Energy Conversion and Management 95 (2015) 352–370

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Numerical simulation of heat dissipation processes in underground

power cable system situated in thermal backfill and buried
in a multilayered soil
Paweł Ocłoń a,⇑, Piotr Cisek a, Marcin Pilarczyk a, Dawid Taler b
a
Institute of Thermal Power Engineering, Faculty of Mechanical Engineering, Cracow University of Technology, al. Jana Pawła II 37, PL-31-864 Kraków, Poland
b
Institute of Thermal Engineering and Air Protection, Faculty of Environmental Engineering, Cracow University of Technology, ul. Warszawska 24, PL-31-155 Kraków, Poland

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents the thermal analysis of the underground transmission line, planned to be installed in
Received 4 November 2014 one of the Polish power plants. The computations are performed by using the Finite Element Method
Accepted 31 January 2015 (FEM) code, developed by the authors. The paper considers a system of three power cables arranged in
Available online 3 March 2015
flat (in-line) formation. The cable line is buried in the multilayered soil. The soil layers characteristic
and thermal properties are determined from geological measurements. Different conditions of cable
Keywords: bedding are analyzed including power cables placement in the FTB or direct burial in a mother ground.
Underground power cables
The cable line burial depth, measured from the ground level, varies from 1 m to 2.5 m. Additionally, to
Finite Element Method
Thermal conductivity
include the effect of dry zones formation on the temperature distribution in cable line and surroundings,
Fluidized thermal backfill soil and FTB thermal conductivities are considered as a temperature-dependent. The proposed approach
for determining the temperature-dependent thermal conductivity of soil layers is discussed in detail. The
FEM simulation results are also compared with the results of the simulation that consider soil layers as
homogeneous materials. Therefore, thermal conductivity is assumed to be constant for each layer. The
results obtained by using the FEM code, developed by the authors, are compared with the results of
ANSYS simulations, and a good agreement was found.
 2015 Elsevier Ltd. All rights reserved.

1. Introduction assumes that the soil is a homogeneous medium with constant

The underground power cable transmission line operates at the
maximum possible electric current. Cable ampacity is a maximum
electrical current that may be safely transferred without exceeding
temperature limitations. Cable ampacity may be also described as
current carrying capacity. The current carrying capacity mostly
depends on the conductor temperature. Thus, heat dissipation pro-
cess from the cable conductor to the surrounding soil plays a cru-
cial role in evaluating the performance of buried cable systems
[1,2]. First Joule’s law states that cable electric resistance and cur-
rent load affects heat generation rate from the cable core. When
the cable core temperature is too high, the cable conductor over-
heat may experience. Thus, cable insulation meltdown may occur
what leads directly to the transmission line malfunction.
The traditional method (IEC and IEEE-Standards [3,4]) is often
used in calculating the thermal resistance between the cable sys-
tem and the external environment. The standard procedure
⇑ Corresponding author.
E-mail address: poclon@mech.pk.edu.pl (P. Ocłoń).
thermal conductivity. In fact, the soil is multilayered and consists
of quartz, organic matter, and other materials which thermal prop-
erties differ from each other. In the general case soil shall be
regarded as a porous medium. Heat conduction from the hot cable
to surroundings depends on the thermal conductivity of each soil
layer, which in turn depends on the porosity, liquid–vapor trans-
port and temperature [5]. When the porosity is considerable, and
the pores are saturated with water, the thermal conductivity of soil
increases [6,7]. Furthermore, the soil layers thermal conductivity
decreases with a rise in temperature [8]. When liquid water, which
is existing in the soil pores, evaporates then the soil is drying out
and loses its ability to conduct heat. Moreover, the trench shape,
power cables arrangement type and configuration of the soil layers
influence the temperature distribution in the ground, bedding lay-
er and cable core.
Many studies have been performed recently to understand the
heat transfer in the porous media. The conducted analysis include,
among others: liquid–vapor heat and mass transfer [5,9]; flow and
thermal non-equilibrium through a porous medium [10]; the local
thermal non-equilibrium condition of porous media [11]; heat

http://dx.doi.org/10.1016/j.enconman.2015.01.092
0196-8904/ 2015 Elsevier Ltd. All rights reserved.
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370 353

Nomenclature

A cross-sectional area, m2 DQ heat losses from the power cable, W/km

a soil thermal conductivity equation coefficient DT cable core and soil temperature difference, C
C capacitance for the cable core situated in a vacuum, F/m e dielectric loss factor, –
d diameter, m q density, kg/m3
E total number of finite elements in FEM model qd soil dry density, kg/m3
f alternating current (AC) frequency, Hz qs solids unit weight, kg/m3
fe element load vector, W/m
H burial depth, m Subscripts
I current load, A AC alternating current
gidf grid independence factor, – c cable conductor
Ke Kersten number, – DC direct current
k thermal conductivity, W/(m K) dry dry soil
l distance between conductor axes, m e external
n porosity, – IEC according to IEC Standards
N total number of nodes in FEM model, – ins cable insulation
q quartz content, – j cable jacket
qv heat source per cable core unit volume, W/m3 opt optimal value
Re electric resistance, X/km ref reference value
Rh thermal resistance, (m K)/W s solid
r radius, m sat saturated state
Sr degree of saturation, – sh cable sheath
T temperature at any point in the x–y plane around the w water
underground cable, C wet soil in a moist state
Te average temperature within a triangular finite element,
C Superscript
Tmax central cable core maximum temperature, C
e element number
Tmax,p maximum allowable cable core temperature, C i iteration number
tan d insulation loss factor, –
U peak voltage, kV
Ud RMS voltage, kV Matrices and vectors (Appendix A)
Wd dielectric losses, W/m {f} global load vector, W/m
{J} Jacobian matrix
wn gravimetric natural water content, %
wsat gravimetric saturation water content, % [K] global stiffness matrix, W/(m K)
x distance from the symmetry plane, m {q} nodal coordinates vector, m
y distance from the ground surface, m {T} nodal temperatures vector, C
{DT}i difference between nodal temperatures in i-th and
ys skin effect factor, –
yp proximity effect factor, – i  1-th iterations
{U} functions vector used in Newton–Raphson algorithm
Greek symbols
a temperature coefficient, –
 absolute value of relative error, %

transfer through the porous extended surfaces [12–15]; flow
through porous channels, pipes and cavities [16–21]; and combus-
tion applications [22]. The modeling of flow and heat transfer pro-
cesses in porous materials is a very complicated and challenging
task. Therefore, the presented study attempts to propose an alter-
native approach that includes the effect of dry zones formation on
soil thermal conductivity. The conductivity drop in a proximity of
cables is modeled via the exponential function of temperature
dependent thermal conductivity.
Heat dissipation process through the underground power cables
and the surrounding soil was considered in [23–29], among others.
Different authors performed both the numerical simulations and
the experimental investigations on underground power cable sys-
tems. Hwang and Jiang [23] presented a combined nonlinear mag-
netothermal analysis, including radiation effects, for calculating
the thermal fields of an underground cable system. Al-Saud et al.
[24] performed the numerical computations of temperature distri-
bution in soil and underground power cables. A new perturbed
finite-element analysis technique was used. Al-Saud’s approach
involves the use of derived sensitivity coefficients associated with
various cable parameters of interest, and use these coefficients to
achieve optimal cable performance. De Lieto et al. [25] performed
a numerical study based on a Control-Volume Formulation of the
Finite-Difference Method. That technique was used to determine
the thermal resistance existing between an underground electrical
power cable and the ground surface. The thermal behavior of the
cables system was studied for various dimensions of the trench,
cable burial depth, two backfill layers thicknesses, and the cable
bedding. The authors developed a semi-empirical correlating equa-
tion using multiple regression procedures and verified proposed
model experimentally in [26].
The comprehensive approach to determining the cable ampa-
city was presented by Kroener et al. [5]. The authors developed
the new numerical model of coupled liquid water, vapor and heat
flow in a thermal system consists of underground cable buried in
the soil. The transient computations performed using the Finite
Element Method were verified experimentally and demonstrated
the strong relation of the cable temperature on soil water content.
The serviceable simplified analytical model of transient heat
dissipation from underground cable to surroundings was devel-
oped by Papagiannopoulos et al. [27]. The authors considered both
the thermal impedances and the thermal resistances that influence
the dynamic behavior of the thermo-electrical system. The analyti-
cal model results were further compared with the results obtained
354 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370
using the commercial COMSOL software. Moreover, the authors
verified their computations with the experimental procedure,
and a satisfactory agreement was found. In Ref. [28] Chatziathana-
siou et al. presented an experimental study, performing a dynamic
thermal analysis of underground cables. The thermal impedance
and the two new representations: the thermal time constant distri-
bution and the structure functions were utilized. In [29] Wie˛cek
et al. focused on the harmonic analysis of dynamic thermal prob-
lems in an overhead transmission line and a buried power cable.
The authors calculated the temperature delay time with respect
to the power peaks.
Based on the performed literature survey, it may be concluded
that many different approaches are used to analyze the thermal
behavior of underground power cables. The associated physics is
complex involving the thermal, electrical and moisture migration
processes. Therefore, still the development of alternative methods
to analyze the heat dissipation processes from the underground
power cables to their surroundings is necessary.
This paper presents thermal analysis of the underground trans-
mission line, planned to be installed in the Polish CHP plant that
delivers 600 MW of electrical power. The paper considers the pow-
er cable system placed in thermal backfill and buried in the multi-
layered soil in an in-line arrangement. In the presented case study,
the burial depth of the underground cable line, measured from the
ground level, varies from 1 m to 2.5 m.
The numerical analysis performed in this paper compares two
different cables laying conditions:
1. The first considered case includes power cables system
arranged in flat formation, situated in the Fluidized Thermal
Backfill (FTB), and buried in the multilayered soil.
2. The second solution, economically less expensive, considers the
cables placement directly in a multilayered soil. In this case the
FTB bedding is not applied.
This study proposes a simplified Finite Element Method (FEM)
computational methodology applied to underground power cable
system. The analysis considers the multilayered soil as a material
with temperature dependent thermal conductivity. Thus, the effect
of a decrease in soil and FTB bedding layers thermal conductivities
with temperature rise is included. Therefore, the analytical formula
to determine the thermal conductivity of each layer is proposed.
The obtained computational results are compared with those
obtained under the assumption that soil layers are made of a
homogeneous material with constant (temperature independent)
thermal conductivity. It should be noted that the thermal proper-
ties of each soil layers are determined on the basis of the ground
geological measurements. The geological analysis was carried out
in a location where the power transmission line is planned to be
installed. The procedure for determining the temperature-depen-
dent thermal conductivity considers the empirical formulas that
may be easily found in the literature [6,43,47]. Therefore, the pro-
posed computational approach can be directly implemented in
FEM procedures for heat conduction.
Based on the performed thermal analysis, the best configuration
of cable placement is selected. The aim of the conducted computa-
tions is to achieve the lowest conductor temperature determined
using the Finite Element Method [30].
2. Underground power cables installation types
The power cables connected to a 3-phase circuit transmission
line, are installed in a different configurations. Typically designed
formations include trefoil (triangular) and flat (in-line) arrange-
ments (Fig. 1). The choices between the presented types of power
Fig. 1. Installation types for buried cables; 1 – warning tape; 2 – warning grid; 3 –
mother ground; 4 – concrete protective layer; 5 – thermal backfill, 6 – power cable.
cable placement depend on several factors like screen bonding
method, conductor cross-section area and available space for
installation.
Heat transfer conditions enhancement allow to decrease the
power cable conductor cross-sectional area. Thereby it significant-
ly reduce the underground power cable installation costs, which
are greater when total length of the underground transmission
lines is taken into account.
The primary method to increase the maximum allowable elec-
trical load is enhancing the amount of heat dissipated from the
cables to its surroundings. That may be achieved e.g. by replacing
the mother ground around the cable with a thermal backfill mate-
rial, with significantly higher thermal conductivity. What should
be mentioned, the gain is larger for a dry thermal backfill, e.g.
dry FTB exhibits up to five times higher thermal conductivity than
the dry soil. For safety reasons, the top surface of the backfill layer
is usually protected with a concrete slab, and buried in the native
soil [31].
FTB is an engineered slurry backfill mixture which, when solidi-
fied, develops into an efficient heat conducting medium with the
following specific thermal and mechanical properties [32]. Pro-
posed FTB composition is given in [33] and comprises of: fine
and coarse natural mineral aggregates constituting the bulk vol-
ume of the mixture; cement providing the interparticle bond and
strength; fluidizer to impart a homogeneous fluid consistency for
ease of placement. Moreover, other additives may provide
improvement of the thermal properties.
The power cables laying technique, involving the use of the FTB
bedding, is the most traditional method of protecting the under-
ground transmission line from overheating. That type of arrange-
ment in used when the cable line is laying in non-urbanized
areas, thus vibration, and heavy load do not occur in ground.
Otherwise, e.g. under road crossings, the cables are laying inside
concrete encased PVC or PE ducts.
3. Materials and methods
In this section, the method for determination of the multilay-
ered soil thermal conductivity is presented. Moreover, the thermal
analysis of underground power cables system is discussed in detail.
3.1. Material properties and computational cases
This paper performs thermal analysis of three 400 kV single
core, XLPE insulated, underground power cables arranged in flat
formation. All parameters of the considered power cable are speci-
fied by the producer [34] and given in Table 1.
Fig. 2 depicts the computational cases considered during the
performed analysis. The underground power cables placement
conditions are presented. Three power cables are buried at the
depth H from 1 m to 2.5 m below the ground level in the trench
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370
Table 1
400 kV high voltage power cable characteristic [34].
Conductor cross-sectional area Ac, mm2
Conductor diameter (dc), mm
Insulation total thickness, mm
Cable external diameter (de), mm
DC resistance Re,DC at Tref = 20 C, X/km
AC resistance Re,AC at Topt = 65 C/50 Hz, X/km
Current load at 65 C, A
The maximum allowable temperature of cable operation Tmax,p, C
of 1.4 m width. The spacing between the subsequent cables cores is
equal to 0.4 m.
In the first of the considered laying conditions, the power cables
are situated in FTB bedding of rectangular cross-section with a
height of 0.4 m. Furthermore, the FTB bedding is buried in the
mother ground (Fig. 2a).
In the second considered case, the power transmission line is
situated directly in the mother ground (Fig. 2b). A comparison
between the presented types of cable installation allows to predict
further maximum cable temperature drop when replacing
the mother ground with FTB backfill. Soil thermal conductivity
[35–38] is considered as temperature dependent. The cable
location directly in the mother ground (Fig. 2b) decreases the
installation costs considerably. However, the cable temperature
may increase significantly when compared with a cable line laying
technique using FTB bedding (Fig. 2a). In order to simplify the com-
putations, it is assumed that the symmetry of thermal solution pat-
tern exists. Therefore, only the half of the cables system is
considered in the computations.
The performed geological measurements indicated that the soil
need to be considered as multilayered. Fig. 2 presents a variety of
soil layer types for the examined geological cross-section. The pre-
sented multilayered soil model comprises of, respectively:
(a) surface layer  native soil (NS), with a thickness of 0.8 m,
(b) medium sand (MSa), with a thickness of 0.9 m,
(c) gravel (Gr), with a thickness of 0.2 m,
(d) sand clay loam with gravel (grsaCl), with a thickness of
0.7 m,
Fig. 2. Power cables laying conditions: (a) cables buried in FTB and placed in a multilayered soil, (b) cables buried directly in a multilayered soil.
355
(e) sand clay loam (saCl) which is the bottom layer. It is
assumed that the bottom layer do not change further with
1600 depth.
49.6
27 Soil classification is based on ISO 14688 [45].
133.8
0.0113
As MSa and Gr porous layers (relatively large grain size) are per-
0.0157 meable to water (aquifers), thus their water content is very high,
1145 reaching saturation level: 23.0% and 18.5% respectively. Other ana-
90 lyzed soil layers, saCl and grsaCl, with a significantly smaller grain
size, exhibits their natural water content lower than the saturation
level. Thus, saCl and grsaCl water content equals to 20.5% and
17.0%, with saturation in: 24.0% and 18.0%, respectively. Therefore,
the performed analysis shall include that soil thermal conductivity
is not only a function of temperature but also depends on soil
moisture, among others.
For the first considered type of the underground cable installa-
tion (Fig. 2a), cables are arranged at the bottom of the trench and
placed in the FTB layer. In this case, authors proposed the use of
SGFC mixture (Sand, Gravel, Fly ash, and Cement-mix), according
to [33]. SGFC is one of the FTB types and consists of 41% of fine
aggregate, 49% of coarse aggregate, 2.5% of cement and 7.5% of
fly ash. The fly ash is used as a fluidizer. The proposed mixture
takes the given dry density of 2187 kg/m3. Moreover, the SGFC
thermal conductivity equals to 1.54 W/(m K), while totally dry,
and 4.35 W/(m K) when reaching saturation level [33].
3.1.1. Soil thermal conductivity calculation methods
Soil thermal conductivity is a parameter that is in many cases
difficult to estimate. Therefore, the geological analysis shall be per-
formed in order to determine the soil layers properties. Farouki
[48] indicated that soil thermal conductivity depends not only on
water content, but also on mineral composition (grain size and
quartz content), natural porosity, dry density and temperature.
Thermal conductivities and also other physical properties of differ-
ent soil constituents are given in Table 2.
In the literature, many semi-empirical methods to determine
soil thermal conductivity may be found. Farouki [48] presented a
comprehensive review of 11 methods including: Kersten [49], de
Vries and Afgan [50,51], Johansen [44], Sepaskah and Boersman
[52] and Kasubuchi [53] methods, among others. In [48], Farouki
compared predicted and measured values for a broad range of
356 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370

Table 2 Table 3
Thermal properties of various soil constituents (after Farouki [48]) [6]. Median quartz content for different soil types [6].

Material Density Specific heat (J/ Heat capacity Thermal Soil texture Sand Gravel Sand clay loam with gravel Sand clay loam
(kg/m3) (kg K)) (kJ/(m3 K)) conductivity
Quartz content (%) 0.92 0.92 0.60 0.52
(W/(m K))
Quartz 2650 733 1942 8.4
Soil minerals 2650 733 1942 2.9 3. Determination of the Kersten number Ke
Soil organics 1300 1926 2503 0.25
Water 1000 4186 4186 0.6
The Kersten number is a function only of the degree of satura-
Ice 900 2093 1883 2.5
Air 1.20 1005 1.20 0.026
tion Sr. For unfrozen soils:

1 þ 0:7log10 Sr for Sr > 0:05; coarse soil;
Ke ¼ ð5Þ
1 þ log10 Sr for Sr > 0:10; fine soil:
grain size (fine and coarse soil) across a range of soil water content.
The performed analysis indicated that a modified Johansen method Degree of saturation for analyzed soil types is calculated as:
[44] is superior to all other methods [6]. However, for dry fine soils wn
Sr ¼ ; ð6Þ
de Vries [51] approach may be applied and gives satisfactory wsat
agreement with experiment.
where wn and wsat are, respectively, the gravimetric water content
According to Farouki, for fine unfrozen soils the Johansen’s
in naturally occurred soil state and in a saturated state, both given
method is the most accurate. For saturations under 20%, Farouki
in %.
indicated that Johansen’s original method underestimates thermal
conductivity by about 5–15%. Therefore, a slightly modified
4. Calculating of the soil thermal conductivity kwet from Eq. (1).
method in this case is recommended and presented by Peters-
Soil layers properties required to determine the soil thermal
Lidard et al. [6].
conductivity are listed in Table 4. All the parameters for various
Johansen [44] considered soil thermal conductivity as a func-
soil types were obtained from the experiment.
tion of water content, porosity, quartz content and dry density.
Geological analysis was performed to assess ground conditions
Proposed thermal conductivity kwet correlation is a combination
at place of power cable installation.
of the dry kdry and saturated ksat thermal conductivities, weighted
by a normalized thermal conductivity Ke, known as the Kersten
3.1.2. Thermal conductivity of a multilayered soil
number:
IEC standard [3] for Poland region assumes constant thermal
kwet ¼ kdry þ K e ðksat  kdry Þ: ð1Þ conductivity of a native soil (kIEC) equal to 1.00 W/(m K). In fact,
soil shall be considered as a porous material with temperature-
The modified Johansen’s method for determining soil thermal
dependent thermal properties.
conductivity includes four steps [6]:
The preliminary thermal analysis of the studied high voltage
transmission line have been performed in [46]. Cisek et al. consid-
1. Determination of the dry thermal conductivity kdry
ered soil as a homogeneous material with constant thermal prop-
erties. This paper presents a different, more comprehensive
Johansen [44] proposed a semi-empirical equation to predict
approach for determining the thermal conductivity of a multilay-
the dry thermal conductivity for natural soils kdry:
ered soil. Water existing in the pores may locally evaporate near
0:135qd þ 64:7 the cable when the cable heat load and temperature are high. Ther-
kdry ¼ ; ð2Þ
qs  0:947qd mal conductivity of the vapor phase is up to 20-times lower than
for liquid. Hence, the heat transfer conditions deteriorate sig-
where qd and qs are, respectively, the soil dry density and the solids
nificantly with an increase in vapor phase content in the soil or
unit weight, both given in kg/m3.
FTB bedding pores. Kroener et al. [5] presented the complex and
extensive numerical approach. The thermal performance assess-
2. Determination of the saturated thermal conductivity ksat
ment of underground power cables, including liquid–vapor trans-
Saturated thermal conductivity ksat depends on the soil porosity
port in the soil, was analyzed. However, due to the vast
n and quartz content q:
complexity of the extensive numerical approach, other simplified
ksat ¼ ks
1n n
kw ; ð3Þ models may be used.
The underground transmission line operation under loading
where kw = 0.57 W/(m K) represents the thermal conductivity of conditions may influence moisture migration in the soil. Joule’s
water existing in soil pores. Moreover, the solids thermal conduc- heat generated in the power cable conductor may induce the tem-
tivity ks is calculated as follows: perature gradient in the ground. Therefore, water migrates from
q 1q
ks ¼ kq k0 ; ð4Þ Table 4
Soil properties required to determine thermal conductivity – based on geological
In Eq. (4) kq = 7.7 W/(m K) and denotes the thermal conductivity analysis.
of quartz grains. The thermal conductivity of other soil minerals
Layer Natural Saturation Porosity, Natural Dry Solids unit
k0 = 2.0 W/(m K) is given for q > 0.2, where q is the quartz volume
water water n density, q density weight qs
fraction. content, wn content, wsat (kg/m3) qd (kg/ (kg/m3)
Farouki [48] concluded that only quartz crystals exhibit a sig- (%) (%) m3)
nificant thermal conductivity. However, the conductivity of quartz FTBa 7.0 12.0 (slurry)   2187 
or silicate material bound inside clay or silt particles is similar to MSa 23.0 23.0 0.44 1930 1569 2822
that of other soil materials. Peters-Lidard et al. [6] assumed that Gr 18.5 18.5 0.38 2060 1738 2809
the quartz content for each soil type is related to the sand content grsaCl 20.5 24.0 0.38 2100 1795 2886
saCl 17.0 18.0 0.49 2050 1701 3349
in the investigated soil. Median quartz contents for various soil
a
texture classes, considered in this study, is given in Table 3. Radhakrishna et al. [33].
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370 357

the proximity of the power cable and dry zones are formed. Dry
zones are primarily occurred around the power cables and result-
ing in a significant decrease in soil thermal conductivity. It leads to
deterioration of heat dissipation process and, consequently to
increase in cable core temperature. This paper considers the dry
zones formation in the proximity of the power cable [8]. Dry zones
are formed when the temperature of the mother ground or FTB
bedding layer approaches the maximum allowable cable tem-
perature Tmax,p (see Table 1). The rapid decrease in soil thermal
conductivity k(T), while the cable temperature approaches Tmax,p
may be achieved by utilizing the following relationship [47]:
h  2 i
kðTÞ ¼ kdry þ ðkwet  kdry Þ  exp a1 ðT  T ref Þ=ða2  T max;p Þ ; ð7Þ

where kdry and kwet are the thermal conductivities in dry and wet
conditions (Table 2); Tref = 20 C is the temperature at a ground level
(H = 0 m), and Tmax,p = 90 C is the cable maximum allowable oper-
ating temperature, as specified by the producer. The a1 and a2 coef-
ficients, appearing in Eq. (7), are defined as:
Fig. 3. Soil layers and FTB thermal conductivities as a function of temperature T.
a1 ¼ T max;p =T ref ; ð8Þ Presented correlations are used in simulations for cases with a multilayer ground
profile (Fig. 2a and b).

a2 ¼ 1  ð1=a1 Þ:
Table 5 presents the calculated thermal conductivities and dry
densities of the corresponding soil layers obtained from the mod-
ified Johansen’s method (discussed in details in Section 3.1.1).
For the thermal conductivity model, given by Eq. (7), if T = Tref,
then k(T) = kwet. Moreover, when soil temperature T > Tmax,p, then
k(T) decreases and tends to kdry (see Fig. 3). Proposed formula
simulates dry zone formation in the case when the soil tem-
perature increases, and moisture migration occurs in the immedi-
ate proximity of the cable line. Since the proposed formula for
thermal conductivity takes only the Tref and Tmax,p as input para-
meters, its implementation in the FEM code is straightforward.
In general, the predicted temperature values, using Eq. (7), are
higher than calculated under assumption that k(T) = const. The pro-
posed approach results in ensuring a safety margin, especially
when the soil properties are not sufficiently recognized. Locally,
the soil mean thermal conductivity may be significantly lower than
k(T) = const. The simplified approach for determining the soil ther-
mal conductivity also helps to avoid difficulties associated with
modeling of the moisture transport in the soil. Thus, it may be
directly implemented in the FEM procedures for heat conduction.
The numerical simulation results, including temperature
dependent model of thermal conductivity (Eq. (7)), are compared
with that considering homogeneous conditions: k(T) = const. This
comparison allows to observe the effect of employing FTB bedding
in the analyzed underground power system at the maximum tem-
perature of the cable core. The temperature-independent thermal
ð9Þ
conductivity is assessed for all soil layers and FTB bedding. The cor-
responding values of thermal conductivities, used in this case, are
given in column 6 of Table 5. Soil layers thermal conductivities,
shown in Table 5, were obtained on the basis of the geological ana-
lysis. Soil top layer (NS) has a non-homogeneous structure and
composition (which strongly depends on local geological condi-
tions) which is practically undetermined. Therefore, authors
assume that the thermal conductivity of native soil equals to
kIEC = 1.00 W/(m K), according to IEC standards [3]. Furthermore,
the mother ground, covering the bedding layer, is considered as
non-specified soil with thermal conductivity also equal to
kIEC = 1.00 W/(m K). Thermal conductivity of dry sand varies from
0.2 to 0.4 W/(m K) depending on the literature sources [55]. Thus,
authors assumed that the average thermal conductivity of the dry
native soil (NS) and mother ground (MG) are both equal to 0.3 W/
(m K) (see Table 5). Since the dry zones may form in the vicinity of
underground cable line, the thermal conductivity of FTB bedding
layer is calculated for the dry conditions and equals to 1.54 W/
(m K).
3.2. Numerical determination of the power cable conductor
temperature
The numerical computations are performed to determine the
maximum conductor temperature for the cable installation types,
shown in Fig. 2. Computations are carried out using the FEM code
developed in MATLAB [39]. The following modeling assumptions
are made:

Table 5 1. thermal conductivity of a multilayered soil is temperature

Thermal conductivity and density of applied materials.
dependent;
Layer Dry Solids unit Dry thermal Moist soil Homogeneous 2. the formula for determining the volumetric heat source is pro-
density weight qs conductivity thermal conditions, vided according to IEC 287 Standards [42].
qd (kg/m3) kdry conductivity kwet k(T) = const.
(kg/m3) (W/(m K)) (W/(m K))
Since the cable line length is considerable, thus only the cross-
FTBa 2187  1.54 4.35 1.54
section of the multilayered soil, FTB bedding and underground
NSb   0.30c 1.00 1.00
MGb   0.30c 1.00 1.00 power cables is considered. In practice, due to the various compo-
MSa 1569 2822 0.21 2.28 2.28 sition of a multilayered soil along the transmission line length, the
Gr 1738 2809 0.26 2.67 2.67 resultant temperature field may be three-dimensional. However,
grsaCl 1795 2886 0.26 2.01 2.01 the largest temperature gradients occur across the cross-section
saCl 1701 3349 0.17 1.45 1.45
of the multilayered soil and underground power cable system.
a
Radhakrishna et al. [33]. Therefore, it is assumed that the obtained thermal field do not vary
b
IEC Standard 60287 [3]. along the transmission line length. Hence, the two-dimensional
c
Assumption.
heat transfer problem is formulated. A similar approach was
358 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370

proposed in other studies e.g. [5,8,23,24] and validated experimen- Table 6

tally with a good agreement [5]. The two-dimensional steady-state Thermal properties and thicknesses of 400 kV power cable layout materials.

heat conduction equation [30]: Cable Material Thickness Layer radius Thermal conductivity
    layout (mm) (mm) (W/(m K))
@ @T @ @T
kðTÞ þ kðTÞ ¼ qv ðT c Þ ð10Þ Conductor Copper de = 49.6 rc = 24.8 400
@x @x @y @y Insulation XLPE 30.5 rins = 55.3 0.2875
Sheath Copper 6.4 rsh = 61.7 400
is solved to determine the temperature distribution in the analyzed
Jacket HDPE 5.1 rj = 66.8 0.2875
thermal system. In Eq. (10), T denotes the temperature at any point
in the x–y plane around the underground cable; k(T) is the tem-
perature-dependent thermal conductivity and qv(Tc) is the tem-
perature-dependent volumetric heat source. Its value is calculated The Root-Mean-Square (RMS) voltage Ud is calculated as
according to the procedure presented below for the determined follows:
cable core temperature Tc. U
U d ¼ pffiffiffi ; ð15Þ
2
3.2.1. Determination of volumetric heat source per cable core unit
volume where U = 400 kV is a peak voltage for the considered AC circuit.
The volumetric heat source qv per the cable conductor unit vol- Thermal resistances of XLPE insulation, jacket and cable sur-
ume used in the temperature field calculation in the FEM model is roundings, respectively: Rh,ins, Rh,j and Rh,e, are calculated as:
defined as:


1 r ins 1 rj
Rh;ins ¼ ln ; Rh;j ¼ ln ; Rh;e
DQ 2pkins rc 2pkj r sh
qv ðT c Þ ¼ ; ð11Þ

Ac 1 2H
¼ ln : ð16Þ
where Ac = 0.25pdc2 denotes the cross-sectional area of the cable 2pksoil rj
conductor, calculated based on the conductor diameter dc. The
The cable heat losses may be expressed by introducing alternat-
volumetric heat source is calculated for the cable core temperature
ing current (AC) resistance to first Joule’s law:
Tc determined from Eq. (10). The method of determining qv(Tc)
employs the simplified geometrical model of 400 kV, XLPE insulat- DQ ¼ I2 Re;AC ; ð17Þ
ed, single core underground power cable buried at the depth of H
under the ground level (Fig. 4b). Symbols rc, rins, rsh and rj denote, where
respectively, the cable conductor, insulation, sheath and jacket Re;AC ¼ Re;DC ð1 þ ys þ yp Þ;
outer radiuses (Fig. 4a). The cable external diameter is defined as   ð18Þ
with Re;DC ¼ Re;ref 1 þ aref ðT c  T ref Þ :
de = 2rj, and the conductor diameter is given by dc = 2rc.
Structural material types, thermal conductivities and thickness In Eq. (18), Re,ref and aref are, respectively, the cable conductor
of the individual cable layers are shown in Table 6. electric resistance and the temperature coefficient, both given at
Eq. (12) provides the current rating formula for a buried cable the reference temperature Tref = 20 C. For described computational
without including the effect of dry zone formation [41]: cases (braided copper conductor) Re,ref = 0.0113 X/km and
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi aref = 0.00393 (Table 1). Skin and proximity effects factors are taken
DT  W d ð0:5Rh;ins þ Rh;j þ Rh;e Þ
I¼ ; ð12Þ into account by introducing the ys and yp coefficients:
Re;AC ðRh;ins þ Rh;j þ Rh;e Þ
x4s 8pf
where DT = Tc  Tref is a difference between the cable core Tc and ys ¼ ; x2s ¼ 107 ks ; ð19Þ
192 þ 0:8x4s Re;DC
ground level Tref temperatures. The dielectric losses Wd for the cable
conductor are calculated as: 2 3

2
2
x4p dc 6 dc 1:18
7
W d ¼ 2pf e CU 2d tan d: ð13Þ yp ¼ 40:312 þ 5;
192 þ 0:8x4p l l x4p
192þ0:8x4
þ 0:27
The dielectric loss factor e is equal to 3; the insulation loss fac- p

tor tan d is 0.005 (according to [42,43]); AC frequency f is 50 Hz. 8p f

x2p ¼ 107 kp ; ð20Þ
The reference capacitance C for the cable core situated in a vacuum Re;DC
as a medium (instead of insulation) is defined as:
where ks and kp values are equal to 0.435 and 0.37, respectively,
109 according to [39]; l = 0.4 m is a cable spacing (Fig. 2).
C¼  : ð14Þ
18 ln rrinsc Eqs. (12)–(16) reveal that the current loading I depends on the
temperature difference DT and the cable line burial depth H.

Fig. 4. Simplified model of 400 kV, single core, XLPE insulated underground power cable considered to determine volumetric heat transfer rate qv: (a) power cable cross-
section, (b) underground transmission line laying conditions.
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370 359

Fig. 5 shows the current loading I as a function of DT and H.

The computations performed in the present study consider the
cable burial depth in a range of 1 m to 2.5 m. Therefore, the opti-
mum current loading I varies for different burial depths (see
Fig. 5). However, the practical computations of underground cable
lines consider the operating load as the maximum power cable
load specified by cable producer, i.e. 1145 A (Table 1). The load cor-
responds to the optimum operational temperature of Topt = 65 C
(Table 1). The constant value of I = 1145 A is used in the calculation
of heat losses DQ according to Eq. (17). An alternating current (AC)
resistance Re,AC, that is also needed to determine DQ, is calculated
from Eq. (18) for a given cable conductor temperature Tc. In the
FEM code the Tc temperature is determined from Eq. (10) by using
the FEM method. Based on the obtained values of DQ the qv(Tc)
function is determined from Eq. (11). The cable conductor tem-
perature in the range of 20–100 C is considered in the present cal-
culations. Therefore, the form of qv(Tc) function shown in Fig. 6 is
obtained.
The value of Tc shall be calculated from Eq. (10) for every finite
element that represents the cable conductor layer. The developed
FEM code uses a temperature dependent volumetric heat source
function qv(Tc) when solving the Eq. (10) iteratively.
Fig. 6. Volumetric heat generation rate qv(Tc) as a function of cable core
temperature Tc for I = 1145 A (current loading at Topt = 65 C).
3.2.2. Boundary conditions and computational cases
The appropriate boundary conditions shall be specified to solve
the steady-state heat conduction Eq. (10).
Fig. 7 depicts the types of boundary conditions and computa- ground level given in IEC standards. Since the cable line project
tional cases performed in this study. Equations (21) summarizes is at the design stage, thus the real temperature distributions along
the types of applied boundary conditions: the heat transfer domain boundaries (right and bottom edges of

@T 
the boundary region) are unknown. Therefore, the thermal insula-
@x x¼10m
¼ 0; for the right edge of the boundary region; tion boundary conditions are assumed at a given domain bound-

@T 
@x x¼0
¼ 0; for symmetry plane; aries. Since no heat reception from the surroundings is

@T  considered, thus the thermal insulation boundary condition may
@y
¼ 0; for the bottom edge of the boundary region;
y¼10m
Tðx;y ¼ 0Þ ¼ T ref ¼ 20  C; for the top edge of the boundary region:
ð21Þ
Soil temperature at the ground level is assumed to be Tref = 20 -
C. This value is specified by cable producer [34] and given in IEC
standards for Poland region [3]. In the summer, the temperature
at the ground level may exceed 20 C [40]. However, the computa-
tions presented in this study consider only the temperature at
Fig. 5. Power cable current load I as a function of cable core and soil temperature
difference DT = Tc  Tref.
result in a slightly higher cable conductor temperature than in rea-
lity. The similar approach in specifying boundary conditions was
successfully applied by Kroener et al. in [5], and the experimental
verification confirmed the correctness of the boundary conditions
type.
The right edge of the boundary region is located at a distance of
10 m from the symmetry plane (the distance is measured in a per-
pendicular direction to the symmetry plane). The distance from the
ground level to the bottom edge of the boundary region is also
equal to 10 m (see Fig. 7). A further increase in this distance has
a negligible influence on the maximum temperature of cable con-
ductor. Similarly, increasing the distance from the right boundary
region edge to the symmetry plane does not affect the obtained
maximum temperature of cable conductor.
The performed computational cases (Fig. 7) assumed that:
(a) the multilayered soil thermal conductivity is temperature
dependent (k(T) – const., determined using the formula
(7)) and the cables are laid within FTB bedding layer
(Fig. 7a);
(b) the multilayered soil thermal conductivity is temperature
dependent (k(T) – const.) and the cables are situated in
mother ground (Fig. 7b);
(c) the multilayered soil thermal conductivity is temperature
independent (k(T) = const., Table 5 – homogeneous condi-
tions) and the cables are laid within FTB bedding (Fig. 7c);
(d) the multilayered soil thermal conductivity is temperature
independent (k(T) = const., Table 5 – homogeneous condi-
tions) and the cables are laid directly in mother ground
(Fig. 7d).
Solution algorithm used in the FEM code in order to solve the
heat conduction Eq. (10) is presented in Appendix A.
360 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370

Fig. 7. Boundary conditions for the presented computational cases: (a) cables buried in FTB and placed in a multilayered soil with k(T) – const. – Eq. (7), (b) cables buried
directly in a multilayered soil and placed in a mother ground with k(T) – const., (c) cables buried in FTB and placed in a multilayered soil with k(T) = const., (d) cables buried
directly in a multilayered soil and placed in a mother ground with k(T) = const. (the dimensions shown in Fig. 7 are given in meters).

4. Results and discussion lES

This section presents the computational cases results, shown in
Fig. 7. The temperature distributions are determined using FEM
method for the computational domain, that consists of under-
ground power cables, multilayered soil and FTB bedding, for cable
burial depth H from 1 m to 2.5 m range. In order to provide the grid
independent solution, the grid independence study have been
performed.
4.1. Grid independence study
The grid independence study is performed to determine the
appropriate size of the finite element. The FEM model mesh is gen-
erated by using the ANSYS software [54]. A sample mesh of the
region around the power cables is shown in Fig. 8.
Table 7 lists the average sizes of the finite element edges (lES).
The edge sizes were set for cable layers (conductor, insulation,
sheath, jacket), FTB bedding layer and the multilayered soil (MG,
NS, MSa, Gr, grsaCl, and saCl). Given lES values allowed to obtain
the grid independent solution.
In order to control the mesh size the grid independence factor
gidf defined as:
gidf ¼ ; ð22Þ
lE
is used. The gidf values from 0.05 to 1.5 are considered during the
grid independence study. It allows to define the corresponding
average size of finite element edge lE calculated for each layer.
The ANSYS command AESIZE is applied for this purpose. As a result,
the FEM models of underground power cable system are created.
The exemplary numbers of nodes and finite elements used in the
grid independence studies are listed in Table 8 for two different
cable burial depths: H = 1 m and H = 2.5 m.
The central cable conductor maximum temperature (Tc1,max) is
calculated from FEM model during the grid independence study.
Fig. 9a–d shows how the obtained values of Tc1,max vary with the
increasing total number of nodes N. The cable line burial depth H
from 1 m to 2.5 m is considered.
Two cable placement types: in FTB bedding (‘with FTB’) and
directly in the mother ground (‘without FTB’) are analyzed. The
computational cases that assume the temperature-dependent
thermal conductivities of the multilayered soil and FTB bedding
layer are denoted as (‘k – const.’). Cases assuming the tem-
perature-independent thermal conductivities are denoted as
(‘k = const.’).
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370 361

Fig. 8. Part of the numerical grid, determined for gidf = 1, used in the computation – the close-up on the cable location.

Table 7 maximum temperature is lower when the mother ground is

Average sizes of finite element edge specified for each layer in FEM model for gidf = 1. replaced with FTB bedding. The largest gain is obtained when the
Layer Finite elements edge average size (lES) (m) soil thermal conductivity decreases locally in a proximity of the
Conductor 0.0025
cable line, and the burial depth is large, i.e. H = 2.5 m. In this case
Insulation 0.0025 the (Tc1,max) temperature is up to 7.2 C lower when compared with
Sheath 0.0025 the cable laid directly in mother ground. The before mentioned
Jacket 0.0025 heat transfer conditions are modeled via the temperature-depen-
MG 0.01
dent thermal conductivity of the multilayered soil and FTB bedding
NS 0.02
FTB 0.006 layer, i.e. k(T) – const, using the formula (7).
Msa 0.025 When the thermal conductivity of soil and FTB bedding layers
Gr 0.02 are assumed to be temperature independent, i.e. k(T) = const
grsaCl 0.02
(Table 5 – homogeneous conditions), FTB backfill application does
saCl 0.02
not improve the heat dissipation conditions considerably. When
compared to the cable laying in mother ground, the FTB backfilling
reduces the (Tc1,max) temperature by approx. 2 C.
Table 8 The computational results are obtained by using the FEM code,
The total number of nodes N and elements E created for different values of grid
developed based on the algorithm described in Appendix A. The
independence factor gidf.
same calculations are performed by using the commercial ANSYS
code. The code validation proved that the (Tc1,max) temperatures
obtained using FEM code do not differ more than 0.01 C from
those calculated using ANSYS. In a further part of the results ana-
lysis, the detailed plots are presented for the cable line burial
depths of H = 1 m and H = 2.5 m. Temperature variations in under-
ground power cable system cross-section obtained for computa-
tional cases, shown in Fig. 7 are discussed.

4.2. Temperature distribution in the underground power cable system

The obtained temperature distributions are depicted in Fig. 11,

The corresponding values of gidf are denoted using the dashed
lines. One can observe in Fig. 9 that the gidf equal to 1 provide
the grid independent solution. The obtained maximum tem-
perature of cable conductor for gidf = 1 do not differ more than
0.05 C while increasing the number of nodes in an FEM model.
Therefore, the results obtained using FEM model for which gidf = 1
(Table 8), are presented in this study.
Fig. 10 depicts the central cable conductor maximum tem-
perature (Tc1,max) variations with burial depth H. The four compu-
tational cases shown in Fig. 7 are considered. It can be seen that
the (Tc1,max) temperature increases with burial depth. Application
of FTB backfilling improves the heat dissipation conditions from
the cable to the surroundings. Therefore, the cable conductor
for H = 1 m, and H = 2.5 m. The computational cases are shown in
Fig. 7. Firstly, the computations results considering the tem-
perature-dependent multilayered soil and FTB layer thermal con-
ductivity (i.e. k(T) – const.) and H = 1 m are discussed. Comparing
Fig. 11a with Fig. 11b one can observe that the use of FTB backfill
material leads to lower temperature in the region of the bedding
layer (Fig. 11a). Moreover, a more uniform temperature field with-
in bedding layer can be observed when the FTB bedding is used
instead of the mother ground.
The heat dissipation process enhancement nearby the cable
line, that results in lower temperatures of bedding layer, may be
also seen for the burial depth H = 2.5 m. The bedding layer tem-
peratures are significantly lower when the FTB (Fig. 11e) is applied
instead of mother ground (Fig. 11f). Moreover, the cable conductor
temperatures are lower when the cables are backfilled in FTB.
Dry zone formation effect is simulated by the rapid decrease in
soil and FTB thermal conductivities (Fig. 3) when k(T) – const. The
thermal conductivity drop within the bedding layer region may
result in a considerable increase in temperature. As it can be seen
362 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370

Fig. 9. Results of grid independence studies performed for the computational cases shown in Fig. 7.

in Fig. 11a and e, the application of FTB results in a significantly
lower temperatures in the cable bedding region. Dry zone effect
is not simulated when k(T) = const. (Fig. 11c and g). In this case
larger temperature gradients occur, comparing to k(T) – const.
Considering mother ground as a bedding layer leads to sig-
nificant increase in temperature in the proximity of cable line.
When the dry zone formation occur (k(T) – const.), the tem-
perature gradiends in mother ground layer are higher since the
thermal conductivity is lower than kIEC = 1.00 W/(m K) (compare
Fig. 11b and f with Fig. 11 d and h).
The results of FEM code, developed in this study, are compared
with the ANSYS program. To confront the results directly at the
specified nodal positions three evaluation lines are created
(Fig. 12):
 line 1 that is a vertical line coincident with a symmetry plane;
the line crosses the central cable,
 line 2 that is a vertical line located at the distance of x = 0.4 m
from the symmetry plane; the line crosses the side cable;
 line 3 that is a horizontal line located at a distance of y = H from
the ground level.
The evaluation lines allow to create the temperature profiles
helpful in comparing the exact values of central cable (c1) and side
cable (c2) temperatures. In order to compare directly the maxi-
mum central and side cable temperatures the temperature differ-
ence DTc,max = Tc1,max  Tc2,max is depicted in Fig. 12.
At first, the computational cases shown in Fig. 7a and b are ana-
lyzed. The temperature-dependent thermal conductivity of the
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370
Fig. 10. The maximum conductor temperature as a function of burial depth H –
calculated for the power cable laying conditions shown in Fig. 7 according to the
proposed computational approach.
multilayered soil and FTB bedding material, i.e. k(T) – const., is
considered. Fig. 12a and b compares the temperature distributions
along evaluation lines 1 and 2 for cable line burial depth H equal to
1 m. The temperature variations along line 3 are shown in Fig. 12 c
and d.
For a cable burial depth of H = 1 m (Fig. 12a–d) it can be seen
that:
– maximum temperature of the central cable (c1) conductor is
higher than the determined for the side cable (c2) conductor;
– the temperature difference DTc,max is lower when the cable line
is situated in FTB bedding than in the mother ground (compare
Fig. 12b with Fig. 12c);
– the largest temperature drop occurs across the cable insulation
(approx. 10 K when FTB backfilling is used, and approx. 12 K
363
Fig. 13a–h present the temperature variations along the eval-
uation lines (1–3) for the cable line burial depths H of 1 m
(Fig. 13a–d) and 2.5 m (Fig. 13e–h). The homogenous conditions
k(T) = const. (Table 5), are considered when specifying the thermal
conductivities of multi-layered soil and FTB bedding layer.
The thermal conductivity of FTB bedding material (1.54 W/
(m K)) and mother ground (1.00 W/(m K)) differ slightly. Thus,
the temperature differences calculated at the corresponding loca-
tions are low (Fig. 13). For the cable line burial depth of H = 1 m
the Tc1,max temperature is equal to 42 C when the cables are situat-
ed in FTB (Fig. 13a and c), while approx. 44 C for a direct cable
burial in the mother ground (Fig. 13b and d). Meanwhile, the cor-
responding Tc2,max temperatures are equal to approx. 41 C and
42.5 C. For the cable line burial depth of H = 2.5 m the Tc1,max tem-
perature is equal to approx. 50.5 C for the cable placement in
mother ground (Fig. 13f and h), while approx. 48.5 C for the cable
line burial in FTB bedding (Fig. 13e and g). Therefore, in the case
when the homogenous conditions of thermal conductivity exists,
the cable line backfilling in FTB does not produce the significant
cable conductor temperature drop. In practice, the ground thermal
conductivity exhibit changes with the temperature and porosity.
This situation may result in a considerable temperature increase
in a proximity of the underground cable line. Therefore, the non-
homogeneous conditions of multilayered soil and FTB bedding lay-
er thermal conductivity (similar to those shown in Fig. 3) should be
considered during the design of underground power cables system.
In this case the cable backfilling in FTB bedding instead of mother
ground improves the heat dissipation from the cables to their sur-
rounding, especially when the high cable conductor temperatures
are expected.
Figs. 12 and 13a–h reveal a good agreement between the results
obtained from the developed FEM code and ANSYS program. The
values of temperatures calculated for the corresponding x and y
coordinates remain practically the same.
4.3. Error analysis
The error analysis is performed to validate the temperature
fields obtained using the FEM code and shown in Fig. 11 with the
results received using ANSYS program. The absolute value of rela-
tive error:
when the cables are placed in mother ground); 

 T i;FEMcode  T i;ANSYS 
– at the evaluation line 3 the lowest temperature in the region of 2¼    100%;
cables bedding layer occurs at the point P. The point is located
in the middle of the horizontal distance (xP = 0.2 m) between
the cables (Fig. 12c and d);
– the temperatures evaluated at the end of evaluation lines i.e.
y = 10 m (line 1 and 2; Fig. 12a and b) and x = 10 m (line 3;
Fig. 12c and d) are approaching asymptotically the constant val-
ues of 22.5 C (lines 1 and 2) and of 20.5 C (line 3).
For a cable burial depth of H = 2.5 m (Fig. 12e–h) it can be seen
that:
– the central cable conductor temperature is higher than in the
case when H = 1 m. The cable core temperature is about 7 C
higher in the case when the FTB bedding layer is applied (com-
pare Fig. 12a with Fig. 12e). When the cables are situated in the
mother ground, the cable core temperature increase is 8.5 C
(compare Fig. 12b with Fig. 12f);
– the temperature differences DTc,max are higher than for H = 1 m
(compare Fig. 12a and b with Fig. 12e and Fig. 12f);
– difference between the Tc1,max and the TP (xP = 0.2 m, line 3) tem-
peratures remain practically the same as for burial depth of
H = 1 m (compare Fig. 12c and d with 12 g and 12 h).
 for i ¼ 1; . . . ; N ð23Þ
T i;ANSYS
is calculated for each i-th node in the FEM model. Therefore, the
nodal temperatures obtained from the code Ti,FEMcode, may be direct-
ly compared with ANSYS results Ti,ANSYS. The relative error  distri-
butions are depicted in Fig. 14a–h for the temperature fields
shown in Fig. 11a–h. The temperature fields are obtained for non-
homogeneous, i.e. k(T) – const. (Fig. 11a–b and e–f), and homoge-
neous, i.e. k(T) = const. (Fig. 11c–d and g–h) conditions. It can be
seen in Fig. 14a–b and e–f that the relative error is higher when
the temperature-dependent thermal conductivity of the multilay-
ered soil and FTB bedding layer is considered (i.e. k(T) – const.).
However, values of  do not exceed 0.008%.
When the constant thermal conductivities of a multilayered soil
and FTB bedding layer are considered, i.e. k(T) = const. (see
Fig. 14c–d and g–h), then  is less than 0.0001%. The largest values
of  occur in the region of cable bedding. Slight differences
between the results obtained by using the ANSYS code and the
developed FEM code may be explained as follows. When the vari-
able thermal properties of a multilayered soil are considered in
computations then, the heat conductivity function (7) is incorpo-
rated directly in an FEM code to determine the corresponding ther-
mal conductivities kðT e Þ within the finite elements. However, the
364 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370

Fig. 11. Temperature distributions in the underground power cable system obtained for burial depth of H = 1 m (a–d) and H = 2.5 m (e–h) for the computational cases shown
in Fig. 7.

ANSYS program used the tabulated kðT e Þ values within the range of
temperature changes from 20 C to 90 C. The temperature incre-
ment between the tabulated thermal conductivity data is set to
1 C in this study. Then, the thermal conductivities are linearly
interpolated by the ANSYS solver.
The different procedure of thermal conductivity interpolation
might result in the maximum relative error  of 0.008% in a prox-
imity of the cable line location. In this region, the temperature gra-
dients are the highest and thermal conductivity exhibits
considerable variations. The slight differences in predicted nodal
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370 365

Fig. 12. Temperature variations along the evaluation lines (1–3) obtained for computational cases shown in Fig. 7a and b for cable line burial depths of H = 1 m (a–d) and
H = 2.5 m (e–h).
366 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370

Fig. 13. Temperature variations along the evaluation lines (1–3) obtained for computational cases shown in Fig. 7c and d for cable line burial depths of H = 1 m (a–d) and
H = 2.5 m (e–h).
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370 367

Fig. 14. Absolute value of the relative  error distributions in the underground power cable system obtained for burial depth of H = 1 m (a–d) and H = 2.5 m (e–h) for the
temperature fields shown in Fig. 11a–h.
368 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370

temperatures may also result from the adopted computational The proposed computational approach, which considers the
algorithm. The ANSYS solver employs the full Newton–Raphson temperature dependent thermal conductivity of a multilayered
(N–R) method and solves Eq. (A16) at every iteration while the soil, is planned to be validated experimentally in the future.
FEM code used the Jacobi method, given by Eq. (A20) (see Appen-
dix A). The Jacobi method is simpler in implementation within Acknowledgement
numerical code and exhibits a better numerical stability than N–
R method. In the supplementary materials section the authors pro- Authors would like to acknowledge the ENERGOPROJEKT-
vide the FEM code and the input file for ANSYS software used in KRAKÓW S.A. for a creative support and advices helpful in writing
this study. Therefore, it is possible for the readers to analyze the this paper.
developed code and compare the obtained results with ANSYS.
Appendix A
5. Conclusions
The local FEM formulation of the heat conduction Eq. (10) relat-
This paper presents the thermal analysis of underground 400 kV ed to an individual finite element results in [47,57,58]:
high-voltage power cable system. An in-line arrangement of three  e e  e 
underground power cables is considered. The burial depths vary K ðT Þ 33 fT e g31 ¼ f ðT e Þ 31 : ðA1Þ
from 1 m to 2.5 m. The computations are performed using the The triangular finite elements with linear shape functions are used
FEM code developed in MATLAB and then validated by comparison in this study to determine the temperature distribution in the ana-
with ANSYS simulations. The two-dimensional heat conduction lyzed power cables system. Pairing the nodal coordinates of finite
equation is solved to determine the temperature distribution in element nodes i = 1, . . . , 3 as:
cables and the surrounding multilayered ground. Two different
types of cable laying technique are studied: fqi g12 ¼ fxi ; yi g; for i ¼ 1; . . . ; 3; ðA2Þ
allows to define the following vectors:
1. Multilayered soil is considered; cables are placed in a mother
ground. fug12 ¼ fq2 g12  fq3 g12 ;
2. Multilayered soil is considered; cables are placed in FTB bed- fv g12 ¼ fq3 g12  fq1 g12 ; ðA3Þ
ding layer. fwg12 ¼ fq1 g12  fq2 g12 :

Thermal properties of a multilayered soil are determined based Subsequently, the stiffness matrix of triangular finite element
 e e 
on the parameters obtained from the geological measurements, K ðT Þ can be calculated as [56]:
such as water content, porosity, density and the type of layer. 0 1
fugfug fugfv g fugfwg
The geological measurements were performed at the location   kðT e
Þ B C
where the power transmission line is planned to be installed. In K e ðT e Þ 33 ¼ @ fv gfug fv gfv g fv gfwg A: ðA4Þ
4Ae
order to include the effect of dry zones formation, the temperature fwgfug fwgfv g fwgfwg
dependent thermal conductivity is considered at first. Subsequent-
Where e.g. {u}{v} denotes a dot product of vectors {u} and {v}; Ae is
ly, the computations are repeated with the assumption that the
the surface area of triangular element e:
thermal conductivity of soil and FTB layers is temperature inde-
pendent. For analyzed underground power cable system, the per- 1
Ae ¼ ½ðx2 y3  x3 y2 Þ þ ðy2  y3 Þx1 þ ðx3  x2 Þy1 : ðA5Þ
formed case studies produced the following conclusions: 2
The mean thermal conductivity across the finite element kðT e Þ is
1. maximum temperature of the cable conductor increases with
calculated using the average element temperature T e given by:
the increase in burial depth;
2. the highest temperature values are obtained for the centrally T e1 þ T e2 þ T e3
Te ¼ : ðA6Þ
located cable; 3
3. the underground power cables placement in FTB bedding layer
where the nodal temperatures T e1 , T e2 and T e3 are the components of
improves the heat dissipation from cable to surroundings and
nodal temperatures vector {Te}:
decreases the temperature of the cable core; 8 e9
4. applied temperature dependent thermal conductivity formula < T1 >
> =
for the soil and FTB bedding layers resulted in lower cable core fT e g31 ¼ T e2 : ðA7Þ
temperatures (in comparison to homogenous conditions) when >
: e> ;
T3
FTB bedding is applied. Higher cable core temperatures are
reached when the mother ground is used as a cable bedding. Assuming the uniform heat generation rate within the finite
The difference between the obtained values of cable core tem- element one can obtain the element load vector {f e} [57]:
perature (for temperature dependent and independent thermal 8 9
e> 1>
conductivity approaches and application of FTB bedding) e e  q ðT ÞA < =
e

increases with a burial depth; f ðT Þ 31 ¼ v 1 : ðA8Þ

3 >
: > ;
5. use of the FTB backfill layer is the most favorable when sig- 1
nificant changes of thermal conductivity are expected in the
It is assumed that the total number of nodes and finite elements in
vicinity of the cable.
the FEM model is N and E, respectively. The element stiffness matri-
ces [Ke] and element load vectors {f e} determined for e = 1, . . . , E are
The thermal analysis approach, proposed in this study, is simple
assembled into the global stiffness matrix [K] and load vector { f }
in the numerical implementation and requires only the basic geo-
defined as [56,57]:
logical measurements performed in a location where the electricity
transmission line is installed. The composition and the water con- X
E

tent of each soil layer need to be specified in order to the proposed ½KNN ¼ ½K e 33 ; ðA9Þ
e¼1
model might be applied.
 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370 369

and until the convergence criterion:


X
E
ff gN1 ¼
e
ff g31 : ðA10Þ jdj ¼ max jfDTgiN1 j < 0:0001 K; ðA18Þ
e¼1
is satisfied. During the initial iteration (i = 0), the constant value of
Then the system of nonlinear equations:
volumetric heat source qv = 10,000 W/m3 and homogenous thermal
½KðfTgÞNN fTgN1 ¼ ff ðfTgÞgN1 ; ðA11Þ conductivities of soil layers and FTB bedding (Table 5), are assumed.
Therefore, a nonlinear system (A11) simplifies to a linear system of
must be solved to determine the nodal temperatures vector equations, and the starting values vector {T}0 can be found directly
{T}N1 = {T1, T2, . . . , TN1, TN}T. The nonlinearity is a result of tem- by applying the appropriate numerical method (e.g. LU factoriza-
perature-dependent thermal conductivity kðT e Þ of a multilayered tion). To avoid the numerical calculation of derivatives in Eq.
soil and FTB bedding material (Eq. (7)). Moreover, the temperature (A15) the following simplification are made. The average element
dependent volumetric heat source qv ðT e Þ, given by Eq. (11) also temperatures T e are calculated for each finite element e = 1, . . . , E.
introduces nonlinearity to the system (A11). In Eq. (A11), fTg Hence, the element stiffness matrices [Ke] and load vectors {f e}
denotes a vector of average element temperatures calculated for are determined directly by applying Eqs. (7) and (11), respectively
e = 1, . . . , E using Eq. (A6): for kðT e Þ and qv ðT e Þ. The mean temperatures T e are calculated based
8 9 on the nodal temperatures from the previous iteration {T}i1. Taking
>
>
> T1 > >
>
>
> > into account the abovementioned simplifications the Jacobian
>
>
> T2 > >
>
> matrix [J({T}i1)] is simplified directly to the stiffness matrix
>
< : > = i1
fTgE1 ¼ : ðA12Þ ½KðfTg Þ and {U({T}i1)} to:
>
> : >>
>
> >
>
>
> T E1 >
> i1 i1 i1 i1
>
>
>
>
> fUðfTgi1
N1 ÞgN1 ¼ ½KðfTg
i1
ÞNN fTgN1  ff ðfTg ÞgN1 ðA19Þ
: E > ;
T
Therefore, the Jacobi iterations:
The Newton–Raphson method can be applied for solving the non-
i1 i1
linear system (A11). To achieve this goal the functions /i({T}), cor- ½KðfTg ÞNN fDTgiN1 ¼ fUðfTgN1
i1
ÞgN1 ðA20Þ
responding for i = 1, . . . , N equations, are created.
8 are performed to determine the {DT}i and {T}i vectors from
>
> /1 ðfTgÞ ¼ fK 1 ðfTgÞgfTg  f 1 ðfTgÞ ¼ 0 Eqs. (A20) and (A17), respectively.
>
>
>
> /2 ðfTgÞ ¼ fK 2 ðfTgÞgfTg  f 2 ðfTgÞ ¼ 0
>
>
>
< / ðfTgÞ ¼ fK ðfTgÞgfTg  f ðfTgÞ ¼ 0
3 3 3 Appendix B. Supplementary material
; ðA13Þ
>
> :
>
>
>
> Supplementary data associated with this article can be found, in
>
> :
>
: the online version, at http://dx.doi.org/10.1016/j.enconman.2015.
/N ðfTgÞ ¼ fK N ðfTgÞgfTg  f N ðfTgÞ ¼ 0
01.092.
where {Ki} denotes, i-th row of [K] matrix and fi is the right side vec-
tor of the i-th equation in a nonlinear system (A11). Denoting References
{U({T})} as:
8 9 [1] Lobão JA, Devezas T, Catalão JPS. Reduction of greenhouse gas emissions
> /1 ðfTgÞ > resulting from decreased losses in the conductors of an electrical installation.
>
> >
>
>
> /2 ðfTgÞ >
> Energy Convers Manage 2014;87:787–95.
>
> >
> [2] Tian Q, Lin X. Winding capacitance dividing scheme for a high-voltage cable-
>
< >
=
: wound generator. Energy Convers Manage 2010;51:428–33.
fUðfTgÞgN1 ¼ ; ðA14Þ [3] IEC 60287-2-1: Electric Cables-Calculation of the current rating – Part 2:
>
> : >
>
>
> >
> Thermal resistance – Section 1: calculation of the thermal resistance; 1995.
>
> / ðfTgÞ > > [4] IEEE 835-1994: IEEE Standard Power Cable Ampacity Tables; 1994.
>
> >
>
: N1 ; [5] Kroener E, Vallati A, Bittelli M. Numerical simulation of coupled heat, liquid
/N ðfTgÞ water and water vapor in soils for heat dissipation of underground electrical
power cables. App Therm Eng 2014;70:510–23.
the Jacobian [J] matrix as: [6] Peters-Lidard CD, Blackburn E, Liang X, Wood EF. The effect of soil thermal
2 @/1 @/1 @/1 @/1
3 conductivity parameterization on surface energy fluxes and temperatures. J
@T @T 2
:: :: @T N1 @T N Atmos Sci 1998;55:1209–24.
6 @/1 7 [7] Rees SW, Adjali MH, Zhou Z, Davies M, Thomas HR. Ground heat transfer
6 2 @/2
:: :: @/2 @/2 7
6 @T 1 @T 2 @T N1 @T N 7 effects on the thermal performance of earth-contact structures. Renew Sust
6 7 Energy Rev 2000;4:213–65.
6 : : : : : : 7
½JNN ¼6
6 :
7; ðA15Þ [8] Gouda OE, El Dein AZ, Amer MG. Effect of the formation of the dry zone around
6 : : : : : 7 7 underground power cables on their rating. IEEE Trans Power Deliver
6 @/N1 @/N1 @/N1 @/N1 7 2011;26:972–8.
6 @T :: :: 7
4 1 @T 2 @T N1 @T N 5 [9] Chengyun X, Zhonghao R, Xinyu Y, Zhengchang S, Dongtai H. Numerical
@/N @/N @/N @/N investigation of vapor–liquid heat and mass transfer in porous media. Energy
@T 1 @T 2
:: :: @T N1 @T N Convers Manage 2014;78:1–7.
[10] Costa SC, Barreno I, Tutar M, Esnaola JS, Barrutia H. The thermal non-
and fDTgiN1 ¼ fTgiN1  fTgi1
N1 as the difference between the nodal equilibrium porous media modelling for CFD study of woven wire matrix of a
temperatures in two successive iterations, the i-th iteration of New- Stirling regenerator. Energy Convers Manage 2015;89:473–83.
[11] Dehghan M, Milad Tajik JA, Saman R. Analytical interpretation of the local
ton–Raphson algorithm can be performed:
thermal non-equilibrium condition of porous media imbedded in tube heat
h i n o exchangers. Energy Convers Manage 2014;85:264–71.
i1
JðfTgN1 Þ fDTgiN1 ¼  UðfTgi1
N1 Þ : ðA16Þ [12] Dipankar B, Balaram K, Abdul A. Enhancement of heat transfer from a
NN N1
continuously moving porous fin exposed in convective–radiative environment.
The nodal temperatures are updated in the following manner: Energy Convers Manage 2014;88:842–53.
[13] Hatami M, Mehdizadeh Ahangar GHR, Ganji DD, Boubaker K. Refrigeration
fTgiN1 ¼ fDTgiN1 þ fTgN1
i1
; ðA17Þ efficiency analysis for fully wet semi-spherical porous fins. Energy Convers
Manage 2014;84:533–40.
370 P. Ocłoń et al. / Energy Conversion and Management 95 (2015) 352–370
[14] Moradi A, Hayat T, Alsaedi A. Convection-radiation thermal analysis of
triangular porous fins with temperature-dependent thermal conductivity by
DTM. Energy Convers Manage 2014;77:70–7.
[15] Hamdan M, Al-Nimr MA. The use of porous fins for heat transfer augmentation
in parallel-plate channels. Transport Porous Med 2010;84:409–20.
[16] Alkam M, Al-Nimr MA, Hamdan MO. Enhancing heat transfer in parallel-plate
channels by using porous inserts. Int J Heat Mass Transf 2001;44:931–8.
[17] Mohamad AA. Heat transfer enhancements in heat exchangers fitted with
porous media Part I: constant wall temperature. Int J Therm Sci
2003;42:385–95.
[18] Bogdan IP, Abdulmajeed AM. An experimental and numerical study on heat
transfer enhancement for gas heat exchangers fitted with porous media. Int J
Heat Mass Tran 2004;47:4939–52.
[19] Merrikh AA, Mohamad AA. Non-Darcy effects in buoyancy driven flows in an
enclosure filled with vertically layered porous media. Int J Heat Mass Tran
2002;45:4305–13.
[20] Ucar E, Mobedi M, Ozerdem B, Pop I. A comment on change of Nusselt number
sign in a channel flow filled by a fluid-saturated porous medium with constant
heat flux boundary conditions. Transport Porous Med 2013;96:97–103.
[21] Çekmer Ö, Ozerdem B, Mobedi M, Pop I. Fully developed forced convection
heat transfer in a porous channel with asymmetric heat flux boundary
conditions. Transport Porous Med 2011;90:791–806.
[22] Abdul Mujeebu M, Abdullah MZ, Mohamad AA. Development of energy
efficient porous medium burners on surface and submerged combustion
modes. Energy 2011;36:5132–9.
[23] Hwang CC, Jiang YH. Extensions to the finite element method for thermal
analysis of underground cable systems. Electr Pow Syst Res 2003;64:159–64.
[24] Al-Saud MS, El-Kady MA, Findlay RD. A new approach to underground cable
performance assessment. Electr Pow Syst Res 2008;78:907–18.
[25] De Lieto VR, Fontana L, Vallati A. Thermal analysis of underground electrical
power cables buried in non-homogeneous soils. Appl Therm Eng
2011;31:772–8.
[26] De Lieto VR, Fontana L, Vallati A. Experimental study of the thermal field
deriving from an underground electrical power cable buried in non-
homogeneous soils. Appl Therm Eng 2014;62:390–7.
[27] Papagiannopoulos I, Chatziathanasiou V, Exizidis L, Andreou GT, De Mey G,
Wie˛cek B. Behaviour of the thermal impedance of buried power cables. Electr
Pow Energy System 2013;44:383–7.
[28] Chatziathanasiou V, Chatzipanagiotou P, Papagiannopoulos I, De Mey G,
Wie˛cek B. Dynamic thermal analysis of underground medium power cables
using thermal impedance, time constant distribution and structure function.
Appl Therm Eng 2013;60:256–60.
[29] Wie˛cek B, De Mey G, Chatziathanasiou V, Papagiannakis A, Theodosoglou I.
Harmonic analysis of dynamic thermal problems in high voltage overhead
transmission lines and buried cables. Electr Pow Energy System
2014;58:199–205.
[30] Taler J, Ocłoń P. Finite element method in steady state and transient heat
conduction. In: Hetnarski RB, editor. Encyclopedia of thermal stresses, vol.
4. Dordrecht, Heidelberg, New York, London: Springer; 2014. p. 1604–33.
[31] Saleeby KE, Black WZ, Hartley JG. Effective thermal resistivity for power cables
buried in thermal backfill. IEEE Trans Power Ap Syst 1979;98:2201–14.
[32] Hawaiian Electric Co. INC.: Specification No.CS7001-17 for Construction of
underground facilities; 2006.
[33] Radhakrishna HS. Fluidized cable thermal backfill. In: Boggs SA et al., editors.
Underground cable thermal backfill. Pergamon Press; 1982. p. 34–53.
[34] TELE-FONIKA Company. High voltage cables. 1st ed. Cracow; 2012 (in polish).
[35] Campbell GS, Jungbauer JR JD, Bidlake WR, Hungerford RD. Predicting the
effect of temperature on soil thermal conductivity. Soil Sci 1994;5:307–13.
[36] Usowicz B, Lipiec J, Usowicz JB, Marczewski W. Effects of aggregate size on soil
thermal conductivity: Comparison of measured and model-predicted data. Int
J Heat Mass Tran 2013;2:536–41.
[37] Bittelli M, Ventura F, Campbell GS, Snyder RL, Gallegati F, Pisa PR. Coupling of
heat, water vapor, and liquid water fluxes to compute evaporation in bare
soils. J Hydrol 2008;3–4:191–205.
[38] Gori F, Corasaniti S. New model to evaluate the effective thermal conductivity
of three-phase soils. Int Commun Heat Mass 2013;47:1–6.
[39] MathWorks Inc., Matlab R2011b; 2011.
[40] Popiel CO, Wojtkowiak J, Biernacka B. Measurements of temperature
distribution in ground. Exp Therm Fluid Sci 2001;25:301–9.
[41] Metwally IA, Al-Badi AH, Al Farsi AS. Factors influencing ampacity and
temperature of underground power cables. Electr Eng 2013;95:383–92.
[42] IEC 287, Calculation of the continuous current rating of cables (100%) load
factor, IEC Publication 287; 1982.
[43] Wang T. Thermal performance of 69 kV underground cables. A Thesis
Presented in partial Fulfillment of the Requirements for the Degree Master
of Science, Arizona State University; 2013.
[44] Johansen O. Thermal conductivity of soils, Ph.D. thesis, Trondheim, Norway;
1975.
[45] ISO 14688-2, Geotechnical investigation and testing – Identification and
classification of soil – Part 2: Principles for a classification.
[46] Cisek P, Ocłoń P, Pilarczyk M. Thermal analysis of operating conditions for The
400 kV underground power cable transmission line as a power plant delivery
system. Rynek Energii 2014;114:70–7 [in polish].
[47] Ocłoń P, Taler D, Cisek P, Pilarczyk M. FEM based thermal analysis of
underground power cables situated in backfills made of different materials.
Mater Res Innov; 2015 (in press).
[48] Farouki OT. Thermal properties of soils. Series Rock Soil Mech 1986;11:1–136.
[49] Kersten MS. Thermal properties of soil. Bull Univ Minn Inst Technol
1949;52:1–225.
[50] De Vries DA, Afgan NH. Heat and mass transfer in the biosphere. John Wiley
and Sons; 1975.
[51] De Vries DA. Thermal properties of soils. In: Van Wijk WR, editor. Physics of
plant environment. Amsterdam: North-Holland Publishing Company; 1963. p.
210–35.
[52] Sepaskhah AR, Boersma L. Thermal conductivity of soils as a function of
temperature and water content. Soil Sci Soc Am J 1979;43:439–44.
[53] Kasubuchi T. Heat conduction model of saturated soil and estimation of
thermal conductivity of soil solid phase. Soil Sci 1984;138:240–7.
[54] ANSYS, ANSYS documentation, Manual Set and Theory Reference Manual,
ANSYS Inc.; 2009.
[55] Tarnawski VR, Mosome T, Leong WH, Bovesecchi G, Coppa P. Thermal
conductivity of standard sands. Part I. Dry-state conditions. Int J
Thermophys 2009;30:949–68.
[56] Cuvelier F, Japhet C, Sarella G. An efficient way to perform the assembly of
finite element matrices in Matlab and Octave. Technical Report. 8305,
Universit́e Paris 13; 2013.
[57] Rao SS. The finite element method in engineering. 4th
ed. Burlington: Butterworth-Heinemann; 2005.
[58] Taler J, Duda P. Solving direct and inverse heat conduction
problems. Berlin: Springer; 2006.