Electrokinetic treatment of desiccated expansive clay

Item Type Article

Authors Hamza, Omar; Ikin, Jamie

Citation Hamza, O. and Ikin, J., (2019). 'Electrokinetic Treatment of

Desiccated Expansive Clay'. Géotechnique, pp.1-34. DOI: 10.1680/
jgeot.18.p.266.

DOI 10.1680/jgeot.18.p.266

Publisher ICE Publishing

Journal Géotechnique

Download date 10/03/2021 13:47:32

Link to Item http://hdl.handle.net/10545/623763

 Hamza, O. & Ikin, J. Géotechnique [https://doi.org/10.1680/jgeot.18.P.266]

Electrokinetic treatment of desiccated expansive clay

OMAR HAMZA
and JAMIE IKIN†

Structures built on shallow foundations in expansive clay, where vegetation and weather conditions can
significantly reduce the moisture content, must overcome the risk associated with excessive desiccation
conditions, such as shrinkage and settlement. For many existing structures, removal of trees may not
necessarily be an effective solution, as it can cause swelling and heave in the long term. This study
explores a system that could be operated during prolonged drought periods to rehydrate and neutralise
the negative pore pressures of expansive clay, avoiding the development of excessive desiccation.
Electrokinetics is a proven method that has increasingly been adopted for dewatering and consolidation
of fine-grained soils. The hypothesis presented is to reverse the electrokinetics process by extracting
water from a saturated stratum below the groundwater level and force it into the ‘early-stage’ desiccated
area, accelerating rehydration and reducing suction. The theory was examined by a laboratory model
using moderately expansive clay. After draining the soil for 23 days to reach a moderate desiccation
state, water was allowed into the lower part of the soil while applying the EK treatment using nine
electrodes, placed in a radial pattern and inserted 110 mm into the soil with approximately 30% of their
lengths below the water level. As a result of the electro-osmotic process, the average moisture content
increased in the model by more than two-fold within 8 h. The suction measurement (taken in the
middle of the desiccated area) showed an initial slow response followed by fast and consistent reduction
rate, where suction dropped by 93% within 5 h and ultimately down to 1 kPa at the end of the
treatment. This paper offers some insight into the factors controlling the proposed system and provides
a good basis for future research to manage the risks associated with desiccation of expansive soils.

KEYWORDS: electrokinetics; expansive soils; footings/foundations; vegetation

INTRODUCTION zone (suction) or desiccated area. Environmental, geological,

Expansive soils in many countries pose a significant hazard
to light buildings in response to the seasonal fluctuation of
moisture. In the UK, damage to properties became a major
issue following the drought of 1975/1976 with a significant
increase in insurance claims, where there was an average of
one claim per 1000 dwellings (Clark, 1980). The majority of
damage is attributed to shrinking and swelling of expansive
soils affecting shallow foundations in response to changes in
water content and pore water pressure. With future scenarios
of climate change, such as severe alterations in rainfall
intensity causing wetter or hotter seasons (Kilsby et al., 2009;
Moss et al., 2010), the scale of the problem is expected to
increase.
In the shrinking case, this is caused by soil desiccation –
that is, drying of the soil resulting in an increase in suction (or
decrease in pore water pressure over the normal, ‘equili-
brium’ values), as shown in Fig. 1 (BRE Digest 412 (BRE,
1996)). The suction referred to throughout this paper is
sometimes known as the matrix suction or capillary tension.
Soil desiccation (associated with shrinkage) is caused by a
combination of surface evaporation and evapotranspiration
from vegetation demanding water from roots (Lawson, 2004;
Biddle, 1998). Water molecules are bound to the clay
structure and, as vegetation extracts these molecules, the
clay volume is reduced, creating a local negative pressure
Manuscript received 15 October 2018; revised manuscript accepted
17 April 2019.
Discussion on this paper is welcomed by the editor.

The University of Derby, College of Engineering and Technology,
Derby, UK (Orcid:0000-0002-5048-0423).
† C2C Consulting Engineers Ltd, Stoke-on-Trent, UK.
climatic, hydrogeological, chemical and ecological are just
some of the factors that influence the desiccation model,
which is why it is so difficult to predict and therefore no two
sites will ever be the same (BRE Digest 298 (BRE, 1999)).
This may explain the level of ambiguity displayed in the
current guidelines and standards when it comes to building in
such soils.
Desiccation from trees can be up to eight metres
(Chandler et al., 1992; Blight, 2005) and not all movement
from settlement is vertical; lateral shrinkage, propagation of
desiccation cracks or simply decreased saturation all need to
be accounted for. Rehydration occurs either from rain
infiltration at the surface or hydrogeological influences.
However, due to the extremely low permeability of clay, it
can take years for the moisture levels to return to a stable
level. Yet the removal of trees or large shrubs might not
always be a feasible option because of the following reasons:
(a) soil swelling (expansion) following the removal of a tree or
vegetation is a very common problem as the moisture
recovers, resulting in foundation heave; (b) in some situations,
changing or removal of trees is prohibited by law, as many
trees are classified as protected species under Tree
Preservation Orders (Profous & Loeb, 1990).
The Building Research Establishment (BRE) digest 240
(BRE, 1993) notes that the minimum foundation depth is
0·9 m to avoid the effects of moisture variations, with the
upper limit for mass trench fill foundations in the National
House Building Council (NHBC) guidelines (NHBC, 2018)
being 2·5 m. Such deep excavation requires a large quantity
of construction materials, making this solution inefficient,
complex and expensive. Although considered as cheaper
alternatives, trench fill or strip foundations are subject to
loads in all directions: lateral, vertical and possibly rotational
movements, as shown in Fig. 1. The use of proprietary

Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
 2 HAMZA AND IKIN
Foundation of an
existing building
Ground shrinkage or expansion
changes vertical and lateral
pressure on foundation unless it
is deepened below the
influence zone
Fig. 1. Effects of desiccation with depth on shallow foundations
products on the faces of the foundation provides some
tolerance to these pressures and is recommended by both
BRE and NHBC guidelines, although representing an extra
additional cost.
In this study, a system is explored that utilises the inno-
vative electrokinetics (EK) technology to rehydrate and
neutralise the negative pore pressures of expansive clay,
avoiding the development of excessive desiccation and manag-
ing the risk of shrinkage and settlement during prolonged
drought periods. EK treatment of soil involves passing a
direct current (d.c.) between electrodes and invoking the
movement of charged ions that drag water molecules with
them (Casagrande, 1949, 1952, 1983; Casagrande et al.,
1961). The technique is typically used for dewatering and
consolidation of fine-grained soils (Hamir et al., 2001;
Lamont-Black et al., 2012, 2016; Jones et al., 2017).
The current authors’ hypothesis is to reverse the EK
process by extracting water from a stratum of lower saturation
(or below the groundwater table) and forcing it up into the
‘early-stage’ desiccated area, accelerating rehydration and
reducing suction. Bringing the pore water pressure back to
near equilibrium state, especially where an existing building is
within the influence zone of trees, would reduce the settlement
potential and negate the need for costly solutions.
The aim of this paper is to explore whether the proposed
EK system can move the water from the deep saturated layer
laterally and vertically into desiccated areas in an accelerated
way to bring pore water pressure back to a near-equilibrium
state. This was investigated on a small-scale laboratory model
using a purpose-built testing apparatus and moderately
expansive clay. The physical model allowed good control of
the boundary conditions throughout the experimental
phases. The authors were able to produce desiccated soil
(i.e. representing the short-term effect of plant roots and dry
season) before introducing water to the lower part of the soil
(i.e. representing the groundwater table). The EK treatment
was applied through nine small electrodes, where anodes and
cathodes were placed in a radial pattern. Then the authors
looked at how the electro-osmotic process changed soil
conditions, including suction, average moisture content, and
vertical and horizontal displacement of the soil surface in
relation to time, to evaluate the effectiveness of the proposed
EK treatment.
BACKGROUND TO ELECTROKINETICS
Observations of the electrokinetic phenomenon are
reported as early as 1809, when Reuss carried out several
Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
Weather
Tree condition
Pore water pressure
– ve + ve
Equilibrium
Desiccation
Influence zone limit
Depth
experiments and noticed that when clay was placed under
direct current, water collected at the negative pole. At a later
stage, several tests were carried out by Casagrande (1952),
where steel piles had been driven into saturated silt that was
impeding the excavation of a railway cutting due to extensive
flow slides. From these early experiments, practical ideas for
applying EK treatment to soil have been developed. The
technique has significant advantages when the electro-
osmosis effect is required – most significantly, the speed of
the process and the ability to overcome permeability
limitations of fine-grained soils.
Electrokinetic treatment involves passing d.c. through
a material, provoking three main processes: electro-
osmosis, electrophoresis and electro-migration (Abdullah &
Al-Abadi, 2010). The present study considered electro-
osmosis, which, across a voltage gradient between two
electrodes – the anode (positive) and the cathode (negative) –
induces ion migration. As the ions migrate, they provide
hydraulic drag that forces pore water to flow (Gray &
Mitchell, 1967). The flow is initiated in a diffuse double
layer, where there are significantly more cations than anions
due to the proximity of the negatively charged clay surface
(Glendinning et al., 2005). In the process, oxygen is released
at the anode and hydrogen at the cathode, as well as
generating heat as the energy is transferred.
Natural water flow has a direct relationship to the
hydraulic head, creating a difference in pressures and particle
size. As fine-grained soils have a very low hydraulic
permeability it is extremely difficult to move water in, out
or through the material; consequently, the effects of volume
change can be incredibly slow. Unlike hydraulic permeability,
electro-osmotic permeability is relatively unaffected by pore
size, therefore under a d.c. voltage water flows at rates 100 to
10 000 times greater than hydraulic flow (Jones et al., 2008).
The flow of water under an electrical potential gradient can
be described in analogy with Darcy’s law as follows
Qe ¼ Ke ie A ð1Þ
where Ke is the coefficient of electro-osmotic permeability
(m2/s V); ie is the electrical potential gradient (V/m); and A is
the cross-sectional area (m2). According to equation (1), if an
effective voltage gradient is maintained, the electro-osmotic
flow will largely depend on the material having an acceptable
coefficient of electro-osmotic permeability (Ke).
The link between electro-osmotic permeability and elec-
trical conductivity (reciprocal of electrical resistivity), water
content, porosity, particle size and particle structure is a
complex model to predict. Yeung & Mitchell (1993) provided
 ELECTROKINETIC TREATMENT OF DESICCATED EXPANSIVE CLAY 3
9
a Ke value of 5·7  10 m /s V for kaolin clay at 67·7% water
2
60
CE
content. Dias (2015) explored the relationship between CV
electrical resistivity in kaolin clay and the link with void 50
ratio, water content and suction. Dias (2015) proposed that, CH
40 ME
in cohesive soils like clay, the soil’s internal particle structure

is linked to the electrical conductivity, and the geometry of
the voids should be considered to restrict the flow. Fukue
et al. (1999) observed that, in unsaturated clays, the resistivity
is very high for low water contents (lower than a critical
value) and very low for higher water contents. The abrupt rise
in the resistivity shows a clear zone of discontinuity of the
pore water that is a direct result of the reduction of voids due
to desiccation. Such a rise in resistivity would require an
increase in voltage and current that would make EK
treatment unviable past this limit.
Several practical issues associated with the EK technique,
such as the lack of control over boundary conditions, which
might not have adequately appreciated in the past (Hamir
et al., 2001), have been recently addressed (Lamont-Black
et al., 2012, 2016). Most industrial applications involve
the removal of water, such as decontamination or con-
solidation. Typical solutions for ground engineering
application would be consolidation of weak soil or the
stabilisation of embankments (Pugh, 2002; Chew et al., 2004;
Kulathilaka et al., 2004; Glendinning et al., 2005; Abdullah
& Al-Abadi, 2010; Lamont-Black & Weltman, 2010; Jones
et al., 2011).
Research investigating an enhanced efficiency of electro-
osmotic processes for ground improvement and remediation
has focused on several techniques, including polarity reversal
(Luo et al., 2005; Ou et al., 2009) and injecting chemical
solution (Alshawabkeh & Sheahan, 2003; Asavadorndeja &
Glawe, 2005). In the present paper an attempt is made to
expand the applications of the EK techniques and examine
whether the reversal of the mainstream dewatering electro-
osmotic process is suitable for treating desiccated expansive
soils.
MATERIALS AND METHODS
Materials characterisation
The components of the soil mixture selected for the
experimental programme were commercially sourced and
then mixed in the laboratory. The particle size distribution
(PSD) of the soil mixture was 100% smaller than 63 μm
including 26% clay (i.e. ,2 μm). The clay contained
kaolinite and bentonite (6 : 1), which both have been widely
used in fundamental studies of soil behaviour and in physical
model tests (e.g. Seed & Lundgren, 1962; Horpibulsuk et al.,
2011). To determine the shrinkage and swelling potential of
the soil, Atterberg limits tests (plastic limit (PL) and liquid
limit (LL)) were conducted on nine specimens according to
BS 1377 (BSI, 2016). These tests allow the determination of
the plasticity index (PI), which has been widely used to assess
the swelling and shrinkage behaviour of expansive soils
(Jones & Jefferson, 2012).
The results showed the soil had an average liquid limit
(LL) = 52%, PL = 25% and PI = 27, confirming a high-
plasticity clay (CH), as shown on the plasticity chart in
Fig. 2. Also, as the soil contained 26% clay, the activity was
determined as 0·96. Soil with activity between 0·75 and 1·0 is
considered as active, signifying a high potential for volume
change (Seed et al., 1960).
According to NHBC (2018) and BRE (1993), and based
on the average PI value, the soil has medium volume change
potential (VCP). This is also consistent with other methods
for determining swell potentials, such as the correlation
between the soil activity and the percentage of clay (Yilmaz,
2006). Because of the (relatively) small scale of the physical
Plasticity index
CI
30
MV
20
MH
10
ML MI
0
20 30 40 50 60 70 80 90 100
Liquid limit: %
Fig. 2. Plasticity chart showing the class of the tested soil: mainly
high-plasticity clay (CH)
model adopted for the experimental programme, moderately
expansive soil is considered suitable for this study. Ideally, the
testing would have been conducted on highly to very highly
expansive soil; however, such soils can generate significant
suction, causing large-scale desiccation cracks, which would
require a larger model and special instrumentation to
measure the suction – and this was not possible with the
available experimental equipment.
The coefficient of hydraulic permeability (K) for the soil
was determined by falling head permeameter and indicated
that the soil has K = 4·2  108 m/s at a void ratio of
approximately 1·5.
Assembly of the testing box and model preparation
A testing box of 300 mm  300 mm  300 mm was con-
structed from 10 mm thick poly(methyl methacrylate)
(PMMA) and tested for watertightness (see Fig. 3).
A perforated tube and outlet at the base were added for
controlled drainage. The set-up allowed one-dimensional
vertical movement to occur while providing a full view of the
sample surfaces. A degree of lateral displacement and
movement was expected due to shrinkage, and the clear
PMMA allowed this to be monitored. A granular layer of fine
gravel was added with a Camlab grade 4 filter paper
separation to avoid the migration of fines. This layer was
essential to provide drainage to speed up the initial
consolidation.
To prepare the soil in the testing cube, the soil was weighed
dry and mixed with distilled water into a workable slurry of
approximately 80% moisture content. This is much higher
than the LL (52%) to avoid any suction and to improve the
workability while the soil is being prepared. Four 30–40 mm
layers of the slurry were poured into the cube; each layer was
treated with a vibrating poker to remove air and provide a
homogeneous material.
Measurement of moisture content, suction and displacement
As the clay was prepared from ‘oven-dry’ condition
(powder) and the mass of the testing cube and other
materials was known, the initial average moisture content
(ω) could be calculated from the following equation
Wω
ω¼  100 ð2Þ
WS
where Wω is the weight of water mixed with the dry soil and
WS is the weight of soil solids.

Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
 4 HAMZA AND IKIN
(a)
Fig. 3. Testing box: a cube of 300 mm3 constructed from 10 mm thick PMMA: (a) the box tested for water tightness; (b) granular layer at the
base used for drainage; (c) some of the soil (slump) added within box
The loss of water (and thus the average moisture content
change) was monitored by weighing the whole box at regular
intervals and any changes in the overall weight/mass
throughout the experiment were accounted for – that is,
removal or addition of material or tools. These adopted
procedures were able to capture the change of the moisture
content during all testing phases, as well as any moisture loss
due to evaporation and electrolysis. The measurement of the
moisture content was further confirmed on samples recov-
ered from the soil at the beginning and the end of testing.
Negative pore water pressure (suction) was monitored
using a UMS Miniature-Tensiometer T5X connected to a
compatible data logger. The probe was located centrally in
the surface and inserted to a depth of 50 mm where it would
measure matric potential within the range of expected water
movement in the test.
The vertical displacement of the soil surface at the corners
and the centre was measured directly using a displacement
gauge, which could slide on a steel frame fixed on the edges
of the box, reaching any desired location. The horizontal
displacement of the soil surface was determined from
photographs taken consistently from set locations by a
digital single-lens reflex camera. Target points and texture
were generated on the soil surface by sprinkling a small
amount of leaves (of Sawara cypress). Measurements taken
from the photographs were calibrated against the square
edges of the testing box and a small lightweight Lego® house
was placed on the soil for a short period while taking the
photographs.
Phases of the model test
The model test was conducted in three phases: (a) phase 1,
soil consolidation to neutralising the pore water pressure;
(b) phase 2, desiccating the soil; and (c) phase 3, electro-
kinetic treatment. These phases are presented and discussed
in the next following sections.
Soil consolidation – neutralising the pore water pressure
(phase 1). The aim of this phase is to eliminate any
excess pore water pressure and bring the soil over the point of
the onset of suction through self-consolidation. Following
the initial set-up, the slurry was left to consolidate under its
own weight, where excess water could drain away by way of
the perforated tube in the gravel layer, which could be sealed
off as required. However, sealing was only required in
phase 3, when a small reservoir with a constant head
Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
(b) (c)
(simulating groundwater table) was created (see the section
entitled ‘Electrokinetic treatment (phase 3)’ below). A plastic
sheet was placed on the top of the box to minimise surface
evaporation. The average moisture content of the soil at
the end of this stage was recorded to become the baseline and
the target values of the EK treatment.
Desiccating the soil (phase 2). The aim of this phase is to
drain the soil further and reduce the average moisture
content, ensuring the soil is in a low to moderate desiccation
condition (Driscoll, 1983). Soils, when affected by trees, are
generally considered to be in a desiccation state when the
moisture content decreases to a value ranging from 0·4  LL
to 2% above the PL (Driscoll, 1983), which is (for the tested
soil) equal to 20% and 27%, respectively. For this study, 25%
was selected as a target moisture content, which is equal to
the soil PL. Although this has been considered as a rough
guide, a tensiometer was used in the test to ensure the soil has
increased in suction (or decrease in pore water pressure) well
over the normal, ‘equilibrium’ values. The selected moisture
content of 25% reflects the recommended range for effective
electric conductivity (Chandler et al., 1992) where it is
proposed that clay soil maintains a water content of between
25 and 35%.
To continue drying up the soil to the desired moisture
content (25%) in a reasonable time frame, it was deemed
necessary to force the water out of the soil by vacuuming,
owing to the poor permeability of the clay. A small
cylindrical porous stone (average pore size of 179 μm) was
inserted under the soil surface at one corner of the box and
sealed in with the clay (Fig. 4). A vacuum pressure of up to
28 kPa was then applied to the stone by way of a tube, forcing
water out of the soil. With this experimental setting, it was
possible to reduce the water content from 55 to 25% within
5 days as discussed in the ‘Conclusions’ section.
Electrokinetic treatment (phase 3). Once the soil model
had reached the target water content (25%), the process of
EK treatment had begun. Nine graphite rods, 10 mm
dia.  110 mm long, were used as electrodes: three for the
anodes and six for the cathodes. Each was prepared for use by
drilling and tapping into the core and attaching a ring
connector using a stainless steel M4 bolt. Graphite was used
owing to its inert qualities, which would not contaminate the
soil or affect the results. Fig. 5 shows a schematic diagram of
the experimental layout of phase 3.
 ELECTROKINETIC TREATMENT OF DESICCATED EXPANSIVE CLAY 5

B C
Soil

Filter layer

Geotextile
300 mm

Porous stone
(vacuum)

Direction of
A D flow

300 mm 300 mm Tensiometer

(a) (b)

Fig. 4. Schematic diagram of the model in phase 2 (desiccating the soil)

B C
208 mm 10 mm Water
R 300 mm –
– Soil
+ –
– R 200 mm Filter layer
50
s)

– 300 mm

135 mm
diu

Geotextile

110
(ra

+ – 25
Tensiometer
0

+ 35
15

+ R 85
R

25 Electrodes:
R 70 + anode
A D
– cathode
300 mm 300 mm
(a) (b)

Fig. 5. Schematic diagram of phase 3 showing the electrodes’ layout where anodes and cathodes are placed in a radial pattern (not to scale)

To provide the EK system with a water source, a small
reservoir with a constant head (simulating the groundwater
table) was created at the same corner as the cylindrical
porous stone (i.e. used in the previous stage), which was
carefully removed and replaced with a perforated section of
PMMA with filter paper separation (Fig. 5). Distilled water
was used to fill the section to a certain level, where the base of
the electrodes was approximately 35 mm below the water
level (i.e. about 30% of the electrode’s length). Any water
added to the reservoir to create and maintain its constant
level was carefully measured.
It was noticed that water started to seep from the reservoir
into the lower part of the soil at a slow rate before starting the
EK operation; however, the rate of water flow significantly
increased when the electrodes started.
The electrodes were inserted into the surface to a depth of
110 mm in the arrangement shown in Fig. 5 and connected
to the d.c. supply (Fig. 6). Voltage was then applied with the
bench test supply unit allowing up to 30 V of d.c. supply.
During this process, pore water pressures (suction), water
content and surface displacement were monitored at
intervals.
in the soil would overcrowd and disturb the model further.
This issue was addressed by conducting the following stages.
(a) Eleven soil samples across the model were recovered at
the end of phase 3 to measure the moisture contents at
different locations and depths. The testing revealed that
the variation in the moisture content was less than 5·8%,
with maximum values recorded at the central lower part
of the model and minimum values at the top edges of
the soil surface. The average value from the 11 samples
was worked out and compared with the average
moisture content obtained by the adopted approach
(discussed in the earlier section ‘Measurement of
moisture content, suction and displacement’); the
maximum difference was less than 5%.
(b) Visual inspection of the soil surfaces was conducted
during all the testing phases. The appearance of cracks
and the change in soil colour corresponding to dry/wet
conditions provided a reasonable indication of the
variation of moisture condition. The appearance of the
soil at the end of each phase was generally consistent,
suggesting that the moisture varied little across the model.

Variation of moisture contents across the model. The suction

(in the middle of the model, 50 mm below the surface), as
well as the average moisture content, were measured during
the three testing phases (as discussed in the earlier section
entitled ‘Measurement of moisture content, suction and
displacement’). Ideally, the use of more instruments would
have provided better information about the variation of
moisture contents and suction values across the model.
However, given the small scale of the model and the number
of electrodes, the authors believed that inserting more probes
Provisional and control model tests
Conducting provisional and control model tests was
necessary to optimise the methods and understand the
effect of EK operation on the desiccated soil.
Control model tests. The control model tests were con-
ducted using the typical experimental procedures explained
above but without inserting the electrode or applying EK

Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
 6 HAMZA AND IKIN

Lego® house
for scaling

Tensiometer

(a) (d)
Anode
(black cable) D.C. bench
power
Cathode supply
(red cable)

Water Drainage
reservoir layer

(b) (c) (e)

Fig. 6. Photographs taken in phase 3 where electrodes are wired and in operation: (a) top view of the box; (b) anodes and cathodes are inserted;
(c) general view of the testing box; (d) single electrode; (e) bundle of electrodes

treatment. In phase 3, the water was introduced in the

reservoir to create and maintain a low constant level – that is,
to saturate the lower 65 mm of the soil model. It was noticed
that a small rate of seepage continued for 5 days and then
almost stopped; nevertheless the upper part of the soil
remained in a desiccated condition with only 23% reduction
in suction associated with the capillary. Compared with the
results of the main model test (discussed in the later section
‘Results and discussion’), the EK made a significant impact
on the suction and the amount of water delivered into the
desiccated soil.
In a second control test, where the reservoir contained a
high water level, almost near the soil surface, the soil had
naturally (without the aid of EK) regained moisture,
although it required more than 9 days for the suction to
drop to near equilibrium.

Depth of electrodes. It was important to place the tips of the

electrodes adequately below the level of water for an effective
electro-osmotic process. This was based on provisional
testing, where the effect of the depth of the electrodes’ tips
in relation to the water level was examined. The tests
indicated that, when the electrodes had less than 25% of
their length below the water level, the EK did not operate
effectively, causing significant shrinkage around the anodes
and insufficient reduction in suction. The effect of the depth
of electrodes is further discussed in the later section entitled
‘The depth of electrodes’.
Voltage adjustment. In the provisional testing, some obser-
vations were made based on alterations in the voltage and the
polarity of electrodes. The voltage was adjusted between 7 V
and 30 V, which provided a constant current of between
100 mA and 300 mA. It was observed that the higher voltage
resulted in an increase in water flow at the cathodes but also
an increase in the amount of gas produced, as shown in
Fig. 7. The response resulted in water flowing back across the
surface as it was expelled and a reduction of the soil integrity
around the cathodes. Based on the provisional testing, it was
decided to conduct the main model test at no more than 12 V.
This proved to induce adequate flow for the scale of the test
without damaging the soil structure.
Fig. 7. High voltage caused disturbance to the soil around the cathodes
because of the significant water and gas expelled from the surface
RESULTS AND DISCUSSION
The three phases of the experiment lasted approximately
23·5 days. The four data sets recorded over the period of the
experiment were suction (in the middle of the model at
50 mm below the surface), average moisture content and
vertical and horizontal (radial) displacement of the soil
surface. These data are shown in Fig. 8.
Phase 1: soil consolidation
The first phase (Fig. 8(a)) took about 18 days; during this
period drainage through the granular layer allowed a rapid
settlement of the highly saturated clay. Water continued to be
expelled from the drainage pipe for the first 5–6 days, after
which it gradually slowed down and ultimately stopped. This
would suggest that the hydrostatic pressure defined by the
bottom boundary conditions reached an equilibrium point
where water would no longer drain freely. At the end of this
phase, the tensiometer started to pick up small values of
suction (Fig. 8(a)).

Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
 ELECTROKINETIC TREATMENT OF DESICCATED EXPANSIVE CLAY 7
100 100 100

80 80 80

Moisture: % 60 60 60

40 40 40

20 20 20
0 3 6 9 12 15 18 18 19 20 21 22 23 0 2 4 6 8
Time: d Time: d Time: 23 d + h

120 120 120

100 100 100
Suction: kPa

80 80 80
60 60 60
40 40 40
20 20 20
0 0 0
0 3 6 9 12 15 18 18 19 20 21 22 23 0 2 4 6 8
Time: d Time: d Time: 23 d + h
0 3 6 9 12 15 18 18 19 20 21 22 23 0 2 4 6 8
0 0 0
Average vertical

–5 –5 –5
distance: mm

–10 (Settlement) –10 –10 (Swelling)

(Settlement)

–15 –15 –15

–20 –20 –20

Time: d Time: d Time: 23 d + h

0 3 6 9 12 15 18 18 19 20 21 22 23 0 2 4 6 8
0 0 0
Max. (total) horizontal

–5 –5 –5
distance: mm

–10 –10 –10

Radial
–15 Radial shrinkage –15 shrinkage toward the –15 Radial expansion
toward the centre porous stone (corner A) away from corner A
–20 –20 –20
(a) (b) (c)

Fig. 8. Measurements of average moisture content, suction (in the middle of the model at 50 mm below surface) and displacement of the soil
surface during the three phases: (a) phase 1: soil consolidation, where soil surface developed almost uniform vertical displacement, but insignificant
horizontal displacement; (b) phase 2: soil desiccation; and (c) phase 3: electrokinetic treatment. Note: the horizontal displacement of the soil
surface was measured radially in reference to corner A of the testing box. Maximum values were found to be within the proximity of corner C of the
testing box – see Figs 4 and 5

With an initial thickness of 154 mm, the surface level
dropped almost uniformly across the soil surface during the
consolidation, with a maximum settlement of nearly 16 mm
over the 18 days. The reduction in settlement rate can be seen
after day six, which matches the period when drainage from
the granular layer started to reduce. Insignificant horizontal
displacement was noticed during this phase, showing a small
radial shrinkage toward the centre.
During this stage, the average moisture content decreased
from 80 to 55%, with a steady rate of approximately 3·6%
reduction per day in the first 6 days, followed by a slowing
down rate until water completely stopped coming out of the
soil. The tensiometer did not pick up any significant value of
suction within the first 8–9 days, as shown in Fig. 8(a), until
the moisture content reached approximately 55%, where the
suction started to increase at a measurable rate.
Phase 2: Soil desiccation
Mechanically induced suction proved a success and
provided the result needed within a reasonable time frame.
With the vacuuming aid through porous stone (at corner A in
Fig. 4), the average moisture content continued to drop
steadily in phase 2 (Fig. 8(b)), from 55% to reach the targeted
value of 25% within 5 days at a rate of 4% reduction per day.
At the same time, suction increased from 9 to 98 kPa,
indicating that the soil was well within the condition of
desiccation.
Unlike the previous phase, the horizontal displacement
was dominant in this phase (Fig. 8(b)) and showed a radial
pattern of shrinkage, which seemed to be following the
direction of water flow toward the porous stone located at
corner A. The radial horizontal displacement of several
markers distributed over the soil surface was measured in
reference to corner A (see Fig. 5) and the measured values
were interpolated to plot displacement contours using R3·2·1
(R Core Team (RDCT, 2015)). Contours of the (total, radial)
horizontal displacement at the end of this phase are shown in
Fig. 9(a), where a maximum net value of 11 mm is
encountered around corner C (opposite to A, i.e. the location
of the porous stone). Surface settlement continues during this
phase with an additional (average) value of approximately

Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
 8 HAMZA AND IKIN
B C 300 B C
300
–8 –9 –10 6 7 8
–1 9
–7 1
–6
5
–5 4
200 200 3
2
1

–4 100
100 –3

0 D 0
A A D
0 100 200 300 0 100 200 300
(a) (b)

Fig. 9. Contours of (total, radial) horizontal displacement at the end of (a) phase 2 (radial shrinkage) and (b) phase 3 (radial expansion).
Displacement is measured in millimetres and in reference to corner A

4 mm, which suggests further consolidation associated with
the reduction in voids by water removal. A larger value of
settlement (8 mm) was recorded locally around the porous
stones.
During this phase, shrinkage cracks in the side surfaces of
the soil were noticed through the PMMA walls of the testing
box. Fig. 10 shows photographs taken on day 3 of phase 2 (or
day 21 from the beginning of phase 1), which have been
filtered to show the extent of the shrinkage cracking. Most of
the cracks propagated from the corners and extended
horizontally. The two sides furthest from the porous stone,
B–C and C–D, showed more cracking. To some extent, the
cracks on these two surfaces are consistent, both with depth
and direction, which suggests it is happening axisymmetri-
cally around the porous stone. As surface evaporation was
prevented during this phase, these desiccation cracks were
mainly due to suction while the soil was shrinking toward the
porous stone.
The macrostructures left in the form of cracks are an
artefact of the preparation method (the way the water was
forced out of the soil), which might have largely influenced
the displacement pattern of the soil. Nevertheless, the radial
movement of water toward the porous stone (and the
associated radial soil displacements) can partially represent
field conditions around tree roots, where some of the
groundwater moves radially toward the root system causing
desiccation cracks. Moreover, the pattern and density of
cracks generated in the model might have an influence on the
electrical conductivity of soil, particularly at the beginning of
EK treatment (as discussed in the later section entitled ‘The
relationship between suction and moisture content’).
At the end of this phase, there was an increase of vertical
cracks (see Fig. 11) and, in the area circled on the side surface
C–B, what seemed to be a large void where the surface had
moved away from the PMMA. There was a slight increase in
the size of the horizontal cracks, yet no greater amount. At
the end of phase 2, the vacuum pressure was stopped and the
EK set-up was made ready for the next phase. Suction
remained steady and did not fall during the 2 h needed to
start the EK treatment.
Phase 3: EK treatment
The EK treatment commenced on day 24 and lasted for
8 h. Suction pressures, average moisture contents and surface
displacement were measured at regular intervals over the
entire period. As shown in Fig. 8(c), once the treatment had
commenced there was an immediate increase in moisture
content, but a very slow response in the tensiometer readings.
After the first hour, the suction started to decrease at a fast
rate for 5 h while the moisture content continued to rise
steadily. The slow response of suction at the beginning of the
test may indicate an initial influence of the low moisture
content (and possibly the desiccation cracks) on the electro-
osmosis process within the first hour. This may also be
attributed to the location of the tensiometer tip, which was at

C B D C

B C

Surface CB Surface DC

B A A D
Porous
stone

A D
Plan view
Surface BA Surface AD

Fig. 10. Desiccation cracks propagating from the corners. Note: the photographs are filtered to show the extent of the shrinkage cracking

Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
 ELECTROKINETIC TREATMENT OF DESICCATED EXPANSIVE CLAY 9

Surface CB Surface DC

Fig. 11. Shrinkage cracks noticed from the PMMA walls of the testing box at the end of phase 2. The photographs are filtered to show the extent
of the shrinkage cracking

25 mm above the water level, where it might have taken some 100
time for the electrodes to move the water horizontally to the
lower parts of the soil (below water level) before moving the Phase 1
80
water vertically to the upper parts, where the tensiometer was Phase 2
located.

Suction: kPa
Phase 3
Following the first hour, the suction dropped by more than 60
93% within 5 h; after that the rate of suction drop stabilized,
where it was left to continue running for two more hours. 40
There was a very small change of 5·8 kPa between 6 and 8 h,
which suggests that the desired result had already occurred
20
within the first 6 h.
Horizontal displacement showed radial expansion away
from corner A in the direction of water flow from the anodes 0
to the cathodes, as shown in Fig. 9(b). Most of the shrinkage 20 30 40 50 60 70
experienced during desiccation in phase 2 was recovered Moisture: %
in phase 3. However, the recovery pattern is irregular, as
shown in the displacement contours plotted in Fig. 9(b), Fig. 12. The relationship between the measured suction and average
moisture during all experimental phases
which might have been influenced by the locations of the
electrodes.
Vertical expansion is also evident in the vertical displace-
ment measurement where the soil surface showed a net 120
average swell of 3·4 mm, which is almost equal to the net ER = 15%
settlement recorded in phase 2 during desiccation. 100
ER = 23%
Interestingly, there was a small local settlement around the
anodes, which might be associated with water flow toward 80 ER = 30%
the cathodes.
Suction: kPa

60

The relationship between suction and moisture content 40

The laboratory results of the suction and moisture content
during the three phases were analysed and are plotted in
Fig. 12, which can be reasonably considered as an overall
approximation of the soil-water characteristics curve. As
shown in Fig. 12, the suction depends primarily on the
moisture content and increased with soil dehydration in
phases 1 to 2, but decreased with electro-osmosis rehydration
during phase 3.
The effect of depth of electrodes
As mentioned in the earlier section ‘Phases of the model
test’, the provisional testing indicated that the depth of
electrodes below the water level can affect the ability of the
electro-osmotic process to move water into the soil and
reduce suction. Fig. 13 shows the tensiometer measurements
taken in phase 3 (during EK operation – see Fig. 5) in three
different model tests. In each test, the electrodes have
different embedment ratio (ER) below the water level,
where ER is defined as
D considerable role in the success of the EK treatment. In
ER ¼ ð3Þ
L actual field conditions, the groundwater table could be very
20
0
0 2·5 5·0 7·5 10·0 12·5 15·0
Time: h
Fig. 13. Suction measurements in phase 3 for three different model
tests with different embedment ratio (ER) of the electrodes below water
level
where D is the depth of electrode below the water level and L
is the total length of electrode within the soil.
As can be seen in Fig. 13, within 8–10 h of EK operation,
the suction decreased in all three model tests and reached
almost stable values. These values were found to be a
function of the electrode embedment ratio (ER). Only when
the electrode embedment ratio was large enough (ER = 30%)
did the suction drop down to almost zero.
These findings indicate that the groundwater table plays a

Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
 10
low in the dry season and therefore very long anodes might
be required. Moreover, the electrical field in a large area is
different from that in a small model, and this might influence
the pattern of water movement differently, as well as any
change in the microstructure of expansive soils during the
electro-osmotic process. Therefore, it is necessary to conduct
large-scale testing to validate the proposed technique.
CONCLUSIONS
The purpose of the current study was to investigate the
suitability of the EK technique to mobilise groundwater from
deep saturated strata and rehydrate and neutralise the
negative pore pressures in the desiccated expansive clay,
avoiding the development of excessive desiccation during
drought periods and thus managing the risk of structural
damage to buildings.
The results, from the laboratory model, showed that EK is
an option worth considering and the electro-osmotic process
is capable of moving water molecules into clay horizontally
and vertically, restoring pore water pressures and moisture
content in the areas that have been affected by desiccation.
The EK treatment successfully increased the average moist-
ure content by more than two-fold within 8 h. The suction
measurement (taken in the middle of the desiccated area)
showed an initially slow response, which might be attributed
to the location of the tensiometer and the possible influence
of the low moisture and desiccation cracks on the electro-
osmosis process within the first hour. However, this was
followed by a fast and consistent response, where suction
dropped by 93% within 5 h and ultimately down to 1 kPa at
the end of the treatment period.
The horizontal displacements of the soil surface indicated
that the initial shrinkage, generated by desiccation, was
mostly recovered during the EK treatment; however, the
recovery pattern is irregular and largely influenced by the
lateral distribution of the electrodes.
Despite its exploratory nature, this study offers some
insight into the factors controlling the proposed system,
including the following points.
(a) Adjustment of voltage is required to suit the soil
condition, where higher voltage can increase water flow
at the cathodes but also may result in water flowing back
across the surface, causing reduction of the soil integrity
around the cathodes. This finding is in agreement with
earlier literature.
(b) Insufficient depth of the electrodes below the water level
is a detrimental factor to the EK efficiency; a minimum
depth of 30% of the length of the electrode provided the
best response. In the field, the groundwater table could
be very low in the dry season and therefore longer
anodes might be required.
(c) The proposed system would be more effective when the
treated soil is in an early stage of desiccation, when the
shrinkage is tolerable and the electrical conductivity is
not highly affected by the desiccation cracks and low
moisture content.
The results provide a good basis for future research to
manage the risks associated with desiccated expansive soils.
Further experimental investigations are required to monitor
the variation in soil conditions by continuously measuring
the pore water pressure and moisture content at multiple
locations across the model – both vertically and horizontally.
The study might also benefit from further research into the
short- and long-term effects of the electro-osmotic process on
the microstructures and VCP of expansive soil. Element
Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
HAMZA AND IKIN
testing would also be required to examine the effect of
moisture and dissociation cracks on electro-osmotic per-
meability, and to support the development of a closed-form
solution describing the re-hydration path of the EK.
Another future direction for this study would be to carry
out full-scale in situ testing of the proposed system within a
desiccated soil with established trees. This would be essential
to create a model predicting the electro-osmotic rehydration
process in expansive soils following natural desiccation.
Moreover, the full-scale in situ testing would generate an
electrical field in a large area which is different from a small
model, and this might differently influence the pattern of
water movement, as well as any change in the microstructure
of the expansive soils during the electro-osmotic process.
ACKNOWLEDGEMENT
The authors would like to acknowledge the School of
Mechanical Engineering and Built Environment at the
University of Derby (EAT2017IS2), which has internally
funded this project.
NOTATION
A cross-sectional area
a inverse of the air entry pressure
D depth of electrode below the water level
ER electrode embedment ratio below water level
ie electrical potential gradient
K coefficient of hydraulic permeability
Ke coefficient of electro-osmotic permeability
L total length of electrode within the soil
n pore size distribution parameter
Qe flow of water under electrical potential gradient
WS weight of soil solids
Wω weight of water within the soil
ω moisture content of soil
REFERENCES
Abdullah, W. & Al-Abadi, A. (2010). Cationic–electrokinetic
improvement of an expansive soil. Appl. Clay Sci. 47, No. 3–4,
343–350.
Alshawabkeh, A. N. & Sheahan, T. C. (2003). Soft soil stabilization
by ionic injection under electric fields. Ground Improvement 7,
No. 4, 177–185.
Asavadorndeja, P. & Glawe, U. (2005). Electrokinetic strengthening
of soft clay using the anode depolarization method. Bull. Int.
Ass. Engng Geol. 64, No. 3, 237–245.
Biddle, P. G. (1998). Tree root damage to buildings – vol. 1 causes,
diagnosis and remedy. Wantage, UK: Willowmead Publishing.
Blight, G. E. (2005). Desiccation of a clay by grass, bushes and trees.
Geotech. Geol. Engng 23, No. 6, 697–720.
BRE (Building Research Establishment) (1993). Digest 240 –
low-rise buildings on shrinkable clay: part 1. Garston, UK:
Construction Research Communications.
BRE (1996). Digest 412 – desiccation in clay soils. Garston, UK:
Construction Research Communications.
BRE (1999). Digest 298 – low-rise building foundations: the influence
of trees on house foundations in clay soils. Garston, UK:
Construction Research Communications.
BSI (2016). BS 1377: Methods of test for soils for civil engineering
purposes – part 2: Classification tests. London, UK: BSI.
Casagrande, L. (1949). Electro-osmosis in soils. Géotechnique 1,
No. 3, 159–177, https://doi.org/10.1680/geot.1949.1.3.159.
Casagrande, L. (1952). Electro-osmotic stabilisation of soils.
J. Boston Soc. Civ. Engrs 39, No. 1, 51–83.
Casagrande, L. (1983). Stabilisation of soils by means of electro-
osmosis state-of-the-art. J. Boston Soc. Civ. Engrs 69, No. 2,
255–302.
Casagrande, L., Loughney, R. W. & Matich, M. A. J. (1961).
Electro-osmotic stabilisation of a high slope in loose saturated
silt. Proceedings of the 5th international conference on soil
 ELECTROKINETIC TREATMENT OF DESICCATED EXPANSIVE CLAY
mechanics and foundation engineering, Paris, France, vol. 2,
pp. 555–561.
Chandler, R. J., Crilly, M. S. & Montgomery-Smith, G. (1992).
A low-cost method of assessing clay desiccation for low-rise
buildings. Proc. Instn Civ. Engrs – Civ. Engng 92, No. 2, 82,
https://doi.org/10.1680/icien.1992.18771.
Chew, S. H., Karunaratne, G. P., Kuma, V. M., Lim, L. H.,
Toh, M. L. & Hee, A. M. (2004). A field trial for soft clay
consolidation using electric vertical drains. Geotextiles and
Geomembranes 22, No. 1–2, 17–35.
Clark, M. J. (1980). Property damage by foundation failure. In Atlas
of drought in Britain (eds J. C. Doornkamp, K. J. Gregory and
A. S. Burn), pp. 63–64. London, UK: Institute of British
Geographers.
Dias, A. S. A. (2015). Electrical resistivity of compacted kaolin and
its relation with suction. Doctoral Thesis, Instituto Superior
Técnico, University of Lisbon, Lisbon, Portugal.
Driscoll, R. (1983). The influence of vegetation on the swelling and
shrinking of clay soils in Britain. Géotechnique 33, No. 2,
93–105, https://doi.org/10.1680/geot.1983.33.2.93.
Fukue, M., Minato, T., Horibe, H. & Taya, N. (1999). The
micro-structures of clay given by resistivity measurements.
Engng Geol. 54, No. 1–2, 43–53.
Glendinning, S., Jones, C. J. & Lamont-Black, J. (2005). The use of
electrokinetic geosynthetics (EKG) to improve soft soils. In
Ground improvement – case histories (eds B. Indraratna and
J. Chu), Elsevier geo-engineering book series vol. 3,
pp. 997–1043. Amsterdam, the Netherlands: Elsevier.
Gray, D. H. & Mitchell, J. K. (1967). Fundamental aspects of
electroosmosis in soils. J. Soil Mech. Found. Engng 93, No. 6,
209–236.
Hamir, R. B., Jones, C. J. F. P. & Clarke, B. G. (2001). Electrically
conductive geosynthetics for consolidation and reinforced soil.
Geotextiles Geomembranes 19, No. 8, 455–482.
Horpibulsuk, S., Yangsukkaseam, N., Chinkulkijniwat, A. &
Du, Y. J. (2011). Compressibility and permeability of Bangkok
clay compared with kaolinite and bentonite. Appl.Clay Sci. 52,
No. 1–2, 150–159.
Jones, L. D. & Jefferson, I. (2012). Expansive soils. ICE manual
of geotechnical engineering. In Geotechnical engineering prin-
ciples, problematic soils and site investigation (eds J. Burland,
T. Chapman, H. Skinner and M. Brown), vol. 1, ch. 33,
pp. 413–438. London, UK: ICE Publishing.
Jones, C., Lamont-Black, J., Glendinning, S., Bergado, D., Eng, T.,
Fourie, A., Liming, H., Pugh, C., Romantshuk, M.,
Simpanen, S. & Yan-Feng, Z. (2008). Recent research and
applications in the use of electrokinetic geosynthetics. In
Proceedings – EuroGeo 4, the fourth European geosynthetics
conference: Edinburgh, UK, keynote lecture 2, CD-ROM.
Loughborough, UK: UK Chapter, International Geosynthetics
Society.
Jones, C., Lamont-Black, J. & Glendinning, S. (2011). Electrokinetic
geosynthetics in hydraulic applications. Geotextiles
Geomembranes 29, No. 4, 381–390.
Jones, C., Lamont-Black, J., Huntley, D., Alder, D. & Glendinning, S.
(2017). Electrokinetic geosynthetics from research to hype
to practice. Proc. Instn Civ. Engrs – Civ. Engng 170, No. 3,
pp. 127–134, https://doi.org/10.1680/jcien.16.00039.
Downloaded by [ University of Derby] on [31/05/19]. Copyright © ICE Publishing, all rights reserved.
11
Kilsby, C., Glendinning, S., Hughes, P. N., Parkin, G. &
Bransby, M. F. (2009). Climate-change impacts on long-term
performance of slopes. Proc. Instn Civ. Engrs – Civ. Engng 162,
No. 2, 59–66, https://doi.org/10.1680/ensu.2009.162.2.59.
Kulathilaka, S. A. S., Sagarika, D. K. N. S. & Perera, H. A. C.
(2004). The parameters affecting the electro-osmotic consolida-
tion of peaty clays. Ann. Trans. Inst. Engrs Sri Lanka 2004,
184–193.
Lamont-Black, J. & Weltman, A. (2010). Electrokinetic strengthen-
ing and repair of slopes. Ground Engng, April, pp. 28–31.
Lamont-Black, J., Hall, J. A., Glendinning, S., Jones, C. J. F. P.
& White, C. (2012). Stabilisation of a railway embankment
using electrokinetic geosynthetics. In Earthworks in Europe:
proceedings of the Geological Society, London (ed.
T. A. Radford), special publication no. 26, pp. 125–139.
London, UK: Geological Society.
Lamont-Black, J., Jones, C. J. F. P. & Alder, D. (2016). Electrokinetic
strengthening of slopes–case history. Geotextiles Geomembranes
44, No. 3, 319–331.
Lawson, M. (2004). Tree related subsidence of low rise buildings and
the management options. Arboric. J. 27, No. 3, 191–219.
Luo, Q., Zhang, X., Wang, H. & Qian, Y. (2005). The use of
non-uniform electrokinetics to enhance in situ bioremediation of
phenol-contaminated soil. J. Hazardous Mater. 121, No. 1–3,
187–194.
Moss, R. H., Edmonds, J. A., Hibbard, K. A., Manning, M. R.,
Rose, S. K., Van Vuuren, D. P., Carter, T. R., Emori, S.,
Kainuma, M., Kram, T. & Meehl, G. A. (2010). The next
generation of scenarios for climate change research and
assessment. Nature 463, No. 7282, 747–756.
NHBC (National House Building Council) (2018). NHBC
Standards Chapter 4.2. Building near trees. London, UK:
National House Building Council.
Ou, C. Y., Chien, S. C. & Chang, H. H. (2009). Soil improvement
using electroosmosis with the injection of chemical solutions:
field tests. Can. Geotech. J. 46, No. 6, 727–733.
Profous, G. V. & Loeb, R. E. (1990). The legal protection of urban
trees: a comparative world survey. J. Environ. Law 2, No. 2,
179–193.
Pugh, R. C. (2002). The application of electrokinetic geosynthetic
materials to uses in the construction industry. Doctoral thesis,
University of Newcastle, Newcastle, UK.
RDCT (R Development Core Team) (2015). R: a language and
environment for statistical computing. Vienna, Austria: R
Foundation for Statistical Computing. See http://www.
R-project.org (accessed 10/06/2018).
Seed, H. B. & Lundgren, R. (1962). Prediction of swelling
potential for compacted clays. J. Soil Mech. Found. Div. 88,
No. 3, 53–88.
Seed, H. B., Mitchell, J. K. & Chan, C. K. (1960). The strength of
compacted cohesive soils. J. Soil Mech. Found. Div. 88,
No. SM-3, 53–87.
Yeung, A. T. & Mitchell, J. K. (1993). Coupled fluid, electrical and
chemical flows in soil. Géotechnique 43, No. 1, 121–134,
https://doi.org/10.1680/geot.1993.43.1.121.
Yilmaz, I. (2006). Indirect estimation of the swelling percent and
a new classification of soils depending on liquid limit and cation
exchange capacity. Engng Geol. 85, No. 3–4, 295–301.