water

Article
Numerical Simulation and Experiment Study on the
Characteristics of Non-Darcian Flow and Rheological
Consolidation of Saturated Clay
Zhongyu Liu 1, *, Yangyang Xia 1,2 , Mingsheng Shi 2, *, Jiachao Zhang 1 and Xinmu Zhu 1
1 School of Civil Engineering, Zhengzhou University, Zhengzhou 450001, China
2 School of Water Conservancy and Environment, Zhengzhou University, Zhengzhou 450001, China
* Correspondence: zhyliu@zzu.cn (Z.L.); sms315@126.com (M.S.)




Received: 18 May 2019; Accepted: 29 June 2019; Published: 5 July 2019


Abstract: To investigate the characteristics of the non-Darcian water flow through saturated clay and
one-dimensional rheological consolidation behaviors of the soil in the Henan Province, we conducted
constant-head permeability and one-dimensional rheological consolidation tests with one-way
drainage using improved permeameter and oedometer tests, respectively. We then used Hansbo’s
flow equation to classify the permeability test results and one-dimensional rheological consolidation
equation combined with unified hardening (UH) constitutive model considering time effect was
introduced to simulate the oedometer test results. The obtained results showed that the improved
constant-head permeability test device was suitable for saturated clays, and that the UH constitutive
model and Hansbo’s flow equation had good applications for the saturated clays investigated in
this experiment.

Keywords: soil mechanics; saturated clay; non-Darcian flow; rheological consolidation; permeability

1. Introduction
Terzaghi’s [1] one-dimensional consolidation theory has been widely employed by researchers
since its introduction. Especially in areas with saturated clay, researchers have found that settlement
values predicted by Terzaghi’s theory have large deviations from measured values, mainly because this
theory ignores the influence of non-Darcian flow of pore water seepage and the rheological properties
of saturated clays. Therefore, many researchers have conducted studies on saturated clays considering
the above two factors. As early as 1898, King [2] found a deviation from Darcy flow in saturated clays,
which was unfortunately unrecognized by the academic community at the time. In 1960, Hansbo [3]
also witnessed this phenomenon, which he called the non-Darcian flow, and proposed a corresponding
mathematical model using experimental results. The non-Darcian flow phenomenon in saturated
clays was subsequently confirmed by many researchers around the world, and different mathematical
equations were proposed for it. For example, Swartzendruber [4] proposed the non-Newtonian
index flow model, Slepicka [5] proposed the exponential flow model, Miller et al. [6] proposed the
threshold gradient flow model, and Qi et al. [7] proposed the broken line flow model. At the same
time, Dubin et al. [8], Wang et al. [9], Deng et al. [10], Sun et al. [11], performed experiments showing
that seepage in the relevant clays deviated from Darcy’s law in varying degrees. Wang et al. [9]
attributed the deviation to the combined action of gravity water, capillary water, and weak bound
water in saturated clay under different hydraulic gradients. Currently, Hansbo’s [3] flow theory is
most commonly used, and has been applied to one-dimensional [12–15] and two-dimensional [16]
consolidations, as well as sand drains [17], further confirming the necessity of considering non-Darcian
flow in the analysis of saturated clay consolidation.

Water 2019, 11, 1385; doi:10.3390/w11071385 www.mdpi.com/journal/water

Water 2019, 11, 1385 2 of 16

All these previous studies related to clay consolidation mostly focused on non-linearity [18,19],
anisotropy [20], large-strain [21], time-dependent loading [22], and so on, while relatively few studies are
related to the rheological properties of clay. Early studies [23–25] believe that the rheological properties
of clay are caused by the viscosity of water around soil particles, the plastic adjustment of soil particles’
structure, the retention and creep due to shear stress, and hydrostatic pressure. However, there was
a disagreement as to whether rheology should be taken into account during consolidation. Buisman [23]
believed that rheological or secondary consolidation occurred after completing the main consolidation,
while Tayor et al. [24] suggested that secondary consolidation occurred throughout the consolidation
process. Bjerrum [26] confirmed Tayor’s theory through experiments and described the creep
deformation of soil using a simple element model. Then, with continuous improvement in research
methods, researchers around the world carried out in-depth research on the rheological properties
of saturated clay based on one-dimensional rheological consolidation tests considering the effects of
external load, loading ratio, preloading, drainage conditions, etc. First, Bjerrum [27] believed that the
secondary consolidation coefficient was linearly related to the external load. However, Nash et al. [28]
and Chen et al. [29] showed that the secondary consolidation coefficient first increased and then
decreased with the increase in external load. Yin et al. [30] and Reddy et al. [31] further confirmed the
above conclusion. Unlike Nash et al. [28], Chen et al. [29], Yin et al. [30], and Reddy et al. [31], it was
found that the peak value of the secondary consolidation coefficient did not necessarily appear near
pre-consolidation pressure, which Yin et al. [30] attributed to the effect of loading ratio and preloading.
More in depth research ultimately demonstrated a high degree of nonlinearity in the deformation
of saturated clay. Yin [32,33], Yao [34,35], and others have proposed some elastic visco-plastic (EVP)
models. The unified hardening (UH) constitutive model was applied in one-dimensional rheological
consolidation analysis, and successfully explained the increase in excess pore pressure near impervious
surfaces at the initial stage of rheological consolidation. Compared to the EVP model proposed by
Yin [32], the UH model proposed by Yao [34] has many advantages, such as fewer model parameters
and a clearer physical concept, and thus could be simply determined by routine experiments.
In summary, in order to further explore the permeability and rheological consolidation
characteristics of saturated clay, traditional permeability and one-dimensional consolidation test
devices, as well as saturated remolded clay were improved based on previous works. A series of
permeability and rheological consolidation tests were carried out, and their results were compared
with fitting results obtained from Hansbo’s flow model and theoretical prediction results from the UH
rheological consolidation model in order to verify the applicability of both models to saturated clay.

2. Experimental Study on the Permeability Characteristics of Saturated Clay

2.1. Constant-Head Permeability Test Device

Currently, there are two main types of permeability test methods, namely constant-head and
falling-head [36]. Previous studies on saturated clay permeability employ the falling-head tests,
but as shown in Figure 1, because of the compactness and low permeability of clay, the falling-head
permeability test requires a long time, and the errors caused by human and environmental factors
cannot be avoided during the test. However, it is noteworthy that the constant-head permeability
test requires a very short time, but this test is primarily used to measure sand or granular soil with
better permeabilities, and mostly to determine water volume using measuring cylinder or beaker.
For clays with insignificant amounts of water flow and low permeability, it is impossible to measure
water volume cannot be accurately measured in a short time; therefore, we improved the conventional
constant-head permeability test device.
Water 2019, 11,
Water 11, x FOR
FOR PEER
PEER REVIEW
REVIEW 333 of
of 16
16
Water 2019,
2019, 11, x1385 of 16

12,000
12,000 Inital void
Inital void ratio
ratio ee0=1.34
=1.34
0

(min)
Inital void ratio e =1.32

Time(min)
10,000 Inital void ratio e 0=1.32
0
10,000 Inital void
void ratio
ratio ee0=1.27
=1.27
Inital

TestTime
0
8000
8000 Inital void
Inital void ratio
ratio ee0=1.22
=1.22
0

AccumulativeTest
6000
6000

Accumulative
4000
4000
2000
2000

00

00 20
20 40
40 60
60 80
80 100
100
Hydraulic Head
Hydraulic Head (cm)
(cm)

Figure 1.
Figure Falling-head permeability
1. Falling-head permeability test
test results.
results.
Figure 1. Falling-head permeability test results.
Figure 22 shows
Figure shows a schematic
schematic diagram
diagram of of the
the improved
improved constant-head
constant-head test
test device
device proposed
proposed in in this
this
Figure 2 shows aa schematic diagram of the improved constant-head test device proposed in this
paper. Its main
paper. Its
Its main difference
difference from
from the
the conventional constant-head permeability
conventional constant-head
constant-head permeability device
device is is that
that inin the
the
paper. main difference from the conventional permeability device is that in the
modified device,
modified device, a graduated
device, aa graduated U-tube
graduated U-tube with
U-tube with a 4 mm
with aa 44 mm inner
mm inner diameter
inner diameter was
diameter was equipped
was equipped with
equipped with a video
with aa video device
video device
deviceto
modified
achieve accurate
to achieve
achieve accurate measurement
accurate measurement of
measurement of insignificant
of insignificantamounts
insignificant amounts of
amounts ofwater flow
of water in
water flow a
flow in short period
in aa short
short periodof
period oftime. On
of time. the
time. On
On
to
other
the hand,
other test test
hand, head measurement
head measurement waswasrealized by combining
realized by the constant-head
combining the constant-head control device
control and
device
the other hand, test head measurement was realized by combining the constant-head control device
calibration
and plate,
calibration which
plate, reduces
which test time
reduces and test
test time
time and error, and greatly
test error,
error, improves
and greatly
greatly test efficiency.
improves test efficiency.
efficiency.
and calibration plate, which reduces test and test and improves test
Constant-head
Constant-head Screw truss
truss
Screw Graduated
control device
control device Graduated
U-tube
U-tube
Permeameter
Permeameter Permeable
Permeable
cover
cover plate
plate

Video device
Video device

Filter
Filter
paper
paper
Soil sample
Soil sample
Ring
Ring
knife
knife Fixing frame
Fixing frame

Calibration
Calibration Permeameter
Permeameter Permeameter Rubber seals
seals Valve
plate Permeameter
plate sleeve Rubber Valve
base
base sleeve

Figure 2.
Figure Schematicdiagram
2. Schematic
Schematic diagramof
diagram ofthe
of the improved
the improved constant-head
constant-head permeability
permeability test
test device.
device.

2.2. Permeability
2.2. Permeability Test
Test and Data
and Data Processing
Data Processing
Processing
2.2. Permeability Test and
The clay
The clay sample
sample used
usedininthis
thispaper
paperwas
wascollected
collectedfrom
from certain
a certain place
in in
thethe Henan Province,
The clay sample used in this paper was collected from aa certain place
place in the Henan
Henan Province,
Province, at
at
at
a a
2.52.5
m m depth.
depth. TheThe basic
basic physical
physical parameters
parameters of of the
the soil
soil are
are summarized
summarized in
in Table
Table 1.
1. Since
Since natural
natural
a 2.5 m depth. The basic physical parameters of the soil are summarized in Table 1. Since natural
undisturbed soil
undisturbed soil contains
contains more
more impurities
impurities and
and it
it was
was not
not uniform,
uniform, saturated
saturated remolded
remolded clay
clay was
was used
used
undisturbed soil contains more impurities and it was not uniform, saturated remolded clay was used
in this
in this paper.
this paper.
paper.
in
Table 1. Physical properties of soils.
Table 1.
Table 1. Physical
Physical properties
properties of
of soils.
soils.
Gs
Specific Gravity, Specific Limit, wLimit,
Liquid Liquid L (%) Limit,wP (%)
Plastic Limit,
Plastic Plastic Index, IP (%)
Plastic Index,
Specific Liquid Limit, Plastic Limit, Plastic Index,
2.71 Gravity, Gss
Gravity, G 39.4
w L (%)
wL (%) w 22.3
P (%)
wP (%) IIPP (%)
(%) 17.1
2.71
2.71 39.4
39.4 22.3
22.3 17.1
17.1
First, six permeability samples with different void ratios were prepared using a self-made saturated
First,
remolded six permeability
First, clay
six permeability samples
sampler [37],samples
accordingwith
with different
different
to the void
void
Standard ratios
forratios were
were
Soil Test prepared
prepared
Method using aa self-made
using self-made
(GB/T50123-1999) [36].
saturated
saturated remolded
The heightremolded clay sampler
clay sampler
and diameter [37], according
[37], according
of the sample to the
were 4 andto6.18 Standard
the cm,
Standard for Soil Test
for Soiland
respectively, TesttheyMethod
Method (GB/T50123-
were (GB/T50123-
labeled SL-1
1999)
1999) [36].
[36].
to SL-6. The vacuum
The
After height and
height and diameterpermeability
diameter
saturation, of the
of the sample
sampletests
were
were 44 and
were and 6.18 cm,
6.18
carried cm, respectively,
outrespectively,
using the deviceand they
and they were
were
described
labeled SL-1 to SL-6. After vacuum saturation, permeability tests were carried out
labeled SL-1 to SL-6. After vacuum saturation, permeability tests were carried out using the device using the device
 permeameter sleeve, the excess Vaseline was wiped out and upper and lower filter papers were
removed. Finally, permeameter sleeve was installed on the base and tighten the screw was tightened.
(2) Constant head control device was adjusted and excess air inside the permeameter was discharged.
Then, liquid levels were adjusted to calibration lines in the graduated U-tubes. (3) The left valve on
Water
the 2019, 11, 1385
bottom of the permeameter was opened and the video device on the right side was turned4 of 16 it
on.
was left until liquid level in the U-tube reached the next calibration line and the test was
instopped(where
Figure 2. The one scale
details of was 0.034procedures
the test mL, thereforewere theastest time was
follows: shorter).
(1) Coat Then,
a layer of the constant
Vaseline on head
the
control device was adjusted using the same method and the corresponding
inner wall of the permeameter sleeve, and attach a filter paper to the upper and lower surfaces seepage velocities under
of
different constant pressure heads were recorded. Finally, v–i curves
the saturated sample with cutting ring (to avoid being covered by Vaseline when the sample was were obtained by connecting
these experimental points.
installed, which could affect the permeability of the sample). Then, the sample was loaded into
Using the nonlinear
the permeameter sleeve, thefitting
excesstoolbox
Vaselineinwas Origin
wiped software
out andand Hansbo’s
upper and lower [3] filter
flow papers
mathematical
were
equations shown in Equation (1), permeability test data were segmentally fitted as shown in Figure
removed. Finally, permeameter sleeve was installed on the base and tighten the screw was tightened.
(2)3.Constant
Test resultsheadwere compared
control devicewith those obtained
was adjusted and excessfromair Hansbo’s flow
inside the fitting. The was
permeameter basicdischarged.
parameters
of soil sample and Hansbo’s flow are summarized in Table 2.
Then, liquid levels were adjusted to calibration lines in the graduated U-tubes. (3) The left valve
on the bottom of the permeameter was opened ci m and, ithe  ivideo device on the right side was turned
v U-tube

1
(1)
on. it was left until liquid level in the
 k i  i 
reached
, i  i
the next calibration line and the test was
stopped(where one scale was 0.034 mL, therefore the test time was shorter). Then, the constant head
0 1

control
wheredevice
i0 = i1(mwas adjusted
− 1)/m, using
c = k/(mi the same method and the corresponding seepage velocities under
1m−1); c and k are the coefficient of permeability of the exponential flow
different constant pressure heads
at low gradients and the linear flow at were recorded. Finally, v–i
high gradients, curves were
respectively; obtained
m is by connecting
the exponent these
of exponential
experimental points.
flow at low gradients; i0 is the initial the hydraulic gradient of the linear percolation section; and i1 is
the threshold nonlinear
Using the hydraulic fitting
gradient toolbox
for thein linear
Originrelationship.
software andWhen Hansbo’s
i0 = 0[3]or flow
m = mathematical
l, Equation (1)
equations
degeneratesshownintoinDarcy’s
Equation flow(1),law.
permeability test data were segmentally fitted as shown in Figure 3.
Test results were compared
The developed fitting with
method thosewasobtained from
as follows: (1)Hansbo’s
If there wereflownfitting. The basic
experimental parameters
points, of
first Origin
soil sample and Hansbo’s flow are summarized in Table 2.
software was used for non-linearly fitting of 0 to g (where 0 < g < n) experimental points based on
v  ci m to obtain the values of c and (m. (2)m Equation v  cmi1m 1 i  i1  m  1 m  was obtained by
ci , i < i1
substituting c and m into formula vv= k  i ki(0i−, then (1)
i0 ), theig≥toi1 n experimental points were non-linearly
fitted through equation v  cmi1m 1 i  i1  m  1 m  by Origin software, and the value of i1 was obtained.
where i0 = i1 (m − 1)/m, c = k/(mi1 m−1 ); c and k are the coefficient of permeability of the exponential flow
(3) By substituting the values of c, m, and i1 into the Equation (1), the calculation points of Hansbo’s
at low gradients and the linear flow at high gradients, respectively; m is the exponent of exponential
flow mathematical model were obtained, and the correlation between experimental and the
flow at low gradients; i0 is the initial the hydraulic gradient of the linear percolation section; and i1
calculation values was compared to obtain maximum R2. Otherwise, the above methods were
is the threshold hydraulic gradient for the linear relationship. When i0 = 0 or m = l, Equation (1)
repeated at 0 to g + 1 and g + 1 to n experimental points.
degenerates into Darcy’s flow law.

7
Experimental 6
6 Hansbos flow fitting 5
5
4
v (10-5cm/s)

4 3
3 2

1 1

0
0 10 20 30 40 50 60 70 80
i
Figure 3. Comparison of experimental results and those obtained from Hansbo’s flow fitting.
Figure 3. Comparison of experimental results and those obtained from Hansbo’s flow fitting.
Table 2. Hansbo’s flow fitting parameters.

Soil Sample Number e0 k (cm/s) i0 m R2

SL-1 1.26 6.73 ×10−8 22.20 1.45 0.985
SL-2 1.29 1.28 × 10−7 23.54 1.58 0.978
SL-3 1.36 2.31 × 10−7 27.81 1.63 0.964
SL-4 1.44 6.85 × 10−7 31.25 1.81 0.972
SL-5 1.47 8.95 × 10−7 24.37 1.64 0.967
SL-6 1.50 1.29 × 10−6 30.41 1.83 0.956
where e0 is the initial void ratio , k is the permeability coefficient of soil samples, R2 is the correlation coefficient.
Water 2019, 11, 1385 5 of 16

Water 2019, 11, x FOR PEER REVIEW 5 of 16

The developed fitting method was as follows: (1) If there were n experimental points, first Origin
software was used for non-linearly fitting of 0 to g (where 0 < g < n) experimental points based on
Table 2. Hansbo’s flow fitting parameters.
v = cim to obtain the values of c and m. (2) Equation v = cmim−1 1
[i − i1 (m − 1)/m] was obtained by
substituting Soil Sample
c and m into formula v = k(i − i0 ), then the g to n experimental points were non-linearly
e0 k (cm/s) i0 m R2
fitted through equation v = cmim−1
Number
1
[i − i1 (m − 1)/m] by Origin software, and the value of i1 was
obtained. (3) By SL-1substituting the 1.26 values of c, m, 6.73
and × 10 −8
i1 into 22.20 (1), the
the Equation 1.45calculation0.985
points of
Hansbo’s flowSL-2 mathematical model 1.29 were obtained,
1.28 × 10 −7
and 23.54
the correlation 1.58 experimental
between 0.978 and
the calculationSL-3 values was compared 1.36 to obtain 2.31 × 10−7
maximum R2 . Otherwise,
27.81 the1.63 0.964 were
above methods
repeated at 0 to g + 1 and g + 1 1.44
SL-4 to n experimental 6.85 ×points.
10 −7
where 31.25 1.81 void ratio,
e0 is the initial 0.972k is the
SL-5
permeability coefficient 1.47 R2 is the
of soil samples, 8.95correlation
× 10−7 24.37
coefficient. 1.64 0.967
SL-6Table 3 show 1.50
Figure 3 and that the results 1.29obtained
× 10 −6
from 30.41
Hansbo’s flow 1.83equation 0.956
fitting and
test results had a high degree of fitting with a correlation coefficient R2 above 0.95. Hansbo’s flow
where e0 is
parameter mthe
wasinitial void ratio
also within , k is thesuggested
the ranges permeability coefficient
by Hansbo [3], of soil samples,
Dubin et al. [8], R2 is the correlation
and Sun et al. [11]
coefficient.
(between 1 and 2). However, due to the lack of relevant experimental data, there is no unified statement
on the range of i0 in current studies, Hansbo [3] stated that i0 ranged from 1 to 4; Dubin et al. [8]
reportedFigure 3 and
it was Table 33 show
between and 25, thatandtheSun
results obtained
et al. from Hansbo’s
[11] reported flow equation
it to be between 0.5 and fitting andthe
14, and test
results had a high degree of fitting with a correlation coefficient R 2 above 0.95. Hansbo’s flow
results obtained in this paper were basically within the above range, which proves the applicability of
parameter
Hansbo’s m was
flow modelalso
towithin the ranges suggested by Hansbo [3], Dubin et al. [8], and Sun et al. [11]
clay samples.
(between 1 and 2). However, due to the lack of relevant experimental data, there is no unified
statement on the range ofBasic
Table 3. i0 in properties
current studies, Hansbo [3]
of the rheological stated thattest
consolidation i0 ranged
sample.from 1 to 4; Dubin et
al. [8] reported it was between 3 and 25, and Sun et al. [11] reported it to be between 0.5 and 14, and
Sample Number Specific Gravity, Gs Bulk Density, ρ (g/cm3 ) Water Content, w (%) Void Ratio, e0
the results obtained in this paper were basically within the above range, which proves the
LB-1 2.71 1.738 47.77 1.30
applicability
LB-2 of Hansbo’s flow2.71 model to clay samples. 1.864 34.14 0.95
Taylor [38] proposed a linear relationship between the logarithm of k and void ratio e as shown
in Equation (2):proposed a linear relationship between the logarithm of k and void ratio e as shown in
Taylor [38]
Equation (2): log k  log k0  e0  e Ck (2)
log k = log k0 − (e0 − e)/Ck (2)
where k0 is the initial permeability coefficient corresponding to a initial void ratio e0 and Ck is the
where k0 is the initial permeability coefficient corresponding to a initial void ratio e0 and Ck is the
permeability index.
permeability index.
In order to prove the applicability of Equation (2) to the clay samples studied in this paper, void
In order to prove the applicability of Equation (2) to the clay samples studied in this paper,
ratio e and permeability coefficient k listed in Table 2 were nonlinearly fitted according to Equation
void ratio e and permeability coefficient k listed in Table 2 were nonlinearly fitted according to
(2), and the obtained results are shown in Figure 4. The experimental data fit values satisfied the e-
Equation (2), and the obtained results are shown in Figure 4. The experimental data fit values satisfied
logk relationship, and the permeability index Ck = 0.193 was also in the reasonable range of
the e-logk relationship, and the permeability index Ck = 0.193 was also in the reasonable range of
permeation index given by Berry [39] (Ck value between 0.02 and 5.00). Therefore, it was confirmed
permeation index given by Berry [39] (Ck value between 0.02 and 5.00). Therefore, it was confirmed
that Equation (2) is applicable to the clay samples studied in this paper, i.e., logk and the void ratio e
that Equation (2) is applicable to the clay samples studied in this paper, i.e., logk and the void ratio e
were linearly related.
were linearly related.

1.55
Experimental fitting value
1.50 Eq.(2) fitting

1.45
void ratio e

1.40

1.35 e=2.36+0.193log k
1.30 R2=0.992

1.25

110-7 110-6
logk(cm/s)

Figure 4. The relationship between the permeability coefficient k and void ratio e.
Figure 4. The relationship between the permeability coefficient k and void ratio e.

3. Experimental Study on Rheological Consolidation Characteristics of Saturated Clay

saturated clay layer, the one-dimensional rheological consolidation test with upward drainage was
used in this work. The loading and unloading scheme shown in Table 3 were applied based on Yin
et al. [32] and Feng et al. [33]. Detailed test procedures were as follows: (1) the prepared saturated
clay samples LB-1 and LB-2 were loaded into the consolidation cylinder, respectively, as shown in
Figure 5. Filter paper and permeable plate are placed on the top and bottom of the samples. (2)
Water 2019, 11, 1385 6 of 16
Displacement meter and pore pressure sensor were adjusted using the data collection sofware. Then,
loading was started when the displacement sensor gave values lower than 0.1 mm in a day or the
pore
3. pressure dissipated
Experimental Study on more than 95%,
Rheological the next stage
Consolidation of loading or
Characteristics of unloading was carried out
Saturated Clay
according to Table 4. (3) According to the displacement sensor reading, the vertical settlement
3.1. One-Dimensional
deformation Rheological
S of the soil sampleConsolidation Testand then the corresponding void ratio e was obtained
was calculated,
according
In orderto Equation
to improve(3).test accuracy, the TWJ-1 data collector device was introduced based on

log k  log k0  e0  e Ck  
Sun’s [11] permeability-consolidation cross-test device. Displacement and pore water pressure at the
(3)
bottom of the sample were measured by connecting the device to a computer. Figure 5 shows the
schematic
where e0 isdiagram of void
the initial the improved consolidation
ratio corresponding device.
to the initial thickness H0 of the soil sample.

Lever loading
Fixed table
frame
Displacement Dowel steel
meter

Oedometer
Computer Fixing
cover
bolts

Permeable
plate

Soil sample
Data collector Filter paper

Pore pressure Consolidation Valve

sensor Oedometer
cylinder
base

Figure 5.
Figure 5. Schematic
Schematic diagram
diagram of
of the
the improved
improved oedometer.
oedometer.

Table 3. Basic
First, two consolidated properties
samples of the rheological
were prepared consolidation
as described test2.2
in Section sample.
with sample height and
diameter of 2 and 6.18 cm, respectively. The basic physical parameters of the sample are summarized
Sample Specific Gravity, Bulk Density, ρ Water Content, w Void Ratio,
in Table 2. In order to simulate the rheological consolidation process of a natural saturated clay
Number Gs (g/cm3) (%) e0
layer, the one-dimensional rheological consolidation test with upward drainage was used in this
LB-1 2.71 1.738 47.77 1.30
work. The loading and unloading scheme shown in Table 3 were applied based on Yin et al. [32] and
Feng et al. [33].LB-2 2.71
Detailed test procedures 1.864 (1) the prepared
were as follows: 34.14saturated clay
0.95
samples LB-1
and LB-2 were loaded into the consolidation cylinder, respectively, as shown in Figure 5. Filter paper
and permeable plate are placed Table 4. Loading
on the top andprocedures
bottom of intheoedometer
samples. tests.
(2) Displacement meter and pore
pressure sensor were adjusted using the data collection sofware.
Order LoadingThen,(kPa) loading was started when the
displacement sensor gave First
values lower
loading than 0.1 mm in a day or the pore pressure dissipated more than
0-25-50-100-200
95%, the next stage of loading or unloading was carried out200-100-50
First Unloading according to Table 4. (3) According to the
displacement sensor reading, the vertical
Second Loading settlement deformation S of the soil sample was calculated,
50-100-200-400
and then the corresponding void ratio
Second Unloading e was obtained according to
400-200-100-50Equation (3).

log k = log k0 − (e0 − e)/Ck (3)

where e0 is the initial void ratio corresponding to the initial thickness H0 of the soil sample.

Table 4. Loading procedures in oedometer tests.

Order Loading (kPa)

First loading 0-25-50-100-200
First Unloading 200-100-50
Second Loading 50-100-200-400
Second Unloading 400-200-100-50
Third Loading 50-100-200-400-800
Third Unloading 800-400-200-100-50
Fourth Loading 50-100-200-400-800-1600
 Water 2019, 11, x FOR PEER REVIEW 7 of 16

Third Loading 50-100-200-400-800

Third Unloading 800-400-200-100-50
Water 2019, 11, 1385 Fourth Loading 50-100-200-400-800-1600 7 of 16

3.2. One-Dimensional Rheological Consolidation Test Data Processing

3.2. One-Dimensional Rheological Consolidation Test Data Processing
Figure 6 shows the relationship between void ratio and time-dependent loading under first-
Figure 6 shows the relationship between void ratio and time-dependent loading under first-stage
stage loading conditions. As can be seen in the figure, the obtained results were consistent with those
loading conditions. As can be seen in the figure, the obtained results were consistent with those
obtained by Yin et al. [32] and Feng et al. [33] for Hong Kong clay samples; the e-logt curve had an
obtained by Yin et al. [32] and Feng et al. [33] for Hong Kong clay samples; the e-logt curve had an
inverse “S” shape and its tail section was elongated with the increase of vertical load. At the same
inverse “S” shape and its tail section was elongated with the increase of vertical load. At the same
time, the e-logt curve depended to some extent on the initial void ratio of the sample. As can be seen
time, the e-logt curve depended to some extent on the initial void ratio of the sample. As can be seen
from Figure 6, by the decrease of initial void ratio, the tail section of e-logt curve became steeper and
from Figure 6, by the decrease of initial void ratio, the tail section of e-logt curve became steeper and
longer times were required for deformation stabilization. In addition, the slope of the tail section of
longer times were required for deformation stabilization. In addition, the slope of the tail section of
the e-logt curve reflected the secondary consolidation coefficient. Table 5 summarizes the secondary
the e-logt curve reflected the secondary consolidation coefficient. Table 5 summarizes the secondary
consolidation coefficient values obtained by fitting the tail section of the e-logt curve under different
consolidation coefficient values obtained by fitting the tail section of the e-logt curve under different
loads. These results are similar to those obtained by Nash et al. [28] and Chen et al. [29], where the
loads. These results are similar to those obtained by Nash et al. [28] and Chen et al. [29], where the
secondary consolidation coefficient was first increased and then decreased with the increase of
secondary consolidation coefficient was first increased and then decreased with the increase of external
external load. Smaller initial void ratios of the sample gave greater consolidation coefficients. This
load. Smaller initial void ratios of the sample gave greater consolidation coefficients. This could be
could be attributed to the stability of clay structure at the particle’s contact surface [40].
attributed to the stability of clay structure at the particle’s contact surface [40].

(a) Sample LB-1 (b) Sample LB-2

1.3 25kPa 1.0 25kPa
1.2 50kPa 50kPa
0.9 100kPa
1.1 100kPa 200kPa
1.0
void ratio e
void ratio e

200kPa 0.8
400kPa
0.9
0.8 400kPa 0.7
800kPa
0.7 800kPa
0.6
0.6 1600kPa 1600kPa
0.5 0.5
10-1 100 101 102 103 104 10-1 100 101 102 103 104
logt (min) logt (min)

Figure 6. e-logt curves of samples LB-1 and LB-2 under primary loading condition.
Figure 6. e-logt curves of samples LB-1 and LB-2 under primary loading condition.
Table 5. Fitting results of secondary consolidation coefficient Cα under primary loading conditions.
Table 5. Fitting results of secondary consolidation coefficient Cα under primary loading conditions.
Loading (kPa)
Soil Sample Number
Soil Sample 25 50 100 Loading200 (kPa) 400 800 1600
Number
LB-1 25 0.0040 50
0.0050 100
0.0120 200
0.0110 400
0.0091 800
0.0085 1600
0.0080
LB-2
LB-1 0.00400.0009 0.0015
0.0050 0.0030
0.0120 0.0092
0.0110 0.0135
0.0091 0.0208
0.0085 0.0134
0.0080
LB-2 0.0009 0.0015 0.0030 0.0092 0.0135 0.0208 0.0134
Figure 7 shows the e-log p curves of samples LB-1 and LB-2. The curve was fitted according to the
StandardFigure 7 shows
for Soil the e-log(GB/T50123-1999)
Test Method p curves of samples LB-1 and
to obtain the LB-2.
valuesThe
of C curve
c andwas fittedit according
Cs , and was foundto
thethe
that Standard
averagefor Soil of
value Test
Cc Method (GB/T50123-1999)
for the two samples was 0.42.[Error!
TheReference
values of C c /Cs for
source not found.]LB-1
samples to obtain
and
the were
LB-2 values of Cc and
4.32–4.79 andCs,5.20–7.65,
and it was found that The
respectively. the average value of
corresponding CcC/Cc sfor the of
value two samplesclay
saturated waswas0.42.
The values
between of C7.65,
4.32 and c/Cs which
for samples
was alsoLB-1 and LB-2
obtained by Li were 4.32–4.79
[41] based on theandGDS 5.20–7.65,
consolidationrespectively.
instrument The
who found Cc /Cs C
corresponding c/Cs value
value of 3.65oftosaturated clay
7.88 for clay was between
samples collected4.32 and
from the7.65, whichin
Xiaoshan was also obtained
Zhejiang province. by
Li [41] based on the GDS consolidation instrument who found C /C value
Based on the above analysis, the effectiveness of the improved oedometer proposed in this paper
c s of 3.65 to 7.88 for clay
samples
was collected from the Xiaoshan in Zhejiang province. Based on the above analysis, the
verified.
effectiveness of the improved oedometer proposed in this paper was verified.
Water
Water2019, 11,11,
2019, 1385
x FOR PEER REVIEW 8 of 16 16
8 of

(a) Sample LB-1 (b) Sample LB-2

1.0

1.2
0.9
CC=0.39, R2=0.996
2
CS=0.103, R =0.998 CS=0.051, R2=0.889
1.0 0.8

void ratio e
void ratio e

CC=0.45, R2=0.998
CS=0.070, R2=0.835
0.7
0.8 CS=0.104, R2=0.962
CS=0.075, R2=0.896
2
CS=0.094, R =0.925 0.6
0.6
0.5

100 1000 100 1000

log p (kPa) log p (kPa)

Figure 7. e-log p (vertical load) curves for samples LB-1 and LB-2.
Figure 7. e-log p (vertical load) curves for samples LB-1 and LB-2.
4. Derivation of Governing Equations
4. Derivation of Governing Equations
4.1. UH Model Considering Time Effect
4.1. UH Model Considering Time Effect
The above tests show that saturated soft clays have obvious rheological properties. In order to
use thisThe above in
property tests show that
predicting saturated soft
consolidation clays have
behavior, obvious
a unified rheological
hardening (UH)properties. In order to
model considering
useeffect
time this property in predicting
was introduced [34,35]consolidation
to describe thebehavior, a unified
stress–strain hardening
relationship (UH)
of soil. model
These considering
relationships
time effect was
are expressed as: introduced [34,35] to describe the stress–strain relationship of soil. These relationships
are expressed as:  dεev + dεsp tp
dσ0 ≥ 0

v + dεv ,
dεv =  (4)

tp
 dεee + dεsp , dσ0 < 0
d v  d vv  d v , d   0
tp

d v   ve (4)
dand
v  dvv  vare
sp tp
where dεv is the vertical strain increment dεetp, dεv and, ddε 0 elastic, plastic and time-dependent
viscous strain increments caused by stress, respectively, and can be expressed as:
where d v is the vertical strain increment and d ve , d vsp and d vtp are elastic, plastic and time-
0
dependent viscous strain incrementsdεev caused
= Cs dσby/[stress, + e0 )σ0 ]
ln 10(1respectively, and can be expressed as: (5)

s d / cln 10s1  e0  
d vedεspC=  C − C M4 dσ0 (5)
v (6)
ln 10(1 + e0 ) M4 σ0
f
tp C  C M d  4

v = 
dεd sp C dt/c[ln 10
s (1 + e )(t + t )] (7)
ln 101  e0  M f4  
α 0 α 0 (6)
v

where Cc is the compressed index, Cs is the swelling index, Cα is the secondary consolidation coefficient,

d vtp  Cα dt /ln 101  e0 tα  t0 

tα is the aging time, t0 is the unit time which can be 1 s, 1 min, 1 h, etc. According to actual situations,
(7)
M is the stress ratio at critical state and Mf is the potential failure stress ratio, which can be expressed as:
where Cc is the compressed index, Cs is the swelling index, C α is the secondary consolidation
M = 6 sin ϕ/(3 − sin ϕ) (8)
coefficient, t α is the aging time, t 0 is the unit time which can be 1 s, 1 min, 1 h, etc. According to
q 
actual situations, M is the stressMfratio
= 6 at χ/Rcritical
(1 +state
χ/Rand M f is the potential failure stress ratio,
) − χ/R (9)
which can be expressed as:
(ta + t0 )/t0 = R−α (10)
M  6 sin  3  sin   (8)
p
σ0 ln 10·(1 + e0 )εv
!
R= exp −
M f  6p0 c  / R1   / C
Rc −
Cs/ R  (11)
(9) 
where ϕ is the internal friction angle of soil, χ = M2 /12(3 − M), α = (Cc − Cs )/Cα , and R is a parameter
reflecting the degree of over-consolidation. ta When  R effect
t0 t0time α
 
is not considered, the initial value R0(10)
is
p
reciprocal of the over-consolidation ratio (OCR). εv is the plastic strain produced by stress and time,

R 
  of rebound
p0 c is the vertical stress at the intersection
exp  
1  einstantaneous
 ln 10  and  p 
0 v
compression lines, similar to
pre-consolidation pressure. p  C C  (11)
c  c s 
where  is the internal friction angle of soil,   M 2 123  M  ,   Cc  Cs  Cα , and R is a
parameter reflecting the degree of over-consolidation. When time effect is not considered, the initial
Water 2019, 11, 1385 9 of 16

4.2. One-Dimensional Rheological Consolidation Equation of UH Model

First, the principle of effective stress is obtained as:

σ0 = σ0 0 + p0 − u (12)

where σ0 0 is the initial effective stress, p0 is the external loads, and u is the excess pore water pressure.
For both sides of Equation (12), partial derivatives could be obtained at the same time as:

∂σ0 ∂u
=− (13)
∂z ∂z
Based on continuous conditions of permeability we had:

∂q ∂εv
=− (14)
∂z ∂t
where q is the change of water quantity within the unit cell during an infinitesimal time period of dt
can be expressed as q = vA, A is the area of water flowing through the unit cell.
Substituting Equations (1), (2), (4), and (13) into Equation (14) gives:

∂ ∂σ0 ∂σ0
!
ψ =β +η (15)
∂z ∂z ∂t
 exp[−ln 10ε (1+e )/C ]  0 m−1
v 0 k  ∂σ 
 · ∂z  , |i| ≤ i1
where η = γw Cα /[ln 10k0 (1 + e0 )(tα + t0 )], ψ = m−1

m ( γ i ) ,

 w 1
 exp[−ln 10ε (1 + e )/C ], |i| > i1

v 0 k
[CS +(CC −CS )M4 /M4f ]γw

dσ0 ≥ 0


(1+e0 )k0 σ0 ln 10
,
β=


CS γw
, dσ0 < 0


(1+e0 )k0 σ ln 10
0

However, the initial and boundary conditions of the device shown in Figure 4 are expressed as:

σ0 (z, 0) = σ0 0 , 0 ≤ z ≤ H (initial condition) (16)

σ0 (0, t) = p0 + σ0 0 , t > 0 (permeable boundary) (17)

∂σ0 

= 0, t > 0 (imperious boundary) (18)
∂z z=H


4.3. Discretization of the Governing Equation

Due to the complexity of Equation (15), it was very difficult to obtain its analytical solution.
Therefore, the numerical solutions can be obtained using the finite difference method (FDM),
finite volume method (FVM), or finite element method (FEM). It was found that among these
methods, the FVM was a numerical solution based on FDM, and absorbs the advantages of the
FEM. Some researchers have studied this method [15,42] and their results showed that FVM as more
advantages than FDM and FVM in terms of convergence and accuracy. Therefore, FVM was used in
this paper to solve the problem.
To obtain numerical solutions for Equation (15) using FVM, as shown in Figure 8, the spatial
domain 0 ≤ z ≤ H was firstly divided into n thin layers with equal length ∆z. Meanwhile, the time
axis is divided by a series of time increments ∆t. Where bth is the number of spatial nodes, jth is time
discretized point. Then, within any time interval ∆t (tj to tj+1 ), Equation (15) was integrated in the
control volume of bth:
Z t j+1 Z Z t j+1 Z
∂ ∂σ0 ∂σ0
! !
ψ dzdt = β + η dzdt (19)
tj ∂z ∂z tj ∂t
∆z ∆z
 t j 1       t j 1    
tj

 z B
 dzdt  t j   
z D  z 
t
  dzdt

(20)

where the 11,

Water 2019, subscripts
1385 B and D represent the lower and upper boundary points of the control volume,
10 of 16
respectively.

permeable top
1 z

b-1 z
H b z
b+1 z

n z

impervious bottom
Discretizedschematic
Figure8.8.Discretized
Figure schematicofofthe
thesoil
soillayer.
layer.

Equation
For (19)(20),
Equation can the
be rewritten
first-orderas:forward difference is used to approximate the right side of the
equal sign   t andZthe t j+1
central0 difference  is
! used toZ approximate
t j+1 Z
the !left side of the equal sign
∂σ  ∂σ0  ∂σ0
  z . To describe the relationship
ψ ψ
 −between  , =
 dzdt  b ,  b1 , and + ηthe
β time, dzdt
effective stresses at (20)
tj
tj ∂z B ∂z D b 1 tj ∂t
and t j 1 are usually weighted as a combination and then taken ∆z as the average value of the time from

j to t j the
twhere 1 . Then, the timeBintegral
subscripts and D term on the the
represent left side
lower of Equation
and upper (20)boundary
is calculated. Where
points the weight
of the control
of t j is 0 respectively.
volume, and that of j 1 is 1. That is to say, Equation (20) can be written as:
t
For Equation (20), the first-order forward difference is used to approximate the right side of the
 j sign1  ∂σ  j0/∂t
b 1   b
1
 j 1 the j 1  
 central  j 1   bjis
1
  to approximate
 
equal and difference used thej left sidej of
1 the equal sign

∂σ /∂z. To describe
0       
b 1
0    t
0  
0
j 1
  j 1
    z   zt
σ b−1 , σ b , σ b+1 , and time, the effective stresses at t(21)
 the relationship between  j and
b 1 2 b 1 2 b b b b
  z   z  
t j+1 are usually weighted as a combination and then taken as the average value of the time from t j to
t j+1 .Equation
Then, the(21)is time reorganized
integral termas: on the left side of Equation (20) is calculated. Where the weight of t j
is 0 and that of t j+1 is 1. That is to say, Equation (20) can be written as:
     
 b j 1   b j   bj11 2  bj 11   b j 1  bj11 2  b j 1   bj11  t bj 1z 2 bj 1t bj 1
 j+1
(22)
 j+1   j+1 j+1

 σ0
 b+1 − σ b 
0

j+1 

j+1 
 σ b − σ b−1 
0 0  
j+1 0 j+1j j j+1
where b is ψthe number  − ψ
of spatial nodes,  j is ∆t = β point,
time discretized σ0 b )∆z + ηbthe∆z∆t
(σ b xb −represents value of (21)
x
 b+1/2  ∆z b−1/2 ∆z j b
  
variable at z  b 1 2 z and t  j t , xb 1 2 represents the value of x variable at
 

z  bEquation  and
1 2  1 2(21) t  j t , the as:
is reorganized same below.
With
j+1
the above-mentioned
j
h j+1 j+1
methodj+1
(FVM),
j+1
Equations
j+1
(16)–(18)
j+1
i canj+be
1
discretized
j+1
as: j+1
σ0 b = σ0 b + ψb+1/2 (σ0 b+1 − σ0 b ) − ψb−1/2 (σ0 b − σ0 b−1 ) ∆t/(βb ∆z2 ) − ηb ∆t/βb (22)
 b0   0 , b  1, 2,3 n (23)
j
where b is the number of spatial nodes, j is time discretized point, xb represents the value of x variable
j
at z = (b − 1/2)∆z and t = j∆t, xb±1/2 represents the value of x variable at z = (b − 1/2 ± 1/2) and
t = j∆t, the same below.
With the above-mentioned method (FVM), Equations (16)–(18) can be discretized as:

σ0 0b = σ0 0 , b = 1, 2, 3, . . . , n (23)

j+1 j
h j+1 j+1 j+1 j+1 j+1
i j+1 j+1 j+1
σ0 1 = σ0 1 + ψ3/2 (σ0 2 − σ0 1 ) − 2ψ1/2 (σ0 1 − p0 − σ0 0 ) ∆t/(β1 ·∆z2 ) − ∆t·η1 /β1 (24)
j+1 j j+1 j+1 j+1 j+1 j+1 j+1
σ0 n = σ0 n − ψn−1/2 (σ0 n − σ0 n−1 )∆t/(βn ∆z2 ) − ηn ∆t/βn (25)

Thus, Equations (22)–(25) composed a closed system of equations, and the exact solutions for
j+1
unknown variables σ0 b can be obtained using the method of iteration.
Water 2019, 11, 1385 11 of 16

In order to analyze the settlement deformation law of foundation, settlement at time t is

expressed as:
Z H n
X
j+1
St = εv (z, t) dz = ∆z· εv b (26)
0 b=1

where H is the thickness of foundation soil or the soil sample’s initial height, and εv (z, t) is the vertical
strain of foundation or soil sample at any depth z and time t.

5. Verification of the Suitability of UH Model

5.1. Verification of the Experimental Results in this Paper

First, a code is written in the Fortran software according to the above algorithm. The test results
obtained for samples LB-1 and LB-2 are then used as input. Since the test sample’s thickness is small, the
effect of Hansbo’s flow can be neglected [14] and analysis can be performed based on Darcy’s flow; i.e.,
m = 1.0 or i1 = 0. The compression index Cc , swelling index Cs , and secondary consolidation coefficient
Cα are taken according to the above test results. Sorensen et al. [43] summarized some clay experimental
data on clay and provided the relationship between the internal friction angle ϕ and plasticity index IP ,
as shown in Equation (27). Therefore, ϕ = 25◦ can be estimated based on Equation (27) and Table 1
(IP = 17.1), and then M = 0.984 can be obtained by substituting ϕ into Equation (8). Moreover, the initial
permeability coefficient k0 , permeation index Ck , and initial over-consolidation parameter R0 were
selected by trial calculation, as shown in Tables 6 and 7. Numerical simulation results are shown in
Figure 9, which shows that the predicted results from the UH model, considering the time effect, are in
good agreement with the experimental results. However, the predicted results of Terzaghi’s theory are
quite different from the experimental results in the later stage of loading. The above results fully prove
the necessity of considering the soil’s rheological characteristics in predicting foundation settlement in
saturated clay areas, and also proved the UH model’s applicability and the necessity of considering
variable permeability coefficient.
(
43 − 10· log(IP ), (mean)
ϕ= (27)
39 − 11· log(IP ), (lower bound)

Table 6. UH model parameters of sample LB-1.

Loading Interval/(kPa) H/mm e0 CC Ck CS Cα R0 k0 /(m/min)

1.2 ×
50–100 18.99 1.184 0.0120 0.65
10−8
0.70 ×
100–200 18.01 1.070 0.42 0.66 0.100 0.0110 0.92
10−8
0.40 ×
200–400 16.63 0.913 0.0091 0.83
10−8
0.25 ×
400–800 15.44 0.776 0.0085 0.79
10−8
0.12 ×
800–1600 14.29 0.644 0.0080 0.76
10−8

Table 7. UH model parameters of sample LB-2.

Loading Interval/(kPa) H/mm e0 CC Ck CS Cα R0 k0 /(m/min)

50–100 1.988 0.940 0.0030 0.58 1.0 × 10−8
100–200 1.959 0.911 0.0092 0.53 0.85 × 10−8
200–400 1.899 0.852 0.42 0.66 0.065 0.0135 0.65 0.39 × 10−8
400–800 1.789 0.745 0.0190 0.60 0.19 × 10−8
800–1600 1.512 0.622 0.0135 0.66 0.06 × 10−8
 50–100 1.988 0.940 0.0030 0.58 1.0 × 10−8
100–200 1.959 0.911 0.0092 0.53 0.85 × 10−8
200–400 1.899 0.852 0.42 0.66 0.065 0.0135 0.65 0.39 × 10−8
400–800 1.789 0.745 0.0190 0.60 0.19 × 10−8
Water 2019, 11, 1385
800–1600 1.512 0.622 0.0135 0.66 0.06 × 1012−8 of 16

0.0 0.0

0.2
0.2
0.4
0.4 0.6
S/mm

S/mm
0.6 0.8
Experimental points 1.0 Experimental points
0.8 Predicted(CUHM) Predicted(CUHM)
Predicted(VUHM) 1.2 Predicted(VUHM)
Predicted(Terzaghi′s theory) Predicted(Terzaghi′s theory)
1.0 1.4
0.1 1 10 100 1000 10000
10,000 0.1 1 10 100 1000 10000
10,000
t/min t/min
(a) Sample LB-1(50-100kPa) (b) Sample LB-1(100-200kPa)

0.0 0.0

0.2 0.2
0.4
0.4
0.6
S/mm

S/mm
0.6
0.8
Experimental points 0.8 Experimental points
1.0
Predicted(CUHM) Predicted(CUHM)
1.2 Predicted(VUHM) 1.0 Predicted(VUHM)
Predicted(Terzaghi′s theory) Predicted(Terzaghi′s theory)
1.4 1.2
0.1 1 10 100 1000 10000
10,000 0.1 1 10 100 1000 10000
10,000
t/min t/min
(c) Sample LB-1(200-400kPa) (d) Sample LB-1(400-800kPa)

0.0 0.0

0.2
0.1
0.4
S/mm

S/mm

0.6 0.2

0.8 Experimental points Experimental points

Predicted(CUHM) 0.3 Predicted(CUHM)
1.0 Predicted(VUHM) Predicted(VUHM)
Predicted(Terzaghi′s theory) Predicted(Terzaghi′s theory)
1.2 0.4
0.1 1 10 100 1000 10000
10,000 0.1 1 10 100 1000 10000
10,000
t/min t/min
(e) Sample LB-1(800-1600kPa) (f) Sample LB-2(50-100kPa)

Figure 9. Cont.
Water 2019, 11,
Water 2019, 11, 1385
x FOR PEER REVIEW 13
13 of
of 16
16
Water 2019, 11, x FOR PEER REVIEW 13 of 16

0.0 0.0
0.0 0.0
0.1 0.2
0.1 0.2
0.2 0.4
0.2 0.4
0.3 0.6
S/mm

S/mm
0.3 0.6
S/mm

S/mm
0.4 0.8
0.4 Experimental points 0.8 Experimental points
0.5 1.0
Predicted(CUHM)
Experimental points Predicted(CUHM)
Experimental points
0.5 1.0
0.6 Predicted(VUHM)
Predicted(CUHM) 1.2 Predicted(VUHM)
Predicted(CUHM)
0.6 Predicted(Terzaghis
Predicted(VUHM) theory) 1.2 Predicted(Terzaghis
Predicted(VUHM) theory)
0.7 1.4
0.1 Predicted(Terzaghis
1 10 theory)
100 1000 10000
10,000 0.1 Predicted(Terzaghis
1 10 theory)
100 1000 10000
10,000
0.7 1.4
0.1 1 10 t/min 100 1000 10000
10,000 0.1 1 10 t/min 100 1000 10000
10,000
(g) Sample LB-2(100-200kPa)
t/min (h) Sample LB-2(200-400kPa)
t/min
(g) Sample LB-2(100-200kPa) (h) Sample LB-2(200-400kPa)
0.0 0.0
0.0 0.0
0.2 0.2
0.2 0.2
0.4 0.4
0.4 0.4
0.6 0.6
S/mm

S/mm
0.6 0.6
S/mm

S/mm
0.8 0.8
0.8 Experimental points 0.8 Experimental points
1.0 1.0
Predicted(CUHM)
Experimental points Predicted(CUHM)
Experimental points
1.0 1.0
1.2 Predicted(VUHM)
Predicted(CUHM) 1.2 Predicted(VUHM)
Predicted(CUHM)
1.2 Predicted(Terzaghis
Predicted(VUHM) theory) 1.2 Predicted(Terzaghis
Predicted(VUHM) theory)
1.4 1.4
0.1 Predicted(Terzaghis
1 10 theory)
100 1000 10000
10,000 0.1 Predicted(Terzaghis
1 10 theory)
100 1000 10000
10,000
1.4 1.4
0.1 1 10 t/min 100 1000 10000
10,000 0.1 1 10 t/min 100 1000 10000
10,000
(i) Sample LB-2(400-800kPa)
t/min (j) Sample LB-2(800-1600kPa)
t/min
(i) Sample LB-2(400-800kPa) (j) Sample LB-2(800-1600kPa)
Figure 9. Comparison of theoretical and experimental settlement values for samples LB-1and LB-2.
9.unified
Figure 9.
CUHM: Comparison of theoretical
hardening model withand experimental
constant settlement
settlement
permeability values for
for samples
values VUHM:
coefficient, LB-1and
unified LB-2.
hardening
CUHM: unified hardening model with constant
model with variable permeability coefficient. permeability coefficient, VUHM: unified hardening
coefficient.
model with variable permeability coefficient.
5.2. Verification
5.2. Verification of of Experimental Results obtained
Experimental Results obtained by
by Li's
Li's
5.2. Verification of Experimental Results obtained by Li's
Li
Li [41]
[41] conducted
conducted one-dimensional
one-dimensional consolidation
consolidation tests
tests on
on clay
clay samples
samples collected
collected from
from Xiaoshan,
Xiaoshan,
Li [41]
Zhejiang conducted
province one-dimensional
using an advanced consolidation
GDS consolidationtestsinstrument,
on clay samples collected
introduced the from Xiaoshan,
three-element
Zhejiang province using an advanced GDS consolidation instrument, introduced the three-element
Zhejiang
model province
(TEM) and using an advanced
four-element model GDS consolidation
(FEM) beeninstrument,
(can been seen Figure
Figureintroduced
into the
10) into three-element
one-dimensional
model (TEM) and four-element model (FEM) (can seen 10) one-dimensional
model (TEM) and four-element model (FEM) (can been seen Figure
consolidation analysis, and verified the obtained results by comparing them to the experimental 10) into one-dimensional
consolidation analysis, and verified the obtained results by comparing them to the experimental results.
consolidation
results. Hu et al. analysis, and
[35] used theverified
test datatheof obtained
samples 3results by comparing
and 4 under them of
loading stages to800–1,600
the experimental
kPa [41]
Hu et al. [35] used the test data of samples 3 and 4 under loading stages of 800–1,600 kPa [41] to verify
results. Hu et al. [35] used the test data of samples 3 and 4 under loading stages
to verify the UH model’s applicability, but did not consider the permeability coefficient’s influence.of 800–1,600 kPa [41]
the UH model’s applicability, but did not consider the permeability coefficient’s influence. The same
to verify the UH model’s applicability, but did not consider the permeability
The same UH model parameters as those listed in [35] were used in this work as H = 14.8 mm, e0 = coefficient’s influence.
UH model parameters as those listed in [35] were used in this work as H = 14.8 mm, e0 = 0.82, CC = 0.356,
The same
0.82, UH model
CC = 0.356, parameters
Ck = 0.528, as those
CS = 0.073, Cα =listed
0.011,ink0[35]
= 2.0were
× 10used in M
−11 m/s, this= work
0.984, as
andHR = 014.8 mm,The
= 0.575. e0 =
Ck = 0.528, CS = 0.073, Cα = 0.011, k0 = 2.0 × 10−11 m/s, M = 0.984, and R0 = 0.575. The comparison of
0.82, C C = 0.356, Ck = 0.528, CS = 0.073, Cα = 0.011, k0 = 2.0 × 10−11 m/s, M = 0.984, and R0 = 0.575. The
comparison of theoretical and experimental values based on the proposed model is shown with
theoretical and experimental values based on the proposed model is shown with Figure 11. It can be seen
comparison
Figure 11. Itofcan theoretical and the
be seen that experimental values from
results obtained basedthe on UHthe proposed model is the
model considering shown with
variable
that the results obtained from the UH model considering the variable permeability coefficient are closer
Figure 11. It can be seen that the results obtained from the UH model
permeability coefficient are closer to experimental values compared to those from the component considering the variable
to experimental
permeability values compared to those from the component model, which fullyfrom
confirms superiority
model, whichcoefficient are closer
fully confirms to experimental
superiority of the UH values
modelcompared
and the to those
necessity the component
of considering the
of the
model, UH model
which and
fully the necessity
confirms of considering
superiority of the the
UH permeability
model and coefficient
the in
necessity similar
of studies.
considering the
permeability coefficient in similar studies.
permeability coefficient in similar E1 studies. E1
E1 E1
E0 E0 0
 E0   E0 0

 1   1 
1 1
(a) (b)
(a) (b)
Figure 10.
10. Schematic diagram
Schematic of rheological
diagram modelmodel
of rheological (a): three-element rheological
(a): three-element model, (b)model,
rheological Four-
Figure
element 10. Schematic
rheological diagrammodel.
model. of rheological
where Emodel (a): three-element rheological model, (b) Four-
(b) Four-element rheological 0 and E1 are the moduli of the independent spring
element
and rheological
the spring model.
of the Kelvin–Voigt body, respectively; η0 and η1 are the viscosity coefficient of the
independent dashpot and the dashpot of the Kelvin–Voigt body, respectively.
Water 2019, 11, x FOR PEER REVIEW 14 of 16

where E0 and E1 are the moduli of the independent spring and the spring of the Kelvin–Voigt body,
respectively; η0 and η1 are the viscosity coefficient of the independent dashpot and the dashpot of the
Water 2019, 11, 1385 14 of 16
Kelvin–Voigt body, respectively.

0.0

0.2

S/mm
0.4 Experimental data [41]
Predicted (TEM) [41]
Predicted (FEM) [41]
0.6 Predicted (CUHM)
Predicted (VUHM)
Predicted (Terzaghis theory)
0.8
1 10 100 1000 10000
10,000 100000
100,000
t/s

Figure 11. Comparison of prediction values and experimental results of the settlement theory under
Figure 11. Comparison of prediction values and experimental results of the settlement theory under
different models.
different models.
6. Conclusions
6. Conclusions
In this paper, we used an improved constant-head permeameter to measure the permeability
In thisofpaper,
coefficient we usedsoil
fine-grained anrapidly
improved andconstant-head
accurately at permeameter to measure
insignificant amounts of the
waterpermeability
flow rate.
coefficient
The improved of fine-grained
oedometer soil rapidly
is used and accurately
to verify at insignificant
the significant amounts
rheological of water
properties flow rate.clay.
of saturated The
improved oedometer is used to verify the significant rheological properties of saturated
Then the one-dimensional rheological consolidation equation based on Hansbo’s flow and the UH model clay. Then
the one-dimensional
considering rheological
the time effect consolidation
is derived, equation
and the numerical basedof
solution onFVM
Hansbo’s flow
is given. and the
Finally, the UH model
numerical
considering
simulation the time
results effect is derived,
are compared and the numerical
to the experimental solution
results, and of FVMconclusions
the following is given. Finally,
are drawn:the
numerical simulation results are compared to the experimental results, and the following conclusions
(1) Compared to the traditional falling-head permeability experiment, the improved constant-head
are drawn:
experimental device for insignificant amounts of water flow saturated clays requires shorter time
(1) and
Compared
has lesstoexperimental
the traditional falling-head permeability experiment, the improved constant-head
error.
experimental device for insignificant
(2) The water flow of saturated clay pores amounts of water
deviated flow saturated
from Darcy flow andclays requires
Hansbo’s flowshorter time
equations
and has less experimental error.
had good applicability to clay samples used in this experiment.
(2) The water flow of saturated clay pores deviated from Darcy flow and Hansbo’s flow equations
(3) Saturated clay had significant rheological properties. The predicted results of Terzaghi’s theory
had good applicability to clay samples used in this experiment.
were quite different from the experimental results, and the UH constitutive model agreed well
(3) Saturated clay had significant rheological properties. The predicted results of Terzaghi’s theory
with experimental results and had good applicability in predicting the rheological deformation of
were quite different from the experimental results, and the UH constitutive model agreed well
saturated clay samples.
with experimental results and had good applicability in predicting the rheological deformation
The graduated
of saturated U-tube
clay with a 4 mm inner diameter and a video device were used to improve the
samples.
conventional constant-head permeability test device to provide accurate and fast measurement
The graduated U-tube with a 4 mm inner diameter and a video device were used to improve the
of permeability coefficient of saturated clays under the condition of insignificant amounts of
conventional constant-head permeability test device to provide accurate and fast measurement of
water flow, saves considerable experimental time, and avoids the experimental errors caused by
permeability coefficient of saturated clays under the condition of insignificant amounts of water flow,
environmental and human factors. The unified hardening (UH) constitutive model considering
saves considerable experimental time, and avoids the experimental errors caused by environmental
the time effect was introduced to describe the rheological properties of saturated clay soils,
and human factors. The unified hardening (UH) constitutive model considering the time effect was
and Hansbo’s equation was employed to describe the non-Darcian flow in the process of consolidation.
introduced to describe the rheological properties of saturated clay soils, and Hansbo’s equation was
Accordingly, Terzaghi’s one-dimensional consolidation theory of saturated clay was modified and
employed to describe the non-Darcian flow in the process of consolidation. Accordingly, Terzaghi’s
numerical analysis was conducted using FVM. Comparing the numerical simulation and experimental
one-dimensional consolidation theory of saturated clay was modified and numerical analysis was
results, the applicability of the UH model considering time effects and the necessity of considering
conducted using FVM. Comparing the numerical simulation and experimental results, the
variable permeability coefficient were verified.
applicability of the UH model considering time effects and the necessity of considering variable
permeability
7. Patents coefficient were verified.

1.7. Patents
Zhongyu Liu; Yangyang Xia; Jiachao Zhang; Xinmu Zhu; Jiandong Wei. A device for preparing
saturated remolded clay ring knife sample. China Patent, CN201,721,820,336.8, 2018-08-07.
1. Zhongyu Liu; Yangyang Xia; Jiachao Zhang; Xinmu Zhu; Jiandong Wei. A device for preparing
2. Jiandong Wei; Yangyang Xia; Zhongyu Liu; Jiachao Zhang; Xinmu Zhu. A triple saturated clay
saturated remolded clay ring knife sample. China Patent, CN201,721,820,336.8, 2018-08-07.
constant-head permeability test device. China Patent, CN201,821,335,366.4, 2019-04-12.
Water 2019, 11, 1385 15 of 16

Author Contributions: Conceptualization, Z.L., M.S., and Y.X.; investigation, Y.X., J.Z., and X.Z.; methodology,
Z.L., M.S., Y.X., J.Z., and X.Z.; formal analysis, Z.L., M.S., and Y.X.; writing, Y.X. All the authors approved the
submission of this manuscript.
Funding: This experiment was supported by a research grant (Project No. 51,578,511 and 51,679,219) from the
National Natural Science Foundation of China (NSFC).
Acknowledgments: Special thanks to Yuke Wang, Jiandong Wei, and Chaojie Wang for providing instructions
and suggestions on this work.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Terzaghi, K. Theoretical Soil Mechanics; Wiley and Sons: New York, NY, USA, 1943.
2. King, F.H. Principles and Conditions of the Movement of Ground Water; United States Geological Survey: Reston,
VA, USA, 1898.
3. Hansbo, S. Consolidation of clay with special reference to influence of vertical sand drain. Swed. Geotech. Inst.
1960, 18, 45–50.
4. Swartzendruber, D. Modification of darcy’s law for the flow of water in soils. Soil Sci. 1962, 93, 22–29.
[CrossRef]
5. Slepicka, F. Contribution to the Solution of the Filtration Law; International Union of Geodesy and Geophysics,
Commission of Subterranean Waters: Helsinkl, Finland, 1960; pp. 245–258.
6. Miller, R.J.; Low, P.E. Threshold gradient for water flow in clay systems. Soil Sci. Soc. Am. J. 1963, 27, 605–609.
[CrossRef]
7. Qi, T.; Xie, K.H.; Hu, A.F.; Zhang, Z.Q. Laboratorial study on non-Darcy permeability in Xiaoshan clay.
J. Zhejiang Univ. Eng. Sci. 2007, 41, 1023–1028.
8. Dubin, B.; Moulin, G. Influence of a critical gradient on the consolidation of clays. In Consolidation of Soils:
Testing and Evaluation (STP 892); ASTM: West Conshochen, PA, USA, 1986; pp. 354–377.
9. Wang, X.Y.; Liu, C.L. New understanding of the regularity of water permeability in cohesive soil.
Acta Geosci. Sin. 2003, 24, 91–95.
10. Deng, Y.E.; Xie, H.P.; Huang, R.Q.; Liu, C.Q. Law of nonlinear flow in saturated clays and radial consolidation.
Appl. Math. Mech. 2007, 28, 1427–1436. [CrossRef]
11. Sun, L.Y.; Yue, J.C.; Zhang, J. Experimental study on non-Darcian permeability characteristics of saturated
clays. J. Zhengzhou Univ. Eng. Sci. 2010, 31, 31–34.
12. Hansbo, S. Deviation from Darcy’s law observed in one dimensional consolidation. Geotechnique 2003, 53, 601–605.
[CrossRef]
13. E, J.; Chen, G.; Sun, A.R. One-dimensional consolidation of saturated cohesive soil considering non-Darcy
flows. Chin. J. Geotech. Eng. 2009, 31, 1115–1119.
14. Li, C.X.; Xie, K.H. One-dimensional nonlinear consolidation of soft clay with the non-Darcian flow. J. Zhejiang
Univ. Sci. 2013, 14, 435–446. [CrossRef]
15. Liu, Z.Y.; Sun, L.Y.; Yue, J.C.; Ma, C.W. One-dimensional consolidation theory of saturated clay based on
non-Darcy flow. Chin. J. Rock Mech. Eng. 2009, 28, 973–979.
16. Liu, Z.Y.; Zhang, J.C.; Zheng, Z.L.; Guan, C. Finite element analysis of two-dimensional Biot’s consolidation
with Hansbo’s flow. Rock Soil Mech. 2018, 39, 4617–4626.
17. Liu, Z.Y.; Jiao, Y. Consolidation of ground with ideal sand drains based on Hansbo’s flow. Chin. J. Geotech. Eng.
2015, 37, 792–801.
18. Davis, E.H.; Raymond, G.P. A non-linear theory of consolidation. Géotechnique 1965, 15, 161–173. [CrossRef]
19. Zhuang, Y.C.; Xie, K.H.; Li, X.B. Nonlinear analysis of consolidation with variable compressibility and
permeability. J. Zhejiang Univ. Eng. Sci. 2005, 6, 181–187. [CrossRef]
20. Conte, E. Consolidation of anisotropic soil deposits. Soils Found. 1998, 38, 227–237. [CrossRef]
21. Bartholomeeusen, G.; Sills, G.C.; Znidarčiċ, D.; Van Kesteren, W.; Merckelbach, L.M.; Pyke, R.; Carrier, W.D.;
Lin, H.; Penumadu, D.; Winterwerp, H.; et al. Sidere: Numerical prediction of large-strain consolidation.
Géotechnique 2002, 52, 639–648. [CrossRef]
22. Conte, E.; Troncone, A. Nonlinear consolidation of thin layers subjected to time-dependent loading.
Can. Geotech. J. 2007, 44, 717–725. [CrossRef]
Water 2019, 11, 1385 16 of 16

23. Buisman, A.S. Results of long duration settlement tests. In Proceedings of the 1th International Conference
on Soil Mechanics and Foundation Engineering, Cambridge, MA, USA, 22–26 June 1936; pp. 103–105.
24. Taylor, D.W.; Merchant, W. A theory of clay consolidation accounting for secondary compression. J. Math. Phys.
1940, 19, 167–185. [CrossRef]
25. Chen, Z.J. One dimensional problems of consolidation and secondary time effects. China Civ. Eng. J. 1958, 5, 1–3.
26. Bjerrum, L. Engineering geology of normally consolidated marine clays as related to the settlement of
buildings. Geotechnique 1967, 17, 83–118. [CrossRef]
27. Bjerrum, L. Embankments on soft ground: State of threat report. In Proceedings of the Specialty Conference
on Performance of Earth and Earth Supported Structures, Lafayette, Indiana, 11–14 June 1972; pp. 1–54.
28. Nash, D.F.T.; Sills, G.C.; Davison, L.R. One-dimensional consolidation testing for soft clay from Bothkennar.
Geotechnique 1992, 42, 241–256. [CrossRef]
29. Chen, X.P.; Luo, Q.Z.; Zhou, Q. Time-dependent behaviour of interactive marine and terrestrial deposit clay.
Geomech. Eng. 2014, 7, 279–295. [CrossRef]
30. Yin, Z.Z.; Zhang, H.B.; Zhu, J.G.; Li, G.W. Secondary consolidation of soft soils. Chin. J. Geotech. Eng. 2003, 25, 521–526.
31. Reddy, B.K.; Sahu, R.B.; Ghosh, S. Consolidation behavior of organic soil in normal Kolkata Deposit.
Indian Geotech. J. 2014, 44, 341–350. [CrossRef]
32. Yin, J.H. Non-linear creep of soils in oedometer tests. Geotechnique 1999, 49, 699–707. [CrossRef]
33. Feng, W.Q.; Lalit, B.; Yin, Z.Y.; Yin, J.H. Long-term Non-linear creep and swelling behavior of Hong Kong
marine deposits in oedometer condition. Comput. Geotech. 2017, 84, 1–15. [CrossRef]
34. Yao, Y.P.; Kong, L.M.; Hu, J. An elastic-viscous-plastic model for overconsolidated clays. Sci. China Tech. Sci.
2013, 56, 441–457. [CrossRef]
35. Hu, J.; Yao, Y.P. One-dimensional consolidation analysis of UH model considering time effect. J. Beijing Univ.
Aeronaut. Astronaut. 2015, 41, 1492–1498.
36. GB/T50123-1999. Standard for Soil Test Method; China Planning Press: Beijing, China, 1999.
37. Liu, Z.Y.; Xia, Y.Y.; Zhang, J.C.; Zhu, X.M.; Wei, J.D. Preparation Device for Saturated Remodeling Clay Soil
Ring Knife Sample. China Henan CN207703568U, 7 August 2018.
38. Taylor, D.W. Fundamentals of Soil Mechanics; John Wiley & Sons Inc.: New York, NY, USA, 1948.
39. Beery, P.L.; Wilkinson, W.B. The radial consolidation of caly soils. Geotechnique 1969, 19, 253–254. [CrossRef]
40. Wong, R.C.; Varatharajan, S. Viscous behaviour of clays in one-dimensional compression. Can. Geotech. J.
2014, 51, 795–809. [CrossRef]
41. Li, X.B. Theoretical and Experimental Studies on Rheological Consolidation of Soft Soil; Zhejiang University:
Hangzhou, China, 2005.
42. Chen, F.; Drumm, E.C.; Guiochon, G. Coupled discrete element and finite volume solution of two classical
soil mechanics problems. Comput. Geotech. 2011, 38, 638–647. [CrossRef]
43. Sorensen, K.K.; Okkels, N. Correlation between Drained Shear Strength and Plasticity Index of Undisturbed
Overconsolidated Clays. In Proceedings of the 18th International Conference on Soil Mechanics and
Geotechnical Engineering, Paris, France, 2–6 September 2013; Volume 1, pp. 423–428.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).