Evaluation of Electric Field in a DC Model Cable

During Superimposed Impulse Voltage Tests

Anh Hoang1*, Amirhossein Abbasi1, Ajith J. Thomas2, Mikael Unge2, Claire Pitois2
1
NKT HV Cables AB, R&D, SE-371 23 Karlskrona, Sweden
2
NKT HV Cables AB, Technology Consulting, SE-722 26 Västerås, Sweden
*[email protected]

Abstract—Superimposed impulse voltage tests are II. SIMULATION STUDY

performed as a part of the qualification of HVDC cable systems.
The electro-thermal preconditioning prior to the impulse tests A. BCT model
may lead to the accumulation of space charges in cable BCT models with inclusion of ionic charges have been
insulation, which eventually affects the resulting electrical field. developed and discussed in several publications [6]-[9]. In this
Such an impact was evaluated in this work using the bipolar work, we utilized the BCT model that is able to reproduce the
charge transport model with inclusion of ionic charge carriers. experimental evidence of EF enhancement near the interfaces
The results show that the preconditioning gives rise to the between cable insulation and the inner and outer
inhomogeneity of the electrical stress, specifically it leads to an
semiconductors [6]. Some modifications were introduced to
electric field enhancement near inner radial position during the
impulse tests of opposite polarity and near outer radial position
the model as described below.
under impulse tests of same polarity. Contribution of cable The electrical conduction in pure polymeric insulation has
temperature and the generation and transport of ionic charges been formulated relying on the injection of electronic charges
were thereafter studied and discussed. (electrons and holes – e and h) at electrodes and their transport
under EF and trapping in localized states to form trapped
Keywords—DC cable, superimposed impulse voltage tests, charges (etr and htr) [10]. For extruded DC cables, the
electric field, bipolar charge transport model, ionic charge
presence of impurities gives rise to the formation of ionic
I. INTRODUCTION charges. The generation and transport of positive (p) and
negative (n) ions are governed by the continuity equations
The ongoing transition from fossil fuels to renewable
energy sources requires the construction of new transmission +∇∙ − − ∇ = , (1)
lines to integrate, e.g., offshore wind farms and solar farms
into the electrical transmission system. As HVDC cables play +∇∙ − ∇ = . (2)
a key role in such a transition, the development of HVDC
cable systems that work reliably under electrical, thermal, and Here, n and μ are respectively the density and mobility of
mechanical stresses is of utmost importance. As a part of the charge carriers, E is the vector of EF, D is diffusion
qualification of HVDC cable systems, switching impulse and coefficient, and subscripts n and p denote the type of charge
lightning impulse withstand tests are performed. During these carriers. The source terms Sn and Sp on the right-hand side of
tests, a cable system is exposed to multiple switching impulses the equations account for the generation and annihilation of
(SI) or lightning impulses (LI) superimposed on DC stress at ionic charges
nominal voltage and a temperature gradient [1]. The presence
of a temperature gradient in combination with DC stress , = exp − − ! . (3)
facilitates the formation of space charges in polymeric The first term represents the ionization due to dissociation of
insulation of DC cables [2]-[6], and thus, the impact of the impurities (concentration nm) and the second term accounts for
electro-thermal preconditioning prior to the impulse tests must the recombination of positive and negative ionic charges. The
not be overlooked while evaluating electrical stress in cable dissociation rate is temperature dependent according to
insulation. Arrhenius expression with an activation energy WD. Note that
In this study, bipolar charge transport (BCT) model [6] has k is the Boltzmann constant (k = 1.38·10–23 JK–1), T is the
been employed for calculating the electric field (EF) and space absolute temperature, and q is the elementary charge (q =
charge accumulation in DC model cable insulation during the 1.6·10–19 C). The recombination rate Rnp is defined by the
preconditioning. The BCT model includes the generation and Langevin recombination
transport of ionic charges that account for the formation of "# $"%
= , (4)
hetero-charges and EF enhancement [6] and is able to &' &(
elucidate experimental evidence [3]. Electrical stress in cable where ε0 is the vacuum permittivity (ε0 = 8.854 Fm ) and εr is–1

insulation during superimposed switching impulse 250/2500 the relative permittivity of the insulation. The mobility of
μs of same or opposite polarity (SISP or SIOP) and ionic charges is thermally activated
superimposed lightning impulse 1.2/50 μs of opposite polarity #,%
(LIOP) with regard to the applied nominal voltage has been , = ), ) exp − , (5)
then calculated. Impact of the dissociation rate of impurities
with the activation energy Wn,p. The diffusion coefficient of
and ion mobility as well as cable temperature on the EF in
charge carriers is defined by the Einstein relation
cable insulation has been investigated.

979-8-3503-0897-6/24/$31.00 ©2024 IEEE

 , = , . (6) to the reduced transmission power (cable temperature),
* reduced dissociation rate and reduced ion mobility were then
Space charges include contribution from all types of investigated.
electronic and ionic charge carriers:
TABLE I. BCT MODEL PARAMETERS [6], [10], [11]
+ = ! , + ,-. + − / − /-. − . (7)
Electrical field is calculated by using the Gauss’ law Charge Negative Positive
Parameter
carriers carriers carriers
∇ ∙ 01 2 = +, (8) Injection barrier eV e, h 1.16 1.27
= −∇3, (9) Hopping barrier eV e, h 0.71 0.65
Distance between
where V is electrical potential. nm e, h 1 1
traps
Trapping
Model parameters accounting for the electronic s−1 e, h 0.1 0.2
coefficient
conduction were taken from literature, e.g., [10]-[11], whereas Deep trap density C·m–3 e, h 100 100
model parameters for the ionic conduction that are either De-trapping
eV e, h 0.96 0.99
identical with or derived from the validated model parameters barrier height
e-h 0
[6]. All BCT model parameters are provided in Table I. Recombination
m3C−1s−1 e-htr, etr-h,
coefficients 4∙10–3
etr-htr
B. Superimposed impulse voltage tests 2
µn,p0, m V s −1 −1
n, p 2.8·10 –9
2.8·10–8
The study was carried out for a DC model cable of cross- Ion mobility
Wn,p, eV n, p 0.4 0.4
linked polyethylene (XLPE) insulation working at a nominal
SD, s−1 n, p 3.68·10–2
voltage U0 = 80 kV. The cable parameters are according to the Dissociation rate
MV DC cable investigated in [3]. The cable insulation system WD, eV n, p 0.65
is made by triple extrusion that forms cable insulation, inner Impurity –3
nm, mol·m – 23
concentration
and outer semiconducting layers simultaneously. The
insulation thickness is 5 mm and the radial position of the TABLE II. SUPERIMPOSED IMPULSE VOLTAGE TESTS
interfaces between cable insulation and the inner and outer
Type Wave shape Peak-to-ground voltage
semiconductors are respectively 5 mm (ri) and 10 mm (ro).
Prior to the superimposed impulse voltage tests (SIVTs), the SISP a. 250/2500 μs 2.1U0
cable is preconditioned for at least 10 h at the nominal DC SIOP b. 250/2500 μs 1.2U0
voltage and under required conductor temperature and a LIOP c. 1.2/50 μs 2.1U0
temperature gradient. In this study, the preconditioning a.
SISP – switching impulse of same polarity
includes applying negative DC voltage of –80 kV and b.
SIOP – switching impulse of opposite polarity
temperature Ti (at ri) and To (at ro). Parameters of impulse c.
LIOP – lightning impulse of opposite polarity
voltages applied during SIVTs according to CIGRE
recommendation [1] are shown in Table II and the LIOP is
3.1U0
illustrated by the solid curve in Fig. 1, where the peak-to-
ground voltage of 2.1U0 is achieved by superimposing LI of
3.1U0 on DC voltage –U0. Simulation was also carried out 2.1U0
while applying LI of 3.1U0 without DC superimposed voltage,
which is illustrated by the dashed curve in the same figure.
C. Model implementation
The described set of partial differential equations of the
BCT model was solved using the commercial software
COMSOL Multiphysics©. The model was developed in one
dimensional (1D) axisymmetric geometry thanks to the –U0
symmetry of cable geometry. Temperature Ti and To were set
as boundary conditions and the radial distribution of
temperature was calculated using the software’s heat transfer Fig. 1. Lightning impulse superimposed on nominal DC voltage of –80 kV
and lightning impulse without superimposed DC voltage. The DC
module. A given superimposed impulse voltage was applied stress has been preconditioned 10 h prior to zero time.
at the inner radial position ri and ground potential was set at
the outer position ro. TABLE III. STUDY SCENARIOS. VALUES OF DISSOCIATION RATE
AND ION MOBILITY FROM TABLE I.
As it has been noted that the blocking of the ionic fluxes
Scenario Cable temperature Dissociation Ion
as the boundary condition at insulation–semiconductor rate mobility
interfaces can lead to numerical artifact [6], and hence, the
interfacial processes for ionic charges deserve additional 1. Reference Ti = 70°C, To = 50°C SD µn0 & µp0
investigations. In this work, non-blocking ionic fluxes were 2. Reduced Ti = 50°C, To = 30°C SD µn0 & µp0
set at boundaries. temperature

D. Study scenarios 3. Reduced Ti = 70°C, To = 50°C SD/3 µn0 & µp0

dissociation rate
Four scenarios were considered in the study as shown in
Table III. The first scenario is a so-called reference case for 4. Reduced ion Ti = 70°C, To = 50°C SD µn0/3
mobility µp0/3
which insulation temperature was Ti = 70 °C and To = 50 °C
whereas the model parameters for the ionic conduction from
Table I were used. Three additional scenarios corresponding
III. SIMULATION RESULTS EF distribution under impulse voltages without
superimposed DC voltage is depicted in Fig. 4 (note that peak-
A. Reference scenario
to-ground impulse voltages are indicated in the legend). The
Fig. 2 shows the EF distribution in cable insulation at the calculation was done for the same peak-to-ground voltage as
start and the end of the preconditioning in scenario 1. The EF for the SIVTs (see Table II). An additional case for a LI of
has been inverted where the EF enhancement at outer 3.1U0 (i.e., the dashed curve in Fig. 1) was also considered. As
positions can be found at the end of the preconditioning. It is expected, EF is highest at the interface with inner
explained by the accumulation of hetero-charges at positions semiconducting layer under impulse voltage solely. By
close to insulation edges [6]. A strong EF suppression near the comparing the results shown in Fig. 3 and 4, it is noted that
interface with the electrode of higher temperature (i.e., inner the formation of space charges due to the DC stress and the
semiconductor) is elucidated by the formation of homo- temperature gradient during the preconditioning of SIVTs
charges accumulated due to the injection at the electrode alters the location of the maximum EF for SISP vs. SI from
followed by their trapping in localized states [6]. the interface with the inner semiconducting layer (during SI)
Fig. 3 shows the EF distribution in SIVTs for the time to that closer to the outer semiconducting layer (SISP). In case
corresponding the peak value of impulses. The EF distribution impulse voltages of opposite polarity were applied (SIOP and
under DC stress at the end of the preconditioning was the LIOP), the lower EF near the electrode of higher temperature
initial condition for the SIVTs. Thus, the electrical stress (i.e., inner semiconductor) turns out to be an unfavorable
during SIVTs gradually built up on top of the initial EF and initial condition as it enhances the EF during SIVTs. In short,
the maximum EF was achieved at the time instant at which the the preconditioning increases the inhomogeneity of EF
impulse voltages reaching their peaks. distribution. The latter can be featured by the discrepancy
between maximum and minimum EF for a given test.

(a) Scenario 2 – Reduced temperature

Fig. 2. EF distribution in cable insulation at the start and the end of the
preconditioning under DC voltage –U0 and temperature at insulation
edges of 70 °C and 50 °C (scenario 1).
(b) Scenario 3 – Reduced
dissociation rate

Fig. 3. EF distribution in cable insulation in superimposed impulse voltage

tests at time of the peak of the impulse voltages (scenario 1).
(c) Scenario 4 – Reduced ion
mobility

Fig. 5. EF distribution in cable insulation in superimposed impulse voltage

Fig. 4. EF distribution in cable insulation exposed to impulse voltages tests at time of the peak of the impulse voltages (scenarios 2–4).
without superimposed DC voltage.
 attributed to the temperature dependency of electronic charge
injection at electrodes. Under DC stress, a higher cable
conductor temperature facilitates charge injection at the
insulation–inner semiconductor interface that enhances homo-
charge accumulation and suppresses EF near the inner
semiconductor, and therefore, pushes a higher EF outward.
This eventually increases EF near the inner semiconductor in
SIVTs of opposite polarity and EF near the outer
semiconductor in SIVTs of same polarity (Fig. 3 and 5a).
A lower dissociation rate of impurities reduces the
accumulation of space charges that leads to a more
homogeneous EF distribution under DC as shown in Fig. 6.
The impact is however less recognizable in the SIVTs (Fig. 3
Fig. 6. EF distribution in cable insulation at the end of the preconditioning in vs. 5b) due to the large discrepancy in magnitude of the DC
scenarios 1–4 (DC voltage –U0, ΔT = 20 K). and impulse voltage applied (U0 vs. 2.1U0). The impact of the
B. Impact of cable temperature dissociation rate on EF distribution in SIVTs is less noticeable
as compared to the cable temperature contribution. A more
Cable temperature alters due to cable workload. A higher pronounced effect was achieved in case the dissociation rate
transmission power is associated with a higher conductor was reduced by a factor of 10 as compared to that of the
current that results in a larger joule heating, thus increasing reference model (results were not shown here). In a similar
conductor temperature as well as temperature in all cable manner, the reduction of ion mobility by a factor of 3 has a
layers. Preconditioning temperature of a cable in qualification weak impact on the EF distribution in preconditioning (Fig. 6)
test defined by its transmission power can be calculated [12]. as well as in SIVTs (Fig. 5c).
For the sake of simplicity, different preconditioning
temperatures were considered in scenarios 1 and 2 and the Preconditioning DC cables of XLPE insulation using DC
respective EF distributions are illustrated in Fig. 3 and 5a. electrical stress and a temperature gradient has a significant
Under a lower operating temperature, fewer ionic charges are impact on the resulting EF distribution in cable insulation
generated as compared to the reference scenario owing to the during SIVTs, for which EF can be up to 70 kV/mm for the
temperature dependency of the impurity dissociation, see eqn. studied model cable within this work. Successful qualification
(3). Thus, less space charge accumulation results in a more of a cable system, of which SIVT is an integrated part, implies
homogeneous EF distribution under DC stress as well as the high quality of the cable system, including cable
during SIVTs. insulation, cable accessories and their sufficient compatibility.
Further investigations on this topic, which is currently still
C. Impact of dissociation rate and ion mobility limited in literature, should be carried out.
Fig. 5b-c demonstrates the impact of dissociation rate and
mobility of ionic charges under DC stress to the EF REFERENCES
distribution in SIVTs. When the dissociation rate of impurities [1] Working Group WG B1.62 CIGRE, “Recommendations for testing DC
and ion mobility decrease, a similar trend of space charge extruded cable systems for power transmission at a rated voltage up to
accumulation is expected as for lowering cable temperature. and including 800 kV,” 2021.
The impact is anyhow less discernable in the final EF [2] M. Fu et al., “Influence of thermal treatment and residues on space
charge accumulation in XLPE for DC power cable application,” IEEE
distribution during SIVTs. The observation can be elucidated Trans. Dielectr. Electr. Insul., vol. 14, no. 1, pp. 53−64, 2007.
by comparing the DC stress at the end of the preconditioning [3] T.T.N. Vu et al., “Field distribution under temperature gradient in
in different scenarios, which is replotted in Fig. 6. polymeric MV-HVDC model cable: simulation and space charge
measurements,” Eur. J. Electr. Eng. vol. 17, pp. 307−325, 2014.
IV. DISCUSSION AND CONCLUSION [4] C. Chen et al., “A comparison of space charge behaviors in coaxial
cable and film sample under temperature gradient,” IEEE Trans.
The parameters of the BCT model [6] have been verified Dielectr. Electr. Insul., vol. 26, no. 6, pp. 1941−1948, 2019.
with experimental results on the MV DC cable shown in [3]. [5] X. Wang et al., “Space charge characteristics in 160 kV DC XLPE
Therefore, it is advantageous to employ the verified BCT cable under temperature gradient,” IEEE Trans. Dielectr. Electr. Insul.,
model for predicting the DC EF distribution at the end of the vol. 25, no. 6, pp. 2366−2374, 2018.
preconditioning of SIVTs. The obtained results indicate that [6] A. Hoang et al., “Simulation of electric fields in insulation of a DC
the DC EF distribution plays a vital role in determining the model cable under temperature gradient,” in 2022 IEEE 4th
International Conference on Dielectrics (ICD), pp. 214-217.
electrical stress of high frequencies in SIVTs. [7] S. Kumara et al., “Simulation of ionic contribution on space charge
EF is low (close to zero) near the inner semiconductor in characteristics of XLPE insulations,” in IEEE ICD 2022, pp. 218-221.
[8] S. Kumara, Y. V. Serdyuk, and M. Jeroense, “Calculation of electric
all studied scenarios, which results in a high maximum EF in fields in HVDC cables: Comparison of different models,” IEEE Trans.
SIVTs of opposite polarity. The magnitude of the EF is as high Dielectr. Electr. Insul., vol. 28, no. 3, pp. 1070-1078, 2021.
as the maximum EF obtained by the application of impulse [9] Y. Zhan et al., “Space charge measurement and modelling in cross-
voltages of higher peak-to-ground level. In fact, maximum EF linked polyethylene,” Energies, vol. 13 (8), p. 1906, 2020.
is approximately 71 kV/mm for both cases of SIVT LIOP of [10] S. Le Roy et al., “Description of charge transport in polyethylene using
2.1U0 and solely LI of 3.1U0. a fluid model with a constant mobility: fitting model and experiments,”
J. Phys. D: Appl. Phys., vol. 39, pp. 1427−1436, 2006.
Cable temperature defined by its transmission power has a [11] S. Le Roy, G. Teyssèdre and C. Laurent, “Modelling space charge in a
crucial contribution to the EF stress in SIVTs. Lowering cable cable geometry,” IEEE Trans. Dielectr. Electr. Insul., vol. 23, no. 4,
pp. 2361-2367, 2016.
temperature by 20 K at both inner and outer radial positions of
[12] A. Abbasi et al., “Performance evaluation of 525 kV and 640 kV
insulation resulted in a more homogeneous EF distribution extruded DC cable systems,” 10th International Conference on Power
throughout cable insulation as illustrated in Fig. 5a. This is Insulated Cables (Jicable 2019), paper A9.4.