Vous êtes sur la page 1sur 9

198

A practical design approach for reducing the occurrence of shallow slides in cohesionless till highway embankments
J.-M. Konrad and G. Grondin

Abstract: Shallow slides in cohesionless till occurred in several highway fills in the province of Quebec, Canada, following spring thaw and sometimes during rainy autumns. This study has shown that the strength parameters of cohesionless tills can be related to the liquid limit for both drained and undrained conditions. These relationships can be used as input in stability analyses to readily yield design values of stable slopes. Tills with liquid limits between 15% and 25% are all very susceptible to liquefaction flow failures if loosely compacted and saturated. To reduce the risk of shallow slides in the cohesionless till embankments used for highways, a stabilizing coarse-grained fill placed atop of the till embankment appears to be an efficient technique. Design charts of minimum thickness of the stabilizing fill at the base of the till slope were developed using available stability programs and an empirically derived strength of liquefied tills. Key words: cohesionless, till, liquefaction, stability, slope, skin. Rsum : Des instabilits superficielles des talus routiers en till pulvrulent ont t observes au Qubec, Canada, lors du dgel printanier et quelques fois durant des automnes pluvieux. Cette recherche a dmontr que les paramtres de rsistance draine et non draine des tills pulvrulents pouvaient tre relis la limite de liquidit. Ces relations peuvent tre utilizes pour tablir linclinaison de pentes stables. Des tills ayant des limites de liquidit entre 15 % et 25 % compacts dans un tat lche et saturs sont trs susceptibles de se liqufier. Afin de rduire le risque de coule de peau dans les talus routier en till, le tapis drainant plac par-dessus le remblai est une technique efficace. Des abaques de dimensionnement ont t dvelopps pour dterminer la largeur minimale du tapis drainant la base du talus en fonction de la rsistance au cisaillement du till liqufi. Mots cls : pulvrulent, till, liqufaction, stabilit, pente, peau. Konrad and Grondin 206

Introduction
Shallow slides in till occurred in several highway fills in the province of Quebec, Canada, following spring thaw and displayed typical flow slide features resulting from static liquefaction (Konrad et al. 2000). Poor compaction control near the embankment slopes owing to the lack of confinement and to the fact that heavy compaction equipment usually do not approach too close from the edge of the slope is mainly responsible for these shallow slides. Liquefaction failures were also reported in Scandinavian tills, particularly in Asele (Ekstrm and Olofsson 1985) where loose compaction was a result of winter placement. This paper presents the results of a comprehensive study aimed at the development of a design methodology to minimize the risk of shallow slides in till embankments. After a
Received 4 November 2003. Accepted 3 September 2004. Published on the NRC Research Press Web site at http://cgj.nrc.ca on 1 March 2005. J.-M. Konrad.1 Department of Civil Engineering, Universit Laval, Sainte-Foy, QC G1K 7P4, Canada. G. Grondin. Ministre des Transport, 930 chemin Sainte-Foy, Qubec, G1S 4X9, Canada.
1

brief review of the behaviour of saturated cohesionless soils subjected to undrained loading, the paper presents the conceptual mechanisms involved in shallow slides using critical state soil mechanics principles. The results of an extensive laboratory study are given for various tills from Quebec and were used to establish empirical relationships between the strength parameters of the till and index soil properties. These relationships, in turn, constitute key input parameters for the design of the slope angle of till highway embankments to minimize the risk of shallow sliding as the upper layers become progressively saturated during the infiltration of water.

Liquefaction in saturated cohesionless soils


As stated by Poulos (1988), any given soil element in situ subjected to a particular loading condition has two welldefined strengths, the peak strength, tp, and the steady state strength, tF. The peak strength is the maximum shear stress in the soil element equal to one-half of the maximum deviator stress, qp. The steady state strength is defined as the strength that exists after the shear loading has eliminated all effects of the original structure of the element and after any orientation of grains that might affect strength has occurred. At the steady state of deformation, soil is being sheared at
2005 NRC Canada

Corresponding author (e-mail: jmkonrad@gci.ulaval.ca).

Can. Geotech. J. 42: 198206 (2005)

doi: 10.1139/T04-095

Konrad and Grondin

199 Fig. 1. State and stress conditions in a dense till slope.

constant volume, constant effective stress, and constant strain rate (Poulos 1981). The peak strength and the peak strength envelope depend on the soil composition, the structure of the sample or fabric anisotropy, the void ratio and effective confining stress (often referred to as state), rate of dilatancy and loading method (Poulos 1981, 1988; Bolton 1986). In contrast, Poulos (1981) states that the slope of the steady state strength envelope, , is solely a function of the soil composition. Similar views were developed by Bolton (1986), who states that the critical state angle of shearing resistance of soil. which is shearing at constant volume. is principally a function of mineralogy. It was also well-established that in order for liquefaction flow failure to occur, the soil must be contractive when subjected to undrained shear and, moreover, the driving shear stress must be larger than the steady state strength of the soil. A number of researchers (Vasquez-Herrera et al. 1988; Negussey et al. 1988; Konrad 1993) showed that triggering of flow failure in monotonic and cyclic undrained tests occurred when the effective stress path crossed the peak strength envelope, which is a straight line passing through the origin and with a slope referred to as pu . Work by Doanh et al. (1999) has shown that pu was essentially a function of soil fabric and stress anisotropy.

Conceptual failure mechanism for shallow slides


Till in dense states The fill for embankment highways is usually specified to be compacted at least 95% of the maximum dry density obtainable according to standard test methods. These specifications provide adequate strength and compressibility characteristics in tills and often result in relatively dense compaction states associated with a dilative response during undrained shearing, and thus generally, are not prone to develop liquefaction flow failures. This is well-illustrated in Fig. 1, which is a 3D representation of the state and stress conditions in a soil element located in an infinite slope at a depth z. Initially, the soil is unsaturated, and the water table is located at a depth greater than the soil element. Rather than using void ratio to describe the physical state of the soil element, it is more practical to use water content in unsaturated and compacted soils since the latter can vary substantially, while void ratio essentially remains constant for these dense tills. For dense compaction states, the soil element in an infinite slope at depth z is represented by point A in Fig. 1: its water content is equal to the water content at optimum Proctor, wopt, and its stress conditions can be approximated by eq. [1] assuming that the unit weight is constant: [1] n = [ z (ua uw)] cos 2 and = n tan line whose inclination is . At point B, the soil element is fully saturated, its water content now becomes equal to wsat, and, since its state is below the critical state line (CSL), the response of the soil element during undrained loading is dilative, the soil element cannot thus liquefy. If the water table progressively raises above the soil element, the effective stress path is a horizontal line passing through point B. Ultimately failure is reached at point C when the water table is equal to: [2] hw = ( nB nC)/ w cos 2

At failure, slope movement is not dramatic since the soil behaviour is dilative during undrained shear. Often, water table fluctuations induce creep movements of these dense till slopes. This is not the case when the till is placed in a loose state. Till in loose states Highways embankments are usually placed in lifts that are successively compacted using heavy vibratory rollers to achieve the target density. However, this heavy compaction equipment does not approach too close from the edge of the embankment sloped face, which in turn, often results in poorly compacted zones near the slope face. For these poorly compacted layers, undrained collapse can be triggered only if the soil in the vicinity of the potential failure surface (usually at the interface between wel- compacted and poorly compated layers) is at a compaction state above the critical state and becomes saturated as a result of snow melt infiltration or sustained rainy periods.
2005 NRC Canada

where n is the effective stress normal to the soil element, is the shear stress, z is the depth, is the unit weight of the soil, is the slope angle, and (ua uw) is the matric suction. With increasing water content in the soil element as a result of water infiltration, the suction decreases and the effective stress path moves towards the stress space origin along a

200 Fig. 2. Poorly compacted till for the case > . pu

Can. Geotech. J. Vol. 42, 2005 Fig. 3. Poorly compacted till for the case < . pu

The analysis of a failure by undrained collapse for the simplified case of an infinite slope must therefore consider the relationship between shear stress, q, mean principal effective stress, p, and void ratio. For the particular case of skin flows in saturated till initially placed in unsaturated conditions, it is convenient to consider the relationship between shear stress, , normal effective stress, n and water content, w, for a soil element parallel to the slope as indicated in Fig. 2. Two cases must be analysed: (i) the slope angle, , is greater than the slope of the trigger surface, pu ; and (ii) < pu . Case 1: > pu In Fig. 2, point I represents the state and stress conditions of a soil element at the potential failure surface parallel to the face of a till embankment with a slope . Water infiltration causes a progressive increase in the degree of saturation, Sr , and the representative point moves from I to i as the effective stresses decrease since the suction decreases with increasing values of Sr . With a further increase in Sr , the soil element reaches point j, which is located just above CSL, i.e., in the contractive domain. However, liquefaction flow failure cannot occur since the soil element is not yet sufficiently close to fully saturated conditions. At point J, the soil element is fully saturated. In the stress space (, n ), the slope of line OJI is equal to and in the state space (w, n ), point J is located above CSL, which indicates that the conditions for liquefaction flow failure are fully met. Since > pu , liquefaction is triggered as soon as the soil element becomes fully saturated or near it. During undrained flow

failure, the water content remains constant, and the value of the induced pore pressure is given by the value of segment JK. In the stress diagram, the strength of the liquefied till, (K), corresponds thus to the failure envelope for the effective confining stress n (K). Case 2: < pu In Fig. 3, point M represents the initial state and stress conditions of a soil element at the potential failure surface. As the degree of saturation increases with infiltration, the stress and state conditions move progressively towards point N. At point N, the soil element is fully saturated, but since < pu , the effective stress path has not crossed the trigger line and liquefaction cannot occur. At point N, the water table is at the potential failure surface, i.e., at the interface between the denser till and poorly compacted till. With further infiltration, the water table rises progressively and the effective stress path in the (, n ) space is a horizontal line pass ing through point N. When the effective stress path crosses the trigger line, i.e., at point O, liquefaction is possible. The pore pressure required to reach point O can be related to the height of the water table above the potential failure surface, hw and as: [3] u = whw cos 2 Once liquefaction is triggered, the state conditions of the liquefied till are given by point P in the state diagram (w, n ), and the corresponding strength (P) is obtained from
2005 NRC Canada

Konrad and Grondin Table 1. Grain size characteristics of the till studied. Localization Tmiscaming 1 Tmiscaming 2 Saint-Ren-de-Matane St-Georges 1 St-Georges 2 St-Martin 1 St-Martin 2 Beauceville Weedon Sherbrooke 1 Sherbrooke 2 Fig. 4. Grain size distribution of till studied. d60 (mm) 0.18 0.29 0.065 0.16 0.07 0.0205 0.031 0.11 0.88 0.09 0.18 d50 (mm) 0.11 0.19 0.022 0.075 0.04 0.011 0.015 0.068 0.405 0.062 0.12 d10 (mm) 0.09 0.04 0.00045 0.0025 0.0025 0.001 0.0013 0.0025 0.008 0.0047 0.0065 % <80 m 43 31 64 52 62 87 72 56 36 58 40

201

Table 2. Index properties of the till studied. Sites Tmiscaming 1 Tmiscaming 2 Saint-Ren-deMatane St-Georges 1 St-Georges 2 St-Martin 1 St-Martin 2 Beauceville Weedon Sherbrooke 1 Sherbrooke 2 wL (%) 18 16 33 19 20.5 30 23 25 20 15.5 s (t/m3) 2.70 2.70 2.74 2.71 2.73 2.75 2.74 2.69 2.72 2.72 2.71 % <80 m 43 31 64 52 62 87 72 56 36 58 40 Ss (m2/g) 3.7 3.7 36.8 8.9 18.5 16.5 4.4

the failure envelope in the stress space (, n ). The undrained excess pore pressure during flow liquefaction is equal to the value of segment OP.

may not be sufficient to classify adequately the till with respect to its liquefaction potential. Table 2 summarizes some of the index properties of the till studied, and highlights the tills used for triaxial testing. The liquid limit was determined using the fall cone test, according to the BNQ2501-092 standard (CAN/BNQ 1986), on the fine fraction smaller than 400 m. The density of solid particles, s, was determined according to the BNQ 2501-070 standard (CAN/BNQ 1986). The specific surface area, Ss, of the fraction passing through the No. 200 sieve (80 m) was obtained using the methylene blue technique. Critical state characteristics from triaxial tests Unlike uniform sands, till has lower hydraulic conductivities, which leads to more difficulties in preparing a fully saturated sample for triaxial testing. In the field, the till near the embankment face is placed in horizontal layers at or near optimum water content, but is lightly compacted for the reasons discussed previously. The in situ mean effective stress ranges between 15 and 20 kPa if the denseloose interface is located at a depth of about 1 m. Critical state characteristics should therefore be determined for these stress conditions using lightly compacted samples at different initial void ratios. These low stresses, in turn, require precise instrumentation for adequate monitoring of pore pressure and deviatoric stress during monotonic loading. The samples were prepared by the moist tamping method to study representative field conditions. This, however, often leads to sudden collapse during saturation under water per 2005 NRC Canada

Properties of till studied


The analyses described above, however simplified, have shown that undrained collapse in loosely compacted tills depends on pu and the ultimate or critical state characteristics of the saturated till. Both characteristics can be obtained from triaxial laboratory tests. Till was collected from 10 different sites from four geographic regions in Quebec (Abitibi, Gaspsie, Beauce, and Estrie). Index properties were obtained for all soils and strength characteristics from triaxial tests were obtained for four different tills, where slope instabilities were observed. Index properties The grain size distribution for each till is presented in Fig. 4 and details are given in Table 1. Till, defined as a well-graded soil, may be characterized by a wide range of fines content, indicating that grain size distribution alone

202 Fig. 5. CU triaxial test results for Weedon till.

Can. Geotech. J. Vol. 42, 2005 Table 3. Strength parameters for four tills. Till Tmiscaming St-Martin Sherbrooke Weedon RTE 175 Mf 1.65 1.35 1.5 1.6 1.67 () 40.3 33.4 36.9 39.2 40.8 Mp 0.6 0.8 0.63 0.81 0.7 pu () 15.8 20.7 16.6 20.9 18.3

Fig. 6. Critical state characteristics of tills studied.

colation. During saturation with increasing backpressures, sufficient time must be allowed to reach a uniform water content distribution within the sample. Generally, 1 to 2 days are required to obtain satisfactory conditions. Once saturated, rate of shearing must be adapted to the samples swelling characteristics to ensure uniform pore pressure distribution. At the end of testing, a water content of the entire sample is determined and compared to the calculated changes in void ratio during the different steps of the testing (saturation by percolation, backpressure saturation, consolidation). Figure 5 presents typical CU test data obtained for till from Weedon tested for different void ratio and stress conditions. The failure envelope in the Cambridge plot, i.e., p = (1 + 2 3 )/ 3 and q = 1 3, has a slope of 1.6, and a peak strength envelope of 0.81. In a (, n ) stress space, the fail ure envelope is thus = 39.2, and the peak strength envelope pu = 20.9. Table 3 presents the results from the triaxial tests in terms of strength parameters, while Fig. 6 summarizes the critical state characteristics of the till studied in a conventional void ratio mean effective stress plot. Shear strength of till from empirical relationships The design of stable slopes in till requires as an input parameter the shear strength along the potential slip surface. The mobilization of shear can occur in drained or undrained conditions. Triaxial testing is a common approach to determine these strengths, but may not be practical nor economical for preliminary studies to assess the various design options. The use of empirical relationships between the strength parameters and more readily obtained index properties of tills is an interesting and practical alternative. Drained conditions For drained conditions, triaxial tests on saturated samples have shown that all the till studied behave as cohesionless soils, i.e., with an effective cohesion equal to 0. The drained strength is thus simply related to the effective friction angle, . For the till studied, the best correlation between and index soil properties was obtained with the liquid limit of the fraction smaller than 400 m, wL. As shown in Fig. 7, this relationship can be expressed as: [4] = 66 1.4wL

Fig. 7. Empirical relationships for strength parameter of tills.

with R2 = 0.95 and 15% < wL < 25%.

Undrained conditions For undrained conditions, two cases must be considered: dilative and contractive behaviour during shear. For the problem at hand, the important aspect is related to contractive behaviour of poorly compacted tills near the slope face becoming saturated with water infiltration. For flow sliding to occur, the stress path must cross the trigger envelope, pu . Since the liquid limit appears to correlate well with the friction angle, it should also be an acceptable indicator for the peak undrained strength as indicated by Fig. 7, and the best fit equation given below:
2005 NRC Canada

Konrad and Grondin Table 4. Water content values at given undrained strength of liquefied tills. Till Tmiscaming St-Martin Sherbrooke Weedon Water content % for Cu = 3 kPa 17.3 24.5 21.8 24.5 Water content % for Cu = 5 kPa 16.9 22.0 18.6 22.5 Water content % for Cu = 6 kPa 16.5 21.5 17.5 21.4 Water content % for Cu = 8 kPa 16.0 20.2 17.0 20.0

203

[5]

pu = 0.72wL + 3.4

Fig. 8. Characteristics of Weedon till in a (w, , n ) space.

with R2 = 0.82 and 15% < wL < 25%. The undrained shear strength, Cu, mobilized along the slip surface depends on the type of till and its void ratio. For Weedon till, the test results shown in Figs. 5 and 6 can be combined to produce a complete state and stress diagram in an arithmetic scale as illustrated by Fig. 8. The undrained shear strength of liquefied Weedon till placed at a void ratio of 0.54 (i.e., a water content in a saturated state of 20%) would be 8.4 kPa. Furthermore, the same diagram also gives the minimum depth of the failure surface to ensure that the driving shear stress is at least equal to the liquefied shear strength. For instance, if the till embankment has a slope 2H:1V, the vertical effective stress at the failure surface must be equal or greater than 17 kPa (i.e., a minimum depth of about 1.2 m considering that z = n / cos 2 ) and for a slope 1.5H:1V, the stress at the failure surface must be greater than 12.5 kPa, i.e., a minimum depth of about 0.85 m. Figure 8 also illustrates that Weedon till compacted to 95% of its Proctor value, i.e., a void ratio of 0.32 (wsat = 12%) will display a dilative response during undrained shear, and thus never experience liquefaction flow failures. For the case of shallow slides, it is convenient to develop a relationship between Cu, wL, and wsat. Figure 8 can be used to determine, for a given soil, wsat values for different values of undrained shear. Path 2 in Fig. 8 illustrates clearly that, for Weedon till, a shear strength value of 5 kPa is directly associated with a water content of a saturated sample of 22.5%. As anticipated, for a given soil, the undrained strength increases with decreasing water content of the saturated till (i.e., decreasing void ratios) (see Table 4). As shown above, the effect of soil type is well-reflected by its liquid limit. Figure 9 indicates that, for a given water content, the liquefied strength increases with increasing values of wL. The relationships presented in Fig. 9 can be used to readily estimate the undrained shear strength of liquefied till given its liquid limit and its in situ compaction state, usually given in terms of dry unit weight, d. Obviously, fully saturated conditions are a prerequisite for liquefaction, and wsat can be calculated from d as: [6] 1 1 w sat = w d s w

safety as a function of various soil properties and environmental conditions. Design charts For the sake of simplicity, it is considered that the slip surface is approximately parallel to the ground surface (infinite slope procedure), and that the safety factor is given by: [7] F= (h whw) cos 2 tan h sin cos

A practical design approach


The empirical relationships established above can be used to obtain the required input parameters in stability analyses to readily yield design values of stable slopes. Design charts considering that the embankment face is approximated by an infinite slope provide a rapid means to assess the factor of

in which h is the depth of the slip surface, is the soil unit weight, hw is the elevation of the water table above the slip surface, is the bedding of the slip surface, and is the friction angle of the soil. Figure 10 presents the relationship between slope angle, liquid limit and safety factor for drained conditions for a failure surface and a water table located a depth of 1 m. These charts can be used to determine the slope of dense till embankments. For different conditions, eq. [7] readily gives the value of the safety factor. In the poorly compacted zone near the face, it can be argued that the slope should be less than pu to prevent the onset of liquefaction. This, in turn, signifies that extremely flat slopes must be constructed as shown on Fig. 10 (open squares). For instance, for a till that has a liquid limit of 15%, the minimum slope inclination should be 4H:1V, and
2005 NRC Canada

204 Fig. 9. Undrained shear strength of liquefied till.

Can. Geotech. J. Vol. 42, 2005

Fig. 10. Design charts for till slopes used for highway embankments.

characterized by a liquid limit equal to 17%, thus with a stable slope of 3.5H:1V to prevent liquefaction flow failure, while a slope of 1.35H:1V would display a safety factor of 1.2 for drained conditions. Design of stabilization fill It is well-known that moist tills can be placed at relatively loose states. The study presented above clearly indicates that for those loose conditions, most of the tills (15% < wL < 25%) will be very susceptible to liquefaction when progressively saturated. To reduce the susceptibility of liquefaction, several remedial measures can be considered: Improved compaction near the face of the slope (costly and fairly difficult). Construction of a wider crest with subsequent cut to final design width (costly, time consuming, not always feasible if right-away too narrow). Slope stabilization by vegetation (often skin failures occur in the first spring before vegetation roots have grown sufficiently deep). Geosynthetics (costly). Control of drainage, i.e., directing surficial water to armoured channels. Coarse-grained stabilization fill. This latter remedial measure is often used by the Quebec Ministry of Transportation. It has several advantages because it acts as a draining layer and reduces water infiltration, especially at the level of the loosedense interface parallel to the till slope. It also acts as a mechanical stabilization since the shear strength mobilized in the coarse-grained fill is the drained strength characterized by its effective friction angle. When placed immediately during construction, it is the most cost-effective solution. Stability calculations with SLOPE/W were conducted to establish the minimum thickness for the granular pad to ensure slope stability once the till has liquefied along the loosedense interface parallel to the till slope. The minimum thickness is associated with the parameter L shown in
2005 NRC Canada

for wL = 25%, it becomes 2.5H:1V. These values are too low to be of practical interest. Figure 10 is also indicative of the liquefaction susceptibility of the various tills if placed loosely, and which become saturated with time. The difference between the slope angle determined for drained and undrained conditions is directly related to the loss of strength during liquefaction flow failure. It appears thus that tills with low liquid limit values are more susceptible to develop significant flows than tills with higher liquid limits. This is confirmed by the failure observed in RTE 175 till (Konrad and al. 2000) where liquefied till flowed over a distance of 60 m. The till at RTE 175 was

Konrad and Grondin Fig. 11. Design chart for a stabilizing fill.

205

The stability analyses also showed that the width of the stabilizing fill was a function of slope inclination of the till embankment, the position of both the water table and the loosedense interface. Light-weight dynamic penetrometer testing may prove to be a useful tool to obtain adequate data on the position of the loosedense interface near the slope face. Density determinations in the loose zone are also of utmost importance to estimate the undrained shear strength of the saturated till.

Conclusions
Flow slides in cohesionless soils can be related to an undrained collapse phenomenon that depends upon many factors such as the initial void ratio, the effective stress path during loading, the initial stress state, and obviously the type of soil. Only soils that tend to decrease in volume during shear, i.e., contractive behaviour, can suffer the loss of shearing resistance that results in a flow slide when the driving shear stress is larger than the ultimate undrained strength of the saturated soil. Loosely compacted tills may be extremely sensitive to undrained collapse depending on their critical state characteristics. These conditions exist near the face of highway embankments owing to difficulties of adequate compaction near the slope edge, especially with increasing slope heights. Once the loosely placed tills become saturated as a result of infiltration in the spring or fall, undrained collapse and liquefaction flow failures may be triggered. This study has shown that the strength parameters of cohesionless tills can be related to the liquid limit for both drained and undrained conditions. These relationships can be used to readily provide the necessary input parameters in stability analyses and yield design values of stable slopes. Tills with liquid limits between 15% and 25% are all very susceptible to liquefaction flow failures if loosely compacted and saturated. To reduce the risk of shallow slides in the till embankments used for highways, a stabilizing coarsegrained fill placed atop of the till embankment appears to be an efficient technique. Design charts of the minimum thickness of the stabilizing fill at the base of the till slope considering the empirical relationships for the strength of liquefied tills were developed using available slope stability programs.

Fig. 11. The conditions of the analysis are well-summarized by the insert in Fig. 11. The geometry considered herein comprises a till slope at an inclination of 2H:1V in which the loose-dense interface is located at a depth of 1 m. The height of the water table above this interface is considered to be 0.3 m. The granular fill placed above the till slope is defined by its minimum length, L, at the base of the till slope whose height is H. The stability analyses were conducted for two potential failure surfaces, referred to as surface S1 and S2. S1 is parallel to the till slope, coinciding with the actual loosedense interface with the interface between the stabilizing fill and the till. S2 is considered as a more realistic failure surface since it assumes that it exits through the stabilizing fill near the base of the till slope. S2 corresponds more closely to actual failure surfaces observed in the field for till embankment slopes. The strength parameters in the stabilizing fill were taken as c = 0 and = 45. The undrained strength of the liquefied till was taken as 5 kPa. It is emphasized that the actual value will be related to the liquid limit and the actual density in the loose zone as illustrated by the relationships presented in Fig. 9. The results of the stability analysis show that the minimum width, L, of the stabilizing fill is related to the slope height, H, and the geometry of the failure surface (Fig. 11). For the realistic S2 failure surface, a stabilizing fill is required when the height of the till embankment is larger than 4 m. For a height of 10 m, the width of the stabilizing fill is close to 5 m, for a strength of the liquefied till equal to 5 kPa.

Acknowledgements
This research was supported by the Quebec Ministry of Transportation (MTQ). The technical assistance of Franois Gilbert, Jean Ct, Olivier Turcotte, and Isabelle Leclerc is gratefully acknowledged.

References
Bolton, M.D. 1986. The strength and dilatancy of sands. Gotechnique, 36: 6578. CAN/BNQ. 1986. Soils Determination of liquid limit by the Swedish fall cone penetrometer method and determination of plastic limit. Canadian Standards Association and Bureau de normalisation du Qubec. CAN/BNQ 2501-092-M-86. Doanh, T., Ibraim, E., Dubujet, P., Mariotti, R., and Herle, I. 1999. Static liquefaction of very loose Hostun sand: Experiments and modelling. In Proceedings of the International Workshop on the
2005 NRC Canada

206 Physics and Mechanics of Soil Liquefaction, Baltimore, Md. 1011 September 1998. Edited by P.V. Lade and J.A. Yamamuro. pp. 1728. Ekstrm, A., and Oloffson, T. 1985. Water and frost-stability risks for embankments of fine grained soils. In Proceedings of the Symposium on Failures in Earthworks., I.C.E. London, UK. Thomas Telford, London. pp. 155166. Konrad, J.-M. 1993. Undrained response of loosely compacted sands during monotonic and cyclic compression tests. Gotechnique, 43: 6989. Konrad, J.-M., Grondin, G., and Saim, Y. 2000. Conditions for flow slides in compacted till highway embankments. In Proceedings of the 8th International Symposium on Landslides, Cardiff, Wales.

Can. Geotech. J. Vol. 42, 2005 Negussey, D., Wijewickreme, W.K.D., and Vaid, Y.P. 1988. Constant-volume friction angle of granular materials. Canadian Geotechnical Journal, 25: 5055. Poulos, S.J. 1981. The steady state of deformation. Journal of the Geotechnical Engineering Division, ASCE, 107(GT5): 553562. Poulos, S.J. 1988. Strength for static and dynamic stability analysis. In Proceedings of the Conference on Hydraulic Fill Structures, ASCE, Fort Collins, Colo. pp. 452474. Vasquez-Herrera, A., Dobry, R., and Ng, T.-T. 1988. Pore pressure buildup and liquefaction failure of anisotropically consolidated sand due to cyclic straining. In Proceedings of the Conference on Hydraulic Fill Structures, ASCE, Fort Collins, Colo. pp. 346366.

2005 NRC Canada