You are on page 1of 9

Engineering Geology 209 (2016) 21–29

Contents lists available at ScienceDirect

Engineering Geology

journal homepage: www.elsevier.com/locate/enggeo

Variability in unsaturated hydraulic properties of residual soil


in Singapore
Qian Zhai a, Harianto Rahardjo b,⁎, Alfrendo Satyanaga a
a
School of Civil and Environmental Engineering, Nanyang Technological University, Block N1, B4B-07, Nanyang Avenue, 639798, Singapore
b
School of Civil and Environmental Engineering, Nanyang Technological University, Block N1, #1B-36, 50 Nanyang Avenue, Singapore 639798, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: Residual soil covers most of the land in Singapore and is commonly found in unsaturated conditions due to the
Received 9 August 2015 deep ground water table. Matric suction in residual soil plays an important role on the stability of residual soil
Received in revised form 21 March 2016 slopes against rainfall. Matric suction in the soil will change due to infiltration during rainfall, and infiltration is
Accepted 27 April 2016
mainly controlled by the unsaturated hydraulic properties (i.e., soil-water characteristic curve (SWCC) and hy-
Available online 10 May 2016
draulic conductivity) of the residual soil. Variability in SWCC and hydraulic conductivity of residual soils in
Keywords:
Singapore, such as Jurong Formation, Bukit Timah granite and Old Alluvium, were quantified and compared
Soil-Water characteristic curve with the uncertainty of saturated soil properties. A new framework for the estimation of SWCC from the saturat-
Residual soil ed hydraulic conductivity, ks, is proposed in this paper. The upper bound from the proposed framework is recom-
Hydraulic property mended for the infiltration analyses and the lower bound from the proposed framework is recommended for the
Slope stability slope stability analyses.
Infiltration analyses © 2016 Elsevier B.V. All rights reserved.

1. Introduction physical meaning, are compared with the COV of other saturated soil
properties. A new framework for estimation of the SWCC for residual
Singapore has a tropical rainforest climate with no distinctive sea- soil in Singapore from the saturated hydraulic conductivity, ks, is pro-
sons, uniform temperature and air pressure, high humidity, and abun- posed in this paper. The upper limit from the framework is suggested
dant rainfall. Rainfall-induced slope failure is common in tropical areas for the analyses of rainfall-induced slope failures as the upper limit of
and many cases have been reported in Singapore in the past few SWCC will result in the most conservative result. On the other hand,
years. The stability of slopes against rainfall is highly dependent on rain- the lower limit of SWCC is recommended to be used for the estimation
water infiltration which is controlled by the unsaturated hydraulic of the shear strength as the lower limit of SWCC will result in the most
properties, such as SWCC and the permeability function, of the soil. conservative estimation.
Therefore, the variability of SWCC and permeability function is impor-
tant for the rainfall-induced slope stability analyses.
Residual soil is developed or weathered from the underlying parent 2. Literature review
material and has similar general chemistry to the parent material
(Brady and Weil, 1996). Weathering of the parent material can occur Tsaparas et al. (2002) performed numerical analyses on rainfall-
through physical weathering (disintegration), chemical weathering induced landslides and concluded that the permeability function and
(decomposition) or chemical transformation. Therefore, the properties rainfall pattern could significantly affect the seepage pattern in the
of residual soil can vary drastically due to the varying parent materials soil. Rezaur et al. (2003) carried out field monitoring of the hydrologic
and the different forms of weathering. The different textures and ar- behavior of residual soil slopes in Singapore and indicated that the in-
rangements of soil particles can result in different SWCCs and hydraulic crease in pore-water pressure in the soil can be roughly estimated
conductivities. The variability in the unsaturated hydraulic properties of from daily rainfall. Tsaparas et al. (2003) concluded that changes in
residual soil is quantified using Zhai and Rahardjo's (2013, 2014) pore-water pressure in the soil were not only dependent on total rain-
equations. The coefficient of variations (COV) for SWCC variables, such fall but also on the initial pore-water pressure. Rahardjo et al. (2005)
as air-entry value, residual suction, and residual saturation which have studied the response of a residual soil slope to rainfall and suggested
that infiltrated rainwater rather than total rainfall, was important in
rainfall-induced slope failures. Rahimi et al. (2010) presented the effect
⁎ Corresponding author.
of variability in fitting parameters (i.e., a, n and m in Fredlund and Xing's
E-mail addresses: zhaiqian@ntu.edu.sg (Q. Zhai), chrahardjo@ntu.edu.sg (1994) equation) and saturated hydraulic conductivity on the variabili-
(H. Rahardjo), Alfrendo@ntu.edu.sg (A. Satyanaga). ty in simulated infiltration and factors of safety for slope stability.

http://dx.doi.org/10.1016/j.enggeo.2016.04.034
0013-7952/© 2016 Elsevier B.V. All rights reserved.
22 Q. Zhai et al. / Engineering Geology 209 (2016) 21–29

Rahardjo et al. (2010) also studied three response parameters for char-
acterization of infiltration: the depth of the wetting front, the matric
suction reduction depth and the sectional infiltration rate. These
research works indicate that unsaturated hydraulic properties can sig-
nificantly affect rainwater infiltration into residual soil. Therefore, un-
derstanding the variability in unsaturated hydraulic properties is very
important.
There are various factors that may lead to uncertainty in SWCC, and
these factors have been discussed by many researchers. Zapata (1999),
Zapata et al. (2000) examined uncertainty arising due to different best
fit equations, different operators, different numbers of data points and
different suction ranges measured for the regression process and differ-
ent methods used to obtain the SWCC data. Yaldo (1999) conducted
laboratory testing on four compacted cohesive soils to assess the impact
of soil type and compaction condition on SWCC. The results indicated
that a water content difference of 2% on either side of the optimum
water content did not have a significant effect on SWCC. These results
also indicated that the residual volumetric water content, θr, is a soil Fig. 1. The framework for prediction of SWCC from Zapata et al. (2000).
characteristic and not a function of the soil compaction condition, and
that θr increases with increasing plasticity. In order to minimize uncer- characteristics of residual soil could be related to the degree of
tainty in SWCC due to the soil specimen itself, Yaldo (1999) suggested weathering. Based on later work by Rahardjo et al. (2012), the averages
using a large-scale specimen for SWCC measurements (i.e., 45 cm diam- value and the coefficients of variation for index properties, such as
eter and 30 cm height). Gaharagheer (2009) also reported that the com- water content, w%, plastic limit, PL, liquid limit, LL, void ratio, e, effective
paction effort and molding water content did not have any significant cohesion, c’ (kPa), effective friction angle ϕ’ (degree), angle indicating
effects on SWCC for non-plastic soils. Gaharagheer's (2009) results are the rate of increase in shear strength relative to matric suction , ϕb,
in agreement with Yaldo's (1999) report. Furthermore, Tinjum et al. air-entry value, (AEV), residual suction, ψr, and residual saturation, Sr,
(1997) found that the air-entry value increased as the molding water of residual soil including Jurong sedimentary formation, Bukit Timah
content and compaction effort increased; the slope of the SWCC was granite and Old Alluvium are computed and presented in Table 1 As
steeper for soil compacted at the dry optimum than for soil compacted shown in Table 1, the COV of unsaturated soil properties is much higher
at the wet optimum; the dry unit weight had little effect on the SWCC than the index properties and saturated soil properties. Therefore, the
for compacted clay; and the shape of the SWCC was a function of the variability of unsaturated soil properties should be studied and
soil type. A statistical study by Gurdal et al. (2003) showed that param- quantified.
eter “a” in van Genuchten's (1980) equation had a strong relationship Tukey (1977) proposed the box-plot method to display robust sta-
with the saturated hydraulic conductivity (ks) and was inversely related tistics. The box-plot method is a non-parametric statistical method
to the clay content and remolding water content. However, parameter that displays variation in samples of a population without making any
“n” was less sensitive to the physical and basic properties than ks. Dye assumptions of the underlying distribution. The box-plot is a graphical
et al. (2011) presented the effect of uncertainty in SWCC on seepage technique that depicts, in its traditional form, five numeric summaries
analysis results and concluded that variability in SWCC had a significant about a data set in order to visualize its dispersion and skewness
effect on the computed pore-water pressure. Ye et al. (2012) also indi- (McGill et al., 1978). In this paper, the method of the confidence limit
cated that variability in the SWCC and hydraulic conductivity may result from Zhai and Rahardjo (2013, 2014) and the box-plot method are
from temperature variation. adopted for quantification of variability in the SWCC and hydraulic con-
Fredlund and Houston (2009) provided recommendations and sug- ductivity of residual soil.
gestions for the determination and use of the soil–water characteristic Zapata (1999); Zapata et al. (2000) and Perera et al. (2005) proposed
curve and consequently, for the computation of the unsaturated soil the framework, as shown in Fig. 1, for estimation of the SWCC from the
property function (USPF). Winn et al. (2001) presented the saturated index properties such as D60, where D60 = grain diameter correspond-
soil properties of two residual soils, which were derived from Bukit ing to 60% passing by weight or mass (mm), for the non-plastic soil and
Timah granite and Sedimentary Jurong Formation, in Singapore. Agus wPI, where wPI = % passing #200 xPI, PI = the plastic limit, for the plas-
et al. (2001) presented 12 sets of soil-water characteristic curve tic soil. As shown in Fig. 1, a single curve is correlated with D60 and wPI.
(SWCC) data for the same two residual soils in Singapore. By performing Due to the high variability in SWCC for the residual soil in Singapore as
a series of tests on residual soils from Bukit Timah granite and Jurong presented in Table 1, a single curve may not be sufficient to predict the
Formation, including tests of the index properties and engineering SWCC for residual soil in Singapore. Therefore, the single curve in
properties, mercury porosimetry tests and scanning electron micro- Zapata's (1999); Zapata et al.’s (2000) and Perera et al.’s (2005) frame-
scope (SEM) examination, Rahardjo et al. (2004) concluded that work is modified as the upper bound and the lower bound for estimation
variation in the engineering properties as well as micro structural of the SWCC of the residual soil in Singapore to represent its variability.

Table 1
Average value (AV) and coefficient of variation (COV) for properties of residual soil in Singapore (computed from Rahardjo et al., 2012).

Index properties Saturated soil properties Unsaturated soil properties

w% e c′ ϕ′ ϕb AEV ψr
in situ PL LL In situ (kPa) (degree) (degree) (kPa) (kPa) Sr

Jurong Formation AV 17.36 20.54 42.54 0.49 11.58 32.33 28.28 41.66 4187.37 0.20
COV 29.92% 19.63% 18.72% 23.58% 29.68% 9.15% 22.72% 70.19% 49.73% 58.58%
Bukit Timah granite AV 32.00 33.69 43.38 0.51 8.83 36.83 31.60 16.74 2142 0.22
COV 26.03% 21.62% 17.52% 18.58% 43.74% 9.12% 23.28% 127.22% 98.32% 93.37%
Old Alluvium AV 33.00 35.82 56.82 0.59 20.17 36.17 31.30 18.90 2754.39 0.22
COV 28.29% 24.58% 13.35% 16.50% 36.98% 10.79% 28.42% 58.30% 106.39% 76.31%
Q. Zhai et al. / Engineering Geology 209 (2016) 21–29 23

Table 2
Soil properties of residual soil in Singapore.

S/N Soil Name: USCS ks (m/s) Gs LL(%) PL(%) PI(%) Sands(%) Fines(%) Formation Reference
d(%)

1 NTU-1a CL 1.24E-08 2.77 39 29 10 14 86 Jurong Formation Agus et al. (2001), (2005)


2 NTU-1b CL 1.00E-08 2.63 32 5 27 25 75 Jurong Formation Agus et al. (2001), (2005)
3 NTU-1c SM 3.65E-08 2.73 28 18 10 63 37 Jurong Formation Agus et al. (2001), (2005)
4 NTU-2a CL 4.53E-08 2.63 39 23 16 28 72 Jurong Formation Agus et al. (2001), (2005)
5 NTU-2b CL 3.59E-09 2.67 47 26 21 23 77 Jurong Formation Agus et al. (2001), (2005)
6 NTU-2c CL 5.73E-10 2.72 34 21 13 16 84 Jurong Formation Agus et al. (2001), (2005)
7 NTU-3a MH 3.05E-09 2.52 55 33 22 27 73 Jurong Formation Agus et al. (2001), (2005)
8 NTU-3b ML 1.18E-09 2.56 36 26 10 50 50 Jurong Formation Agus et al. (2001), (2005)
9 NTU-3c CL 4.96E-09 2.72 44 30 14 8 92 Jurong Formation Agus et al. (2001), (2005)
10 NTU-4a ML 2.33E-09 2.67 42 27 15 44 56 Jurong Formation Agus et al. (2001), (2005)
11 NTU-4b ML 1.04E-09 2.76 43 28 15 44 56 Jurong Formation Agus et al. (2001), (2005)
12 NTU-4c ML 2.89E-09 2.63 35 26 9 79 21 Jurong Formation Agus et al. (2001), (2005)
13 JK-1 SC 8.21E-06 2.53 47 33 14 69 27 Jurong Formation Rahardjo et al. (2011)
14 JK-2 SC 8.21E-06 2.64 30 13 17 73 27 Jurong Formation Rahardjo et al. (2011)
15 OSC-1 CL 5.18E-06 2.67 47 34 13 – – Jurong Formation Rahardjo et al. (2005)
16 OSC-2 CL 5.18E-06 2.67 47 34 13 – – Jurong Formation Rahardjo et al. (2005)
17 PCS-1 CL 1.67E-07 2.72 34 13.5 20.5 – – Jurong Formation Rahardjo et al. (2005)
18 PCS-2 CL 1.67E-07 2.72 34 13.5 20.5 – – Jurong Formation Rahardjo et al. (2005)
19 BM1 CL 1.30E-06 2.63 49 24 25 0 100 Jurong Formation Rahardjo et al. (2010)
20 BM2 CL 2.39E-06 2.67 45 25 20 5 95 Jurong Formation Rahardjo et al. (2010)
21 BM3 CL 2.39E-06 2.68 39 19 20 3 97 Jurong Formation Rahardjo et al. (2010)
22 HR MH 9.20E-07 2.62 62 36 26 37 63 Jurong Formation Rahardjo et al. (2010)
23 CC CL 8.58E-06 2.72 29 18 11 14 86 Jurong Formation Rahardjo et al. (2010)
24 DR MH 7.00E-07 2.69 45 24 21 20 80 Jurong Formation Rahardjo et al. (2010)
25 TB CL 3.60E-05 2.67 41 21 20 1 99 Jurong Formation Rahardjo et al. (2010)
26 BP CL 5.60E-06 2.7 – – – 25 75 Jurong Formation Rahardjo et al. (2012)
27 JKL CL 1.40E-05 2.69 38 18 20 2 98 Jurong Formation Rahardjo et al. (2012)
28 JBM CL 3.30E-05 2.68 39 19 20 3 97 Jurong Formation Rahardjo et al. (2012)
29 SLR-1a SM 2.50E-08 2.6 108 47 61 61 39 Bukit Timah Granite Agus et al. (2001), (2005)
30 SLR-1b SM 7.36E-08 2.61 64 34 30 72 28 Bukit Timah Granite Agus et al. (2001), (2005)
31 SLR-1c SM 2.88E-07 2.55 60 35 25 75 25 Bukit Timah Granite Agus et al. (2001), (2005)
32 SLR-2a MH 2.78E-08 2.6 105 47 58 38 62 Bukit Timah Granite Agus et al. (2001), (2005)
33 SLR-2b MH 3.00E-08 2.61 116 48 68 38 62 Bukit Timah Granite Agus et al. (2001), (2005)
34 SLR-2c CH 1.36E-08 2.55 61 36 25 43 57 Bukit Timah Granite Agus et al. (2001), (2005)
35 YS-a MH 7.27E-08 2.61 54 33 21 44 56 Bukit Timah Granite Agus et al. (2001), (2005)
36 YS-b MH 4.80E-07 2.53 68 49 19 32 68 Bukit Timah Granite Agus et al. (2001), (2005)
37 LA-a SC 9.51E-07 2.53 48 37 11 66 34 Bukit Timah Granite Agus et al. (2001), (2005)
38 LA-b SM 5.37E-08 2.6 97 38 59 72 28 Bukit Timah Granite Agus et al. (2001), (2005)
39 LA-c SM 1.86E-08 2.54 61 4 57 62 38 Bukit Timah Granite Agus et al. (2001), (2005)
40 LA-d SM 3.57E-08 2.6 97 38 59 75 25 Bukit Timah Granite Agus et al. (2001), (2005)
41 MR1 SM 6.00E-06 2.66 52 31 21 53 46 Bukit Timah Granite Rahardjo et al. (2011)
42 MR2 MH 3.30E-05 2.58 54 39 15 40 60 Bukit Timah Granite Rahardjo et al. (2011)
43 SS SM 2.20E-06 2.59 48 27 21 61 38.9 Bukit Timah Granite Tami et al. (2007)
44 TR1 CH 3.19E-08 2.59 71 34 37 27 73 Bukit Timah Granite Rahardjo et al. (2012)
45 TR2 MH 5.65E-05 2.66 51 29 22 30 70 Bukit Timah Granite Rahardjo et al. (2012)
46 BBA1 CL 5.21E-08 2.82 31 20 11 43 54 Bukit Timah Granite Rahardjo et al. (2012)
47 BBA2 CL 8.28E-08 2.64 34 20 14 20 80 Bukit Timah Granite Rahardjo et al. (2012)
48 BB1 MH 1.20E-08 2.65 55 32 23 3 97 Bukit Timah Granite Rahardjo et al. (2012)
49 BB2 SM 2.15E-08 2.71 49 30 19 51 49 Bukit Timah Granite Rahardjo et al. (2012)
50 AMK1 SM 8.52E-06 2.54 56 38 18 58 42 Bukit Timah Granite Rahardjo et al. (2012)
51 AMK2 SM 3.96E-06 2.58 58 36 22 53 47 Bukit Timah Granite Rahardjo et al. (2012)
52 AMA MH 2.52E-07 2.7 54 29 25 29 70 Bukit Timah Granite Rahardjo et al. (2012)
53 TWL MH 5.60E-06 2.66 72 45 27 5 95 Old Alluvium Rahardjo et al. (2012)
54 JS1 SM 1.61E-06 2.66 72 28 44 62 6 Old Alluvium Rahardjo et al. (2012)
55 JS2 SM 4.73E-06 2.61 54 28 26 62 11 Old Alluvium Rahardjo et al. (2012)
56 BN1 CL 1.40E-08 2.72 49 27 22 36 64 Old Alluvium Rahardjo et al. (2012)
57 BN2 CL 1.40E-08 2.71 57 27 30 7 93 Old Alluvium Rahardjo et al. (2012)
58 BN3 SC 5.00E-05 2.67 34 18 16 61 20 Old Alluvium Rahardjo et al. (2012)
59 Tam CL 9.00E-05 2.68 60 36 24 73 25 Old Alluvium Rahardjo et al. (2012)

3. Theory (2013, 2014) proposed equations for determination of the confidence


limits of the best fitted SWCC as follows:
The method for estimation of the confidence limits of the best fitted
SWCC, which defines the upper and lower bounds of the curve, from ob-
served experimental data is presented. The estimation of hydraulic con-
ductivity (i.e., kw function) from the SWCC using the statistical method
is also presented.
θs
when 0 b ψ b a max ; θupper ¼ C ðψÞ n h  n max iom min
ψ
3.1. Confidence limits of best fitted SWCC ln e þ a max
ð1Þ
θs
when ψNa max ; θupper ¼ C ðψÞ n h  n min iom min
By adopting a first-order error analysis approach and the t-statistic ψ
ln e þ a max
for estimation of variation in the fitting parameters, Zhai and Rahardjo
24 Q. Zhai et al. / Engineering Geology 209 (2016) 21–29

θs
when 0 b ψ b a min ; θlower ¼ C ðψÞ n h 
n min iom max
ψ
ln e þ a min
ð2Þ
θs
when ψNa max ; θupper ¼ C ðψÞ n h  n min iom min
ψ
ln e þ a max

where,

a, n, m fitting parameters in Fredlund and Xing's (1994) equation


C(ψ) correction factor in Fredlund and Xing's (1994) equation
θ volumetric water content
θs saturated volumetric water content.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a max ¼ a þ t α=2 Var ðaÞ a min ¼ a−t α=2 Var ðaÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n max ¼ n þ t α=2 Var ðnÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n min ¼ n−t α=2 Var ðnÞ m max ¼ m þ t α=2 Var ðmÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi :
m min ¼ m−t α=2 VarðmÞ

3.2. Estimation of kw function from SWCC

Zhai and Rahardjo (2015) presented SWCC in the form of degree of


saturation, which could be considered analogous to the pore-size distri-
bution function. By considering the pores in the soil as series of capillary
tubes and these tubes to be randomly connected, Zhai and Rahardjo
(2015) derived the equation correlating the SWCC and saturated hy-
draulic conductivity, ks, and unsaturated hydraulic conductivity, k(ψ),
as follows:
( )
N h i
1 2 X
ks ¼ n ½1−Sðψi Þ2 −½1−Sðψi−1 Þ2 r 2i ; ð3Þ
8 i¼1

8    2 2 9
> >
< S ψmþi −S ψmþiþ1 r mþi þ
>

>
=
nmþi 2 XN     2     2
>
> S ψmþi −S ψ j − S ψmþi −S ψ j−1 rj >
2
>
  : ;
j¼mþiþ1
k ψmþi ¼ kðψm Þ 8  2 2 9
> >
< Sðψm Þ−S ψmþ1 r m þ
> >
=
N h
X i
nm 2
>
> ðS ðψ Þ−S ðψ ÞÞ2
−ðS ðψ Þ−S ðψ ÞÞ2 2
r >
i >
: m i m i−1 ;
i¼mþ1

ð4Þ

where,

ks = saturated hydraulic conductivity;


k(ψm + i), k(ψm) = hydraulic conductivity corresponding to suctions
ψm+i and ψm, respectively;
ψm+i, ψm = suction in the soil, ψm+i, Nψm;
nm + i, nm = porosity corresponding to suctions of ψm + i and ψm,
respectively;
Fig. 2. SWCCs for residual soil in Singapore.
rm.+i, rm. = equivalent pore radius of pores corresponding to suction of
ψm+i and ψm using Kelvin's capillary law, respectively and (a) SWCCs for residual soil from Jurong Formation.
(b) SWCCs for residual soil from Bukit Timah granite.
S(ψm+i), S(ψm) = degree of saturation in the soil corresponding to suc-
(c) SWCCs for residual soil from Old Alluvium.
tion of ψm+i and ψm, respectively.

Eq. (3) correlates the SWCC and saturated hydraulic conductivity, ks


which suggests the possibility that SWCC can be estimated from ks. formations: (i) igneous rocks of granite (Bukit Timah Granite) in the
Eq. (4) suggests that unsaturated hydraulic conductivity k(ψ) could be center and northwest, (ii) sedimentary rocks (Jurong Formation) in
estimated from a measured hydraulic conductivity k(ψm). Zhang et al. the west, and (iii) a semi-hardened alluvium (Old Alluvium) which
(2015) suggest k(ψm) to be measured near the point of the air-entry covers older rocks beneath in the east of Singapore (PWD, 1976). Gran-
value. ite occurs in two separate masses. The larger one is found in the central
and northern areas, the smaller one in north eastern parts of Singapore.
4. Unsaturated hydraulic properties of residual soil in Singapore Granite or igneous rocks underlie the Bukit Timah Nature Reserve and
Central Catchment Area in the centre of the island. The granite in
The Singapore Island is situated around 15 m above sea level Singapore, according to radioactive age determination, is more than
(PWD, 1976). The geology of Singapore consists essentially of three 200 million years old. The Sedimentary rocks of Jurong Formation
Q. Zhai et al. / Engineering Geology 209 (2016) 21–29 25

form extensive areas in southern, south western and western parts of Formation and 19 datasets for Bukit Timah granite analyzed by Rahardjo
Singapore. These variations of conglomerate, sandstone and shale are et al. (2012) while 28 datasets for Jurong Formation and 25 datasets of
also observed on the island to the south to west. The semi-hardened Bukit Timah granite were analyzed in Table 3.
Old Alluvium was deposited by an ancient river system, probably in
the Pleistocene epoch, during a low stand of the sea. (PWD, 1976). 5.1. Assessment of variability in SWCCs for residual soil using Zhai and
A total of 59 datasets for residual soils in Singapore were collected Rahardjo's (2013, 2014) method and the box-plot method
from the published literature. The soil properties are summarized in
Table 2. The SWCCs of the residual soils are illustrated in Fig. 2. The assessment of the variability in SWCCs for residual soil — such as
Jurong Formation, Bukit Timah granite and Old Alluvium — using the
5. Quantification of variability in saturated and unsaturated soil confidence limits from Zhai and Rahardjo (2013, 2014) and the box-
properties of residual soil plot method is illustrated in Fig. 3.
Fig. 3 indicates that the variability in SWCCs as quantified using the
The variability in SWCC for residual soils from Jurong Formation, confidence limits from Zhai and Rahardjo (2013, 2014) agrees with the
Bukit Timah granite and Old Alluvium were assessed using Zhai and variability quantified using the box-plot method. Fig. 3 also illustrates
Rahardjo's (2013, 2014) equations and presented using the box-plot that the median is not located at the center of the box, which means
method. The air-entry values, residual suction and residual saturation that the water contents at different suctions follow a biased distribution
of the residual soils were calculated from the fitting parameters using rather than an unbiased distribution (e.g., normal distribution, t-
Zhai and Rahardjo's (2012) equations, as given in Eqs. (5) to (7). distribution).

1−Si
ψb ¼ ψi 0:1 s1
ð5Þ 5.2. Assessment of variability in hydraulic conductivity for residual soil
using Zhai and Rahardjo's (2013, 2014) method and the box-plot method
Si −S0 þs1 logðψi Þ−s2 logðψ0 Þ

ψr ¼ 10 s1 −s2
ð6Þ As very limited experimental data have been reported on unsaturat-
  ed hydraulic conductivity, the hydraulic conductivities of unsaturated
Sr ¼ Si −s1 log ψr ψ ð7Þ residual soils were predicted using the statistical model with Zhai and
i
Rahardjo's (2015) equation. The assessment of the variability in the pre-
dicted hydraulic conductivities for residual soils from Jurong Formation,
where,
Bukit Timah granite and Old Alluvium using confidence limits obtained
from Zhai and Rahardjo (2013, 2014) and the box-plot method are illus-
ψb = air-entry value, (kPa),
trated in Fig. 4.
s1 = slope at the inflection point,
Fig. 4 indicates that the variability in the predicted hydraulic conduc-
Si = degree of saturation corresponding to the inflection point,
tivity using the confidence limits from Zhai and Rahardjo (2013, 2014)
ψi = matric suction corresponding to the inflection point,
agrees with that obtained using the box-plot method. Therefore, the
s2 = the slope of the curve on the point where the slope behaves
method using the confidence limits from Zhai and Rahardjo (2013,
linearly,
2014) is a reliable method for the indirect estimation of variability in
S′ = degree of saturation corresponding the point where the slope be-
unsaturated hydraulic properties.
haves linearly,
ψi = matric suction corresponding to the inflection point where the
5.3. Effect of variability of SWCC on the permeability function and shear
slope behaves linearly,
strength
ψr = residual suction (kPa) and
Sr = residual saturation.
The upper band, the lower band and the best fitted SWCC for Jurong
Formation as illustrated in Fig. 5 (a) is used as a demo for the prediction
The comparison of the average values and coefficients of variation
of the permeability function and the shear strength. Zhai and Rahardjo's
(COV) of saturated soil properties such as water content, w%, plastic
(2015) equation as illustrated in Eq. (4) is used for prediction of the per-
limit, PL, liquid limit, LL, void ratio, e, effective cohesion, c’, effective fric-
meability function. It is assumed that hydraulic conductivity of soil is
tion angle, ϕ’, and saturated hydraulic conductivity, ks, and unsaturated
constant and equal to ks = 6 × 10−6m/s when the suction is less than
soil properties — such as air-entry value, AEV, residual suction, ψr, and re-
the air-entry value, which is a reasonable assumption for the soil with
sidual saturation, Sr, — for the residual soils is summarized in Table 3. As
insignificant soil volume change (Zhai et al., 2015). The predicted per-
illustrated in the table, the COV of unsaturated soil properties is much
meability function is illustrated in Fig. 5 (b). Vanapalli et al.’s (1996)
higher (i.e., around 3 to 5 times) than that of saturated soil properties.
equation, as shown in Eq. (8), is used for the estimation of the shear
The AV and COV of the air-entry value, AEV, residual suction, ψr, and
strength for the unsaturated soil.
residual saturation, Sr, for Jurong Formation and Bukit Timah granite as
shown in Table 3 are different from that presented by Rahardjo et al.
(2012) as illustrated in Table 1. There were total 18 datasets for Jurong τ ¼ c0 þ ðσ −ua Þ tanϕ0 þ ðua −uw ÞΘκ tanϕ0 ð8Þ

Table 3
Comparison of average value and COV of soil properties for residual soil in Singapore.

Index properties Saturated soil properties Unsaturated soil properties

w% PL LL e c′ (kPa) ϕ′ Log(ks) ϕb Air-entry value, Residual suction, Residual


in situ In situ (degree) (degree) AEV (kPa) ψr (kPa) saturation, Sr

Jurong Formation AV 17.36 20.54 42.54 0.49 11.58 32.33 −6.71 28.28 34.18 3238.25 0.30
COV 29.92% 19.63% 18.72% 23.58% 29.68% 9.15% 22.46% 22.72% 96.95% 72.37% 77.12%
Bukit Timah granite AV 32.00 33.69 43.38 0.51 8.83 36.83 −6.62 31.60 16.58 2189.76 0.24
COV 26.03% 21.62% 17.52% 18.58% 43.74% 9.12% 17.05% 23.28% 126.73% 100.49% 83.74%
Old Alluvium AV 33.00 35.82 56.82 0.59 20.17 36.17 −5.86 31.30 18.90 2754.39 0.22
COV 28.29% 24.58% 13.35% 16.50% 36.98% 10.79% 28.51% 28.42% 58.30% 106.39% 76.31%
26 Q. Zhai et al. / Engineering Geology 209 (2016) 21–29

1.2 -4
Box plot of 28 sets of Jurong Formation
Upper limit from Zhai and Rahardjo (2013) Box plot from 28 sets of Jurong formation
Lower limit from Zhai and Rahardjo (2013) Lower limit from Zhai and Rahardjo (2013)
Experimental data Upper limit from Zhai and Rahardjo (2013)
1.0 -6
Degree of saturation, S

0.8
-8

log(k(ψ))
0.6

-10
0.4

-12
0.2

0.0 -14
10-2 10-1 100 101 102 103 104 105 106 10-2 10-1 100 101 102 103 104 105 106
Matric Suction, ψ (kPa) Matric Suction, ψ (kPa)
(a) Variability in SWCC for residual soil from Jurong Formation; (a) Variability in permeability function for residua
soil from Jurong Formation;
1.2 -2
Box plot of 25 sets of Bukit Timah granite
Box plot from 25 sets of Bukit Timah granite
Upper limit from Zhai and Rahardjo (2013)
Upper limit from Zhai and Rahardjo (2013)
Lower limit from Zhai and Rahardjo (2013)
1.0 -4 Lower limit from Zhai and Rahardjo (2013)
Experimental data
Degree of saturation, S

0.8 -6
log(k(ψ))

0.6 -8

0.4 -10

0.2 -12

0.0 -14
10 -2 10 -1 10 0 10 1 10 2 10 3 10 4 10 5 10 6 10-2 10-1 100 101 102 103 104 105 106
Matric Suction, ψ (kPa) Matric Suction, ψ (kPa)
(b) Variability in SWCC for residual soil from Bukit Timah granite; (b) Variability in permeability function for residua
soil from Bukit Timah granite;
1.2 Box plot of 6 sets of Old Alluvium -2
Upper limit from Zhai and Rahardjo (2013) Box plot from 6 sets of Old Alluvium
Lower limit from Zhai and Rahardjo (2013) Upper limit from Zhai and Rahardjo (2013)
Experimental data Lower limit from Zhai and Rahardjo (2013)
1.0 -4
Degree of saturation, S

0.8 -6
log(k(ψ))

0.6 -8

0.4
-10

0.2
-12

0.0
-14
10 -2 10 -1 10 0 10 1 10 2 10 3 10 4 10 5 10 6
10-2 10-1 100 101 102 103 104 105 106
Matric Suction, ψ (kPa)
Matric Suction, ψ (kPa)
(c) Variability in SWCC for residual soil from Old Alluvium; (c) Variability in permeability function for residua
soil from Old Alluvium;
Fig. 3. Variability in SWCC for residual soils in Singapore.
Fig. 4. Variability in hydraulic conductivity for residual soils in Singapore.
τ = shear strength of unsaturated soil,
c′ = effective cohesion,
(σn − ua) = net normal stress, ua = pore-air pressure,
(ua − uw) = matric suction, uw = pore-water pressure and
ϕ’ = effective internal friction angle of saturated soil, Θ = normalized volumetric water content defined as θθ−θ r
s −θr
.
Q. Zhai et al. / Engineering Geology 209 (2016) 21–29 27

Table 4
Best fitted SWCC
1.2 Upper limit from Zhai and Rahardjo (2013) Results of regression analyses between air-entry values of residual soil and physical soil
Lower limit from Zhai and Rahardjo (2013) properties.

1.0 Regression Degree of Degree of R2


Degree of saturation, S

freedom for freedom for


regression dfR residual dfE
0.8
y = cx1 + b (x1 = LL) 1 53 7.40e-5
y = cx2 + b (x2 = PL) 1 53 0.019
0.6 y = cx3 + b (x3 = PI) 1 53 0.007
y = cx4 + b (x4 = sand%) 1 53 0.005
0.4 y = cx5 + b (x5 = fine%) 1 53 0.002
y = cx6 + b (x6 = log(ks)) 1 53 0.140
y = cx7 + b(x7 = sand% ∗ PI) 1 53 3.66e-05
0.2 y = cx8 + b(x8 = fine% ∗ PI) 1 53 0.018
y = cx9 + b(x9 = log(ks) ∗ PI) 1 53 0.029
y = cx10 + b(x10 = sand% ∗ log(ks)) 1 53 0.001
0.0
y = cx11 + b (x11 = fine% ∗ log(ks)) 1 53 0.068
10-2 10-1 100 101 102 103 104 105 106
y = cx12 + b (x12 = PI2) 1 53 0.012
Matric suction, ψ (kPa) y = cx13 + b(x13 = (log(ks))2) 1 53 0.155
(a) Illustration of the best fitted, upper band and lower bounds
of SWCC for residual soil from Jurong formation.
As shown in Fig. 5 (b), the predicted permeability function from the
10-3
upper bound is much higher than that predicted results from the other
Computed from best fitted SWCC
10-4 Computed from upper limit two. Higher permeability means easier to be infiltrated during the rain-
Computed from lower limit fall. In other words, upper bound can be used for infiltration analysis
Permeability function, k(ψ) (m/s)

10-5
and conservative result (i.e., lower suction profile) is expected to be ob-
10-6 tained. Fig. 5 (c) indicates that estimation of the unsaturated shear
10-7 strength from the lower bound gives most conservative result. In
other word, lower bound can be used for the slope stability analysis
10-8
and conservative factor of safety is expected to be obtained.
10-9

10-10
5.4. Regression analysis for correlation of SWCC and soil basic properties
10-11

10-12 Thirteen sets of variables (i.e., x1 to x13) as illustrated in Table 4, were


defined from the physical properties of soil, including LL, PL, PI, sand%,
10-13
fine%, log (ks) and their combinations. Regression analyses between
10-2 10-1 100 101 102 103 104 105 106 107
the air-entry value of each residual soil and the physical properties of
Matric suciton, ψ (kPa)
the soil were carried out. The results of the regression analyses
(b) Illustration of predicted permeability function from the (i.e., coefficient of determination, R2) are illustrated in Table 4.
best fitted, upper bound and lower bound of SWCC. Table 4 shows that the coefficients of determination R2 for the re-
gression between the air-entry value, and x6 (log(ks)) and x13
500 (log(ks)2) were the highest (i.e. 0.140 and 0.155, respectively). On the
Estimated unsaturated shear strength, τ (kPa)

Estimated from the best fitted SWCC other hand, the R2 for the regression between the air-entry values and
Estimated from the upper limit
Estimated from the lower limit x8 (fine%PI) was only 0.018. Table 4 indicates that the air-entry value
400
is most related to saturated hydraulic conductivity, ks, based on correla-
tion equations presented in Table 4. In a previous study, Zhai and
300
Rahardjo (2015) derived an equation that correlates the SWCC and ks
in a highly non-linear relationship. Therefore, it is reasonable to make
a prediction of the SWCC from the saturated hydraulic conductivity,
200 ks, for residual soil in Singapore.

100 5.5. A new framework for prediction of SWCC from saturated hydraulic
conductivity, ks

0 Many factors can affect the shape of the SWCC. If the SWCC is pre-
100 200 300 400 500
dicted from one basic soil property, then variability is expected in the
Matric suction, ψ (kPa) results. Confidence limits were incorporated to set the boundaries of
(c) Illustration of estimated shear strength from the best fitted, the predicted SWCC from ks . The SWCCs for residual soils in
upper bound and lower bound of SWCC. Singapore were correlated with ks and the results are illustrated in
Fig. 6.
Fig. 5. Illustration of the variability in predicted permeability function and estimated shear Fig. 6 indicates that the variability in the predicted SWCCs
strength due to the variability in the SWCC.
(i.e., the bands of the confidence limits) is a function of ks: a higher
or lower ks value results in a narrower band of the predicted SWCC.
σ = 10 kPa, c’ = 10 kPa, ϕ’ = 35 kPa κ = 1 is used for the estimation In other words, low variability in the SWCC is expected for pure
of the unsaturated shear strength and the estimated results are illustrat- clay or pure sand, whereas high variability in the SWCC is expected
ed in Fig. 5 (c). for mixed soil.
28 Q. Zhai et al. / Engineering Geology 209 (2016) 21–29

(a) (b)

(c) (d)

(e) (f)
Fig. 6. SWCCs of residual soil with respect to saturated hydraulic conductivity, ks.

6. Conclusions of the box-plot. The method of confidence limits from Zhai and
Rahardjo (2013, 2014) was verified as a reliable indirect method for
The variability in SWCC and hydraulic conductivity for residual soils the estimation of variability in unsaturated hydraulic properties. A
in Singapore was presented in the form of confidence limits from the new simple framework for prediction of SWCCs for residual soils in
model presented by Zhai and Rahardjo (2013, 2014) and in the form Singapore from saturated hydraulic conductivity, ks, was proposed in
Q. Zhai et al. / Engineering Geology 209 (2016) 21–29 29

this paper. In order to get conservative results, the upper limit of the Rezaur, R.B., Rahardjo, H., Leong, E.C., Lee, T.T., 2003. Hydrologic behaviour of residual soil
slopes in Singapore. J. Hydraul. Eng. ASCE 8 (3), 133–144 May/June.
SWCC from the proposed framework is recommended for the infiltra- Tami, D., Rahardjo, H., Leong, E.C., 2007. Characteristics of Scanning Curves of Two soils.
tion analysis while the lower limit of the SWCC from the proposed Soils Found. 47 (1), 97–109 (Japanese Geotechnical Society, February).
framework is recommended for the slope stability analysis. Tinjum, J.M., Benson, C.H., Bolt, L.R., 1997. Soil-water characteristic curves for compacted
clays. J. Geotech. Geoenviron. 1060–1069.
Tsaparas, I., Rahardjo, H., Toll, D.G., Leong, E.C., 2002. Controlling parameters for rainfall-
References induced landslides. Comput. Geotech. 29 (1), 1–27.
Tsaparas, I., Rahardjo, H., Toll, D.G., Leong, E.C., 2003. Infiltration characteristics of two in-
Agus, S.S., Leong, E.C., Rahardjo, H., 2001. Soil-water characteristic curves of Singapore re- strumented residual soil slopes. Can. Geotech. J. 40 (5), 1012–1032 October.
sidual soils. J. Geotech. Geoenviron. Eng. 19 (3–4), 285–309 (Special Issue on “Unsat- Tukey, J.W., 1977. Exploratory Data Analysis. Addison-Wesley, Reading, MA.
urated and Collapsible Soils,” September). Van Genuchten, M.T., 1980. A close form equation predicting the hydraulic conductivity of
Agus, S.S., E.C. Leong and H. Rahardjo (2005). “Estimating permeability functions of unsaturated soil. Soil Sci. Soc. Am. J. 44, 892–898.
Singapore residual soils“. J. Eng. Geol., April, Vol. 78, No. 1–2, pp. 119–133. Vanapalli, S.K., Fredlund, D.G., Pufahl, D.E., Clifton, A.W., 1996. Model for the prediction of
Brady, N., Weil, R., 1996. The nature and properties of soils. Upper Saddle River. Prentice shear strength with respect to soil suction. Can. Geotech. J. 33, 379–392.
Hall, NJ. Winn, K., Rahardjo, H., Seh, C.P., 2001. Characterization of residual soils in Singapore.
Dye, H.B., Houston, S.L., B.D, W., 2011. Influence of unsaturated soil properties uncertainty J. Geotech. Eng. 32 (1), 1–13 April.
on moisture flow modeling. Geotech. Geol. Eng. 29 (161–169). Yaldo, K., 1999. Impact of soil type and compaction conditions on soil-water characteristic
Fredlund, D.G., Houston, S.L., 2009. Protocol for the assessment of unsaturated soil prop- curves Ph.D. Dissertation Wayne state University, Detriot, Michigan.
erties in geotechnical engineering practice. Can. Geotech. J. 46, 694–707. Ye, W.M., Wan, M., Chen, B., Chen, Y.G., Cui, Y.J., Wang, J., 2012. Temperature effects on the
Fredlund, D.G., Xing, A., 1994. Equations for the soil-water characteristic curve. Can. unsaturated permeability of the densely compacted GMZ01 bentonite under con-
Geotech. J. 31 (3), 521–532. fined condition. Eng. Geol. 126, 1–7.
Gaharagheer, F.S.T., 2009. Estimating hydraulic properties of unsaturated compacted cohe- Zapata, C.E., 1999. Uncertainty in Soil-Water Characteristic Curve and Impacts on Unsat-
sive soils from basic geotechnical soil properties PhD thesis The University of Toledo. urated Shear Strength Predictions pH.D. Dissertation Arizona State University,
Gurdal, T., Benson, C.H., Albright, W., 2003. Hydrologic properties of final cover soils from Tempe, United States.
the alternative cover assess program. Geo Engineering Report #03–02. University of Zapata, C.E., Houston, W.N., Houston, S.L., Walsh, K.D., 2000. Soil-water characteristic
Wisconsin_Madison. curve variability. In: Shackelford, C.D., Houston, S.L., Chang, N.-Y. (Eds.), Advances
McGill, R., Tukey, J.W., Larsen, W.A., 1978. Variation of box plots. Am. Stat. 32 (1), 12–16 in Unsaturated Geotechnics (GSP 99), Proceedings of the GeoDenver Conference,
(Feb). Denver, Colo., 5–8 August 2000. American Society of Civil Engineers, Reston, Va.,
Perera, Y.Y., Zapata, Z.E., Houston, W.N., Houston, S.L., 2005. Prediction of the soil-water pp. 84–124.
characteristic curve based on grain-size distribution and index properties. Proceed- Zhai, Q., Rahardjo, H., 2012. Determination of soil–water characteristic curve variables.
ings of Geo-Frontiers Austin Texas UAS January 24–26. Comput. Geotech. 42, 37–43.
PWD, 1976. Geology of the Republic of Singapore. Public Works Department, Singapore. Zhai, Q., Rahardjo, H., 2013. Quantification of uncertainty in soil-water characteristic
Rahardjo, H., Ong, B.H., Leong, E.C., 2004. Shear strength of a compacted residual soil from curve associated with fitting parameters. Eng. Geol. 163, 144–152.
consolidated drained and constant water content Triaxial tests. Can. Geotech. J. 41 Zhai, Q., Rahardjo, H., 2014. Variability in soil–water characteristic curve associated with
(3), 421–436 June. fitting parameters. Conference Paper, UNSAT 2014. Sydney, Australia, 02–04-July
Rahardjo, H., Lee, T.T., Leong, E.C., Rezaur, R.B., 2005. Response of a residual soil slope to 2014.
rainfall. Can. Geotech. J. 42 (2), 340–351 April. Zhai, Q., Rahardjo, H., 2015. Estimation of permeability function from the soil–water char-
Rahardjo, H., Ong, T.H., Rezaur, R.B., Leong, E.C., Fredlund, D.G., 2010. Response parame- acteristic curve. Eng. Geol. 199, 148–156.
ters for characterization of infiltration. Environ. Earth Sci. 60 (7), 1369–1380. Zhai, Q., Rahardjo, H., Satyanaga, A., 2015. Effects of variability of unsaturated hydraulic
Rahardjo, H., Melinda, F., Leong, E.C., Rezaur, R.B., 2011. Stiffness of a compacted residual properties on stability of residual soil slopes. Conference Paper, UNSAT 2015. Guilin,
soil. Eng. Geol. 120, 60–67 June. China, 23–26-October 2015.
Rahardjo, H., Satyanaga, A., Leong, E.C., Ng, Y.S., 2012. Variability of Residual Soil Proper- Zhang, F., Fredlund, D.G., Wilson, G.W., 2015. Hydraulic properties for soil that undergo
ties. J. Eng. Geol. 141–142 (124–140) June. volume change as soil suction is increased. Unsaturated Soils: Research & Applica-
Rahimi, A., Rahardjo, H., Leong, E.C., 2010. Effect of hydraulic properties of soil on rainfall- tions China Guilin 23–26 Oct 2015.
induced slope failure. J. Eng. Geol. 114, 135–143.

You might also like