Professional Documents
Culture Documents
Cable system, capacitive coupling, complex impedance Results in all cases are in very good agreement, validating
matrix (CIM), electromagnetic transient program (EMTP), the applicability of the abovementioned models. It is con-
finite element method (FEM), inductive coupling. cluded that CIM is the most straightforward approach, lim-
ited though for cases of inductive coupling. On the other
hand, the EMTP-like software can be used taking into ac-
INTRODUCTION count both capacitive and inductive coupling for most typi-
In the process of designing cable systems, several issues cal cases. In more complicated geometries and configura-
are related to the electromagnetic interference (EMI) and tions, FEM can be employed solving the generic electro-
especially to the capacitive and inductive coupling between magnetic formulation.
conductors and sheaths. EMI is associated with the current
rating calculation, where the power losses on cable SYSTEM UNDER STUDY
sheaths during normal operation must be taken into ac-
count [1]. EMI problem is also related to the calculation of An underground cable of nominal voltage 87/150 kV and
overvoltages on cable sheaths, which should be kept under cross-section 1000 mm2 is examined. The cable follows the
acceptable limits satisfying health and safety standards [2]. design of Fig. 1, where the various layers are indicated.
Their properties, including the surrounding media, are
The calculation of voltages and currents in the abovemen- given in Tables A.I-A.III in the Appendix. The whole model
tioned scenarios can be performed using a variety of meth- is assumed to be infinitely long in the cable axis direction,
ods, mostly depending on the tradeoff between the accu- neglecting any end effects so as to render the 2D analysis
racy and computation burden. The three most common nu- sufficient. Considering sheath bonding types, the most
merical approaches are: the Complex Impedance Matrix commonly used in transmission systems are examined.
(CIM) method, a numeric solution based on the self and These include: solid-bonding, single-point-bonding, and
mutual impedances of the cable system taking into account sectionalized cross-bonding.
the inductive coupling; the ElectroMagnetic Transient Pro-
gram (EMTP-like) software, using distributed impedances
Air
and admittances for the modeling of the cable system and
relevant surrounding, which also takes into account the ca- Soil
Conductor
Inner semi-
pacitive coupling of the cable system; and the finite element conductive
a variety of approaches. In this paper, three different meth- The bonding type is taken into account by an external elec-
ods are examined, namely FEM, EMTP-like software and trical circuit that models currents and voltages in circuits [5].
CIM method. This interface has connections to the distributed field model
of (4)-(7) and is solved simultaneously based on Kirchhoff’s
FEM conservation laws for the voltages, currents and charges
The computational scheme of FEM is based on Maxwell’s associated with the circuit elements.
equations, explicitly formulated by the electromagnetic
Coordinate
problem that is to be solved. Considering the capacitive scaling
coupling, FEM computes electric field, current, and poten- ñ×Ā=0
tial distributions in conducting media where inductive ef- Region of interest
Air
fects are assumed negligible [5]. Specifically, the fre-
Soil
quency-domain steady-state problem can be described
with the following equations:
r→∞
∇ ⋅ J = − j ωρ (1)
J =σ E + j ω D + Je (2)
E = −∇V (3) Fig. 2: Boundary conditions of inductive coupling in
2D FEM.
where J the current density, ρ the space charge density,
σ the material conductivity, E the electric field, D the elec-
EMTP-like software
EMTP-like software is a category of programs capable of
tric displacement field, ω the angular frequency, Je the ex-
analysing electromagnetic transients in power systems us-
ternally imposed current density and V the voltage potential ing a variety of sophisticated models both in time- and fre-
which is also the dependent variable to be solved. quency-domain. Focusing on cables under the steady-state
The boundary conditions include the electric potential at conditions, telegrapher’s equations in frequency-domain
each conductor and sheath, taking into account the can be employed:
adopted bonding type of each case. Since the electric field dI
is confined between conductor and sheath at power fre- − = YV = (G + j ωC ) V (8)
quency, it is not necessary to model the remaining layers dz
and the surrounding media, thereby simplifying the prob-
dV
lem. − =ZI =( R + j ω L ) I (9)
dz
For the inductive coupling, FEM computes magnetic field
and induced current distributions in and around conduc- where I, V the N × 1 current and voltage vectors, while Y
tors [5]. The corresponding equations in frequency-domain and Z the per-unit-length (pul) N × N shunt admittance and
under steady-state conditions are formulated below: series impedance matrices, respectively, which consist of
conductance G, capacitance C, resistance R, and induct-
∇×H =J (4) ance L matrices.
B = ∇× A (5) Eqs. (8) and (9) can be modelled by the N-element homog-
enous transmission line with the assumption of plane-wave
J =σ E + j ω D + Je (6) propagation along z direction. In this paper, the lumped PI
equivalent model is employed, since it has been proven to
E = − jω A (7) be very accurate and stable under steady-state power fre-
quency conditions [3].
where H the magnetic field intensity, B the magnetic field, Since (8) and (9) are in principle mutually coupled, attention
and A the magnetic vector potential which is also the de- must be paid to proper modelling of capacitive and induc-
pendent unknown variable to be determined. It is noted that tive coupling. For the former, ideal voltage sources are em-
the second term in (6) is negligible at power frequency. ployed with the resulting charging current flowing through
the shunt admittance branch Y of the PI equivalent in (8).
The imposition of boundary conditions in inductive coupling
For the latter, ideal current sources are used flowing
is somewhat more challenging compared to the capacitive
through the series impedance branch Z of (9). In both
coupling, since the magnetic field can be considered un-
cases, bonding types are implemented by proper circuit el-
bounded. In order to limit the extent of the FEM model to a
ements and connections, while the use of multiple cas-
manageable region of interest with reasonable execution
caded PI equivalents allows for the derivation of current
time, a coordinate scaling is adopted to layers of virtual do-
and voltage profiles with respect to cable length.
mains surrounding the physical region of interest, i.e., to air
and soil as indicatively shown in Fig. 2 [5]. These virtual CIM method
layers can be mathematically stretched out towards infinity,
CIM method calculates currents and voltages while ne-
where n × A = 0 is imposed with n being the normal unit glecting the influence of capacitive currents and assuming
vector. As a result, the model is computationally efficient, that the earth return path is represented by an equivalent
while the solution inside the region of interest is not affected conductor. The following set of equations represents the
by the artificial geometric boundaries. cable system:
∆V =
ZI (10) to be solved N times, with one conductive layer having non-
zero potential each time. From the N solutions and by su-
− ( RG,1 + RG,2 )( IS, A + IS,B + IS,C )
∆VG = (11) perimposing every two solutions, the whole matrix Y can be
found.
where ∆V the N × 1 vector of voltage drop along all con- In EMTP-like software, the admittance matrix of the exam-
ductive layers, Z part of the full pul series impedance ma- ined cable is evaluated from the internal potential coeffi-
trix Z, the cable length, RG,1 , RG,2 the grounding re- cient matrix, ignoring any leakage current, i.e. G = 0 [7]:
sistances at both cable ends, and IS , A , IS ,B , IS ,C the cur- −1
Pi 1 0 0
rents at cable sheaths. = ω Pi −1 j ω 0 Pi 2 0
Y j= (15)
The number of unknowns in (10) are twice the number of 0 0 Pi 3
conductive layers. In order to get a unique solution (11) is
applied, which imposes the cable sheaths as the current Each diagonal submatrix is given by the following form:
return path, thus equalizing the longitudinal voltage drop on
pcj + psj psj
the sheaths to the voltage drop ∆VG on the grounding re- Pij = (16)
sistances. For solid-bonding type, a single section is con- psj psj
sidered, described by its Z . For single-point-bonding type, where:
(10) and (11) are solved by imposing the three sheath cur-
rents equal to zero. Finally, for cross-bonding type, the total ln ( rsemi − out rc ) ln ( rover − s rs )
impedance matrix is calculated as the vectorial sum of the pcj = and psj = (17)
2πεins 2πε over − s
partial impedance matrices of the three minor sections.
In (17), rc, rsemi-out, rs, rover-s, are the radii of conductor, outer
PUL PARAMETERS semi-conductive layer, sheath and oversheath, respec-
tively. εins is the compensated permittivity of the insulation
A first comparison between the considered approaches can
be made by calculating the pul parameters, which are fur- triplex [8], while ε over − s is the permittivity of the oversheath.
ther employed by the EMTP-like software and the CIM
method for the calculation of sheath currents and voltages. Due to the symmetry in matrix Y, only the elements Y11, Y14
Although FEM does not rely on these parameters for the and Y44 are examined at 50 Hz, referring to one of the three
problem formulation, it is possible to calculate them for power cables. Results obtained by the different models are
comparison and validation purposes. compared in Fig. 3, where the relative difference is also
shown. Results present very small deviations with the max-
Shunt admittance matrix imum relative error being less than 0.02%.
Matrix Y is calculated only by FEM and EMTP-like soft- 5
10
-7
0.02
ware, since CIM method does not model the shunt admit- EMTP
FEM 0.018
Error
method and energy method [6]. The latter approach, which 0.012
Relative difference [%]
0.004
1 1 0
(12)
2 S∫∫ ∫∫ ε∇V ∇VdS
=
WE =
DEdS 0.002
cab
2 Scab -1 0
Y Y Y ΔY ΔY ΔY
11 14 44 11 14 44
FEM
trix Z is only part of the full Z in (19). A good agreement is
50
Current magnitude [A]
0.5
error being less than 2% and 2.2% for EMTP-like software -50
1.5 FEM
Voltage angle [deg]
90
CIM
0.01
1
89.9
0.005
0.5
0 89.8
0 0 75 150 225 300 375 450 0 75 150 225 300 375 450
R R R R R R Length [m] Length [m]
11 12 41 42 44 45
Single-point-bonding type
4
difference being 2.76%. It is shown that the charging cur- The same cable lengths as in the capacitive coupling are
rent accumulates inside the sheath at the bonded end. At considered. A current of 1000 A (rms) is assumed for the
the other end, the current is zero and the sheath voltage energized cable, while a value of 5 Ω is used again for
reaches a maximum. grounding resistances.
Solid-bonding type
(a) (b)
3
EMTP
-90
FEM
The sheath current and voltage profiles of first (left) cable
Current magnitude [A]
90.2
Voltage angle [deg]
718 FEM
with the maximum relative error being 3.41%. The charging Length [m]
(c)
Length [m]
(d)
current for the three minor sections shows a 120o phase 2
100
1.4 100
Current angle [deg]
1.2 0
0.06
(c)
100
(d)
Single-point-bonding type
Fig. 9 presents the results for the case of single-point-bond-
50
Voltage magnitude [V]
0.04
0.02
maximum relative difference being 0.58% and 0.60% for
-50
-100
0
0 225 450 675 900 1125 1350 0 225 450 675 900 1125 1350
EMTP-like software and CIM method compared to FEM.
Length [m] Length [m] Single-point-bonding is considered less complicated for
managing losses in current rating purpose. It provides im-
Fig. 7: Sheath current and voltage profiles for capaci- proved cable current capacity by eliminating the circulating
tive coupling and cross-bonding type. current losses in the sheaths. The induced voltage on a
sheath is proportional to the cable length, thus limiting the
INDUCTIVE COUPLING applicable cable length due to the safety limits.
The inductive coupling in FEM is formulated in a 2D plane 70
EMTP
(a)
-104.9
(b)
40
Voltage angle [deg]
-105.1
For the EMTP-like software, ideal current source is con- 20
equivalents is constructed to derive the profiles along cable Length [m] Length [m]
has the advantage of not limiting the length of the cable British Columbia, Vancouver, Canada, 1990.
system. The induced voltages in sheaths present offset in [7] A. Ametani, T. Ohno, N. Nagaoka, Cable system
time by one-third of the period and are added up, thus re- transients. Singapore: IEEE Wiley, 2015.
ducing the sheath circulating current.
[8] B. Gustavsen, J. A. Martinez, D. Durbak, “Parameter
determination for modelling system transients - Part II:
(a) (b)
0.65 146
0.645 Insulated cables,” IEEE Trans. Power Del., vol. 20, no.
Current magnitude [A]
60
[10] Modeling cables in COMSOL Multiphysics®: 6-part tu-
Voltage magnitude [V]
20
0
AB, Stockholm, Sweden, 2017.
0
0 225 450 675 900 1125 1350
-100
0 225 450 675 900 1125 1350
APPENDIX
Length [m] Length [m]
In Table A.I, the outer diameter of each layer is given in
Fig. 10: Sheath current and voltage profiles for induc- mm. In Table A.II the electrical properties are presented,
tive coupling and cross-bonding type. where ρ is the electrical resistivity in Ω·m, εr is the relative
permittivity, and μr is the relative permeability. In Table A.III,
DISCUSSION AND CONCLUSIONS the properties of air and soil are given, respectively.