Professional Documents
Culture Documents
Abstract
Full-scale test embankments, with and without geotextile reinforcement, were constructed
on soft Bangkok clay. The performances of these embankments are evaluated and compared
with each other on the basis of field measurements and FEM analysis. The analyses of failure
mechanisms and the investigations on the embankment stability using undrained conditions
were also done to determine the critical embankment height and the corresponding geotextile
strain. The high-strength geotextile can reduce the plastic deformation in the underlying
foundation soil, increase the collapse height of the embankment on soft ground, and produce a
two-step failure mechanism. In this case study, the critical strain in the geotextile
corresponding to the primary failure of foundation soils may be taken as 2.5–3% irrespective
of the geotextile reinforcement stiffness.
r 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Soft soil model; Mohr–Coulomb model; Critical height; Collapse height; Critical strain
1. Introduction
0266-1144/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 6 - 1 1 4 4 ( 0 2 ) 0 0 0 3 2 - 8
344 D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365
Table 1
Properties of the geotextile reinforcements
layers of Polyfelt TS420. The HGE embankment was reinforced by a single layer of
high-strength, nonwoven/woven geotextile (PEC200) placed directly on the natural
ground surface. The properties of the geotextile are presented in Table 1.
The general soil profile and the basic soil properties are given in Fig. 3. The
top 12 m depth can be divided into four layers. The weathered crust consisting of
346 D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365
vertically at the toe of each embankment down to the depth of 12 m into the stiff clay
layer.
6.5 m from the center of embankment corresponding to notations L1, L2, L3 and L4,
respectively.
Before construction of the test embankments, the ground surface was excavated to
about 0.20 m depth and leveled. The canal with dimensions of 2 m depth from the
original ground surface and 7.5 m width at the bottom was excavated along the
proposed failure side of all the embankments.
D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365 349
The embankment loading was calculated as the product of the fill thickness and
the total unit weight considering the effects of rainfall during construction. The net
embankment height is defined as the difference between the current elevation of
embankment crest and the original elevation of the embankment base. The CE
embankment was constructed in layers with compaction lift thickness of about
0.33 m. The dry density of 17 kN/m3 and average moisture content of 9% were
maintained. The total unit weight was 18.5 kN/m3. In all the three cases, very small
deformations were observed at embankment heights lower than 2.5 m. The rate of
settlement and lateral movement increased significantly when the embankment
heights exceeded 3.5 m. The CE embankment reached a net height of 4 m (i.e. 4.32 m
fill thickness or 80 kPa embankment loading) 24 days after the start of embankment
construction. The corresponding maximum settlement and lateral movement
measured at this time were 0.32 and 0.17 m, respectively. No other construction
activity was done and the embankment was stable until the morning of the 25th day.
A crack of about 5 mm wide and 3 m long was observed in the morning of the 25th
day along the bottom of the canal near the settlement plate, S6. The width of the
crack opened quickly to about 10 mm wide, and then the embankment collapsed
within 10 min. The vertical crack and the collapse of embankment crest are
illustrated in Fig. 6. The failure surface was estimated with the bamboo sticks
embedded during construction, approximated as a circular surface.
Subsequently, the MGE and HGE embankments were built at the same time and
at almost the same rate of filling. Construction procedure and quality control were
also kept the same as that of the CE embankment. When the two embankments
reached the net height of 3.75 m, one small crack of about 5 mm wide was observed
near the surface settlement plate S6 in the bottom of the canal. In contrast with the
case of the unreinforced embankment, the crack gradually developed to about
10 mm wide and 3 m long the next day with no further development observed in the
later days. At this embankment height of 3.75 m, the embankment loading was about
75 kPa and the measured strains in geotextile were smaller than 1.0%. The rates of
settlement, lateral displacement and geotextile strain increased significantly with
further increase in height. When the embankments reached 4.2 m net height, the
measured maximum settlements were 0.33 and 0.40 m for HGE and MGE
embankments, respectively. The corresponding maximum strains in the geotextile
were about 2.3% and 3%, and the most strained point in the geotextile is located
about 3.5 m from the centerline of both embankments. The failure of MGE
embankment that concurrently induced failure of HGE embankment occurred on
the 62nd day. All instruments of HGE embankment were still functioning except the
inclinometer casing. The maximum strain in geotextile of HGE embankment
measured after the failure was 8.5%, which indicated that the failure has caused an
additional 6% strain in the geotextile. At this stage, the geotextile reinforcements in
the HGE embankment were not ruptured while in the MGE embankment they were
completely ruptured. The HGE embankment construction was then continued to a
net height of 6 m at which stage it collapsed. The corresponding settlement measured
at the center of the embankment was 0.67 m. The maximum strain in geotextile of
HGE embankment was in the order of 12–14%, which is the range of rupture strain
from wide-width in-air tensile tests. A comparison of the failure planes of CE and
reinforced embankments for both MGE and HGE embankments are presented in
Figs. 6 and 7.
D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365 351
It is noted that there were 35 rainy days during the construction of MGE and
HGE embankments. Consequently, the measured average field moisture content of
MGE and HGE fill was about 13% corresponding to the total unit weight of
19.2 kN/m3, and was about 4% higher than that of the CE embankment with total
unit weight of 18.5 kN/m3. Therefore, the actual collapse loads of MGE and HGE
embankments are 88 kPa (i.e., 4.2 m net embankment height plus 0.4 m maximum
settlement=4.6 m fill thickness) and 128 kPa (i.e., 6 m net embankment height plus
0.67 m maximum settlement=6.67 m fill thickness), which are 1.1 and 1.6 times
higher than that of the CE embankment (80 kPa and 4.32 m fill thickness),
respectively.
The finite element meshes are presented in Figs. 8 and 9 for CE and HGE
embankments, respectively. Both consolidation and undrained analyses were carried
out. The consolidation analysis simulating the actual construction conditions was
conducted in order to consider the effects of consolidation process due to the
difference in construction sequences of the test embankments.
The 6-noded triangle, linear-strain elements are employed for both foundation
soils and embankment fill. The geotextile reinforcement and soil-geotextile interface
are simulated by 3-noded geotextile elements and 6-noded flat elements, respectively.
The finite element mesh configuration is based on the actual soil profile (Fig. 3)
which consisted of four layers: weathered crust from 0 to 2 m, soft clay from 2 to
8.5 m, medium clay from 8.5 to 11 m, and stiff clay from 11 to 14 m. The soft clay is
then divided into three sublayers from 2 to 4 m, 4 to 6 m and 6 to 8.5 m to better
simulate the depth effect. The piezometric drawdown due to excessive groundwater
pumping (Bergado et al., 1988) is considered in calculating the in situ stresses and
pore pressures of the subsoils. The lower boundary with fixed displacements and free
drainage is set at 14 m depth from the ground surface where the sand layer was
encountered.
The embankment fill is modeled by three and five layers of elements for CE and
HGE embankments, respectively (Figs. 8 and 9). The canal excavation and the
embankment construction are taken into account by removing or adding the
corresponding elements according to the construction sequence. In the PLAXIS
program, the load caused by the self-weight of removing or adding elements are
automatically applied by small increments in order to satisfy the tolerable errors
(smaller than 5%) for both local and global equilibrium.
Table 2
Selected parameter for subsoils in FEM analyses
embankment loading is far from failure. The model parameters of foundation soils
for FEM analyses that are basically interpreted from the soil investigation data are
given in Table 2.
The SSM, for soft and medium clay layers, requires six parameters: friction angle,
f; cohesion c0 ; modified compression ratio, l ; modified swelling ratio, k ; at-rest
lateral earth pressure coefficient in normally consolidated state, K0ðNCÞ ¼
ð1 sin f Þ; and Poisson’s ratio, n0 : The FEM prediction of embankment failure
height is very sensitive with the selected strength parameters, f0 and c0 : The cohesion
depends on the current overburden, s0vo ; the overconsolidation ratio, OCR, and the
friction angle, f0 : In this study, the friction angle of the subsoils is adopted as 231
based on extensive test results on soft Bangkok clays (Balasubramaniam, 1991). The
cohesion values are back-calculated from the failure of CE embankment in addition
to deriving from the standard field vane shear tests by the following approximation:
1 þ 2K0 0
c0 ¼ msuv svo tan f0 ; ð1Þ
3
where suv is the average field vane strength, m is the correction factor, and K0 is the
at-rest earth pressure coefficient, estimated as follows:
In Eq. (2), the value of m ¼ 0:3 (Ladd, 1991) and the earth pressure coefficient in
normally consolidated state, K0ðNCÞ ¼ 1 sin f0 ; as given by Jaky (1948), are used.
The correction factor, m; in Eq. (1), is obtained by matching the calculated and
measured failure heights of CE embankment and a value of m of 0.77 was obtained.
Then, the corresponding apparent cohesion, c0 ; can be obtained as presented in
Table 2.
The modified compression ratio, l ; is computed from the compression index, Cc ;
which is obtained from the conventional oedometer test results. The empirical
correlation using the initial water content (Bergado et al., 1995a) has been used to
further modify l to account for the variation with depth. The value of k is taken as
0.2 times that of l as suggested by Vermeer and Brinkgreve (1995). Poisson’s ratio
for a soft clay depends on the confining stress status level and hence a variable value
with depth has been adopted.
354 D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365
Table 3
Selected parameters for embankment fill in FEM consolidation analysis
the range of strain from 0% to 12%. The creep strains and the influence of
temperature on the in-soil stiffness of the reinforcements in the short period of the
embankment construction were ignored.
The FEM undrained analyses for embankment stability were carried out for all the
three test embankments. These are parametric studies using the same properties of
embankment fill to verify the improvement of stability in the geotextile-reinforced
embankments and to further investigate the failure mechanisms and to directly
compare the stability behavior of the three test embankments. In FEM modeling of
MGE embankment, the three layers of geotextile TS420, which are spaced at one
compaction lift thickness of 0.33 m, are simulated as one layer having an equivalent
stiffness of the three layers (70 3=210 kN/m).
From the case histories of HGE embankment and Sackville embankment (Rowe
et al., 1995; Rowe and Hinchberger, 1998), it appears that the height of a geotextile-
reinforced embankment can be increased after the failure of foundation soil (primary
failure). Therefore, it is necessary to distinguish the collapse height, Hf; and the
critical height, Hc ; of an embankment on soft ground. The collapse height (of
reinforced or unreinforced embankment) can be defined as the height at which any
attempt to increase the height of the fill will yield no increase in the net embankment
height. In other words, the collapse height corresponds to the maximum net
embankment height that can be constructed. Thus, the collapse height can be
obtained directly from the plot of net embankment height versus embankment
settlement. The net embankment height is defined as the difference between the
current elevation of the embankment crest and the original elevation of the
embankment base. The critical height of a reinforced embankment corresponds to
the collapse height of an unreinforced embankment on the same foundation subsoil.
Therefore, from Figs. 10a and b the net collapse heights of CE, MGE and HGE
embankments would be 4.04, 4.65, and 5.96 m, respectively, while the corresponding
net critical heights of MGE and HGE embankments are 4.40 and 4.80 m.
Furthermore, as observed from these figures, using higher geotextile stiffness, more
reduction in foundation deformations and higher critical height can be achieved. It
should be noted that the critical height and the collapse height of a reinforced
embankment defined herein can be considered as the heights corresponding to the
failure of foundation soil (primary failure) and the collapse of embankment and its
foundation as a whole (secondary failure), respectively. For unreinforced CE
embankment, the FEM results indicate that the critical height coincides with the
collapse height.
356 D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365
Fig. 10. FEM undrained analyses for the three test embankments. (a) Net embankment height versus
maximum settlement. (b) Net embankment height versus maximum lateral displacement.
Fig. 11. Net embankment height and geotextile strain versus settlement in FEM undrained analysis of
AIT test embankments.
Table 4
Comparison of critical strains for different test embankments
height and the critical height is dependent on both geotextile stiffness and foundation
soil.
The typical profiles of lateral displacement under the toe of HGE and CE
embankments are presented in Figs. 12 and 13, respectively. Generally, the FEM
results are comparable with the monitored data except for the prior failure case of
CE embankment, which may be due to excess plastic zone formation.
Both FEM and measured data show the zone of highest lateral displacement
that occurred within the depth of 2–4 m coinciding with the weakest soil layer. The
lateral displacement in an HGE embankment is substantially smaller than that of the
CE embankment at embankment loading higher than 65 kPa. The lateral
displacements were almost the same as the embankment loading increase up to
65 kPa. Thus, the geotextile reinforcement mainly reduced the plastic lateral
displacement. This is consistent with the interpretation of settlements. It should be
mentioned that the maximum lateral and vertical displacements in HGE embank-
ment at 4.2 m net height (q ¼ 88 kPa) are almost the same as that of CE embankment
prior to collapse at 4.0 m height (q ¼ 80 kPa). This demonstrated that the foundation
of the HGE embankment had failed at an embankment height of 4.2 m. The failure
of HGE embankment at this height, therefore, may also be considered as the primary
failure.
D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365 359
In FEM analysis, the collapse load can be determined as the load that triggered
the continuous lateral displacements. Consequently, the loading of 78 and 123 kPa
are the FEM calculated collapse loads of CE and HGE embankments, respectively
(Fig. 10a). The collapse loads are calculated based on the fill thickness, i.e., net
embankment height plus maximum settlement of the embankment. The correspond-
ing actual collapse loads of CE and HGE embankments were 80 and 128 kPa. Thus,
good agreement between calculated and actual collapse loads has been obtained.
The variation of geotextile strain versus time combined with the corresponding
FEM results are presented in Fig. 14. Up to day 61, the calculated strains agreed well
with the mean values of measured strains. Beyond this point the FEM results are
smaller than the measurements. This is because the induced failure on day 61
resulted in an additional strain of 6%, which has not been included in the FEM
analysis.
360 D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365
Fig. 15. (a) Shear stresses at interfaces of PEC200/soil at Hnet ¼ 4:2 m. (b) Shear stresses at interface of
PEC200/soil at Hnet ¼ 6:0 m.
The shear stress distribution along the soil–geotextile interface obtained by FEM
analysis are plotted in Fig. 15a and b corresponding to the net embankment heights
of 4.2 m (q ¼ 88 kPa) and 6.0 m (q ¼ 128 kPa) of the HGE embankment. At an
embankment load of 88 kPa, which can be considered as the maximum service load,
the interaction mechanism between the soil and geotextile is the direct shear mode, as
manifested by the opposing directions of the mobilized interface shear between the
upper interface and the lower interface of the geotextile (Fig. 15a). However, prior to
collapse at q ¼ 128 kPa, the pullout mode dominated, as manifested by the same
362 D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365
Fig. 16. (a) Contours of relative shear stress in CE embankment at q ¼ 61 kPa (qf =q ¼ 1:3). (b) Contours
of relative shear in CE embankment at failure q ¼ qf ¼ 80 kPa, contour interval=0.1, maximum
value=1.0.
direction of the mobilized interface shear between the upper and lower interfaces of
the geotextile reinforcement (Fig. 15b). Considering the magnitude of mobilized
shear stresses along the interfaces, it can be found that the interface shear stresses are
much smaller than the shear strength of the surrounding soils even at collapse load.
With the undrained shear strength of the weathered crust beneath the geotextile of
37.5 kPa, the maximum value of interface shear stresses is less than one-half of the
available shear strength. Therefore, there was no slippage between soil and geotextile
and the pullout failure did not occur in this test embankment.
The contours of relative shear stress, defined as the ratio of shear stress to the
available shear strength, are illustrated in Figs. 16a and b for CE embankment
corresponding to embankment load of q ¼ 61 kPa and at failure (q ¼ qf ¼ 80 kPa),
respectively. The darkest shade zones in these figures represent the zones with unity
value of relative shear stress ratio i.e. the soil is considered as at plastic state or
failure. Fig. 16a indicates that the local failures appear in the soft clay layer at an
embankment load of 61 kPa which is in accordance with the interpretation from the
measured pore pressures and deformation where the local failure load was found to
be 65 kPa. According to q ¼ 61 kPa, the value of qf =q is 1.3. Thus, a safety factor
against bearing failure of foundation soil of 1.3, as used in conventional design, can
D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365 363
be considered as the limit to avoid the large deformation due to the development of
plastic zone in foundation soils. Moreover, the local failures that started at qf =q ¼
1:3 as obtained herein also implied that the lower bound limit analyses, in which the
failure load is defined as the load corresponding to the failure at the first point in the
soil mass, underestimated the embankment stability. Increasing the embankment
loading to the failure load of qf ¼ 80 kPa, the plastic zones progressively developed
to be a continuous band as seen in Fig. 16b. Thus, the failure surface has been
formed in the foundation soil while the shear strength in the embankment fill has not
fully mobilized yet. However, even without the increase of embankment load, the
deformation in the embankment fill may continue because of the continuous
displacement of foundation soils at plastic state together with the undrained creep
deformations at a high stress level. Consequently, the shear stress in embankment fill
continued to mobilize until softening. Subsequently, the complete collapse of the
embankment occurred. This explained the delay in the failure of CE embankment
that occurred about 14 h after the failure load had been fully reached. Such failure
mechanism suggested that the strength parameters used in stability analyses of
unreinforced embankment on soft Bangkok clay should be selected corresponding to
large strain at critical state for both foundation soils and embankment fill.
The progressive mobilization of relative shear stress of the HGE embankment are
reflected in Figs. 17a and b corresponding to embankment loads of q ¼ 88 and
Fig. 17. (a) Contours of relative shear stress in HGE embankment at q ¼ 88 kPa (Hnet ¼ 4:2 m). (b)
Contours of relative shear in HGE embankment at failure q ¼ 128 kPa, contour interval=0.1, maximum
value=1.0.
364 D.T. Bergado et al. / Geotextiles and Geomembranes 20 (2002) 343–365
128 kPa. At embankment loading of 88 kPa, the continuous band of plastic zone was
formed as seen in Fig. 17a. Increasing the embankment fill, the failure zone
developed further as reflected in Fig. 17b. Considering the failure zone in HGE
embankment, it is confirmed that even without the failure of MGE embankment on
day 61, the primary failure in HGE embankment might have also taken place at the
embankment load around 88 kPa. Associated with the primary failure is the
mobilization of the localized geotextile strength that resulted in the increase of
embankment height until the embankment collapsed (secondary failure) accom-
panied by the rupture of geotextile. This can be referred to as the two-step failure
mechanism of high-strength geotextile-reinforced embankment on soft ground. This
failure mechanism is similar to the case history at Sackville embankment described
by Rowe et al. (1995). In this embankment, the primary failure was observed at 5.7 m
fill thickness with a sudden settlement of about 0.5 m associated with the additional
strain in geotextile of about 3%. Afterwards, the embankment fill was successfully
added up to the embankment collapse at 8.5 m fill thickness. The mobilized strain in
geotextile prior to primary failure was in the range of 2.3–3% in both HGE and
Sackville embankments, even though their geometry and foundation soil profile are
not the same. At this strain level, the improvement in embankment stability may not
be considerable. Nevertheless, both case histories indicated that the high-strength
geotextile reinforcement can significantly increase the collapse height after the
primary failure of foundation soils. This is an important aspect that should be
considered in practical applications.
6. Conclusions
(4) The results of this study as well as other three case histories presented have
substantiated that the critical strain, elc ; in geotextile corresponding to the
primary failure of foundation soils is not much dependent on the geotextile
stiffness and the subsoil conditions. In practical design, it can be taken as
2.5–3%.
Acknowledgements
With deep appreciation and gratitude, the authors would like to acknowledge the
financial and technical support from Polyfelt Geosythetics of Austria. Sincere thanks
are due to Dr. P. Delmas for his technical assistance in the project.
References
Balasubramaniam, A.S., 1991. Contributions in geotechnical engineering soil mechanics and foundation
engineering. Inaugural Lecture, Asian Institute of Technology, Bangkok, Thailand, p. 86.
Bergado, D.T., Long, P.V., Lee, C.H., Loke, K.H., Werner, G., 1994. Performance of reinforced
embankment on soft Bangkok clay with high-strength geotextile reinforcement. Geotextiles and
Geomembrances. 13, 403–420.
Bergado, D.T., Polinar, L.T., Long, P.V., 1995a. Prediction and performance on consolidation settlement
of AIT full scale test embankment. Proceedings of International Symposium on Compression and
Consolidation of Clayey Soils, IS-Hiroshima’95.
Bergado, D.T., Werner, G., Tien, M.H., Zhou, X.H., 1995b. Interaction between geotextiles and silty sand
by large direct shear and triaxial tests. Proceedings of Geosynthetics’95 Vol. 3, Nashville, TN, USA,
pp. 1097–1109.
Bergado, D.T., Long, P.V., Werner, G., 1996. Deformation behavior of geotextile reinforcement at
vicinity of shear surface. Proceedings of the First European Geosynthetics Conference and Exhibition,
Maasstricht, The Netherlands, pp. 177–182.
Bergado, D.T., Nutalaya, P., Balasubramaniam, A.S., Apaipong, W., Chang, C.C., Khaw, L.G., 1998.
Cause, effects and predictions of land subsidence in AIT campus, Chao Phraya Plain, Thailand.
Bulletin of the Association of Engineering Geologists 25 (1), 57–81.
Burland, J.B., 1965. The yielding and dilation of clay (correspondence). Geotechnique 15, 211–214.
Chai, J.C., Miura, N., Bergado, D.T., Long, P.V., 1997. Finite element analysis of embankment failure on
soft subsoil. Geotechnical Engineering. 28 (2), 249–276.
Delmas, P., Queroy, D., Quaresma, M., De Saint, A., 1992. Failure of an experimental embankment:
Guiche. In: de Hoedt (Ed.), Geotextiles, Geomembranes and Related Products. Balkema, Rotterdam.
Jaky, J., 1948. Pressure in soil. Proceedings of the Second ICSMFE 2, pp. 42–494.
Ladd, C.C., 1991. Stability evaluation during staged construction. Journal of Geotechnical Engineering,
ASCE 4 (117), 537–615.
Rowe, R.K., 1992. Reinforced embankments on soft cohesive deposits. Proceedings of the International
Symposium on Applications of Geosynthetics Technology, Jakarta, pp. 1–19.
Rowe, R.K., Hinchberger, S.D., 1998. The significance of rate effects in modeling the Sackville test
embankment. Canadian Geotechnical Journal 35, 500–516.
Rowe, R.K., Granendran, C.T., Landva, A.O., Valsangkar, A.J., 1995. Construction and performance of
a full scale geotextile reinforced test embankment. Sackville, New Brunswick. Canadian Geotechnical
Journal 32, 223–237.
Vermeer, P.A., Brinkgreve, R.B.J., 1995. Finite element code for soil and rock analyses. Balkema,
Rotterdam.