LC. (1974). “Viscopastcty- plasticity and erep in laste sls nied ‘umericl slation approach uJ. Nan Meu gr. 8, 93-848. ical dys and the tay datum 56 JOURNAL OF GEOTECHNICAL ENGINEERING

Вам также может понравиться