Settlement Prediction of the Ballina Embankment,

Australia, Considering Creep

Shan Huang 1; Jinsong Huang 2; Richard Kelly, Ph.D. 3; Merrick Jones 4;
and A. H. M. Kamruzzaman 5
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Abstract: The prediction of time-dependent deformations of embankments constructed on soft soils is essential for preloading or surcharge
design. The predictions can be obtained by Bayesian back analysis methods progressively based on measurements so that practical decisions
can be made after each monitoring round. However, the effect of creep is typically ignored in previous settlement predictions based on
Bayesian back analysis to avoid the heavy computational costs. This study aims to fill this gap by combining the Bayesian back analysis
with a decoupled consolidation constitutive model, which accounts for creep to perform long-term settlement predictions of the trial embank-
ment with prefabricated vertical drains (PVDs) constructed in Ballina, Australia. The effect of creep on settlement predictions is illustrated by
the comparisons of the cases with and without considering creep. The results show that good settlement predictions could be obtained if creep
is ignored and could be further improved if creep is incorporated when the monitoring settlement data is applied in the Bayesian back analysis.
Ignoring creep could lead to an underestimation of the ultimate consolidation settlement. The swelling index κ and the compression index λ
need to be adjusted to larger values to match the measurements if creep is ignored. Four updating schemes (using surface settlement data only,
using settlement data at all monitoring depths, using pore water pressure data only, and using both settlement and pore water pressure data) are
applied to study the effects of monitoring data on the accuracy of settlement prediction. The results show that the variability introduced by the
noisy pore water pressure data result in fluctuating settlement predictions. Incorporating both settlement and pore water pressure observations
into the Bayesian updating process reduces the variability in the updated soil parameters. DOI: 10.1061/JGGEFK.GTENG-11261. © 2024
American Society of Civil Engineers.
Author keywords: Embankment; Settlement prediction; Soft soil creep; Bayesian back analysis; Prefabricated vertical drains (PVDs).

Introduction performance is worse than expected, additional surcharge or a lon-

ger period of consolidation time can be adopted. The earlier in the
Soft soil is often characterized by high compressibility, low bearing construction period that this information is known with confidence
capacity, and low permeability. It is well known that preloading is leads to the greatest advantage or best risk mitigation.
an effective means for soft soil ground improvement. Typically, a Soft soil deposits formed by long-term natural sedimentation
design is performed where the thickness of surcharge and spacing or reclamation exhibit time-dependent deformations, such as creep,
of vertical drains is determined to achieve a target post construction relaxation, and strain rate effects. The time-dependent behavior
settlement and a time period for construction. The design is vali- (soft soil creep) can accumulate for years after the commencement
dated by monitoring settlement and pore water pressure during the of construction, which may not be acceptable over the design life.
construction period. If the performance is assessed to be better There are two predominant methods for calculating consolidation
than expected, the construction period can be reduced. If the settlement of soft soils exhibiting creep, which are Hypothesis A
(Ladd et al. 1977; Mesri and Godlewski 1977) and Hypothesis B
1
Ph.D. Candidate, Discipline of Civil, Surveying, and Environmental (Bjerrum 1967; Garlanger 1972; Yin and Graham 1996; Yin and
Engineering, College of Engineering, Science, and Environment, Univ. Feng 2017; Feng and Yin 2017; Chen et al. 2021). Hypothesis
of Newcastle, Callaghan, NSW 2308, Australia. ORCID: https://orcid A assumes that creep only occurs after the “primary” consolidation
.org/0000-0003-1825-6365 stage, whereas Hypothesis B assumes that creep commences at the
2
Professor, Discipline of Civil, Surveying, and Environmental beginning of consolidation. Bjerrum (1967) introduced the concep-
Engineering, College of Engineering, Science, and Environment, Univ. of
tual time line model, which conforms to Hypothesis B. Based on
Newcastle, Callaghan, NSW 2308, Australia (corresponding author). Email:
[email protected]
Bjerrum’s time line model, Yin and Graham (1989) developed a
3
Chief Technical Principal and General Manager Technical Excellence, one-dimensional elastic viscous-plastic (1D EVP) model for the
SMEC, Australia and New Zealand Div., Level 6, 480 St Pauls Terrace, general stress-strain-time relationship of clayey soils. Nash and
Fortitude Valley, QLD 4006, Australia. Ryde (2001) employed Yin and Graham’s 1D EVP model to ana-
4
Ph.D. Candidate, Discipline of Civil, Surveying, and Environmental lyze the consolidation settlement of an embankment on soft soils
Engineering, College of Engineering, Science, and Environment, Univ. with vertical drains. The 1D EVP model has been implemented by
of Newcastle, Callaghan, NSW 2308, Australia. the finite difference method (Yin and Graham 1996) and the finite
5
Principal Engineer Geotechnical, Transport for NSW, Sydney, NSW element method (Zhu and Yin 2000). Apart from preloading, instal-
2145, Australia.
lation of PVDs is the most common geotechnical technique
Note. This manuscript was submitted on August 16, 2022; approved on
November 28, 2023; published online on February 24, 2024. Discussion to improve the soft soils by providing the horizonal drainage path
period open until July 24, 2024; separate discussions must be submitted to speed up the dissipation of the pore water pressure. Hansbo
for individual papers. This paper is part of the Journal of Geotechnical (1981) proposed the radial consolidation governing equations based
and Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. on Darcy’s law. Walker (2006) extended the Hansbo’s equation

© ASCE 04024025-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 (Hansbo 1981) by combining the radial and vertical flow to perform to ensure numerical stability, which is time-consuming and not ef-
the consolidation analysis for the multilayered soil. Based on the ficient for the recursive consolidation analysis in MCMC simula-
radial consolidation equation proposed by Hansbo (1981) and tion. The simplified Hypothesis B method is an ideal choice to be
Walker (2006), Huang et al. (2023) incorporated the radial drainage incorporated into Bayesian back analysis algorithms as no PDEs
into the 1D EVP model and applied the extended 1D EVP model to needs to be solved, which can guarantee the numerical stability
perform settlement prediction of a PVD-improved embankment and the computational efficiency. In this study, a visual basic (VB)
based on Bayesian back analysis. To avoid solving complicated par- program, 1DSimp_B, based on a decoupled consolidation constit-
tial differential equations (PDEs) involved in the rigorous 1D EVP utive model based on a general simplified Hypothesis B method is
model, a decoupled simplified Hypothesis B method based on the developed to perform one-dimensional consolidation analysis.
equivalent time concept was proposed by Yin and Feng (2017) Bayesian back analysis is adopted and solved by a differential evo-
based on the analytical solution for pore water pressure proposed lution adaptive metropolis [DREAMðzsÞ ] algorithm based on the
by Walker et al. (2009) using the spectral method. This general multi-chain MCMC method, which was developed by Vrugt et al.
simplified Hypothesis B method has been applied for calculating (2008) to calibrate the input parameters of the prediction model
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

consolidation settlement of a single soil layer (Yin and Feng 2017), based on the measurements. Four updating schemes, using monitor-
double soil layers (Feng and Yin 2017), and multilayers with PVDs ing surface settlement data only (‘sur_sett’), using monitoring set-
(Chen et al. 2021). The simplified Hypothesis B method was veri- tlement data at various depths (‘sett_only’), using monitoring pore
fied by Yin et al. (2022) against the results from the rigorous water pressure data at various depths (‘epp_only’), and using both
Hypothesis B finite element method under various loading condi- monitoring settlement and pore water pressure data at various depths
tions. Chen et al. (2022) extended the simplified Hypothesis B (‘sett_epp’), are applied to study the effect of the monitoring data on
method into two-dimensional (2D) conditions by incorporating the settlement prediction, which can provide the geotechnical de-
the stress diffusion and buoyancy effects. signers with practical references for long-term behavior prediction
In addition, various uncertainties, such as the uncertainties in of embankments constructed on soft soils.
soil properties, contribute to the discrepancy between the model
predictions and monitoring performance. Therefore, calibrating
model parameters based on monitoring data in situ is essential Bayesian Back Analysis
to improve the fidelity of long-term embankment prediction. Back
analysis methods, which include traditional manual calibrations, Basic Theory
maximum likelihood method (Gong et al. 2014), least squares
The uncertainties in the input soil properties, which are obtained
method (Gioda and Maier 1980), maximum a posteriori (Zhao et al.
from lab or in situ tests, should be considered and incorporated into
2021), and Bayesian back analysis method (Zhang et al. 2010b;
the settlement and pore water pressure predictions. The soil param-
Kelly and Huang 2015; Zheng et al. 2018; Huang et al. 2021,
eters can be calibrated by Bayesian back analysis by considering
2023), are widely applied to calibrate soil parameters based on
them as random variables x. When the model prediction FðxÞ is
measurements. Among these methods, the traditional manual cal-
available, it will then have
ibration, such as trial and error approach, is employed based on
the results from Class A predictions (Chai et al. 2018). However, d ¼ FðxÞ þ e ð1Þ
the accuracy of the manual calibration cannot be ensured due to the
high nonlinearity of the elastic viscoplastic characteristics of soft where e = the difference between the actual performance d and the
soils and the essential correlations among the soil parameters. model prediction FðxÞ.
The Bayesian updating approach treats the soil parameters as Based on the Bayesian theorem, posterior information about soil
random variables described in terms of the probability density func- and model parameters is inferred by updating the prior probability
tions. The new monitoring information can be incorporated into the distribution with measurements. This process can be expressed by
Bayesian back analysis progressively, which makes the Bayesian
PðxjdÞ ¼ cLðxjdÞPðxÞ ð2Þ
updating approach a rational and robust means to accurately calibrate
the soil parameters and predict the long-term behavior. There are where c = normalizing constant; LðxjdÞ is the likelihood function;
many successful applications of the Bayesian approach in geotech- PðxÞ is the prior distribution reflecting the knowledge about x be-
nical engineering (Kelly and Huang 2015; Hsein Juang et al. 2013; fore obtaining the field-monitored data; and the prior information is
Zhang et al. 2010a; Miranda et al. 2009; Honjo et al. 1994). How- usually obtained from site investigations, engineering judgment,
ever, due to the high dimension of the posterior distribution of the and experience. The posterior information PðxjdÞ is obtained by
updated parameters and the nonlinearity of the numerical model, updating x while incorporating both the prior information and
the analytical form of the posterior distribution cannot be computed. the field-monitored behaviors. Considering high dimension of
The Markov chain Monte Carlo (MCMC) method was introduced by the posterior distribution in this study, the posterior probability den-
Metropolis et al. (1953). Hastings (1970) further developed the sity function (PDF) of x, PðxjdÞ, will be solved by using the sam-
method by introducing the Metropolis-Hastings algorithm, which pling method. An MCMC algorithm, which is called DREAMðzsÞ
is designed to sample from probability distributions that are difficult and developed by Ter Braak et al. (2008) and Vrugt et al. (2009), is
to directly sample from. This algorithm can be applied to estimate the then adopted.
average properties of complex systems, and for posterior inference in
a Bayesian framework (Gelfand and Smith 1990).
Although previous work has dealt with embankment settlement Likelihood Function
predictions (Zheng et al. 2018; Chai et al. 2018; Le et al. 2018; The likelihood function LðxjdÞ presents the difference between the
Huang et al. 2021), there are almost no studies incorporating measurements d, which includes all the monitoring data at various
creep/viscoplastic behavior into the framework of Bayesian back time points, and the model predictions FðxÞ. The error e in Eq. (1)
analysis with MCMC simulation. The main reason is that the rig- is assumed to follow a zero-mean Gaussian distribution and can be
orous elastic viscoplastic constitutive model needs to be solved by modeled explicitly through the PDF. The likelihood function can be
numerical methods, and sometimes small time steps are required presented by

© ASCE 04024025-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 1 The calculation of Sfj , Screep;f , and Screep;d are given in Yin et al.
LðxjdÞ ¼ N d =2
ð2πÞ det ðRÞ1=2 (2022) and can be presented by


1 8
× exp − ½d − FðxÞT R−1 ½d − FðxÞ ð3Þ > κ σf
2 >
> Hj V ln σ ðfor overconsolidation stateÞ
< 0
Sfj ¼
where N d = number of points for a specific type of measurement >
> κ σp λ σf
>
: Hj V ln σ þ H j V ln σ ðfor normal consolidation stateÞ
(settlement or pore water pressure in this study); R represents the 0 p
variance matrix corresponding to the coefficient of variation (COV)
ð6Þ
of the measurement error and the monitoring data; and det repre-
sents the determinant.
2  
3
If multiple types of measurement are applied, the measurement
φ t0 þte
errors for various measurement types are assumed to be indepen- 6 V ln t0 7
Screep;fj ðtÞ ¼ Hj 4 f 5 ðfor t ≥ t0 Þ ð7Þ
− Δεvp
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

dent. Therefore, the likelihood function of all measurements can be 1þ φ 

obtained by VΔεl ln
t0 þte
t0

Y
n
LðxjdÞ ¼ Li ðxjdi Þ ð4Þ 2  
3
i φ t0 þte
6 V ln tEOP 7
where i = number of measurement data types. For example, i ¼ 1 if Screep;dj ðtÞ ¼ H j 4 φ f 5
− Δεvp ðfor t ≥ tEOP Þ
1þ 
only the monitoring settlement or pore water pressure data is used, VΔεl ln
t0 þte
tEOP
and i ¼ 2 if both the monitoring settlement and pore water pressure
data are applied; Li ðxjdi Þ represents the likelihood function ob- ð8Þ
tained by Eq. (3) for ith monitoring data type; di represents the
measurements of the ith data type and varies with the updating where λ and κ = compression index and swelling index respec-
scheme. For example, if the updating scheme ‘sur_sett’ is applied, tively; H j = thickness of the layer j; σ0 and σf = initial and final
di then represents the monitoring settlement data at surface only. effective stress respectively; σp = preconsolidation pressure; tEOP =
elapsed time t at the end of primary consolidation; t0 = reference
time as the starting point of creep calculation and t0 ¼ 1 day in this
One-Dimensional Consolidation Analysis for study; te = equivalent time and more details can be found in Yin
Multilayered Soils with PVD and Graham (1989) and Yin et al. (2002); φ=V = creep coefficient;
Δεl = creep strain limit; Δεvp f = difference between the targeted
In this study, the model prediction FðxÞ in Eq. (1) is obtained by the strain εf and the reference strain εrf on the reference time line; and
one-dimensional consolidation analysis based on the decoupled 8
simplified Hypothesis B method proposed by Yin and Feng <0 for normal consolidation state
(2017) and Yin et al. (2022). The basic assumptions are made as Δεf ¼ λ − κ σp
vp
ð9Þ
: ln for over consolidation state
follows: V σf
1. Soil is assumed to be isotropic and homogeneous within each
soil layer; The average degree of consolidation for the soil layer j, U j , in
2. Soil deformation is solely due to the volume change and only Eq. (5) can be determined by
occurs in the vertical direction;
3. Soil is fully saturated and no free water on the soil surface; and ūj ðtÞ
U j ðtÞ ¼ 1 − ð10Þ
4. Darcy’s law is valid, and the effective stress theorem is Δσj ðtÞ
applicable.
As this method is based on Hypothesis B, creep occurs in the where ūj ðtÞ = average pore water pressure; and Δσj ðtÞ = load
whole consolidation period, both within and after the primary con- increment.
solidation. According to Yin et al. (2022), for all t > tEOP;lab , According to Yin et al. (2022), if there were more than one load-
the main equation of the simplified Hypothesis B method for the ing stage, the total Sprimary should include the settlements produced
calculation of the total consolidation settlement, Stotal , can be ex- by all loading stages, which can be presented by
pressed as
X
n
Stotal ¼ Sprimary þ Screep Sprimary ¼ ðU j;i Sfj;i Þ ð11Þ
i¼1
X
j¼n X
j¼n
¼ U j Sfj þ ½αU βj Screep;fj þ ðð1 − αU βj ÞScreep;dj Þ ð5Þ
where i = ith loading stage; n = total number of loading stages;
j¼1 j¼1
U j;i = degree of consolidation of layer j at loading stage i, which
where U j and Sfj = average degree of consolidation and the final can be calculated by Eq. (10); Sfj;i is the ultimate primary settle-
primary consolidation settlement of the layer j respectively; α and ment of layer j at loading stage i and can be determined by Eq. (6).
β are constants to reasonably consider the creep compression The creep for layer j, Screep;j , should be calculated from the cur-
coupled with consolidation, and α ¼ 0.8, β ¼ 0.3, suggested by rent stress-strain state under the total loading increment. The degree
Yin et al. (2022), are adopted in this study; Screep;f = creep settle- of consolidation for layer j at loading stage i, U multi;i;j ðtÞ, should
ment under the final effective vertical stress without excess pore then be applied in Eq. (5) to replace U j to calculate Screep;j by
water pressure coupling; and Screep;d = “delayed” compression set- Pi
ūj;k ðtÞ
tlement and only occurs when the time t ≥ tEOP;field , where tEOP;field U multi;i;j ðtÞ ¼ 1 − k¼1 ð12Þ
is the time at Uj ¼ 98%. σj;i ðtÞ − σi0

© ASCE 04024025-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 Load Table 2. Time intervals, vertical pressure changes, and loading sequence
Permeable types
rd rs Vertical
0
Layer 1 Crust layer Starting Ending pressure
1.5 Layer 2 Loading time time changed
2
sequence (day) (day) (kPa) Loading type
Layer 3
Depth (m)

4 1 0 1 12.6 Linearly varying with time

5 Layer 4
Estuarine clay 2 1 12 0 Uniform load
Layer 5
6 layer 3 12 20 21.6 Linearly varying with time
4 20 42 0 Uniform load
Smear
zone
5 42 55 61.5 Linearly varying with time
6 55 974 0 Uniform load
Layer 6
10.5
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Vertical drain Permeable

where σ̄ = total stress at depth z; H = total depth of the soil pro-

file; mv = volume compressibility; and for the layer j, mvj , can be
Fig. 1. Soil deposit and vertical drain with smear drain. calculated by mvj ¼ Sfj =ðHj Δσ 0 Þ; kv and kh = vertical and hori-
zontal permeability respectively; re = drain influence radius; μ is
the parameter corresponding to the drain and smear zone [more
details can be found in Walker et al. (2009) and Walker and
where σ j;i ðtÞ and σi0 = total loading increment and the initial ver-
tical stress of layer j respectively; and ūj;k ðtÞ = average excess pore Indraratna (2006)]; and mv =m̄v , kv =k̄v , and η=η̄ = normalized vol-
water pressure of j-layer at loading stage k. ume coefficient mv , permeability kv, and η to simplify the equa-
The average pore water pressure, ū, in Eqs. (10) and (12) can be tions and remove the influence of units, where m̄v , k̄, and η̄
estimated by the governing equation for consolidation incorporat- are considered as the mv , kv , and η of the first soil layer in this
ing radial and vertical flow derived by Walker et al. (2009). It can study.
be presented by Eqs. (5) and (13) and can be considered as the decoupled
governing equations for solving the 1D consolidation problem
   of the soft soils improved with PVDs under any loading condi-
mv ∂ ū η ∂ kv ∂ ū m ∂ σ̄
¼ − dT h ū − dT v þ v ð13Þ tions. The analytical solution of Eq. (13) was derived by Walker
m̄v ∂t η̄ ∂Z k̄v ∂Z m̄v ∂t
et al. (2009) based on the spectral method and can be presented by

z c̄v 2η̄ ūðZ; tÞ ¼ u0 ΦvEðΓvÞ−1 θ ð14Þ

Z¼ ; dT v ¼ ; dT h ¼ ;
H H2 γ w m̄v
k̄v kh
c̄v ¼ ; η¼ where u0 = initial pore water pressure; E is a diagonal matrix, which
γ w m̄v r2e μ depends on the loading conditions; v represents the eigenvector

Table 1. Basic model parameters and prior distribution for the random parameters in Bayesian back analysis of Ballina Embankment
Crust Estuarine clay
Basic model parameters Layer 1 2 3 4 5 6
Layer thickness H j (m) 1.5 0.5 2 1 1 4.5
Overconsolidation ratio OCR 4.8 3.98 2.33 2.32 2.07 1.84
Unit weight of soil layer γ (kN=m3 ) 18.753 15.965 14.700 13.700 13.971 13.673
Initial void ratio e0 0.81 1.63 2.31 3.16 2.89 3.19
Compression index λ 0.043 0.348 0.587 1.304 0.913 1.061
Creep index φ 0.002 0.006 0.006 0.007 0.007 0.007
Reference time t0 (day) 1 1 1 1 1 1
Ratio of vertical to horizontal permeability kv =kh 1 1 1 1 1 1
Drain spacing Sp (m) 1.2 1.2 1.2 1.2 1.2 1.2
Drain pattern — Square Square Square Square Square Square
Drain radius rd (m) 0.025 0.025 0.025 0.025 0.025 0.025
Ratio of smear to drain radius rs =rd 5 5 5 5 5 5
Ratio of undisturbed horizontal permeability kh =ks 5 5 5 5 5 5
to smear zone permeability
The prior distribution for the random variables in Bayesian back analysis
Swelling index κ 0.013 0.028 0.061 0.087 0.087 0.107
Vertical permeability kv (m=day) 2.30 × 10−3 2.57 × 10−4 6.53 × 10−4 1.04 × 10−4 8.93 × 10−5 9.6 × 10−5
Ratio of the compression index λ to the R1 Uniform distribution—U[5,10]
recompression index κ (λ=κ)
Ratio of the creep index φ to the compression R2 Uniform distribution—U[0.02,0.08]
index λ (φ=λ)
Coefficient of variation COV (κ) 0.3 0.3 0.3 0.3 0.3 0.3
COV (kv =γ w ) 3.5 3.5 3.5 3.5 3.5 3.5

© ASCE 04024025-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 1.6 150
0m 1.5m 4.5m 7.5m
2m 6m 10m

120
1.2

Pore water pressure (kPa)

90
Settlement (m)

0.8

60

0.4
30
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

Fig. 2. Field-monitored settlement and pore pressure at various depths.

1.8
1.8

M0 prior 12d 42d M0 prior 12d 42d

76d 117d 216d 292d 76d 117d 216d 292d
1.5 1.5
384d 496d 797d 974d 384d 496d 797d 974d

1.2 1.2
Settlement (m)
Settlement (m)

0.9 0.9

0.6 0.6

0.3
0.3

0.0
0 200 400 600 800 1000 0.0
0 200 400 600 800 1000
Time (day)
(a) (b) Time (day)

1.6 1.8
M0 prior 12d 42d
76d 117d 216d 292d M0 prior 12d 42d
384d 496d 797d 974d 76d 117d 216d 292d
1.5 384d 496d 797d 974d
1.2
1.2
Settlement (m)

Settlement (m)

0.8 0.9

0.6
0.4

0.3

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(c) Time (day) (d) Time (day)

Fig. 3. The predicted settlement at surface based on the four updating schemes with creep ignored: (a) sur_sett; (b) sett_only; (c) epp_only; and
(d) sett_epp.

© ASCE 04024025-5 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 2.0 1.8
M0 prior 12d 42d M0 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d 1.5 384d 496d 797d 974d
1.6

1.2

Settlement (m)
1.2
Settlement (m)

0.9

0.8
0.6

0.4
0.3
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000

(a) Time (day) (b) Time (day)

2.5 2.0
M0 prior 12d 42d M0 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d 384d 496d 797d 974d
2.0
1.5

1.5
Settlement (m)

Settlement (m)
1.0
1.0

0.5
0.5

0.0
0 200 400 600 800 1000 0.0
0 200 400 600 800 1000
Time (day)
(c) (d) Time (day)

Fig. 4. The predicted settlement at surface based on the four different updating schemes with creep considered: (a) sur_sett; (b) sett_only;
(c) epp_only; and (d) sett_epp.

associated with each eigenvalue of the matrix Γ −1 Ψ ; Γ and Ψ Case Study

involve the governing partial differential equations and vary with
time; the matrix θ is determined by the combined distribution of Model Description
mv and ∂ σ̄=∂t; and Φ is a matrix corresponding to the drainage
conditions, for soil layer j The Ballina embankment is located in the National Soft Soil Field
Testing Facility at Ballina (NSW), Australia. It was constructed in
2013 to study the long-term consolidation settlement of soft soils in
ϕj ðZÞ ¼ sinðM j ZÞ ð15Þ
coastal areas. Based on Pineda et al. (2016) and Kelly et al. (2018),
the deposit of the Ballina embankment is comprised of approxi-
where mately 1.5 m of clay crust, followed by a 9-m-deep layer of estua-
rine clay and 4.5-m transition layer, and underlain by clayey sand
8 to medium sand layers down to a depth of 19.3 m. Only the crust
< jπ for pervious top pervious bottom
clay and the estuarine clay are considered in this study due to their
Mj ¼ π ð16Þ
: ð2j − 1Þ for pervious top impervious bottom high plasticity and compressibility (Yang and Carter 2018). Based
2 on the soil parameter profile reported in Kelly et al. (2018) and to
represent the variation of soil properties of the crust clay and
An open source spreadsheet called ‘SPECCON’ was devel- the estuarine clay layer, six sublayers are adopted based on the in-
oped by Walker et al. (2009) for solving Eq. (13). In this formation from geotechnical investigations as depicted in Fig. 1.
study, based on the Visual Basic Applications (VBA) code in The drainage boundary conditions can also be found in Fig. 1. The
‘SPECCON’ and (Yin et al. 2022), an in-house 1D consolidation thickness for each soil layer, the prior mean and the COVof the soil
analysis program in vb.net, which is called ‘1D_SimpB,’ was parameters of Ballina embankment are summarized in Table 1.
developed. The prior mean and COV for the soil parameters are obtained from

© ASCE 04024025-6 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 1.6 1.5
M1.5 prior 12d 42d
M1.5 prior 12d 42d
76d 117d 216d 292d
76d 117d 216d 292d
384d 496d 797d 974d
384d 496d 797d 974d
1.2
1.2

0.9
Settlement (m)

Settlement (m)
0.8

0.6

0.4
0.3
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000

(a) Time (day) (b) Time (day)

1.6 1.5
M1.5 prior 12d 42d
M1.5 prior 12d 42d 76d 117d 216d 292d
76d 117d 216d 292d 384d 496d 797d 974d
384d 496d 797d 974d 1.2
1.2

0.9
Settlement (m)
Settlement (m)

0.8

0.6

0.4
0.3

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(c) Time (day) (d) Time (day)

Fig. 5. The predicted settlement at 1.5 m based on the four different updating schemes with creep ignored: (a) sur_sett; (b) sett_only; (c) epp_only;
and (d) sett_epp.

the lab and in situ tests which can be found in Pineda et al. (2016) soil layer are considered as random variables in Bayesian back
and Kelly et al. (2018). The OCR for each soil layer is based on the analysis and are assumed to be characterized statistically by log-
results from the CRS tests and the strain rate effects are not con- normal distributions to remain positive. Considering the variables
sidered in this case study. As reported in Kelly et al. (2018), the are sampled randomly in the MCMC simulation, the infinite pre-
PVDs of the Ballina embankment are installed in a square pattern consolidation pressure might occur when λ is roughly equal to κ,
and the detailed parameters of the PVDs are provided in Table 1. and the sampled λ might be smaller than κ, which is unrealistic. In
The estimation of the radius of the smear zone and the undisturbed addition, according to Mesri and Godlewski (1977), the ratio of the
zone are based on (Walker et al. 2009). As indicated in Kelly et al. creep index to the compression index, φ=λ, is constant. Therefore,
(2017), the difference between the horizontal and vertical per- two factors, R1 and R2 , are set to represent the ratio λ to κ, and the
meability (kh and kv ) can be ignored for the Ballina embankment. ratio of μ to λ respectively. The factors R1 and R2 are also deemed
The ratio of kv and kh is thus considered as 1. Based on Basu et al. to be random variables in Bayesian back analysis, and are assumed
(2010) and Kelly et al. (2018), the ratio of kh and ks , where ks is the to be uniformly distributed with a range of 5 to 10 and 0.02 to 0.08,
permeability of smear zone, is adopted as 5. More details of the site respectively (Pineda et al. 2016; Kelly et al. 2018). φ is considered
conditions, properties of soils, construction process, and parame- as 0 when creep is ignored in the consolidation analysis. Therefore,
ters of Ballina embankment can be found in Pineda et al. (2016) there are 14 random variables in total when creep is considered, and
and Kelly et al. (2017, 2018). The starting and ending time, vertical 13 random variables when creep is ignored. It is assumed that there
pressure changes and loading sequence types are summarized in are no correlations between the random variables in the prior
Table 2. The monitoring settlement at 0 (surface), 1.5, 4.5, and distributions.
7.5 m and pore water pressure at 2, 6, and 10 m in situ are shown To ensure the robust posterior distribution of the soil properties,
in Fig. 2. three Markov chains and 80,000 simulations are adopted for each
The swelling index (κ) and the ratio of the vertical permeability MCMC simulation. The first 20,000 generations are considered
of the undisturbed zone to the unit weight of water (kv =γ w ) for each as the “burn-in” period to ensure the stability and independence

© ASCE 04024025-7 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 1.8 1.5
M1.5 prior 12d 42d M1.5 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d 384d 496d 797d 974d
1.5
1.2

1.2
0.9

Settlement (m)
Settlement (m)

0.9

0.6
0.6

0.3
0.3
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

2.5 1.8
M1.5 prior 12d 42d M1.5 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d 1.5 384d 496d 797d 974d
2.0

1.2
1.5
Settlement (m)

Settlement (m)
0.9

1.0
0.6

0.5
0.3

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(c) Time (day) (d) Time (day)

Fig. 6. The predicted settlement at 1.5 m based on the four different updating schemes with creep considered: (a) sur_sett; (b) sett_only; (c) epp_only;
and (d) sett_epp.

of the posterior distribution. The posterior distribution of the ran- Results

dom variables is determined by four different schemes. The first
updating scheme only applies the settlement monitored at the Settlement Prediction
surface and is represented by ‘sur_sett.’ The second updating In this section, the settlements at various depths are predicted by
scheme, which is denoted by ‘sett_only’ only employs the moni- the four updating schemes. The settlement versus time curves at
toring settlements at depths 0, 1.5, 4.5, and 7.5 m. The third one is the surface (0 m), 1.5, 4.5, and 7.5 m using the monitoring data
denoted by ‘epp_only’ as only the pore water pressure data moni- prior to day 12, 42, : : : , and 974 with creep ignored and with
tored at depths 2, 6, and 10 m are used. Both settlement measure- creep considered are presented in Figs. 3–10. The monitoring set-
ments at depths 0, 1.5, 4.5, and 7.5 m, and monitored pore water tlements (denoted by asterisks with ‘Mx’ which represents that the
pressure at 2, 6, and 10 m are applied in the fourth updating monitoring data at depth ‘xm,’ where x represents 0, 1.5, 4.5, and
scheme, which is represented by ‘sett_epp.’ In addition, the stan- 7.5 for settlement prediction and 2, 6, and 10 for pore water pres-
dard deviation of the measurement errors for pore pressure and sure prediction) and the settlement prediction using prior (denoted
settlement are initially assumed to be 0.02 m and 1 kPa, respec- by the dashed line with ‘prior’) are also presented for comparison.
tively (Kelly and Huang 2015). The measurement errors for settle- It should be noted that the legend “nd” represents the predicted
ments and pore water pressure are assumed to be uncorrelated settlement based on the monitoring data from 0 to nth day, where
for mathematical simplification. It should be noted that there is n represents 12, 42, 76, : : : , and 974.
underlying physical correlation between the measured magnitudes As presented in Figs. 3–10, incorporating creep would improve
of settlement and pore water pressure, i.e., if the pore water pres- the accuracy of the settlement prediction when the updating scheme
sure is low, the settlement should then be high. However, this cor- ‘sett_only’ is applied. For example, as shown in Figs. 3–9(b) and in
relation does not apply to the measurement errors associated with Figs. 4–10(b), the settlement predictions with creep considered at
settlements and pore water pressure as they are measured by inde- all monitoring depths match the measurements better than the ones
pendent devices. when creep is ignored. Additionally, when creep is ignored, the

© ASCE 04024025-8 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 1.2 1.0
M4.5 prior 12d 42d
M4.5 prior 12d 42d
76d 117d 216d 292d
76d 117d 216d 292d
384d 496d 797d 974d
0.8 384d 496d 797d 974d
0.9

Pore water pressure (kPa)

Settlement (m)

0.6

0.6

0.4

0.3
0.2
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

1.0 1.0
M4.5 prior 12d 42d
M4.5 prior 12d 42d
76d 117d 216d 292d
76d 117d 216d 292d
384d 496d 797d 974d
0.8 384d 496d 797d 974d 0.8

0.6
Settlement (m)

Settlement (m) 0.6

0.4 0.4

0.2 0.2

0.0
0.0
0 200 400 600 800 1000
0 200 400 600 800 1000
Time (day)
(c) (d) Time (day)

Fig. 7. The predicted settlement at 4.5 m based on four different updating schemes with creep ignored: (a) sur_sett; (b) sett_only; (c) epp_only; and
(d) sett_epp.

potential increase of settlement cannot be predicted well in the fluctuation in the predicted settlement curves can be found in
long-term. This would result in underestimating the final settle- Fig. 10(b) where creep is considered.
ment. However, this long-term increasing trend can be well pre- For the settlement prediction with creep considered, the settle-
dicted when creep is considered. ments at the surface and 1.5 m can be well predicted by using
As shown in Fig. 4(a), the settlement at the surface can be well 76 days of monitoring data when updating scheme ‘sett_epp’ is
predicted by the updating scheme ‘sur_sett’ based on 496 days applied. However, slight overestimation can be observed if 117,
of monitoring settlement at surface when creep is considered. 216, and 292 days of monitoring data are used, and good predic-
However, the deviation between the settlement predicted by tions can be obtained again if more than 384 days of monitoring
‘sur_sett’ and the actual measurements becomes more significant data are applied. The possible reason why the accuracy of the set-
when it goes deeper as shown in Fig. 6(a) at depth 1.5 m, Fig. 8(a) tlement prediction does not increase with the increasing monitoring
at depth 4.5 m, and Fig. 10(a) at depth 7.5 m. The reason data for this case is that the variability caused by the incorporation
why using ‘sur_sett’ cannot give good settlement prediction of pore water pressure data would result in fluctuating predicted
at deeper locations is that the monitoring surface data are not settlement curves. It would then need more monitoring data to ob-
sufficient to get accurate settlement predictions at all monitoring tain stable predicted settlement curves. The same reason could also
depths. be used to explain why the settlement at 7.5 m cannot be well pre-
As shown in Figs. 3(b)–10(b), the settlement predicted by the dicted by ‘sett_epp’ whether creep is considered or not as shown in
updating scheme ‘sett_only’ where only monitoring settlements at Figs. 9(d)–10(d). Additionally, comparing Figs. 3(b)–10(b) with
various depth are used, agrees well with the measurements at Figs. 3(d)–10(d), the updating scheme ‘sett_only’ performs better
depths 0, 1.5, and 4.5 m by using only 76 days of monitoring data. than ‘sett_epp’ for settlement prediction. The monitoring pore
The settlement at 7.5 m can be well predicted, provided that water pressure data is subject to greater uncertainty than the mon-
216 days of monitoring settlement are applied, whether creep is itoring settlement data due to potential issues such as the clogging
considered or not as shown in Figs. 9(b) and 10(b). However, less of the fine particles, bending of vertical drains, groundwater table

© ASCE 04024025-9 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 1.5 1.0
M4.5 prior 12d 42d M4.5 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d 384d 496d 797d 974d
1.2 0.8

0.9 0.6
Settlement (m)

Settlement (m)
0.6 0.4

0.3 0.2
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000

(a) Time (day) (b) Time (day)

1.6 1.0
M4.5 prior 12d 42d M4.5 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d
384d 496d 797d 974d
0.8
1.2
Settlement (m)

0.6
Settlement (m)

0.8

0.4

0.4
0.2

0.0
0 200 400 600 800 1000 0.0
0 200 400 600 800 1000
Time (day)
(c) (d) Time (day)

Fig. 8. The predicted settlement at 4.5 m based on four different updating schemes with creep considered: (a) sur_sett; (b) sett_only; (c) epp_only; and
(d) sett_epp.

changes and sensor movements. These factors are not accounted for scheme ‘epp_only’ whether creep is considered or not. It can be
in the simplified Hypothesis B method, which may explain why concluded that only incorporating the pore water pressure data into
incorporating pore water pressure data does not improve the accu- the Bayesian updating process is not sufficient to obtain accurate
racy of the settlement predictions. settlement predictions.
When ‘sett_epp’ is applied, Figs. 3(d) and 5(d) show better set-
tlement predictions when creep is ignored, compared to Figs. 4(d) Pore Water Pressure Prediction
and 6(d) where creep is considered. It indicates that the settlement In this section, the pore water pressure predictions based on the
prediction is more sensitive to the variability resulting from the four different updating schemes are summarized. The pore water
incorporation of pore water pressure when creep is considered. pressure predictions at 10 m using the monitoring data prior to
The settlement predictions at the surface and 1.5 m need more 12, 42, : : : , and 974 days based on the 4 updating schemes are
monitoring data to reach the convergence to the measurements presented in Fig. 11 with creep ignored and in Fig. 12 with creep
when creep is considered than the ones when creep is ignored. considered respectively. The comparison of the pore water pressure
However, this influence of the variability introduced by the pore predictions at depths 2 and 6 m based on the four updating schemes
water pressure data on settlement prediction could be reduced using the monitoring data prior to the 496th day are presented
with the increase of depth when creep is considered. As shown in Fig. 13(a) with creep ignored and in Fig. 13(b) with creep
in Figs. 7(d) and 9(d) where creep is ignored and in Fig. 8(d) considered. In Fig. 13, ‘sur_sett2’ and ‘sett_only6’ represent the
and in Fig. 10(d) where creep is considered, the settlement predic- pore water pressure prediction at 2 and 6 m obtained by ‘sur_sett’
tions at 4.5 and 7.5 m considering creep perform better than the where the surface settlement data are applied, and ‘sett_only’ where
ones when creep is ignored. the monitored settlement at various depths are used respectively.
As presented in Figs. 3(c)–10(c), the settlements at all mon- The meaning can be extended to ‘sur_sett10,’ ‘sett_only2,’ : : : ,
itoring depths cannot be accurately predicted by the updating ‘sett_epp10.’

© ASCE 04024025-10 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 0.6 0.5
M7.5 prior 12d 42d
M7.5 prior 12d 42d 76d 117d 216d 292d
76d 117d 216d 292d 384d 496d 797d 974d
0.5
384d 496d 797d 974d 0.4

0.4
0.3

Settlement (m)
Settlement (m)

0.3

0.2
0.2

0.1
0.1
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

0.5 0.5
M7.5 prior 12d 42d M7.5 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d 384d 496d 797d 974d
0.4 0.4

0.3
Settlement (m)

0.3
Settlement (m)

0.2 0.2

0.1 0.1

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(c) Time (day) (d) Time (day)

Fig. 9. The predicted settlement at 7.5 m based on four different updating schemes with creep ignored: (a) sur_sett; (b) sett_only; (c) epp_only; and
(d) sett_epp.

As shown in Fig. 11 where creep is ignored, the predicted pore 974th day and it still shows a significant pore water pressure of
water pressure by the updating schemes ‘sur_sett’ in Fig. 11(a) and over 110 kPa, while less than 105 kPa on the 974th day is observed
‘sett_only’ in Fig. 11(b) deviate from the actual measurements the in Figs. 12(a and b). It can be concluded that the dissipation rate of
most, followed by the ones updated by ‘sett_epp’ in Fig. 11(d). It the pore water pressure is overestimated if only the monitoring
indicates that the pore water pressure cannot be well predicted by settlement data is incorporated. In comparison to Fig. 11 where
only incorporating the monitoring settlement, but the prediction can creep is ignored, the pore water pressure predictions using the
be improved by incorporating the monitoring pore water pressure. four updating schemes in Fig. 12 are greatly improved. The pore
The prediction for the dissipation rate of the pore water pressure water pressure at 10 m can be well predicted by ‘epp_only’ as
can also be slightly improved by incorporating the monitoring shown in Fig. 12(c) by using 496 days of monitoring pore water
data in longer time intervals. For example, for the updating scheme pressure data. The pore water prediction after 500th day match
‘sett_only’ as shown in Fig. 11(b) where only the monitoring set- well with those predicted by ‘sett_epp’ using prior to 496 days
tlement are used, the pore water pressure has fully dissipated on of monitoring settlement and pore water pressure data. In com-
around the 200th day when 76 days of monitoring data are applied, parison with Figs. 11(a and b), the pore water pressure predictions
and it extends to the 800th day when 797 days of monitoring data in Figs. 12(a and b) are much closer to the measurements. This
are used. The pore water pressure predicted by the updating scheme indicates that the incorporation of creep can significantly improve
‘epp_only’ is not stable, and a good match between the prediction the pore water pressure prediction.
and the actual measurement cannot be observed until 974 days of As shown in Fig. 13, the pore water pressure at depths 2 and 6 m
monitoring data are applied. predicted by ‘epp_only’ is the closest to the measurements, while
When creep is considered, the dissipation rate of the pore water the greatest deviation between the predictions and the measure-
pressure is still overestimated by ‘sur_sett’ and ‘sett_only’ as ments are observed when the updating schemes ‘sur_sett’ and
shown in Figs. 12(a and b) respectively. The dissipation of the mon- ‘sett_only’ are applied. Comparing Fig. 13(a) where creep is
itoring pore water pressure has not been completed even on the ignored with Fig. 13(b) where creep is considered, the incorporation

© ASCE 04024025-11 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 2.0 0.6
M7.5 prior 12d 42d
M7.5 prior 12d 42d
76d 117d 216d 292d
76d 117d 216d 292d
384d 496d 797d 974d 0.5 384d 496d 797d 974d
1.6

0.4
1.2
Settlement (m)

Settlement (m)
0.3

0.8
0.2

0.4
0.1
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

0.9 0.6
M7.5 prior 12d 42d M7.5 prior 12d 42d
76d 117d 216d 292d 76d 117d 216d 292d
384d 496d 797d 974d 0.5 384d 496d 797d 974d

0.6 0.4

Settlement (m)
Settlement (m)

0.3

0.3 0.2

0.1

0.0 0.0
200 400 600 800 1000 0 200 400 600 800 1000

(c) Time (day) (d) Time (day)

Fig. 10. The predicted settlement at 7.5 m based on four different updating schemes with creep considered: (a) sur_sett; (b) sett_only; (c) epp_only;
and (d) sett_epp.

of creep can improve the pore water pressure predictions obtained be noted that ‘/’ in Tables 3 and 4 represents ‘not applicable.’ The
by the updating schemes except ‘sur_sett.’ The dramatic increase of COV of the posteriors are presented in Table 5 with creep ignored
the pore water pressure at 2 m caused by the last loading is better and in Table 6 with creep considered.
captured in Fig. 13(b) where creep is considered. The pore water It can be seen from Tables 3 and 4 that the updated λ and φ
pressure predictions at 6 m shown in Fig. 13(b) are much closer values are within the range obtained from the test results when
to the measurements than the ones in Fig. 13(a). the monitoring settlement data at various depths (‘sett_only’) are
The pore water pressure is difficult to predict well. As shown in applied. When only the monitoring pore water pressure data at vari-
Fig. 11, the monitoring pore water pressure at 2 m increases around ous depths are applied, the updated λ and φ values are unrealistic.
the 300th day and after the 500th day, which might be caused by Also, when the updating scheme ‘sett_epp’ where both the mon-
seasonal groundwater table changes as reported by Pineda et al. itoring settlement and pore water pressure are applied, the updated
(2016) or caused by sensor movements, the clogging of the fine λ and φ values are too small at deeper layers (layers four and five
particle, and the bending of vertical drains. Identifying the actual from depths 4 to 6 m), falling outside the range. This indicates that
reasons for the discrepancies between the prediction and the mon- incorporating pore water pressure measurements would lead to
itoring pore water pressure data is difficult and requires more spe- unrealistic posterior set of soil parameters. This can be explained
cific field information. by the significant fluctuations in the monitoring pore water pressure
measurements, which might be caused by the changes in the
Posterior Parameters groundwater table caused by the seasonal rainfall, sensor move-
In this section, the posterior distribution of the soil parameters ob- ments and the clogging of fine particles, etc. In addition, the reason
tained by the four updating schemes are summarized in Table 3 for the superior performance of the updating scheme ‘sett_only’
with creep ignored and in Table 4 with creep considered. For com- over ‘sett_epp’ on updated soil parameters is possibly because the
parison, the prior, the lower and upper bound for λ and φ from the uncoupled consolidation analysis (the simplified Hypothesis B ap-
lab tests reported in Kelly et al. (2018), and the corresponding proach) where consolidation settlement would not affect the excess
parameters in Buttling et al. (2018) are also presented. It should pore water pressure dissipation is applied in this study.

© ASCE 04024025-12 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 150 150
M10 prior 12d 42d
M10 prior 12d 42d
76d 117d 216d 292d
140 140 76d 117d 216d 292d
384d 496d 797d 974d
384d 496d 797d 974d
Pore water pressure (kPa)

Pore water pressure (kPa)

130 130

120 120

110 110

100 100
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

90 90
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

160 150
M10 prior 12d 42d M10 prior 12d 42d
150 76d 117d 216d 292d 76d 117d 216d 292d
140 384d 496d 797d 974d
384d 496d 797d 974d
140
Pore water pressure (kPa)

Pore water pressure (kPa)

130
130
120
120

110
110

100 100

90 90
0 200 400 600 800 1000 0 200 400 600 800 1000
(c) Time (day) (d) Time (day)

Fig. 11. The predicted pore water pressure at 6 m based on four different updating schemes with creep ignored: (a) sur_sett; (b) sett_only;
(c) epp_only; and (d) sett_epp.

The updated κ, λ, and φ are correlated with the updated per- adjusted to larger values to match the measurements if creep is
meability kv =γ w, although they are originally assumed to be inde- ignored. Also, the updated parameters obtained by the updating
pendent in Bayesian back analysis. For example, when the updating scheme, ‘sett_only,’ in Table 4 where creep is incorporated are
scheme ‘sett_only’ is applied, the settlement at all monitoring closer to the corresponding values in Buttling et al. (2018) than the
depths can be well predicted by using either 216 or 797 days mon- ones in Table 3 where creep is ignored. For example, when the up-
itoring settlement data at various depths. The updated κ, λ, and φ dating scheme ‘sett_only’ is applied, the updated κ for the first
for the first soil layer using 216 days of monitoring data are 0.0182, layer using 76, 216, 496, and 797 days of monitoring data when
0.1747, and 0.0095, respectively. These values are much larger than creep is considered are 0.0062, 0.0073, 0.0182, and 0.0094 respec-
the ones (0.0094, 0.0914, and 0.0048, respectively) using 797 days tively match better with 0.0039, which is suggested in Buttling et al.
of monitoring data. In contrast, the updated permeability kv =γ w (2018) comparing with the ones (0.0072, 0.0093, 0.0125, and
(1.71 × 10−6 ) is much smaller than the one (1.46 × 10−3 ) obtained 0.0124, respectively) when creep is ignored. This indicates that
using 797 days of monitoring data. This indicates that if the com- incorporating creep would improve the accuracy of the updated soil
pressibility of the soil is high, then the permeability should be low parameters.
to match the measurements. Based on Tables 5 and 6, the small COV of the soil parameters
Comparing Table 3 where creep is ignored with Table 4 where can be obtained whether creep is considered or not. The COVs
creep is considered, the updated κ and λ values by the updating of the random variables in Bayesian back analysis obtained by
schemes (‘sur_sett,’ ‘sett_only’ and ‘sett_epp’) in Table 3 are all ‘sett_only’ are small. However, the COVs obtained by ‘sett_epp’
much larger than the ones in Table 4. For example, when 496 days are even smaller than the ones obtained by ‘sett_only,’ especially
of both monitoring settlement and pore water pressure data are used for R1 (λ=κ) and R2 (φ=λ) whether creep is considered or not. This
(‘sett_epp’), the updated κ and λ of the first soil layer are 0.0193 suggests that the incorporation of both the settlement and pore
and 0.1932 in Table 3, while are 0.0092 and 0.0924 in Table 4, water pressure data can reduce the variability in the updated soil
respectively. It can be concluded that the κ and λ need to be parameters. It can also be inferred from Tables 5 and 6 that the

© ASCE 04024025-13 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 150 150

M10 prior 12d 42d M10 prior 12d 42d

140 76d 117d 216d 292d 140 76d 117d 216d 292d
384d 496d 797d 974d 384d 496d 797d 974d
Pore water pressure (kPa)

Pore water pressure (kPa)

130 130

120 120

110 110

100 100
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

90 90
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

160 160

M10 prior 12d 42d M10 prior 12d 42d

150 76d 117d 216d 292d 150 76d 117d 216d 292d
384d 496d 797d 974d 384d 496d 797d 974d
140 140

Pore water pressure (kPa)

Pore water pressure (kPa)

130 130

120 120

110 110

100 100

90 90
0 200 400 600 800 1000 0 200 400 600 800 1000

(c) Time (day) (d) Time (day)

Fig. 12. The predicted pore water pressure at 6 m based on four different updating schemes with creep considered: (a) sur_sett; (b) sett_only;
(c) epp_only; and (d) sett_epp.

140 140
M2 sur_sett2 sett_only2 epp_only2 sett_epp2 M2 sur_sett2 sett_only2 epp_only2 sett_epp2
M6 sur_sett6 sett_only6 epp_only6 sett_epp6 M6 sur_sett6 sett_only6 epp_only6 sett_epp6
120 120

100
Pore water pressure (kPa)

100
Pore water pressure (kPa)

80 80

60 60

40 40

20 20

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
(a) Time (day) (b) Time (day)

Fig. 13. The predicted pore water pressure at 2 and 6 m based on four different updating schemes with (a) creep ignored; and (b) with creep
considered using the monitoring data prior to 496th day.

© ASCE 04024025-14 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 Table 3. The prior and the posterior mean of the soil properties based on four updating schemes with creep ignored
Updating scheme
Soil Source of Data applied
variables parameter (day) sur_sett sett_only epp_only sett_epp
Soil layer 1
κ Prior — 0.013 0.013 0.013 0.013
Buttling et al. (2018) — 0.0039 0.0039 0.0039 0.0039
Posterior 76 0.0054 0.0072 1.92 × 10−6 0.0239
216 0.0010 0.0093 0.0001 0.0119
496 0.0018 0.0125 0.0223 0.0193
797 0.0106 0.0124 0.0233 0.0204
λ Prior — 0.0435 0.0435 0.0435 0.0435
Range — — — — —
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Buttling et al. (2018) — 0.0865 0.0865 0.0865 0.0865

Posterior 76 0.0538 0.0717 1.92 × 10−5 0.2387
216 0.0098 0.0926 0.0008 0.1190
496 0.0182 0.1244 0.2231 0.1932
797 0.1055 0.1243 0.2333 0.2038
kv =γ w Prior — 2.35 × 10−4 2.35 × 10−4 2.35 × 10−4 2.35 × 10−4
Buttling et al. (2018) — 1.56 × 10−4 1.56 × 10−4 1.56 × 10−4 1.56 × 10−4
Posterior 76 2.36 × 10−3 2.05 × 10−3 1.26 × 10−3 4.37 × 10−6
216 1.84 × 10−3 1.37 × 10−3 1.34 × 10−3 2.62 × 10−3
496 1.51 × 10−3 2.13 × 10−3 9.99 × 10−7 2.68 × 10−3
797 1.56 × 10−3 2.32 × 10−3 6.43 × 10−7 2.68 × 10−3
Soil layer 2
κ Prior — 0.028 0.028 0.028 0.028
Buttling et al. (2018) — 0.013 0.013 0.013 0.013
Posterior 76 0.0041 0.0044 0.0172 0.0003
216 0.0037 0.0031 0.0005 0.0007
496 0.0042 0.0027 1.92 × 10−6 0.0524
797 0.0014 0.0361 1.19 × 10−5 0.0517
λ Prior — 0.3478 0.3478 0.3478 0.3478
Range — (0.130, 0.782) (0.130, 0.782) (0.130, 0.782) (0.130, 0.782)
Buttling et al. (2018) — 0.4887 0.4887 0.4887 0.4887
Posterior 76 0.0408 0.0441 0.1720 0.0025
216 0.0373 0.0312 0.0051 0.0075
496 0.0417 0.0266 1.92 × 10−5 0.5240
797 0.0136 0.3608 1.18 × 10−4 0.5172
kv =γ w Prior — 2.62 × 10−5 2.62 × 10−5 2.62 × 10−5 2.62 × 10−5
Buttling et al. (2018) — 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5
Posterior 76 2.63 × 10−4 2.53 × 10−4 4.25 × 10−5 2.44 × 10−5
216 2.38 × 10−4 1.75 × 10−4 1.84 × 10−4 2.53 × 10−4
496 2.14 × 10−4 2.88 × 10−4 7.21 × 10−10 2.99 × 10−4
797 1.96 × 10−4 2.84 × 10−4 3.73 × 10−10 2.99 × 10−4
Soil layer 3
κ Prior — 0.061 0.061 0.061 0.061
Buttling et al. (2018) — 0.013 0.013 0.013 0.013
Posterior 76 0.1001 0.0920 0.1113 0.1065
216 0.0951 0.1076 0.1016 0.1158
496 0.0971 0.1150 1.2 × 10−6 0.1158
797 0.1090 0.1155 3.73 × 10−5 0.1158
λ Prior — 0.587 0.587 0.587 0.587
Range — (0.130, 0.782) (0.130, 0.782) (0.130, 0.782) (0.130, 0.782)
Buttling et al. (2018) — 0.4887 0.4887 0.4887 0.4887
Posterior 76 0.9971 0.9143 1.1110 1.0647
216 0.9475 1.0699 1.0096 1.1575
496 0.9666 1.1488 1.20 × 10−5 1.1582
797 1.0875 1.1535 3.72 × 10−4 1.1583
kv =γ w Prior — 6.66 × 10−5 6.66 × 10−5 6.66 × 10−5 6.66 × 10−5
Buttling et al. (2018) — 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5
Posterior 76 7.44 × 10−4 7.34 × 10−4 6.17 × 10−5 7.12 × 10−4
216 6.91 × 10−4 5.35 × 10−4 3.03 × 10−5 2.34 × 10−4
496 5.33 × 10−4 1.88 × 10−4 1.83 × 10−6 1.19 × 10−4
797 6.14 × 10−6 1.09 × 10−4 6.79 × 10−7 1.02 × 10−4

© ASCE 04024025-15 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 Table 3. (Continued.)
Updating scheme
Soil Source of Data applied
variables parameter (day) sur_sett sett_only epp_only sett_epp
Soil layer 4
κ Prior — 0.087 0.087 0.087 0.087
Buttling et al. (2018) — 0.045 0.045 0.045 0.045
Posterior 76 0.0036 0.0030 0.0038 0.1496
216 0.0444 0.0843 0.1346 0.1649
496 0.1615 0.1623 1.34 × 10−5 0.1652
797 0.1564 0.1612 5.70 × 10−6 0.1651
λ Prior — 1.3043 1.3043 1.3043 1.3043
Range — (0.261, 1.825) (0.261, 1.825) (0.261, 1.825) (0.261, 1.825)
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Buttling et al. (2018) — 1.0999 1.0999 1.0999 1.0999

Posterior 76 0.0363 0.0302 0.0379 1.4957
216 0.4421 0.8376 1.3380 1.6485
496 1.6079 1.6211 1.34 × 10−4 1.6514
797 1.5611 1.6102 5.70 × 10−5 1.6513
kv =γ w Prior — 1.06 × 10−5 1.06 × 10−5 1.06 × 10−5 1.06 × 10−5
Buttling et al. (2018) — 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6
Posterior 76 8.79 × 10−5 6.08 × 10−5 6.63 × 10−5 3.74 × 10−5
216 1.87 × 10−5 3.91 × 10−5 9.45 × 10−7 6.16 × 10−8
496 5.83 × 10−6 3.06 × 10−6 1.01 × 10−4 7.65 × 10−9
797 3.78 × 10−5 3.59 × 10−5 4.50 × 10−5 1.10 × 10−8
Soil layer 5
κ Prior — 0.087 0.087 0.087 0.087
Buttling et al. (2018) — 0.045 0.045 0.045 0.045
Posterior 76 0.0111 0.0228 0.0031 0.0004
216 0.1535 0.1501 0.0001 0.1646
496 0.1599 0.1576 6.10 × 10−6 0.1650
797 0.1444 0.1586 1.45 × 10−6 0.1650
λ Prior — 0.913 0.913 0.913 0.913
Range — (0.261, 1.825) (0.261, 1.825) (0.261, 1.825) (0.261, 1.825)
Buttling et al. (2018) — 1.0999 1.0999 1.0999 1.0999
Posterior 76 0.1106 0.2266 0.0308 0.0037
216 1.5290 1.4915 0.0006 1.6454
496 1.5922 1.5737 6.10 × 10−5 1.6502
797 1.4413 1.5839 1.45 × 10−5 1.6501
kv =γ w Prior — 9.11 × 10−6 9.11 × 10−6 9.11 × 10−6 9.11 × 10−6
Buttling et al. (2018) — 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6
Posterior 76 2.99 × 10−5 5.84 × 10−6 3.81 × 10−5 6.16 × 10−5
216 8.30 × 10−5 1.15 × 10−5 3.93 × 10−6 8.37 × 10−5
496 6.33 × 10−5 3.55 × 10−5 2.05 × 10−6 4.97 × 10−5
797 7.05 × 10−6 6.34 × 10−6 5.72 × 10−7 3.41 × 10−5
Soil layer 6
κ Prior — 0.107 0.107 0.107 0.107
Buttling et al. (2018) — 0.045 0.045 0.045 0.045
Posterior 76 0.1959 0.0967 0.2028 0.2031
216 0.1982 0.1814 3.87 × 10−5 0.2032
496 0.2022 0.1674 0.1008 0.2032
797 0.1972 0.1804 0.0354 0.2033
λ Prior — 1.0609 1.0609 1.0609 1.0609
Range — (0.261, 1.825) (0.261, 1.825) (0.261, 1.825) (0.261, 1.825)
Buttling et al. (2018) — 1.0999 1.0999 1.0999 1.0999
Posterior 76 1.9518 0.9609 2.0239 2.0303
216 1.9737 1.8031 0.0004 2.0318
496 2.0135 1.6713 1.0073 2.0321
797 1.9683 1.8014 0.3543 2.0325
kv =γ w Prior — 9.80 × 10−6 9.80 × 10−6 9.80 × 10−6 9.80 × 10−6
Buttling et al. (2018) — 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6
Posterior 76 2.19 × 10−5 6.05 × 10−5 1.17 × 10−5 1.19 × 10−5
216 2.27 × 10−5 1.65 × 10−5 7.42 × 10−10 5.60 × 10−6
496 2.00 × 10−5 1.55 × 10−5 1.79 × 10−6 5.97 × 10−6
797 6.02 × 10−5 1.37 × 10−5 6.01 × 10−7 7.15 × 10−6

© ASCE 04024025-16 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 Table 4. The prior and the posterior mean of the soil properties based on four updating schemes with creep considered
Updating scheme
Soil Source of Data applied
variables parameter (day) sur_sett sett_only epp_only sett_epp
Soil layer 1
κ Prior — 0.013 0.013 0.013 0.013
Buttling et al. (2018) — 0.0039 0.0039 0.0039 0.0039
Posterior 76 0.0018 0.0062 3.09 × 10−6 0.0055
216 0.0012 0.0073 4.44 × 10−6 0.0074
496 0.0016 0.0182 0.0001 0.0092
797 0.0037 0.0094 0.0155 0.0085
λ Prior — 0.0435 0.0435 0.0435 0.0435
Range — — — — —
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Buttling et al. (2018) — 0.0865 0.0865 0.0865 0.0865

Posterior 76 0.0173 0.0596 3.09 × 10−5 0.0549
216 0.0119 0.0717 4.43 × 10−5 0.0742
496 0.0155 0.1747 0.0008 0.0924
797 0.0358 0.0914 0.1547 0.0850
φ Prior — 0.002 0.002 0.002 0.002
Range — — — — —
Buttling et al. (2018) — 0.0054 0.0054 0.0054 0.0054
Posterior 76 0.0010 0.0035 1.75 × 10−6 0.0029
216 0.0007 0.0041 2.02 × 10−6 0.0059
496 0.0009 0.0095 4.69 × 10−5 0.0045
797 0.0020 0.0048 0.0088 0.0044
kv =γ w Prior — 2.35 × 10−4 2.35 × 10−4 2.35 × 10−4 2.35 × 10−4
Buttling et al. (2018) — 1.56 × 10−4 1.56 × 10−4 1.56 × 10−4 1.56 × 10−4
Posterior 76 2.10 × 10−3 1.14 × 10−3 1.11 × 10−3 2.60 × 10−3
216 1.73 × 10−3 8.04 × 10−4 4.90 × 10−4 2.69 × 10−3
496 1.69 × 10−3 1.71 × 10−6 1.55 × 10−4 2.57 × 10−3
797 1.50 × 10−3 1.46 × 10−3 9.85 × 10−6 2.51 × 10−3
Soil layer 2
κ Prior — 0.028 0.028 0.028 0.028
Buttling et al. (2018) — 0.013 0.013 0.013 0.013
Posterior 76 0.0188 0.0181 0.0007 0.0001
216 0.0353 0.0221 1.17 × 10−5 0.0001
496 0.0079 0.0452 0.0031 0.0010
797 0.0180 0.0099 0.0001 0.0005
λ Prior — 0.3478 0.3478 0.3478 0.3478
Range — (0.130, 0.782) (0.130, 0.782) (0.130, 0.782) (0.130, 0.782)
Buttling et al. (2018) — 0.4887 0.4887 0.4887 0.4887
Posterior 76 0.1833 0.1749 0.0066 0.0008
216 0.3465 0.2175 0.0001 0.0006
496 0.0756 0.4342 0.0311 0.0102
797 0.1733 0.0964 0.0005 0.0047
φ Prior — 0.006 0.006 0.006 0.006
Range — (0.004, 0.053) (0.004, 0.053) (0.004, 0.053) (0.004, 0.053)
Buttling et al. (2018) — 0.0208 0.0208 0.0208 0.0208
Posterior 76 0.0108 0.0102 0.0004 4.24 × 10−5
216 0.0196 0.0124 5.33 × 10−6 4.78 × 10−5
496 0.0046 0.0237 0.0018 0.0005
797 0.0099 0.0051 2.96 × 10−5 0.0002
kv =γ w Prior — 2.62 × 10−5 2.62 × 10−5 2.62 × 10−5 2.62 × 10−5
Buttling et al. (2018) — 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5
Posterior 76 1.69 × 10−4 1.66 × 10−4 1.04 × 10−4 5.23 × 10−5
216 1.85 × 10−4 1.89 × 10−4 4.98 × 10−8 4.71 × 10−5
496 2.02 × 10−4 1.77 × 10−6 6.70 × 10−5 2.82 × 10−4
797 2.07 × 10−4 2.61 × 10−4 1.58 × 10−10 2.87 × 10−4
Soil layer 3
κ Prior — 0.061 0.061 0.061 0.061
Buttling et al. (2018) — 0.013 0.013 0.013 0.013
Posterior 76 0.0371 0.0518 0.1157 0.1085
216 0.0490 0.0593 0.0003 0.1156
496 0.1086 0.0610 0.0900 0.0944
797 0.0177 0.0723 0.0005 0.0971

© ASCE 04024025-17 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 Table 4. (Continued.)
Updating scheme
Soil Source of Data applied
variables parameter (day) sur_sett sett_only epp_only sett_epp
λ Prior — 0.587 0.587 0.587 0.587
Range — (0.130, 0.782) (0.130, 0.782) (0.130, 0.782) (0.130, 0.782)
Buttling et al. (2018) — 0.4887 0.4887 0.4887 0.4887
Posterior 76 0.3622 0.4993 1.1558 1.0844
216 0.4817 0.5849 0.0035 1.1560
496 1.0365 0.5853 0.8896 0.9435
797 0.1704 0.7012 0.0046 0.9705
φ Prior — 0.006 0.006 0.006 0.006
Range — (0.004, 0.053) (0.004, 0.053) (0.004, 0.053) (0.004, 0.053)
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Buttling et al. (2018) — 0.0208 0.0208 0.0208 0.0208

Posterior 76 0.0213 0.0291 0.0656 0.0572
216 0.0273 0.0334 0.0002 0.0924
496 0.0632 0.0319 0.0508 0.0458
797 0.0097 0.0370 0.0003 0.0498
kv =γ w Prior — 6.66 × 10−5 6.66 × 10−5 6.66 × 10−5 6.66 × 10−5
Buttling et al. (2018) — 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5 5.83 × 10−5
Posterior 76 6.64 × 10−4 6.64 × 10−4 1.73 × 10−4 1.43 × 10−4
216 5.20 × 10−4 2.61 × 10−4 5.59 × 10−9 5.32 × 10−6
496 4.43 × 10−6 5.58 × 10−4 2.22 × 10−11 7.47 × 10−4
797 2.80 × 10−4 1.16 × 10−4 4.27 × 10−7 7.60 × 10−4
Soil layer 4
κ Prior — 0.087 0.087 0.087 0.087
Buttling et al. (2018) — 0.045 0.045 0.045 0.045
Posterior 76 0.1012 0.1310 0.0605 0.0003
216 0.0318 0.1509 0.0002 0.1645
496 0.1505 0.1429 0.1648 0.0004
797 0.0783 0.1580 0.0001 0.0003
λ Prior — 1.3043 1.3043 1.3043 1.3043
Range — (0.261, 1.825) (0.261, 1.825) (0.261, 1.825) (0.261, 1.825)
Buttling et al. (2018) — 1.0999 1.0999 1.0999 1.0999
Posterior 76 0.9886 1.2627 0.6041 0.0031
216 0.3126 1.4872 0.0015 1.6439
496 1.4364 1.3720 1.6294 0.0039
797 0.7560 1.5331 0.0012 0.0029
φ Prior — 0.007 0.007 0.007 0.007
Range — (0.006, 0.079) (0.006, 0.079) (0.006, 0.079) (0.006, 0.079)
Buttling et al. (2018) — 0.0207 0.0207 0.0207 0.0207
Posterior 76 0.0581 0.0736 0.0343 0.0002
216 0.0177 0.0850 0.0001 0.1314
496 0.0876 0.0749 0.0930 0.0002
797 0.0431 0.0809 0.0001 0.0001
kv =γ w Prior — 1.06 × 10−5 1.06 × 10−5 1.06 × 10−5 1.06 × 10−5
Buttling et al. (2018) — 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6
posterior 76 1.38 × 10−5 4.58 × 10−7 5.58 × 10−5 9.44 × 10−5
216 3.52 × 10−5 3.41 × 10−6 1.14 × 10−4 5.15 × 10−6
496 3.24 × 10−5 4.80 × 10−6 1.33 × 10−6 1.07 × 10−4
797 5.59 × 10−5 2.24 × 10−6 6.20 × 10−5 9.17 × 10−5
Soil layer 5
κ Prior — 0.087 0.087 0.087 0.087
Buttling et al. (2018) — 0.045 0.045 0.045 0.045
Posterior 76 0.1235 0.0964 0.0431 0.0001
216 0.0595 0.1013 0.0001 3.89 × 10−5
496 0.1557 0.0803 0.0002 1.19 × 10−5
797 0.0850 0.0736 0.0001 1.47 × 10−5
λ Prior — 0.913 0.913 0.913 0.913
Range — (0.261, 1.825) (0.261, 1.825) (0.261, 1.825) (0.261, 1.825)
Buttling et al. (2018) — 1.0999 1.0999 1.0999 1.0999
Posterior 76 1.2071 0.9293 0.4302 0.0007
216 0.5851 0.9982 0.0013 0.0004
496 1.4861 0.7704 0.0024 0.0001
797 0.8204 0.7139 0.0009 0.0001

© ASCE 04024025-18 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 Table 4. (Continued.)
Updating scheme
Soil Source of Data applied
variables parameter (day) sur_sett sett_only epp_only sett_epp
φ Prior — 0.007 0.007 0.007 0.007
Range — (0.006, 0.079) (0.006, 0.079) (0.006, 0.079) (0.006, 0.079)
Buttling et al. (2018) — 0.0207 0.0207 0.0207 0.0207
Posterior 76 0.0709 0.0542 0.0244 3.47 × 10−5
216 0.0332 0.0570 0.0001 3.10 × 10−5
496 0.0906 0.0420 0.0001 5.78 × 10−6
797 0.0468 0.0377 4.96 × 10−5 7.54 × 10−6
kv =γ w Prior — 9.11 × 10−6 9.11 × 10−6 9.11 × 10−6 9.11 × 10−6
Buttling et al. (2018) — 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Posterior 76 2.05 × 10−5 3.81 × 10−6 6.08 × 10−5 1.80 × 10−5

216 2.83 × 10−5 2.29 × 10−6 6.19 × 10−6 2.32 × 10−5
496 7.63 × 10−6 6.80 × 10−6 3.46 × 10−6 3.57 × 10−6
797 9.50 × 10−6 1.19 × 10−5 9.28 × 10−7 2.69 × 10−6
Soil layer 6
κ Prior — 0.107 0.107 0.107 0.107
Buttling et al. (2018) — 0.045 0.045 0.045 0.045
Posterior 76 0.1690 0.1085 0.1910 0.1108
216 0.1733 0.0863 0.1902 0.0831
496 0.0472 0.0950 3.36 × 10−5 0.1707
797 0.1445 0.0909 0.0859 0.1585
λ Prior — 1.0609 1.0609 1.0609 1.0609
Range — (0.261, 1.825) (0.261, 1.825) (0.261, 1.825) (0.261, 1.825)
Buttling et al. (2018) — 1.0999 1.0999 1.0999 1.0999
Posterior 76 1.6514 1.0458 1.9073 1.1072
216 1.7030 0.8508 1.8983 0.8304
496 0.4507 0.9120 0.0003 1.7063
797 1.3944 0.8818 0.8580 1.5848
φ Prior — 0.007 0.007 0.007 0.007
Range — (0.006, 0.079) (0.006, 0.079) (0.006, 0.079) (0.006, 0.079)
Buttling et al. (2018) — 0.0207 0.0207 0.0207 0.0207
Posterior 76 0.0971 0.0610 0.1083 0.0584
216 0.0965 0.0486 0.0866 0.0664
496 0.0275 0.0498 1.90 × 10−5 0.0828
797 0.0795 0.0465 0.0486 0.0814
kv =γ w Prior — 9.80 × 10−6 9.80 × 10−6 9.80 × 10−6 9.80 × 10−6
Buttling et al. (2018) — 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6 7.3 × 10−6
Posterior 76 6.04 × 10−6 5.78 × 10−6 1.04 × 10−5 6.37 × 10−6
216 4.28 × 10−6 5.82 × 10−6 4.71 × 10−6 1.94 × 10−6
496 8.52 × 10−5 4.39 × 10−6 1.07 × 10−9 3.01 × 10−6
797 1.47 × 10−5 5.55 × 10−6 9.43 × 10−7 2.69 × 10−6

incorporating more monitoring data could decrease the COVs 1. The long-term settlement predictions can match well with the
whether creep is considered or not. For example, when ‘sett_epp’ measurements whether creep is considered or not. However,
is applied with 76 days of monitoring data applied, the COV is the ultimate consolidation settlement is slightly underestimated.
2.32 × 10−2 with creep ignored and 9.73 × 10−2 with creep consid- The κ and λ values need to be adjusted to larger values to
ered. However, when 797 days of monitoring data are used, the match the measurements if creep is ignored. Incorporating
COVs decrease to 2.02 × 10−2 with creep ignored and 1.44 × 10−2 creep can improve the accuracy of the predictions of the long-
with creep considered. term behaviors of Ballina embankment if the noisy monitoring
pore water pressure data is not incorporated.
2. A single type of observations, whether settlement or pore water
Conclusions pressure, is not sufficient to accurately predict the long-term per-
formance of Ballina embankment in terms of both settlement
In this study, Bayesian back analysis is combined with a decoupled and pore water pressure. If only the monitoring pore water pres-
consolidation constitutive model to calibrate the soil parameters sure data is incorporated, the settlement prediction would devi-
based on the in situ measurements and to conduct long-term per- ate significantly from the measurements. Similarly, if only the
formance predictions for the Ballina embankment. The effects of monitoring settlements are used, the dissipation rate of pore
creep and the monitoring data on the settlement prediction are in- water pressure would be overestimated.
vestigated. Based on the obtained results, the following conclusions 3. The variability in the updated soil parameters can be reduced by
have been drawn: incorporating more types of observations and increasing the

© ASCE 04024025-19 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 © ASCE

Table 5. The COV of posteriors for each soil layer with creep ignored
Data applied (day)
76 216 496 797
Soil Soil
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

variables layer sur_sett sett_only epp_only sett_epp sur_sett sett_only epp_only sett_epp sur_sett sett_only epp_only sett_epp sur_sett sett_only epp_only sett_epp
κ 1 2.12×10−1 1.12×10−1 2.21×10−3 2.32×10−2 7.20×10−1 1.23×10−1 1.89×10−1 5.65×10−2 9.85×10−1 4.09×10−2 4.73×10−2 2.63×10−2 1.42×10−1 4.37×10−2 1.55×10−2 2.03×10−2
2 8.92×10−1 6.83×10−1 5.86×10−3 6.76×10−1 7.74×10−1 6.71×10−1 5.21×10−1 6.53×10−1 6.49×10−1 6.91×10−1 9.45×10−1 2.20×10−2 5.39×10−1 2.05×10−1 1.99×10−1 1.69×10−2
3 4.19×10−2 1.57×10−2 5.94×10−4 1.66×10−2 1.04×10−1 3.63×10−2 9.93×10−2 7.54×10−4 1.05×10−1 2.80×10−3 8.43×10−1 4.67×10−4 2.87×10−2 2.24×10−3 6.13×10−2 3.30×10−4
4 1.04 8.02×10−1 2.45×10−3 4.31×10−2 7.56×10−1 4.93×10−1 8.14×10−3 2.26×10−3 3.01×10−2 2.23×10−2 7.04×10−1 7.49×10−4 7.06×10−2 1.20×10−2 7.26×10−1 5.72×10−4
5 1.01 7.29×10−1 5.73×10−3 8.75×10−1 8.23×10−2 4.91×10−2 4.77×10−1 3.74×10−3 2.40×10−2 3.13×10−2 6.37×10−1 1.28×10−3 6.54×10−2 2.45×10−2 6.56×10−1 9.62×10−4
6 5.53×10−2 1.77×10−1 1.69×10−4 9.26×10−4 2.68×10−2 5.32×10−2 8.28×10−2 3.13×10−4 3.67×10−3 2.61×10−2 1.29×10−1 1.26×10−4 8.68×10−3 1.08×10−2 7.77×10−2 1.42×10−4
kv =γ w 1 7.98×10−2 1.39×10−1 3.27×10−3 1.93×10−1 2.15×10−1 4.09×10−1 1.55×10−1 2.24×10−2 5.28×10−1 9.14×10−2 6.38×10−2 2.88×10−3 2.32×10−1 9.87×10−2 2.38×10−2 3.76×10−3
2 9.37×10−2 9.55×10−2 4.84×10−3 1.85×10−1 2.13×10−1 5.32×10−1 1.77×10−1 8.69×10−2 3.21×10−1 4.64×10−2 7.76×10−1 3.46×10−3 3.46×10−1 4.52×10−2 5.98×10−1 4.45×10−3
3 1.47×10−2 2.53×10−2 1.51×10−3 1.99×10−2 9.69×10−2 1.79×10−1 1.65×10−1 4.93×10−2 2.44×10−1 6.91×10−2 7.71×10−2 5.64×10−2 1.88×10−1 1.21×10−1 4.82×10−2 3.87×10−2
4 3.05×10−1 5.65×10−1 2.25×10−3 1.01×10−1 1.17 3.60×10−1 7.44×10−2 9.02×10−1 9.16×10−1 8.15×10−1 6.97×10−2 6.42×10−1 2.69×10−1 2.86×10−1 6.49×10−2 6.58×10−1
5 6.04×10−1 1.44 1.99×10−4 3.19×10−2 2.59×10−1 2.90×10−1 1.37×10−1 1.33×10−2 2.32×10−1 2.23×10−1 1.40×10−1 9.27×10−3 4.94×10−1 2.69×10−1 8.00×10−2 1.24×10−2
6 1.82×10−1 2.61×10−1 1.02×10−3 9.45×10−3 2.02×10−1 7.75×10−2 1.76×10−1 1.04×10−2 1.46×10−1 6.92×10−2 1.29×10−1 6.46×10−3 1.04×10−1 6.64×10−2 7.80×10−2 9.99×10−3
R1 — 2.42×10−3 4.03×10−3 7.97×10−5 1.38×10−4 3.88×10−3 4.32×10−3 1.26×10−3 1.95×10−4 5.25×10−3 8.00×10−4 3.00×10−4 1.21×10−4 4.41×10−3 1.02×10−3 2.08×10−4 4.96×10−5
04024025-20

Table 6. The COV of the posteriors for each soil layer with creep considered
Data applied (day)
76 216 496 797
Soil Soil
variables layer sur_sett sett_only epp_only sett_epp sur_sett sett_only epp_only sett_epp sur_sett sett_only epp_only sett_epp sur_sett sett_only epp_only sett_epp
κ 1 5.43×10−1 1.11×10−2 1.39×10−3 9.73×10−2 4.87×10−1 5.76×10−2 4.02 5.66×10−2 5.78×10−1 3.66×10−2 2.18×10−2 4.01×10−2 3.48×10−1 1.36×10−2 5.97×10−2 1.44×10−2
−1 −1 −2 −1 −1 −1
2 3.59×10 1.94×10 1.40×10 6.97×10 3.22×10 5.23×10 3.06×10−1 5.68×10 −1
8.64×10 −1
9.82×10−2
4.13×10−2
6.36×10 −1
3.04×10 −1
2.13×10−1
2.31×10−1
1.23×10−1
3 1.78×10−1 1.89×10−2 1.44×10−4 1.84×10−2 1.73×10−1 5.93×10−2 5.83×10−1 1.19×10−3 8.71×10−2 4.92×10−2 3.53×10−3 1.05×10−2 2.15×10−1 1.76×10−2 1.75×10−2 1.64×10−3
4 3.33×10−1 4.59×10−2 7.21×10−3 5.99×10−1 9.48×10−1 6.36×10−2 5.00×10−1 1.34×10−3 5.62×10−2 4.51×10−2 1.58×10−3 4.93×10−1 2.40×10−1 1.61×10−2 3.39×10−1 6.78×10−2
5 2.78×10−1 1.66×10−1 6.41×10−4 7.81×10−1 3.22×10−1 2.01×10−1 3.34×10−1 5.05×10−1 8.33×10−2 9.69×10−2 1.66×10−2 9.35×10−1 3.33×10−1 2.98×10−2 6.02×10−2 2.99×10−1
6 8.43×10−2 5.35×10−2 5.81×10−4 4.35×10−2 7.32×10−2 6.81×10−2 1.86×10−2 3.66×10−3 1.59×10−1 3.40×10−2 3.94×10−3 1.37×10−2 9.33×10−2 1.62×10−2 6.25×10−2 1.44×10−2
kv =γ w 1 1.99×10−1 2.61×10−1 7.33×10−3 2.97×10−2 2.34×10−1 1.88×10−1 4.21×10−1 5.22×10−3 2.96×10−1 1.28×10−1 1.02×10−2 2.57×10−2 1.90×10−1 1.55×10−1 1.87×10−1 9.55×10−3
2 3.38×10−1 9.46×10−2 9.88×10−3 5.07×10−2 1.85×10−1 2.35×10−1 1.02 2.40×10−2 3.45×10−1 1.52 1.67×10−2 3.71×10−2 2.45×10−1 3.61×10−2 6.73×10−1 2.04×10−2
8.49×10−2 4.27×10−2 1.00×10−3 4.55×10−2 1.28×10−1 1.13×10−1 5.87×10−1 2.94×10−2 2.65×10−1 6.59×10−2 9.30×10−2 1.53×10−2 2.98×10−1 4.97×10−2 5.09×10−2 2.52×10−4
J. Geotech. Geoenviron. Eng.

3
4 7.50×10−1 4.41×10−1 1.87×10−2 9.43×10−2 4.69×10−1 3.13×10−1 2.80×10−2 4.05×10−2 1.72×10−1 2.61×10−1 1.47×10−2 8.42×10−2 2.61×10−1 3.05×10−1 1.06×10−1 2.01×10−2
5 3.77×10−1 2.40 1.04×10−3 5.13×10−2 6.69×10−1 5.83×10−1 4.51×10−2 1.13×10−2 3.23×10−1 2.56×10−1 1.60×10−2 1.99×10−2 6.52×10−1 1.44×10−1 6.39×10−2 2.02×10−2
6 6.84×10−1 1.71×10−1 2.75×10−3 4.19×10−2 1.25×10−1 1.32×10−1 7.35×10−2 8.67×10−3 1.13×10−1 8.40×10−2 1.78×10−2 1.35×10−2 1.54×10−1 2.62×10−2 6.44×10−2 8.83×10−3
R1 — 1.57×10−2 3.58×10−3 6.11×10−4 2.32×10−4 5.79×10−3 1.08×10−2 2.04×10−3 1.47×10−4 3.30×10−2 1.68×10−2 5.01×10−3 1.43×10−4 3.92×10−2 7.66×10−3 8.61×10−4 1.30×10−5
R2 — 4.94×10−2 2.22×10−2 4.05×10−1 3.88×10−2 2.38×10−2 2.72×10−2 3.70×10−1 8.80×10−4 4.44×10−2 2.85×10−2 1.80×10−1 1.83×10−2 3.06×10−2 1.09×10−2 2.30×10−1 1.63×10−2

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 amount of monitoring data. The soil parameters updated by Chai, J., J. Shen, M. Liu, and D. Yuan. 2018. “Predicting the performance
‘sett_epp’ where both the monitoring settlement and pore water of embankments on PVD-improved subsoils.” Comput. Geotech.
pressure data are applied can converge to the stable values if 93 (Feb): 222–231. https://doi.org/10.1016/j.compgeo.2017.05.018.
Chen, Z., W. Feng, and J. Yin. 2021. “A new simplified method for cal-
496 days of monitoring data are used whether creep is consid-
culating short-term and long-term consolidation settlements of multi-
ered or not. layered soils considering creep limit.” Comput. Geotech. 138 (Aug):
4. The incorporation of the pore water pressure data introduces 104324. https://doi.org/10.1016/j.compgeo.2021.104324.
variability into settlement predictions. Settlement predictions Chen, Z. J., W. Q. Feng, J. H. Yin, and X. S. Shi. 2022. “Finite element
that consider creep and those that ignore creep may respond dif- model and simple method for predicting consolidation displacement of
ferently to this variability. When creep is ignored, the impact of soft soils exhibiting creep underneath embankments in 2-D condition.”
this variability would be more significant at deeper locations, Acta Geotech. 18 (5): 2513–2528. https://doi.org/10.1007/s11440-022
while the impact could be reduced for the settlement prediction -01741-z.
that consider creep. One possible reason could be that the de- Feng, W., and J. Yin. 2017. “A new simplified Hypothesis B method for
coupled constitutive model is used to perform consolidation calculating consolidation settlements of double soil layers exhibiting
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

creep.” Int. J. Numer. Anal. Methods Geomech. 41 (6): 899–917.

analysis in this study. https://doi.org/10.1002/nag.2635.
5. The pore water pressure is difficult to predict well as the mon- Garlanger, J. E. 1972. “The consolidation of soils exhibiting creep under
itoring pore water pressure data incorporated is noisy. Huge constant effective stress.” Géotechnique 22 (1): 71–78. https://doi.org
fluctuations in the monitoring pore water pressure data, which /10.1680/geot.1972.22.1.71.
include the changes in the groundwater table caused by the sea- Gelfand, A. E., and A. F. Smith. 1990. “Sampling-based approaches to cal-
sonal rainfall, sensor movements and the clogging of fine par- culating marginal densities.” J. Am. Stat. Assoc. 85 (410): 398–409.
ticles, etc., could also explain the discrepancies between the https://doi.org/10.1080/01621459.1990.10476213.
model predictions and the measurements. Gioda, G., and G. Maier. 1980. “Direct search solution of an inverse prob-
6. Considering that some unrealistic values could be sampled lem in elastoplasticity: Identification of cohesion, friction angle and
in situ stress by pressure tunnel tests.” Int. J. Numer. Methods Eng.
when incorporating the pore water pressure data, constrained
15 (12): 1823–1848. https://doi.org/10.1002/nme.1620151207.
sampling ranges will be applied in the MCMC sampling process
Gong, W., Z. Luo, C. Juang, H. Huang, J. Zhang, and L. Wang. 2014.
to improve the accuracy of the updated soil parameters in our “Optimization of site exploration program for improved prediction
future study. Also, a correction factor that is used to correct the of tunneling-induced ground settlement in clays.” Comput. Geotech.
yield pressure obtained from the CRS test and the initial void 56 (Apr): 69–79. https://doi.org/10.1016/j.compgeo.2013.10.008.
ratio will be introduced as a random variable in Bayesian back Hansbo, S. 1981. “Consolidation of fine-grained soils by prefabricated
analysis. drains.” In Vol. 3 of Proc., 10th Int. Conf. Soil Mechanics and Foun-
dation Engineering, 677–682. Rotterdam, Netherlands: A. A. Balkema.
Hastings, W. K. 1970. “Monte Carlo sampling methods using Markov
chains and their applications.” Biometrika 57 (1): 97–109. https://doi
Data Availability Statement .org/10.1093/biomet/57.1.97.
Honjo, Y., W. Liu, and S. Guha. 1994. “Inverse analysis of an embankment
All data, models that support the findings of this study are available on soft clay by extended Bayesian method.” Int. J. Numer. Anal.
from the corresponding author upon reasonable request. Methods Geomech. 18 (10): 709–734. https://doi.org/10.1002/nag
.1610181004.
Hsein Juang, C., Z. Luo, S. Atamturktur, and H. Huang. 2013. “Bayesian
Acknowledgments updating of soil parameters for braced excavations using field observa-
tions.” J. Geotech. Geoenviron. Eng. 139 (3): 395–406. https://doi.org
This work was supported by the Australian Government through /10.1061/(ASCE)GT.1943-5606.0000782.
the Australian Research Council’s Linkage Projects funding Huang, S., J. Huang, R. Kelly, M. Jones, and A. H. M. Kamruzzaman.
2023. “Predicting settlement of embankments built on PVD-improved
scheme (Project G2000327), National Natural Science Foundation
soil using Bayesian back analysis and elasto-viscoplastic modeling.” Com-
of China (Project Nos. 41972280, 42272326), Chinese Govern- put. Geotech. 157 (May): 105323. https://doi.org/10.1016/j.compgeo.2023
ment through the China Scholarship Council (201906370054), .105323.
and the University of Newcastle through CSC top-up scholarship Huang, S., J. Huang, R. B. Kelly, C. Zeng, and J. Xie. 2021. “Settlement
(3332480). predictions of a trial embankment on Ballina Clay.” ISSMGE Int. J.
Geoeng. Case Hist. 6 (Feb): 101–114. https://doi.org/10.4417/IJGCH
-06-04-07.
Kelly, R. B., and J. Huang. 2015. “Bayesian updating for one-dimensional
References consolidation measurements.” Can. Geotech. J. 52 (9): 1318–1330.
https://doi.org/10.1139/cgj-2014-0338.
Basu, D., M. Prezzi, and M. R. Madhav. 2010. “Effect of soil disturbance
Kelly, R. B., J. A. Pineda, L. Bates, L. P. Suwal, and A. Fitzallen. 2017.
on consolidation by prefabricated vertical drains installed in a rectan-
“Site characterisation for the Ballina field testing facility.” Géotechnique
gular pattern.” Geotech. Geol. Eng. 28 (1): 61–77. https://doi.org/10 67 (4): 279–300. https://doi.org/10.1680/jgeot.15.P.211.
.1007/s10706-009-9279-7. Kelly, R. B., S. W. Sloan, J. A. Pineda, G. Kouretzis, and J. Huang. 2018.
Bjerrum, L. 1967. “Engineering geology of Norwegian normally- “Outcomes of the Newcastle symposium for the prediction of embank-
consolidated marine clays as related to settlements of buildings.” Géo- ment behaviour on soft soil.” Comput. Geotech. 93 (Jan): 9–41. https://
technique 17 (2): 83–118. https://doi.org/10.1680/geot.1967.17.2.83. doi.org/10.1016/j.compgeo.2017.08.005.
Braak, T., J. F. Cajo, and J. A. Vrugt. 2008. “Differential evolution Markov Ladd, C. C., R. Foott, F. Schlosser, K. Ishihara, and H. G. Poulos. 1977.
chain with snooker updater and fewer chains.” Stat. Comput. 18 (4): “Stress-deformation and strength characteristics, state of the art report.”
435–446. https://doi.org/10.1007/s11222-008-9104-9. In Proc., 9th Int. Conf. on SMFE, 421–494. London: International So-
Buttling, S., R. Cao, W. Lau, and D. Naicker. 2018. “Class A and Class C ciety for Soil Mechanics and Geotechnical Engineering.
numerical predictions of the deformation of an embankment on soft Le, T., D. Airey, and J. Surjadinata. 2018. “Modelling the behaviour of
ground.” Comput. Geotech. 93 (Jan): 191–203. https://doi.org/10 the Ballina test embankment.” Comput. Geotech. 93 (Dec): 115–122.
.1016/j.compgeo.2017.06.017. https://doi.org/10.1016/j.compgeo.2017.06.009.

© ASCE 04024025-21 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025

 Mesri, G., and P. M. Godlewski. 1977. “Time-and stress-compressibility Ballina embankment.” Comput. Geotech. 93 (Jan): 42–60. https://doi
interrelationship.” J. Geotech. Eng. Div. 103 (Aug): 417–430. https:// .org/10.1016/j.compgeo.2017.05.004.
doi.org/10.1061/AJGEB6.0000421. Yin, J., and W. Q. Feng. 2017. “A new simplified method and its verifi-
Metropolis, N., A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, and cation for calculation of consolidation settlement of a clayey soil with
E. Teller. 1953. “Equation of state calculations by fast computing ma- creep.” Can. Geotech. J. 54 (3): 333–347. https://doi.org/10.1139/cgj
chines.” J. Chem. Phys. 21 (6): 1087–1092. https://doi.org/10.1063/1 -2015-0290.
.1699114. Yin, J., and J. Graham. 1989. “Viscous–elastic–plastic modelling of one-
Miranda, T., A. G. Correia, and L. R. e Sousa. 2009. “Bayesian method- dimensional time-dependent behaviour of clays.” Can. Geotech. J.
ology for updating geomechanical parameters and uncertainty quanti- 26 (2): 199–209. https://doi.org/10.1139/t89-029.
fication.” Int. J. Rock Mech. Min. Sci. 46 (7): 1144–1153. https://doi.org Yin, J., and J. Graham. 1996. “Elastic visco-plastic modelling of one-
/10.1016/j.ijrmms.2009.03.008. dimensional consolidation.” Géotechnique 46 (3): 515–527. https://doi
Nash, D. F. T., and S. J. Ryde. 2001. “Modelling consolidation accelerated .org/10.1680/geot.1996.46.3.515.
by vertical drains in soils subject to creep.” Géotechnique 51 (Nov): Yin, J., J. Zhu, and J. Graham. 2002. “A new elastic viscoplastic model for
257–273. https://doi.org/10.1680/geot.2001.51.3.257. time-dependent behaviour of normally and overconsolidated clays:
Downloaded from ascelibrary.org by University of Central Florida on 04/13/24. Copyright ASCE. For personal use only; all rights reserved.

Pineda, J. A., L. P. Suwal, R. B. Kelly, L. Bates, and S. W. Sloan. 2016. Theory and verification.” Can. Geotech. J. 39 (1): 157–173. https://doi
“Characterisation of Ballina clay.” Géotechnique 66 (7): 556–577. .org/10.1139/t01-074.
https://doi.org/10.1680/jgeot.15.P.181. Yin, J. H., Z. J. Chen, and W. Q. Feng. 2022. “A general simple method for
calculating consolidation settlements of layered clayey soils with ver-
Vrugt, J. A., P. H. Stauffer, T. Wohling, B. A. Robinson, and V. V.
tical drains under staged loadings.” Acta Geotech. 17 (8): 3647–3674.
Vesselinov. 2008. “Inverse modeling of subsurface flow and transport
https://doi.org/10.1007/s11440-021-01318-2.
properties: A review with new developments.” Vadose Zone J. 7 (2):
Zhang, J., W. H. Tang, and L. M. Zhang. 2010a. “Efficient probabilistic
843–864. https://doi.org/10.2136/vzj2007.0078.
back-analysis of slope stability model parameters.” J. Geotech. Geoen-
Vrugt, J. A., C. J. F. Ter Braak, C. G. H. Diks, B. A. Robinson, J. M.
viron. Eng. 136 (1): 99–109. https://doi.org/10.1061/(ASCE)GT.1943
Hyman, and D. Higdon. 2009. “Accelerating Markov chain Monte -5606.0000205.
Carlo simulation by differential evolution with self-adaptive random- Zhang, L. L., J. Zhang, L. M. Zhang, and W. H. Tang. 2010b. “Back analysis
ized subspace sampling.” Int. J. Nonlinear Sci. Numer. Simul. 10 (3): of slope failure with Markov chain Monte Carlo simulation.” Comput.
273–290. https://doi.org/10.1515/IJNSNS.2009.10.3.273. Geotech. 37 (Apr): 905–912. https://doi.org/10.1016/j.compgeo.2010
Walker, R. 2006. “Analytical solutions for modeling soft soil consolidation .07.009.
by vertical drains.” Ph.D. thesis, Dept. of Civil Engineering, Univ. of Zhao, C., W. Gong, T. Li, C. Juang, H. Tang, and H. Wang. 2021.
Wollongong. “Probabilistic characterization of subsurface stratigraphic configuration
Walker, R., and B. Indraratna. 2006. “Vertical drain consolidation with with modified random field approach.” Eng. Geol. 288 (Nov): 106138.
parabolic distribution of permeability in smear zone.” J. Geotech. Geo- https://doi.org/10.1016/j.enggeo.2021.106138.
environ. Eng. 132 (7): 937–941. https://doi.org/10.1061/(ASCE)1090 Zheng, D., J. Huang, D. Li, R. B. Kelly, and S. W. Sloan. 2018.
-0241(2006)132:7(937). “Embankment prediction using testing data and monitored behaviour:
Walker, R., B. Indraratna, and N. Sivakugan. 2009. “Vertical and radial A Bayesian updating approach.” Comput. Geotech. 93 (Apr): 150–162.
consolidation analysis of multilayered soil using the spectral method.” https://doi.org/10.1016/j.compgeo.2017.05.003.
J. Geotech. Geoenviron. Eng. 135 (5): 657–663. https://doi.org/10.1061 Zhu, G., and J. Yin. 2000. “Elastic visco-plastic consolidation modelling
/(ASCE)GT.1943-5606.0000075. of clay foundation at Berthierville test embankment.” Int. J. Numer.
Yang, C., and J. P. Carter. 2018. “1-D finite strain consolidation analysis Anal. Methods Geomech. 24 (5): 491–508. https://doi.org/10.1002
based on isotach plasticity: Class A and Class C predictions of the /(SICI)1096-9853(20000425)24:5<491::AID-NAG78>3.0.CO;2-V.

© ASCE 04024025-22 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., 2024, 150(5): 04024025