1-s2.0-S0013795221001186-main

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Engineering Geology 288 (2021) 106107

Contents lists available at ScienceDirect

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

Modeling thermal conductivity of clays: A review and evaluation of 28


predictive models
Lanmin Liu a, Hailong He a, b, *, Miles Dyck c, Jialong Lv a, d
a
College of Natural Resources and Environment, Northwest A&F University, Yangling 712100, China
b
State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, CAS&MWR, Northwest A&F University,
Yangling, Shaanxi 712100, China
c
Department of Renewable Resources, University of Alberta, Edmonton, Alberta T6G 2E3, Canada
d
Key Laboratory of Plant Nutrition and the Agri-Environment in Northwest China (Ministry of Agriculture), Northwest A&F University, Yangling 712100, China

A R T I C L E I N F O A B S T R A C T

Keywords: Effective thermal conductivity of clay (λeff) has essential applications in disciplines such as environmental and
Effective soil thermal conductivity earth science, agriculture and engineering. Much work has been done pertaining to model thermal conductivity
Normalized model of clay, but no work was found to collate and synthesize these works. This study aimed to conduct an extensive
Transient heat pulse method
review of the predictive models for thermal conductivity of clay and evaluate their performance with a large
Literature review
compiled dataset. A total of 28 models were collated and categorized and their performance was evaluated with a
Model comparison
large dataset consisting of 1250 measurements made on 65 clays from 21 studies. The result showed that the
DF1979 was the best performing model but not satisfactory, with root-mean-square-error (RMSE) = 0.35 W
m− 1 ◦ C− 1, average deviation (AD) = − 0.04 W m− 1 ◦ C− 1 and Nash-Sutcliff Efficiency (NSE) = 0.54. Further
investigations showed that these models are dataset dependent and fairly good performance might be found on
certain soils/dataset. Therefore, cares should be taken when selecting the most appropriate model for predicting
λeff in practice. Limitations of the thermal conductivity models for clays have been stated and perspectives on
future studies were also presented. This work would provide the novice or expert alike information on the
advantage and limitations on modeling thermal conductivity of clays for various purposes.

1. Introduction and landfill (Ghorbel-Abid and Trabelsi-Ayadi, 2015; Kaya and Duru­
kan, 2004).
Clays differ from other soils in many aspects including particle size Effective thermal conductivity of clay (λeff), reciprocal of thermal
and shape, specific surface area and mineralogy (Farrar and Coleman, resistivity, is of critical importance and interest among these thermo-
1967; Hepper et al., 2006; Maček et al., 2013; Paterson, 2018), which hydro-mechanical characteristics. Because λeff of clay is required to de­
make it special in the thermal, hydraulic and mechanical characteristics termines heat transmission through soil in order to optimize engineering
(Kang et al., 2015; María and Villar, 2004; Mozumder and Laskar, 2015; design (He et al., 2021; Reno and Winterkorn, 1967; Wang et al., 2016;
Tang et al., 2011; Zhang et al., 2017a). With respect to this, clay has Xu et al., 2019b). Although great progress has been made in techniques
been widely used in disciplines such as environmental and earth science, of λeff measurement (Bristow et al., 1994; Campbell et al., 1991; He et al.,
agriculture and geotechnical/geological engineering. These applications 2018a; He et al., 2015; He et al., 2018b; Ren et al., 2003; Wang et al.,
include water treatment (Chang and Juang, 2004), backfilling material 2020a; Zhang et al., 2017b), λeff data are still rare and sparse compared
for buried high-voltage power cable and oil, gas or hot-water pipes to that of soil hydraulic and mechanical properties. Therefore, many
(Newson et al., 2002; Reno and Winterkorn, 1967; Ye et al., 2010) or efforts have been made to model λeff of clays (Abuel-Naga et al., 2009;
borehole heat exchanger (Luo et al., 2019; Yu et al., 2015), natural Akinyemi et al., 2011; Barry-Macaulay et al., 2011; Caridad et al., 2014;
geological engineering barrier for hazardous waste disposal including Dao et al., 2014; de Zárate et al., 2010; Garnier et al., 2010; Jougnot and
radioactive nuclear waste (Gens et al., 2002; Madsen, 1998; Mügler Revil, 2010; Kang et al., 2015; Sun et al., 2020; Tang et al., 2008; Xu
et al., 2006; Tang and Cui, 2007; Villar et al., 2005; Yoon et al., 2018) et al., 2019d). However, it is interesting to note that there is a lack of

* Corresponding author.
E-mail addresses: liulanmin@nwafu.edu.cn (L. Liu), hailong.he@hotmail.com (H. He), miles.dyck@ualberta.ca (M. Dyck), ljlll@nwsuaf.edu.cn (J. Lv).

https://doi.org/10.1016/j.enggeo.2021.106107
Received 2 April 2020; Received in revised form 5 December 2020; Accepted 22 March 2021
Available online 2 April 2021
0013-7952/© 2021 Published by Elsevier B.V.
L. Liu et al. Engineering Geology 288 (2021) 106107

study that tends to collate and synthesize these works. = [0, 0, 1] is used in this study for predicting thermal conductivity of
It is also noted that many of these thermal conductivity models for clay. In addition, Yan et al. (2019) evaluated the use of air, water and
clay were commonly developed or tested on a small dataset composing 2 solid as the continuous phase and their results showed that the de Vries
to less than 200 λeff measurements (Cai et al., 2015; de Zárate et al., (1952b) model with solid as the continuous phase performed the best
2010; Xu et al., 2019d). In addition, there is a lack of study assessing and this will used in this study.
these models with a large dataset consisting of clays with wide range soil
physic properties (e.g. water content, mineralogy, and bulk density). 2.1.2. Modified de Vries, 1963 (DV1963II) model
Previous studies (de Vries, 1952a; He et al., 2015; He et al., 2018b; Wang Tian et al. (2016) simplified the de Vries (1963) model, Eq. (1), by
et al., 2020a) have shown that soil thermal conductivity measured with introducing thermal conductivities of sand, silt, and clay (λsa, λsi and λcl)
transient methods (e.g., heat pulse probe, thermal needle, and hot-wire and shape factors of sand, silt, clay, air, and ice particles (ga(sand), ga(silt),
method) is more accurate for soils from dryness to saturation compared ga(clay), ga(air) and ga(ice)). While the water was assumed to be continuum,
to steady-state method (e.g., guarded hot plate, divided bar method, and and ellipsoidal particles were confocal.For the soil solid phase,
heat flux meter). The steady-state methods result in phase change of ice
for frozen soils or water redistribution in unfrozen soils under temper­ λs = λfsasa λfsisi λfclcl (3)
ature gradient, especially in unsaturated soils. Therefore, λeff measured
ga(solid) = ga(sand) fsa + ga(silt) fsi + ga(clay) fcl (4)
with the transient methods are recommended for development and
calibration of predictive models for soil thermal conductivity (He et al.,
where fsa, fsi and fcl are the fraction of sand, silt, and clay for soil solids
2018b; Zhang et al., 2017b).
(gravimetric %). For this work, λsa = 7.70 (W⋅m− 1⋅K− 1), λcl = 1.93
The objectives of this study, therefore, were threefold: (1) to conduct
(W⋅m− 1⋅K− 1), λsi = 2.74 W⋅m− 1⋅K− 1, ga(sand) = 0.182, ga(clay) = 0.00775
an extensive review of the predictive models for thermal conductivity of
and ga(silt) = 0.0534.. For the soil air phase,
clay; (2) to compile a large dataset consisting of clays of various types
from dry to full saturation states and at different bulk densities; (3) to 1( θair )
ga(air) = 1− (5)
evaluate the performance of the thermal conductivity models for clay 3 P
with the compiled dataset. It is hoped that this work would provide the
novice or expert alike information on the advantage and limitations on where P is porosity and θair is volumetric content of air. The simplified
modeling thermal conductivity of clays for various purposes. model presented by Tian et al. (2016) does not require calibration
because all of the parameters are specified priori.
2. A review of thermal conductivity models for clays
2.1.3. Sakashita and Kumada, 1998 (SK1998) model
Sakashita and Kumada (1998) developed a heat transfer model for
The extensive literature search returned 28 thermal conductivity
compacted bentonite accounting for effects of soil’s microstructure. The
models for clays. These models were categorized into three groups: (1)
unknown constants were determined empirically and the model was
theoretical/semi-theoretical models (N = 2) and mixing models (N = 4);
then modified as
(2) normalized models (N = 6); and (3) linear and non-linear regression
{ }
models (N = 14). The initials and year of publication were used to
λeff = λdry 1 + [(9.75P − 0.706)Sr ]0.285P+0.731 (6)
represent the model for better science communications. It should note
that only models developed for clays were presented and evaluated in
where P is porosity, Sr is the degree of saturation, λdry is thermal con­
this study.
ductivity of dry bentonite that is calculated by

2.1. Theoretical/semi-theoretical and mixing models λdry = 0.0479 + 0.222(1 − P) + 0.968(1 − P)3 (7)

2.1.1. de Vries, 1963 (DV1963) model 2.1.4. Jougnot and Revil, 2010 (JR2010) model
_and de Vries (1952b) adapted the Maxwell-Eucken equation Given the analogy between thermal conductivity and electrical
(Eucken, 1932; Maxwell, 1881) for electric conductivity (of solute so­ conductivity, Revil (2000) developed a differential effective medium
lutions) to predict λeff. The de Vries model assumes that ellipsoidal approach to evaluate λeff of 2-phase system by applying the Bruggeman-
particles randomly distributed in continuous phase and it is expressed Hanai-Sen approach (Sen et al., 1981), which was extended for 3-phase
as: system by Jougnot and Revil (2010).
/ [ ( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) ]
∑N -1 N -1
∑ λf 1
λeff = Fi θ i λ i Fi θ i (1) λ= FΘ + (1 − Θ) 1 − Θ + (1 − Θ)2 + 4FΘ (8)
F 2
i=0 i=0

where F = P m/(1-m) and Θ = λs/λf, λf is the thermal conductivity of the


where N is the number of soil components (e.g. solid, water and air) with
fluid phase, which is calculated by a generalized second Archie law
subscript i = 0 representing the continuous phase; θi is the volume
(Jougnot and Revil, 2010)
fraction of each phase; λi is the thermal conductivity of each phase,
W⋅m− 1⋅K− 1; Fi is the percentage of the average temperature gradient
( )
λf = λw Srn + 1 − Srn λair (9)
across each phase, which is estimated by the following equation (de
Vries, 1963) where λw and λair represent thermal conductivity of water and air, m is
[ ( ) ]− 1 the second Archie exponent, m = 2.0 and n = 2 were recommended in
1 ∑ λi
Fi = 1+ − 1 gj (2) their original study.
3 j=a,b,c λ0
2.1.5. Donazzi et al., 1979 (DF1979) model
where ga, gb, and gc are depolarization factors (particle shape factors) Donazzi et al. (1979) proposed an exponential relationship for esti­
that depend on relative lengths of the major and minor axes of the mating λeff of various soils including clays
dispersed soil particles. For instance, special cases of gj = [1/3, 1/3, 1/ [ ]
3], gj = [0,0,1], and gj = [0.5, 0.5, 0] represent spherical (e.g., sand), λeff = λPw λ1−s P ⋅exp − 3.08P⋅(1 − Sr )2 (10)
discs-shaped (e.g., clay), and needle-shaped (e.g., silt) particles,
respectively (He and Dyck, 2013; Sihvola and Kong, 1988). Therefore, gj where λs indicates thermal conductivity of soil particles, Donazzi et al.

2
L. Liu et al. Engineering Geology 288 (2021) 106107

1 ◦ -1
(1979) used λs = 2.3 W m− C for clays. m− 1 ◦ C− 1 (λother = 3 for fquartz ≤ 20% at 20 ◦ C) (Johansen, 1977). It was
assumed that λeff = λdry when Sr < 0.1 in this study to avoid numeric
2.1.6. Kiyohashi and Banno, 1995 (KB1995) model error of the logarithm function used to calculate Ke.
Kiyohashi and Banno (1995) extended the Woodside and Messmer
(1961) model for 3-phase bentonite system (solid, water, and air) (Ould- 2.2.2. Knutsson, 1983 (KS1983) model
Lahoucine et al., 2002): Based on the model of Johansen (1975), Knutsson (1983) investi­
gated the measurement of metal film for measuring of λeff of highly
λeff = (1 − P)− θair /P
⋅[λs (1 − P) + λw P ]1− P ⋅λθwv ⋅λθairair (11) compacted pure Mx-80 bentonite, assuming λw = 0.56 W m− 1 ◦ C− 1 and
λs = 2 W m− 1 ◦ C− 1, which resulted in
where θv and θair are volumetric fraction of water and air, respectively, λs
( )
is calculated based on temperature (◦ C) and bulk density (kg m− 3) of λeff = 0.034P− 2.1 + (1 + log10 Sr ) 0.56P 21− P − 0.034P− 2.1 (16)
bentonite.
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ 2.2.3. Côté and Konrad, 2005 (CK2005) model
λs = 5.71 exp( − 0.00129T)⋅exp( − 0.679ρb ) (12)
Côté and Konrad (2005) improved the Johansen (1975) model by
where T is temperature (◦ C). introducing an texture-dependent factor (k). The resulted equation is
k⋅Sr ( P 1− P )
λeff = λ λ − χ 10− η⋅P
+ χ 10− η⋅P
(17)
2.1.7. Gens et al., 2009 (GA2009) model 1 + (k − 1)Sr w s
Gens et al. (2009) proposed a geometric mean model to estimate λeff
of clays where the parameter k = 1.9 and 0.85 for silts and clays at unfrozen and
frozen conditions, respectively. χ = 0.75 W m− 1 ◦ C− 1, η = 1.2 for natural
λeff = λSsatr λ1−drySr (13)
mineral soils, and λs was calculated in the same way as Johansen (1975).
Eq. (17) was incorporated to DOS-TEM to estimate λeff by Yi et al. (2013)
where λsat is thermal conductivity of saturated clays. Values of λdry =
on the Qinghai-Tibetan Plateau.
0.5205 W m− 1 ◦ C− 1, λsat = 1.234 W m− 1 ◦ C− 1 were recommended by
Yoon et al. (2018) with R2 = 0.83.
2.2.4. Lee et al., 2016 (LJ2016) model
Lee et al. (2016) proposed the following equation for estimating λeff
2.1.8. Cho et al., 2011 (CW2011) model
of compacted bentonite
Cho et al. (2011) proposed a modified geometric mean model for
estimating λeff of compacted bentonite and bentonite-sand mixture: λeff =
λdry − λsat
[( )/ ] + λsat (18)
1 + exp Sr − Savg A
2.099ϕbentonite ( θv θair )A
(14)
1.303ϕ
λeff = λsand sand λbentonite λw λair
where Savg is the degree of saturation when λeff = 1/2(λdry + λsat), and A is
where λsand and λbentonite are thermal conductivity of sand and bentonite, fitting parameter. Values of Savg = 0.6906, λdry = 0.6608 W m− 1 ◦ C− 1,
respectively, λsand = 4.302 W m− 1 ◦ C− 1 and λbentonite = 1.232 W m− 1 ◦ C− 1 λsat = 1.2410 W m− 1 ◦ C− 1 and A = 0.0878 were recommended by Yoon
were used by Cho et al. (2011), ϕsand = fsand ρb/ρs and ϕbentonite = fbentonite et al. (2018) with R2 = 0.91.
ρb/ρs are volumetric fraction of sand and bentonite, respectively; fsand
and fbentonite are mass fraction of sand and bentonite, respectively; ρs is 2.2.5. RÓżański and Stefaniuk, 2016 (RS2016) model
particle density (g cm− 3); A = 0.564 were suggested with R2 = 0.83 in RÓżański and Stefaniuk (2016) modified Johansen (1975) model
the original study. with a new equation for calculating λs

λs = a1 aSSA
2 + a3 (19)
2.2. Normalized models
where SSA is specific surface area (m2 g− 1), a1 = 5.7, a2 = 0.988, a3 =
2.2.1. Johansen, 1975 (JO1975) model 2.0 are the best fitting parameters used by RÓżański and Stefaniuk
Johansen (1975) proposed the normalized concept for estimating λeff (2016).
( )
λeff = Ke λsat − λdry + λdry (15) 2.2.6. Zhang et al., 2016 (ZN2016) model
Zhang et al. (2016) further modified the Côté and Konrad (2005)
where, Ke = Log10Sr + 1 is the Kersten number or the normalized thermal
model with experimental data measured on sand-kaolin clay mixtures
conductivity, Sr = θv/P, λdry λdry = (0.135ρb + 0.0647)/(ρs − 0.947ρb)
with thermo-TDR method, the resulted formula is
and λsat λsat = λPwλ1− s
P
. λs is calculated based on mass fraction and thermal
conductivity of quartz (fquartz, λquartz) and other minerals (λother),λs =
λfquartz
quartz
λ1− fquartz
other with λquartz = 7.7 W m− 1 ◦ C− 1 at 20 ◦ C and λother = 2.0 W

[ ]
2.168 × 10− 5 e(fquartz /7.903) + 1.252 Sr
λeff = [ ] ⋅
1 + 2.168 × 10− 5 e(fquartz /7.903) + 0.252 Sr
{ [ ] }
( )1− P [ ] − 0.003exp(
fquartz /16.452) − 1.84 P (20)
1.216 × 10 exp(fquartz /6.599) + 3.034 × 10
fquartz 1− fquartz
λPw λquartz λkaolin − − 6

[ ]
[ ] − 0.003exp(
fquartz /16.452) − 1.84 P
+ 1.216 × 10 e(fquartz /6.599)− 6
+ 3.034 × 10

3
L. Liu et al. Engineering Geology 288 (2021) 106107

where λquartz = 7.5 W m− 1 ◦ C− 1 and λkaolin is thermal conductivity of 2.3.6. Campbell, 1985 (CG1985) model
kaolin (2.9 W m− 1 ◦ C− 1), Kaolin is assumed to be clay in this study, and McInnes (1981) measured λeff of five field soils from the eastern
λs was calculated in the same way as Johansen (1975). Washington state (USA) and developed an empirical function:
[ ]
λeff = A + Bθv − (A − D)exp − (Cθv )E (26)
2.3. Linea/non-linear thermal conductivity models
where A, B, C, D, and E are empirical parameters. Campbell (1985)
2.3.1. Kersten, 1949 (KM1949) model demonstrated that parameters of Eq. (26) are dependent on and can be
Kersten (1949) proposed an empirical model to predict λeff of un­ related to the volumetric fraction of soil components (e.g., minerals,
frozen silt/clay: organic matter, and clay), they can be calculated by (Campbell, 1985;
( )
λeff = UC1 0.9logθg − 0.2 × 100.01UC2 ⋅ρb (21) Tarnawski and Leong, 2012):
0.57 + 1.73ϕquartz + 0.93ϕother
where UC1 = 0.144228 and UC2 = 62.427961 are the parameters used to A=
1 − 0.74ϕquartz − 0.49ϕother
− 2.8ϕs (1 − ϕs )
convert from “Btu in ft-2 h-1 ◦ F-1” to “W m-1 ◦ C-1” and from “g cm− 3” back
to “lb ft.− 3”, respectively; θg is the total soil water content on gravimetric B = 2.8ϕs D = 0.03 + 0.7ϕ2s (27)
basis (in %mass); ρb is dry bulk density (g cm− 3), volumetric water /√̅̅̅̅̅̅̅
content θv = 0.01 θg/ ρb. λeff = 0.25 W m− 1 ◦ C− 1 was assumed when θg < C = 1 + 2.6 fclay E=4
1.7% to avoid negative values of estimated λeff.
where and total volume fraction of solids ϕs = ϕquartz + ϕother. For soils
2.3.2. Makowski and Mochlinski, 1956 (MM1956) model with no or negligible quartz content and soil particle density ρs = 2.65 g
Makowski and Mochlinski (1956) presented an empirical model cm− 3, Eq. (27) can be simplified to A = 0.65 − 0.78ρb + 0.60ρ2b , B =
similar to the Kersten (1949) model, 1.06ρb, and D = 0.03 + 0.10ρ2.b Campbell (1985) model has been widely
[ ] used (Ross, 2014; Steele-Dunne et al., 2010) and been incorporated into
λeff = alog10 θg + b 100.01UC2 ρb (22)
numerical simulation models such as the Soil–Vegetation–Atmosphere
Scheme (SVATS), HYDRUS (Simunek et al., 1997), and Soil-Litter-Iso
where, a and b are related to clay content (fclay, %), a = 0.142408 − 4.65
(SLI) (Haverd and Cuntz, 2010).
× 10− 4fclay, b = 0.04192 − 3.13 × 10− 4fclay. λeff = 0.25 W m− 1 ◦ C− 1 was
assumed when θg < 1.7% to avoid negative values of estimated λeff with
2.3.7. Chung and Horton, 1987 (CH1987) model
the MM1956 model.
Chung and Horton (1987) proposed an empirical relationship for λeff
√̅̅̅̅̅
2.3.3. van Rooyen and Winterkorn, 1959 (VW1959) model λeff = a + bθv + c θv (28)
van Rooyen and Winterkorn (1959) developed an empirical equation
to predict λeff as a function of the Sr, ρb, mineral type, and particle shape where, a, b, and c are parameters pertaining to soil physical properties
(Fricke et al., 2002) including bulk density, porosity, and organic matter, with a = − 0.197, b
( ) = − 0.962, c = 2.521 for clay. The scheme of Chung and Horton (1987)
λ−eff1 = 0.01 A10− BSr + S (23)
has been incorporated in HDS-SPAC, with one parameter for loam was
used (Yang, 2015) and HYDRUS (Simunek et al., 1997). λeff = 0.25 W
where coefficient 0.01 was to convert “cm ◦ C W-1” to “m ◦ C W-1”,
m− 1 ◦ C− 1 was assumed when θv < 0.04 to avoid negative values of
parametersA = 10a1− 0.44ρb2, B = b1 − 5.5ρb, S = S1 − S2ρb, a1, b1, S1, and
estimated λeff.
S2 are functions of soil texture and mineral components. For clays, a1 =
3.55, b1 = 5.6 × 10-4fclay + 9.58, S1 = 317 and S2 = 155, b1 is assumed to
2.3.8. Becker et al., 1992 (BB1992) model
be 9.58 if fclay is not available (Farouki, 1982).
Becker et al. (1992) developed an unified methodology for predicting
λeff of gravel, sand, silt, clay and peat in both the frozen and unfrozen
2.3.4. Oriol et al., 1978 (OF1978) model
states using literature data (Dissanayaka et al., 2012)
Oriol et al. (1978) proposed an empirical model for expanded clay
[ ( ) ]
with R = 0.73 Sr = 0.01a1 sinh 6.9348a2 ⋅λeff + a3 − sinh(a4 ) (29)
λeff = 0.1UC3 (0.013θv + 0.558) (24)
or

{ [ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ] }/
λeff = UC1 log [(100Sr )/a1 + Sinh(a4 ) ] + [(100Sr )/a1 + Sinh(a4 ) ]2 + 1 − a3 a2 (30)

where coefficient 0.1 is used to convert “mW cm-1 ◦ C -1”to “W m-1 ◦ C -1”, where 0.01 is the coefficient to covert the Sr to be decimal value ranging
UC3 = 4.1868 is used to covert “cal cm-1 ◦ C-1 s-1” to “W m− 1 ◦ C -1”. from 0 to 1, λeff is multiplied by UC1 = 1/6.9348 = 0.1442 to convert the
SI unit (W m− 1 ◦ C− 1) back to “Btu in. ft-2 hr-1 ◦ F-1” for consistence with
2.3.5. Kahr and Müller-Vonmoos, 1982 (KV1982) model the original study. a1 to a4 are coefficients depending on soil types. For
Kahr and Müller-Vonmoos (1982) investigated λeff of Mx-80 Na- clay, the values of parameter a1 to a4 are 27.0, 0.265, − 1.5 and − 0.97,
bentonite and Montigel Ca-bentonite and developed the following respectively.
expression
2.3.9. Thomas et al., 1994 (TH1994) model
λeff = − 0.56 + 0.6ρb + 0.004θg ⋅ρ3b (25) Thomas et al. (1994) presented an empirical equation for analysis of
coupled water and heat transfer in unsaturated clay

4
Table 1

L. Liu et al.
Selected soil properties of the compiled dataset.
No. Literature source Soil Name No. Meas. Temp Soil texture fquartz ρs ρb P Water content Method

fsand fsilt fclay θg θv Sr


3

C —%— % —g cm− — % cm3cm− 3
1 Barry-Macaulay et al. (2013) Basaltic Clay 15 25.00ab 0.00 40.00 60.00 57.00 2.67 1.04–1.71 0.36–0.58 3.7–32 0.04–0.46 0.07–0.99 Thermal needle probe
2 Beziat et al. (1988) Ca-Smectite 19 25.00a 0.00a 0.00a 100a 7.00 2.65a 1.40–2.00 0.25–0.55 6.00–17.00 0.10–0.31 0.24–0.95 hot-wire method
3 Beziat et al. (1988) Illite 5 25.00a 0.00a 0.00a 100a 7.00 2.65a 1.20–1.60 0.40–0.55 12.00–17.00 0.17–0.27 0.36–0.69 hot-wire method
4 Beziat et al. (1988) Palygorskite 4 25.00a 0.00a 0.00a 100a 10.00 2.65a 1.40–1.80 0.32–0.47 6.00–12.00 0.10–0.22 0.24–0.67 hot-wire method
5 Beziat et al. (1988) Ca-Smectite 52 25.00a 0.00a 0.00a 100a 7.00 2.65a 0.90–2.10 0.25–0.58 5.00–18.00 0.06–0.33 0.11–1.04 hot-wire method
6 Beziat et al. (1988) Na-Smectite 20 25.00a 0.00a 0.00a 100a 15.00 2.65a 1.35–1.96 0.26–0.49 6.00–17.00 0.10–0.30 0.26–0.86 hot-wire method
7 Beziat et al. (1988) Illite 14 25.00a 0.00a 0.00a 100a 7.00 2.65a 1.36–1.67 0.37–0.49 6.00–12.00 0.08–0.19 0.17–0.49 hot-wire method
8 Beziat et al. (1988) Palygorskite 20 25.00a 0.00a 0.00a 100a 10.00 2.65a 1.12–1.57 0.45–0.58 8.00–25.00 0.09–0.36 0.16–0.80 hot-wire method
9 Börgesson et al. (1994) Mx-80 11 25.00a 0.00 0.00 100.00 3.00a 2.65a 1.69–1.97 0.45–0.47 13.20–28.40 0.24–0.56 0.45–0.97 Thermal probe
10 Brigaud and Vasseur (1989) Clayey sandstones 2 25.00a 0.00a 0.00a 100a 59.70 2.67 2.18–2.45 0.09–0.18 0.00 0.00 0.00 needle probe
11 Brigaud and Vasseur (1989) Limestones 12 25.00a 0.00a 0.00a 100a 0.00 2.69 1.41–2.69 0.01–0.48 0.00 0.00 0.00 needle probe
12 Brigaud and Vasseur (1989) Dolomites 7 25.00a 0.00a 0.00a 100a 0.00 2.86 2.46–2.82 0.02–0.14 0.00 0.00 0.00 needle probe
13 Brigaud and Vasseur (1989) Anhydrites 2 25.00a 0.00a 0.00a 100a 0.00 2.96 2.89–2.92 0.01–0.03 0.00 0.00 0.00 needle probe
14 Brigaud and Vasseur (1989) Smectite 9 25.00a 0.00a 0.00a 100a 27.90 2.64 1.87–1.97 0.25–0.27 10.25–14.71 0.220–0.28 0.75–0.95 needle probe
15 Brigaud and Vasseur (1989) Illite-Kaolinite 10 25.00a 0.00a 0.00a 100a 7.00 2.65 1.66–1.87 0.30–0.37 9.97–17.25 0.17–0.29 0.52–0.94 needle probe
16 Brigaud and Vasseur (1989) Kaolinite 10 25.00a 0.00a 0.00a 100a 26.00 2.65 1.65–1.94 0.27–0.38 13.78–22.73 0.27–0.38 1.00 needle probe
17 Caridad et al. (2014) Hectorite 18 20.05 0.00 0.00 100.00 0.00a 2.65a 1.04–1.09 0.59–0.61 90.00–94.00 0.97–0.98 1.60–1.67 KD2 Pro
18 Caridad et al. (2014) Sodium bentonite 12 20.15 0.00 0.00 100.00 0.00a 2.65a 1.07–1.12 0.58–0.60 85.00–90.00 0.95–0.96 1.62–1.65 KD2 Pro
19 Caridad et al. (2014) Magnesium bentonite 24 20.27 0.00 0.00 100.00 0.00a 2.65a 1.17–1.42 0.46–0.56 55.00–75.00 0.78–0.88 1.55–1.68 KD2 Pro
20 Caridad et al. (2014) Common clay 24 19.97 0.00 0.00 100.00 0.00a 2.65a 1.12–1.77 0.33–0.58 35.00–80.00 0.62–0.90 1.44–1.87 KD2 Pro
21 Caridad et al. (2014) Clay rosa 18 19.94 0.00 0.00 100.00 0.00a 2.65a 1.32–1.45 0.45–0.50 45.00–65.00 0.65–0.86 1.44–1.71 KD2 Pro
22 Caridad et al. (2014) Clay cocoa 18 20.17 0.00 0.00 100.00 0.00a 2.65a 1.34–1.72 0.35–0.49 35.00–55.00 0.60–0.74 1.49–1.72 KD2 Pro
23 Caridad et al. (2014) Clay sabhasana 18 20.07 0.00 0.00 100.00 15.00a 2.65a 1.35–1.64 0.38–0.49 35.00–55.00 0.58–0.74 1.51–1.56 KD2 Pro
24 Cho et al. (2011) Sand-bentonite mix 1 18 25.00 10.00 0.00 90.00 10.90 2.65a 1.60 0.32–0.40 13.08–20.00 0.21–0.32 0.57–0.97 QTM-500
5

25 Cho et al. (2011) Sand-bentonite mix 2 15 25.00 20.00 0.00 80.00 20.80 2.65a 1.60 0.32–0.40 13.08–20.00 0.21–0.32 0.58–0.97 QTM-500
26 Cho et al. (2011) Sand-bentonite mix 3 14 25.00 30.00 0.00 70.00 30.70 2.65a 1.60 0.32–0.40 13.08–20.00 0.21–0.32 0.58–0.97 QTM-500
27 Cho et al. (2011) Sand-bentonite mix 4 74 25.00 0.00 0.00 100.00 1.00 2.65a 1.20–1.80 0.32–0.55 10.65–20.00 0.13–0.36 0.26–1.10 QTM-500
28 Chu (2009) Kaoline 27 25.00a 32.00 44.00 24.00 0.00a 2.65a 1.30–1.80 0.32–0.51 0.73–16.98 0.01–0.29 0.02–0.81 Thermal probe
29 Chu (2009) ZH-bentonite 25 25.00a 27.50 37.50 35.00 0.00a 2.65a 1.40–1.80 0.32–0.47 0.61–7.17 0.01–0.13 0.02–0.40 Thermal probe
30 Chu (2009) Wyoming bentonite 27 25.00a 3.00 22.00 75.00 0.00a 2.65a 1.30–1.70 0.32–0.51 0.84–15.49 0.01–0.26 0.02–0.73 Thermal probe
31 de Zárate et al. (2010) Magnesium bentonite 2 38.90 0.00 0.00 100.00 0.00 2.65a 1.27–1.56 0.41–0.56 40.00–660.00 0.62–0.76 1.46–1.52 hot-wire method
32 JNC (2000) Kunigel V1 bentonite 14 25.00a 30.00a 0.00a 70.00a 60.00a 2.65a 1.80 0.35 0.05–18.51 0.00–0.33 0.00–0.94
33 Kahr and Müller-Vonmoos (1982) MX-80 30 32.00 0.00 0.00 100.00 3.00a 2.65a 1.45–2.36 0.11–0.45 0.00–14.00 0.00–0.331 0.00–1.86 Hot-wire method
34 Kahr and Müller-Vonmoos (1982) Montigel 9 35.00 0.00 0.00 100.00 0.00a 2.65a 1.86–2.14 0.19–0.30 0.00–6.00 0.00–0.15 0.78 Hot-wire method
35 Kasubuchi et al. (2007) Kuroboku 9 25.00a 27.70 14.30 58.00 45.00 2.44 0.85 0.65 0.12–76.11 0.01–0.65 0.00–1.00 DPHP-Twin-heat probe
36 Knutsson (1983) MX-80 bentonite4 12 25.00a 0.00 15.00 85.00 3.00a 2.65a 2.17 0.18 1.89–4.47 0.04–0.10 0.23–0.54 transient hot strip
37 Ould-Lahoucine et al. (2002) B-1 14 25.00a 0.00 0.00 100.00 60.00a 2.65a 1.35–2.17 0.30–0.50 0.00–25.69 0.00–0.47 0.00–0.93 Point heat source
38 Ould-Lahoucine et al. (2002) M-1 8 25.00a 13.60 0.00 86.40 60.00a 2.65a 2.04–2.30 0.30 0.00–11.22 0.00–0.26 0.00–0.86 Point heat source
39 Ould-Lahoucine et al. (2002) M-9 6 25.00a 21.20 0.00 78.80 60.00a 2.65a 2.12–2.40 0.30 0.00–12.00 0.00–0.29 0.00–0.96 Point heat source
40 Ould-Lahoucine et al. (2002) M-15 8 25.00a 38.60 0.00 61.40 60.00a 2.65a 2.32–2.56 0.30 0.00–9.65 0.00–0.25 0.00–0.82 Point heat source
41 Ould-Lahoucine et al. (2002) M-23 3 25.00a 13.20 0.00 86.80 60.00a 2.65a 1.84–2.17 0.39 0.42–17.79 0.01–0.39 0.02–0.99 Point heat source
42 Ould-Lahoucine et al. (2002) M-26 6 25.00a 11.86 0.00 88.14 60.00a 2.65a 1.78–2.08 0.40 0.00–14.54 0.00–0.30 0.00–0.76 Point heat source

Engineering Geology 288 (2021) 106107


43 Ould-Lahoucine et al. (2002) M-32 13 25.00a 19.70 0.00 80.30 60.00a 2.65a 1.64–2.36 0.42–0.63 0.73–25.69 0.01–0.53 0.02–0.99 Point heat source
44 Reno and Winterkorn (1967) Natural kaolinite 33 25.00a 0.00a 0.00a 100a 0.00a 2.61 0.65–1.25 0.52–0.72 3.80–32.30 0.03–0.40 0.03–0.77 Transient needle probe, SPHP
45 Reno and Winterkorn (1967) Sodium kaolinite 26 25.00a 0.00a 0.00a 100a 0.00a 2.61 0.90–1.09 0.53–0.66 1.24–30.56 0.01–0.38 0.02–0.72 Transient needle probe, SPHP
46 Reno and Winterkorn (1967) Calcium kaolinite 25 25.00a 0.00a 0.00a 100a 0.00a 3.29 0.90–1.31 0.48–0.74 1.44–33.66 0.01–0.42 0.02–0.88 Transient needle probe, SPHP
47 Reno and Winterkorn (1967) Aluminum Kaolinite 26 25.00a 0.00a 0.00a 100a 0.00a 2.61 0.83–1.25 0.52–0.68 1.23–34.78 0.01–0.43 0.01–0.83 Transient needle probe, SPHP
48 Singh and Devid (2000) Black cotton clay 35 25.00a 0.00 4.00 96.00 0.00a 2.72 1.00–1.40 0.49–0.63 0.00–30.00 0.00 0.00–0.20 SPHP
49 Tien et al. (2005) Pure B.H. Bentonite 27 25.00a 0.00a 0.00a 100a 64.50a 2.65 1.43–1.84 0.31–0.46 0.03–20.04 0.00–0.33 0.00–0.96 Thermal needle probe
50 Tien et al. (2005) Kinmen-betonite mix 1 27 25.00a 12.50 0.00 87.50 64.50a 2.65 1.43–1.84 0.31–0.46 0.03–20.04 0.00–0.33 0.00–0.96 Thermal needle probe
51 Tien et al. (2005) Kinmen-betonite mix 2 27 25.00a 25.00 0.00 75.00 64.50a 2.66 1.43–1.84 0.31–0.46 0.03–20.04 0.00–0.33 0.00–0.84 Thermal needle probe
52 Tien et al. (2005) Kinmen-betonite mix 3 24 25.00a 37.50 0.00 62.50 64.50a 2.66 1.43–1.84 0.31–0.46 0.03–20.04 0.00–0.31 0.00–0.81 Thermal needle probe
53 Tien et al. (2005) Kinmen-betonite mix 4 23 25.00a 50.00 0.00 50.00 64.50a 2.66 1.43–1.84 0.31–0.46 0.03–20.04 0.00–0.31 0.00–0.85 Thermal needle probe
(continued on next page)
L. Liu et al. Engineering Geology 288 (2021) 106107

λeff = 0.62 + 3.976Sr + 2.687Sr ln(Sr ) − 3.313Sr2 (31)

Hot-wire method (Kemtherm


QTM-D3 conductivity meter)
λeff = 0.25 W m− 1 ◦ C− 1
was assumed when Sr < 0.1 to avoid un­
Hot-wire method (KD2)
reasonable values of estimated λeff.

2.3.10. Singh and Devid, 2000 (SD2000) model


KD2 Pro SH-1
KD2 Pro SH-1
KD2 Pro SH-1
KD2 Pro SH-1

KD2 Pro SH-1


KD2 Pro SH-1
KD2 Pro SH-1
Thermo-TDR
Based on the work of Gangadhara Rao and Singh (1999), Singh and
Devid (2000) proposed a series of empirical models to predict λeff as
KD2 Pro

ISOMET
Method


⎨ a100.6243ρb − 3 θg < 2% (a)
λeff = (b100.6243ρb − 3 ) 2% ≤ θg ≤ 5% (b) (32)
0.29–0.86
0.00–0.01

0.31–0.97
0.04–0.84
0.10–0.78
0.03–0.99
0.02–1.00
0.04–1.00
0.00–0.83
0.01–0.91
0.02–0.97
0.00–1.00

1.07logθg + c 100.6243ρb − 3 θg > 5% (c)
Sr

where parameters a, b, and c are soil texture and water content depen­
dent. For clays: a = 0.219 for θg < 2%, b = 0.243 for 2% ≤ θg < 4% and b
0.13–0.30
0.00–0.56

0.12–0.35
0.02–0.50
0.10–0.47
0.01–0.37
0.01–0.41
0.02–0.42
0.00–0.26
0.01–0.38
0.01–0.40
0.00–0.75

= 0.276 for 4% < θg ≤ 5%, and c = − 0.73 for θg > 5% (Singh and Devid,
2000).
θv
Water content

2.3.11. Newson et al., 2002 (NT2002) model


0.00–115.20
9.00–17.92
0.10–27.40

7.20–21.20
2.17–45.69
5.18–42.44

0.40–25.40
1.00–26.20
0.00–14.50
0.40–23.90
0.40–24.70

Newson et al. (2002) proposed a regression model to estimate λeff of


0.80–0.26

consolidating offshore clayey backfill


θg

( )− 0.316
λeff = 3.674 θg (33)
Symbols: ρs – Particle density; ρb–Bulk density; P—porosity; θg –gravimetric water content; θv—volumetric water content; Sr—degree of saturation.
0.35–0.47
0.35–0.45

0.31–0.38

0.35–0.64
0.33–0.63
0.28–0.64
0.28–0.49
0.35–0.57
0.33–0.56
0.72–0.75

Eq. (33) gives unreasonable high value close to zero water content
0.60
0.60

and θg = 0.03% was used as arbitrary bound to calculate λdry and to avoid
P

numeric error.
1.45–1.83
1.53–1.76

1.65–1.83

1.00–1.80
1.00–1.80
1.00–1.80
1.00–1.80
1.20–1.80
1.20–1.80
0.65–0.71

2.3.12. Tang and Cui, 2006 (TC2006) model


1.10
1.10
ρb

Tang and Cui (2006) estimated λeff of clay from air filled porosity
2.65a

(34)
2.76
2.70

2.74
2.73
2.76
2.70
2.76
2.76
2.76
2.70
2.58

λeff = α(θair /P) + λsat


ρs

where α ≈ − λsat is the slope of λeff vs θair/P. α = − 2.22 W m− 1 ◦ C− 1


and
12.45
44.00
10.00

15.00
15.00
10.00
0.00a
fquartz

3.00
2.00

7.00

7.00
0.00

λsat = 1.15 W m− 1 ◦ C− 1 were suggested by Tang and Cui (2006).


100.00

2.3.13. Caridad et al., 2014 (CV2014) model


67.00

82.58
53.50
47.59
43.60
81.90
40.10
80.20
43.60
81.90
100a
fclay

Caridad et al. (2014) proposed an empirical model to estimate λeff of


Values estimated based on the literature description or the soil information from other literature.

seven clays from Spain


22.00

32.20
40.70
56.40
18.10
59.90
19.80
56.40
18.10
0.00a

0.00

0.00
fsilt
Soil texture

λeff = 0.5873ρb − 0.0036 (35)


11.00

17.42
14.30
11.71
0.00a

0.00
0.00
0.00
0.00
0.00
0.00
0.00
fsand

2.3.14. Yoon et al., 2018 (YS2018) model


Yoon et al. (2018) proposed a regression model to estimate λeff of KJ-I
and KJ-II Ca-Bentonite from Korea with R2 = 0.74
25.00a

25.00a
25.00a
25.00a
25.00a

25.00a
20.00

21.00
Temp

5.00
5.00

5.00
5.00

λeff = 0.641ρb + 0.624Sr − 0.510 (36)


No. Meas.

3. Data compilation and analysis


24
59

37
20
18
17
16
16
12
32
33
15

3.1. Published datasets


Liuzhou lateritic clay
Guilin lateritic clay

To evaluate the various models to estimate λeff of clay, experimental


MX-80 bentonite
FEBEX bentonite

Genhe silty clay

data were collated by following several important criteria: (1) reliable


Kaolin clay
Soil Name

and reproducible experimental techniques/setup. λeff were measured on


GMZ07

GMZ01

GMZ07

soil samples with transient method (e.g., heat pulse method, hot-wire
MX80

MX80

MX80

method and thermal probe) at room temperature under atmosphere


pressure condition and thereby temperature effects on the λeff will not be
data were from Tang et al. (2008).

considered in this study; (2) measurements were taken on pure clays or


clay content is greater than 50%; and (3) detailed description of the
specimens preparation and sufficient soil physic properties including
grain/particle size distribution (fclay, fsilt, and fsand), P, ρb, ρs and mineral
Villar Galicia (2002)
Tong et al. (2009) b

Zhang et al. (2016)


Xu et al. (2019d)
Xu et al. (2019d)
Xu et al. (2019d)
Xu et al. (2019d)
Xu et al. (2019d)
Xu et al. (2019d)

composition are available. The screen procedure ended up with a total of


Xu et al. (2019a)
Literature source

Xu et al. (2019c)
Xu et al. (2019c)
Table 1 (continued )

1250 measurements made on 65 soils from 21 studies performed around


the world, including Australia (Barry-Macaulay et al., 2013), France
(Beziat et al., 1988; Brigaud and Vasseur, 1989; Tong et al., 2009), Spain
(Caridad et al., 2014; de Zárate et al., 2010; Villar Galicia, 2002),
Sweden (Börgesson et al., 1994; Knutsson, 1983), Swiss (Kahr and
No.

54
55

56
57
58
59
60
61
62
63
64
65

Müller-Vonmoos, 1982), USA (Reno and Winterkorn, 1967; Zhang et al.,


b
a

6
L. Liu et al. Engineering Geology 288 (2021) 106107

Fig. 1. Performance of two theoretical/semi-theoretical models: (a) DV1963 (de Vries, 1963)_ENREF_27_ENREF_27; (b) DV1963II (Tian et al., 2016); (c) SK1998
(Sakashita and Kumada, 1998); (d) JR2010 (Jougnot and Revil, 2010) and four mixing models: (e) DF1979 (Donazzi et al., 1979); (f) KB1995 (Kiyohashi and Banno,
1995); (g) GA2009 (Gens et al., 2009); and (h) CW2011 (Cho et al., 2011) for predicting thermal conductivity of clays. Data greater than 3 W m− 1 ◦ C− 1
were truncated.

2016), Japan (JNC, 2000; Kasubuchi et al., 2007; Ould-Lahoucine et al., compared with measured and predicted λeff values using a 1:1 plot
2002), Korea (Cho et al., 2011), China (Chu, 2009; Tien et al., 2005; Xu (Piñeiro et al., 2008); ±10% and ± 50% lines off the 1:1 line and three
et al., 2019a; Xu et al., 2019c, 2019d), and India (Singh and Devid, additional statistical indices are also used:
2000). Selected soil properties for the compiled dataset were tabulated
(Table 1). (1) root mean square error (RMSE)
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√∑
3.2. Values for unavailable physical properties √n (
√ X − Xm
)2
√i=1 p
RMSE = (40)
For dataset without particle density, ρs = 2.65 g cm− 3 was assumed. n
Values of λw = 0.6 W m− 1 ◦ C− 1and λair = 0.025 W m− 1 ◦ C− 1were used for
thermal conductivity of soil water and air, respectively. For model that
requires input of thermal conductivity of solid (λs) but did not provide a
way to calculate it, the method of Johansen (1975) was used (2) the average deviations (AD)

f 1− f
(37) 1∑ n
( )
(41)
quartz
λs = λquartz λotherquartz AD = Xm − Xp
n i=1
where fquartz, λquartz λother are mass fraction of quartz, thermal conduc­
tivity of quartz and other minerals, respectively. Average value of fquartz
= 7.7 W m− 1 ◦ C− 1 at 20 ◦ C and λother = 2.0 W m− 1 ◦ C− 1 (λother = 3 for
fquartz ≤ 20% at 20 ◦ C) (Johansen, 1975, 1977). If quartz content was not (3) the Nash-Sutcliff Efficiency (NSE)
provided in the dataset, fquartz = 0 is assumed because clays tend to n (
∑ )2
consists of less quartz compared to sands and silts. Xp − Xm
For the RS2016 model (RÓżański and Stefaniuk, 2016) that requires NSE = 1-∑ i− 1
n ( )2 (42)
the input of specific surface area (SSA, m2 g− 1), which is calculated by
Xp − Xmean− m
i− 1
(Sepaskhah et al., 2010)
where, Xp values are predicted by different thermal conductivity models,
SA = 3.89dg− 0.905 (38)
Xm are measured values, Xmean-m are mean-measured values, n is the
( ) number of measured values and Pn is the number of parameters in the
dg = exp -0.069078fsand − 0.036497fsilt − 0.0247fclay (39) model.
RMSE describes the deviation between the predicted values and the
where fsand, fsilt, and fclay are mass fraction (%) of sand, silt and clay, measured values. AD represents the deviation from the predicted value
respectively. and the measured value. NSE determines the relative magnitude of the
residual variance between measured and predicted values (Nash and
3.3. Model performance metrics Sutcliffe, 1970). A smaller difference between the predicted and
measured values is expected when NSE approaches 1.
The performance of different thermal conductivity models was

7
L. Liu et al. Engineering Geology 288 (2021) 106107

Table 2 is concave, which is opposite to the convex or sigmoidal shape of λeff(Sr)


Performance measures of the 28 thermal conductivity models for clays. or λeff(θv) relationships commonly appeared in literature (He et al.,
No. Models Abbrev. Perfromance Meas. 2020e; Wang et al., 2020b; Yan et al., 2019).
RMSE AD NSE
4.2. Normalized thermal conductivity models
1 ◦ − 1
W m− C

Theoretical/semi-theoretical models The overall performance of LJ2016 model (Fig. 2d) is the best and
1 de Vries (1963) DV1963 4.21 − 0.66 − 64.94 the KS1983 model is the worst among the six normalized thermal con­
2 Tian et al. (2016) DV1963II 1.90 − 0.53 − 12.48
ductivity models (Fig. 2b and Table 2). Although the KS1983 model
3 Sakashita and Kumada SK1998 0.51 0.07 0.03
(1998) worked best for most of the soils investigated as will discussed in section
4 Jougnot and Revil (2010) JR2010 0.62 − 0.03 − 0.41 4.4, unreasonable performance on soils from Brigaud and Vasseur
Mixing models
(1989) significantly lowered its overall performance.
5 Donazzi et al. (1979)a DF1979 0.35 ¡0.04 0.54 The commonly used JO1975 model (Fig. 2a) generally over­
6 Kiyohashi and Banno KB1995 0.50 0.12 0.07 estimated λeff, which echoes the findings of He et al. (2020e) who
(1995) demonstrated that the JO1975 model could give satisfactory estimates
7 Gens et al. (2009) GA2009 0.63 0.06 − 0.50
for coarse- and medium-textured soils but not on fine-textured soils as
8 Cho et al. (2011) CW2011 1.23 0.94 − 4.63
clays. Similar overestimation can also be found for the rest normalized
Normalized models thermal conductivity models (Fig. 2c, e and f and Table 2). This agrees
9 Johansen (1975) JO1975 0.55 0.22 − 0.14
10 Knutsson (1983) KS1983 54.79 3.27 − 11,152.25
well with the study of He et al. (2017) who found many of the
11 Côté and Konrad (2005) CK2005 0.59 0.17 − 0.29 normalized model failed to model λeff of coarse- or fine-textured soils.
12 Lee et al. (2016) LJ2016 0.49 − 0.01 0.11
13 RÓżański and Stefaniuk RS2016 0.87 0.47 − 1.81 4.3. Linea/non-linear thermal conductivity models
(2016)
14 Zhang et al. (2016) ZN2016 0.53 0.29 − 0.05
For the 14 linear/non-linear thermal conductivity models, the best
Linear/non-linear models performing model is CV2014 (Fig. 4g) with RMSE = 0.42 W m− 1 ◦ C− 1,
15 Kersten (1949) KM1949 0.43 0.21 − 0.51
16 Makowski and Mochlinski MM1956 0.57 0.16 − 0.22
AD = 0.03 W m− 1 ◦ C− 1 and NSE = 0.35, followed by YS2018 (Fig. 4h)
(1956) and KV982 (Fig. 3e). However, the CV2014 model only considered the
17 van Rooyen and VW1959 8.05 − 0.03 − 239.64 effects of bulk density and it generally gave overestimates of λeff for λeff
Winterkorn (1959) < 1 W m− 1 ◦ C− 1. The YS2018 model accounts for effects of bulk density
18 Oriol et al. (1978) OF1978 0.81 − 0.62 − 1.43
and water content, is simple and easy to be applied. The KV1982 model
19 Kahr and Müller-Vonmoos KV1982 0.49 − 0.28 0.11
(1982) was only applicable to clays with gravimetric water contents ranging
20 Campbell (1985) CG1985 0.62 − 0.33 − 0.41 from 0 to 12% and bulk density ranging from 1.4 to 2.3 g cm− 3 (Ould-
21 Chung and Horton (1987) CH1987 0.56 − 0.13 − 0.17 Lahoucine et al., 2002).
22 Becker et al. (1992) BB1992 0.51 0.08 0.03 The VW1959 model (Fig. 3c) performed the worst with RMSE =
23 Thomas et al. (1994) TH1994 0.53 − 0.17 − 0.03
8.05 W m− 1 ◦ C− 1, AD = − 0.03 W m− 1 ◦ C− 1 and NSE = − 239.64, fol­
24 Singh and Devid (2000) SD2000 0.41 − 0.33 − 0.43
25 Newson et al. (2002) NT2002 2.77 1.46 − 27.49 lowed by the NT2002 model (Table 2). This agrees with previous studies
26 Tang and Cui (2006) TC2006 0.56 − 0.12 − 0.18 that the VW1959 model was complicated and the choice of parameters is
27 Caridad et al. (2014) CV2014 0.42 0.03 0.35 uncertain (Farouki, 1981). Its applications is limited to unfrozen gravels
28 Yoon et al. (2018) YS2018 0.44 − 0.05 0.29
and sands with 0.015 < Sr < 0.1 (Becker et al., 1992; Farouki, 1986;
a
the best performing model for the compiled dataset. Fricke et al., 2002; Nikolaev et al., 2013). The values of λeff estimated
with NT2002 model decreases with increase of water content (Newson
4. Results and discussion et al., 2002), which is opposite to the widely accepted fact that λeff in­
creases with water content and bulk density. The OF1968 (Fig. 3d) gave
4.1. Theoretical/semi-theoretical and mixing models nearly constant λeff values, because it assumed that λdry = 0.23 W m− 1 ◦ C
-1
and λsat = 0.32 W m− 1 ◦ C -1 (Oriol et al., 1978). Therefore, the resulted
For the four theoretical/semi-theoretical models, the SK1998 model values of λeff ranged between 0.23 and 0.32 W m− 1 ◦ C -1 and significantly
performed better than the JR2010 model (Fig. 1c and b, Table 2). underestimated the measured λeff at high water content range.
Although the de Vries (1963) model has been widely applied, it cannot The models of BB1992 (Fig. 4b), CH1987 (Fig. 4a), CG1985 (Fig. 3f)
be used to well predict thermal conductivity of clay (Fig. 1a, Table 2). and KM1949 (Fig. 3a) have been widely used or incorporated into
The performance of its simplified from proposed by Tian et al. (2016) is various land surface models (Haverd and Cuntz, 2010; He et al., 2020c;
slightly better but still not satisfactory (Fig. 1b, Table 2). Theoretical Simunek et al., 1997; Yang, 2015), but they could not well predict λeff.
models attract attention modeling soil physical properties including The KM1949 model generally overestimated λeff, while the models of
thermal conductivity (de Vries, 1963) and dielectric permittivity (He CG1985 and CH1987 generally underestimated λeff. This differs from the
and Dyck, 2013; He et al., 2016). However, their applications were studies of Tien et al. (2005) and Abu-Hamdeh (2000) who reported the
restricted by the fact that the models are complicated and the parame­ CG1985 model could be used to well predict λeff.
ters are difficult to be determined (Dyck et al., 2019; Tian et al., 2016).
The DF1979 model performed the best among the four mixing 4.4. Limitations and perspectives
models with RMSE = 0.35 W m− 1 ◦ C− 1, AD = − 0.04 W m− 1 ◦ C− 1 and
NSE = 0.54 (Fig. 1c~f, Table 2). While the CW2011 model performed The DF1979 is the best performing models among the 28 thermal
the worst with RMSE = 1.23 W m− 1 ◦ C− 1, AD = 0.94 W m− 1 ◦ C− 1 and conductivity models investigated although its performance is not satis­
NSE = − 4.63. The CW2011 model generally overestimated the factory on the whole dataset. This is reasonable because currently there
measured λeff of clays and most of the estimated λeff are above the +50% is no universal thermal conductivity model has been reported (Cheng
line, which indicates they are greater than 1.5 times of measured λeff. and Hsu, 1999; Gangadhara Rao and Singh, 1999; He et al., 2021; Lu and
However, the DF1979 model was found to perform less satisfactorily in Dong, 2015). He et al. (2020b) included 24 soil thermal conductivity
sandy soils (Haigh, 2012), because the curvature of λeff as a function of Sr models widely used by land-surface, hydrological and soil-vegetation-
atmosphere transfer models into the Community Land Model

8
L. Liu et al. Engineering Geology 288 (2021) 106107

Fig. 2. Performance of six normalized models for predicting thermal conductivity of clays: (a) JO1975 (Johansen, 1975); (b) KS1983 (Knutsson, 1983); (c) CK2005
(Côté and Konrad, 2005); (d) LJ2016 (Lee et al., 2016); (e) RS2016 (RÓżański and Stefaniuk, 2016); and (f) ZN2016 (Zhang et al., 2016). Data greater than 3 W
m− 1 ◦ C− 1 were truncated.

9
L. Liu et al. Engineering Geology 288 (2021) 106107

Fig. 3. Performance of six linear/non-linear models for predicting thermal conductivity of clays: (a) KM1949 (Kersten, 1949); (b) MM1956 (Makowski and
Mochlinski, 1956); (c) VW1959 (van Rooyen and Winterkorn, 1959); (d) OF1978 (Oriol et al., 1978); (e) KV1982 (Kahr and Müller-Vonmoos, 1982); and (f) CG1985
(Campbell, 1985). Data greater than 3 W m− 1 ◦ C− 1 were truncated.

10
L. Liu et al. Engineering Geology 288 (2021) 106107

Fig. 4. Performance of eight linear/non-linear models for predicting thermal conductivity of clays: (a) CH1987 (Chung and Horton, 1987); (b) BB1992 (Becker et al.,
1992); (c) TH1994 (Thomas et al., 1994); (d) SD2000 (Singh and Devid, 2000); (e) NT2002 (Newson et al., 2002); (6) TC2006 (Tang and Cui, 2006); (f) CV2014
(Caridad et al., 2014); and (g) YS2018 (Yoon et al., 2018). Data greater than 3 W m− 1 ◦ C− 1 were truncated.

(CLM4.5) and compared the simulated land surface temperature (LST) models cannot predict λeff of clay at full range of water content or bulk
and LST from the moderate resolution imaging spectrometer (MODIS). density. In Addition, numeric errors may arise due to the selections of
Their results demonstrated that all 24 models performed similarly un­ basic model functions (e.g., logarithm or exponential functions). For
satisfactorily in clay and overestimated MODIS LST, while under­ instance, the models of JO1975, KS1983, MM1956 and TH1994 cannot
estimated MODIS LST in loam and sand. However, it should be noted be used for zero water content, because zero is an invalid input for the
that many of the models performed quite well on certain dataset. Good logarithm function. The other models may give negative values (e.g.,
performing models with NSE ≥ 0.8 are highlighted in Table 3 with bold KM1949 and CH1987) or unreasonable high or low values (e.g.,
font and shadow. Generally, the model performed best on the dataset NT2002) at low water content. This phenomena was found to be com­
used to develop the model, for example, little difference was found be­ mon in thermal conductivity modeling (He et al., 2020b) and future
tween the measured λeff of Kahr and Müller-Vonmoos (1982) and the effort developing closed-form equation with wide applications is highly
predicted λeff with KV1982 model, with NSE = 0.98 (Table 3), the similar recommended (He et al., 2020a).
was found for the TC2006 model on dataset of Tong et al. (2009). Moreover, hysteresis of water retention characteristics is more sig­
Meanwhile, some models including SK1998, DF1979, KB1995, JO1975, nificant for clays than sands (Pham et al., 2005; Wang et al., 2020b).
KS1983, CK2005, LJ2016, RS2016, ZN2016, KV1982, CG1985, Researches have shown the effects of hysteresis on the estimated λeff
CH1987, TH1994, TC2006, CV2014 and YS2018 have good perfor­ (Feng and Fredlund, 2003; Tang et al., 2008; Yoon et al., 2018). How­
mance on other datasets, which strengthen that the predictive capability ever, no study has attempted to include the effects of hysteresis on
of the model is dataset dependent. Only models of DF1979, CK2005, thermal conductivity models for clays (He et al., 2020a). The swell and
TH1994 and TC2006 have good performance (with NSE ≥ 0.8) on three shrinkage of some clays may make things more complicated, because
or more datasets. The feature of soil or dataset specific thermal con­ cracks develop when soil swells/shrinks (Christopher and Nwonu, 2019)
ductivity model for clays is also apparent in the development of the and significantly reduce the connectivity of heat transfer path, which in
model. For example, Yoon et al. (2018) defined the upper limit of λsat = turn influence the measured soil thermal conductivity. Therefore, none
1.2410 W m− 1 ◦ C− 1 for model YS2018, while Tang and Cui (2006) of these models would adequately describe thermal conductivity of clay
assumed λsat = 1.15 W m− 1 ◦ C− 1 for the TC2006 model. A model may unless the fitting model accounts for changes in pore geometry as a
work well on one dataset but badly on another dataset, thereby, care function of soil moisture. Similarly, the effects of temperature (Kurz,
should be taken by the researcher or practitioner when select the most 2014; Smits et al., 2013; Xu et al., 2019a; Xu et al., 2019c, 2019d; Zhang
appropriate model for predicting λeff of the clays they investigate. Model et al., 2018a), aging time (Villar et al., 2005; Xu et al., 2019b), additive
that was developed by previous researchers on the same soils, if any, (Beziat et al., 1992; Kaya and Durukan, 2004; Liu et al., 2019), pressure
from the investigated region is recommended if there is no access to take (Beziat et al., 1988; Wiebe et al., 1998) on λeff of clay and other soils
measurement of λeff. were reported but less studied nor modelled. Future studies on these
It is well known that empirical model cannot be extended to soils topics should be of interest.
beyond those used to develop the model (He et al., 2017; Lu et al., 2014; Furthermore, previous studies have shown that the value of solid
Yan et al., 2019). Therefore, many of the empirical thermal conductivity thermal conductivity (λs) largely affects the performance of a thermal

11
L. Liu et al. Engineering Geology 288 (2021) 106107

Table 3
Dataset dependent performance measure of the 28 thermal conductivity models for clays.

12
L. Liu et al. Engineering Geology 288 (2021) 106107

conductivity model (He et al., 2020e; Tarnawski et al., 2009; Wang required. The possible solutions may include: (1) the development of a
et al., 2020b; Yan et al., 2019). λs is largely determined by soil miner­ global soil mineralogy database (e.g.,) from scattered data in literature
alogical information (especially quartz content) that requires special in order to calculate λs from soil mineralogy using geometric mean
instruments including X-ray fluorescence (XRF) and X-ray diffraction model (Schönenberger et al., 2012; Tarnawski et al., 2012), because
(XRD), but these measurements are not commonly made (Tarnawski there is already dataset on soil mineralogy of fine soil (Journet et al.,
et al., 2009; Whittig, 1965). The occurrence of quartz content depends 2014) that can be expanded to meet the requirement of calculating λs;
mainly on the parent material, which make it highly unpredictable and (2) the development of a global database of λs that was calculated from
there is no effective way to model it as of date (Tarnawski et al., 2012; the measured saturated soil thermal conductivity using inversed geo­
Tarnawski et al., 2011). Therefore, instead of using the Johansen (1975) metric mean model or the three-point method proposed by He et al.
method, simplified approaches were taken by not accounting for λs, (2020d) from unsaturated soil thermal conductivity; (3) the use of ma­
assuming constant values of λs or estimating from other physic proper­ chine learning or artificial intelligence (Rizvi et al., 2020; Rizvi et al.,
ties (e.g., bulk density, texture or specific surface area). For example, 2019; Wen et al., 2020; Zhang and Shi, 2020) to establish soil thermal
none of the 14 linear/non-linear model considered effects of λs, but it conductivity models from known soil information in the similar ap­
should pointed out that the empirical model of TC2006 related λeff to air proaches used by soil hydraulic properties (Zhang and Schaap, 2019;
filled porosity is promising approach to get around the inclusion of λs Zhang et al., 2020; Zhang et al., 2018b): and (4) the development of soil
(Ochsner et al., 2001; Xie et al., 2020). Knutsson (1983) used λs = 2 W thermal conductivity products (Dai et al., 2019) for direct use in land-
m− 1 ◦ C− 1 for KS1983 model and Donazzi et al. (1979) used λs = 2.3 W surface modeling, similar to the driven data of air temperature, pre­
m− 1 ◦ C -1 for DF1979 model, while Cho et al. (2011), Kiyohashi and cipitation, solar radiation and wind etc. (Lembrechts et al., 2020; Peng
Banno (1995), and RÓżański and Stefaniuk (2016) defined different et al., 2019; Zhao et al., 2019).
equations to calculate λs. However, none of these approaches were
approved to be effective for wide applications and more studies are

13
L. Liu et al. Engineering Geology 288 (2021) 106107

RMSE–root mean square error (W m− 1 ◦ C− 1); AD–average deviations (W m− 1 ◦ − 1


C ); NSE–Nash-Sutcliff Efficiency.
Bold and shadowed values of NSE were greater than or equal to 0.8.

5. Conclusion Abu-Hamdeh, N.H., 2000. Effect of tillage treatments on soil thermal conductivity for
some Jordanian clay loam and loam soils. Soil Tillage Res. 56, 145–151. https://doi.
org/10.1016/S0167-1987(00)00129-X.
In this study, 28 thermal conductivity models for clays were Akinyemi, O.D., Sauer, T.J., Onifade, Y.S., 2011. Revisiting the block method for
reviewed and evaluated with a large compiled dataset from 21 studies. evaluating thermal conductivities of clay and granite. Int. Commun. Heat Mass
The resulted showed that the DF1979 model had the best performance Transf. 38, 1014–1018. https://doi.org/10.1016/j.icheatmasstransfer.2011.05.016.
Barry-Macaulay, D., Bouazza, A., Singh, R.M., 2011. Study of thermal properties of a
on the whole compiled dataset but still not satisfactory. Performance of basaltic clay. Geo-Frontiers 2011.
the thermal conductivity models may be dataset or soil dependent and Barry-Macaulay, D., Bouazza, A., Singh, R.M., Wang, B., Ranjith, P.G., 2013. Thermal
cares should be taken when choosing the most appropriate model for conductivity of soils and rocks from the Melbourne (Australia) region. Eng. Geol.
164, 131–138. https://doi.org/10.1016/j.enggeo.2013.06.014.
prediction purpose in practice. Future efforts focusing on the develop­ Becker, B.R., Misra, A., Fricke, B.A., 1992. Development of correlations for soil thermal
ment of more accurate thermal conductivity models for wider applica­ conductivity. Int. Commun. Heat Mass Transf. 19, 59–68. https://doi.org/10.1016/
tions are recommended. 0735-1933(92)90064-O.
Beziat, A., Dardaine, M., Gabis, V., 1988. Effect of compaction pressure and water
content on the thermal conductivity of some natural clays. Clay Clay Miner. 36,
Declaration of Competing Interest 462–466. https://doi.org/10.1346/CCMN.1988.0360512.
Beziat, A., Dardaine, M., Mouche, E., 1992. Measurements of the thermal conductivity of
clay-sand and clay-graphite mixtures used as engineered barriers for high-level
The author declares no conflict of interest. radioactive waste disposal. Appl. Clay Sci. 6, 245–263. https://doi.org/10.1016/
S0169-1317(09)90001-1.
Acknowledgements Börgesson, L., Fredrikson, A., Johannesson, L.-E., 1994. Heat Conductivity of Buffer
Materials. Svensk kärnbränslehantering AB.
Brigaud, F., Vasseur, G., 1989. Mineralogy, porosity and fluid control on thermal
Datasets used in this study were digitalized from the published conductivity of sedimentary rocks. Geophys. J. Int. 98, 525–542. https://doi.org/
literature. 10.1111/j.1365-246X.1989.tb02287.x.
Bristow, K.L., Horton, R., Kluitenberg, G.J., 1994. Measurement of soil thermal
Funding for this research was provided in part by the National Nat­
properties with a dual-probe heat-pulse technique. Soil Sci. Soc. Am. J. 58,
ural Science Foundation of China [Grant No. 41877015, 42077135], 1288–1294. https://doi.org/10.2136/sssaj1994.03615995005800050002x.
Natural Science Foundation of Shaanxi Province [2020JM-169], Young Cai, G., Zhang, T., Puppala, A.J., Liu, S., 2015. Thermal characterization and prediction
science and technology star of Shaanxi Province, China Postdoctoral model of typical soils in Nanjing area of China. Eng. Geol. 191, 23–30. https://doi.
org/10.1016/j.enggeo.2015.03.005.
Science Foundation [Grant No. 2018M641024], the Training Program Campbell, G.S., 1985. Soil Physics with BASIC: Transport Models for Soil-Plant Systems,
Foundation for the Young Talents of Northwest A&F University, and the 3rd ed. Elsevier, New York.
111 project [Grant No. B12007]. Campbell, G.S., Calissendorff, C., Williams, J.H., 1991. Probe for measuring soil specific
heat using a heat-pulse method. Soil Sci. Soc. Am. J. 55, 291–293. https://doi.org/
10.2136/sssaj1991.03615995005500010052x.
References Caridad, V., Ortiz de Zárate, J.M., Khayet, M., Legido, J.L., 2014. Thermal conductivity
and density of clay pastes at various water contents for pelotherapy use. Appl. Clay
Abuel-Naga, H.M., Bergado, D.T., Bouazza, A., Pender, M.J., 2009. Thermal conductivity Sci. 93–94, 23–27. https://doi.org/10.1016/j.clay.2014.02.013.
of soft Bangkok clay from laboratory and field measurements. Eng. Geol. 105,
211–219. https://doi.org/10.1016/j.enggeo.2009.02.008.

14
L. Liu et al. Engineering Geology 288 (2021) 106107

Chang, M.-Y., Juang, R.-S., 2004. Adsorption of tannic acid, humic acid, and dyes from He, H., Dyck, M., Zhao, Y., Si, B., Jin, H., Zhang, T., Lv, J., Wang, J., 2016. Evaluation of
water using the composite of chitosan and activated clay. J. Colloid Interface Sci. five composite dielectric mixing models for understanding relationships between
278, 18–25. https://doi.org/10.1016/j.jcis.2004.05.029. effective permittivity and unfrozen water content. Cold Reg. Sci. Technol. 130,
Cheng, P., Hsu, C., 1999. The effective stagnant thermal conductivity of porous media 33–42. https://doi.org/10.1016/j.coldregions.2016.07.006.
with periodic structures. J. Porous Media 2, 19–38. https://doi.org/10.1615/ He, H., Zhao, Y., Dyck, M., Si, B., Jin, H., Lv, J., Wang, J., 2017. A modified normalized
JPorMedia.v2.i1.20. model for predicting effective soil thermal conductivity. Acta Geotech. 12,
Cho, W.-J., Lee, J.-O., Kwon, S., 2011. An empirical model for the thermal conductivity 1281–1300. https://doi.org/10.1007/s11440-017-0563-z.
of compacted bentonite and a bentonite–sand mixture. Heat Mass Transf. 47, He, H., Dyck, M., Horton, R., Li, M., Jin, H., Si, B., 2018a. Distributed temperature
1385–1393. https://doi.org/10.1007/s00231-011-0800-1. sensing for soil physical measurements and its similarity to heat pulse method. In:
Christopher, I., Nwonu, D., 2019. Emerging trends in expansive soil stabilisation: A Sparks, D.L. (Ed.), Advances in Agronomy. Academic Press, Cambridge,
review. J. Rock Mech. Geotech. Eng. 11 https://doi.org/10.1016/j. pp. 173–230.
jrmge.2018.08.013. He, H., Dyck, M.F., Horton, R., Ren, T., Bristow, K.L., Lv, J., Si, B., 2018b. Development
Chu, C.-A., 2009. Measurement and Modeling for Thermal Conductivity of Geomaterials. and application of the heat pulse method for soil physical measurements. Rev.
National Central University. Geophys. 56, 567–620. https://doi.org/10.1029/2017rg000584.
Chung, S., Horton, R., 1987. Soil heat and water flow with a partial surface mulch. Water He, H., Dyck, M., Lv, J., 2020a. A new model for predicting soil thermal conductivity
Resour. Res. 23, 2175–2186. https://doi.org/10.1029/WR023i012p02175. from matric potential. J. Hydrol. 589, 125167. https://doi.org/10.1016/j.
Côté, J., Konrad, J.-M., 2005. A generalized thermal conductivity model for soils and jhydrol.2020.125167.
construction materials. Can. Geotech. J. 42, 443–458. https://doi.org/10.1139/t04- He, H., He, D., Jin, J., Smits, K.M., Dyck, M., Wu, Q., Si, B., Lv, J., 2020b. Room for
106. improvement: A review and evaluation of 24 soil thermal conductivity
Dai, Y.J., Xin, Q.C., Wei, N., Zhang, Y.G., Wei, S.G., Yuan, H., Zhang, S.P., Liu, S.F., Lu, X. parameterization schemes commonly used in land-surface, hydrological, and soil-
J., 2019. A global high-resolution data set of soil hydraulic and thermal properties vegetation-atmosphere transfer models. Earth Sci. Rev. 211, 103419. https://doi.
for land surface modeling. J. Adv. Model. Earth Syst. 11, 2996–3023. https://doi. org/10.1016/j.earscirev.2020.103419.
org/10.1029/2019ms001784. He, H., He, D., Smits, K., Dyck, M., Si, B., Lv, J., 2020c. A Review and Evaluation of Soil
Dao, L., Delage, P., Tang, A.-M., Cui, Y.-J., Pereira, J.-M., Li, X., Sillen, X., 2014. Thermal Conductivity Models for Land Suface Modelling [Submitted].
Anisotropic thermal conductivity of natural Boom Clay. Appl. Clay Sci. 101, He, H., Li, M., Dyck, M., Si, B., Wang, J., Lv, J., 2020d. Modelling of soil solid thermal
282–287. https://doi.org/10.1016/j.clay.2014.09.003. conductivity. Int. Commun. Heat Mass Transf. 116, 104602. https://doi.org/
de Vries, D.A., 1952a. A nonstationary method for determining thermal conductivity of 10.1016/j.icheatmasstransfer.2020.104602.
soil in situ. Soil Sci. 73, 83–90. He, H., Noborio, K., Johansen, Ø., Dyck, M., Lv, J., 2020e. Normalized concept for
de Vries, D.A., 1952b. The Thermal Conductivity of Soil. effective soil thermal conductivity modelling from dryness to saturation. Eur. J. Soil
de Vries, D.A., 1963. Thermal properties of soil. In: van Dijk, W.R. (Ed.), Physics of Plant Sci. 71, 27–43. https://doi.org/10.1111/ejss.12820.
Environment. North Holland Publishing, Amsterdam, pp. 210–235. He, H., Flerchinger, G., Kojima, Y., Dyck, M., Lv, J., 2021. A review and evaluation of 39
de Zárate, J.M.O., Hita, J.L., Khayet, M., Legido, J.L., 2010. Measurement of the thermal thermal conductivity models for frozen soils. Geoderma 382, 114694. https://doi.
conductivity of clays used in pelotherapy by the multi-current hot-wire technique. org/10.1016/j.geoderma.2020.114694.
Appl. Clay Sci. 50, 423–426. https://doi.org/10.1016/j.clay.2010.08.012. Hepper, E.N., Buschiazzo, D.E., Hevia, G.G., Urioste, A., Antón, L., 2006. Clay
Dissanayaka, S.H., Hamamoto, S., Kawamoto, K., Komatsu, T., Moldrup, P., 2012. mineralogy, cation exchange capacity and specific surface area of loess soils with
Thermal properties of peaty soils: Effects of liquid-phase impedance factor and different volcanic ash contents. Geoderma 135, 216–223. https://doi.org/10.1016/j.
shrinkage. Vadose Zone J. 11 https://doi.org/10.2136/vzj2011.0092. geoderma.2005.12.005.
Donazzi, F., Occhini, E., Seppi, A., 1979. Soil thermal and hydrological characteristics in JNC, 2000. Safety assessment of the geological disposal system. In: Supporting Report,
designing underground cables. In: Proceedings of the Institution of Electrical p. 3.
Engineers. IET, pp. 506–516. Johansen, O., 1975. Varmeledningsevne Av Jordarter (Thermal Conductivity of Soils).
Dyck, M., Miyamoto, T., Iwata, Y., Kameyama, K., 2019. Bound Water, phase PhD, University of Trondheim, Trondheim, Norway.
configuration, and dielectric damping effects on TDR-measured apparent Johansen, O., 1977. Thermal conductivity of soils. In: Cold Regions Research and
permittivity. Vadose Zone J. 18, 1–14. https://doi.org/10.2136/vzj2019.03.0027. Engineering Laboratory Report Report, 637.
Eucken, A., 1932. Thermal conductivity of ceramics refractory materials. Forsch. Geb. Jougnot, D., Revil, A., 2010. Thermal conductivity of unsaturated clay-rocks. Hydrol.
Ing., B-3, Forschungsheft 53, 6–21. Earth Syst. Sci. 14, 91–98. https://doi.org/10.5194/hess-14-91-2010.
Farouki, O.T., 1981. Thermal properties of soils. U.S. Army Corps of Engineers, Cold Journet, E., Balkanski, Y., Harrison, S.P., 2014. A new data set of soil mineralogy for
Regions Researchand Engineering Laboratory, Hanover, N.H, p. 1981. dust-cycle modeling. Atmos. Chem. Phys. 14, 3801–3816. https://doi.org/10.5194/
Farouki, O.T., 1982. Evaluation of methods for calculating soil thermal conductivity. In: acp-14-3801-2014.
DTIC Document Report CRREL Report, pp. 82–88. Kahr, G., Müller-Vonmoos, M., 1982. Wärmeleitfähigkeit von Bentonit MX 80 und von
Farouki, O.T., 1986. Thermal Properties of Soils. Trans Tech Publ, Zurich, Switzerland. Montigel nach der Heizdrahtmethode. NAGRA Technisher Bericht document ID:
Farrar, D.M., Coleman, J.D., 1967. The correlation of surface area with other properties 82–06.
of nineteen british clay soils. J. Soil Sci. 18, 118–124. https://doi.org/10.1111/ Kang, X., Ge, L., Kang, G.-C., Mathews, C., 2015. Laboratory investigation of the strength,
j.1365-2389.1967.tb01493.x. stiffness, and thermal conductivity of fly ash and lime kiln dust stabilised clay
Feng, M., Fredlund, D.G., 2003. Calibration of thermal conductivity sensors with subgrade materials. Road Mater. Pavement Design 16, 928–945. https://doi.org/
consideration of hysteresis. Can. Geotech. J. 40, 1048–1055. https://doi.org/ 10.1080/14680629.2015.1028970.
10.1139/T03-046. Kasubuchi, T., Momose, T., Tsuchiya, F., Tarnawski, V., 2007. Normalized thermal
Fricke, B.A., Misra, A., R., B.B. & Stewart, J.W.E., 2002. Soil Thermal Conductivity: conductivity model for three Japanese soils. Transactions of the Japanese Society of
Effects of Saturation and Dry Density. University of Missouri-Kansas City. Irrigation. Drain. Rural Eng. (Japan) 251, 529–533.
Gangadhara Rao, M., Singh, D.N., 1999. A generalized relationship to estimate thermal Kaya, A., Durukan, S., 2004. Utilization of bentonite-embedded zeolite as clay liner.
resistivity of soils. Can. Geotech. J. 36, 767–773. https://doi.org/10.1139/t99-037. Appl. Clay Sci. 25, 83–91. https://doi.org/10.1016/j.clay.2003.07.002.
Garnier, J.P., Gallier, J., Mercx, B., Dudoignon, P., Milcent, D., 2010. Thermal Kersten, M.S., 1949. Thermal Properties of Soils. Minnesota University Institute of
conductivity measurement in clay dominant consolidated material by Transient Hot- Technology.
Wire method. EPJ Web Conf. 6, 38001. https://doi.org/10.1051/epjconf/ Kiyohashi, H., Banno, K., 1995. Effective thermal conductivity of compact bentonite as a
20100638001. buffer material for high level radioactive waste. High Temp. High Pressures 27 (28),
Gens, A., Guimaraes, L.D.N., Garcia-Molina, A., Alonso, E.E., 2002. Factors controlling 653–663.
rock–clay buffer interaction in a radioactive waste repository. Eng. Geol. 64, Knutsson, S., 1983. On the thermal conductivity and thermal diffusivity of highly
297–308. https://doi.org/10.1016/s0013-7952(02)00026-1. compacted bentonite. In: SKB Report, Swedish Nuclear Fuel and Waste Management
Gens, A., Sánchez, M., Guimarães, L.D.N., Alonso, E.E., Lloret, A., Olivella, S., Villar, M. Co. SKB, 83-72.
V., Huertas, F., 2009. A full-scale in situ heating test for high-level nuclear waste Kurz, D., 2014. Understanding the effects of temperature on the behaviour of clay.
disposal: observations, analysis and interpretation. Géotechnique 59, 377–399. J. Clin. Investig. 122, 3051–3053.
https://doi.org/10.1680/geot.2009.59.4.377. Lee, J.O., Choi, H., Lee, J.Y., 2016. Thermal conductivity of compacted bentonite as a
Ghorbel-Abid, I., Trabelsi-Ayadi, M., 2015. Competitive adsorption of heavy metals on buffer material for a high-level radioactive waste repository. Ann. Nucl. Energy 94,
local landfill clay. Arab. J. Chem. 8, 25–31. https://doi.org/10.1016/j. 848–855. https://doi.org/10.1016/j.anucene.2016.04.053.
arabjc.2011.02.030. Lembrechts, J.J., Aalto, J., Ashcroft, M.B., De Frenne, P., Kopecky, M., et al., 2020.
Haigh, S.K., 2012. Thermal conductivity of sands. Géotechnique 62, 617–625. https:// SoilTemp: a global database of near-surface temperature. Glob. Chang. Biol. 26,
doi.org/10.1680/geot.11.P.043. 6616–6629. https://doi.org/10.1111/gcb.15123.
Haverd, V., Cuntz, M., 2010. Soil–Litter–Iso: A one-dimensional model for coupled Liu, X., Cai, G., Liu, L., Liu, S., Puppala, A., 2019. Thermo-hydro-mechanical properties
transport of heat, water and stable isotopes in soil with a litter layer and root of bentonite-sand-graphite-polypropylene fiber mixtures as buffer materials for a
extraction. J. Hydrol. 388, 438–455. https://doi.org/10.1016/j. high-level radioactive waste repository. Int. J. Heat Mass Transf. 141, 981–994.
jhydrol.2010.05.029. https://doi.org/10.1016/j.ijheatmasstransfer.2019.07.015.
He, H., Dyck, M., 2013. Application of multiphase dielectric mixing models for Lu, N., Dong, Y., 2015. Closed-form equation for thermal conductivity of unsaturated
understanding the effective dielectric permittivity of frozen soils. Vadose Zone J. 12 soils at room temperature. J. Geotech. Geoenviron. 141, 04015016 https://doi.org/
https://doi.org/10.2136/vzj2012.0060. 10.1061/(ASCE)GT.1943-5606.0001295.
He, H., Dyck, M., Wang, J., Lv, J., 2015. Evaluation of TDR for quantifying heat-pulse- Lu, Y., Lu, S., Horton, R., Ren, T., 2014. An empirical model for estimating soil thermal
method-induced ice melting in frozen soils. Soil Sci. Soc. Am. J. 79, 1275–1288. conductivity from texture, water content, and bulk density. Soil Sci. Soc. Am. J. 78,
https://doi.org/10.2136/sssaj2014.12.0499. 1859–1868. https://doi.org/10.2136/sssaj2014.05.0218.

15
L. Liu et al. Engineering Geology 288 (2021) 106107

Luo, J., Zhang, Y., Rohn, J., 2019. Analysis of thermal performance and drilling costs of Singh, D.N., Devid, K., 2000. Generalized relationships for estimating soil thermal
borehole heat exchanger (BHE) in a river deposited area. Renew. Energy 151, resistivity. Exp. Thermal Fluid Sci. 22, 133–143. https://doi.org/10.1016/S0894-
392–402. https://doi.org/10.1016/j.renene.2019.11.019. 1777(00)00020-0.
Maček, M., Mauko, A., Mladenovič, A., Majes, B., Petkovšek, A., 2013. A comparison of Smits, K.M., Sakaki, T., Howington, S.E., Peters, J.F., Illangasekare, T.H., 2013.
methods used to characterize the soil specific surface area of clays. Appl. Clay Sci. Temperature Dependence of Thermal Properties of Sands across a Wide Range of
83-84, 144–152. https://doi.org/10.1016/j.clay.2013.08.026. Temperatures (30–70◦ C).
Madsen, F.T., 1998. Clay mineralogical investigations related to nuclear waste disposal. Steele-Dunne, S.C., Rutten, M.M., Krzeminska, D.M., Hausner, M., Tyler, S.W., Selker, J.,
Clay Miner. 33, 109–129. https://doi.org/10.1180/000985598545318. Bogaard, T.A., de Giesen, N.C.v., 2010. Feasibility of soil moisture estimation using
Makowski, M.W., Mochlinski, K., 1956. An evaluation of two rapid methods of assessing passive distributed temperature sensing. Water Resour. Res. 46, W03534 https://doi.
the thermal resistivity of soil. In: Proceedings of the IEE - Part A: Power Engineering, org/10.1029/2009wr008272.
103, pp. 453–470. https://doi.org/10.1049/pi-a.1956.0123. Sun, Q., Lyu, C., Zhang, W., 2020. The relationship between thermal conductivity and
María, V., Villar, M., 2004. Thermo-hydro-mechanical characteristics and processes in electrical resistivity of silty clay soil in the temperature range − 20 C to 10 C. Heat
the clay barrier of a high level radioactive waste repository. State of the art report. Mass Transf. https://doi.org/10.1007/s00231-020-02813-0.
Maxwell, J.C., 1881. A Treatise on Electricity and Magnetism, 1. Clarendon press. Tang, A., Cui, Y., 2006. Determining the thermal conductivity of compacted MX80 clay.
McInnes, K.J., 1981. Thermal Conductivities of Soils from Dryland Wheat Regions of In: Unsaturated Soils 2006. ASCE, pp. 1695–1706.
Eastern Washington. M.Sc. Washington State University. Tang, A.M., Cui, Y.J., 2007. Experimental evidences on thermo-hydro-mechanical
Mozumder, R.A., Laskar, A.I., 2015. Prediction of unconfined compressive strength of coupling in engineered clay barrier for deep nuclear waste disposal. In: Geo-Denver
geopolymer stabilized clayey soil using Artificial Neural Network. Comput. Geotech. 2007. Denver, Colorado, United States, pp. 1–10.
69, 291–300. https://doi.org/10.1016/j.compgeo.2015.05.021. Tang, A.-M., Cui, Y.-J., Le, T.-T., 2008. A study on the thermal conductivity of compacted
Mügler, C., Filippi, M., Montarnal, P., Martinez, J.M., Wileveau, Y., 2006. Determination bentonites. Appl. Clay Sci. 41, 181–189. https://doi.org/10.1016/j.
of the thermal conductivity of opalinus clay via simulations of experiments clay.2007.11.001.
performed at the Mont Terri underground laboratory. J. Appl. Geophys. 58, Tang, C.-S., Shi, B., Liu, C., Suo, W.-B., Gao, L., 2011. Experimental characterization of
112–129. https://doi.org/10.1016/j.jappgeo.2005.05.002. shrinkage and desiccation cracking in thin clay layer. Appl. Clay Sci. 52, 69–77.
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part I - https://doi.org/10.1016/j.clay.2011.01.032.
A discussion of principles. J. Hydrol. 10, 282–290. https://doi.org/10.1016/0022- Tarnawski, V.R., Leong, W., 2012. A series-parallel model for estimating the thermal
1694(70)90255-6. conductivity of unsaturated soils. Int. J. Thermophys. 33, 1191–1218. https://doi.
Newson, T., Brunning, P., Stewart, G., 2002. Thermal conductivity of consolidating org/10.1007/s10765-012-1282-1.
offshore clayey backfill. In: ASME 2002 21st International Conference on Offshore Tarnawski, V.R., Momose, T., Leong, W.H., 2009. Assessing the impact of quartz content
Mechanics and Arctic Engineering. American Society of Mechanical Engineers, on the prediction of soil thermal conductivity. Géotechnique 59, 331–338. https://
pp. 17–24. doi.org/10.1680/geot.2009.59.4.331.
Nikolaev, I., Leong, W., Rosen, M., 2013. Experimental investigation of soil thermal Tarnawski, V.R., Momose, T., Leong, W., Piper, D.W., Gliński, J., 2011. Estimation of
conductivity over a wide temperature range. Int. J. Thermophys. 34, 1110–1129. quartz content in mineral soils. In: Horabik, J., Lipiec, J. (Eds.), Encyclopedia of
https://doi.org/10.1007/s10765-013-1456-5. Agrophysics. Springer Netherlands, pp. 275–280.
Ochsner, T.E., Horton, R., Ren, T., 2001. A new perspective on soil thermal properties. Tarnawski, V.R., McCombie, M.L., Leong, W.H., Wagner, B., Momose, T.,
Soil Sci. Soc. Am. J. 65, 1641–1647. https://doi.org/10.2136/sssaj2001.1641. Schönenberger, J., 2012. Canadian field soils II. Modeling of quartz occurrence. Int.
Oriol, F., De Boodt, M., Verdonck, O., Cappaert, I., 1978. Thermal properties of two J. Thermophys. 33, 843–863. https://doi.org/10.1007/s10765-012-1184-2.
organic (peat, pine bark) and two inorganic (perlite, clay) horticultural substrates. Thomas, H.R., He, Y., Ramesh, A., Zhou, Z., Villar, M.V., Cuevas, J., 1994. Heating
Catena 5, 389–394. https://doi.org/10.1016/0341-8162(78)90021-8. unsaturated clay-an experimental and numerical investigation. In: Smith, I.M. (Ed.),
Ould-Lahoucine, C., Sakashita, H., Kumada, T., 2002. Measurement of thermal Proc. 3rd European Conference on Numercal Methods in Geotechnical Engineering.
conductivity of buffer materials and evaluation of existing correlations predicting it. ECONMIG 94, Manchester, UK, pp. 181–186.
Nucl. Eng. Des. 216, 1–11. https://doi.org/10.1016/S0029-5493(02)00033-X. Tian, Z., Lu, Y., Horton, R., Ren, T., 2016. A simplified de Vries-based model to estimate
Paterson, E., 2018. Specific surface area and pore structure of allophanic soil clays. Clay thermal conductivity of unfrozen and frozen soil. Eur. J. Soil Sci. 67, 564–572.
Miner. 12, 1–9. https://doi.org/10.1180/claymin.1977.012.1.01. https://doi.org/10.1111/ejss.12366.
Peng, S.Z., Ding, Y.X., Liu, W.Z., Li, Z., 2019. 1 km monthly temperature and Tien, Y.M., Chu, C.A., Chuang, W.S., 2005. The Prediction Model of Thermal
precipitation dataset for China from 1901 to 2017. Earth Syst. Sci. Data 11, Conductivity of Sand-Bentonite Based Buffer Material.
1931–1946. https://doi.org/10.5194/essd-11-1931-2019. Tong, F., Jing, L., Zimmerman, R.W., 2009. An effective thermal conductivity model of
Pham, H.Q., Fredlund, D.G., Barbour, S.L., 2005. A study of hysteresis models for soil- geological porous media for coupled thermo-hydro-mechanical systems with
water characteristic curves. Can. Geotech. J. 42, 1548–1568. https://doi.org/ multiphase flow. Int. J. Rock Mech. Min. Sci. 46, 1358–1369. https://doi.org/
10.1139/t05-071. 10.1016/j.ijrmms.2009.04.010.
Piñeiro, G., Perelman, S., Guerschman, J.P., Paruelo, J.M., 2008. How to evaluate van Rooyen, M., Winterkorn, H.F., 1959. Structural and Textural Influences on Thermal
models: Observed vs. predicted or predicted vs. observed? Ecol. Model. 216, Conductivity of Soils.
316–322. https://doi.org/10.1016/j.ecolmodel.2008.05.006. Villar Galicia, M.V., 2002. Thermo-Hydro-Mechanical Characterisation of a Bentonite
Ren, T., Ochsner, T.E., Horton, R., 2003. Development of thermo-time domain from Cabo de Gata A Study Applied to the Use of Bentonite as Sealing Material in
reflectometry for vadose zone measurements. Vadose Zone J. 2, 544–551. https:// High Level Radioactive Waste Repositories. Universidad Complutense de Madrid, Ph.
doi.org/10.2136/vzj2003.5440. D.
Reno, W.H., Winterkorn, H.F., 1967. The Thermal Conductivity of Kaolinite Clay as a Villar, M.V., García-Siñeriz, J.L., Bárcena, I., Lloret, A., 2005. State of the bentonite
Function of Type of Exchange Ion, Density and Moisture Content. barrier after five years operation of an in situ test simulating a high level radioactive
Revil, A., 2000. Thermal conductivity of unconsolidated sediments with geophysical waste repository. Eng. Geol. 80, 175–198. https://doi.org/10.1016/j.
applications. J. Geophys. Res. Solid Earth 105, 16749–16768. https://doi.org/ enggeo.2005.05.001.
10.1029/2000JB900043. Wang, Y., Cui, Y.-J., Tang, A.M., Tang, C.-S., Benahmed, N., 2016. Changes in thermal
Rizvi, Z., Zaidi, H., Akhtar, S., Shoarian Sattari, A., Wuttke, F., 2019. Soft and hard conductivity, suction and microstructure of a compacted lime-treated silty soil
computation methods for estimation of the effective thermal conductivity of sands. during curing. Eng. Geol. 202, 114–121. https://doi.org/10.1016/j.
Heat Mass Transf. 56, 1947–1959. enggeo.2016.01.008.
Rizvi, Z., Akhtar, S., Sabeeh, W., Wuttke, F., 2020. Effective Thermal Conductivity of Wang, J., He, D., Dyck, M., He, H., 2020a. Theory and solutions of heat pulse method for
Unsaturated Soils Based on Deep Learning Algorithm. determining soil thermal properties. IOP Conf. Series Earth Environ. Sci. 440 https://
Ross, P.J., 2014. Estimation of soil thermal properties in two and three dimensions. doi.org/10.1088/1755-1315/440/5/052039.
Vadose Zone J. 13 https://doi.org/10.2136/vzj2013.07.0143. Wang, J., He, H., Dyck, M., Lv, J., 2020b. A review and evaluation of predictive models
RÓżański, A., Stefaniuk, D., 2016. On the prediction of the thermal conductivity of for thermal conductivity of sands at full water content range. Energies 13, 1083.
saturated clayey soils: Effect of the specific surface area. Acta Geodyn. Geomater. 13, https://doi.org/10.3390/en13051083.
184. https://doi.org/10.13168/AGG.2016.0016. Wen, H., Bi, J., Guo, D., 2020. Calculation of the thermal conductivities of fine-textured
Sakashita, H., Kumada, T., 1998. Heat transfer model for predicting thermal conductivity soils based on multiple linear regression (MLR) and artificial neural networks (ANN).
of highly compacted bentonite. J. Atomic Energy Soc. Japan 40, 235–240. Eur. J. Soil Sci. https://doi.org/10.1111/ejss.12934. n/a.
Schönenberger, J., Momose, T., Wagner, B., Leong, W.H., Tarnawski, V.R., 2012. Whittig, L.D., 1965. X-Ray diffraction techniques for mineral identification and
Canadian field soils I. Mineral composition by XRD/XRF measurements. Int. J. mineralogical composition. Methods Soil Anal. 671–698.
Thermophys. 33, 342–362. https://doi.org/10.1007/s10765-011-1142-4. Wiebe, B., Graham, J., Tang, G.X., Dixon, D., 1998. Influence of pressure, saturation, and
Sen, P.N., Scala, C., Cohen, M.H., 1981. A self-similar model for sedimentary rocks with temperature on the behaviour of unsaturated sand-bentonite. Can. Geotech. J. 35,
application to the dielectric constant of fused glass beads. Geophysics 46, 769–783. 194–205. https://doi.org/10.1139/t97-093.
Sepaskhah, A.R., Tabarzad, A., Fooladmand, H.R., 2010. Physical and empirical models Woodside, W., Messmer, J.H., 1961. Thermal conductivity of porous media. I.
for estimation of specific surface area of soils. Arch. Agron. Soil Sci. 56, 325–335. Unconsolidated sands. J. Appl. Phys. 32, 1688–1699.
Sihvola, A.H., Kong, J.A., 1988. Effective permittivity of dielectric mixtures. Geosci. Xie, X., Lu, Y., Ren, T., Horton, R., 2020. Thermal conductivity of mineral soils relates
Remote Sens. IEEE Trans. on 26, 420–429. linearly to air-filled porosity. Soil Sci. Soc. Am. J. https://doi.org/10.1002/
Simunek, J., Huang, K., Van Genuchten, M.T., 1997. The HYDRUS-ET software package saj2.20016.
for simulating the one-dimentional movement of water, heat and multiple solutes in Xu, X., Zhang, W., Fan, C., Li, G., 2019a. Effects of temperature, dry density and water
variably-saturated media, Version 1.1. Bratislava: Inst. Hydrology Slovak Acad. Sci. content on the thermal conductivity of Genhe silty clay. Res. Phys. 16, 102830.
https://doi.org/10.1016/j.rinp.2019.102830.

16
L. Liu et al. Engineering Geology 288 (2021) 106107

Xu, Y., Sun, D.A., Zeng, Z., Lv, H., 2019b. Effect of aging on thermal conductivity of Zhang, Y., Schaap, M.G., 2019. Estimation of saturated hydraulic conductivity with
compacted bentonites. Eng. Geol. 253, 55–63. https://doi.org/10.1016/j. pedotransfer functions: A review. J. Hydrol. 575, 1011–1030. https://doi.org/
enggeo.2019.03.010. 10.1016/j.jhydrol.2019.05.058.
Xu, Y., Sun, D.A., Zeng, Z., Lv, H., 2019c. Effect of temperature on thermal conductivity Zhang, M., Shi, W., 2020. Systematic Comparison of Five Machine-Learning Methods in
of lateritic clays over a wide temperature range. Int. J. Heat Mass Transf. 138, Classification and Interpolation of Soil Particle Size Fractions Using Different
562–570. https://doi.org/10.1016/j.ijheatmasstransfer.2019.04.077. Transformed Data.
Xu, Y., Sun, D.A., Zeng, Z., Lv, H., 2019d. Temperature dependence of apparent thermal Zhang, N., Yu, X., Pradhan, A., Puppala, A.J., 2016. A new generalized soil thermal
conductivity of compacted bentonites as buffer material for high-level radioactive conductivity model for sand–kaolin clay mixtures using thermo-time domain
waste repository. Appl. Clay Sci. 174, 10–14. https://doi.org/10.1016/j. reflectometry probe test. Acta Geotech. 12, 739–752. https://doi.org/10.1007/
clay.2019.03.017. s11440-016-0506-0.
Yan, H., He, H., Dyck, M., Jin, H., Li, M., Si, B., 2019. A generalized model for estimating Zhang, M., Zhang, X., Xu, X., Lu, J., Pei, W., Xiao, Z., 2017a. Water–heat migration and
effective soil thermal conductivity based on the Kasubuchi algorithm. Geoderma frost-heave behavior of a saturated silty clay with a water supply. Exp. Heat Transf.
353, 227–242. https://doi.org/10.1016/j.geoderma.2019.06.031. 30, 1–13. https://doi.org/10.1080/08916152.2017.1312639.
Yang, Y., 2015. Evapotranspiration over Heterogeneous Vegetated Surfaces: Models and Zhang, N., Yu, X., Pradhan, A., 2017b. Application of a thermo-time domain
Applications. Springer. reflectometry probe in sand-kaolin clay mixtures. Eng. Geol. 216, 98–107. https://
Ye, W., Chen, Y., Chen, B., Wang, Q., Wang, J., 2010. Advances on the knowledge of the doi.org/10.1016/j.enggeo.2016.11.016.
buffer/backfill properties of heavily-compacted GMZ bentonite. Eng. Geol. 116, Zhang, M., Lu, J., Lai, Y., Zhang, X., 2018a. Variation of the thermal conductivity of a
12–20. https://doi.org/10.1016/j.enggeo.2010.06.002. silty clay during a freezing-thawing process. Int. J. Heat Mass Transf. 124,
Yi, S., Li, N., Xiang, B., Wang, X., Ye, B., McGuire, A.D., 2013. Representing the effects of 1059–1067. https://doi.org/10.1016/j.ijheatmasstransfer.2018.02.118.
alpine grassland vegetation cover on the simulation of soil thermal dynamics by Zhang, Y., Schaap, M.G., Zha, Y., 2018b. A high-resolution global map of soil hydraulic
ecosystem models applied to the Qinghai-Tibetan Plateau. J. Geophys. Res. properties produced by a hierarchical parameterization of a physically-based water
Biogeosci. 118, 1186–1199. https://doi.org/10.1002/jgrg.20093. retention model. Water Resour. Res. 54, 9774–9790. https://doi.org/10.1029/
Yoon, S., Cho, W., Lee, C., Kim, G.-Y., 2018. Thermal Conductivity of Korean Compacted 2018wr023539.
Bentonite Buffer Materials for a Nuclear Waste Repository, 11, p. 2269. Zhang, Y., Schaap, M.G., Wei, Z., 2020. Development of hierarchical ensemble model and
Yu, K.L., Singh, R.M., Bouazza, A., Bui, H.H., 2015. Determining soil thermal estimates of soil water retention with global coverage. Geophys. Res. Lett. 47,
conductivity through numerical simulation of a heating test on a heat exchanger e2020GL088819 https://doi.org/10.1029/2020gl088819.
pile. Geotech. Geol. Eng. 33, 239–252. https://doi.org/10.1007/s10706-015-9870-z. Zhao, B., Mao, K., Cai, Y., Shi, J., Li, Z., Qin, Z., Meng, X., 2019. A combined Terra and
Aqua MODIS land surface temperature and meteorological station data product for
China from 2003–2017.

17

You might also like