Soils and Foundations ]]]];](]):]]]–]]]

HOSTED BY The Japanese Geotechnical Society

Soils and Foundations

www.sciencedirect.com
journal homepage: www.elsevier.com/locate/sandf

Estimation methods for thermal conductivity of sandy soil with electrical

characteristics
Tetsuya Tokoroa,n, Tatsuya Ishikawab, Shoya Shiraic, Tsutomu Nakamurad
a
Department of Civil Engineering, National Institute of Technology, Tomakomai College, 443 Nishikioka, Tomakomai 059-1275, Japan
b
Faculty of Engineering, Hokkaido University, Japan
c
Graduate School of Engineering, Nagaoka University of Technology, Japan
d
Department of Civil Engineering, Tomakomai National College of Technology, Japan
Received 17 March 2015; received in revised form 8 June 2016; accepted 12 June 2016

Abstract

This study estimates the water dependence of the thermal conductivity of soils and proposes a new thermal conductivity model and empirical
equation for soil. Thermal conductivity tests for three different soil types were conducted under different moisture contents using the thermal
probe method. Electrical conductivity tests were also performed to obtain the electrical resistibilities of the soils, and the influence of the moisture
content on thermal conductivity was examined using the electrical conductivity data. The results show that the water dependence of thermal
conductivity strongly correlates to the electrical resistibility. The relationships between the moisture contents and the thermal conductivities of the
three soil types are found to be nonlinear, with an inflection point in each relationship. The moisture contents corresponding to these inflection
points approximately agree with each other. The provision of a current flow largely along the pore water indicates that the thermal conductivity of
a soil depends on whether or not the pore water, which significantly contributes to heat conduction, is continuous. The presence of pore water in
soils can explain the water dependence of thermal conductivity. This study further proposes a model to estimate the thermal conductivity of soil
and an empirical equation based on electrical conductivity test results.
& 2016 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Thermal conductivity; Thermal property; Thermal probe method; Electrical resistibility; Saturated-unsaturated soil (IGC: D08)

1. Introduction experimental and theoretical approaches have been proposed

The freeze–thaw action is the key triggering factor in many
geotechnical disasters in cold regions during both winter and
spring. Estimations of the temperature distribution underneath
soil layers and the frost-penetration depth, in consideration of
the thermal conductivities of soils, are critical for the preven-
tion of such disasters. In the last few decades, a number of
n the experimental determination of thermal conductivity.
Corresponding author. Fax: þ81 144 67 8062.
E-mail address: t-tokoro@civil.tomakomai-ct.ac.jp (T. Tokoro).
Peer review under responsibility of The Japanese Geotechnical Society.
for the measurement of thermal conductivity. The thermal
probe method was widely used in the 1950s (De Vries, 1952;
Hopper and Chang, 1953) to measure the thermal conductivity
of soils after its development in the 1940s (Kersten, 1952;
De Vries and Peck, 1958). Kasubuchi (1977) subsequently
developed a twin-thermal probe method as an improvement
over the original method. Many empirical equations and
theoretical models have also been proposed as alternatives to
Kersten (1949) proposed an empirical equation for determining
the thermal conductivity of soils after analyzing the outcomes

http://dx.doi.org/10.1016/j.sandf.2016.08.016
0038-0806/& 2016 The Japanese Geotechnical Society. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
2 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]]
of thermal conductivity experiments on 19 different soil types.
Johansen (1975) established an empirical equation that con-
sidered the quartz content, which typically leads to larger
predicted thermal conductivities for soils.
Many of the theoretical models for estimating the thermal
conductivities of soils apply the same analogy, i.e., electrical
resistance in series, parallel, and series–parallel flow models
(e.g., Woodside and Messmer, 1961; Kasubuchi, 1975). The
thermal conductivity models proposed above and the empirical
equations will be discussed in detail in the next chapter. These
models and empirical equations, however, are unable to
quantitatively explain the thermal conductivities of soils
because the factors that influence thermal conductivity, such
as dry density, moisture content, and soil mineral composition,
are intertwined in a complex and causal correlation with
thermal conductivity, and the factors have not been evaluated
to a sufficient degree. Thus, it is necessary to reveal the
mechanism of the thermal flow in soil in order to speculate on
the thermal conductivity of soil.
The purpose of this study is to clarify the impact of the
water content on the thermal conductivity of sandy soils. In
this study, thermal conductivity tests were conducted on three
different soil types using the thermal probe method. Electrical
conductivity tests were also conducted to examine the water
dependence on the thermal conductivity in detail. A new
model and empirical equation are proposed after an analysis of
the experimental results.
2. Review of thermal conductivity models and empirical
Equations
Although a number of difficulties are encountered when
estimating the thermal conductivity of sandy soils empirically,
many theoretical thermal conductivity models have been
proposed to estimate this property over the last few decades,
and most of them are based on resistor equations. Woodside
and Messmer (1961) used a series–parallel model, suggested
by Wyllie and Southwick (1954), to determine the electrical
conductivity of granular materials, similar to the thermal
conductivity models for soils that have been fully saturated
by water, oil, or gas. Fig. 1 shows the heat flow of Woodside's
Fig. 1. Thermal conductivity model.
Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
model, visualized as a unit cube of soil. The model has three
paths of heat flow, as shown in Fig. 1. The values of a, b, and,
c are the proportion of the heat flow through each cuboid to the
total heat flow, and each cuboid has the following meaning:
(a) The cuboid with length a represents the heat flow through
the fluid and solid in a series.
(b) The cuboid with length b represents the heat flow through
a continuous solid.
(c) The cuboid with length c represents the heat flow through
a continuous fluid.
Although it is difficult to determine the parameters, the
thermal conductivity of fully saturated and dry soil can be
expressed using Woodside's model. Subsequently, many
researchers have extended Woodside's model to a three-
phase model in order to estimate the thermal conductivities
of saturated and unsaturated soils. Kasubuchi (1975) devel-
oped a qualitative model, in which thermal conduction takes
place along soil particles and water. This model is not capable
of explaining the non-linear relationship between thermal
conductivity and moisture content. Matsumoto and Ohkubo
(1977) modified Kasubuchi's model by addressing the non-
linear thermal conductivity of soils. This model was not,
however, quantitative, since the parameters were not linked
with either thermal conductivity or physical properties.
Mitsuno et al. (1983) developed a new model to account for
the nonlinearity of the thermal conductivity of soil by
separating the void into three parts to account for the pore-
water contribution in the thermal conductivity calculation. For
this model, it was necessary to set the sizes of the two voids.
On the basis that suction values of 6 and 980 kPa were used as
an indication of macropores and the beginning point of
capillary condensation, respectively, the sizes of the voids
were determined by the water volumes during outflow to these
suction magnitudes using the drainage-path soil water-
characteristic curve.
As the inflection points depend on the soil under considera-
tion, these values could not be set as constant and there is no
well-established theory to explain them. Tarnawski and Leong
(2012) concluded that a series–parallel model-extension to
three-phase porous media did not produce a satisfactory
outcome. Kersten (1949) proposed empirical equations devel-
oped from thermal conductivity tests using 19 different soil
types. The thermal conductivity, λ, for the unfrozen condition
for silt and clay and for only sand is expressed in Eqs. (1) and
(2), respectively:

λ ¼ ð0:9 log w  0:2Þ  100:6242ρd  3:4628
418:6ðfor silt and clayÞ ð1Þ

λ ¼ ð0:7 log w þ 0:4Þ  100:6242ρd  3:4628
418:6ðfor sandÞ ð2Þ
where λ is the thermal conductivity of the soil (Wm  1 K  1),
w is the moisture content (%), and ρd is the dry density
(g/cm3).
 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]] 3

Johansen (1975) proposed an empirical equation that con-
siders the content of quartz, a mineral with high thermal
conductivity. The thermal conductivity of the soil particles, λs,
is defined using the quartz content by volume, θq, as
λs ¼ 7:7θq U21  θq
Thermal conductivity at the saturated condition, λsat, and that
after air drying, λdry, are calculated using Eqs. (4) and (5),
respectively. The thermal conductivity of soil under unsatu-
rated conditions is defined in Eq. (6) using the thermal
conductivities of soil under saturated and air-dried conditions,
and Kersten number Ke, namely,
λsat ¼ 0:57n U λ1s  n
therefore, the thermal probe method (an unsteady-state
method) is adopted.
Fig. 2(a) shows a schematic diagram of the thermal conductiv-
ity testing device, which consists of a thermal probe, a data
logger, and a DC power supply. A detailed drawing of the thermal
ð3Þ
probe is shown in Fig. 2(b), which consists of a copper pipe
(150 mm in height and 3 mm in diameter) and constantan wire as
a heater, and a T-type thermocouple. The remaining space was
ð4Þ

137ρd þ 64:7
λdry ¼ ð5Þ
2700  947ρd


λunsat ¼ λsat  λdry K e þ λdry ð6Þ

K e ¼ 0:7logðSr = 100Þþ 1:0

ðfor coarseÞ logðSr = 100 Þþ 1:0 ðfor fineÞ ð7Þ
Here, λsat, λdry, and λunsat are the thermal conductivity of the 75mm 3mm
saturated condition, that of the air-dried condition, and that of 150mm
the unsaturated condition, respectively (Wm  1 K  1), Ke is
the Kersten number, n is the porosity, ρd is the dry density Fig. 2. Thermal conductivity test device. (a) Schematic of thermal conductivity
(g/cm3), and Sr is the degree of saturation (%). test device. (b) Detailed drawing of thermal probe.
Johansen's equation has been improved by modifying the
Kersten number and the thermal conductivity for dry soil. For
instance, Côte and Konrad (2005) modified the Kersten
number using a soil-texture-dependent parameter and the
degree of saturation to exclude logarithmic functions. Lu et al.
(2007) also improved the Kersten number and the thermal
conductivity of dry soil.
Furthermore, new approaches have recently been developed
for predicting thermal conductivity using soil–water character-
istic curves. For example, Lu and Dong (2015) focused
attention on the strong relationship between thermal conduc-
tivity and the soil–water characteristic curve, and proposed a
closed-form equation by applying the van Genuchten model
(Van Genuchten, 1980). Fig. 3. Electric conductivity test device.
Yamazaki et al. (2003), however, indicated that these
empirical equations are not appropriate for the volcanic soils
on Hokkaido Island, Japan. Further studies are needed in order
to reveal the thermal conductivity of these soils and to
establish a method for estimating thermal conductivity,
because the applicability of these equations is limited.

3. Testing apparatus

3.1. Thermal conductivity tests

Steady-state and unsteady-state are the major categories of

thermal conductivity testing methods; the steady-state method
is not appropriate for measuring the thermal conductivity of
soils, since water redistribution due to the temperature gradient
in a specimen causes thermal conduction. In this study, Fig. 4. Grain size distribution.

Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
4 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]]

filled with paraffin. Paraffin has been used by some researchers to
keep the wire and thermocouple in position (Bi et al., 2005;
Nagano and Ochifuji, 1993). In this study, before paraffin was
adopted, it was confirmed that the paraffin would not melt under
the test temperature. The probe causes heat by applying a voltage
to the constantan wire, and the rise in temperature of the probe
can be measured by inserting a thermocouple. The thermal probe
method is based on the principle that the change in temperature of
the probe is a function of the specimen's thermal conductivity.
Thermal conductivity is calculated using Eq. (8):
 
Q t2
λ¼ ln ð8Þ
4πΔθ t1
where λ is the thermal conductivity (Wm  1 K  1), Q is the heat
per unit length (Wm  1), t1 and t2 are the time (s), and Δθ is the
increase in temperature between t1 and t2 (1C).
3.2. Electrical conductivity tests
The four-electrode technique is used to measure the elec-
trical resistibility. Fig. 3 illustrates the electrical conductivity
test device. The cylindrical cell, made of polyvinyl chloride, is
150 mm in height and 56 mm in diameter. Two supplying
electrodes (current electrodes) and two measuring electrodes
(potential electrodes) are used separately. The length between
the two measuring electrodes, L, is 100 (mm). The electrical
resistibility is calculated using Eq. (9), namely,
VA
r¼
IL
where r is the electric resistibility (Ωm), V is the voltage
measured between the two measuring electrodes (V), I is the
current measured between the measuring electrodes (A), and A
is the cross section of the area (mm2).
4.2. Thermal conductivity tests
The soils with adjusted moisture contents are put in stainless
steel cylindrical containers (180 mm in height and 72 mm in
diameter) and compacted to achieve the prescribed dry
densities as Toyoura sand: ρd ¼ 1.59 g/cm3, Touryo soil:
ρd ¼ 1.28 g/cm3, and Atsuma sand: ρd ¼ 1.83 g/cm3. To
achieve a constant temperature, each specimen is kept in a
constant-temperature water tank at 20 1C. After a constant
temperature is obtained, a voltage is supplied to the constantan
wire for 2 min and the temperature-increment of the probe is
measured. The heat per unit length is set to 12 W/m to control
the temperature-change to less than 6 1C during the test. An
example of temperature rising is shown in Fig. 5. The
relationship between the increase in temperature, Δθ, and
ln (t) is linear. The gradient of the relationship between the
increase in temperature and the elapsed time (Δθ/ln (t2/t1)) is
obtained by the least-squares method. Thermal conductivity is
calculated by substituting the gradient and the heat per unit
length into Eq. (8). Even if the duration is set to be over 2 min,
the gradients, which are calculated by the least-squares
method, are found to be almost the same. Therefore, the
duration is set to be 2 min in this study.
4.3. Electrical conductivity tests
Similar to the thermal conductivity tests, soil with an
adjusted moisture content is placed in a cylindrical cell
(150 mm in height and 56 mm in diameter) and compacted
ð9Þ
to obtain the same dry density as the thermal conductivity test

4. Testing method

4.1. Test samples

The material used in this study is Japanese standard sand

(Toyoura sand), volcanic cohesive soil of Kitami, Hokkaido,
Japan (hereafter referred to as Touryo soil), and sand of
Atsuma, Hokkaido, Japan (hereafter referred to as Atsuma
sand). The grain-size distribution curves and the physical
properties of the materials are shown in Fig. 4 and Table 1. Fig. 5. Increase in temperature.

Table 1
Physical properties.

Sample ρs ρd max ρd min emax emin D50 Fc θq Sm

Soil particle Maximum dry Minimum dry Maximum void Minimum void Average grain Fine Quartz Specific
density density density ratio ratio diameter content content surface
g/cm3 g/cm3 g/cm3 – – mm % % m2/g

Toyoura sand 2.65 1.65 1.35 0.96 0.61 0.18 0 69 0.5

Touryo soil 2.56 1.34 0.84 2.05 0.91 0.09 46.98 7 18.9
Atsuma sand 2.89 1.85 1.71 0.69 0.56 0.65 0 34 3.4

Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]]
specimen. After the sample-specimen preparation, (a) voltage
is applied to the electrodes (current electrodes) at both ends of
the specimen, and the current and the voltage are measured by
the current and the potential electrodes (Fig. 3). The electrical
resistibility is calculated by Eq. (9) using the measured voltage
and current.
5. Test results and discussion
Fig. 6 shows the thermal conductivities and electrical
resistibilities of three soil types: Toyoura sand, Touryo soil,
and Atsuma sand. The thermal conductivities of the soils
Fig. 6. Thermal conductivity and electrical resistibility: (a) Toyoura sand,
(b) Touryo soil, and (c) Atsuma sand.
Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
5
increase with the moisture-content growth. The thermal con-
ductivity of Touryo soil is lower than those of Toyoura and
Atsuma sand over the entire range of moisture contents. This is
thought to be due to the Touryo soil having a low quartz
content, compared to Toyoura or Atsuma sand (see Table 1),
as quartz is a high thermal-conductivity mineral. However, the
relationships between thermal conductivity and moisture con-
tent are nonlinear, and the change rate of the thermal
conductivity along with the moisture content is not constant,
as previous studies have shown. The thermal flow was
conducted through soil particles and pore water because the
thermal conductivity of air is low. Therefore, thermal con-
ductivity is likely to be proportional to the moisture content,
since the thermal conductivity of soil depends on its moisture
content, even with identical dry densities. For example,
Kamoshida et al. (2013) indicated results that were similar to
those of this study for the relationship between thermal
conductivity and the moisture content, and they presumed that
water at lower moisture contents does not contribute to thermal
Fig. 7. Definition of the inflection point of electrical resistivity.
Fig. 8. Thermal conductivity model.
6 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]]
conduction. This is thought to be because most pore water
does not continue at moisture contents lower than the
inflection point.
In this study, therefore, electrical conductivity tests were
conducted to examine the dependence of soil thermal con-
ductivity upon the moisture content, and thermal conductivity
is further compared with electrical resistibility. Even the
corresponding inflection points of the moisture content are
slightly different, but the inflection points of both thermal
conductivity and electrical resistibility approximately agree
with each other. The inflection point of electrical resistivity is
used in the thermal conductivity model developed in this
study. Details will be given in the next chapter. Similar
mechanisms are behind the nonlinear relationship between
moisture content and either thermal conductivity or electrical
resistibility. Considering the fact that the electrical current flow
along the pore water and the changing rate of electric
resistibility at an inflection point change significantly, pore
water seems to be continuous with the thin water layer in this
region. Furthermore, the inflection point of the thermal
conductivity of Touryo soil occurs at a higher water content
than in Toyoura or Atsuma sand. It can be presumed that a
great volume of water is required for most of the pore water to
be continuous, due to its high specific surface (see Table 1).
The conductivities of heat and electric currents depend on the
structural configuration of the pore water, even when the heat
and electric currents of the transmitting materials are different.
6. Estimation of thermal conductivity
The thermal conductivity and electrical resistibility of soils
are strongly connected, as shown in Section 5. In this section, a
thermal conductivity model and an estimation equation, which
have different roles, are proposed on the basis of test results.
6.1. Thermal conductivity model
Even with an inflection point (as shown in Fig. 6), some
thermal conductivity models are capable of expressing the
Fig. 9. For the case when: (a) θ is lower than θinf. and (b) θ is higher than θinf,.
Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
nonlinear thermal conductivity of soils due to the complicated
relationship between thermal conductivity and the moisture
content. However, the reason for this inflection point in the
thermal conductivity–moisture content relationship was not
properly clarified in previous studies. In this study, therefore,
the inflection point of thermal conductivity is defined as the
moisture content corresponding to the inflection point of
electrical resistibility. This definition is based on consideration
of the electrical resistibility inflection point as a water-
continuous point, because electricity mainly flows along the
pore water in the soil. The method for obtaining the inflection
point is as follows. Fig. 7 shows the change ratio of the electric
resistibility. Although the change ratio is slightly smaller in the
high-moisture region, there is a point at which the ratio
drastically changes. Starting from the higher moisture contents
and working down, the first point at which the gradient begins
to change is defined as an inflection point. Assuming that the
structural configuration of pore water changes at the electrical
resistibility inflection point, two types of pore water behavior
can be defined, namely, 1) water located near soil particles and
2) bulk water. A new series–parallel model for thermal
conductivity has been proposed using this definition. Fig. 8
shows the thermal conductivity model proposed in this study.
The model has three types of voids, defined as void 1(e1), void
Fig. 10. Thermal resistance.
 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]] 7

2(e2), and void 3(e3), respectively, to represent the three types as that used to calculate the electrical resistance, namely,
of heat flows related to water. The arrows in Fig. 8 illustrate
1 1 1 1
the heat flow in the model, where heat is conducted in 4 ways λ¼ þ þ þ
from Part 1 to Part 4. The heat flow in Part 1 is the heat flow R1 R2 þ R3 R4 þ R5 R6
between the soil particles, while that in Part 2 is the heat flow 1 1
¼ λs d 1 þ D 1 þ D3 þ λw d 4w þ λa d4a
between the soil particles and the nearby water. The heat flow þ D2
þ D4
λs d 2 λw d 2w þ λa d2a λs d 3 λw d3w þ λa d 3a
in Part 3 is the heat flow between the soil and the bulk water,
ð12Þ
while that in Part 4 is the heat flow only through the bulk
water. In this way, it is possible to represent three types of heat
flow related to water by having three different voids. The
electrical resistibility inflection point is defined as the pore
water continuous point. The moisture content corresponding to
this point is defined as θinf. When θ is lower than θinf, pore
water fills in e1 (Fig. 9(a)). In particular, e1 is obtained by
θinf/100; when θ is higher than θinf, e1 has already filled with
water, and water seems to become somewhat thicker for heat
conduction. In this study, assuming that the heat flows of Parts
3 and 4, shown in Fig. 8, occur simultaneously, the distribution
of water is set so as to fill both e2 and e3 evenly and
simultaneously (Fig. 9(b)). Therefore, in this study, the size
of e2 is assumed to be equivalent to that of e3. These are just
assumptions to simplify the model, since they cannot be
determined at the present stage.
The thermal conductivity of the soil can be calculated in
terms of the thermal conductivity models shown in Figs. 8–10.
Using a cube (d in width (m), D in length (m), unit length in
depth and thermal conductivity: λ (Wm  1 K  1)), thermal
resistance, R, can be calculated as follows:

D
R¼ ð10Þ
λUd

The thermal conductivities of soil particles, water, and air

are required to calculate the thermal conductivity using this
model. Even if the thermal conductivities of water and air are
known, the thermal conductivity of the soil particles cannot be
measured. This study uses Johansen's equation (1975) to
obtain the thermal conductivity of soil particles (Eq. (3)).
The thermal resistances in Fig. 10 are as follows:

1
R1 ¼
λs d 1
D1
R2 ¼
λs d 2
D2
R3 ¼
λw d2w þ λa d2a
D3
R4 ¼
λs d 3
D4
R5 ¼
λw d3w þ λa d3a
1
R6 ¼ ð11Þ
λw d4w þ λa d4a
The thermal conductivity of this model, shown in Fig. 8, is
an inverse function of the series–parallel thermal resistance,
since the model has the dimensions of a unit cube. Therefore, Fig. 11. Thermal conductivity calculated by Eq. (12): (a) Toyoura sand,
thermal conductivity can be calculated by the same procedure (b) Touryo soil, and (c) Atsuma sand.

Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
8 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]]

This equation depends on only two variables because which are not common tests. This is the disadvantage of the
porosity, n, and e1 are known (e1 ¼ θinf/100). In this study, proposed model. Even though there are some problems to solve,
d1 and D1 are treated as two variables which are estimated to fit these results suggest that the thermal conductivity model proposed
the experimental results in calculating the thermal conductivity in this study can explain the thermal conductivity of the soil.
with Eq. (12). The other parameters are obtained in the order
shown below: 6.2. Empirical equation from electrical resistibility
D2 ¼ 1 D1
The equation proposed by Kersten (1949), using dry
e1
d2 ¼ d 4 ¼ e3 density, is commonly used to estimate the thermal conductivity
D2
of soil in Japan. This study proposes an empirical equation
d3 ¼ 1  ðd1 þ d2 þ d 4 Þ
e2 based on electrical resistibility, using the strong relationship
D4 ¼ between thermal conductivity and the electrical resistibility of
d3
the soil:
D3 ¼ 1 D4 ð13Þ
λ ¼ α þ βr γ ð15Þ
where it is presumed that e2 ¼ e3. e 2 and e 3 are given by
where α, β, γ are the variable numbers and r is the electrical
1 n 
e2 ¼ e3 ¼  e1 ð14Þ resistibility.
2 100
Each thermal resistance is calculated by the above variables,
and the thermal conductivity can be obtained by Eq. (12).
Fig. 11 shows a comparison between the thermal conductivities
calculated by Eq. (12) and the experimental results. All of the
parameters are shown in Table 2. As in Fig. 11, the model
introduces an inflection point and the thermal conductivity can be
calculated by Eq. (12) using appropriate parameters. The para-
meters of the equations are further examined, comparing the
physical properties of the soils. The d1 and D1 values are compared
with the thermal conductivity of soil particles, λs, and the soil
particle surface area per unit volume, S, respectively, as shown in
Fig. 12. Soil particle surface area per unit S can be calculated as
S¼ Sm  ρd  106 using specific surface Sm. A linear approxima-
tion by the least-squares method is similarly inserted into Fig. 12.
d1 and D1 values are strongly correlated with the thermal
conductivity of the soil particles and their surface area, respectively,
because the determination coefficients are very high for both. The
proportion of heat flow in Part 1 was assumed to be a function of
the thermal conductivity of soil particles for d1. The surface area of
the soil particle was considered to be related to the amount of water
located near the soil particle for D1 in this study. As for the soils
used in this study, therefore, it is practical to assume that
parameters, which have been established to match the experimental
results, are appropriate values due to their correlations with soil
physical properties. As mentioned in Chapter 2, it is difficult to
determine the size of the model. One of the objectives of this paper
is to present a way to determine the size of the void using electrical
resistivity. Moreover, this study aims to associate values with
physical properties. Together, these constitute the advantages in
comparison with previous studies. On the other hand, obtaining S Fig. 12. Comparison of physical property with: (a) the value of d1 and (b) the
and λs requires XRD and the measurement of the specific surface, value of D1.

Table 2
Parameters.

Sample λs d1 d2 d3 d4 D1 D2 D3 D4 ρd (g/cm3) n(%) e1 e2 þ e3 θinf (%)

Toyoura sand 5.07 0.04 0.63 0.18 0.15 0.84 0.16 0.19 0.81 1.59 40.0 0.10 0.15 10
Touryo soil 2.20 0.10 0.70 0.10 0.10 0.57 0.43 0.07 0.93 1.28 50.0 0.30 0.10 30
Atsuma sand 3.16 0.07 0.48 0.32 0.13 0.77 0.23 0.60 0.40 1.83 36.7 0.11 0.13 11

Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]] 9

α was defined as a fitting parameter with the assumption that

β and γ are constant values. In particular, α is treated as a soil-
dependent variable and β and γ are treated as soil-independent
constants. Fig. 13 explains the best-fitted thermal conductiv-
ities calculated by Eq. (15). The parameters, by which best-
fitted thermal conductivity was obtained, will be discussed
later by comparing them with physical properties to consider

Fig. 14. Comparison of electric resistibility with the value of α.

the applicability of the proposed equation. The thermal

conductivities calculated by Kersten's and Johansen's equations
are also shown in Fig. 13 for comparison with previous
empirical equations. Fig. 14 shows the relationship between
α, which was first examined, and the electrical resistibility
under saturated conditions. There is considerable validity in
the empirical equation using electrical resistibility, even
though constant values for α, β, γ should be investigated to
reveal in detail the relationship between the thermal conduc-
tivity and the electrical resistibility of the soil.
Empirical equations for the three soil types were also
examined in this study. The thermal conductivity of Atsuma
sand is overestimated by Kersten's equation. The use of only
dry density to estimate thermal conductivity may be the key
reason for this overestimation. The dry density of Atsuma sand
is high compared to that of normal soil due to the high soil-
particle density. The difference between measured and esti-
mated values may occur due to high dry density. Johansen's
equation agrees well with the results for the three soil types
used in this study. Johansen's equation, however, requires the
quartz content, which must be measured by X-ray diffraction.
Therefore, it is not convenient to estimate thermal conductivity
using Johansen's equations. Thermal conductivity can poten-
tially be obtained by the proposed equation using only the
electrical conductivity test, which is simple and low cost.
According to the results of this study, thermal conductivity
strongly correlates to electrical resistibility and, in fact, can be
estimated with electrical resistibility.

7. Conclusions

The following findings have been obtained in this study:

– The relationship between thermal conductivity and the

Fig. 13. Thermal conductivity calculated by Eq. (15): (a) Toyoura sand,
(b) Touryo soil, and (c) Atsuma sand.
moisture content has an inflection point, and the values of
the inflection points correspond well with each other for
different soil types. The water dependencies of thermal
conductivity and electrical resistibility have similar
mechanisms.

Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
10 T. Tokoro et al. / Soils and Foundations ] (]]]]) ]]]–]]]
– The moisture content corresponding to the electrical resistibility
inflection point is defined as a pore water continuous point. A
series–parallel model was proposed using this definition. As for
the soil used in this study, the proposed model with appropriate
parameters is able to express experimental thermal conductiv-
ity, which has an inflection point.
– An empirical equation, based on the electrical resistibility,
was also proposed. According to the results, thermal
conductivity can be calculated only with knowledge of the
electrical resistibility for the soils used in this study.
A direction for future research will be to provide more
evidence for equations and to conduct thermal conductivity
and electrical resistibility tests for more soil types, especially
fine soils. Verification of the generality of the equations
proposed in this study will be an issue in the future because
the vapor pressure technique or other techniques for unsatu-
rated soil, which take a long time, will be adopted to obtain
specimens of fine soils in homogenous water distributions.
References
Bi, C., Suzuki, T., Sawada, S., Yamashita, S., 2005. Study on thermal
conductivity of granular materials used for a countermeasure against frost
heave. Proc. Jpn. Soc. Civil. Eng. 785, 83–92 (in Japanese).
Côte, J., Konrad, J.M., 2005. A generalized thermal conductivity model for
soils and construction materials. Can. Geotech. J. 42, 443–458.
De Vries, D.A., 1952. A non-stationary method for determining thermal
conductivity of soil in situ. Soil Sci. 73, 83–89.
De Vries, D.A., Peck, A.J., 1958. On the cylindrical probe method of
measuring thermal conductivity with special references to soils. Aust.
J. Phys. 11, 255–271.
Hopper, F.C., Chang, S.C., 1953. Development of the thermal conductivity
probe. Trans. Am. Soc. Heat. Vent. Eng. 59, 463–472.
Johansen, O., 1975. Thermal Conductivity of Soils (PhD thesis). Trondheim
University, Norway.
Please cite this article as: Tokoro, T., et al., Estimation methods for thermal conductivity of sandy soil with electrical characteristics. Soils and Foundations
(2016), http://dx.doi.org/10.1016/j.sandf.2016.08.016
Kamoshida, T., Hamamoto, S., Kawamoto, K., Sasaki, T., Komatsu, T., 2013.
Thermal properties of sands at variably-saturated condition: effects of
particle size and shape, and quartz content. J. Jpn. Soc. Soil Phys. 124,
11–16.
Kasubuchi, T., 1975. The effect of soil moisture on thermal properties in some
typical Japanese upland soils. Soil Sci. Plant Nutr. 21 (2), 107–112.
Kasubuchi, T., 1977. Twin transient-state cylindrical-probe method for
determination of the thermal conductivity of soils. Soil Sci. 124 (5),
255–258.
Kersten, M.S., 1949. Thermal Properties of Soils. University of Minnesota
Institute of Technology, Engineering Experiment Station, Bulletin, 28.
Kersten, M.S., 1952. Thermal Properties of Soils. 2. Highway Research Board
Special Report161–166.
Lu, N., Dong, Y., 2015. Closed-form for thermal conductivity of unsaturated
soils at room temperature. J. Geotech. Geoenviron. Eng. 141 (6),
04015016.
Lu, S., Ren, T., Gong, Y., Horton, R., 2007. An improved model for predicting
soil thermal conductivity from water content at room temperature. Soil Sci.
Soc. Am. J. 71 (1), 8–14.
Matsumoto, J., Ohkubo, T., 1977. Experimental study on heat transfer
characteristics of soils. Proc. Jpn. Soc. Civil Eng. 257, 53–60 (in Japanese).
Mitsuno, T., Ishida, T., Maruyama, T., 1983. Model for soil heat conduction
under a three-phase condition and calculation of soil heat conductivity.
Trans. Jpn. Soc. Irrig. Drain. Reclam. Eng. 103, 35–43.
Nagano, K., Ochifuji, K., 1993. Changes of Thermal Properties of Soils
Caused by Freezing and Melting and a Method for Calculating the Thermal
Conductivity, 52. Transactions of the Society of Heating, Air-Conditioning
and Sanitary Engineers of Japan91–102 (in Japanese).
Tarnawski, V.R., Leong, W.H., 2012. A series-parallel model for estimating
the thermal conductivity of unsaturated soil. Int. J. Thermophys. 33 (7),
1191–1218.
Van Genuchten, M.T., 1980. A closed-form equation for predicting the
hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44 (5),
892–898.
Woodside, W., Messmer, J.H., 1961. Thermal conductivity of porous media.
J. Appl. Phys. 32 (9), 1688–1699.
Wyllie, M.R., Southwick, P.F., 1954. An experimental investigation of the S.P.
and resistivity phenomena in dirty sands. J. Pet. Technol. 6 (2), 44–57.
Yamazaki, Y., Tsuchiya, F., Tsuji, O., 2003. Measurement of thermal
conductivity of quarts-containing frozen and unfrozen soils. Trans. Jpn.
Soc. Irrig. Drain. Reclam. Eng. 226, 43–51 (in Japanese).