Improvement of dynamic soil properties induced by preloading verified by a
field test

C. Stamatopoulos, P. Petridis, M. Bassanou, S. Allkja, N. Loukatos, A.

Small

PII: S0013-7952(13)00186-5
DOI: doi: 10.1016/j.enggeo.2013.06.003
Reference: ENGEO 3616

To appear in: Engineering Geology

Received date: 12 October 2012

Revised date: 1 June 2013
Accepted date: 6 June 2013

Please cite this article as: Stamatopoulos, C., Petridis, P., Bassanou, M., Allkja, S.,
Loukatos, N., Small, A., Improvement of dynamic soil properties induced by preloading
verified by a field test, Engineering Geology (2013), doi: 10.1016/j.enggeo.2013.06.003

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
 ACCEPTED MANUSCRIPT
Title: Improvement of dynamic soil properties induced by preloading verified by a field test
1 1 1 2
Authors: C. Stamatopoulos , PhD, P. Petridis , MS, M. Bassanou , PhD, S. Allkja , BS , N.

Loukatos BS, A. Small, BS

T
Affiliation of authors:

P
1: Stamatopoulos and Associates, 5 Isavron st, Athens 114-71, Greece

RI
2: ALTEA & Geostudio 2000, Ruga e Kavages, Ish Mjetet Mesimore, Tirana, Albania

SC
3: GeoSTAND Co, Kalymnou 16,112 51 Athens, Greece

4: Instrumentation Testing & Monitoring (ITM) Limited, Bell Lane, Uckfield, East Sussex TN22

NU
1QL , United Kingdom
MA
Contact address: Stamatopoulos and Associates, 5 Isavron st, Athens 114-71, Greece

Telephone: (30)210-3603911
ED

Fax: (30) 210-3616919, e-mail: k.stam@saa-geotech.gr

PT
CE

Number of words of abstract: 115

Number of words (excluding abstract, keywords, tables and figure captions): 7027
AC

Number o tables: 7

Number of figures: 9

1
 ACCEPTED MANUSCRIPT

ABSTRACT:

Presented here are the results of an elaborate field preloading study on a liquefaction-susceptible

T
site. Preloading was applied by a temporary embankment 9m high. Prior and after preloading,

P
borings with standard penetration tests, cone penetration tests and geophysical studies were

RI
performed. During the process of embankment construction and demolition, settlements, excess

SC
pore pressures and vertical and horizontal stresses were recorded versus time at different

locations. A partial embankment failure occurred during the preloading process. Methods are

NU
proposed and verified which predict (i) risk of failure during the construction of the preload

embankment and (ii) stresses and displacements and (iii) the change in the shear wave velocity
MA
and cyclic liquefaction strength induced by the preload embankment.
ED

Keywords
PT

Soil improvement, preloading, field test, liquefaction, shear wave velocity, horizontal stress
CE

increase, Modified Cam Clay model, embankment failure

AC

2
 ACCEPTED MANUSCRIPT

1. Introduction

Ground improvement methods are frequently used in soft or loose soil to sustain static or seismic

T
loadings (Bell, 1996, Du et al., 1999, Cai et al., 2006, Adalier and Elgamal, 2004 ). One such

P
method is preloading. It consists of a temporary loading, usually a soil embankment, applied to

RI
improve subsurface soils by densification and increasing lateral stress (Alonso et al., 2000, Al-

SC
Shamrani and Dhowian, 1997, Stamatopoulos and Kotzias, 1885, Petridis et al, 2000). Many

cases have been reported where the effectiveness of preloading has been demonstrated

NU
(Stamatopoulos and Kotzias, 1885, Petridis et al, 2000): For a height of the temporary soil

embankment between 7-13m, the preloading-induced settlement was measured of the order of
MA
tens of centimeters, which illustrates soil densification. In addition, construction after preloading

was successful in terms of static settlements. Furthermore, the blow count measured using the
ED

Standard Penetration Test versus depth, both before and after soil improvement, illustrated the

improvement.
PT
CE

The liquefaction cyclic strength is a critical dynamic soil parameter because it determines whether
AC

a site will liquefy under a design earthquake (European Standard, 2003). As a result of the

difficulty of undisturbed sampling of sandy soils, the in-situ liquefaction cyclic strength is mainly

estimated by field tests: the blow count (NSPT) measured in Standard Penetration Tests (SPT), the

cone resistance (qc) measured in Cone Penetration Tests (CPT) and the shear wave velocity (V S)

(Boulanger and Idriss, 2006, European Prestandard, 1994, European Standard, 2003, Idriss and

Boulanger, 2004). The shear wave velocity is also a critical dynamic soil property because it

determines the dynamic amplification response of soils (Kramer, 1996, European Standard, 2003).

A field test with pre- and post improvement SPT, CPT and Vs measurements, as well as

measurements of strains and stresses during the preload process, is the most efficient method of

investigating the effect of preloading in changing the dynamic properties of soils in-situ. e effect of

preloading in changing the dynamic properties of soils in-situ. The reason being that field tests can
3
 ACCEPTED MANUSCRIPT
illustrate (a) the improvement in the field, not only in terms of the in-situ soils and preloading

characteristics, but also in terms of measured in-situ changes in horizontal stress and void ratio,

(b) the ability of methods to predict the measured improvement, and (c) the manner and rate in

which preloading should be applied in the field.

P T
Only one field study was found in the literature available whereby the increase in dynamic soil

RI
properties and quantities affecting them were measured (Stamatopoulos et al., 2005, Raptakis,

SC
2012): The preload embankment was 9m high. SPT and CPT strength and VS versus depth before

and after preloading were measured. The increase in horizontal stress caused by preloading, that

NU
affects the dynamic soil properties, was also measured. However, the site up to 10m depth

consisted mainly of clay soil, and thus was not liquefaction-susceptible. Furthermore, the vertical
MA
strain versus depth was not measured. This is a critical measurement, because it can correlate the

change in dynamic properties of soil layers with the change in void ratio.
ED

According to the aforementioned text, a complete field test on a liquefaction-susceptible site with
PT

data of (a) the SPT and CPT strength and VS versus depth before and after preloading and (b)
CE

changes in the horizontal stress and void ratio of the soil as a result of preloading does not exist

in any literature and is needed to assess the effect of preloading on the dynamic soil properties
AC

and the ability of methods to predict this effect. Such a field test was performed recently during a

project funded by the European Union. What follows shows firstly methods predicting stresses

and displacements induced by the preload embankment and the change in dynamic soil properties

induced by the preload embankment. Then, the field test performed and the measured change in

dynamic and other soil properties is described. Finally, the predictive methods are verified based

on measurements before, during and after the field test. Furthermore, the occurrence of the

embankment failure allowed the verification of a method proposed predicting the risk of failure

during construction of a preload embankment.

2. Predictive methods

4
 ACCEPTED MANUSCRIPT

2.1. Proposed method warning against failure during preload embankment construction

Soil stability failure during the construction of the preload embankment is of considerable concern,

especially when the in-situ soil at shallow depths is not only soft, but also has a low coefficient

T
of consolidation, so that dissipation of excess pore pressures can be slower than the construction

P
rate. Fig 1a gives the approximate undrained strength for failure in terms of the geometry of the

RI
embankment and the depth of the soft layer by modifying the presentation of a chart by Taylor

SC
(1948). Referring to Fig. 1a, the undrained failure strength (sufail) approximately equals

NU
sufail  A γ h (1)
MA
where γ is the unit weight of the soil , h is the height of the embankment and A is a factor that

depends on the height and inclination of the slope of the embankment.

ED

Soft clays are typically normally consolidated. The undrained soil strength of normally-consolidated
PT

soft clay layers , according to Ladd (1991), can be estimated, as a first approximation, as
CE

su  0.2 σ'v (2)

AC

where σ'v is the in-situ effective vertical stress. Using equations (1) and (2), it is inferred that the

factor of safety against stability failure (FS) can be obtained in terms of time from the excess pore

water pressure measurements as

FSt-i  0.2 (σv(t-i) -Pt-i ) / (A γ ht ) (3)

where σv is the total vertical stress, the subscript t indicates variation in terms of time, the

subscript t-i indicates variation in time at location i and Pt-i is the excess pore pressure

measurement at tranducer i versus time t.

5
 ACCEPTED MANUSCRIPT
Equation (3) provides an estimate of the local factor of safety versus construction time in the case

where the soil consists of soft clay near the surface. When this parameter is near unity, danger of

embankment failure exists. Application of equation (3) needs pore pressure measurements, which

can be obtained by the installation of pore pressure transducers, while the total vertical stress can

T
be estimated from solutions of linear elasticity readily available.

P
RI
SC
2.2. Methods predicting maximum and final stresses and displacements induced by preloading

Predictions of settlement and vertical strain induced by preloading are needed, primarily in order to

NU
assess densification, and thus soil improvement achieved by preloading. In addition, for dynamic

problems, prediction of the increase in horizontal stress by preloading is also useful because the
MA
effect of preloading on dynamic soil properties, additionally depends on this parameter, as is

described below
ED

Predictions of the ground settlement and the increase in horizontal stress induced by the
PT

preloading process (a) after embankment construction and (b) after embankment demolition can

be made using both elasto-plastic numerical and simplified procedures. Regarding the simplified
CE

methods, predictions of the vertical (Δσv-max-s) and horizontal (Δσh-max-s) stress induced by the
AC

surcharge versus depth can be estimated by linear elasticity. From the computed distribution of

vertical stresses, settlements can be estimated using one-dimensional consolidation analysis

(Taylor, 1948). When the total surcharge is applied, the change in void ratio (ΔeL) versus depth

under loading equals

ΔeL = -Crj log (OCRbef) - Ccj log(OCRafter/ OCRbef) (4a)

where

OCRbef = Pp/σg (4b)

6
 ACCEPTED MANUSCRIPT
OCRafter = maximum (OCRbef , [(σg +Δσv-max-s)/σg]) (4c)

where σg is the effective geostatic stress prior to the application of preloading, Pp is the

preconsolidation stress and Ccj and Crj are the virgin and swelling coefficients of compressibility

T
respectively of the soil layer (j) at this particular depth.

P
RI
When the surcharge is removed, the change in void ratio (ΔeU) equals

SC
ΔeU = Crj log (OCRafter) (4d)

NU
The total change in void ratio induced by preloading equals
MA

Δe = ΔeL + ΔeU (4e)

ED

Furthermore, the change in vertical strain (Δε) versus depth can be related to the initial void ratio
PT

(e) and its change (Δe) as ,

CE

Δεv= Δe/(1+e) (4f)

AC

The Mayne and Kulhaway (1982) empirical relationship can be used to predict the coefficient of

lateral pressure at rest (ko ) in terms of the overconsolidation ratio (OCR) and the friction angle

(φ) as:

ko = (1 - sinφ) OCRsinφ’ (5)

Using eq. (5), the residual change in the effective horizontal stress (Δσ'h-res ) and the lateral stress

ratio (Δko-res) induced by preloading, can be estimated as:

7
 ACCEPTED MANUSCRIPT
Δσ'h-res = (1 - sinφ) σg (OCRaftersinφ'- OCRbefsinφ') (6a)
sinφ' sinφ'
Δ ko-res = (1 - sinφ) (OCRafter - OCRbef ) (6b)

Regarding the elasto-plastic analysis, the Modified Cam Clay model (Roscoe and Burland 1968).

T
can be used in preload problems, as this model simulates the change of soil properties due to

P
densification. Drained conditions can be assumed in the analyses. The reason is that (i) after

RI
embankment construction, just prior to embankment demolition, and (ii) after embankment

SC
demolition, excess pore pressures induced by preloading are typically small (Stamatopoulos et al.,

2005). The Modified Cam Clay model predicts the maximum and residual stresses, strains and

NU
settlement induced by preloading directly using (i) the model parameters Ccj, Crj, Pp, σg of the

simplified method above and (ii) the elastic soil parameters. Equation (6a) is also needed to
MA
estimate the initial horizontal stresses at the site.
ED

2.3. Equations predicting the change in dynamic soil properties induced by preloading

The shear wave velocity can be estimated from the shear modulus at small strains (Gmax ), the soil
PT

unit weight (γ), and the acceleration of gravity (g) (Kramer, 1996), as
CE

1/2
VS = (g Gmax/γ) (7a)
AC

According to Kramer (1996), laboratory data suggest that Gmax can be expressed as

Gmax= 625 f(e) Pa OCRk (σ'oct/Pa)0.5 (7b)

where f(e) indicates a function in terms of the void ratio, Pa is the atmospheric pressure, σ'oct is the

effective octahedral stress, and the factor k depends on the Plasticity Index (PI). For PI=0, 20, 40,

60, 80, k equals 0, 0.18, 0.30, 0.41 and 0.48 respectively. Jamioljowski et al (1991) suggest that

f(e)= 1/e1.3 (7c)

Using equations (7a), (7b) and (7c) the following equation can be proposed predicting the ratio of

shear wave velocity after and before the application of preloading versus depth:

8
 ACCEPTED MANUSCRIPT
0.65 k/2 0.25
 ebef
VS  after   OCR after   1  2k o after 
      (8)
Vs  bef  eafter 

 OCRbef



 1  2k
 o bef



where the subscripts after and bef indicate the before and after application of preloading.

TP
The cyclic stress ratio is the ratio of the shear stress applied by an harmonic excitation by the in-

RI
situ vertical effective stress. In geotechnical earthquake engineering, cyclic liquefaction strength

SC
(SR15) is defined as the cyclic stress ratio causing liquefaction in the case of an earthquake of

magnitude M equal to 7.5, or equivalently during 15 cycles of uniform harmonic loading (European

NU
Standard. 2003, Idriss and Boulanger. 2004.). A M=7.5 earthquake is used presumably because it

is a representative large event in many seismic regions. Based on the results of cyclic laboratory
MA
tests in samples with or without preloading, and their numerical simulations, Stamatopoulos et al.

(2012) propose the following equation predicting the increase of cyclic liquefaction strength
ED

induced by preloading
PT

0.04 / SR15 bef

SR15-after = SR15-bef PR (9a)
CE

PR=(σg +Δσv-max-e)/σg (9b)

AC

The effect of horizontal stress on the cyclic soil strength has been studied in the tortional-shear

device by Ishihara and Takatsu (1979). They conclude that for samples with the same pre-

consolidation density, the liquefaction cyclic strength is proportional to the consolidation octahedral

effective stress. Based on this, the increase in cyclic strength induced by preloading can be

expressed as

SR15-after / SR15-bef = (1+ 2 ko-after ) / (1+ 2 ko-bef ) (10)

9
 ACCEPTED MANUSCRIPT
Finally, as shown in Fig. 2, the relationship between the shear wave velocity and the liquefaction

cyclic strength given by Eurocode (European Prestandard, 1994) can be simulated in equation

form as

T
SR15= 7.84 10-6 Vs12 (11a)

P
RI
where Vs1 is in m/s and equals

SC
Vs1 =Vs * (Pa/σ'g)0.25 (11b)

NU
The coefficient of correlation of equation (11a) is 0.98, indicating good accuracy. As for sandy
MA
soils which liquify PI is less than 7 (Boulanger and Idriss, 2006), equations (7) and (11) predict that
ED

1.3 0. 5
SR15after  ebef   1  2k o after 
    (12)
SR15bef  eafter   1  2k o bef 
   
PT
CE

In equations (8) and (12) , eafter and ko-after can be obtained from ebef and ko-bef using equations (4)

and (6).
AC

3. The field test

3.1. Site

The site for the field test was in Porto Romano, 10km North of Durress in the Albanian coast (Fig.

3a). The site was rented and four borings up to 15m of depth each along with sampling and

Standard Penetration Tests (SPT) at every meter were performed to verify that the site is suitable

for the purpose of the present research. Then four additional borings up to 15m depth each were

performed with sampling and SPT at every meter at locations corresponding to distances less than

10
 ACCEPTED MANUSCRIPT
6.5m from the centre of the conical embankment to be built. Piezometers were installed in two

borings to measure the elevation of the water table. A standard laboratory testing program was

also performed which included classification, compressibility and strength tests. In addition, three

Cone Penetration Tests, and three down-hole surveys for measuring the shear wave velocity to 15

T
m depth each were performed. The average water table line was measured at depth 1m. Fig. 4a

P
gives the initial (σ'g) and maximum past (Pp) vertical effective stress estimated from oedometer

RI
tests versus depth. Table 1 gives the soil layers which exist in the site and their average Plasticity

SC
Index and fines content based on classification tests, as well as their compressibility estimated

from oedometer tests. Fig. 5 and table 2 give the average (i) NSPT, (ii) qc and (iii) VS

NU
measurements versus depth and layer.
MA
3.2. Instrumentation

Prior to construction of the preload embankment the following instruments were placed at the
ED

location of the field test and worked properly: (a) Vertical pressure cells were installed just below

the ground surface and at three locations which corresponded to different points from the centre to
PT

the edge of the embankment which was to be built. (b) Horizontal pressure cells were placed at 5
CE

locations which corresponded to distances 6.5 and 16.5m from the centre of the embankment and

at depths of approximately 3, 6 and 12m. They were directed in a manner by which the radial total
AC

horizontal stress could be obtained. (c) In each horizontal cell location, pore pressure transducers

were also installed in order to measure the excess pore pressure, and to obtain the effective

horizontal stress from the total horizontal stress. Other pore pressure transducers were also

installed. (d) A horizontal inclinometer was installed at level ground, along a radius of the

embankment prior to construction to measure the ground settlement versus horizontal location.

(e) A magnetic extensiometer was installed very near the center of the base of the embankment

prior to construction in order to measure the ground displacement at 2m increments at depths 0 to

20m. (f) Five settlement plates were placed near to the ground level at different locations of the

embankment base in order to measure the settlement by topographic methods, in order to verify

the settlement measured by the other methods. Fig 6a gives the detailed locations of these

11
 ACCEPTED MANUSCRIPT
instruments. Apart from the settlement plates, the measurements of all the above instruments,

were taken electronically using a data logging system.

T
3.3. Embankment construction and demolition

P
A truncated-cone-shaped preload earth fill was constructed. It was 9 m high and had

RI
50m diameter at the bottom and 13 m diameter at the crest . A ramp was also constructed in

SC
order to perform construction. Construction started on 6/6/2011. This date corresponds to day 0 in

all the graphs and days given below. The soil used to construct the embankment was sandy.

NU
Compaction of the layer was performed with a vibrator. Field density tests were performed in order

to verify compaction and illustrated that the unit density of the soil was 2.03t/m3.
MA

During the placement of the preload embankment, a slide occurred, on day 19. Figs 3b and 3c
ED

give representative photographs at the top and base of the embankment, respectively. Figs. 6b

and 6c give a topographic imprint and schematic illustration of the failure. Fortunately, the
PT

instruments were not damaged during the slide, as they were located in regions not affected by
CE

the slide.
AC

After the slide occurrence, part of the embankment was demolished and reconstructed.

Construction continued at a considerably slower rate. Construction terminated, after reaching an

embankment height of 9m from ground level after settlement, which corresponds to 8.54m above

unsettled ground level, on day 66. Fig. 3c gives a photograph of the embankment at top height.

Fig. 6a gives a cross-section of the embankment at top height. The embankment remained until

day 126 after the start of construction, or 60 days after construction. By then, the rate of

settlement was miniscule , less than 0.001m/day. The embankment was removed in 11 days. Figs.

7a and 7b give the height of the embankment and the corresponding construction rate, both given

versus the days from the start of construction.

3.4. Instrument measurements and discussion

12
 ACCEPTED MANUSCRIPT
Fig. 7c gives the measured vertical stress induced by the preload embankment, both in terms of

time and location. It can be observed that its change in terms of time follows the change in

embankment height. Fig. 8a and table 3 gives the measured maximum vertical stress versus

distance from the center of the cone.

P T
Regarding settlement measurements, it was first observed that settlement measurements of all

RI
devices (horizontal inclinometer, magnetic extensiometer, settlement plates) were consistent. Fig.

SC
7d gives the settlement versus time in terms of distance from the center of the cone measured by

the inclinometer. Figs. 8a and 8b and table 4 give the measured maximum and final settlement

NU
versus distance from the center of the cone. Fig. 4 gives the measured variation of maximum and

final vertical strain versus depth.

MA

In the field test, ground settlement was large, about 0.6m, illustrating the considerable level of
ED

densification, or ground improvement. Maximum settlement occurred at the centerline. The vertical

strain is maximum at the location of the soft clay layer (0-3.5m) and equals 10%. This is
PT

presumably because this layer (i) is near the surface and thus receives larger vertical stress from
CE

the preload embankment, (ii) is not pre-compressed and (iii) has a large coefficient of

compressibility, larger than the other layers which are sandy or silty. At the depths of 0-2m the
AC

vertical strain is less, presumably because the soft clay material was replaced at these depths by

the sandy soil from the preload embankment due to the slide which occurred during construction

as described above. At depths 3.5-15m the vertical strain equals about 1%.

The measured magnitude of settlement and the variation of settlement with the distance from the

centreline were similar to previous preload applications (Stamatopoulos et al., 2005,

Stamatopoulos and Kotzias, 1985). The time required for the soil to settle is also a critical factor in

applications of preloading (Stamatopoulos and Kotzias, 1985). The study illustrated that the

preload procedure, including the generation of the ground settlement (i.e. the settlement rate to

decrease to values less than 0.001m/day) occurred rather quickly, less than five months after the

start of preload construction, even though a clay layer of 3.5m width existed on the ground

13
 ACCEPTED MANUSCRIPT
surface. This is similar to observations of previous field tests in sites containing a considerable

amount of sandy material (Stamatopoulos and Kotzias, 1985, Petridis et al, 2000, Stamatopoulos

et al., 2005).

T
Fig. 7e gives the pore pressure both in terms of time and location measured by the pore pressure

P
transducers. It can be observed that significant excess pore pressures occur only in devices at

RI
depths of 2.5-3.2m. In devices at depths of 5.9-11.7m excess pore pressures are very small. This

SC
is consistent with the soil layers at the site, as, according to table 1, only until a depth of 3.5m

clayey soil exists. Furthermore, at depths of 2.5-3.2m it can be observed that the maximum excess

NU
pore pressures occurred when the rate of embankment construction was maximum, and more

specifically just prior to the failure of the embankment. This is very reasonable and explains failure,
MA
as described in detail below.
ED

Due to disturbance as a result of instrument placement, the initial measurement of horizontal

stress is not reliable. Therefore, only the measured change of horizontal stress due to
PT

embankment construction and demolition is considered. Fig. 7f gives the measured change in
CE

effective horizontal stress in terms of time and device. It can be observed that the measured

response follows that of load application: When loading is applied, the increase in horizontal stress
AC

was almost immediate and follows that of the curve of load application. During constant load

application, horizontal stress did not change considerably. When the load was removed, the

horizontal stress decreased simultaneously with load removal. After removal of the surcharge,

some horizontal stress remained and stayed more or less constant with time. Fig. 9 and table 5

give the measured maximum and final increase in effective horizontal stress due to embankment

construction, Δσ'h-max and Δσ'h-res , versus depth and distance. Fig. 5. gives the measured change in

Δσ'h-res normalized by the initial geostatic effective stress, equal to the measured change in the

coefficient of lateral pressure at rest (Δko-res), versus depth. The measurements indicate that

Δko-res decreases with depth: it equals 0.6 and 0.1 at depths 3m and 12m respectively. This is

presumably because the OCR value induced by the surcharge decreases with depth.

14
 ACCEPTED MANUSCRIPT

3.5. Post-Improvement soil properties and discussion

After the performance of the field test, investigations identical to the pre-improvement field were

performed in order to investigate the post-improvement soil properties. Post-improvement

geotechnical investigations were performed at a distance less than 2m from the corresponding

P T
locations of pre-improvement geotechnical investigations. Fig. 5 gives the average after soil

RI
improvement SPT, CPT and VS measurements versus depth. It can be observed that in almost all

SPT, CPT, VS separate locations versus depth, post-improvement values are larger than the

SC
corresponding pre-improvement values.Table 2 gives the average measured NSPT , qc and VS after

NU
soil improvement in all in-situ soil layers according to the geotechnical characterization and the

corresponding ratio of post to pre-improvement values. It can be observed that the ratio of post to
MA
pre-improvement values is maximum at the upper soft layer. This is presumably a result of the

maximum OCR value induced by preloading, in combination with minimum initial strength at this

layer. However, the increase in SPT, CPT and VS extend at the deeper silt layer up to 15m also,
ED

presumably because this layer is very soft (VS=197m/s, qc=4.3MPa) and the OCR value induced
PT

by preloading is still above 1.3. In particular, as a result of preloading the NSPT and qc values

increased by about 580% and 280% at depths 0-3.5m, by about 10% and 20% at depths 3.5-7m
CE

and by about 20% and 40% at depths 7-15m respectively. Furthermore as a result of preloading

the shear wave velocity increased by about 110% at depths 0-3.5m, by about 10% at depths 3.5-
AC

7m and by about 20% at depths 7-15m . The observation that SPT increases at depths larger

than 10m by more than 20% has been observed at other soft sites which have also been

preloaded (Kotzias-Stamatopoulos Co. 1982). The observation, whereby increases in VS values

more than 20% extend at least at depths up to 15m, has also been observed in a previous preload

field test (Stamatopoulos et al., 2005). Furthermore, it should be noted that the average Vs values

of table 2 hide interesting disparities as those at depths between 7 and 11 m and between 12 and

13 m where Vs pre-improvement values are smaller and their post-improvement increase in Vs is

larger.

3.6. Increase in liquefaction cyclic strength

15
 ACCEPTED MANUSCRIPT

Under conditions of earthquakes, only the lower two layers of table 1, are susceptible to

liquefaction. The upper layer consists of clay of considerable plasticity, and thus is not susceptible

to liquefaction. For the silty sand layer, based on the NSPT and qc measurements the liquefaction

T
cyclic strength is estimated using the state-of-the-art procedures described by Idriss and

P
Boulanger (2004). Based on VS measurement, it was estimated using the relationship given by

RI
European Prestandard (1994). For the non-plastic silt layer, the liquefaction cyclic strength can be

SC
correlated only to the NSPT, according to propositions by Boulanger and Idriss (2006). Table 6

gives the pre- and post- improvement average SR15 , in terms of the measurement used, for the

NU
two soil layers that liquefy. The cyclic liquefaction strength of a silty sand layer at depth 3.5-7m

increased from 0.39-0.50 to 0.46-0.55, or by about 10%. In addition, the cyclic liquefaction
MA
strength of a non-plastic silt layer at depth 7-15m increased from 0.38 to 0.43, or by about 13%. It

can be observed that all field procedures generally produce similar results. The range of variation
ED

of the results is consistent with the non-homogeneous characteristics of soils and the different

methods applied.
PT
CE

4. Predictions of maximum and final measured stresses and displacements

AC

4.1. Description of the elasto-plastic analysis performed

The FLAC (Itasca Consulting Group, 2001) code was used in the elasto-plastic drained analysis.

The assumed drained conditions were verified by the fact that both (i) after embankment

construction, just prior to embankment demolition, and (ii) finally, 50 days after embankment

demolition, excess pore pressures induced by preloading are small: less than 10% the total

vertical stress according to measurements. The soil below the cone was modeled using the

Modified Cam Clay model. The model parameters used per soil layer are those given in table 1. In

addition, a Poisson ratio (ν) of 0.33 was assumed, the elastic shear modulus (G) was taken

according to the pre-improvement VS values of Fig. 3 using eq. (2a) and the maximum past

vertical consolidation stress, Pp was taken according to Fig. 4a. The material of the cone was

modeled assuming isotropic elastic response with model parameters with G and ν equal to

16
 ACCEPTED MANUSCRIPT
20000kPa and 0.3 respectively. The water table was taken, according to the results of the

geotechnical investigation, at a depth of 1.0m. At depths 0-1m the Cc and Cr values were reduced

by 50%. The reason being that the post-improvement borings illustrated that sandy material

replaced the clay material at these depths, presumably due to the embankment stability failure.

T
The depth of the mesh extended 50m below the soil surface, and 100m horizontally from the

P
embankment center. A total of 15,000 axisymetric elements were used to simulate the underlying

RI
soil.

SC
In the numerical analyses, firstly, the initial geostatic state of stress was assigned. The initial

NU
vertical effective stress versus depth was estimated from the soil density, given in table 1 and the

water table depth. The initial horizontal effective stress was estimated using equation (5) and the
MA
measured friction angles given in table 1. Then, cone construction was simulated in five steps,

each with a height of about 2m, by progressively activating rows of elements on which the
ED

embankment is formed. Finally, cone demolition was simulated in five steps, each with a height of

about 2m, by progressively de-activating rows of elements of the embankment.

PT
CE

4.2. Description of the simplified analysis performed

Regarding the simplified method, predictions of the vertical and horizontal stress induced by the
AC

surcharge were estimated by linear elasticity, with the aid of the code FLAC. A shear modulus

equal to 30000 kPa, which corresponds to about half the small-strain shear modulus, was taken

for the in-situ soil. In addition, the Poisson's ratio was taken equal to 0.33. The reason is that long-

term drained conditions are simulated and compared with measurements both after embankment

construction and after embankment demolition. Furthermore it should be noted that a parametric

analysis was performed using different Poisson ratios, but this did not give an overall improvement

in the predictions of stresses. The cone properties were simulated using the parameters used in

the elasto-plastic analyses, as described above. The considered geometry and the cone

construction simulation were identical to those of the elasto-plastic analysis described above.

However, as the analysis was elastic, the initial stresses, prior to the embankment construction,

were not specified in the analysis. From the computed distribution of vertical stresses induced by

17
 ACCEPTED MANUSCRIPT
the surcharge, settlements were estimated using eqs. (4). Similarly to the elasto-plastic analysis,

at depths 0-1m, the Cc and Cr values were reduced by 50%. In addition, equation (6a) was used

to estimate the residual increase in horizontal stress induced by preloading.

T
4.3. Results and discussion

P
The computed maximum vertical stress at level ground below the preload embankment

RI
(Δσv-max) is given in Fig. 8a for both the elastic and elasto-plastic models. Table 3 gives the

SC
computed values at the locations of the measurements. The computed settlement induced by the

surcharge, (δmax) and after its removal (δres) versus location is given in Figures 8b and 8c for both

NU
the elasto-plastic and simplified methods. Table 4 gives the computed values at horizontal

distances r=0, 6.5 and 16.5 from the center of the cone. Fig. 4 gives the computed vertical strain
MA
versus depth at the centerline for both the elasto-plastic and simplified methods. Fig. 9 gives the

increase of horizontal total stress induced by the surcharge (Δσ'h-max) and after its removal (Δσ'h-res)
ED

at distances r=0,6.5, 16.5m from the centerline by both (i) the elasto-plastic and (ii) the simplified

methods. Table 5 gives the computed values at the locations of the measurements.
PT
CE

The predicted maximum vertical stress near the surface using the elasto-plastic and simplified

methods is in accordance with the measured values: the ratio of predicted to measured values
AC

equals 0.96-1.00 in all locations. The maximum and final settlements predicted by the elasto-

plastic method are smaller than those measured at the centerline and larger at the foot of the

embankment. On the other hand, the simplified method predicts more settlement than the one

measured. However, both methods predict settlement within +/-15% of the measured ones at all

locations. The ratio of predicted by measured vertical strain versus depth equals 0.7-1.5 for the

elasto-plastic method and 0.7-1.5 for the simplified method at all depths at the 2m increments of

the measurements. The ratio of predicted to measured maximum horizontal stress of the simplified

and elasto-plastic methods are 0.6-0.9 and 0.7-0.12 in all locations respectively. Regarding the

final horizontal stress, which is of much more interest than the maximum one as it can be related

to changes in dynamic soil properties, the ratio of computed by measured values of the simplified

and elasto-plastic methods are 0.7-1.2 and 1.1-1.7 respectively

18
 ACCEPTED MANUSCRIPT

The above illustrate that the Modified Cam-Clay Model under the assumption of drained conditions

did not provide an improvement to the accuracy of the predictions of the ground settlement and

vertical strain versus depth of the simplified method and predicted the residual increase in

T
horizontal stress induced by preloading with less accuracy. A previous analysis of a preload field

P
test by Stamatopoulos et al. (2005) gave similar conclusions. The reason for the inaccurate

RI
predictions of the change in horizontal stresses as a result of preloading by the Modified Cam clay

SC
model under drained conditions can be attributed to the facts that (a) the initial undrained

response of the soil due to embankment construction are not simulated and (b) this often-used

NU
model is simplified, does not require many model parameters and does not have the capabilities of

more complex recent soil models (Whittle and Kavvadas, 1994). It is inferred that for regular
MA
preload projects the use of Modified Cam-Clay Model under the assumption of drained conditions

is not recommended.
ED

5. Prediction of the embankment failure

PT

o
CE

Referring to Fig. 1a, as in the geometry of the field test i=34 and D=1.5, the factor A equals 0.17

and thus equation (3) becomes

AC

FSt-i  18 (σv(t-i) -Pt-i ) /(γ ht ) =18 [ σ'g + (Δσv-max-s-i ht / (9m) ) - Pt-i ] /(γ ht ) (13)

The factor of safety FSt-i is determined in the locations of the transducers at the upper clay layer

versus time using equation (13), Δσv-max-s estimated previously and the measured pore water

pressure. It is presented in Fig. 1b. It can be observed that FSt-i (a) has a minimum value in all

piezometer locations at the time near failure (or t=19days) and (b) at the time of failure at the

piezometers at depths 2.5 and 2.8m it is less than one. It should be noted that the undrained soil

strength predicted by equation (2) and used in equation (3) is also in general agreement with the

undrained strength of the upper clay measured in samples retrieved from borings in triaxial tests,

19
 ACCEPTED MANUSCRIPT
also taking into account soil anisotropy which reduces soil strength by about 15% (Mayne, 1985).

Concluding, equation (3) predicts the embankment failure.

6. Predictions of the measured increase of dynamic soil properties Evaluation of equations

P T
At the studied site, table 7 gives (a) the measured pre-improvement average void ratio and OCR

RI
value per soil layer, (b) the estimated value of these quantities after the application of preloading

SC
using the simplified method described above and (c) the average ko value both before and after

the application of preloading estimated using the simplified method. Based on these values, table

NU
7 gives the predicted ratio (VS-after/VS-bef) using proposed equation (8) and compares it with the

measured one for all three layers. Equation (8) predicts, with error less than 23%, the ratio of post
MA
by pre-improvement shear velocity. More accurate predictions are observed at the two deeper

layers with VS-bef value greater than 190m/s, where the error in the post-improvement shear
ED

velocity is less than 13%. Closer inspection illustrates that eq. (8) underpredicts the measured

improvement: the increase in shear wave velocity in the top layer is underpredicted by 115% and
PT

at the two deeper layers by 10 and 25%.

CE

Furthermore, it is investigated whether equations (9), (10) and (12) predict the measured increase
AC

in cyclic liquefaction strength induced by preloading. As illustrated in table 6 in both layers that

liquefy (a) proposed equation (12) predicts, with error less than 3%, the ratio of post by pre-

improvement liquefaction strength and (b) both equations (9) and (10) predict the ratio of post by

pre-improvement liquefaction strength with error less than 5%

It is inferred that the field test and its analysis verified that equations (8), (10) and (12) can be

used to evaluate the effect of preloading on the shear wave velocity and cyclic liquefaction

strength by using the proposed simplified methods to predict the effect of preloading on the void

ratio and horizontal stress. Furthermore, it was found that the empirical equation (9), which

provides an easy estimate of the anticipated increase in cyclic strength induced by preloading,

also adequately predicts the measured increase in cyclic liquefaction strength.

20
 ACCEPTED MANUSCRIPT

7. Conclusions

This paper presents a well-organized experiment which provides quantitative results of the effect

T
of preloading in many geotechnical and geophysical parameters. The site consisted of (a) a soft

P
clay layer to depth of 3.5m, (b) a medium-dense silty sand layer at depths 3.5-7m and (c) a soft silt

RI
layer below. Preloading was applied by a temporary embankment 9m high. Measurements

SC
extended to 15m depth. A partial embankment failure occurred during the preloading process. The

experiment illustrated that: (a) Preloading caused settlement of about 0.6m with vertical strain

NU
ranging from 10% at depths above 3.5m to 1% below and till 15m depth. (b) The increase in lateral

stress ratio as a result of the preloading process decreased with depth and equals 0.6 and 0.1 at
MA
depths 3 and 12m respectively. (c) As a result of preloading the shear wave velocity increased by

about 110% at depths 0-3.5m, by about 10% at depths 3.5-7m and by about 20% at depths 7-
ED

15m. (d) The cyclic liquefaction strength of (i) the silty sand layer increased from 0.39-0.50 to

0.46-0.55, or by about 10% and (ii) the non-plastic silt increased from 0.38 to 0.43, or by about
PT

13%.
CE

Predictions of the measured response illustrated that: (i) commonly-used simplified methods
AC

described in section 4.2 predict, with error less than 15%, the settlement induced by preloading,

and with error less than 30%, the increases of horizontal stress induced by preloading, (iii)

proposed equation (8), predicts, with error less than 23%, the ratio of post by pre-improvement

post-improvement shear wave velocity in terms of the change in void ratio and horizontal stress

and (iv) proposed equation (12) and equations (9) and (10) predict the ratio of post by pre-

improvement liquefaction strength with error less than 3%, 5% and 5% respectively in both layers

that liquefy and (v) utilizing the measured excess pore pressures versus time, equation (3)

predicts the stability failure of the embankment. Numerical simulations using the Modified Cam

Clay model under drained conditions did not improve the accuracy of the settlement and horizontal

stress induced by preloading predictions of the simplified methods and thus its use for such

predictions is not recommended.

21
 ACCEPTED MANUSCRIPT

Acknowledgement

The work was funded by the Seventh Framework Programme of the European Community,

T
European Commission Research Executive Agency under grant agreement FP7-SME-2010-1-

P
262161-PREMISERI. Prof. Arezou Modaressi and Dr Fernando Lopez-Caballero assisted in the

RI
interpretation of the field test.

SC
References

NU
Adalier K. and Elgamal A. 2004. Mitigation of liquefaction and associated ground deformations by

stone columns. Journal of Engineering Geology. Volume 72, Issues 3-4, Elsevier, April 2004,
MA
Pages 275-291.
ED

Alonso E. E., Gens A, Lloret A. 2000. Precompression design for secondary settlement reduction,
PT

Geotechnique, 50, No 6, 645-656.

CE

Al-Shamrani M. A. , Dhowian A. W. 1997. Preloading for reduction of compressibility

AC

characteristics of sabkha soil profiles. Journal of Engineering Geology. Volume 48, Issues 1–2, 19

November, Pages 19–41

Bell F.G. 1996. Lime stabilization of clay minerals and soils. Journal of Engineering Geology.

Volume 42, Issue 4, July, Pages 223–237

Boulanger R. W. and Idriss I. M. 2006. Liquefaction Susceptibility Criteria for Silts and Clays. the

Journal of Geotechnical and Geoenvironmental Engineering, Vol. 132, No. 11, November.

22
 ACCEPTED MANUSCRIPT
Cai Y., Shi B., Ng C. W.W, Tang C. 2006. Effect of polypropylene fibre and lime admixture on

engineering properties of clayey soil. Journal of Engineering Geology. Volume 87, Issues 3–4,

November, Pages 230–240

P T
Du Y., Li S., Hayashi S. 1999. Swelling–shrinkage properties and soil improvement of compacted

RI
expansive soil, Ning-Liang Highway, China. Journal of Engineering Geology. Volume 53, Issues

SC
3–4, July, Pages 351–358

NU
European Prestandard. 1994. Eurocode 8 - Design provisions of earthquake resistance of

structures, Part 5: Foundations, retaining structures and geotechnical aspects.

MA

European Standard. 2003. Eurocode 8: Design of structures for earthquake resistance, Final
ED

Draft, prEN 1998-5, December.

PT

Idriss I. M. and R. W. Boulanger. 2004. Semi-empirical procedures for evaluating liquefaction

CE

potential during earthquakes. Invited Paper Presented at The Joint 11th International Conference

on Soil Dynamics & Earthquake Engineering (ICSDEE) and The 3rd International Conference on
AC

Earthquake Geotechnical Engineering (ICEGE), January 7 – 9, Berkeley, California, USA

Itasca Consulting Group, Inc. 2001. Fast Langrangian Analysis of Continua, Use’s Guide.

Minneapolis, Minessota, USA.

Ishihara K. and Takatsu H. 1979. Effects of overconsolidation and Ko conditions on the

liquefaction characteristics of sands. Soils and Foundations, Japanese Society of Soil Mechanics

and Foundation Engineering, 19, No 4, pp 59-68.

23
 ACCEPTED MANUSCRIPT
Kotzias-Stamatopoulos Co. 1982. Report of second phase geotechnical investigation after

preloading. Hellenic Ferroalloys S. A.. Almyros plant. Raw materials storage building. Athens,

Greece, August.

T
Kramer S. L. 1996. Geotechnical earthquake engineering. Prentice Hall, New Jersey.

P
RI
Jamiolkowski, M., Lerouell, S. and Lo Presti, D.C.F. 199). "Design parameters from theory to

SC
practice," Theme Lecture, Geo-Coast '91, Yokohama, pp. 877-917.

NU
Ladd C. C. Stability evaluation during staged construction (The Twenty-Second Karl Terzaghi

Lecture), Journal of Geotechnical Engineering, ASCE, Vol. 117, No. 4, April, 1991
MA

Mayne P. W. and Kulhway F. W. 1982. Ko-OCR relationships in Soils. Journal of the Geotechnical
ED

Engineering Division, ASCE, 108, No 6, pp 851-872.

PT

Mayne Paul W. 1985. Stress Anisotropy Effects on Clay Strength. Journal of Geotechnical
CE

Engineering, Vol. 111, No. 3, March, pp. 356-366

AC

Petridis P., Stamatopoulos C. and Stamatopoulos A. 2000. Soil Improvement by preloading of two

erratic sites. GeoEng2000, International conference on Geotechnical and Geological

Engineering, Melbourne, Australia (in CD-ROM)

Raptakis D. G. 2012. Pre-loading effect on dynamic soil properties: Seismic methods and their

efficiency in geotechnical aspects. Soil Dynamics and Earthquake Engineering. Volume 34, Issue

1, March, Pages 69–77

Roscoe K. K. and Burland J. B. (1968). On the generalised stress-strain behavior of "wet" clay, in

J. Heyman and F. A. Leckie (eds.). Engineeding plastisity (Cambridge:Cambridge University

Press), pp. 535-609.

24
 ACCEPTED MANUSCRIPT

Stamatopoulos A. C., Kotzias P. C. 1985. Soil improvement by preloading. J. Wiley & Sons

Publications.

T
Stamatopoulos C. Petridis P., Bassanou M., Stamatopoulos A. 2005. Increase in horizontal stress

P
induced by preloading. Ground Improvement, 9 (2), pages 47-58.

RI
SC
Stamatopoulos, C.A , Fernando Lopez-Caballero, Arezou Modaressi-Farahmand-Razavi l. 2012

Laboratory tests and numerical simulations giving the effect of preloading on the cyclic liquefaction

NU
strength, 15WCE, Losboa, Portugal
MA
Taylor D.W. 1948. Fundamentals of soil mechanics. John Wiley and Sons Inc. New York
ED

Whittle A. J. and Kavvadas M. J. 1994. Formulation of MIT-E3 constitutive model for

overconsolidated clays. Journal of Geotechnical Engineering, ASCE, 120, No 1, pp 173-198.

PT
CE

Notation

A Factor of eq. (1)

AC

Cc, Cr Virgin and swelling coefficients of compressibility

CPT Cone Penetration Test

d Depth below the ground level

e Void ratio

eL, eU Void ratio during the loading and unloading phases of preloading respectively

FS Factor of safety against embankment failure

g Acceleration of gravity

Gmax Shear modulus at very small strains

h Embankment height

Hi Horizontal pressure cell “i”

ko The coefficient of lateral pressure at rest

25
 ACCEPTED MANUSCRIPT
Ki Settlement plate “i” (i=1 to 5)

NSPT Blow count number of the SPT

OCR Overconsolidation Ratio

P Pore pressure

T
Pa Atmospheric pressure

P
PI Plasticity Index

RI
Pp Maximum past vertical consolidation stress

SC
qc Point resistance of the CPT

PR Prestress ratio defined by equation (9b)

NU
r Horizontal distance from the center of the conical embankment

SR15 Liquefaction cyclic strength

MA
Su Undrained soil strength

Sufail Limit undrained soil strength for failure

ED

SPT Standard Penetration Test

Ti Pore pressure tranducer “i”

PT
CE

VS, VS1 Shear wave velocity and normalized Shear wave velocity

Xi Vertical pressure cell "i"

AC

R SR15-after/ SR15-bef

Greek letters

γ Unit weight

δ Settlement

Δ Change

εv Vertical strain

σh Horizontal stress

σv Vertical stress

σoct Octahedral stress

σg Geostatic stress

φ' Effective friction angle

26
 ACCEPTED MANUSCRIPT
Subscripts

after After the application of preloading

bef Before the application of preloading

ep Predicted by the elasto-plastic method

T
i Index of device/position

P
M Measured

RI
max Maximum value (when all the surcharge is applied)

SC
Pr Predicted

res Residual value (after the application of the preloading process)

NU
s Predicted by the simplified method

t In terms of time
MA

Superscripts
ED

' Effective
PT
CE
AC

27
 ACCEPTED MANUSCRIPT
Table 1. Soil layers that exist in the site and their properties.
Depth Layer PI Fines Initial Total Density φ' (assuming
description content Cc Cr void [t/m3] c=0)
ratio
0-3.5m Silty Clay 30% 77 0.52 0.02 1.40 1.66 28
3.5-7m Medium Gravel <5% 15 0.03 0.007 2.02 39
with silty sand 0.58
7m-15m Fine Sand and Silt <5% 95 0.04 0.015 0.64 2.07 25

T
Table 2. Average measured NSPT ,qc and VS before and after soil improvement
in terms of soil layer and corresponding ratio.

P
Bef. (Pre-improvement) After (Post-improvement) Ratio

RI
Depth NSPT qc VS NSPT qc VS NSPT qc VS
(MPa) (m/s) (MPa) (m/s)
(a) (b) (c) (d) (e) (f) (d)/(a) (e)/(b) (f)/ (c)

SC
0-3.5m 0.6 0.4 94 3.5 1.01 202 5.8 2.5 2.1
3.5-7m 21.7 10.0 192 23.6 11.8 212 1.1 1.2 1.1
7-15m 20.9 4.33 197 25.9 6.11 246 1.2 1.4 1.2

NU
Table 3. Measured (M) and predicted with the elasto-plastic method (ep) and the simplified or
elastic (s) method maximum vertical stresses induced by the surcharge near ground level and
corresponding ratios of predicted by measured values.
Δσν–max-M Δσν-max--ep: Δσν-max-s:
MA
Device r (m) Ratios of predicted
(kPa) (kPa) (kPa) versus measured
No (a) (b) (c) (b/a) (c/a)
1 1.5 160.00 160.29 153.20 1.00 0.96
2 6.5 150.00 148.09 144.90 0.99 0.97
ED

3 16.5 130.00 125.08 127.58 0.96 0.98

Table 4. Measured and predicted settlements induced by preloading and corresponding ratios.
PT

Location Measured Predicted by the Predicted by Ratios of

elasto-plastic the simplified Predicted vs measured
method method
δmax-M δres-M: δmax-ep δres-ep δmax-s δres-s
CE

r (m)
(m) (m) (m) (m) (m) (m)
(a) (b) (c) (d) (e) (f) (c/a) (d/b) (e/a) (f/b)
0 0.61 0.51 0.56 0.44 0.63 0.55 0.92 0.86 1.03 1.08
AC

6.5 0.6 0.48 0.53 0.44 0.60 0.52 0.88 0.92 1.00 1.08
13 0.5 0.4 0.47 0.35 0.53 0.44 0.92 0.86 1.03 1.08

Table 5. Measured and predicted increase of horizontal effective stress due to the preloading
process and corresponding ratios.
Location of Measured Predicted by the elasto- Predicted by the Ratios of
measurement plastic method simplified method Predicted vs measured
d (m) r (m) Δσ'h-max-M Δσ'h-res-M Δσ'h-max-ep Δσ'h-res-ep Δσ'h-max-s Δσ'h-res-s
(kPa) (kPa) k(Pa) (kPa) (kPa) kPa
(a) (b) (c) (d) (e) (f) (c/a) (d/b) (e/a) (f/b)
11.6 16.5 32 15 22 20 22 12.0 0.69 1.33 0.69 0.80
6.40 6.5 70 30 53 32 40 23.0 0.76 1.07 0.57 0.70
5.85 16.5 40 9 47 15 35 8.0 1.18 1.67 0.88 0.78
3.25 6.5 85 18 60 26.7 60 22.4 0.71 1.48 0.71 1.22
2.50 16.5 50 14 40 21.6 42 12.0 0.80 1.54 0.84 0.86

28
 ACCEPTED MANUSCRIPT
Table 6. (a) Pre- and post-improvement cyclic liquefaction strength (SR15-bef and SR15-after) based
on field measurements (NSPT,qc, VS) and corresponding ratio R(=SR15-after/ SR15-bef). (b) Predicted R
values using equations (9), (10) and (12), and corresponding ratios of predicted by measured
values.
(a)

SR15-bef SR15-after RM
Depth NSPT qc VS NSPT qc VS NSPT qc VS Ave

T
3.5-7m 0.50* 0.42 0.39 0.55 0.48 0.46 1.10 1.14 1.19 1.14

P
7-15m 0.38 - - 0.43 - - - - 1.13
*All values given in the table correspond to cyclic liquefaction strength (SR15 ), so they are

RI
dimensionless
(b)

SC
Depth RPr RPr / RM
Eq. Eq. Eq. Eq. Eq. Eq.
(9) (10) (12) (9) (10) (12)
3.5-7m 1.09 1.20 1.13 0.96 1.05 0.99

NU
7-15m 1.07 1.15 1.09 0.95 1.01 0.97
MA
Table 7. OCR value, ko factor and void ratio before and after soil improvement. Measured values
are given underlined. The other values were estimated using the proposed simplified method.
Corresponding predicted (Pr) ratio (VS-after/VS-bef ) according to equation (8) and ratio of measured
by predicted value are also given.
Depth ebef eafter ko-bef ko-after OCRbef OCRafter VS-after/VS-bef
ED

Pr. eq. (8) Pr./Meas

0-3.5m 1.4 0.99 0.59 1.46 1.26 8.02 1.69 0.77
3.5-7m 0.58 0.56 0.45 0.84 1.36 3.47 1.06 0.97
PT

7m-15m 0.64 0.63 0.43 0.65 1.00 2.22 1.05 0.87

CE
AC

29
 ACCEPTED MANUSCRIPT

Figure captions

Fig. 1. (a) Stability chart giving the undrained soil strength for embankment failure (Modified from

Taylor, 1948), (b). Porto Romano field test. Factor of safety versus time at the locations of the

P T
pore pressure transducers using eq. (13).

RI
SC
Fig. 2. (a) The relationship between the normalized shear wave velocity (VS1) and the liquefaction

cyclic strength SR15, according to Eurocode (European Prestandard, 1998) and (b) eq. (11b)

NU
predicting this relationship. MA
Fig. 3. Porto Romano field test. (a) General location of the site, (b)-(c) Photographs at the top and

base of the embankment illustrating the failure at construction day 19, (d) Photograph of the
ED

embankment at top height.

PT

Fig. 4. Porto Romano field test. (a) Initial (σ'g), maximum (σ'g+Δσv-max) and maximum past (Pp)

vertical effective stress versus depth. (b). Maximum and final vertical strain versus depth
CE

measured using the magnetic extensiometer and predictions.

AC

Fig. 5. Porto Romano field test. Average (a) NSPT value of the SPT, (b) qc resistance of the CPT

and (c) VS before and after soil improvement versus depth. (d) Measured final change in radial

effective horizontal stress induced by preloading normalized by the initial geostatic effective stress.

Fig. 6. Porto Romano field test. (a) Cross-section of the embankment used for the preloading

process at maximum height. The location of the settlement plates (Ki), vertical pressure cells (Xi),

pore pressure transducers (Ti) and horizontal pressure cells (Hi) is also given. (b) Topographic

imprint of the failure of the embankment at construction day 18. (c) Schematic illustration of

cross-section II of the topographic imprint.

30
 ACCEPTED MANUSCRIPT

Fig. 7. Porto Romano field test. (a) The height of the embankment, (b) the corresponding

construction rate, (c) the measured vertical stresses just below the embankment, (d) the measured

settlement, (e) the measured excess pore pressures and (f) the measured changes in effective

T
horizontal stress all versus the days from the start of construction.

P
RI
Fig. 8. Porto Romano field test. Measured and computed (a) maximum vertical stress and (b)

SC
maximum and final settlement in terms of distance from centre of the embankment.

NU
Fig. 9. Porto Romano field test. Computed by (i) the elasto-plasic and (ii) the simplified methods

versus the measured (a) maximum and (b) final change in radial horizontal stresses induced by
MA
the surcharge.
ED
PT
CE
AC

31
 ACCEPTED MANUSCRIPT

Figures

T
P
RI
SC
NU
MA
ED
PT
CE
AC

32
 ACCEPTED MANUSCRIPT

(a) (b)

2.0

P T
1.5

RI
FSt-i
SC
1.0

NU d=2.8m, r=1.5m
d=3m, r=6.5m
MA
d=-2.5m, r=16.5m
0.5
0 20 40 60 80 100 120
Days from start
ED

Fig. 1.
PT
CE
AC

33
 ACCEPTED MANUSCRIPT

(a) (b)
0.5
0.7
0.4 0.6
0.5

T
0.3

(SR15)0.5
SR15

0.4 y = 0.0028x

P
0.2 R2 = 0.979
0.3

RI
0.1 0.2
Curve Fig.2a)

SC
0.1
Linear (Curve Fig.2a))
0 0
0 50 100 150 200 250 0 50 100 150 200 250

NU
Vs1 (m/s) MA Vs1 (m/s)

Fig. 2.
ED
PT
CE
AC

34
 ACCEPTED MANUSCRIPT

(a) (b)

P T
RI
SC
(c) (d)

NU
MA
ED
PT

Fig. 3.
CE
AC

35
 ACCEPTED MANUSCRIPT

(a)
σ'g (kPa)
Vertical stress (kPa)

300
σ'g+Δσv-max (kPa)
Pp (kPa)
200

PT
100

RI
0

SC
0 5 10 15 20
Depth (m)
(b) (c)

NU
0.16 0.16

Final vertical strain

Max. vertical strain

Measured Measured
0.12 0.12 Computed by elasto-
Computed by elasto-
MA
plastic method plastic method
0.08 0.08 Computed by simplified
Computed by method
0.04 simplified method 0.04

0.00 0.00
ED

0 5 10 15 20 0 5 10 15 20

Depth (m) Depth (m)

PT

Fig.4.
CE
AC

36
 ACCEPTED MANUSCRIPT

50 25
Before-Average

qc (MPa)
40 Before - Average
After-Average 20

T
After - Average
NSPT

30 15

P
20 10

RI
10 5
0 0

SC
0 5 10 15 20 0 5 10 15 20
Depth (m) Depth (m)
300 0.8

NU
Δko =Δσ'h/σ'g
250 Measured r=6.5m
Measured r=16.5m
Vs (m/s)

0.6
200
150 0.4
MA
100
0.2
50 Before-Average
After-Average 0.0
0
0 5 10 15 20 0 5 10 15 20
ED

Depth (m) Depth (m)

PT

Fig. 5.
CE
AC

37
 ACCEPTED MANUSCRIPT
z

(a) (b)
13m
50m 9m
K3K1 K4 K5
K2 X1 X2 X3 x
T7 H8,T8 H9,T9

T4 H5,T5 H6,T6

T
T1 H2,T2 T3

(c)

P
RI
h=6.35m 1
1.5

h=3.5m 43m

SC
NU
Fig. 6.
MA
ED
PT
CE
AC

38
 ACCEPTED MANUSCRIPT

(a) (b)
10

Construction Rate
1.0
Embankment

8 0.5
height (m)

0.0

T
(m/day)
6 -0.5
-1.0
4
-1.5

P
2 -2.0
-2.5
0

RI
-3.0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
Days from start Days from start

SC
(c) (d)
d=0.5m, r=1m 0.8
Mesured vertical

d=0.5m, r=6.5m r=0 r=18m r=21m

settlement (m)
200
stress (kPa)

d=0.5m, r=11.5m 0.6

160

NU
120 0.4
80
0.2
40
0 0.0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
MA
Days from start Days from start

(e) (f)
100
Pressure (kPa)

90
Excess Pore

Δσ'h (kPa)

70 80
ED

50 60
30 40
10 20

-10 0
PT

0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
Days from start Days from start
d=11.7m, r=1.5m d=11.7m, r=6.5m d=11.6m, r=16.5m d=11.6m, r=16.5m d=6.4m, r=6.5m
d=6.8m, r=1.5m d=6.4m, r=6.5m d=5.9m, r=16.5m d=5.8m, r=16.5m d=3.2m, r=6.5m
d=2.5m, r=16.5m
d=2.8m, r=1.5m d=3.2m, r=6.5m d=2.5m, r=16.5m
CE

Fig. 7.
AC

39
 ACCEPTED MANUSCRIPT

(a) Vertical stress (kPa)

200
160
120
80
40

T
0

P
0 4 8 12 16 20 24
Distance from centre (m)

RI
Measured
Computed by elasto-plastic method
Computed by simplified method
(b)

SC
Final settlement (m)
Max. Settlement (m)

0.8 0.8
0.6 0.6

NU
0.4 0.4
0.2 0.2
0.0 0.0
MA
0 4 8 12 16 20 24 0 4 8 12 16 20 24
Distance from centre (m) Distance from centre (m)
Measured
Computed by elasto-plastic method Measured
Computed by simplified method Computed by elasto-plastic method
Computed by simplifid method
ED

Fig. 8.
PT
CE
AC

40
 ACCEPTED MANUSCRIPT

(ai) (aii)
100 100

Δσ'h-max (kPa)
Δσ'h-max (kPa)

80 80
60 60
40 40

T
20 20
0

P
0
0 5 10 15 0 5 10 15
Depth (m)

Measured r=6.5m
Elasto-plastic r=0m
Elasto-plastic r=16.5m
RI
Depth (m)
Measured r=16.5m Measured r=6.5m Measured r=16.5m
Elasto-plastic r=6.5m Simplified (Elastic) r=0m Simplified (Elastic) r=6.5m
Simplified (Elastic) r=16.5

SC
(bi) (bii)
100 100
Δσ'h-res (kPa)

80 80

Δσ'h-res (kPa)
NU
60 60

40 40

20 20
MA
0 0
0 5 10 15 0 5 10 15
Depth (m) Depth (m)
Measured r=6.5m Measured r=16.5m Measured r=6.5m Measured r=16.5m
Elasto-plastic - r=0 Elasto-plastic - r=6.5m Simplified - r=0 Simplified - r=6.5m
Elasto-plastic - r=16.5m Simplified r=16.5m
ED

Fig. 9.
PT
CE
AC

41
 ACCEPTED MANUSCRIPT
Highlights

- The authors agreed with all changes suggested by the reviewers

- The paper has been improved according to these changes
- These changes are described in detail in the corresponding section.

P T
RI
SC
NU
MA
ED
PT
CE
AC

42