Chapter 27 Case histories of embankments on soft soils and stabilisation with geosynthetics: Canadian experience

Chapter 27 Case Histories of Embankments on Soft Soils and Stabilisation with Geosynthetics: Canadian Experience C. T. Gnanendran ~, A. J. Valsangkar z, and R. Kerry Rowe 3

1School of Aerospace, Civil and Mechanical Engineering, University of New South Wales at ADFA, Canberra, Australia 2Department of Civil Engineering, University of New Brunswick, Fredericton, New Brunswick, Canada 3GeoEngineering Centre at Queen's-RMC, Department of Civil Engineering, Queen's University, Kingston, Ontario, Canada

ABSTRACT The behaviours of Hall's Creek test embankment constructed on an alluvium deposit in Moncton, New Brunswick, and the reinforced test embankment constructed on a soft compressible soil in Sackville, New Brunswick, where a high-strength polyester woven geotextile was used as basal reinforcement are discussed in this chapter. The soils at both these sites have the same geological depositional history. Performance monitoring included the instrumentation of the foundation soil with inclinometers, pneumatic piezometers, settlement plates, settlement augers and heave plates and the geosynthetic reinforcement with different types of strain gauges. Details of the layout, instrumentation, field performance and analyses for behaviour prediction are presented. The observed settlement response of the Hall's Creek embankment could be predicted satisfactorily using Bjerrum's consolidation plus delayed compression approach by considering the immediate, primary and secondary compression; the unusually high pore water pressures that existed even after a 3-year period could not be explained adequately. From further laboratory investigation coupled with the pore pressure and settlement responses, it was concluded that progressive failure of the foundation soil could have been a contributory factor for the observed behaviour. The predictability of the behaviour of the Sackville reinforced embankment under working stress conditions using three types of fully coupled finite element analysis models; namely, a rate-formulated elasto-viscoplastic model with an elliptical cap yield surface, a creep-formulated elasto-viscoplastic model and modified cam clay (MCC) elastoplastic material model 787

788

Chapter 27

for the foundation soil is examined in this chapter. This study suggests that all three FEA models are capable of predicting the performance of this reinforced embankment under working stress conditions reasonably well despite their inability to give accurate predictions of all the behaviour characteristics. The analysis with the creep model gave slightly better overall predictions and that with the rate model predicted the horizontal displacement near the embankment toe and excess pore pressure in the foundation soil better than the MCC model. However, the creep and rate models require additional soil parameters and consume much larger computing resources and longer time. The MCC model could be adequate for predicting the performance of embankments on Sackville-type foundation soils under working stress conditions.

1. INTRODUCTION Soft soils are widely distributed in Canada and other parts of the world and are difficult materials for constructing road embankments due to their low strength, high compressibility and highly non-linear, time-dependent visco-plastic characteristics. Embankments constructed on such soft soils undergo large settlement and lateral deformation during and after construction resulting in a variety of construction and instability problems. Several well-instrumented road embankments have been constructed on soft soils to study their behaviour (La Rochelle et al., 1974; Leroueil et al., 1978a, 1978b; Ortigao et al., 1983; Keenan et al., 1986; Indraratna et al., 1992; Rowe et al., 1995; Crawford et al., 1995; Hussein and McGown, 1998; Bergado et al., 2002; and many others). Two Canadian case histories of such embankments in Eastern Canada that were constructed until failure are discussed in this chapter. The Hall's Creek test embankment was constructed on an alluvium deposit in Moncton, New Brunswick and a reinforced test embankment was constructed on a soft compressible soil in Sackville, New Brunswick where a high-strength polyester woven geotextile was used as basal reinforcement. Performance monitoring included the instrumentation of the foundation soil with inclinometers, pneumatic piezometers, settlement plates, settlement augers and heave plates and the geosynthetic reinforcement with three different types of strain gauges. Details of the layout, instrumentation, field performance and analyses for behaviour prediction are presented. From the available geological evidence, it is concluded that the soils at the Hall's Creek and Sackville site can be characterised as marine interdial deposits formed during the post-glacial period to the present time. These interdial deposits are due to coastal submergence and the erosional and depositional actions of the Bay of Fundy. Case histories generally provide the basis to validate theories and assumptions used in the design and performance prediction. The rheological properties of soft soils and the engineering characteristics of the geosynthetics are such that predicting various aspects of the behaviour of the embankment as a function of time is very complicated. However, significant advances have been achieved recently in predicting the time-dependent behaviour

Case Histories of Embankments on Soft Soils and Stabilisation

789

of reinforced embankments on soft soils using coupled finite element analysis (e.g., Rowe and Hinchberger, 1998; Li and Rowe, 2002; Rowe and Li, 2002; Gnanendran et al., 2005). Predictability of the behaviour of the Sackville reinforced embankment under working stress conditions using selected numerical models are also discussed in comparison with the observed field performance in this chapter.

2.

HALL'S CREEK TEST EMBANKMENT

The New Brunswick (NB) Department of Transportation (DOT) constructed an instrumented test embankment until failure on an alluvium deposit in Hall's Creek, Moncton, New Brunswick in 1977 to study the instability problems of road embankments in this area. The alluvium consisted of organic silt, clayey silt, silty clay and silty sand overlying sandstone bedrock and the geotechnical profile of the foundation soil obtained from site and lab investigations are presented in Figures 1 and 2 (Keenan et al., 1986). The average shear strength parameters of the foundation soil determined from consolidated undrained triaxial compression and direct simple shear tests were reported as c' = 12 kPa and ~' = 22 ~ for the first 6 m depth and that for the lower 3.5 m, c' = 13 kPa and q~' = 16 ~ (Keenan et al., 1986). The test embankment was instrumented with four settlement plates (M1-M4) and three screw-type settlement points (SP1-SP3) for measuring the settlements at the surface and at selected points within the foundation soil (Figures 3 and 4). Heave of the ground outside the embankment were monitored with a number of heave plates installed at different locations.

g.c: ~ < DESCRIPTION F-a

VERTIAL STRESS (kPa) 0

20 i

',

,

"0 -'~f~ FILL t ORGANIC / SILT "CLAY

40

,,

/ % 9 olo

'~

,\

-4 0

Po

WATER CONTENT

60

,,

(%)

We

,,

w

I

40 i I

I

l I

WE 80

i I

I

i I

i I

Cc A i m UNB 1983 0.3 l I

0.5 i I

l I

0.7 iI

| I

o

Pc

*,

-

.

_~

"-I

%%o

\

,

,,J .

-

,

\

o\

~

\ \

!

\ ~ ' : '. .~. . . . '

mt~l

.

I---4o

DENSE SAND Figure 1.

Foundation soil profile of Hall's Creek (from Keenan et al., 1986).

Chapter 27

790

UNDRAINED SHEAR STRENGTH (kPa) 10 20 30 40 50 60 70 80 90

~ O~

LEGEND o UNCONFINED COMPRESSION DOT, N.B., 1977

Z~

o o0

~ o oo

9 UNCONFINED COMPRESSION DOT, N.B., 1978

+ O0

9

~" 5 v

'-r13_ UJ

D

o$

6

9

7 9 O9 oA

zx UNCONFINED COMPRESSION SUB SURFACE SURVEYS,1977

9 ~

+ o

o/&~

10 11

-I-

-I--

zxzx

o ~ + o

,=, UNCONFINED COMPRESSION UNB, 1983

--I-- FIELD VALUE SUB SURFACE SURVEYS, 1977

o

Figure 2. Undrainedshear strength profile of Hall's Creek soil (from Keenan et al., 1986).

~mP,- N

LEGEND SP1 [

STAG E 1 STAG E 2 HEAVE PLATES LINE "C"

(Day

I

I

I

M

] ~ /-

0 1977 09 0 19 77 08 03 19 77 08 08 19 77 08 23 ~" t 0.4 19 77

09

19 78

05 3 1 . ~ f 0 . 2

18 ~..1. o.a

+-o.~

SCALE 0

P3

SP2 SP3

~

ii

.. 9

! P2 Ma, '

P4 "-~

I

HEAVE PLATES LIN E4'A"5

6

11 ~ ..

~ 0.2

~ ' . . ~ i i T ........ 4m

I

10m

Figure 3. Embankment configuration, instrumentation details and measured settlements and heaves (from Keenan et al., 1986). Pore water pressures developed in the foundation soil were monitored with four vibrating wire piezometers (P1-P4) installed at various locations and the horizontal movement in the foundation soil was monitored with an inclinometer installed at the northern toe of embankment.

Case Histories of Embankments on Soft Soils and Stabilisation 10.0

r

_o !--I"<

8.0 0f;

0~

7.0 ~.

SCALE

;m

o "~ 6.0

cc -t ../ -i- m 4.0 ,.,., COT 3.0 ";:""~"~ uJ O x uJ

~"~

791

~ N

........... ...~....,:...~-"'"~':":~....

.......1.1.:..~..~. ~

............i - ~ : - "

LEGEND 1977 08 03

~"~"~. "<": ........

"'~-,-, " ~ .........

2.0

1977 08 08 .....

1977 08 23

. . . .

1977 09 08

..................... 1979 10 10

1.0 0

S :~1

P3

SP2 SP3

~~,r..

?~ -

"1

2.__M3

P4

I

LEGEND STAGE 1 - -

- STAGE 2

Figure 4. Embankmentdetails and measured excess pore pressures (from Keenan et al., 1986). This embankment was constructed in two stages. In the first stage, the fill was raised from elevation 7.47 to 11.3 m at which time the slope failed, as detailed in Figure 5, at its northern slope where no berm was provided (Keenan et al., 1986). The instrumentation was not affected by this failure and was monitored for 36 days including the construction period at which time the second-stage construction commenced. In the second stage, berms were provided on both sides and the embankment was raised to elevation 11.9 m (Figures 3 and 4). All the instruments were monitored periodically afterwards for over 2 years. The vertical displacement profiles of the ground surface at different times obtained from the field monitoring program are shown in Figure 3 and it was observed that the settlements continued to increase even after 2 years. Similarly, the excess pore pressures at elevation 3.35 m obtained from the field measurements shown in Figure 4 also indicated increases in excess pore pressure even after more than 2 years. Keenan et al. (1986) analysed the performance of this embankment and suggested that the method proposed by Foot and Ladd (1981) may lead to conservative estimates of immediate settlements. It was demonstrated that the average settlement- time response of the embankment could be predicted accurately by considering the immediate, primary and secondary compression (creep) of the foundation soil using Bjerrum's (1967) approach for evaluating the combined consolidation plus delayed compression (Figure 6).

Chapter 27

792

Scale I

0

31m

-~ . - ~ , ,

_ --j__ _ _

I !-0 mJ =

:

~1.5 m

Figure 5. Observedfailure of embankment (from Keenan et al., 1986).

0

100 I

v

200 I

TIME (DAYS) 300

400

I

I

E

Fz LU

0.4

uJ ._1

"'-.

f

PREDICTED OBSERVED (FITTED)

w 0.8 7 ..........

----:.-_i_

.. . . . . . .

1.2 Figure 6.

Variation of settlement with time (from Keenan et al., 1986).

Limit equilibrium analysis with undrained shear strengths determined from laboratory tests on undisturbed samples corrected in accordance to Bjerrum (1972) predicted the failure height quite accurately (Keenan et al., 1986). However, there was significant difference between the failure surface predicted by the analysis and the field observations (Figure 7). It was further noted that the analysis with the undrained shear strength from field vane tests indicated a factor of safety of 2.5 for the observed failure surface and would require a 60% reduction factor to account for the failure of this test embankment. Although the observed settlement response and failure of the embankment could be predicted satisfactorily, the existence of unusually high pore water pressures even after a 3-year period could not be adequately explained. Cormier (1986) carried out further laboratory investigation into the behaviour of this soil and concluded that progressive failure of the foundation soil could have been a contributory factor for this phenomenon and recommended further research. The findings from the test embankment were used in designing roadway embankment in the Hall's Creek area.

Case Histories of Embankments on Soft Soils and Stabilisation

793

LEGEND SCALE I 3m I

N

R 137m ~ ..... ~ \',

~

OBSERVED (INFERRED)

R-10 7 m ~ ~ ......... ~ ~ ~ . .'"

UNDRAINED SHEAR STRENGTH (1983) UNDRAINED SHEAR STRENGTH (AVERAGE) L INVESTIGATIONS

~

~9\

, , R=15.7 m . ~

~ . . _ _ . . - . ~ . ~ - - ~ ~ Figure 7. Observedand predicted failure surfaces (from Keenan et al., 1986). 3.

SACKVILLE TEST EMBANKMENT

3.1. Background To investigate the progressive nature of failure of embankments on the soft soils in the Moncton area of eastern Canada in more detail and to study the beneficial effects of using a layer of basal reinforcement for the embankment, a well-instrumented full scale test embankment was designed and constructed in the nearby town of Sackville, New Brunswick in September-October 1989. Brief description of the embankment configuration, instrumentation, observed field behaviour and predictability of its performance from analyses are discussed in the following sections. 3.2. E m b a n k m e n t configuration, foundation soil properties and instrumentation The test site for this embankment was situated in an area of intertidal salt marsh deposit (Hampton and Paradis, 1981), locally known as "Marshland." A summary of the foundation soil profile obtained from the field and lab investigations are shown in Figure 8. The foundation soil was predominantly clayey silt with some organics, fibre and occasional sand lenses at certain depth ranges. It could be observed from Figure 8 that the natural water content was mostly above the liquid limit. The test embankment consisted of a 25-m-long unreinforced section and a 25 m long geotextile reinforced section connected by a reinforced transition but only the performance of the geotextile reinforced embankment section is discussed in this chapter (see Gnanendran, 1993; Rowe et al., 1995, 2001 for further details). The cross section of the geotextile-reinforced embankment with details of the construction sequence and the layout of instrumentation used for monitoring the performance are shown in Figure 9. The instrumentation consisted of piezometers, settlement plates, augers, heave plates, inclinometer casings and a total pressure cell and strain gauges on the geotextile. A total

Chapter 27

794

Water/organic content (%) 0 ~

_ rop,,so,

0 ~

0

0

0 "~"

~

0 ~D

0 O0

0 0 ~--

Clay Crust HeavilyWeathered

2-

Brown Clayey Silt + Organics _

B,ownto~a~'c,ayey

Reddish Clavey silt + Organics + Fiere

E

0 1

1

10

-

7

"

8

8~I

!

0 I ' I "!

A

:.

6

10

B

4-

field CT2 Near Acess Road

0 I'

~, -

---

0

I '!

2-

5

9

v

Stiff Clayey Silt

0

c~

4

6 -~~;~+s~

610

Vane shear (kPa)

O

2

Reddish Clayey Silt

-

O O t'M

o

3

4 " Silt + Shells + Sand lenses

8

qc (kPa)

0 C~l ~--

I~

§

AO0

oJ, 9 a,,o

0

a 12-

12

14-

14

16-

16

1820

9 Vl * V2 9 V3 A V7 Reinforced section

o A 9 +

Natural water content Plastic limit Liquid limit Organic content

Figure 8. Foundationsoil profile at Sackville site (from Rowe et al., 1995). of 32 pneumatic piezometers were installed at various depths and locations in the mid region of the 25-m-long section of the embankment to monitor the pore pressures in the foundation soil. A pneumatic-type total pressure cell was installed close to the centreline of the embankment to measure the total pressure imposed on the foundation soil by the fill. The applied total stresses deduced from the monitoring agreed well with those deduced based on the thickness of fill and the measured unit weight of the fill (Gnanendran, 1993). A total of six settlement plates and eight heave plates were installed to monitor the movement of the ground. In addition, 7 screw-type settlement augurs were installed at various depths to monitor the vertical movements within the foundation soil. Horizontal movements in the foundation soil were monitored with six inclinometer casings installed at various locations up to a depth ranging from 8 to 11 m where a relatively stiff clayey silt/silty clay stratum was encountered.

3.3.

Geotextile reinforcement

A layer of Nicolon (style 68300) polyester multi-filament woven geotextile, with the average properties summarized in Table 1, was used as the reinforcement. The strains developed

795

Case Histories of Embankments on Soft Soils and Stabilisation

,I i I~

9.5m

14110 ol 16=00 ( 9.5m1 t 4 1 1 0 m 1 4 "90 0 ( 9 0"m ) ~ - - ~ !1/10 m 17:00(82m) If/lOot 14:30 ( 3 m ) 3ul~l ( ZOrn) l i l l at I It/R) at 12:~O (6.7m)

! i, ~

~

-

"

24I

!1#10 li, !i=:30 ( r t . ? . m l l ~ ' i . , ,

.._l_ 9.6m I _1_ ........ i - 4 . e i , l ..... [< I

....

.......,........

z:'~tl .,.'~U: " .

.

.

!

:

-'-

............

_l~rn_l=

[I [I

S/lO ( Z 4 m l

/

! I " '

,///N " ""

-

[ 12~l

18.3o o ~ , ? ~ I:lO

o,,II I

,

//#--'1"- 4/i0(2.3m}

~./?,,(,k~.~.mL,

: /,.

10.Sm

sl#'lO (3.5 rnl /#-"'1- tl/lO(3.0ml

////-t-

2 / x ) (Lrm I

/////--[-(O,3m)' / i i l l t - ! - ROAD i l / I / l r P -

3 0 / 9 (1.2 m) 2 9 # 9 (0.9m | 26/9 (0.3ml

"':-~I/-" ~:,*~,=t~r,~mi

. .......

19,~ o ~ l T ~

10.8m I0,

(5.Tm,

10/10 at W:45 (5 ' I / " (5.Oral 0 1 0 a tat l G : 12:00 O 0 ( 5 (4.1 10/t0 (4 m |

, ,-----;',"UJ///--

o:5 .22

_l~4~L

/~_~-'-

.......... ::::~;q S; -- "-'-//-'-'/J-#- AUG.89

i / ~ ! - - n~',ai'XV,..~_'-... ; 22Ii"~ I " ~ " !

o~

8.3m

lII . . i p='il ; -:i::::]::'~ ,......

,~-_

__,_____

~ r - E a s t - West Reference Line

.

" i ET E 1' 3 S ~

,, !!!!/!!k

:/ # # # # 9 ......"~ -#

l

E_

1

";:

i !

! ! 9H 20H 21H

914A ,~...e

05 oe

, 03

o ,ao,

o,

o,

.2

o~ o2t

9 Settlement Plates x Heave Plates 9 Settlement Augers v Total Pressure Cell o Petur Type Pneumatic Piezometer Figure

12A !

I

0

5

~m

9. Details of embankment cross-section, construction sequence and instrumentation layout (from Rowe et al., 1995). Table 1. Properties of geotextile (after Rowe and Gnanendran, 1994) Mass Tensile strength Failure strain Initial modulus Elastic modulus Secant modulus (0- 5% strain) Secant modulus (0- 10% strain)

631 g/m2 216 kN/m 13% 257 kN/m 1920 kN/m 1466 kN/m 1678 kN/m

in the geotextile reinforcement were monitored both in the transverse and longitudinal directions with a total of 38 electrical resistance, 7 electromechanical and 7 mechanical strain gauges (Rowe and Gnanendran, 1994). The strain in the longitudinal direction was monitored with four electrical gauges installed at different locations in the mid-region of the embankment section. These longitudinal gauges indicated almost zero strains in the geotextile indicating that near-plane strain condition existed in the mid-region of the embankment section where most of the monitoring devices were installed. Details concerning the design, configuration and installation of the electromechanical and mechanical gauges were described by Rowe and Gnanendran (1994). 3.4. E m b a n k m e n t construction A locally available fill material, gravelly silty sand with some clay (average unit weight of

19.6 kN/m 3, peak shear strength of c' = 17.5 kPa, r

38 ~ and residual strength c' = 17.5 kPa,

796

Chapter 27

r = 38 ~ determined from direct shear tests on saturated bulk samples), was used for most of the construction. However, to allow adequate interaction between the geotextile and the surrounding soil, a 0.3-0.5 m thick layer of granular fill material ( c ' = 0, ~0' = 42.3 ~ unit weight = 18 kN/m 3) was used both below and above the geotextile. During construction, the fill was spread and compacted using a medium weight bulldozer. The Nicolon style 68300 multifilament polyester woven geotextile that was used as the reinforcement was factory sewn into a 23 m x 30 m rectangular section and it was instrumented with a number of strain gauges. A working platform of 0.4 m average thickness was constructed with good quality granular fill material to provide a level surface for the geotextile. Since the strain gauges were quite delicate, considerable care was taken during transport and placement of geotextile in the field. A 0.4-m-thick layer of granular fill was carefully placed over the geotextile without allowing passage of either the trucks or the bulldozer directly on the geotextile. During the construction of this embankment, the soil deformation became significant at a fill thickness of between 5 and 5.7 m. At a fill thickness of 8.2 m a large heave zone and cracking along the embankment crest was observed but there was no dramatic collapse of the embankment and additional fill could be placed and the embankment height was raised to 9.5 m on 14 October, 1989. The deformations continued and a large depression of about 0.6 m maximum depth and cracks of 4 - 10 cm width were observed on the crest of the embankment close to settlement plate 8S on 16 October, 1989. The embankment had obviously failed and the failure was of visco-plastic type. Rowe et al. (1995) analysed this field behaviour by examining the excess pore pressures in the foundation soil, strain in the geotextile, and settlement and heave of the ground and concluded that the failure thickness of the embankment was 8.2 m.

3.5. Analysis and performance prediction Geosynthetic reinforced embankments constructed on soft soil foundation are often analysed using limit equilibrium and/or finite element methods. Both these methods have advantages and disadvantages and were used to back analyse the performance of Sackville embankment.

3.5.1.

Limit equilibrium analysis. The Sackville reinforced embankment has been backanalysed using the slip circle type limit equilibrium method adopting different approaches assuming the reinforcement force to act horizontally (Rowe et al., 1994; Palmeira et al., 1998; Gnanendran et al., 2000). The limit equilibrium methods employed by these investigators predicted the failure height of the reinforced embankment quite accurately but the predicted failure surfaces differed from that observed in the field (see Figure 10). It is interesting that similar observations were also made at the Hall's Creek site. Gnanendran et al. (2000) suggested that the apparent good prediction of failure height from these analyses could be due to compensating errors rather than correctly predicting the actual behaviour.

797

Case Histories of Embankments on Soft Soils and Stabilisation

--

/

56.2

,.

/

.9.5 m

.1_ u.~lm

,~'~~--'~.

T ~

I

3.4 m I _1

8.3 m

/..//" " " [ " - t ~ 9 ;/" 2 3 [ / 1 El ]J II

/----l~8--[Ls~-Lt~-~-.-~-i

9 .

.

.

.

i

i

5

10 m

.

.

~ ~

.

.

.

I

-I--I-

I

I I

Lt "k-(primary) / / -l'~~61L st~rface crack

FII I

.... ~:--~",~~L~ . .~.~-.-.-~.,-' ~~f "

10.8 m _13rnl_ 10.5 m

i25 ,~-'Surfacee~ck-I-

I

(s~condary) ~

,

,

~t

E,

18s

I

,,,e.l'~:'x~

- ~ I,

I

I

I

Geotextile .15A "' ~ "-reinf~ 16A ~ Secondary 912A failure surface

Primary failure surface Alternate failure surface (Primary)

1.1 m

I "-I

I

I \

LEGEND --Failure surface calculated by palmeria et al. 1998 using USCE approach Failure surface calculated by palmeria et al. 1998 using MBF approach Failure surface calculated by Gnanendran et al. 2000

......

Figure 10.

Observed failure mechanism (Rowe et al. 1995)

Observed and predicted failure surfaces from limit equilibrium analyses (modified from Rowe et al. 1995; Gnanendran et al., 2000).

Although limit equilibrium methods are easy to use and require less time and effort in a design situation, they do not provide any information about deformations of the soil or of the geosynthetic or any information regarding the performance prior to collapse. In addition, limit equilibrium analysis must assume that collapse will be governed by limit equilibrium of both the soil and reinforcement unless other information (such as the strain in the geosynthetic at failure) is provided from other methods of analysis. The tensile force developed in the reinforcement depends on the deformations that occur both within the soil medium and within the geosynthetic. Consequently, a limit equilibrium technique which disregards the deformation characteristics of the soil-reinforcement system cannot be rigorously employed to evaluate the behaviour of the reinforced embankment system. Furthermore, the deformational characteristics of the two elements (i.e. the reinforcement and the soil) are such that concomitant failure will not occur in the two elements. This mathematical inconsistency can be avoided only by recourse to numerical methods of stress analysis such as the finite element analysis (FEA) which takes into account both the constitutive responses of the geosynthetic and the soil mass (i.e. both the embankment and the foundation soil). Sackville embankment was analysed by Rowe et al. (1996) using a fully coupled large-strain elastoplastic finite element formulation adopting a MCC model for the organic clayey silt foundation soil. This study revealed that, although many features of the embankment behaviour could be captured reasonably well

3.5.2.

F E A o f Sackville e m b a n k m e n t .

798

Chapter 27

by this analysis, the MCC elastoplastic formulation is inadequate for predicting the multiple characteristics of the embankment behaviour such as vertical and horizontal deformations, pore pressures and geotextile strains accurately when the embankment approached failure and when its thickness remained constant. It was concluded that the inadequacy might be due to the suspected rate-sensitive nature of the foundation soil. Rowe and Hinchberger (1998) reanalyzed this embankment adopting a fully coupled elasto-viscoplastic formulation with an elliptical cap model for the foundation soft soil. In particular, the foundation soil was modeled using Perzyna's (1963) theory of overstress viscoplasticity and an elliptical yield function of the form proposed by Chen (1982). This analysis was found to capture the multiple characteristics of this embankment behaviour including the vertical deformation near the centreline, horizontal deformation near the toe, geosynthetic strain and the increase in excess pore pressures in the foundation soil even during periods when there was no addition of fill. It was concluded that particular care is required when constructing embankments over rate-sensitive soils as these soils may appear to be performing adequately during construction due to their ability to carry overstress, but may subsequently creep to failure. Embankments are often not constructed to failure heights but are usually designed and constructed with some margin of safety. Therefore, the predictability of the behaviour of an embankment prior to failure under working stress conditions is of prime importance to practicing engineers and would be the focus of discussion in the following sections.

Predictability of SackviUe embankment behaviour under working stress conditions The predictability of the behaviour of Sackville embankment was investigated by Gnanendran et al. (2005) using a creep-based elasto-viscoplastic and MCC elastoplastic material model for the foundation soil adopting the same finite element descretisation with due consideration for geometric nonlinearity and using the same experimental database for the material properties. The elasto-viscoplastic model that take into account creep proposed by Kutter and Sathialingam (1992), which uses the coefficient of secondary consolidation as the additional parameter, was adopted to perform the creep analysis. Hereafter, this is referred to as the "creep model". The results from these analyses and that predicted by Rowe and Hinchberger (1998) using their elasto-viscoplastic elliptical cap model, hereafter referred to as "rate model" are discussed in comparison with the observed field behaviour. It is noted that the discussion is restricted for the behaviour under working stress conditions. The foundation soil properties adopted for the FEA with MCC and creep models are presented in Tables 2 and 3 and those for the rate model are shown in Tables 4 and 5. Properties of the embankment fill adopted in all three analyses are summarised in Table 6.

3.6.

3. 7.

Comparison of calculated and observed performance

The comparison of the settlement at settlement plates 7S and 8S calculated from the three analyses with field measurements are shown in Figure 11. (Note: Time = 0 at 0:00 hour on

799

Case H i s t o r i e s o f E m b a n k m e n t s on Soft Soils a n d Stabilisation Table 2.

Foundation soil parameters assumed for FEA with MCC and creep models (Gnanendran et al., 2005) A,,

A,*

ecs,

eN*

k0'

V

OCR

a

113 113 113 113 113 113

0.055 0.021 0.027 0.045 0.027 0.027

0.242 0.111 0.154 0.224 0.154 0.154

2.210, 1.300, 1.589, 1.799, 1.590, 1.590,

2.339 1.362 1.678 1.924 1.678 1.678

0.68 0.68 0.71 0.77 0.79 0.83

0.3 0.3 0.3 0.3 0.3 0.3

1.0a 3.6 1.2 1.0 1.2 1.2

0.00973 0.00446 0.00619 0.00900 0.00619 0.00619

1 113

0.027

0.154

1.590,

1.678

0.88

0.3

1.2

0.00619

Depth (m)

~'M (kN/m 3)

tr

0.0-1.1 1.1-1.8 1.8-2.7 2.7-4.4 4.4-5.8 5.8-10.0

15.2 17.8 17.8 17.0 17.0 17.0

1 1 1 1 1 1

10-14.0

17.0

aApproximation for the vertical cuts made in the crust as per Rowe et al. (1996). Table 3.

Depth (m)

Permeability variations assumed for the FEA with creep and MCC models (Gnanendran et al., 2005) ke/kv

Normally consolidated, K v = A ( e - C) B

Overconsolidated, K v = A l ( e B1)

A

B

C

A1

B1

0.0-1.0 1.0-3.5 3.5-5.0

10 4 10

0.5769E-3 0.5769E-3 0.5769E-3

5.1033 5.1033 5.1033

0.1006 0.1006 0.1006

0.00864 0.00864 0.00864

0.0 0.0 0.0

5.0-14.0

4

0.7413E-3

4.8574

0.0000

0.00864

0.0

Kv = vertical permeability in m/day.

Table 4.

Foundation soil parameters assumed for FEA with rate model (Rowe and Hinchberger, 1998)

Depth (m)

M*/M

c'k/c'(kPa)/ (kPa)

~'(kN/m3)

ko'

eo

tr

~

?,vp (• 109 s -1)

n

0.0-1.1 1.1-1.8 1.8-2.7 2.7-4.4 4.4-5.8 5.8-10.0

0.75/0.96 0.75/0.96 0.75/0.96 0.75/0.96 0.75/0.96 0.75/0.96

8.0/6.5 8.0/6.5 8.0/6.5 8.0/6.5 8.0/6.5 8.0/6.5

17.8 17.8 17.5 16.5 17.2 17.2

0.68 0.70 0.70 0.75 0.80 0.80

2.2 1.2 1.6 1.6 1.5 1.2

0.055 0.03 0.03 0.05 0.03 0.03

0.28 0.14 0.22 0.15 0.15 0.15

5.6 5.0 5.0 5.6 6.1 5.0

20 20 20 20 20 20

10-14.0

0.75/0.96

8.0/6.5

17.2

0.80

1.2

0.03

0.15

5.0

20

21 September, 1989). The finite element calculations with MCC and creep models overestimated the settlement at settlement plates 7S and 8S until the fill thickness was increased to 5.7 m. However, the analysis with the rate model predicted the settlement at 7S and 8S accurately until the embankment was constructed to 5.7 m thickness. This improvement in predictability of the settlement at low embankment thicknesses could be attributed to the different failure envelope adopted in the overconsolidated stress state in the rate model reported by Rowe and Hinchberger (1998) compared to the elliptical surface adopted for

800 Table 5.

Chapter 27

Permeability variations assumed for FEA with rate model (Rowe and Hinchberger, 1998) kv = kr e x p [ ( e - er) ] Ck]

Depth (m)

kh/kv

er

Ck

0.0-1.1 1.1-1.8 1.8-2.7 2.7-4.4 4.4-5.8 5.8-10.0

10 4 4 4 4 4

2.4 2.4 1.2 2.4 2.4 1.2

0.22 0.22 0.16 0.22 0.22 0.16

1.7 1.7 8.3 1.7 1.7 8.3

10.0-14.0

4

1.2

0.16

8.3 x 10 -8

Table 6.

k r (m]s)

X x X X X X

10 -8 10 -8 10 -8 10 -8 10 -8 10 -8

Properties of embankment fill adopted in the analyses First 0.7 m

Properties of fill material

Janbu's equation, (E/Pa) = K (t73[Pa) m

c (kPa) r ~'(kN/m 3) v K, m

0.0 43 ~ 8~ 18.0 0.35 250, 0.5

Remainder of fill 17.5 38 ~ 7~ 19.6 0.35 250, 0.5

,~

both overconsolidated and normally consolidated stress states in the other two models considered in this chapter. The analysis with rate model significantly underestimated the settlement beyond 5.7 m thickness until the failure thickness of 8.2 m was reached. The predictions from the analyses with MCC and creep model were reasonably good when the fill thickness was increased from 5.7 to 8.2 m. As the embankment reached its failure fill thickness of 8.2 m, the finite element predictions from all three analyses underestimated the settlements at 7S and 8S. Examination of the settlement responses from the three analyses suggest that for an embankment constructed with a margin of safety, i.e. under working stress condition for a 6-7 m thick embankment with an approximate factor of safety of 1.2-1.4, the creep model is capable of giving a slightly better prediction than the rate model and the settlement prediction with the MCC model would also be reasonably good. Examining the settlement responses from FEA with different models for the foundation soil, Gnanendran et al. (2005) concluded that the rapid increase in settlement at constant thickness could not be attributed fully to the suspected rate-sensitive behaviour of the foundation soil, and at least in part was due to creep compression behaviour. The settlement predicted from FEA with each model at settlement augers 9A and 11A are compared with the measured settlements in Figure 12. The settlements at auger 9A predicted from the FEA with creep and MCC models agreed well with the field measurements

Case Histories

1.6

of Embankments

9 ..... o ..... 7S - F i e l d data ( R o w e et al. 1995) . . . . ... -. 8S - F i e l d data ( R o w e et al. 1995) Failure thickness ~... .." 8 .... o .... 7S - M C C m o d e l ( G n a n e n d r a n et al. 2005) - 9 ., .... ~ .... 8S - M C C m o d e l ( G n a n e n d r a n et al. 2005) 7S - C r e e p m o d e l ( G n a n e n d r a n et al. 2005) -..-'-G, 7 8S - C r e e p m o d e l ( G n a n e n d r a n et al. 2005) .."_~& ........... E m b a n k m e n t t h i c k n e s s w i t h time ( R o w e et al. 1995) - ~:/d~) ..... 7S - R a t e m o d e l ( R o w e a n d H i n c h b e r g e r 1998) .- I f ' / ~ , " 6 ' : .................... l ,'t - - . . . . 8 S - R a t e m o d e l ( R o w e and H i n c h b e r g e r , 1998) :: _~tt~- .~

1.4

1.2

1.0

/

o.8 0.6 -

0.2

oo ~ . ~ "

..."

.

".i

3

./

/

2

1

,

400

4

~ . . ~ l ' 9 1 4 9 .... :::p~l

............................................................... _-4 .f" ...... .~.-.Z.~-.- .... o" ~ / -....~ ............... .._,..~,,,,,..,_..,,. . Z . ~ - . Z . = ~ ' ~ : ~ . . - . . ~ . . . - . . . ~ .~" . / . . . . .-...:~'...'=....'=...--'............... t~........ u ................. .~ .... .~./" ....... ~.::t~/"

0.0

..~,r i.

or '

99 0.4

801

on Soft Soils and Stabilisation

0

450

50O

T i m e (hours) F i g u r e 11.

O b s e r v e d and p r e d i c t e d s e t t l e m e n t s at 7S and 8S f r o m F E A ( f r o m R o w e et al., 1995; R o w e a n d H i n c h b e r g e r , 1998; G n a n e n d r a n et al., 2005).

up to about 5.7 m fill thickness and then the analyses tend to underestimate the settlements. The analysis with the rate model always underpredicted the settlement at 9A more than the other two models. Settlement at 11A was reasonably well predicted by the analyses using the creep model compared to the other two models. Again, the analysis with rate model underestimated the settlement at 11A more than the other two models. Therefore, considering the overall settlement predictions in comparison with the field data at settlement plate locations and settlement auger points, it is suggested that the creep model is capable of predicting the settlement near the centreline of the embankment more accurately than the other two models and the MCC model prediction is also reasonably good. Are the measured and calculated vertical displacements at heave plate 2H are presented in Figure. 13. For clarity, the results for only 1 heave plate is given in this figure as the responses were very close and difficult to differentiate if the results of two or more heave plates were shown. The field behaviour indicates an apparent delay in the heave response for the construction loading9 It is seen that the FEA predictions for the heave at this location from all three models are in reasonably good agreement with the field observations until the embankment was constructed to 8.2 m thickness. However, the analyses with MCC and creep models overestimated the heave during the brief construction stoppage at 5.7 m fill thickness whereas the analysis with the rate model predicted the heave more accurately during this period9

Chapter 27

802

,

1.2

,

I

.... 9.... .... 9.... .... o.... . . . . []. . . .

1.o-

............ ......

0 . 8 -

,

,

,

,

I

.

.

.

.

I

,

,

,

,

I

,

,

,

,

I

.

.

.

.

I

9A - Field data (Rowe et al 1995) 11A - Field data (Rowe et al. 1995) Failure thickness 9A - MCC model (Gnanendran et al. 2005) 11A - MCC model (Gnanendran et al. 2005) 9A - Creep model (Gnanendran et al. 2005) 11A - Creep model (Gnanendran et al. 2005) Embankment thickness - time (Rowe et al. 1995) 9A - Rate model (Rowe and Hinchberger 1998) . ....................." 11A - Rate model (Rowe and Hinchberger 1998)

,

,

9

w... ....... .."

.

9

9 . .

-r 6

#

r~ 9

5 0.6-

/ ..."..................... "~

0.4-

..........................................................

4

i

3

":"

0.2-

2 . . . . .

,-

i~

.-...: : .-... z. :#

.........

: : " :: ' " *_

i"

.....

,.2:.; . 2 :-:. !

0.0 "'~':,=::;-~'--~ .... -='i/:~"L~'~-"v-"5'~"-'m~-~"-"r"~,-~~"--"~-'~''---;'''F'" , ' 400

420

440

460

480

0

500

Time (hours) F i g u r e 12.

O b s e r v e d a n d p r e d i c t e d s e t t l e m e n t s at 9 A a n d 1 1 A f r o m F E A ( f r o m R o w e et al., 1995; R o w e a n d H i n c h b e r g e r , 1998; G n a n e n d r a n et al., 2 0 0 5 ) .

The FEA with the rate model indicated slightly less heave than the measured values during the construction of the embankment from 5.7 to 8.2 m thickness but the analyses with MCC and creep models gave accurate predictions for the heave during this construction phase. The rate model analysis indicated large increase in heave than that observed in the field after the embankment was constructed to the thickness of 8.2 m. Therefore, considering the overall heave response near the toe for an embankment of up to about 7 m thickness, it is concluded that all three models are capable of predicting the heave quite well but the prediction with rate model is once again slightly better. Figure 14 shows the comparison between the field data and the calculated horizontal displacement variations with depth obtained from FEA with MCC and creep models at the toe of embankment (221). As the horizontal displacement with depth variation was not reported for the analysis with the rate model, a similar comparison for inclinometer 231 located near the embankment toe is presented in Figure 15 where the FEA results obtained with all three models were available. It is observed that the analysis with MCC model predicted the maximum horizontal displacement at the toe (i.e. at 221) quite well (i.e. for both 3.4 and 5.4 m thicknesses). The analysis with the creep model overestimated the horizontal displacement at 3.4 m thickness and although satisfactory, it slightly underestimated the horizontal displacements at 5.4 m thickness. At depths greater than 2 m, however, the predicted horizontal displacements were generally higher than the measured values.

Case Histories of Embankments on Soft Soils and Stabilisation

1.6

,

I

. . . .

I

. . . .

I

. . . .

I

,

,

,

,

,

,

,

,

,

I

'

'J-

...... 4 ....... 2 H - Field data ( R o w e et al. 1995) Failure thickness-------l~. ..... ...... .o. ...... 2 H MCC model (Gnanendran et al. 2005)

1.4

1.2-

d=

9

803

...... 2H - Creep model (Gnanendran et al. 2005) ......... 2 H - Rate model (Rowe and Hinchberger 1998) ............... Embankment thickness with time (Rowe et al. 1995)

-

8

i ~ _~ 7 ~..."--t,.~

-

_ ................... ...'-

1.0-

9

6

,..~::-.-

r~ r~

,S=

0.8-

r~

~3

>

-

/:

'"'""....................": :

0.6-

................................

0.4-

~

9

_" 4

i"

i

...... o ~

o...~,.o.

3

-

~/' ~..,"~

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

9

:- 2

~,~.,f:.'"" 0.2m=-'~"-'7"~'

9. ~ , . ~ - - - - - = , ~ . . . " r ~ . ~ . . ~ t r ~ . . . ~ d " ~ ~ . ' ~ r - ~ 0.0

..........

400

420

440

~'~.~'#t

i-- 1

":..'Z'....... :: . . . .

~_................... ~..~ .................. ~ ...... t .... ;'"

'

,

460

.

.

.

_

.

I

480

.

.

.

.

,

'

'

0

500

Time (hours) F i g u r e 13.

Observed and predicted heave at 2H from F E A (from Rowe et al., 1995; Rowe and Hinchberger, 1998; Gnanendran et al., 2005).

At inclinometer 231, the analyses with MCC and rate models well predicted the horizontal displacements at 3.4 m thickness while that of creep model overpredicted them. At 5 m thickness, the creep and rate model analyses predicted the maximum horizontal displacements quite accurately but the creep model analysis slightly overpredicted the horizontal displacement profile with depth. The horizontal displacements at different depths predicted by the FEA with the MCC model were slightly higher than those of the creep model at 5 m thickness. In summary, the FEA with rate model well predicted the horizontal displacements at 231 but the MCC and creep models predicted the horizontal displacements at both 221 and 231 quite satisfactorily. Gnanendran et al. (2005) found that the horizontal displacements predicted by the FEA with MCC model agree reasonably well with the field observations and are better than those predicted with the creep model. However, Rowe et al. (1996) reported that the MCC model underestimated the lateral displacements significantly. Figure 16 shows the comparisons between field data and FEA predictions from MCC and creep models for the excess pore pressures at piezometers P12, P28 and P32 installed closer to the centreline of the embankment. The creep model predicted the excess pressures at P12, P28 (i.e. at 4 m depth) reasonably well until the fill thickness reached 5.7 m but overpredicted the excess pore pressures afterwards. However, the excess pore pressure at

804

Chapter 27

0.0

Horizontal displacement (m) 0.2 0.3 0.4

0.1 |

I

,

,

n

'--

i

P,

i

n

i

i

i

i

I

n

i

i

n

0.5

L

I

I

I

I

0.6

I

I

I

I

I

/ j

"'".

.'."

_

9"

9 ,

~

9

.

."

i.

.."

\ ..... .9 9 ........... \

\

.

/

J

\

Ja :

9i 9

."

\J

--9

_

9

/

j

9 ... 9

,, 9149

..9

9149

J

."

39 9

= o

.~

9

i

4-

9

9

a

/

/

.... m "

f . i " ~

/~""

j

,

...."

9

.'" ..........

i"

...

"'/

i "

... 9

. , .

.

/

..'"

/

,

9

.m"

/

"

:"

5-

/ /

~"

.......

............

,"

"

~0

o

9

6O

<~

/.

:

;."l

.."

9

~D

.=. /

9

" I

il'

.~ !

-

l::

.

ili

O

" I

10

/ " . . 9.... . . o..

F i e l d d a t a - e m b . t h i c k n e s s = 3.4 m (449 h) ( R o w e et al. 1995)

......

F i e l d d a t a - e m b . t h i c k n e s s = 5.4 m (473 h) ( R o w e et al. 1995)

../'

9I g -

Z.

,~ .--I

7-

8-

...-"

i

/

9 ..... ..... zx.....

M C C m o d e l - e m b . t h i c k n e s s = 3.4 m ( 4 4 9 h) ( G n a n e n d r a n et al. 2 0 0 5 )

............

M C C m o d e l - e m b . t h i c k n e s s = 5.4 m (473 h) ( G n a n e n d r a n et al. 2 0 0 5 )

"I

.i

C r e e p m o d e l - e m b . t h i c k n e s s = 3.4 m ( 4 4 9 h) ( G n a n e n d r a n et al. 2 0 0 5 )

/

9

C r e e p m o d e l - e m b . t h i c k n e s s = 5.4 m (473 h) ( G n a n e n d r a n et al. 2 0 0 5 )

" ~. #' 9.r

|

i

I

i

i

i

i

I

i

i

i

i

I

i

i

i

i

I

i

i

i

i

I

i

i

i

-

i

Figure 14. Observed and predicted horizontal displacement at 221 from FEA (from Rowe et al., 1995; Gnanendran et al., 2005). piezometer P32 located at 6 m depth, the FEA predictions were always much higher than field measurement. These piezometers were located closer to the centreline and installed near settlement plate 8S where significant cracking and depression of the embankment crest were observed in the field when the embankment was raised above 5.7 m thickness (Rowe et al., 1995). The lack of agreement between predicted and measured excess pore pressures could be due to greater volumetric strain change caused by excess pore water dissipation compared to that caused by creep/rate compression (Yin and Zhu, 1999). The results of FEA analyses are compared against field measurements for piezometers P19, P31, installed at 4 m depth near settlement plate 7S, in Figure 17. Analyses with both creep and rate models captured the pore pressure increase observed in the field during the brief construction period at 5.7 m constant fill thickness quite well and creep model gave the best overall prediction for the excess pore pressure at this piezometer location. The analysis with the MCC model failed to capture the observed pore pressure increase at 5.7 m constant thickness and in general moderately overpredicted the excess pore pressures at P19, 31. The predicted excess pore water pressures are compared with the field measurements in Figure 18 for P24 installed near the embankment toe and P27 just outside the embankment, both installed at 4 m depth. The predicted excess pore pressures from all three models are

Case Histories o f E m b a n k m e n t s on Soft Soils and Stabilisation

805

Horizontal displacement (m) 0.1

0.0 0

0.2

. . . .

-" i.e~. / 9

"I

,

./

o~ =

6 -

7 8 9 10

.

..

/

9

..-'7"~ / /.o ,///.~r ./. !~...-.. / . ~,~, 9

0

.

J

/

./

. .

,\ :~

"

i\

. . . . . . .

I

."

i =

i

9

r

5-

,-.,i

I

/

69 ~

.=

,

I

)

/

"

/

34-

,

t i

6,..1.1 9

2-

'

I

.

di./ ..

0.5

0.4

I.,-1

1 -

= 0 e~O

0.3

J

. i.f

/

/

~ ...... l.. ,<<....

9

/r-i-

I

I

....I,::>"

..

o..:/.. / ~/*'.)P ./ / ..~"ii/ ;/ # .ill" / t /~/":" // .i ~' d / d1 11 /] .'/ / .. J / / ..)/ - / ../ 6-~-~ ' ' u

/.m .....

I

/ ....

..."

.6 .

..... o

9

~ ..... 9 ..... Field data at 3.4 m (449 h) (Rowe et al. 1995) ..... 9 ..... Field data at 5 m (472 h) (Rowe et al. 1995) ..... o ..... MCC model at 3.4 m (449 h) (Gnanendran et al. 2005) ..... []..... MCC model at 5 m (472 h) (Gnanendran et al. 2005) .... Creep model at 3.4 m (449 h)(Gnanendran et al. 2005) Creep model at 5 m (472 h) (Gnanendran et al. 2005) ..... Rate model at 3.4 m (449 h) (Rowe and Hinchberger 1998) . . . . . . Rate model at 5 m (472 h) (Rowe and Hinchberger 1998) '

'

'

'

~

'

'

'

'

,

'

'

'

'

~

'

'

'

'

Figure 15. Observed and predicted horizontal displacement at 231 from FEA (from Rowe et al., 1995; Rowe and Hinchberger, 1998; Gnanendran et al., 2005).

generally in good agreement with the field measurements although the MCC and rate models moderately overpredict the excess pore pressures at lower fill thicknesses at piezometer location P24, whereas creep and rate models slightly underestimate the same at higher fill thicknesses. However, the FEA predictions are generally in good agreement with the field data at P27. Comparison of the calculated and measured geotextile strains at fill thicknesses of 3.4 m (448 h) and 5.7 m (475 h) are shown in the Figures 19 and 20. At 3.4 m fill thickness, the FEA with MCC and creep models predicted the maximum strains reasonably well and the creep model gave the best overall prediction for the strain distribution across the geotextile reinforcement. The rate model overpredicted the maximum geotextile strain at 3.4 m thickness. However, at 5.7 m thickness (475 h), the analysis with the rate model gave the best prediction for the strain while the MCC and creep models significantly underestimated the maximum strain. 3.8. Summary and comments on the FEA predictions The comparative predictions of various behaviour characteristics of the Sackville foundation soil under a typical working stress or loading condition of up to about 7 m thick embankment obtained from the FEA are summarised in Table 7.

Chapter 27

806

200

|

,

I

,

....... 9 ...... m ...... 180 . . . . ....... .zx....... ................ 160 .........

140

,

,

,

I

,

i

,

,

I

,

,

,

,

I

,

,

,

,

I

,

,

,

,

P12,28 - Field data (Rowe et al. 1995) Failure thickness P32 - Field data (Rowe et al. 1995) P12,28 - MCC model (Gnanendran et al. 2005) P32 - MCC model (Gnanendran et al. 2005) Embankment thickness with time P12,28 - Creep model (Gnanendran et al. 2005) P32 - Creep model (Gnanendran et al. 2005)

~J

.

.

.

.

I

,

9

,

~/-/ ~ ...... .A.'..- 9 ,~.

/.-..

9

7

--'

E

2--6

.

~D

:=, 1 2 0 A

~.

_

100

/

2,

i ~.

80

~

60

,_ . . . . . . . . . .

-

///r

. . o ..........

'

, ,'

.O ...........

.. ~, ....................... ax.... . . . ~ . . - - ~ . . . ~ . ' 7 - -

40

'0 ....................

~

9 .

....

.

" " " f ~ -

20

.

.

.

........... .I"

~

...."'

~

. . ""

..........

2

_. " . .... 9

."

4 . ....

..

. . ~

n . , 'A ...... ,,......

_./

5.~

........

~' t

/ -

.

1~-"

/ r,,> / i~"/

/: ~

.-...i

..... I

" ......... i ' " ..... 9 ..........

- I . . . . . . . . . . . . . . . . . . . . . . . . 9 ....... 9 .............. I ' "

0

i

i

I

i

i

i

400

i

I

i

I

i

i

420

. . . . I. . . . . I . . . . . I

I

440

460

Time

480

0

500

(hours)

F i g u r e 16. O b s e r v e d a n d p r e d i c t e d e x c e s s p o r e p r e s s u r e s at P 1 2 , 2 8 a n d P 3 2 f r o m F E A ( f r o m R o w e et al., 1995; G n a n e n d r a n et al., 2 0 0 5 ) .

200

,

,

,

I

,

.

Embankment thickness with time

.

-

120

/.

~.

100

~

-:

~

,

"

o

/.

I.~--.--

6

.

-/~/---5 / 9 / /. i/-------J~_,_/ ........... /f ..--_ g i//~$~ ....... [] . . . .

:

80

9

/ /

_

0 r~

.

-

r~

/ /

_

--

9

> .... ./ /

-

-

~

I

P19, 31 - FEA with Creep model (Gnanendran et al. 2005) P19, 31 - FEA with MCC model (Gnanendran et al. 2005) P19, 31 - FEA with Rate model (Rowe and Hinchberger 1998)

160 140

.

Failure thickness

180 . . . . . . n ..... Field data - P19,31 (Rowe et al. 1995) -

.

i,,o

/

......

r~

J.~

-3-~

........ / .od.)

60

_/

-

/~./.. .......

. . . . . . . . . . . . . . . . . . . . . .

40

-

:,-i//

/.."

~ _ - -

..

-2

d

o

~

............

: ~ ' ' t ~ """S"9 9 9 a . . . . . . . . . . . . . . . . . .

n

-1

20 0

i

400

i

i

i

i

i

i,,

i

450

i

I

500

Time (hours)

Figure

17. O b s e r v e d a n d p r e d i c t e d e x c e s s p o r e p r e s s u r e s at P 1 9 , 3 1

from FEA

R o w e a n d H i n c h b e r g e r , 1998" G n a n e n d r a n et al., 2 0 0 5 ) .

( f r o m R o w e et al., 1995;

Case Histories of Embankments on Soft Soils and Stabilisation I

100 ....

-- --............

60

I

I

I

I

I

I

/

I

/

I

I

I

-

9

I

" ~' 1'~ - ~ .0,I"

40 -

..

o ....... o ................. ..........

~

I

I

I

o"

|

I

. . . . . . . . . . . . .

......... 9v.v.v. . . . . . . . . . . . . . . . . . . . . . .

-

. ......

-

O.-"

/. ..... O- ............... "

5

-----

,

~

~

_

/

4

. . . .

I

@ ....... ~ . . . : . / 6 ~ ~ []a / '

I

~. .r ..~.""

-"

. ......

/ p

.."

.......

/-

~"

"""

"~-

.,,r

~

.o

.;.

~.,~.~,,.C . . . .

I I

r~ c~

-

o ...... o ...... ~ S ,-"--" ./.....~ /

.0'

9

~/" .................

.............. . I~-I~

o/,-

~ . ~ , , , ~ . , ~ . ~ _ . ~ _ ~9_ .'

6

/

. 9 : . : ~ ........ ~*":" . / / .-../ ...-" ---o ..........................

20

I

O .......

.,

... 0

I

I

p 2 4 - F i e l d data ( R o w e et al. 1995)' p27 - Field data ( R o w e et al. 1995) Failure thickness ~.. ....... p 2 4 - M C C m o d e l (Gnanendran et al. 2 0 0 5 ) ." p27 - M C C m o d e l (Gnanendran et al. 2 0 0 5 ) p 2 4 - Creep m o d e l (Gnanendran et al. 2 0 0 5 ) p27 - Creep m o d e l (Gnanendran et al. 2 0 0 5 ) .. E m b a n k m e n t thickness - time ( R o w e et al. 1995) . ......... p 2 4 - Rate m o d e l ( R o w e and Hinchberger 1998):.... .. ~ . - o . ~ . . ...... p27 Rate m o d e l ( R o w e and Hinchberger 1 9 9 . 8 . ~ . . : i ~ i ~ i ! i ~ _

.....

..... 9 ..... ..... o ..... ..... [] .....

80

I

I

I

807

:: . . . . .

9 . . . .~ .,~'.,,~.

.L~r"

3

.--.!"

.

i

_ _ .:~__...,~---.:- . . . . ------" ~

--

.O ......

9 ---- 9 .'~"

-~ .................. i.-7.....--., i~...-7:........i 9 9.n ...........................

0

n

n

~

n

o .................

o .......

~

~

n

400

u

n

~

~

n

420

n-:-: (d - -

0 .......

12[

.o" []""

u

~

n

~

~

440

n

~

~

n

n

460

t

I

~

n

n

480

t

I

t

0

500

T i m e (hours) F i g u r e 18. O b s e r v e d and predicted e x c e s s pore pressures at P 2 4 and P27 f r o m F E A (from R o w e et al., 1995" R o w e and Hinchberger, 1998" Gnanendran et al., 2005).

....

2.0

I , , , , I , , , , I , , , , I , , , , I , , , , I , , , , I , , ,

I , , , , I , , , , I

....

I , , , ,

o []

Field data - average (Rowe and Gnanendran 1994) Field data - lower limit (Rowe and Gnanendran 1994) Field data - upper limit (Rowe and Gnanendran 1994) ................. FEA - MCC model (Gnanendran et al. 2005) FEA - Creep model (Gnanendran et al. 2005) Northern shoulder FEA - Rate model (Rowe and Hinchberger 1998) of embankment

1.5 -

.~

/

.~ .~

....

---.,.,

/ 1.0 -

~..~ ~ ~

/

..

..

i/ . ~ - . . . .

~9 ~

~ " ~

~.

...........

L

~.

/l. ..............

9 ..-:

" ~ . .

........: ~...~.

:.

0,

-

.

0

~

-

0

........................

A,,,,

i.; 0

-

~

. . . .

2

~ . . . .

4

~

6

. . . .

~

8

. . . .

J

10

. . . .

~ .

12

.

.

.

.

.

14

.

.

.

~

16

~ . . . .

18

~

20

.

.

.

.

.

22

.

24

Distance from the northern toe of embankment (m) F i g u r e 19. O b s e r v e d and predicted geotextile strain distributions at 3.4 m (448 h) f r o m F E A (from R o w e et al., 1995; R o w e and Hinchberger, 1998; Gnanendran et al., 2005).

Chapter 27

808 6

, , , , i , , _ , , t , , , , ~

. . . .

5

~"

4-

. . . .

I

. . . .

i

. . . .

I

. . . .

a . . . .

i

. . . .

i

. . . .

~

. . . .

Northern shoulder of embankment

" o

i

----o---- Field data- average (Rowe and Gnanendran 1994) o Fielddata- lower limit (Rowe and Gnanendran 1994) zx Fielddata - upper limit (Rowe and Gnanendran 1994) ............ FEA - MCC model (Gnanendran et al. 2005) FEA - Creep model (Gnanendran et al. 2005) . . . . . . FEA- Rate model (Rowe and Hinchberger 1998)

/!,, a-

/., ~-" ~ ' / f , ~ Z

~

.

0 9

."

...........

~:~, /<:"

i-1

"" ~

"" "~"""

.-...~-:L ........... . . . . ~ : ~ .....

/

"'-"

".

""

9 9 ----- 9

~ []

: :~"'"'-

\

....

9 ,--

9

12

"

/Z /; 0

2

/x

..

....

4

6

8

10

12

14

16

18

20

22

24

Distance from the northern toe of embankment(m) Figure 20. Observed and predicted geotextile strain distributions at 5.7 m (475 h) from FEA (from Rowe et al., 1995" Rowe and Hinchberger, 1998; Gnanendran et al., 2005). The FEA with the creep model predicted the vertical displacements at the settlement plate, augers and heave plate slightly better than the MCC and rate models. Although it was reported that the analysis with the rate model captured the vertical settlements accurately at higher fill thickness of 8.2 m and the large increase in settlement during the construction stoppage period at this fill thickness, the creep model was found to predict the settlements better for the embankment thickness of up to about 7 m. All three analyses predicted the horizontal displacements at or near the toe of embankment satisfactorily but the rate model gave better prediction for the horizontal displacements near the toe. Similarly, FEA with all three models predicted the excess pore pressures in the foundation soil quite satisfactorily but the creep and rate models gave better predictions than the MCC model. The analyses with the MCC and creep models predicted the geotextile strain at low embankment thickness (3.4 m) satisfactorily while the rate model analysis overpredicted it. On the other hand, the analysis with the rate model predicted the geotextile strain at 5.7 m well and the analyses with MCC and creep models underestimated it. An overall examination of the predictions from the three FEA suggests that all three models are capable of predicting the performance of the embankment reasonably well

809

Case Histories of Embankments on Soft Soils and Stabilisation

Table 7. Summaryof FEA predictions with the rate, creep and MCC models Behaviour characteristic

Emb. Thickness/ Location

FEA prediction with MCC model

FEA prediction with creep model

FEA prediction with rate model

Surface settlement near centreline (plate 8S) Settlement at 2 m depth (9A) and 4 m depth (10A) near centreline Heave near the toe (plates 2H)

Up to 5.7 m 5.7 - -7 m

Overestimated Satisfactory

Well predicted Underestimated

Up to 5.7 m 5.7--7 m

Satisfactory Satisfactory

Overestimated Satisfactory, slightly better Satisfactory Satisfactory, slightly better

Up to 5.7 m

At 3.4 m At 5 m

Well predicted Satisfactory

Near the centreline/ shoulder Near the toe At 3.4 m (448 h) At 5.7 m (475 h)

Overestimated

Satisfactory, overestimated Well predicted Overpredicted Satisfactory, slightly underpredicted Satisfactory Satisfactory, slightly better Satisfactory, slightly better

Well predicted

5.7--7 m At 3.4 m At 5.4 m

Satisfactory, overestimated Well predicted Satisfactory Satisfactory

Well predicted Fair-satisfactory

Well predicted Fair-satisfactory

Well predicted Overestimated

Underestimated

Underestimated

Satisfactory-well predicted

Horizontal displacement at the toe (221) Horizontal displacement near the toe (23I) Excess pore water pressure

Geotextile strain

Underestimated Underestimated

Satisfactory

Well predicted Well predicted Satisfactory

despite their inability to give accurate predictions of all the behaviour characteristics for the entire construction (i.e. vertical and horizontal deformations, excess pore pressures and geotextile strains at all the locations and time). The analysis with the creep m o d e l appears to give slightly better overall predictions for the Sackville foundation soil under working stress conditions but it requires an additional parameter, i.e. the coefficient of secondary c o m p r e s s i o n of the foundation soil, and consumes m u c h higher computing resource and time (e.g. over 15,000 load increments for the creep analysis c o m p a r e d to about 5500 for MCC). Similarly, the analysis with rate m o d e l predicted the horizontal disp l a c e m e n t near the e m b a n k m e n t toe and excess pore pressure in the foundation soil better than the M C C m o d e l but requires additional soil parameters such as fluidity constant and rate e x p o n e n t as well as greater computing resources. On the other hand, the M C C m o d e l was found to capture m a n y features of the e m b a n k m e n t behaviour reasonably well and therefore sufficient to predict the performance of Sackville foundation soils under working stress conditions.

Chapter 27

810 4.

CONCLUDING REMARKS

Two well-documented case histories of embankments constructed on soft soils of eastern Canada are presented in this chapter. Details regarding the layout, instrumentation, field performance and analyses were briefly discussed. The observed settlements of Hall's Creek embankment compared well with the predictions from Bjerrum's (1967) approach. The unique observation at the Hall's Creek site was the presence of high excess pore reaction pressures even 2 years after construction. It is speculated that progressive failure is leading to self-generating excess pore water pressures even though the embankment loading has remained constant. Predictability of the Sackville reinforced embankment for its behaviour under working stress conditions was examined in detail using three different finite element models, i.e. MCC, creep and rate model for the foundation soft soil. This study suggests that all three FEA models were capable of predicting the performance of this reinforced embankment under working stress conditions reasonably well despite their inability to give accurate predictions of all the behaviour characteristics for the entire construction (i.e. vertical and horizontal deformations, excess pore pressures and geotextile strains at all the locations and time). The analysis with the creep model gave slightly better overall predictions and that with the rate model predicted the horizontal displacement near the embankment toe and excess pore pressure in the foundation soil better than the MCC model. However, the creep and rate models require additional soil parameters and consume much larger computing resources and time. Therefore, the MCC model could be adequate for practical situations of predicting the performance of Sackville-type foundation soils under working stress conditions.

NOTATION C!

E e

ecs eN* K"

M M*

cohesion Young's modulus void ratio void ratio at unit mean normal pressure on the critical state line void ratio for the reference time at unit mean normal pressure on the isotropic normal consolidation line recompression index in the overconsolidated stress range hydraulic conductivity in the horizontal direction hydraulic conductivity in the vertical direction Coefficient of lateral earth pressure at rest Slope of the critical state line in ( q - p) space effective stress ratio at failure in ( 2 ~ 2 - O'm) space

Case Histories of Embankments on Soft Soils and Stabilisation n

Pa O~

cr~

7 1v V

811

strain rate parameter atmospheric pressure coefficient of secondary compression minor principal stress compression index in the normally consolidated stress range creep inclusive compression index in the normally consolidated stress range fluidity constant unit weight of foundation soil effective friction angle dilation angle poisson's ratio

REFERENCES Bergado, D.T., Long, EV., & Srinivasa Murthy, B.R. (2002) A case study of geotextile reinforced embankment on soft ground, Geotext. Geomembranes, 20, 343-365. Chen, W.F. (1982) Plasticity in Reinforced Concrete, McGraw-Hill, New York. Cormier, R.J. (1986) The Hall's Creek Test Fill, M.Sc.Eng. thesis, Department of Civil Engineering, University of New Brunswick, Canada. Crawford, C.B., Fannin, R.J. & Kern, C.B. (1995) Embankment failures at Vernon, British Columbia, Can. Geotech. J., 32, 271-284. Foot, R. & Ladd, C.C. (1981) Undrained settlement of plastic and organic clays, ASCE J. Geotech. Eng., 107(3), 1079-1094. Gnanendran, C.T. (1993) Observed and Calculated Behaviour of a Geotextile Reinforced Embankment on a Soft Compressible Soil, PhD thesis, University of Western Ontario, London, Ontario. Gnanendran, C.T., Manivannan, G. & Lo, S.-C.R. (2005 - in print). Influence of using a creep, elasto-viscoplastic or an elastoplastic model for predicting the behaviour of embankments on soft soils, Can. Geotech. J. (accepted May 2005). Gnanendran, C.T., Rowe, R.K. & Valsangkar, A.J. (2000) Back analyses of geosyntheticreinforced embankments on soft soils - Discussion, Geotext. Geomembranes J., 18, 63-65. Hussein, A.N. & McGown, A. (1998) The Behaviour of Two Trial Embankments at Perlis, Malaysia with Different Rates of Construction, Proceedings of 4th International Conference on Case Histories in Geotechnical Engineering, St. Louis, Missouri, USA, May 8-12, pp. 454-457. Ed. Shamsher Prakash, GeoResearch Int. Inc., 1633 Meadowbrook Road, Ottawa, Canada K1B 4W6. Indraratna, B., Balasubramaniam, A.S. & Balachandran, S. (1992) Performance of test embankment constructed to failure on soft marine clay, ASCE J. Geotech. Eng., 118(1), 12-33. Keenan, G.H., Landva, A.O., Valsangkar, A.J. & Comer, R.J. (1986) Performance and failure of test embankment on organic silty clay, Proceedings of the Conference on Building on Marginal and Derelict Land, Glasgow, The Institution of Civil Engineers, Vol. 2, pp. 417-428. Kutter, B.L., & Sathialingam, N. (1992) Elastoviscoplastic modelling of the rate-dependent behaviour of clays, Geotechnique, 42, 427-441. La Rochelle, P., Trak, B., Tavenas, E & Roy, M. (1974) Failure of a test embankment on a sensitive champlain clay deposit, Can. Geotech. J., 11, 142-164.

812

Chapter 27

Leroueil, S., Tavenas, E, Mieussens, C., & Peignaud, M. (1978b) Construction pore pressures in clay foundations under embankments. Part II: generalized behaviour, Can. Geotech. J., 15, 66-82. Leroueil, S., Tavenas, E, Trak, B., La Rochelle, P., & Roy, M. (1978a) Construction pore pressures in clay foundations under embankments. Part I: the Saint-Alban test fills, Can. Geotech. J., 15, 54-65. Li, A.L. & Rowe, R.K. (2002) Design considerations for embankments on rate sensitive soils, ASCE J. Geotech. Geoenviron. Eng., 128(11), 885-897. Ortigao, R.J.A., Werneck, M.L.G. & Lacerda, W.A. (1983) Embankment failure on clay near Rio De Janeiro, ASCE J. Geotech. Eng. Div., 109(11), 1460-1479. Palmeira, E.M., Pereira, J.H.E & da Silva, R.L. (1998) Back analyses of geosynthetic reinforced embankments on soft soils, Geotext. Geomembranes J., 16(5), 273-292. Perzyna, P. (1963) The Constitutive Equations for Work-Hardening and Rate Sensitive Plastic Materials, Proceedings of Vibration Problems, Warsaw, vol. 4, pp 281-290. Rampton, V.N. & Paradis, S. (1981) Quarternary Geology of A m h e r s t - Map area 21H, New Brunswick, Mineral Development Branch, Department of Natural Resources, Fredericton, New Brunswick, Map report 81-3. Rowe, R.K. & Gnavnendran, C.T. (1994) Geotextile strain in a full scale reinforced test embankment, Geotext Geomembranes, 13, 781-806. Rowe, R.K., Gnanendran, C.T., Landva, A.O. & Valsangkar, A.J. (1994) Behaviour of a Reinforced and an Unreinfored Test Embankment: A Comparison. Proceedings of the 5th International Conference on Geotextiles, Geomembrances and Related Products, Singapore, u pp. 5-10. Rowe, R.K., Gnanendran, C.T., Landva, A.O. & Valsangkar, A.J. (1995) Construction and performance of a full scale geotextile reinforced test embankment-Sackville, New Brunswick, Can. Geotech. J., 32, 512-534 and erratum, 33, 208. Rowe, R.K., Gnanendran, C.T., Landva, A.O., & Valsangkar, A.J. (1996) Calculated and observed behaviour of a reinforced embankment over soft compressible soil, Can. Geotech. J., 33, 324-338. Rowe, R.K. & Hinchberger, S.D. (1998) The significance of rate effects in modelling the Sackville test embankment, Can. Geotech. J., 35, 500-516. Rowe, R.K., Gnanendran, C.T., Valsangkar, A.J. & Landva, A.O. (2001) Performance of a test embankment constructed on an organic clayey silt deposit. Can. Geotech. J., 38, pp. 1283-1296. Rowe, R.K. & Li, A.L. (2002) Behaviour of reinforced embankments on soft rate sensitive soils, Geotechnique, 52(1), 29-40. Yin, J-H. & Zhu, J-G. (1999) Elastic viscoplastic consolidation modelling and interpretation of pore-water pressure responses in clay underneath Tarsiut Island, Can. Geotech. J., 36, 708-717.