Power System Technical Performance Issues Related To The Application of Long HVAC Cables
Power System Technical Performance Issues Related To The Application of Long HVAC Cables
Working Group
C4.502
October 2013
POWER SYSTEM TECHNICAL
PERFORMANCE ISSUES RELATED
TO THE APPLICATION OF LONG
HVAC CABLES
WG C4.502
Copyright © 2013
“Ownership of a CIGRE publication, whether in paper form or on electronic support only infers right
of use for personal purposes. Are prohibited, except if explicitly agreed by CIGRE, total or partial
reproduction of the publication for use other than personal and transfer to a third party; hence
circulation on any intranet or other company network is forbidden”.
Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept
any responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties
and conditions are excluded to the maximum extent permitted by law”.
ISBN : 978-2-85873-251-7
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Page i
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Page ii
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
BIBLIOGRAPHY/REFERENCES.......................................................................................................................... 88
ANNEX 3. MEASUREMENTS ON THE 100 KM LONG 150 KV AC CABLE CONNECTION OF THE HORNS
REV 2 OFFSHORE WIND FARM ......................................................................................................................... 98
Page iii
EXECUTIVE SUMMARY
This Technical Brochure describes system technical performance study issues important for a power system with
(a large share of) long AC cable lines. The study areas are divided into planning studies, system impact studies
and equipment design studies. The brochure contains examples and explanations of system technical performance
phenomena occurring as a result of various system operations, guidelines for simulation methods, interpretation of
the results and possible suggested countermeasures. As some of the presented examples produce overvoltages of
significant duration and / or amplitude, consideration is given to the adequacy of cable tests recommended by IEC.
The focus of this Technical Brochure is system technical performance issues and modelling of cables and other
equipment is not its main topic. Nevertheless, in order to retain the ability to stand on its own, relevant
recommendations concerning modelling of cables and other pieces of equipment as well as the required extent of
network equivalents are given. Guidelines and recommendations for adequate modelling of AC cable lines for
different types of power frequency studies and for studying phenomena that involve other frequencies are
presented. Practical methods for gaining confidence in the accuracy of cable line models are described.
Calculations of power frequency impedances based on published norms are shown. Available models for transient
studies are identified and their selection for various studies is discussed including the required level of detail,
accuracy and other implementation details. Sensitivity of study results due to variation of cable input parameters is
analysed. Line constant solver results are benchmarked against manual calculations. Also a transient cable model
validation procedure is shown. Derivation of basic computer cable line models, setups for measurement of cable
impedances and results of real-life measurements performed on 100 km long AC cable of an offshore wind farm
are also presented.
Page 1
Chapter 1. INTRODUCTION
In recent years many electricity transmission network operators have been faced with the necessity of upgrading
their transmission networks. This has been due to a variety of reasons, including an increase of cross-border
trades, integration of large scare renewable energy sources, smart grid projects, the replacement of aging facilities,
or a growing demand. Traditionally such upgrades would have been accomplished through the installation of
overhead transmission lines (OHLs). Underground cable systems (UGC) on the transmission voltage level have
been used as well, but their applications have been mainly restricted to densely populated areas. UGC have so far
been widely used on the lower voltage (distribution) levels, typically as radial lines of shorter lengths. This tendency
of not using UGC on the transmission level has been slowly changing over the past ten years due to various factors.
The purpose of high voltage transmission lines is to transmit bulk of electrical energy over longer distances and this
function must be fulfiled by UGC transmission lines as well. As the usage of such long high voltage AC cables on
the transmission level has not been common, network planners, network operators and equipment suppliers do not
typically possess much experience with the technical performance of transmission systems with long AC cables nor
with modelling of cables for studies on the transmission level. The few examples where long EHV AC cables are in
service include Japan, where a length of 40 km at 500 kV is installed, Connecticut, where Northeast Utilities and
United Illuminating have installed 37 km and 15 km double-circuit 345 kV cables into a relatively weak system, or
Denmark, where a 150 kV connection to the offshore wind farm Horns Rev 2 is installed and has a length of 100
km. Denmark has unprecedented plans for underground cabling of the entire existing 132 kV and 150 kV grids as
well as most of the planned new 400kV lines. In 2030 this plan should be completed.
Fundamentally there is nothing new in terms of the technical performance of a transmission system with long cable
lines compared to one with long overhead lines. However, due to the difference in their physical properties cables
have a much larger shunt capacitance compared to overhead lines. This results in a much greater need for reactive
power compensation, observation of resonances at the lower end of the frequency spectrum as well lower
frequency components within current and voltage waveforms generated during transients. Furthermore, modelling
of cables, especially for frequencies other than the system frequency is more complicated and demanding in
comparison to modelling of OHLs.
The main aspects where a system of AC cable lines differs from a system of AC overhead lines at the same
voltage level and transmission capacity are:
1. Level of shunt compensation required is much greater, because the cable capacitance is higher
2. Lower system resonance frequency
3. Lower series reactance, because the phase conductors are closer together
4. Special concerns related to cross-bonding and cable armour
5. For transients, the surge impedance and travelling wave velocity are both lower
6. Electromagnetic field strength can be higher along the line axis or at the edge of a right-of-way, but
decays more quickly with distance from the cable.
7. Cable faults are permanent in nature and usually require long repair time.
Different choices of cable system types may influence the need for shunt compensation or other remedial actions.
For instance, 1-Core and 3-Core solutions, onshore and offshore respectively will give varying needs of additional
equipment depending on parameters like phase separation, transpositions etc. This might give varying constraints
for the cable system depending on the connections points. Further discussions about cable system types can be
found in Chapter 3 of [8].
Page 2
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
of frequencies. This ranges from data gathering to model accuracy, effect of data on model performance,
evaluation of available cable models and their applicability for various studies, methods of validation, etc. However
it must be emphasised that modelling of cables is covered in this Technical Brochure only to the extent as required
to perform the described system studies. Information such as derivations of mathematical equations that constitute
various cable models are not presented. Such information can be found in [8], which is referred where applicable.
Summary of earlier work
This Technical Brochure contains many references to earlier work, particularly in Chapters 2-4. Some of the IEC,
CIGRE, and IEEE publications offer information of broad interest in cable studies. For example, IEC/TR 60071-4 [9]
describes guidelines and test cases for a variety of transient overvoltage studies, over time frames ranging from
temporary overvoltage to very fast-front overvoltage. Some of those test cases involve long OHL or GIL with shunt
compensation, so those results would be similar to those from cable system studies. System islanding overvoltages,
line energization, lightning transients, resonance between line and shunt reactor, and arc extinction following
reactor switching are all covered in [9].
Many CIGRE documents cover cable modelling and studies. Much of this earlier material addresses lightning
transients and the impact of cable bonding methods, so this Technical Brochure focuses on other areas. TB 110 [2]
compares typical impedance and admittance values for OHL and cables. TB 189 [3] covers insulation coordination
for HVAC cables, including the overvoltage types of interest, calculation of the highest theoretically possible
switching overvoltage, and a lightning overvoltage example. TB 189 suggested more work on cables with extruded
insulation, and on developing guidelines for the use of simulation tools, which this Technical Brochure addresses.
TB 250 [4] discusses cable electrical characteristics and critical length, switching overvoltages, lightning
overvoltages, and sheath overvoltages during faults. TB 250 includes an example of EMT simulation of lightning
overvoltage on a mixed OHL/UGC (Underground Cable) line. TB 268 [5] provided more analysis and examples of
lightning transients on long cables, and recommended future work on long-cable switching transients, which this
Technical Brochure addresses. TB 283 [6] discussed the effect of cable bonding techniques on steady-state sheath
overvoltages during faults, and transient sheath overvoltages on mixed OHL/UGC lines during lightning strikes and
auto-reclose operations. It compared sheath overvoltage calculations from theory with various numerical and EMT
models. TB 347 [7] addressed the effect of cable bonding methods on earth potential rise during faults.
From CIGRE General Sessions, [38] described studies for a 39 km, 500 kV UGC in Tokyo with load shedding
overvoltage examples, and [88] described studies for a 43 km, 400 kV UGC in Italy with focus on shunt
compensation and parallel resonance. An IEEE panel session presentation [89] described how to prepare cable
data for use in EMT studies, including model responses to a step voltage excitation. This material was later
expanded into a book chapter [90]. Two IEEE references [13][14] address the special protective relaying issues
that may arise with UGC. All of these references would be helpful to engineers beginning a cable system study.
Page 3
Chapter 2. ANALYSIS OF THE TECHNICAL PERFORMANCE OF POWER
SYSTEMS SPECIFIC FOR SYSTEMS WITH LONG HVAC CABLE LINES
2.1 Introduction
The application of cables in power grids and mixed line-cable-line sections will influence the behaviour of the power
system in several aspects. The steady state, transient and dynamic responses of cables for both small and large
disturbances differ from overhead lines. This has implications for the operation of systems with cables in terms of
technical performance and security of supply. Therefore, all steady state and transient phenomena should be
subject to study before integration of cables is realised in practice.
From the planning and the operational points of view, overhead lines cannot simply be replaced by cables. The
electrical behaviour of cables is different from that of overhead lines. An overhead line is surrounded by air without
other insulation and has a dominant inductive behaviour. A cable consists of an inner and a sheath conductor with
insulating material and semiconductor layers in between and is therefore predominantly capacitive. Cable
capacitance and inductance per unit length differ from the values for overhead lines. The series inductance of a
cable is about five times smaller and the shunt capacitance is around 20 times larger than the corresponding value
for an overhead line. As a result, the surge impedance of cables is roughly 10 times lower and the travelling wave
velocity is about half that of overhead lines. The value of the surge impedance lies between 30 and 70 Ω. This
means the surge impedance loading (SIL) for a cable is several times larger. The cable current rating is usually well
below its SIL and hence behaves like a shunt capacitor.
Reactive power surplus in any operating condition causes a power-frequency voltage rise, not only at the cable
terminations but also at the adjacent nodes in the grid. In normal operating conditions, a certain voltage step is
usually allowed while connecting or disconnecting a cable as prescribed in the grid code or the
planning/operational criteria of the system operator. To keep voltages within acceptable margins, reactive power
compensation is usually necessary. This compensation can be achieved using shunt reactors typically installed at
both ends of a cable.
The capacitive current of a cable depends on the applied voltage and the capacitance per unit length. At the critical
cable length, the cable current rating is completely consumed by the capacitive current and no active power can
flow through the cable. Shunt reactive power compensation installed along the cable route will correct this.
The amount and location of shunt compensation influences the voltage profile along the cable. Theoretically,
uniformly distributed shunt compensation may produce the best voltage profile, but at a high cost. The external
system also plays a role. System studies, as illustrated later in this chapter, identify the optimal compensation
scheme to maintain acceptable voltage profile at a reasonable cost.
Cables and overhead lines can create parallel transmission paths. The lower impedance of cables can lead to an
inequality in the power flow, which could even result in overloading the cable connections. This implies that during
steady state operation, compensation is needed in two different ways: shunt compensation for reactive power
surplus and series compensation for the difference in impedance to control the power flow on parallel paths.
The higher capacitance of a cable also has consequences for the system’s resonant frequencies. Cables are more
likely than overhead lines to create system resonances at low-order harmonic frequencies. Resonances may lead
to overvoltages or steady state harmonic problems. Cable switching transients are in principle similar to overhead
line switching transients. Shunt reactor switching should also be considered. Series and parallel resonance with
transformers may exacerbate cable switching overvoltages.
Another transient issue to consider is the zero-missing phenomenon that can occur while energizing shunt
compensated cables. The decay of the DC component of the current depends on the resistances that are, in
general, small. This results in a current not crossing zero for several cycles. Countermeasures can be taken to
avoid failure of circuit breakers in the event of a fault.
Page 4
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
From the installation point of view, high voltage underground cables longer than 1-2 km require cross-bonding to
reduce induced sheath currents. For shorter lengths, single-point bonding is generally applied. Solid bonding
(earthing in several points) is generally applied to submarine cables but single core cable laying may require
special studies. See Figure 3-2 for an illustration of cross-bonding, and [8] for illustration of other bonding methods.
2. System impact study or detailed system design study, which determines the impact of a cable on the rest
of the power system. At this stage, the high-level specification has been determined. The system impact
study may result in some adjustments, or mitigations applied to the cable or other parts of the system.
Study topics include harmonic resonance, short circuit currents, transient stability, voltage stability, and
system relaying. The study tools include short-circuit, stability, and harmonic analysis programs, and in
some cases an EMT program to explore resonant overvoltages. The modelling needs vary from lumped
parameter to distributed parameter, from positive sequence to three-phase unbalanced representation, and
from DC to a few kHz, depending on the subject. Models are often generic in early studies, later
progressing to specific models for particular equipment.
3. Equipment and system design study, which establishes detailed protection and operating procedures for
the cable, sheath, switchgear, shunt compensation, and related equipment. The basic study tool is an
electromagnetic transients (EMT) program, which can also handle the grounding and bonding connections.
Accurate frequency dependent models are preferable and sometimes necessary for many of these studies.
4. Forensic study, which becomes necessary if a cable or equipment failure occurs during or after
commissioning. This type of investigation is not unique to cables. Information in this Technical Brochure on
modelling and phenomena may help fulfil this type of study.
There may be some iterations between study phases. For example, in an extreme case, resonance problems
identified in phase 2 may change the choice from high-pressure fluid-filled (HPFF) to cross-linked polyethylene
(XLPE) cable, in order to reduce the total capacitance. In another example, zero-miss phenomena may influence
the shunt reactor arrangement. However, usually, the output of one study phase fixes the input to the next phase.
5. Electromagnetic Fields (EMF) and all other environmental issues , as they affect the overhead vs.
underground choice
Page 5
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
2. Self-excitation in synchronous generators may occur when the machine’s internal reactance is greater than
the equivalent external reactance. A system with cables is more likely to meet this condition than a system
with overhead lines, as described in section 2.5.6.
3. Voltage stability – Underground cables (UGCs) provide more reactive power than overhead lines (OHLs).
Switched or variable shunt reactors associated with UGCs may provide more flexibility in reactive power
control. However, the loss of these components must be considered as a contingency in voltage stability
analyses.
4. Transient stability - The lower series reactance of cable lines is favourable to angle stability. On the other
hand, underexcited operation of nearby synchronous generators is detrimental due to smaller internal
electromotive force. and higher load angle. Usually the net effect is favourable.
5. Small signal stability - For due diligence, the small-signal stability of the system should be checked to
determine whether cable installation has any significant impact due to changes in system impedances.
6. Temporary Overvoltages (TOV) are more likely with cables, due to:
b. Transformer energization
7. Ferroresonance is sensitive to the amount of capacitance isolated with a transformer winding. Cables may
provide enough, or too much, capacitance for classical ferroresonance to occur. Transformers and long
cables should not be switched together.
8. Protection issues. References [13]–[15] provide a starting point for considering the impact of cables on
relaying and protection.
2. Energization of long HVAC cables from already energized cables (similar to back-to-back capacitor
switching).
5. Circuit-breaker restrike after interrupting small inductive current (i.e. disconnecting a shunt reactor).
6. Lightning transients and arrester placement at cable junctions, open points and other locations. Arresters
may not be needed at all within the cable system [5].
Page 6
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
7. Auto-reclosing. Impact on system availability and reliability could be considered in the Planning Study. If
reclosing is to be used on a mixed OHL/UGC line, impacts on sheath requirements for current carrying
capability and the effect of mutual couplings of shunt reactors should be considered.
8. Cable discharge of trapped charges through inductive voltage transformers (VTs) and their energy
capability requirements.
9. DC offset in cable and shunt reactor energization and zero-miss phenomenon when high compensation
degrees are deployed.
11. Saturation of current transformers from DC offset currents when energizing shunt reactors and cable.
12. Synchronized (also known as point-on-wave) switching to mitigate inrush currents or overvoltages.
15. Effect of cables on circuit breaker transient recovery voltage (TRV). Because cables have lower surge
impedance and higher capacitance, the TRV rate-of-rise is usually lower than with overhead lines. Fault
currents may be higher with cables due to lower series impedance, but in general, the TRV is less severe
with cables. However, any associated current-limiting reactors (CLR) would require their own TRV study.
16. Effect of cables on incoming surges to the substation. The substation insulation coordination and arrester
placement are usually based on a design incoming surge due to lightning. If the connected circuits are all
underground cables, there will be no direct incoming lightning surges. A different design base for
substation insulation coordination must be selected. This can be an attenuated lightning surge from a
mixed OHL/UGC line, or some type of high-frequency switching surge.
For example, the “active power transmission” issue in the fourth row of Table 2-1 is important, but the concerns are
nearly the same as for overhead lines, resulting in a low weight for the purpose of this Technical Brochure. On the
other hand, “reactive power compensation” in the first row is both important, and of more concern for cables than
for overhead lines, resulting in a high weight.
The study tools mentioned in Table 2-1 are:
Cable Constants – a program that calculates and represents electrical RLC parameters in a matrix form for
a general system of cables and other conductors, over a range of frequencies, and using either transposed
or full unbalanced assumptions. This function may be bundled with another tool, or used separately.
Power flow – calculate steady-state voltages and currents based on a positive sequence model, with non-
linear loads. The cable model is symmetric and transposed. Power flow is the basic tool for transmission
planning.
Short-Circuit – a program that solves for voltage and current during faults, especially three-phase and
single-phase-to-ground faults. The model is linear, symmetric, and assumes phase transposition. An
auxiliary Protection function simulates the response of relays to fault current and voltage.
Page 7
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Dynamics – a time-domain simulator based on numerical integration of differential equations. It differs from
an EMT program in focusing on (slower) electromechanical and control system transients, rather than
electromagnetic transients. The models are sometimes linear and balanced. The program usually includes
eigenvalue analysis, or other functions for small-signal stability.
Harmonics – a frequency domain program that solves voltage and current over a range of frequencies,
using linear or non-linear load and source models, and balanced or unbalanced impedances. The
frequency-scan function outputs driving point impedance, as obtained from the bus voltage for a unit
current injection.
3Φ-sh – three-phase unbalanced distributed parameters, retaining the cable sheaths as separate
conductors.
DC-kHz – parameters used up to switching surge frequencies, and may also be frequency-dependent.
Typical for slow-front overvoltages.
MHz – parameters used up to lightning / fast switching surge frequencies, and may also be frequency-
dependent. Typical for fast-front overvoltages.
Custom tools can also be built on general matrix solvers [17]. These work well in focused studies of different cable
phenomena, but they may be less convenient for system operators to use with existing large-scale models.
Page 8
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1
Notation applied in this drawing is the following: numbers next to generator symbols are their MW generation,
numbers next to load symbols are their MW consumption, numbers next to reactive power compensation symbols
are their MVar generation, and numbers next to transformer symbols are their MVA rating.
2
Numbers next to transformer symbols are their MVA rating (for example 3*500 means 3 transformers of 500 MVA
rating).
Page 9
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
With one cable in each corridor and simple equivalent 230 kV source inductances of 30 mH at Sub A and 25 mH at
Sub C, a frequency scan from Sub B shows a resonant frequency near 100 Hz. This resonance is at the 2nd
harmonic, suggesting possible resonant overvoltages after energizing the transformer at Sub B. RLC ladder
networks may be used to synthesise more complicated network equivalents of the system shown in Figure 2-2.
28.96 km
km
.71
54
96.54 km
107.80 km
19.31 km
43
.4
4
km
54.71 km
46.66 km
km
88.50 km
.44
43
37.01 km
km
.45
80
25.74 km
Page 10
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Page 11
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Page 12
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
2. Degree of shunt compensation. Overhead lines are typically restricted to 60% to 80% shunt compensation
due to risks of line resonance during open-phase conditions [23]-[24]. This phenomenon is caused by
capacitive coupling between the disconnected phase and the remaining energized phases. Given that
there is no capacitive coupling between insulated cables, the restriction on low degree of shunt
compensation does not apply. Compensation degrees close to 100% are typical for underground cables. In
the case of mixed cable-overhead line circuits, inter-phase capacitive coupling will be present in the
overhead line section. Special attention must be taken in these configurations if single-phase autoreclose is
to be applied.
3. Voltage rise at the receiving (open) end must not exceed the maximum continuous operative voltage
(MCOV) or TOV capabilities of shunt reactors, surge arresters, or other equipment. The voltage rise is
often higher in the presence of cables.
4. No-load current of a compensated cable must not exceed the rated capacitive switching current capability
of circuit breakers. If necessary, special-purpose or higher-voltage circuit breakers may be used.
The main questions the system planner needs to answer when faced with a new long cable circuit are:
1. Does the cable circuit need shunt compensation?
2. Does the shunt compensation need to be connected to the cable circuit or can it be placed at the
substation ends (i.e. busbars or tertiary windings of transformers)?
4. What is the optimum shunt compensation scheme for the selected cable circuit?
The following step-by-step procedure helps the system planner answer those questions by using standard power
flow analysis techniques. The following section provides an example. This procedure is based solely on steady-
state analysis looking at voltage rise due to Ferranti effect and voltage step during switching procedures. It applies
to the analysis of a single long HVAC cable circuit. Other limitations on the size and location of the shunt reactors
Page 13
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
may arise from time-domain analysis of resonance conditions. The system planner must be aware of both
limitations and some level of iteration may be required.
Step 1. Establish acceptance criteria for receiving end voltage rise, sending end voltage step, cable active
power reduction due to charging current, and capacitive breaking current of circuit breakers. If
these conditions are not fulfilled then line connected shunt compensation is necessary. Notably, if
the active power reduction is not acceptable, compensation is needed at intermediate locations.
The minimum shunt compensation degree has to be set accordingly, e.g. following the procedure in
[25].
a. Analyse voltage profile using intact network topology. This test may identify the need for shunt
compensation.
b. Analyse voltage profile along cable length assuming open circuit-breaker at one end. This test
determines the need and required location for the shunt reactors – i.e. at the cable ends or at the
substation busbar/transformer tertiary winding.
c. Define various shunt compensation schemes with reactors at either end of the circuit, at both ends
or at different locations along the circuit. Open circuit-breaker at each cable end must be
considered.
d. Assess the preferred compensation scheme using intact network topology. Circuit thermal loading
and system voltages should be within acceptable limits.
e. Assess the preferred compensation scheme using applicable planning criteria (e.g. single circuit
contingency). Post-disturbance circuit thermal loading and system voltages should be within
acceptable limits.
f. If either Step 2d or Step 2e fails, then the scheme is not satisfactory. A new compensation scheme
must be investigated with different sizes of reactors in different locations along the cable circuit.
Step 2 must be repeated with the new preferred option. A number of iterations may be required.
Step 3. Create power flow model representing a maximum demand scenario and verify the scheme
resulting from Step 2.
Step 4. Assess the maximum unit size of shunt reactor that can be switched individually to comply with the
required voltage step limits. This test will also give an indication of requirements for static or
dynamic shunt compensation.
2. Open-circuit voltage profile along the cable circuit under minimum system loading conditions.
3. Normal operating voltage profile under maximum and minimum system loading conditions.
4. Different source impedances that may exist when switching the cable from either end.
Page 14
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
This procedure applies to the analysis of a single long cable. In areas with high concentration of short cables or
lower voltage levels, it may be more appropriate to carry out area-wide analysis to determine the optimum number,
size and location of shunt reactors, taking technical constraints into consideration and with the objective of
minimising system losses and costs.
Figure 2-3: Power Flow Representation of Cable for Shunt Reactive Compensation Analysis
Step 1. Establish acceptance criteria for voltage rise and voltage step.
The assumed Planning Criteria, which may vary among different system operators, are:
Normal voltage limits: 0.95 pu to 1.1 pu
For single contingencies the maximum step change between pre- and post- contingency steady-state
voltages shall not be more than 10%.
Results are plotted in Figure 2-4. Cable connection results in a significant voltage increase when
compared with the equivalent overhead line circuit. In particular, the line voltage along the cable
Page 15
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
circuit is well above the normal limit defined in Step 1 (1.1 pu). This result clearly indicates that
reactive compensation is required for the cable connection.
1.4
1.3
1.2
1.1
1
#17 25% 50% 75% #22
Figure 2-4: Voltage Profile for Intact Network Condition during Minimum Demand
b. Analyse voltage profile along cable length assuming open circuit-breaker at one end.
A fault level analysis is carried out at buses 17 and 22 to identify the weakest source. Results in
Table 2-6 indicate that bus 22 has the lowest short circuit power. Therefore, the highest voltage
rise will occur when the cable circuit is energized from bus 22 with the circuit breaker open at bus
17.
Table 2-6: Short Circuit Power at Circuit Sending and Receiving End
Bus 17 Bus 22
Cable 1963 MVA 1686 MVA
Overhead line 1901 MVA 1475 MVA
Analyse voltage profile assuming circuit energized from weakest source (bus 22) and circuit
breaker open at strongest side (bus 17). Simulate the opening of the circuit breaker at the
strongest cable end (bus 17) and solve load-flow for the cable energized from the weakest
source (bus 22). Results are plotted in Figure 2-5. In this particular example (for the UGC option
only) power flow could not converge due to the nearby generators reaching their under-excitation
limits and being unable to control voltage. A fictitious solution was found in this case, for
illustration purposes only, by increasing the reactive power limits of the generation units. This is a
severe condition and should be avoided. (In Figure 2-4, bus 17 absorbed the excess UGC
reactive power without exceeding any generator under-excitation limits, so that power flow
solution did not have to be modified.) In order to reduce voltages below 1.1 pu and ensure that
generators are operated within the reactive power capabilities, a shunt compensation scheme
must be implemented. The results of the open-circuit analysis indicate that the required shunt
Page 16
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
compensation must be directly connected to the cable, ruling out the bus or transformer location
options.
Open CB Closed CB
Figure 2-5: Voltage Profile Results with Circuit Energized from Weak Source (Bus 22) and
Open-end at Strong Source (Bus 17)
Scheme 1: Shunt reactors installed at the strongest source cable end (bus 17).
The results of the open-circuit test with cable energized from the weaker source are plotted in
Figure 2-6. The minimum size of shunt reactor that guarantees compliance with the voltage limit
criteria is 700 Mvar.
A second open-circuit test is carried out with Scheme 1 (700 Mvar at bus 17) by energizing the
cable from the strong source and leaving the circuit breaker open at the weak source end. The
results of this test are plotted in Figure 2-7. It can be seen that line voltages along the cable are
below the normal and emergency voltage limits defined in Step 1, therefore Scheme 1 is rejected.
Page 17
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1.125
1.100
1.075
1.050
1.025
1.000
#17 25% 50% 75% #22
Figure 2-6: Scheme 1 (Reactors in Cable-end Bus 17) Voltage Profile with Cable Energized
from Weak Source (Bus 22) and Open-end at Strong Source (Bus 17)
0.975
0.950
0.925
0.900
0.875
0.850
0.825
0.800
#17 25% 50% 75% #22
Figure 2-7: Scheme 1 (Reactors in Cable-end Bus 17) Voltage Profile with Cable Energized
from Strong Source (Bus 17) and Open-end at Weak Source (Bus 22)
Page 18
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Scheme 2: Shunt reactors installed at the weaker source cable end (bus 22).
The results of the open-circuit test with cable energized from the weaker source are plotted in
Figure 2-8. It can be seen that the minimum size of shunt reactor that guarantees compliance
with the voltage limit criteria is 600 Mvar.
A second open-circuit test is carried out with Scheme 2 (600 Mvar at bus 22) by energizing the
cable from the strong source and leaving the circuit breaker open at the weak source end. The
results of this test are plotted in Figure 2-9. It can be seen that line voltages along the cable are
below the normal voltage limit defined in Step 1, therefore Scheme 2 is rejected.
Figure 2-8: Scheme 2 (Reactors in Cable-end Bus 22) Voltage Profile with Cable Energized
from Weak Source (Bus 22) and Open-end at Strong Source (Bus 17)
Page 19
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1.025
1.000
0.975
0.950
0.925
0.900
#17 25% 50% 75% #22
Figure 2-9: Scheme 2 (Reactors in Cable-end Bus 22) Voltage Profile with Cable Energized
from Strong Source (Bus 17) and Open-end at Weak Source (Bus 22)
The results from Scheme 1 and Scheme 2 assessments indicate that the shunt compensation
needs to be split between the two cable ends in order to avoid excessive voltage rise or drop.
This is the objective of Scheme 3. The results of the open-circuit test with cable energized from
the weaker source are plotted in Figure 2-10. In this test the reactor size is increased at either
cable end until a solution that meets the Planning Criteria is found. This is initially achieved with
a 200 Mvar + 220 Mvar combination; however detailed inspection of the operating condition for
the generators connected to bus 22 revealed that they are running at their under-excitation limits,
which is an undesirable condition for stability. An increase of size in the reactor connected to the
bus 22 end results in a higher operating power factor in the nearby generators, providing an
adequate stability margin. Further reductions in the reactor size connected to bus 17 give rise to
overvoltages at the receiving cable end or result in the generators reaching their under-excitation
limits again. The selected shunt reactor combination from this test is 200 Mvar + 250 Mvar.
A second open-circuit test is carried out with Scheme 3 (200 Mvar at bus 17 + 250 Mvar at bus
22) by energizing the cable from the strong source and leaving the circuit breaker open at the
weaker source end. The results of this test are plotted in Figure 2-11. It can be seen that line
voltages along the cable are within the normal voltage limits defined in Step 1, therefore Scheme
3 is technically acceptable under open circuit breaker conditions. Scheme 3 becomes the
preferred option.
Page 20
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure 2-10: Scheme 3 (Reactors in Both Cable-ends) Voltage Profile with Cable Energized
from Weak Source (bus 22) and Open-end at Strong Source (Bus 17)
1.075
1.050
1.025
1.000
#17 25% 50% 75% #22
Figure 2-11: Scheme 3 (Reactors in Both Cable-ends) Voltage Profile with Cable Energized
from Strong Source (Bus 17) and Open-end at Weak Source (Bus 22)
Page 21
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Preferred option is Scheme 3 with 200 Mvar at cable-end bus 17 and 250 Mvar at cable-end bus 22.
Power flow analysis for minimum demand scenario indicates that circuit thermal loading and system
voltages are within acceptable limits.
d. Assess the preferred compensation scheme option in single contingency
Contingency power flow analysis for minimum demand scenario indicates that circuit thermal loading
and system voltages are within acceptable limits.
Power flow analysis for maximum demand scenario indicates that circuit thermal loading and system
voltages are within acceptable limits.
b. Assess the preferred compensation scheme option in single contingency
Contingency power flow analysis for maximum demand scenario indicates that circuit thermal loading
and system voltages are within acceptable limits.
The cable voltage profile under normal and open-circuit conditions is plotted in Figure 2-12. These
results indicate that the preferred scheme provides adequate performance under maximum demand
conditions.
1.09
1.08
1.07
1.06
1.05
#17 25% 50% 75% #22
Figure 2-12: Scheme 3 (Reactors in Both Cable-ends) Voltage Profile in Maximum Demand
Conditions
Page 22
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Step 4. Assess the maximum unit size of shunt reactor that can be switched individually.
Calculate the largest reactor size that can be switched individually to comply with the 3% voltage deviation
limit defined in the Planning Criteria. A minimum system loading scenario is used as it leads to the highest
voltage deviations. The results are:
175 Mvar maximum at the bus 17 cable end
Table 2-7 shows the selected compensation scheme out of three considered, accounting for all steady-state
constraints.
MVA s
h , (2.2)
MVA c
where MVAs is the system short circuit strength at the cable and MVAc is the cable’s reactive power at nominal
voltage. Local shunt compensation can be added to MVAs. It must be pointed out that this formula neglects the
capacitance of the existing network and therefore overestimates the frequency of the first parallel resonance.
Page 23
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Table 2-8 shows the positive-sequence line parameters used to illustrate frequency scan analysis on the test
system in Figure 2-1. The 230 kV underground parameters match those used in the previous shunt compensation
example. The 138 kV underground parameters have been modified to yield realistic values for series impedance,
surge impedance and propagation velocity. In [18], the line charging value had been increased without a
corresponding reduction in the series impedance for 138 kV cables. The frequency scan cases cover gradual
conversion of all overhead lines to underground, in the following sequence:
Base case with two existing underground 138 kV lines
Convert all remaining 230 kV lines to underground (no overhead lines left in the system)
At each step, shunt compensation of approximately 90% is added for each cable, split equally between the two
ends.
Table 2-8: Many-cable Example Test System Line Parameters for Frequency Scan Example
Line Type R1’[pu/km] X1’[pu/km] B1’[pu/km]
230 UGC 2.980E-05 2.988E-04 3.763E-02
230 OHL 1.149E-04 8.965E-04 1.883E-03
138 UGC 5.385E-04 4.158E-04 1.520E-02
138 OHL 6.168E-04 2.489E-03 6.775E-04
Figure 2-13 shows the positive sequence driving point impedance from one of the 138 kV transformer terminals on
Bus 10 (“H pu” on the horizontal scale refers to harmonic order, which is also the per-unit frequency). These results
include long-line PI adjustments at each frequency, but no other frequency-dependent line or cable effects. Loads
provide some damping. There are two short cables in the base case, including one that terminates on Bus 10.
These produce a parallel resonance around the 8th harmonic, and the addition of one long 230 kV cable does not
affect this resonance. Conversion of all 138 kV lines to underground will produce parallel resonance near the 2.5,
7.5, and 14th harmonics. Removing all 138 kV shunt reactors has little effect on the frequency scan results at higher
frequencies than the fundamental. Their effects could be more noticeable if the loads were not modelled. When all
overhead lines have been converted to underground, three additional resonant frequencies appear, and two of
them are tuned to the 2nd and 3rd harmonics.
Figure 2-14 shows the positive sequence driving point impedance from the 230 kV side of the same transformer on
Bus 12. The base case impedance is primarily inductive at this bus, even when all 138 kV lines are underground.
The long cable from Bus 17 to 22 also has little impact at Bus 12. However, when all of the 230 kV lines are
underground, at least nine parallel/series resonance points appear in the result.
Many systems have seen growth in background harmonic distortion levels, even at EHV transmission buses. After
converting overhead lines to underground, new parallel resonances at lower frequencies will probably exacerbate
the harmonic distortion. A complete evaluation should consider variable load level because that influences the
damping. Higher damping is indicated with a reduction in magnitude or increase in phase angle of the impedance
at resonance. The study should also consider line outages, transformer outages, and generator outages, because
they can shift the resonances to lower frequencies and increase the peak impedances.
Page 24
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure 2-13: Many-cable Example Test System Frequency Scans from Bus 10
Figure 2-14: Many-cable Example Test System Frequency Scans from Bus 12
Page 25
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
On a system with a long HVAC cable the inductance of the shunt reactors and the cable's distributed capacitance
form a parallel resonance circuit. Parallel resonance is characterized by large impedance at the resonant
frequencies, with relatively low current drawn from the external circuit (current source). Larger currents may
circulate through the resonant L and C and produce higher transient overvoltages [26].
An example situation is the energization of a transformer through a cable. A simple model and its harmonic
equivalent circuit are shown in Figure 2-15. Harmonic current due to the transformer energization is injected to the
parallel resonance circuit composed of a long HVAC cable, its shunt reactor and equivalent source impedance.
Figure 2-15: Parallel Resonance due to a Transformer Energization and its Harmonic
Equivalent Circuit
The parallel resonance frequency of the equivalent circuit in Figure 2-15 can be found as:
1 1 1
f (2.3)
2 LC L0 C
Page 26
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Severe TOV is expected when the parallel resonance frequency found by (2.3) is a multiple of the fundamental
frequency, especially 2nd harmonic. Considering typical compensation level of a long cable around 100 %, it
requires a weak source, namely large L0, to have the parallel resonance frequency at 2nd harmonic.
The parallel resonance caused by a transformer energization is simulated with the long-cable example test system
in Figure 2-2. The 400/230 kV transformer at Sub B is energized from Sub A and Sub C through long cables. As
explained in Section 2.3, a frequency scan from Sub B 400 kV bus shows a resonant frequency near 100 Hz, with
one cable in each corridor and simple equivalent source inductances of 30 mH at Sub A and 25 mH at Sub C. The
result of the frequency scan is shown in Figure 2-16.
Figure 2-16: Impedance of the Long-cable Example Test System Seen from Sub B 400 kV
Bus
Figure 2-17 shows the inrush current (bottom) and the parallel resonance overvoltage caused by the inrush current
(top), reaching a peak of 1.27 pu. This overvoltage has a strong second harmonic component. In a similar case
using overhead line with 80% shunt compensation, parallel resonance overvoltage also occurs, with a strong
seventh harmonic component. The relative voltage peaks depend on the instants of switch pole closings but in the
worst case, peak overvoltages were higher with cable. The overvoltages also decay faster with overhead line.
These observations hold true when the system model includes frequency-dependent loads and/or frequency-
dependent losses in lines and transformers.
Page 27
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
250
125
-125
-250
-375
-500
0.0 0.1 0.2 0.3 0.4 [s] 0.5
(file SimpleModel.pl4; x-var t) v:SUB_BA v:SUB_BB v:SUB_BC
300
[A]
200
100
-100
-200
-300
0.0 0.1 0.2 0.3 0.4 [s] 0.5
(file SimpleModel.pl4; x-var t) c:SUB_BA-X0001A c:SUB_BB-X0001B c:SUB_BC-X0001C
Figure 2-17: Parallel Resonance Overvoltage (Top) and Inrush Current (Bottom)
Page 28
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1
f (2.4)
2 L S C S
Severe TOV is expected when the frequency of a harmonic component in the cable energization current matches
the series resonance frequency found by (2.4).
ZS = 0
Figure 2-18: Series Resonance due to a Cable Energization and its Harmonic Equivalent
Circuit
The series resonance caused by a cable energization is simulated with the long-cable example test system in
Figure 2-2 with some modifications. First, the 400 kV cable between Sub A and Sub B is shortened to 9 km. Also,
the 400 kV cable between Sub B and Sub C is shortened to half of the original length and is energized from Sub B.
After these modifications, the cable energization overvoltage contains a 140 Hz component.
In order to set the series resonance overvoltage at 140 Hz, the total capacitance of 230 kV cable (CS) is set as
287.9 MVA (CS = 17.33 μF), connected at Sub B. Due to this large charging capacity, 230 kV shunt reactors are
required to compensate the charging capacity and thus maintain reasonable steady-state voltage. In this example,
90% compensation is assumed with total 260 MVA shunt reactors.
1 1
f 140 Hz (2.5)
2 LS C S 2 0.07459 (17.33 106 )
Figure 2-19 shows the series resonance overvoltage on the LV side of the 400/230 kV transformer in Sub B.
Severe TOV occurs due to the series resonance.
Page 29
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
360
[kV]
240
120
-120
-240
Figure 2-19: Series Resonance Overvoltage in the LV Side Bus of the 400/230kV
Transformer
TOV are usually evaluated against surge arresters TOV and energy absorption capabilities, because surge
arresters are the most likely to fail due to TOV. These resonance TOVs are weakly damped, because they occur at
low frequencies. These TOV have characteristics of both power frequency TOV and high-energy switching surges,
which are tested separately for surge arresters. A higher line discharge class or multiple columns would help with
energy absorption, but not with sustained TOV. A higher arrester voltage rating could help with both.
Other countermeasures to the parallel and series resonance overvoltage include synchronized (also known as
point-on-wave) switching for the transformer energization and the application of operational restrictions.
2.5.2.3 Ferroresonance
Ferroresonance consists of interactions between capacitors and saturable iron-core inductors [27][28]. It can occur
when the transformer and cable become isolated, and the cable's capacitance is in series with the transformer
magnetizing characteristic. This may lead to sustained temporary overvoltages that damage surge arresters and
other equipment.
If one phase of a circuit breaker opens while the other two remain closed the open phase of the transformer is still
energized through the mutual inductance to the other phases. Overvoltage occurs on all phases, with highest
magnitude occurring on the open phase. This type of ferroresonance is more common on distribution systems,
where fuses may be employed. More information on particular cases can be found in WG C4.307 TB (under
development).
Ferroresonant voltages may also occur even without circuit breaker malfunction, if a transformer and cable are
energized together. To defer switchgear costs, high-voltage systems have sometimes used low-side switching or
transformer-terminated switching, in the early stages of development. In low-side switching, a transformer is
energized from the low side with an unloaded line already connected to the high side. This is often safe with
overhead lines, but should be avoided with long cables. In transformer-terminated switching, a line is energized
with an unloaded transformer already connected to the remote end, as shown in Figure 2-20.
An important factor in a ferroresonance analysis is the point-on-wave switching angle, which affects the voltage
distortion and amplitude, similar to inrush current effect in a transformer energization [29]. Its influence can be seen
in Figure 2-20, which shows the voltage in the transformer secondary during the energization of a transformer-
terminated 108 km cable for two different switching angles (simulation based on the long cable test system).
Page 30
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure 2-21: Equivalent Circuit for TOV after Ground Fault Clearing
200.00
Volt-Seconds Peak
100.00
0.00
-100.00
-200.00
-300.00
-20.00 0.00 20.00 40.00 60.00 80.00
Amperes Peak
Page 31
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure 2-24 shows the ground-fault-clearing TOV from a case based on the long-cable test system in Figure 2-2. All
four cables are in service, with a parallel resonance at 80.5 Hz as determined by frequency scan. The fault occurs
at the Sub A end of a 108 km cable, and is cleared after 3 cycles. With that cable out of service, the first parallel
resonance frequency increases to 92.5 Hz. The shunt reactors have a Q factor of 500, and a 200 MVA load with
0.98 power factor was included at Sub B.
The surge arresters are usually most vulnerable to TOV; they are not intended to mitigate TOV and may fail
thermally during the event. Arresters are tested for TOV capability [30], but the standard test employs a stiff power-
frequency voltage source with constant output. The TOV in Figure 2-24 occurs at higher frequency and variable
magnitude, so both the energy discharge and TOV capabilities of the arrester must be evaluated.
The TOV peaks in Figure 2-24 do not follow a constant or steadily decaying profile, which makes the evaluation
difficult. Arrester switching surge energy and lightning discharge capabilities do not apply to the longer TOV time
frame, because the arrester failure mechanisms are different. In addition to thermal discharge capability, the
dielectric strength also plays a role in determining TOV capability. The arrester TOV capability should not be used
as a withstand envelope. The arrester won’t necessarily withstand a voltage waveform that lies under the TOV
curve, which describes a series of independent tests at different times and constant voltage magnitudes. It has
been suggested that comparing simulated energy discharge to the lightning impulse discharge energy capability,
as defined in Annex N of [31], may give conservative results for harmonic TOV, pending the possible adoption of a
new thermal rating system proposed for [31]. Another option is to define a TOV level and time that the voltage
peaks exceed for 90 % of the TOV time, and convert that to a 10-second equivalent TOV [32]. However, the
waveforms in Figure 2-24 are not steady enough to use that procedure. The arrester vendor should be consulted in
such cases; this was done for a recent study of TOVs in Southwest Connecticut [33].
If an arrester fails during this event, it creates a second fault right after the initial fault. The consequences to
reliability may be severe because at least one more component outage occurs. System loads provide significant
TOV damping; however, on the other hand higher pre-fault voltages exacerbate the TOV by increasing transformer
inrush currents. It’s difficult to control this TOV. To date, ground fault TOV has been mitigated by:
Limiting the cable lengths to avoid resonances at or below the 3rd harmonic
Increasing the arrester voltage ratings, which adversely affects insulation coordination
Page 32
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Reducing system voltage, especially during light load periods. This reduces transformer inrush current
magnitudes.
Other methods have been proposed, including C filters and reactor switching [34][35][36].
A TOV study should include various system contingencies, fault locations, shunt compensation levels, and loading
levels as these affect the resonant frequency and damping rate of TOVs.
Em L 1 1
Vm , 0 , (2.7)
L0 (1 2 CL) L CL0 CL
where L0 is source impedance of the weaker islanded system and Em is the source voltage behind L0. Charging
capacity of the long cable and inductance of the shunt reactors directly connected to the cable are expressed by C
and L respectively.
Page 33
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
2.23 pu
1.0s
1.69 pu
1.0s
Page 34
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
After forced disconnection at the land end of a long cable connecting a large offshore wind farm, the cable system
and the wind farm will be in island operation. Such forced events are rare, but do happen, introducing TOV in the
cable system. In the forced island operation of the 160-MW Horns Rev 1 offshore wind farm during 2006, the
temporary overvoltage was approximately 2 pu of the nominal cable voltage with total duration of a few seconds
[39]. The Horns Rev 1 cable system has no arresters; therefore, this overvoltage stressed the cable. In general, the
overvoltage magnitude and duration depend on the cable capacitance, compensating reactor inductance,
transformer, wind farm operating conditions, and wind turbine technology.
The examples in Sections 2.5.2 to 2.5.4 revealed severe temporary overvoltages related to long or many cable
systems. In order to test the insulation performance of cables, lightning impulse voltages, switching impulse
voltages and power frequency test voltages are applied to cables according to international and national standards.
IEC 62067 defines the test methods and requirements for XLPE cables and accessories for rated voltages above
150 kV up to 500 kV. According to IEC 62067, the test voltage 440 kV is applied to 400 kV cables for 15 minutes,
as power frequency voltage test, after the lightning impulse voltage test.
Considering the voltage level that can be attained by TOVs, the level of applied power frequency voltage appears
to be somehow low even though the duration is sufficient. This mismatch is not an immediate issue as the
overvoltage is normally evaluated against the insulation performance and the energy absorption capability of surge
arresters. However, if one wishes to evaluate the insulation performance of cables, a theoretical conversion of
insulation performance to a shorter duration and a higher voltage level is possible using a voltage-time relationship
of insulation performance. With the voltage-time relationship, a severe temporary overvoltage can be evaluated to
be within the insulation performance of cables and accessories when the following inequality is satisfied.
where V and t are the level and the duration of temporary overvoltages, and Vtest and ttest are the level and the
duration of applied power frequency test voltages. n is a parameter that can generally be obtained from a cable
manufacturer with typical values ranging from 9 up to 15.
Page 35
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1500
reactance x*=2 p.u.
Cable line
Overhead line
2(31600 mm2)
1000
2(3630 mm2)
Size
[MVA] 31600 mm2)
3630 mm2)
500
OHL
0
0 10 20 30 40 50
[km]
Figure 2-27: Minimum Generator MVA to Avoid Self-excitation on an Unloaded Line
1.6
1.4
1.2
VT2 VT1
Voltage [p.u.]
1.0
0.8
Gen Sat
0.6
UGC-w/Rct
0.4 UGC-No Rct
0.2
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Current [p.u.]
Page 36
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
The critical clearing time is mainly dependent on the X/R-ratio and the ratio between line and generator impedance.
Because these parameters are different for cables and overhead lines, the critical clearing time changes when an
overhead line is replaced with a cable. Figure 2-29 shows the critical clearing time for overhead line and
underground cable, based on the 108 km line length. The critical clearing time is slightly higher for the underground
cable, especially for fault locations close to the substations.
These factors influence the critical clearing time in a similar way for both overhead line and underground cable:
Decreasing line length causes the critical clearing time to increase.
Increasing the series inductance decreases the critical clearing time and vice versa. However, for very low
values of series inductance the generator reactance dominates, and further decreasing series inductance
has little effect.
Increasing the series resistance provides more damping and increases the critical clearing time.
OHL UGC
0.24
0.23
Critical Clearing Time [s]
0.22
0.21
0.20
0.19
0.18
0.17
0.16
0.15
0 20 40 60 80 100
Page 37
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1. Is the cable insulation adequate, and if not, is it necessary to use remedial measures such as pre-insertion
resistor, synchronized switching or line surge arresters.
2. If the cable connects two systems with different strengths (i.e. fault levels), one may want an operating
restriction on the end from which the energization will take place (i.e. best-end switching).
Unless there is absolute certainty that there cannot be re-energization with trapped charge, a statistical simulation
of at least 200 cases should be run to assess the maximum overvoltages with trapped charge.
A simulation of overvoltage and overcurrent during cable energization was carried out using the long-cable test
system in Figure 2-2; specifically the 108 km cable connected directly to a 400 kV ideal voltage source. CB1 is
closed to energize the cable, with CB2, CB3, and CB9 open. Transformer models are not included in this test.
Figure 2-30 shows the voltage in the receiving end of the 108 km long cable and the current in the sending end.
The maximum value of overvoltage occurs in phase A (blue line). The dominant transient frequency is
approximately 260 Hz, which is lower than for typical overhead lines. An overvoltage of 2.1 pu is reached. The
sustained large current as a result of the transient might cause overcurrent relays to mis-operate.
Figure 2-30: Receiving End Voltages and Sending End Currents during Energization
The overvoltage caused by cable energization is also affected by cable type and sheath bonding. The next
theoretical example compares a 400 kV single-core cross-bonded cable, a 400 kV single-core two-point bonded
cable, and a 132 kV pipe-type cable. Figure 2-31 shows the cross section diagrams of the single-core cable and
the pipe-type cable. The single-core cable is assumed to be laid in a flat formation with 0.35 m phase separation.
Figure 2-31: Cross Section Diagram of Single Core Cable (Left) and Pipe Type Cable (Right)
The cable length is set to 1.5 km and the length of the minor section for the cross-bonded cable is 0.5 km. One
phase of the cable is energized through 1 Ω resistance. The cable sheath is grounded at both ends through 10 Ω
grounding resistance.
Figure 2-32 compares the energization overvoltages for the three types of cables. The dominant frequency of the
energization overvoltage in the cross-bonded cable is much lower than the other two types of cables, which leads
to lower damping of the overvoltage. In addition, in the cross-bonded cable, the overvoltage propagates to the
other phases (b and c) through sheath cross-bonding.
Page 38
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
600
[kV] 1.67 pu
500
400
Phase a
300
200
100
Phases b and c
0
-100
-200
0.0 0.4 0.8 1.2 1.6 [ms] 2.0
500
[kV]
1.47 pu
400
phase a
300
200
100
phases b and c
0
-100
0.0 0.4 0.8 1.2 1.6 [ms] 2.0
200
[kV]
1.75 pu
160
120 phase a
80
40
phases b and c
0
-40
0.0 0.4 0.8 1.2 1.6 [ms] 2.0
Page 39
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Energization of one cable from a bus requires particular attention. The event is similar to back-to-back switching of
capacitor banks, which can produce high inrush currents. Figure 2-33 shows the CB1 current upon energizing the
first 108-km cable in Figure 2-2. CB2, CB3, CB4, and CB9 are all initially open. This cable has over 90% shunt
compensation, so that relatively large currents flow into the shunt reactors. Each pole of CB1 closes at an instant of
voltage peak to minimize DC offset, but the currents will still take a long time to decay.
Figure 2-34 shows the CB1 currents upon closing CB2, after the currents in Figure 2-33 have reached steady state.
Again, each pole of CB2 closes at the instant of voltage peak, which minimizes the DC offset current into the
second cable but maximizes the capacitive back-to-back current. The transient currents in CB2 are symmetrical to
the transient currents in CB1, due to their back-to-back configuration. The back-to-back currents reach higher peak
values and oscillate at a higher frequency than in Figure 2-33. The first half cycle of phase A current in Figure 2-34
shows the traveling wave nature of back-to-back cable energization currents, but the reactive energization currents
and source impedance effects are still significant. Circuit breakers may have peak inrush current, frequency, and/or
current-frequency product ratings to check. Both single energizations and back-to-back energizations should be
considered.
Figure 2-33: CB1 Currents During First Cable Energization in the Long-Cable Test System
Figure 2-34: CB1 Currents During Second Cable Energization in the Long-Cable Test
System
Page 40
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Attempt auto-reclosing only if the underground cable section is much shorter than the overhead line section.
When single phase auto-reclosing is applied to mixed overhead / cable lines, high degrees of shunt compensation
can cause open-phase resonances due to capacitive coupling between a disconnected phase and the energized
phases in the overhead line section [46].
This can be verified in Figure 2-35 where the voltage during the cable de-energization is shown. The shunt reactor
compensates 46 % of the reactive power of the cable, producing a resonance frequency of approximately 34 Hz. In
Figure 2-35 b) it takes a long time to completely damp the voltage in the cable due to low resistance. Typical cable
discharge times are several minutes with shunt compensation since Q factor of typical shunt reactor ranges around
500. Without shunt compensation, inductive VTs will quickly saturate and cable discharge will be completed in
several hundred milliseconds. A three-phase three-legged shunt reactor will have unequal mutual inductance
between phases that should be included in the model. These mutual inductances may produce different modulating
beat frequencies in the discharge voltage on each phase.
Figure 2-35: Receiving End Cable Voltage during Disconnection: a) Zoom and b) First 5 s
Page 41
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure 2-36: Single Line Diagram of a System with a Cable, Shunt Reactor, and Circuit
Breaker
I SR x IUC (2.9)
where ISR is the AC current component in the shunt reactor, ICB is the AC current component in the circuit breaker,
IUC is the AC current component in the cable and x is the reactive power compensation ratio.
If the reactive power compensation ratio is higher than 50 %, the DC component in the circuit breaker may be
higher than the AC component. In this situation the current does not cross zero during several cycles.
Figure 2-37 shows an example of zero-miss phenomenon for a 150 kV 50 km cable, with 70 % of the reactive
power being compensated by a shunt reactor installed at cable's end. The system is energized when the voltage is
zero and therefore the DC component is at its maximum value. The DC component may saturate the current
transformer effectively blinding the protection devices.
Figure 2-37 a) shows the currents in the circuit breaker (blue), in the shunt reactor (green) and in the cable (red).
The current in the circuit breaker does not cross zero. Figure 2-37 b) shows the current in the circuit breaker for a
period of 5 s, and it takes several seconds before the current crosses zero for the first time. The X/R ratio of a
shunt reactor is typically between 400 and 700, but may be up to 1350 [50] leading to a decay time constant of
4.3 s, neglecting source resistance. The time needed for the current in the circuit breaker to cross zero is
calculated by Equation (2.11), assuming X/R = 1350 on a 50-Hz system.
Zoom
Page 42
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Ls x 1 x 1
t ln t 4.3 ln (2.11)
Rs x x
Energize reactor and cable causes steady state overvoltage or undervoltage and
in sequence voltage step
Page 43
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
The zone of EPR could appear near an earthing electrode used for local grounding of the sheath when the earth
fault location coincides with the earthing location of the sheath. The sheath of a cable connected to the ground grid
of a substation can transfer the ground potential from an earth fault in the station to remote locations.
The countermeasure can preferably be applied at the source cable. For the protection of the environment an
effective solution is the two-point grounding of the cable sheath. This can also be strengthened with conductor laid
along the cable and connected to the sheath at the grounding locations. Local countermeasures such as improved
grounding or filters may also be applied at the influenced system. However, the most efficient countermeasure is to
increase the cross-section of the sheath of the inducing cable, because the inducing effect is then decreased due
to the following reasons:
The sheath - as a current return conductor - leads back an increased part of the return current, which
decreases the portion of the earth current in the zones close to the cable terminations (end effect zones)
and thereby decreases the EPR itself
The sheath - as a compensating conductor - leads an increased part of the return current also along the
stationary section (between the end zones) of the cable line. This is also applicable to the sheath of the
telecom- and signal cables.
Therefore, with a given amount of material the most effective compensation can be achieved when this additional
material is built into the cable as e.g. metallic screen composed of copper wires. This is always more effective than
a separate compensating wire.
Page 44
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
transformers
loads
network equivalents
surge arresters
circuit breakers
affects damping
Loss (magnetic small effect affects damping of oscillations
and copper) on damping of oscillations and time to
current zero
affects zero affects zero affects zero
Mutual (inductive) sequence sequence sequence
coupling impedance*1 impedance*1 impedance*1
*1: three-phase shunt reactor with a three-leg iron core
Page 45
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Magnetizing
impedance and affect amplitude
affect frequency
core non-linearity cause and frequency of
of transient
(knee-point and ferroresonance transient
components
air-core components
inductance)
Losses (magnetic affect damping affect damping
and copper) of overvoltages of overvoltages
affects zero affects zero
Mutual (inductive) sequence sequence
coupling impedance*1 impedance*1
*1: Three-legged three-phase transformer
Shunt reactors, transformers and generators are often modelled without frequency dependence due to
unavailability of precise models. If the precise model is available, the frequency dependent model is useful to yield
reasonable damping in a high frequency region. For a simplified model of transformer damping, an equivalent
network of parallel RL sections can be used to realise a constant X/R ratio at power frequency and above.
In system impact studies discussed in this chapter, assumed load levels have a significant impact on the
magnitude of overvoltages. As the low load conditions generally cause more severe overvoltages, special attention
is required if the cable line is used for the black start restorations. The common practice among utilities is to vary
load levels to find the impact, assuming constant impedance loads. The assumed load power factor and
representation (e.g. series RL vs. parallel RL) also affects the damping provided for transients. For the same real
power load, higher power factors will provide more damping. At minimum load, which is a conservative assumption,
the power factor is likely to be 0.98 or higher. A series RL model is conservative because it provides less damping
than other equivalent circuit load models.
Page 46
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
cross-bonded section. A third zone, with one busbar depth, uses lumped-parameter models. This reduces the
likelihood of erroneous high overvoltage at the expense of a relatively small increase in the computational effort. A
2-port or N-port boundary equivalent network represents the system outside these three zones.
Pi-models
FD-model
Detail Cross-bonded
models
Energised Cable
Page 47
Chapter 3. CABLE MODELLING FOR POWER FREQUENCY STUDIES
3.1 Introduction
In general, modelling of underground cables depends on the type of phenomenon to be considered. Several cable
models have been developed for system simulations. All available models can be broadly divided into two main
classes: lumped parameter models (alternatively referred to as PI-section models) and distributed parameter
models (alternatively referred to as travelling wave models which are covered in Chapter 4). A list of available cable
models for power frequency studies along with their characteristics is shown in Table 3-1.
Models with lumped parameters do not account for an infinite number of incremental cable segments but they
consider the total length of the transmission cable as a single section. Thus the parameters of the PI circuit result
from the multiplication of the cable constants (R’, L’ and C’ in ohms per unit length) by the total length of the
transmission cable. This model is usually referred in the literature as nominal PI model or short cable model and it
provides good steady state accuracy at power frequency. For steady state analysis of power systems, i.e. at
planning stages referred to in Chapter 2, the only frequency component of interest is the nominal power frequency
and it is not necessary to account for full frequency dependence of cable parameters. Therefore lumped parameter
models can be used with sufficient accuracy for studies such as power flow, short circuit, dynamics etc. The term
Nominal PI-model refers to a PI circuit with R, L, C elements calculated at power frequency (50/60 Hz) or at any
other frequency respectively. For relatively long transmission lines, a modification is made to R, L, C lumped
parameters to reduce the error in approximating distributed elements to lumped elements. If the parameters are
modified to compensate long line effects (distributed nature), the PI circuit is called Equivalent PI-Model or Long
Line Corrected PI-Model. An example of corrected pi model application is frequency scans which will be discussed
fully in Chapter 4.
A simplified way of compensating long line effects is to cascade a number of Nominal PI-Models longitudinally. In
such a case total values of line RLC values should be distributed proportionally between each Nominal PI-Model.
Cascading of Nominal PI-Models is always a less accurate line representation in comparison to a single Equivalent
PI-Model and it increases the simulation time depending on the number of cascaded Nominal PI-Models.
This chapter discusses the preparation and use of data for the calculation of parameters for lumped cable models
to be used for system planning studies. It leads the reader through available data and their treatment in the
calculation of long cable parameters for an accurate power frequency computer models based on IEC 60287-1-1
[55] and other references. For the purposes of model validation, it compares IEC 60287-1-1 calculations with
parameters from line constants functions found in commercially available software. It should be noted that a
reverse procedure may also be used where parameters are calculated with a software tool first and then checked
against the same parameters manually calculated with the standard, or other verification method such as
comparison against a test sheet. The process of parameter verification is very important in order to have
confidence in cable model parameters.
The techniques described in this brochure can be applied to all power cable types including solid dielectric type
(XLPE), mass impregnated type (MI), self-contained fluid filled type (SCFF), or high-pressure fluid filled (HPFF)
pipe type cables. However it should be noted that this technical brochure deals largely with long cable installations
with most of installations falling into the XLPE or MI type installations.
Table 3-1: Available Transmission Line Models for Power Frequency Studies
LUMPED PARAMETER MODEL
Equivalent PI-model
Aliases include Exact
R,L,C calculated at a fixed frequency, but
Equivalent PI-model and DC-kHz
corrected for long transmission lines [86][87]
Long Line Corrected PI-
model
Page 48
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Table 3-2: Indicative Cable Data Collected from Three Manufacturer Catalogues
for a 400 kV Cable
Cross- Diameter Inductance Outside Cable Current
Insulation Capacitance Surge weight Rating
section of the diameter
Manufacturer thickness [mH/km] impedance
of core core [μF/km] of cable
[mm] [Ω] [kg/m] [trefoil,A]
[mm2] [mm] trefoil [mm]
A 30.2 32 0.13 0.42 42.2 111 15.7 780
630 B 0.13 0.44 53 109 17 853
C 30 33 0.119 0.46 118 17 594
A 33.7 30 0.14 0.40 38.7 112 17.1 870
800 B 0.15 0.41 48 110 18 954
C 34.2 32 0.134 0.44 118 18 636
A 37.9 29 0.16 0.39 34.5 114 19 960
1000 B 0.17 0.38 45 112 21 1120
C 38.1 29 0.150 0.41 118 20 671
A 44 27 0.18 0.38 30.6 119 20.8 1115
1200 B 0.19 0.38 43 116 24 1209
C 42 27 0.171 0.55 121 22 1001
Details of screen or sheath thicknesses are not available in the catalogue but are important quantities to have for
accurate development of zero sequence impedances of cable systems. The zero sequence impedance is less
important in the planning stages. If necessary, one can use zero sequence data from similar projects and accept
some uncertainty until cable construction and installation details are known.
It should be noted that these are nominal values in dimensions. Cables are engineered for a particular application,
delivery method, installation method, warranty requirements, and manufacturer process. All of these elements may
have an impact on the final delivered product. The core DC resistance should be taken directly as a constant from
IEC 60228 [55] for conductors complying with this standard or otherwise, from the manufacture data sheet to be
provided. It is important to note that the data presented in Table 3-2 are for illustrative purposes only and does not
match the Example model cable developed in this chapter.
As the cable project moves to a more advanced stage a specification is developed to match the requirements of
the application. When more detailed information is available, complex computer models can be developed that can
be used for a broader range of study phenomena. The data from Table 2-2 is derived from various specification
and data sheets for a 400 kV cable, similar to that in Section 3.3. The data provided can be very detailed and is
typically supplied by a cable manufacturer as part of an order.
Page 49
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Page 50
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
R
ln b
Ra
mod_ins inu (3.1)
r
ln b
ra
where rb, ra are inner and outer radii of insulation. Ra is the outer radius of the core and Rb is the sum of the outer
radius of the insulator and thickness of semi-conducting layers.
1
Rs (3.2)
d2
g s .N . . w . cos
4
is the laying angle, N is the number of wires, gs is the wires’ conductivity and dw the wires’ diameter.
The inner and outer radii, respectively (r2) and (r3) of the equivalent sheath being those of the bundle of wires, the
cross-sectional area of the equivalent sheath is:
As' . r32 r22 with r3 r2 d w (3.3)
To get the same resistance, a corrected value of the conductivity has to be used:
d2
N . . w . cos
g s' g s . 4 (3.4)
As'
Page 51
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
The DC resistance per unit length of the conductor at its nominal operating temperature θ is given by:
The following table summarises the typical resistivity and temperature coefficients for different conductor materials
according to IEC 60028, IEC 60889 and IEC 60287-1-1.
When the sheath consists also a metal (but not ferromagnetic) tape wounded around the cable core then the
current in the tape follows the tape. This means that the resistance of the tape shall be calculated according to the
actual length of the tape (Ltape),not according to the length of the cable (L), using the following relationship.
Ltape D
1 ( )2 (3.6)
L L
where D is the diameter of the cable core around which the metal tape is wound.
This correction is not needed for laminated foil, which is not wound.
Page 52
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
(measurement) can be used to obtain the internal impedances of the armouring as function of the frequency and
the pre-magnetization current of the fundamental frequency. The present EMT type program cable models neglect
the ferromagnetic effects.
The earth return resistance increases when the cable is near the surface of earth, and decreases as cable burial
depth increases, and finally at large depths the earth resistance reaches a constant value. The earth return
reactance decreases with increasing depth and eventually reaches a constant value. At relatively low values of
earth resistivity, the earth return resistance increases and then decreases with high values of earth resistivity. The
earth return reactance increases with earth resistivity.
For a simple three-phase cable system (having only three-conductors and no sheaths), the zero sequence
impedance increases with increasing earth resistivity as expected. However, in the presence of sheath, the
magnitude of zero sequence impedance may be decreased with earth resistivity due to the effect of parallel sheath
impedance in the zero sequence circuit. This can be explained using a per phase equivalent circuit of the 3-phase
cable system as shown in Figure 3-3.
The impedances Zp, Zq and Zr can be derived based on basic impedance components in a cable and it can be
shown that Zp and Zq are independent of earth resistivity. However, Zr consists of both self and mutual impedances
between cables. The resulting zero sequence impedance is
ZqZr
Z0 Z p (3.7)
Zq Zr
Since reactance of Zr increases significantly (compared to resistance) with increasing earth resistivity, the
magnitude of Z0 will reduce with increasing earth resistivity.
Page 53
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
If the core is composed of segmented conductors, the resistivity of the core material is increased to take care of the
isolated segments [57]. Hence the effective resistivity is,
r2
eff (3.8)
Ac
where, ρeff is the effective resistivity of the conductor with radius r and Ac is the nominal cross sectional area of the
core. However if the manufacturer directly provides the resistance per unit length (Rdc), then the resistivity is
calculated as,
The conductor AC resistance is essential in cable impedance sequence modelling for system studies, and is
normally not given either in data sheets or manufacturer catalogues. The AC resistance is calculated from the
conductor DC resistance, utilizing IEC 60287-1-1, which gives empirical correction factors to deal with these special
conductor types.
Table 3-4 shows the Rac/Rdc ratio of cores with large cross-sections and different construction types as calculated by
the cable constants [61] and IEC 60287-1-1. While the influence of core stranding does not seem to be pronounced,
the results for round stranded, hollow and helical stranded conductors are in good agreement. The reader can see
that the error in the AC resistance of segmental conductors is excessively high. At higher frequencies the error
becomes even higher. Therefore, it can be concluded that calculations based on cable constants overestimate the
AC resistance of the conductor resulting in an overestimation of the losses and the damping of the transmission
cable. In the case of segmental (Milliken) conductors, manufacturers normally rely on IEC 60287-1-1 for the
calculation of the AC conductor resistance.
Table 3-4: Rac/Rdc Ratio of Conductors of Different Materials and Construction Types under
Consideration of Skin and Proximity Effect
Cross- Rac/Rdc @ 50 Hz Rac/Rdc @ 50 Hz
Material, Type of core
Section IEC 60287-1 Cable Constants
630 mm2 AL, Round stranded 1.051 1.052
1200 mm2 AL, Hollow, helical stranded 1.103 1.104
1200 mm2 AL, Round, 4 segments 1.012 1.144
630 mm2 CU, Round stranded 1.127 1.131
1200 mm2 CU, Hollow, helical stranded 1.236 1.250
1200 mm2 CU, Round, segmental 1.076 1.327
Cable constants programs that rely on a general formulation with solid conductors, typically allow the user to adjust
conductor resistivity in the conducting mediums to approximate this affect.
2
De (3.10)
Page 54
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
In this equation is the angular frequency, is the permeability and is the resistivity of the material. Lower
penetration depth is said to increase the skin effect by restricting current flow to a smaller cross-section. Since the
conductor metal’s conductivity varies with temperature the skin effect is temperature dependent. Higher
temperature increases and the penetration depth. The formulas in IEC standard 60287-1-1 chapter 2.1.2 show
this dependence in another manner.
The temperature dependence can be illustrated with the plot (Figure 3-4) of the skin effect factor of segmented
conductors versus temperature. It can be observed that the skin effect decreases when temperature increases.
1.50
1.45
1.40
1.35
1.30
Rac / Rdc
1.10
1.05
1.00
20 30 40 50 60 70 80 90
CONDUCTOR TEMPERATURE ( °C )
Page 55
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
z sinh l
Z Z C sinh l sinh l zl (3.11)
y l zy
l
tanh
Y 1 cosh l 1 1 l y l yl 2
tanh tanh (3.12)
2 Z C sinh l ZC 2 z 2 2 l zy
2
where z and y are the per unit length impedance and admittance of the line respectively and l is the length of the
line.
z
ZC Z (3.13)
y
zy (3.14)
and
l yl
tanh (3.15)
2 2
and therefore Z z l Z ' zl and Y y l Y ' yl suggests that the PI representation with lumped parameters are
accurate enough. Table 3-5 illustrates cable parameters as calculated in Section 3.5 with modified valued based on
the long line transmission correction for a 36 km cable at nominal frequency. These values will be used for the
basis of validation in Section 3.6.
Page 56
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1. In all cases, it is a best practice to compare and verify results of one method against the other, or from a
measured data sheet. For this chapter, one has calculated the expected results from the IEC standard, and then
repeated the calculation in a software tool with the same parameters to verify the results.
All calculations in this section of the document conform to IEC 60287-1-1, IEC Technical Report 60909-2, CIGRE
Technical Brochures 283 and 347, and the underground transmission system Reference Book by EPRI and are all
referenced where appropriate.
Electrical parameters of cables published in manufacturers’ datasheet are typically calculated according to the
formulation given in IEC 60287-1-1 [55] applicable to steady-state operation of AC cables, where the calculated
parameters are valid at power frequency only. The standard assumes a continuous, constant current operation of
the cable (100% loading) to produce the maximum conductor temperature under stable thermal conditions. The
Technical Brochure 531 “Cable Systems Electrical Characteristics” [8] fully describes these formulas, use, and
application for different types of cable systems and is an excellent reference for the calculation of cable parameters.
IEC 60287-1-1 takes the following into account:
Data related to the construction characteristics of the cable: layout (parallel single core cables and pipe-
type cables), geometry and properties of the conducting and insulating materials (conducting and insulating
layers).
Data related to the surrounding conditions (in air, buried directly underground, in ducts, troughs or in steel
pipes).
Skin and proximity effects for the calculation of AC resistance of conductors.
Dielectric losses, screen and armour losses, losses in steel pipes, etc.
The following sections provide a detailed example in calculating cable parameters based on available cable data
from manufacturer’s specification. The example utilises the 400 kV cable data given in Table 2-2 in flat formation at
a burial depth of 1.5 m and a 500 mm cable spacing between phases, and goes on to calculate the lumped
parameters shown in Figure 3-5 according to IEC 60287-1-1.
All the equations are fully explained in [8].
Rdc = 6.896×10-6 [Ω/m] at 20 °C. As the highest loss occurs at operating temperature 90 °C, the DC resistance used
is
Page 57
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
R dc 90 R dc 20 (1 ( 20 )) (3.17)
It should be noted that the example shown above is based on 90°C but depending on the manufacturer
specifications, utility practices and geographical location other temperatures may be used.
The formulas for ys and yp are given in reference [55]. In calculating the AC resistance, it is assumed that the
conductor has a solid copper core. Hence the kp and ks factors are both unity (to calculate ys and yp) [62]. ys and yp
are described below and earth resistivity is assumed 100 Ω.m.
Re (3.19)
8
Re = 4.9348×10-5 [Ω/m]
earth
De 659 [m] (3.20)
f
Page 58
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
f Frequency [Hz] 50
De = 931.9667 [m]
De
Xa ln (3.21)
2 GMR a
The geometric mean radius for a solid conductor can be shown to be equal to re-1/4 where r is the radius of the
conductor.
Xa = 6.6928×10-4 [Ω/m]
Z a R e R ac jX a (3.22)
The mutual impedance between two adjacent conductors in the flat configuration is given by
D e
Z c Re j ln (3.23)
2 S
Page 59
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
D e
Z l Re j ln (3.24)
2 2 S
The average mutual impedance (Zm,ave) between cables for transposed flat formulation,
2Z c Z l
Z m ,ave (3.25)
3
Since the length of the cable is 36 km, the total positive sequence impedance is,
De
Xs ln (3.27)
2 rs
Z m R e jX s (3.28)
Zm = 4.9348×10-5+j6.0711×10-4 [Ω/m]
Page 60
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
1
Rs (3.29)
g sc Ash
It is assumed that sheath consists of aluminium tape and the sheath resistivity is corrected using
sc90 sc20 (1 ( 20)) (where 4.03 10 3 (Table 1 in [55]) and 90 o C ). The maximum operating
temperature for sheath and core may be different depending on various factors [55]. However in planning stage,
usually the maximum operating temperature of both conductor and sheath is assumed with a typical value such as
90 °C.
Rs = 4.8862×10-5 [Ω/m]
Self impedance of the sheath is given by,
Z e R e R s jX s (3.30)
In the average cross-bonded system, the sheaths of the three cables are transposed and connected to ground
continuously through the total length of the cable system. Assuming average cross-bonded cable, the zero-
sequence impedance is given by,
Z m 2 Z m,ave
Z zero Z a 2 Z m, ave ( Z m 2 Z m ,ave ) (3.31)
Z e 2 Z m,ave
Typically zero sequence impedance is greater than the positive sequence impedance on normal overhead
transmission lines. However for underground or submarine cables, as shown in this example, positive sequence
impedance is often greater than zero sequence impedance. It has to be noted that the grounding resistances are
neglected here.
Page 61
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
2 r 0
C [F/m] (3.32)
D
ln
d
Table 3-6: Comparison of Calculated Power Frequency Cable Parameters (Total Length)
with Average Cross-bonding
Manual Manual Software A Software A
Calculation Calculation (detailed/explicit (average cross- Software Software
(expected) (long line cross-bonding) bonding) B C
corrected)
Rpos [Ω] 0.5055 0.5024 0.4973 0.4995 0.5130 0.5036
Xpos [Ω] 7.5832 7.5700 7.311 7.4353 7.4795 7.4250
B [mS] 2.43 2.40 2.40 2.437 2.45 2.437
Rzero [Ω] 2.257 2.2496 2.2454 2.2513 2.2731 2.2561
Xzero [Ω] 2.2936 2.2870 2.2146 2.1377 2.1354 2.1225
The differences between parameters from software solver and the calculations can be explained. In calculation,
simplified formulas with approximations are used. For example, the depth of cable and actual cross-bonded
configuration are not considered in the manual calculation. However in simulation, the cable is modelled with actual
details such as 1.5 km cable segments with grounding resistances (5 Ω) at each major section. The sheath has an
additional 0.02 mH series inductance.
A check against the anticipated result of the manual calculation provides a form of verification and thus greater
confidence of study results. The results above show that the developed cable model, which includes input
parameters, has been verified at power frequency with reasonable agreement between manual calculations and a
software line constants solver.
Page 62
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
0.5055 Ω 7.5832j Ω
1.215 mS 1.215 mS
2.257 Ω 2.2936j Ω
1.215 mS 1.215 mS
Most simulation software will take these parameters as presented, or they may require modification or data
conditioning depending on the input requirements of your simulation environment.
Page 63
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
4.1 Introduction
Simulation studies of transients in cables are of great importance when analysing the behaviour of cables in power
systems. Transients in power systems can cause steep voltage wave fronts (travelling waves) that contain high
frequency components resulting in overvoltages at bus connections or reflections at open ends. For modelling a
cable for transient simulations, the frequency dependence of cable parameters must be taken into account. Precise
simulations of transients require a detailed cable model.
The major difficulty in modelling underground cables in power system electrical simulation software is that they are
frequency dependent, mainly due to the frequency dependency of conductors (skin effect), as well as the ground or
earth return path. In order to accurately represent a frequency-dependent transmission line when simulating with
time domain EMT-type solvers, the frequency dependent transfer function must be convolved into their equivalent
time-domain characteristics. The techniques required for this convolution are quite complex, and are not a part of
this technical brochure. This chapter is to guide the reader in appropriate selection and use of various cable models
for transient phenomena, and only provides references to cable models that have been developed by various
researchers. No derivations of cable model implementations are described.
Frequency dependent parameters need to be calculated with consideration to the construction of the cable, the
surrounding conditions, the skin effects, the dielectric losses, etc. It is therefore necessary to consider the
parameters in the frequency domain.
Many computer models utilise simplifications in their calculations, for example, assuming constant permeability,
neglecting dielectric losses and proximity effects, and assuming coaxial arrangements of the conducting and
insulation layers. Furthermore, an accurate model for the earth impedance is quite complicated and requires
specific data typically not available. Note that parameter values provided on data sheets by manufacturers are
usually only measured at power frequency and are useful for model benchmarks.
Checking of output parameters calculated by software over a wide frequency range is an important consideration
when determining the suitability of a model for the intended application domain. A table of suggested computer
models for specific applications is provided. Details on the mathematical formulations are provided in Annex 1.
A fundamental description of impedances and admittances has been documented in [61]. These formulations have
been widely used for calculating cable parameters for transient simulations and the model has been verified for
frequencies up to 1 MHz. The frequency range of interest for transient analysis is usually below this frequency.
Page 64
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Table 4-1: Available Transmission Line Cable Models for Transient Studies
A number of terms are used that require some explanation to fully understand the implementation details of the
various models:
- Modal domain. The modal domain is a set of equations used to describe key features of a phenomena
based on physical characteristics of transmission paths. This is very similar in treatment to sequence
components commonly used in power systems theory. For a three phase overhead transmission line,
there are only three modes and they are equivalent to the sequence component definitions. For
underground or submarine cable systems, the modes are defined by the number of conducting elements
including cores and sheaths or screens.
- Curve fitting is another technology used in transmission model construction, and is applied to define an
approximated transfer function to represent a frequency response, or a time domain response.
- Convolution: the theoretical definition is the folding or blending of one function with another. However, in
transmission line construction, convolution is used to transfer a frequency domain response into an
equivalent time domain response.
- Phase domain. In the context of transmission line modelling, phase domain refers to the mathematical
treatment of defining frequency dependent transmission line transfer functions without the use of a modal
transformation, i.e. the transfer function equivalents are realised directly on a per conductor basis or phase
domain basis.
Page 65
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
the order of the curve-fitted transfer function. The Bergeron model computational burden has a linear relationship to
the number of conductors.
⋯
y (4.1)
⋯
The application of rational functions has many advantages. For example, rational functions offer a generalised way
to represent curves plus the flexibility to fit various shapes. One of the limitations of this model is that it uses a
constant transformation matrix to convert from modal domain to phase domain.
Page 66
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
150
Mode 1
Mode 2
100
Mode 3
Mode 4-6
50
0
0 500 1000 1500 2000
Frequency [Hz]
Figure 4-2: Frequency Dependence of Modal Propagation Velocities for the Long-cable
Example
Page 67
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
25
Mode 1
15 Mode 2
Mode 3
Mode 4
10 Mode 5
Mode 6
0
0 500 1000 1500 2000
Frequency [Hz]
Figure 4-3: Frequency Dependence of Modal Characteristic Impedances (Real Part) for the
Long-cable Example
The magnetic field produced by the core current cannot be shielded by the metallic sheath if the transient
frequencies contained in the overvoltages are lower than the characteristic frequency fs calculated by (4.2) [69]. On
the other hand, the magnetic field produced by the core current is completely shielded by the metallic sheath
(coaxial model) if the transient frequencies contained in the overvoltages are higher than the critical frequency fc
calculated by (4.3) [70].
s3/ 4
fs (4.2)
10 s ds
s
fc (4.3)
s d 2
where s : resistivity of metallic sheath [Ωm]
The characteristic frequency and critical frequency are calculated as 55 Hz and 1798 Hz, respectively, for the
example long-cable. It is recommended to select the target frequency at which cable impedances are calculated,
with the following considerations:
- Since the frequency of interest is between the characteristic frequency and the critical frequency in most
analyses of many or long cable systems, the target frequency should be selected between the two
frequencies.
- The target frequency should be selected at the lower end of the two frequencies in order to produce
conservative results with lower damping.
- The cable impedances and other related values are in the transition near the two frequencies. Hence the
target frequency should have some offset from the two frequencies.
Page 68
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
The target frequency 200 Hz, as proposed above, can be selected as a result of these considerations. When
selecting a target frequency which is different from the power frequency, one should confirm that it does not lead to
an unrealistic steady-state condition.
Page 69
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure 4-4 shows the results of an energization overvoltage analysis with different cable models in different EMT-
type simulation software packages. The cable from Sub B to Sub C in the long-cable example of Chapter 2 is
energized from a voltage source behind the source impedance of 1 Ω and 30 mH. The energization occurs at the
voltage peak of phase a, and the voltage (phase a) at the open terminal is monitored.
From the results, the Bergeron model evaluated at 50 Hz produces a similar overvoltage to the Gustavsen model in
Softwares B and C during the first 20 ms after the energization. In finding the maximum overvoltage, the choice of
cable models between the Bergeron model and FD models does not have a noticeable impact. On the other hand,
the overvoltages after the first 20 ms following energization have notable differences, which require the use of FD
models when matching simulation results to field measurements.
600
400
200
Voltage [kV]
-200
-400
-600
-800
0 0.01 0.02 0.03 0.04 0.05 0.06
Time [s]
Page 70
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Table 4-3 shows guideline to select proper fitting parameters for frequency-dependent models.
Page 71
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Page 72
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Since cross-bonding affects cable impedance and propagation velocity, it is highly recommended to model sheath
cross-bonding for switching studies. As the computational burden required by Bergeron models is relatively small,
all cross-bonding points can be modelled without a noticeable effect on computational efficiency. With some
frequency dependent models, however, one may wish to reduce cross-bonding points in order to reduce
computational burden and possible numerical stability issues. The reduction, to some extent, can be made without
a significant impact on the simulation results [74][75].
Bonding of cable sheath is an important factor for the cable models, also when assessing the sheath overvoltage in
transient studies. The sheath overvoltage is generally higher in single point bonded cables.
Page 73
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
maximum tolerances) and the delivered cable may be different. This will have an effect on the accuracy of the
cable models and care must be taken in the use of cable parameter data.
A sensitivity analysis in this section highlights which parameters of the cable are more significant or have more
impact to the accuracy of the cable model for different frequency ranges. This helps simulation software users to
focus more attention to the accuracy of those parameters for various power system studies.
The cable data is same as that of example cable discussed in chapter 2 (Table 2-2) and the cable configuration
used for this analysis is shown in Figure 4-5. The trefoil formation is selected, instead of the flat formation, since
sequence impedances are defined on a balanced three-phase circuit.
In order to determine the effect of cable input parameters for different studies, current and voltage waveforms from
time domain simulations involving typical power system studies can be used. However, it is difficult to conclude the
sensitivity of a particular parameter for the entire frequency range for each type of transients. For example, the
switching transients typically contain frequencies up to 10 kHz, but a particular time domain switching study may
generate transients only up to 1 kHz. A better approach therefore will be to determine the sensitivity of the input
cable parameter in frequency domain using frequency domain transmission line functions.
Many power system studies such as short circuit and fault analysis, dynamic simulations with fault/unbalanced load
shedding etc. involve the analysis using sequence components (positive and zero sequence impedances etc.). So
it is useful for the simulation engineer to know the impact of cable input parameters on sequence components. The
switching studies and lightning studies may not involve direct sequence components analysis and the simulation
depends on phase domain quantities such as phase impedances (relates to phase voltages and currents). Since
sequence impedances are components of phase impedances, a high impact of a parameter on either positive or
zero sequence can cause significant impact on such a study.
In this sensitivity analysis, the sensitivity of a parameter was determined in frequency domain using sequence
impedances and admittances of the cable. This requires a derivation of the relationship between such functions
and the parameter itself. When one parameter is changed (e.g. conductor radius), other parameters are assumed
constant (e.g. thickness of the other layers such as main insulation, sheath, semi-conducting layer etc.)
For a 10% change in the parameter (except for earth resistivity and semi-conducting layer where a 50% change
was applied), the change in the cable model sequence impedances was observed for different frequency ranges,
and checked for any significant impact. It is assumed that the sensitivity of input parameters is classified as high if it
is above 10% for a particular frequency range. This means that a small error in the input parameter may cause a
relatively high voltage or current magnitude change, when exciting the cable with a frequency belonging to that
frequency range.
The cable model sensitivity analysis has identified a set of most sensitive parameters that include conductor radius,
permittivity of inner insulator layer and thickness of inner insulator among others. The following tables summarise
the results with three sensitivity ranges, below 5%, 5% - 10% and above 10%.
Page 74
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
-3
x 10 Resistance of Positive Sequence Impedance [Ohms]
2.5
Default
2 Conductor radius
Conductor Resistivity
1.5 Sheath Resistivity
Earth resistivity
Thickness of the inner insulator layer
1
Thickness of the sheath
0.5
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Difference as a percentage
20
15
10
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Frequency [Hz]
Page 75
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
0.2
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Difference as a percentage
7
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Frequency [Hz]
Page 76
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
-3
x 10 Resistance of Zero Sequence Impedance [Ohms]
2.5
Default
2
Conductor radius
Conductor Resistivity
1.5 Sheath Resistivity
Earth resistivity
1 Thickness of the inner insulator layer
Thickness of the sheath
0.5
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Difference as a percentage
20
15
10
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Frequency [Hz]
Page 77
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Default
0.8 Conductor radius
Conductor Resistivity
0.6 Sheath Resistivity
Earth resistivity
0.4 Thickness of the inner insulator layer
Thickness of the sheath
0.2
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Difference as a percentage
14
12
10
0
-3 -2 -1 0 1 2 3 4 5 6
10 10 10 10 10 10 10 10 10 10
Frequency [Hz]
Page 78
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Parameter Sensitivity
The sensitivity study results presented above suggest that accuracy of the conductor radius is relatively important
for DC or low frequency studies and less important for other studies. The accuracy of conductor resistivity may
have relatively large effect on DC or low frequency studies. The sheath resistivity may have medium effect on
power and switching frequency studies. The accuracy of the inner insulation layer thickness has a medium effect
on all studies except DC or low frequency studies, while the thickness of the sheath has a medium effect on power
and switching frequency studies. The earth resistivity and thickness of inner semi-conducting layer may not cause
considerable impact on all studies (see Section 3.43. for further discussion on effect of earth). The permittivity
value shows a linear relationship with capacitance (as expected from the theory) and the effect is relatively high.
The conductor radius and thickness of inner insulation layer have a medium effect on the capacitance.
It must be underlined that the above observations are based on the relative importance of input parameters on
sequence impedances and admittances for this particular case, and actual effects on voltage and current values
obtained when performing simulations can vary form study to study. For example, Figure 4-10 shows the sensitivity
of the same parameters on cable energization overvoltages. The cable from Sub B to Sub C in the long-cable
example of Chapter 2 is energized from a voltage source behind the source impedance of 1 Ω and 30 mH. The
energization occurs at the voltage peak of phase a, and the voltage (phase a) at the open terminal is monitored.
From the results, it is clear that 10 % or 50 % deviations in conductor, sheath, or earth resistivity do not have
noticeable impact on the energization overvoltages. The deviations in resistivity can affect the damping of the
overvoltages in theory, but the effect cannot be observed in the simulation time 60 ms. The other two parameters
have rather clear impact on the energization overvoltages. The increase of conductor radius leads to lower
reactance and higher capacitance. As the effect on capacitance is more significant than that on reactance, the
results show lower propagation velocity due to the increase of conductor radius. In contrast, the increase in the
thickness of the inner insulator layer leads to lower capacitance, which results in higher propagation velocity as
shown in Figure 4-10. The simulation results demonstrate the importance of modelling the capacitance accurately
for cable energization studies.
Page 79
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
800
600
400
200 Base
Voltage [kV]
Conductor radius
0
Conductor resistivity
-200 Sheath resistivity
-400
-600
-800
0 0.01 0.02 0.03 0.04 0.05 0.06
Time [s]
800
600
400 Base
200
Voltage [kV]
Earth resistivity
0
Thickness of the
-200 inner insulator layer
Thickness of the
-400 sheath
-600
-800
0 0.01 0.02 0.03 0.04 0.05 0.06
Time [s]
Page 80
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
per meter. Second, the positive sequence (or line mode) surge impedance and wave velocity are determined from
(4.4) and (4.5).
(4.4)
1 1
2 (4.5)
Where f is the system power frequency, Xpos is in Ω, and B is in siemens. Xpos and B can both be for the total cable
length, as in Table 3-6, in which case the units of Vpos are cable lengths per second. If Xpos and B are both per meter,
the units of Vpos are meters per second. For Xpos = 7.5832 and B = 0.00243, we have Zpos = 55.86 Ω and Vpos =
2314.31 * 36e3 = 8.332e7 m/s. The surge impedance is about one order of magnitude lower than for an OHL, and
the wave velocity is about fourth the speed of light, both within the range of typical values for UGC. Equations (4.4)
and (4.5) may be used for zero sequence (or ground mode) Bergeron parameters, and to cross-check steady-state
and Bergeron model parameters for consistency. For example, if the derived Bergeron parameters are outside of
typical ranges, it could mean that the planning-model parameters included a series current-limiting reactor, or
described two cables in parallel.
Since it is often difficult to obtain transient measurement data, future challenges should include the development of
a method to validate transient cable models without using transient measurement data. This section introduces one
promising method that can be applied to any time-domain model described in Section 4.2.
The typical frequency domain program is described in this paragraph. The current and voltage relationship at a
terminal of a transmission line can be expressed in terms of transmission line (frequency dependent) parameters
such as series impedance and shunt admittance. If a simple linear network (possibly with current or voltage
sources) is connected to transmission line terminals, (frequency domain) transfer functions for voltages and
currents at terminals can be derived by combining equations for the transmission lines and the external network.
Using techniques such as Numerical Laplace Transform, it is possible to convert the frequency domain transfer
function into a time domain waveform (e.g. for sending-end current).
Indeed, most cable constant programs that generate data for use in time domain simulation use frequency domain
equations to calculate the cable parameters. Typically, the cable being simulated in the time domain would be
cross-bonded and may have additional resistive terminations to ground along its length. Depending on the
Page 81
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
sophistication of the frequency dependent program, it may be possible to validate this entire cable. However, if
such a program is not available, it is still possible to write a simple program (using commercial software) to
calculate the various frequency dependent parameters such as series impedance, shunt admittance as functions of
the cable geometry and material properties such as resistivity of the conductors and ground, relative permittivity etc.
(see Annex 1, frequency dependent phase domain models). Some electromagnetic transients programs may also
directly provide this information.
In the simplest validation approach, a smaller cable segment (in between cross-bonding points) can be considered.
It can be excited on the sending end by a simple voltage or current source, and terminated at the receiving end
with suitable impedance (including open or short circuit). Using the frequency dependent program mentioned in the
previous paragraph, the phase impedances of this cable as seen by the source at any given frequency can be
calculated in a straightforward manner. The magnitudes and phases of these impedances can be plotted or
tabulated as a function of frequency. The same cable segment can then be setup in the EMT program with the
identical terminations as in the frequency dependent analysis, and excited with a sending end source with a given
frequency. Simulation can be conducted till steady state is reached after which the voltages and currents in each
phase are measured from which the impedances can be calculated. The process is repeated with a different
source frequency to yield the impedances at several different frequencies. The impedances are then compared
with the frequency domain solution with a view to validate the EMT model. In order to save validation effort, the
comparisons may be conducted at a handful of frequencies where more accuracy is desired.
A comprehensive validation is possible if a more sophisticated frequency domain formulation is developed, that
allows cable segments to be cascaded. In such cases the cross-bonding and other intermediate terminations can
be represented in the frequency dependent program. The process discussed in the above paragraph can then be
applied to entire cable system rather than to a smaller segment, thereby increasing the confidence in the EMT
model. In order to assess the accuracy of the time domain simulation directly, a time-domain response can be
calculated in the frequency dependent program by using Numerical Laplace Transform techniques [83]. This
resultant waveform can then be used to validate the corresponding EMT simulation response of the full cable.
As an example, consider a 36 km cable (24 segments with 1.5 km each). Each segment is cross-bonded with the
next and there is a 5 Ω resistance to ground after every three segments. The full model is simulated on an EMT
program and also for validation on a frequency domain program with FFT capability. The cable is energized with a
DC voltage step as shown in Figure 4-11. Terminating and source impedances are 1 Ω resistors as shown in the
figure. Figure 4-12 shows the waveforms indicating a good agreement, thereby validating the EMT model.
Page 82
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Main : Graphs
Sending Current (EMT) Sending Current ( FreqDomain Calc)
400
350
300
250
200
150
100
50
0
-50
Sending Voltage (EMT) Sending Voltage ( FreqDomain Calc)
1.2k
1.0k
0.8k
0.6k
0.4k
0.2k
0.0
Figure 4-12: Comparison of Sending End (phase a) Current and Voltage Waveforms for
Step Excitation Obtained from EMT Simulation and FD Model
Figure 4-13 shows the arrangement used for energization with a fundamental frequency (50 Hz) ac source. The
terminations are with 1 Ω resistors at the sending end and essentially open circuit (100 kΩ) on all phase and
sheath conductors at the receiving end. The cable breaker (BRK) is switched on at the peak of the ac voltage
waveform. The corresponding sending end current and voltage waveforms obtained from EMT as well as
frequency domain simulation are shown in Figure 4-14. The agreement is very good, once again confirming the
validity of the EMT model of the cable.
Figure 4-13: Cable Model under Validation (near-open circuit termination on receiving
end)
Page 83
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Main : Graphs
Sending Current (EMT) Sending Current ( FreqDomain Calc)
40
20
-20
-40
Figure 4-14: Comparison of Sending End (phase a) Current and Voltage Waveforms for
Sinusoidal Excitation with Receiving End Open Circuit Obtained from EMT Simulation and
FD Model
Page 84
Chapter 5. Conclusions and Recommendations
This Technical Brochure has looked at the system technical performance issues associated with the installation of
long AC cables. One of the first tasks has been to identify a definition for a long cable that would be acceptable for
the many stakeholders; thus, a long cable has been defined as the one that normally requires shunt reactor
compensation and / or other components installed with the cable.
Studies associated with any cable installation project have been divided into four separate stages; system planning
studies, system impact studies, equipment design studies and forensic studies. Likely issues that may be looked at
for each study stage have been summarised to help identify system technical performance related phenomena.
A weighting factor has been attached to each study type in order to emphasise the importance of the study due to
likely consequences. It is possible that different weighting factors may be more applicable for different power
networks due to some unique peculiarities and characteristics and therefore the given factors should only be taken
as indicative but nevertheless applicable in vast majority of cases. Relevant frequency range and phase along with
appropriate analysis tools have also been drawn for each particular study type.
Two example systems have been presented; one to demonstrate technical performance issues associated with
long cables and one to show issues associated with the presence of many cables. Many cable example test
system has been based on the IEEE 24-bus reliability test system by replacing overhead lines with cables in order
to illustrate issues associated with planning and system impact studies. A new test system has been developed to
show issues associated with long cables.
Although similar in principle to overhead lines, particular attention was given to reactive compensation studies with
illustrative calculations shown in a step by step manner in order to emphasise the differences between overhead
lines and cables.
It has been shown that replacing existing overhead lines with cables will shift parallel resonances to lower
frequencies. This has the potential to exacerbate the existing harmonic voltage distortion on power systems.
Considering that many systems have seen an upward trend in background harmonic distortion levels, it has been
suggested that a complete evaluation at different system loading levels with many system contingencies should be
considered.
Synchronous generator self-excitation with possible uncontrolled voltage rise at generator terminals based on
reported incidents has been highlighted to be another possible issue due to excessive reactive power generated by
cables connecting a power plant to the transmission grid. Connection of generators in parallel, thus reducing their
equivalent reactance or introduction of shunt reactors to the transmission circuit and thereby increasing the
external reactance seen by the generator have been suggested as possible mitigation measures to avoid generator
self-excitation.
Various temporary overvoltage (TOV) phenomena such as those due to ground fault back feed, generator self-
excitation, system islanding and resonance have been shown. Since temporary overvoltages are experienced in a
relatively wide area, the evaluation is often performed against the insulation performance of the weakest equipment
in the area. Therefore particular attention has been drawn to surge arresters due to their vulnerability to TOV.
Although tested for TOV capability under a stiff power frequency voltage source, this capability should not be used
as a withstand envelope. This is because TOVs shown in example cases occur at higher frequencies and with
variable magnitudes. As arresters may not necessarily withstand a voltage waveform that is under their TOV
capability curve it is particularly important to consult equipment vendors/manufacturers in such cases in order or
ascertain capability against calculated values. It is also important that the calculation of TOVs consider various
system contingencies, loading levels and other parameter sensitivities as these affect the resonant frequency and
the damping rate of TOVs.
Example mitigation measures to control TOV especially in the case of ground fault have been drawn. These
measures include limiting cable lengths to avoid resonances at or below 3rd harmonic, increasing arrester voltage
ratings and reducing system voltage under light loading conditions.
For design stage studies, energization of cables has been classified as particularly important as severe TOVs could
be established and therefore assessment to ascertain adequate insulation level is critical. It has been suggested
that unless absolute certainty exists that there can be no re-energization, studies should be run on a statistical
simulation basis to assess the maximum overvoltage with trapped charge. The insulation performance of cables is
normally tested by application of lightning and switching impulse voltages in combination with a power frequency
voltage. IEC 62067 defines the test methods and requirements for XLPE cables and accessories for rated voltages
of 150 kV and up to 500 kV. For example, for a 400 kV cable, a 440 kV power frequency voltage is applied for 15
minutes following a lightning impulse test. Considering the voltage level that can be attained by TOVs, the level of
Page 85
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
applied power frequency voltage appears to be somehow low even though the duration is sufficient. Therefore if a
match between possible TOVs and the applied test voltage is preferred one possibility that may be considered by
the relevant Study Committee on cables could be to shorten the duration of the test while raising the applied
voltage level. In the meantime, it is possible to utilise a theoretical conversion of insulation performance to a shorter
duration but with a higher voltage level using voltage-time relationship of insulation characteristics.
It has been shown that transient frequencies involved in cable energization are lower than for typical overhead lines.
This is more so in cross-bonded cables which leads to lower damping of overvoltages. Cross-bonding has been
shown to be the factor for overvoltages to propagate to other phases through the sheath. It has been concluded
that in the case of cable energization through a cable fed busbar particular attention should be given to high inrush
currents and checking capacitive switching capabilities of breaker as it is analogous to back to back capacitor
energization.
Zero-miss phenomenon characterized by a current not crossing zero for several cycles following energization of a
cable has been identified as another potentially dangerous issue in that if there was a fault on the cable it may be
impossible to open healthy phase poles of a circuit breaker without due damage. Various mitigation measures have
been identified and presented with some explanation.
Inductive coupling to other metallic structures has been suggested as another area of study due to being a possible
safety hazard by introduction of touch voltages as well as contributing to corrosion.
The effect of other peripheral equipment on system impact and equipment design studies has been tabled along
with appropriate size of network model for these studies. In most cases, peripheral power system component
models often lack frequency dependency due to unavailability of data and precise models. However, if available,
inclusion of frequency dependent models is useful as it usually provides reasonable damping in the frequency
region of interest.
Cable modelling has been covered extensively under two main headings: lumped parameter models for use in
steady-state power system analysis and distributed parameter models for use in electromagnetic transient analysis.
Manual calculation of sequence parameters for a lumped model has been illustrated extensively. Results have
been compared against a number of line constant solvers of commercially available software packages with very
good agreement. This in itself forms a good verification tools in terms of the developed cable model that can be
used in studies requiring sequence models.
Reader’s attention has been drawn to some practical consideration while attempting to calculate cable parameters.
It is recommended that semi-conducting layer is treated as part of the insulation by appropriate adjustment of the
permittivity of the insulation. Likewise, due to difficulties in modelling wires that make up the screening of a cable it
has been recommended that screen wires are treated as sheath with the application of approximation equations.
Other practical considerations such as the decrease of skin effect factor with increasing conductor temperature,
and the effect of earth on zero sequence impedance have also been illustrated.
The importance of AC resistance in cable impedance sequence modelling for system studies cannot be ignored. It
has been shown that use of line constant solvers does overestimate AC resistance especially in the case of
segmented or as more commonly referred to Milliken conductors. It has been concluded that the introduced error is
not so pronounced in the case of round stranded, hollow and helical stranded conductors.
As high frequency is intrinsic to most power system transient phenomena, relevant power system component
modelling includes some form of frequency dependence. Cables are no exception to this and the difficulty lies in
the convolution of the frequency dependent transfer function from frequency domain into the equivalent time
domain. Various models that have been developed to account for this are given in a short form without going into
the complexity of their derivation.
A correlation has been drawn between the various issues identified for power system technical performance related
phenomena and the minimum cable model requirement. Bergeron model has been recommended as a minimum in
most cases owing to the fact that it produces relatively conservative results with solid numerical stability and lower
computational burden and hence being the first choice of most engineers.
An alternative method in the selection of target frequency when employing Bergeron model has been presented in
order to minimise the conservative results produced by the use of power frequency for Bergeron models. Choice of
target frequency selection criteria has been drawn based on the calculated characteristic and critical frequencies. It
has been shown that Bergeron model with target frequency set according to the given methodology produces
sufficiently accurate results in the frequency range of interest with errors in the region of 5% on the modal
propagation velocities and modal characteristic impedances. Hence, the use of Bergeron models with this type of
approach is suggested for many transient modelling studies identified in this brochure. A comparative study to
calculate the maximum overvoltage during a cable energization has been included to illustrate the validity of this
Page 86
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
recommendation. However, it must be emphasised that although the usage of Bergeron model is sufficient for
establishing the maximum overvoltage, it has its limitations in producing accurate representation of subsequent
overvoltages when inspected over a range of timeline in a simulation study. Therefore, in cases where higher
accuracy is required, such as in forensic studies where there is usually a need to match measured data with
simulation, frequency dependent models should be used whenever possible.
Cable models in all existing commercial power system analysis tools are based on the formulation of impedance
and admittance matrices for coaxial cables and this trend has been followed in this Technical Brochure. However,
important references were included for those more willing practitioners to attempt modelling of non-standard
configurations by use of either finite element analysis or by employing some form of approximation to fit an arbitrary
shaped conductor into an equivalent cylindrical form.
Various installation and environmental factors that should be taken into account during modelling has been drawn.
For switching studies, modelling of cross-bonding arrangements is highly recommended due to the effect on the
cable impedance and propagation velocity. Same applies during the assessment of sheath overvoltages in
transient studies. Taking into account the environment where cables are installed is equally important. This ranges
from modelling the electrical parameters of a tunnel where the cables are laid to simplifications to reduce a tunnel
into a pipe model. The importance of earth return path was highlighted due to its significant effect on cable
parameters.
A comprehensive sensitivity analysis with respect to the cable model input parameters was performed to establish
their effect on the positive and zero sequence resistance and reactance as well as admittance. This analysis
identified a set of most sensitive parameters that include conductor radius, permittivity of inner insulator layer and
thickness of inner insulator among others. Tables and graphs showing the effects in detail have been included in
Chapter 4 along with the relevant percentage change they introduce.
The best way to verify models is to compare calculated values with field measurements. For verification, a
measurement set up has to be made. For validation of high frequency models, a step function should be supplied
at the cable termination. These measurements should be carried out when the cable is out of service. Measuring
when the cable connection is still under construction is by far the best opportunity for this. Model validation for
higher frequencies is more cumbersome in practice than for power frequency.
Page 87
Bibliography/References
[1] CIGRE Technical Brochure 379, Update of Service Experience of HV Underground and Cable Systems, WG
B1.10, April 2009.
[2] CIGRE Technical Brochure 110, Comparison of High Voltage Overhead Lines and Underground Cables,
JWG 21/22.01, December 1996.
[3] CIGRE Technical Brochure 189, Insulation Coordination for HVAC Underground Cable System, JWG 21/33,
June 2001.
[4] CIGRE Technical Brochure 250, General Guidelines for the Integration of a New Underground Cable System
in the Network, WG B1. 19, August 2004.
[5] CIGRE Technical Brochure 268, Transient Voltages Affecting Long Cables, WG B1.05, April 2005.
[6] CIGRE Technical Brochure 283, Special Bonding of High Voltage Power Cables, WG B1.18, October 2005.
[7] CIGRE Technical Brochure 347, Earth Potential Rises in Specially Bonded Screen Systems, TF B1.26, June
2008.
[8] CIGRE Technical Brochure 531, Cable Systems Electrical Characteristics, WG B1.30, April 2013.
[9] IEC Technical Report 60071-4, Computational Guide to Insulation Co-ordination and Modelling of Electrical
Networks, June 2004.
[10] R. Benato, A. Paolucci: EHV AC Undergrounding Electrical Power. Performance and Planning, SPRINGER,
Series: Power Systems, ISBN: 978-1-84882-866-7, 2010.
[11] R. Benato, A. Paolucci: Operating Capability of Long AC EHV Transmission Cables, Electric Power Systems
Research, Vol. 75/1, luglio 2005, pp. 17-27.
[12] R. Benato, A. Paolucci: "Operating capability of ac EHV mixed lines with overhead and cables links", Electric
Power System Research, Vol. 78/4, aprile 2008, pp. 584-594.
[13] IEEE Working Group Report, “Protective Relaying Considerations for Transmission Lines with High Voltage
AC Cables”, IEEE Trans. on Power Delivery, vol. 12, no. 1, pp. 83-96, January 1997.
[14] J. B. Lee, C. W. Ha, C. H. Jung, “Development of Digital Distance Relaying Algorithm in Combined
Transmission Lines with Underground Cables”, Proc. IEEE PES Summer Meeting, pp. 611-616, 15-19 July
2001.
[15] B. Kasztenny, I. Voloh, J. G. Hubertus, “Applying Distance Protection to Cable Circuits”, 57th Annual
Conference for Protective Relay Engineers, pp. 46-69, 30 March – 1 April 2004.
[16] A. Ametani, C. T. Wan, “Sheath Overvoltages due to Faults on an EHV Cable”, Presented at IPST 1995 in
Lisbon, Portugal.
[17] R. Benato: "Multiconductor Analysis of Underground Power Transmission Systems: EHV AC Cables",
Electric Power System Research, Vol. 79, Issue 1, gennaio 2009, pp. 27-38.
[18] IEEE Task Force Report, "IEEE Reliability Test System", IEEE Trans. on Power Apparatus and Systems, vol.
98, no. 6, pp. 2047-2054, Nov/Dec 1979.
[19] IEEE Task Force Report, “The IEEE Reliability Test System – 1996”, IEEE Trans. on Power Systems, vol. 14,
no. 3, pp. 1010-1020, August 1999.
[20] CIGRE Technical Brochure 30, Reactive Compensation Analyses and Planning Procedure, TF 03.01 of
Committee 38 (Power System Analysis and Technique). 1989.
[21] R. Benato, S. Lauria, F.M. Gatta, L. Colla, M. Rebolini, F. Renaud: "Steady-State And Transient Ehv Ac
Cable Shunt Reactive Compensation Assessment", Proc. of CIGRÉ 2010, Paper C4-109, 2010.
[22] Transmission Line Reference Book 200kV and Above, third edition, EPRI. Palo Alto, USA.2006, 1011974.
[23] M. Val Escudero, M. Redfern. “Effects of Transmission Line Construction on Resonance in Shunt
Compensated EHV Lines”, Presented at IPST 2005 in Montreal, Canada, Paper No IPST05-109.
[24] F. Iliceto, E. Cinieri, A. Di Vita, “Overvoltages due to Open-Phase Occurrence in Reactor Compensated EHV
Lines”, IEEE Trans. On PAS., Vol PAS-103, no. 3, March 1984. pp. 474-482.
[25] S.Lauria, F.M.Gatta, L.Colla, “Shunt compensation of EHV cables and mixed overhead-cable lines” , Proc.
2007 IEEE Lausanne Power Tech Conf., paper n. 562.
[26] R. A. Turner, K. S. Smith, "Resonance Excited by Transformer Inrush Current in Inter-connected Offshore
Power Systems", Industry Applications Society Annual Meeting, 2008.
[27] Surya Santoso, Roger C. Dugan, Thomas E. Grebe, Peter Nedwick, "Modeling Ferroresonance Phenomena
in an Underground Distribution System", International Conference on Power System Transients (IPST), 2001.
Page 88
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
[28] J. R. Marti, A. C: Soudack, "Ferroresonance in power systems: Fundamental solutions", IEE Proceeding-C.,
Vol. 138, No. 4, July 1991.
[29] V. Valverde, A. J. Mazón, I. Zamora, G. Buigues, "Ferroresonance in Voltage Transformers: Analysis and
Simulations", International Conference on renewable Energies and Power Quality, Sevilla, 2007.
[30] IEEE Std. C62.11, IEEE Standard for Metal-Oxide Surge Arresters for AC Power Circuits (> 1 kV).
[31] IEC Std. 60099-4, Surge arresters – Part 4: Metal-oxide surge arresters without gaps for a.c. systems
[32] IEC Std. 60099-5, Surge arresters – Part 5: Selection and application recommendations.
[33] Reliability and Operating Committee (ROC) Final Report, 20 December 2004, available at:
http://www.ct.gov/csc/cwp/view.asp?a=3&q=287674
[34] J. H. R. Enslin, J, Knijp, C. P. J. Jansen, J. H. Schuld, “Impact of Reactive Power Compensation Equipment
on the Harmonic Impedance of High Voltage Networks”, IEEE Bologna Power Tech Conference, June 23-26,
2003, Bologna, Italy.
[35] J. H. R. Enslin, R. A. Wakefield, Y. Hu, S. Eric, “Harmonic Impedance Study for Southwest Connecticut
Phase II Alternatives”, October 18, 2004, available at
http://www.ct.gov/csc/lib/csc/docket_272/kema_harmonic_analysis_report.pdf
[36] A. Ketabi, A. M. Ranjbar, R. Feuillet, “Analysis and Control of Temporary Overvoltages for Automated
Restoration Planning”, IEEE Trans. on Power Delivery, vol. 17, no. 4, pp. 1121-1127, October 2004.
[37] T. Karasaki, T. Goto, A. Ametani, ”An Abnormal Overvoltage due to Load Rejection on EHV Underground
Transmission Lines”, International Conference on Power System Transients (IPST), September 1995.
[38] N. Momose, H. Suzuki, S. Tsuchiya, T. Watanabe, “Planning and Development of 500 kV Underground
Transmission System in Tokyo Metropolitan Area”, CIGRE Session 37-202, 1998.
[39] V. Akhmatov, B. C. Gellert, T. E. McDermott, W. Wiechowski, “Risk of Temporary Over-Voltage and High-
Voltage Fault-Ride-Through of Large Wind Power Plants”, 9th International Workshop on Large-Scale
Integration of Wind Power into Power Systems as well as on Transmission Networks for Offshore Wind
Farms, October 18-19, 2010, Quebec City, PQ..
[40] T. Petersson, “Reactive Power Compensation”. ABB Power Systems. Information NR 500-028E. December
1993.
[41] W. Diesendorf, “Insulation Co-Ordination in High Voltage Electric Power Systems”. Butterworths, 1974. ISBN
0 408 70464 0.
[42] CIGRE Technical Brochure 119, Interaction between HVDC Convertors and nearby Synchronous Machines,
WG 14.05. October 1997.
[43] X. Youfang, “The Practical Criterion of Generator Self Excitation and its Field Example Analysis”,
TENCON’93 Proceedings. Conference on Computer, Communication, Control and Power Engineering, IEEE
Region 10 Conference. DOI 10.1109/TENCON.1993.320572.
[44] C. Concordia, Synchronous Machines, J. Wiley, New York, 1951.
[45] E. W. Kimbark, Power System Stability, J. Wiley, New York, 1948.
[46] F. Iliceto, E, Cinieri, A. Di Vita, “Overvoltages Due to Open-Phase Occurrence in Reactor-Compensated EHV
Lines”, IEEE Trans. on Power Apparatus & Systems, vol. 103, no. 3, pp. 474-482, March 1984.
[47] B. Kulicke, H. H. Schramm, “Clearance of Short-circuits with Delayed Current Zeros in the Itaipu 550 kV-
Substation”, IEEE Trans. on Power Apparatus and Systems, vol. PAS-99, no. 4, July/Aug 1980.
[48] T. Michigami, S. Imai, O. Takahashi, “Theoretical Background for Zero-miss Phenomenon in the Cable
Network and Field Measurements”, IEEJ General Meeting 1459, 1997 (in Japanese).
[49] H. Hamada, Y. Nakata, T. Maekawa, “Measurement of Delayed Current Zeros Phenomena in 500 kV Cable
System”, IEEJ General Meeting 2001 (in Japanese).
[50] K. Søgaard, C. L. Bak, "Analyse og modellering af dynamiske forhold for 400kV linjen FERSLEV-TRIGE",
Institute of Energy Technology, Aalborg University, June 2005 (In Danish???).
[51] ITU-T (International Telecommunication Union, Telecommunication Standardization Sector), Directives
Concerning the Protection of Telecommunication Lines against Harmful Effects from Electric Power and
Electrified Railway Lines; Volumes II and III.
[52] CIGRE Joint Working Group C4.207, Guide for Assessment of Transferred EPR on Telecommunication
Systems due to Faults in A.C. Power Systems: Propagation of EPR (Earth Potential Rise) by metallic
conduction, characteristics of used system configuration and precautions for telecommunication equipment.
Under preparation.
Page 89
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
[53] CIGRE/CIRED Technical Brochure 291, AC Corrosion on Metallic Pipelines due to Interference from AC
Power Lines, Phenomenon, Modelling and Countermeasures, JWG C4.202, April 2006.
[54] A. Ametani, "Wave Propagation Characteristics of Cables", IEEE Transactions on Power Apparatus and
Systems, Vol. PAS-99, No. 2, March/April, 1980.
[55] IEC 60228, “Conductors of insulated cables”, International Standard, Third Edition 2004-11.
[56] IEC 60287-1-1, “Electrical cables- Calculation of the current rating. Part 1: current rating equations and
calculation of losses”, International Standard, First Edition 1994-12.
[57] Bjørn Gustavsen “Panel Session on Data for Modeling System Transients. Insulated Cables”, Proc. IEEE.
Power Engineering Soc. Winter Meeting , 2001.
[58] P. Graneau, “Underground power transmission”, New York : John Wiley & sons, 1979, ISBN 0471057576
[59] M.M. Kashani,”Frequency Dependent Impedance of Stranded Conductors using Subdivision method”, PhD
Thesis, University of British Colombia, Canada, 1990.
[60] CIGRE Technical Brochure 250 Technical and environmental issues regarding the integration of a new HV
underground cable system in the network, 2004.
[61] A. Ametani, “A general formulation of impedance and admittance of cables”, IEEE Transactions on Power
Apparatus and Systems, vol. PAS-99, no. 3, pp. 902-910, May/June 1980.
[62] CIGRE Technical Brochure 272, Large cross-sections and composite screens design, WG B1.03, June 2005.
[63] H. W. Dommel, EMTP Theory Book, 2nd ed. Vancouver, British Columbia Canada: Microtran Power System
Analysis Corporation, Vancouver, British Columbia, 1996.
[64] A. Semlyen and A. Dabuleanu, "Fast and accurate switching transient calculations on transmission lines with
ground return using recursive convolutions", IEEE Transactions on Power Apparatus and Systems, Vol. PAS-
94(2), p. 561-571, 1975.
[65] J. Marti, Accurate Modeling of Frequency Dependent Transmission Lines in Electromagnetic Transients
Simulation, IEEE Transactions on Power Apparatus and Systems, PAS-101, #1, pp. 147-155, Jan. 1982.
[66] T. Noda, N. Nagaoka, A. Ametani, “Phase Domain Modelling of Frequency Dependent Transmission lines by
means of ARMA model”, IEEE transactions on power delivery, Vol. 11, No. 1, January 1996.
[67] L. Marti, “Simulation of transients in underground cables with frequency dependent modal transformation
matrices”, IEEE transactions on Power delivery, Vol. 03, No 03, July 1998.
[68] A. Morched, B. Gustavsen, M. Tartibi, A Universal Line Model for Accurate Calculation of Electromagnetic
Transients on Overhead Lines and Cables, Paper PE-112-PWRD-0-11-1997.
[69] A. Ametani, Distributed-Parameter Circuit Theory, Corona Pub. Co., 1990 (in Japanese).
[70] A. Ametani, “A study of cable transient calculations, Part II Transient calculations”, Science and Engineering
Review of Doshisha University, vol. 24, no. 2, pp. 119-127, July 1983.
[71] C. H. Chien, R. Bucknall, “Harmonic Calculations of Proximity Effect on Impedance Characteristics in Subsea
Power Transmission Cables”, IEEE Trans. on Power Delivery, vol. 24, no. 4, October 2009.
[72] B. Gustavsen, A. Bruaset, J. Bremnes, A. Hassel, ”A Finite-Element Approach for Calculating Electrical
Parameters of Umbilical Cables”, IEEE Trans. On Power Delivery, vol. 24, no. 4, October 2009.
[73] A. Ametani, Ikuko Fuse, “Approximate Method for Calculating the Impedance of Multicounductors with Cross
Sections of Arbitrary Shapes”, Electrical Engineering in Japan, vol. 112, no. 2, 1992.
[74] N. Nagaoka, A. Ametani, “Transient Calculations on Crossbonded Cables”, IEEE Trans. on Power Apparatus
and Systems, vol. PAS-102, no. 4, April 1983.
[75] A. Ametani, Y. Miyamoto, N. Nagaoka, “An Investigation of a Wave Propagation Characteristic on a
Crossbonded Cable”, IEEJ Transaction B, vol. 123, no. 3, 2003 (in Japanese).
[76] N. Nagaoka, K. Yoshida, T. Karasaki, “Transient Sheath Voltage Characteristics of a Crossbonded Cable
Installed within a Tunnel”, IEEJ vol. 112, no. 5, 1992.
[77] L.Wedepohl and D.Wilcox, “Transient analysis of underground power-transmission systems,” IEE
Proceedings, vol. 120, no. 2, pp. 253–260, February 1973.
[78] A. Ametani, “Stratified Effects on Wave Propagation”, IEEE Trans. PAS, col. 93 (5), pp. 1233-1239, 1974.
[79] A. Ametani, “On the Impedance and the Admittance in the EMTP Cable Constants/Parameters”, European
EMTP User Group Meeting, October 2009.
[80] A. Ametani, T. Yoneda, Y. Baba, N. Nagaoka, “An Investigation of Earth-Return Impedance between
Overhead and Underground Conductors and Its Approximation”, IEEE Trans. on Electromagnetic
Compatibility, vol. 51, no. 3, August 2009.
Page 90
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
[81] B. Gustavsen, “A study of overvoltages in high voltage cables with emphasis on sheath overvoltages”,
Trondheim, Norway: PhD. Thesis, NTH, 1986.
[82] A. Ametani, C. T. Wan, “Sheath Overvoltages due to Faults on an EHV Cable”, International Conference on
Power System Transients (IPST), September 1995.
[83] Y. Goto, A. Ametani, T. Kubo, N. Nagaoka, T. Baba, “A Surge Analysis of a Cable System Composed of
Submarine and Underground / Overhead Cables”, UPEC P58-7, September 2009.
[84] A. Ametani, “Problems and Countermeasures of Cable Transient Simulations”, EMTP Journal, vol. 5, April
2000 (in Japanese).
[85] S. Nishiwaki, T. Koshizuka, O. Hosokawa, K. Yamamoto, “Comparison between Measurement and ATP
Computation on Traveling Wave Propagation Characteristics of Long Power Cable”, International Conference
on Power Systems Transients (IPST 2009), available on the web
(http://www.ipst.org/techpapers/2009/papers/25.pdf).
[86] W. I. Bowman and J. M. McNamee, “Development of Equivalent Pi and T Matrix Circuits for Long
Untransposed Transmission Lines”, IEEE Trans. On Power Apparatus and Systems, Vol. 83, Issue 6, pp.
625-632, June 1964.
[87] William D Stevenson Jr., “Elements of power system analysis”, fourth edition, McGraw-Hill, 1982.
[88] L. Colla, M. Rebolini, F. Iliceto, “400 kV AC new submarine cable links between Sicily and the Italian
mainland. Outline of project and special electrical studies”, CIGRE Session C4-116, 2008.
[89] IEEE PES Task Force on Data for Modeling System Transients of IEEE PES Working Group on Modeling
and Analysis of System Transients Using Digital Simulation (General Systems Subcommittee), “Parameter
Determination for Modeling System Transients—Part II: Insulated Cables”, IEEE Trans. on Power Delivery,
vol. 20, no. 3, JULY 2005.
[90] Juan A. Martinez-Velasco, ed., “Power System Transients: Parameter Determination”, CRC Press, 2009.
[91] L. Colla, S. Lauria, F. Palone “Finite Sections Modeling of Power Cable Systems”,International Conference on
Power Systems Transients (IPST2011), paper 139, Delft, Netherlands, June 14-17, 2011
Page 91
Annex 1. CABLE MODEL IMPLEMENTATIONS
Annex 1 describes available computer modelling employed in the typical power system engineering analysis
software:
PI models
Method of finite sections
Bergeron travelling wave models
Frequency dependent mode models
Frequency dependent phase domain models
PI models
Page 92
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
art of the time, the method used passive elements to construct a complex network representing a number of
coaxial cables. Finite sections modelling represents the integration of accurate conductor modelling utilizing
transmission pi-section modelling. Figure A1-2 below shows such an example. The single frequency branches of a
pi-section model are extended to synthesised multi-frequency equivalents.
The addition of earth return complicates the nodal matrix representation of the coaxial cables. For effective use, a
method to transform cable parameters is required to generate the nodal matrix. This method may be the only fully
passive and accurate EMT means of modelling wave propagation in non-linear resistive and/or inductive
conductors. This method has been implemented in an EMT-type software [88].
The disadvantage of this modelling technique is that the nodal matrix expands dramatically impacting simulation
numerical efficiency due to the N2 solution factor of nodal methods [42].
Figure A1-3: Equivalent Circuit of a Differential Length dx Cable Elements for One Cable,
without Mutual Coupling to Other Cables
The Bergeron model is a Norton equivalent model based on travelling wave theory [A1]. This model is often used
when modelling longer power transmission networks and represents the cables inductance and capacitance by its
Page 93
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Norton equivalent with lumped resistance representing the cables distributed series resistance, where the Norton
equivalent is lossless and the resistance represents losses in the transmission line.
The model is usually set up so that the lumped resistance is represented by dividing the line length into only two
sections, where each end has 1/4 part of the resistance and the centre contains 1/2 part. This is shown in Figure 4-
1 in Chapter 4.
The sending end currents in Figure A1-3 can be calculated by using voltage division as shown in equation (1.4)
[A4]. Receiving end currents are obtained similarly.
v1 (t )
i12 (t ) I1 (t / 2)
ZC R / 4
(A1.2)
v (t / 2) Z C R / 4
I1 (t / 2) 2 I 2 (t / 2)
ZC R / 4 ZC R / 4
This type of cable modelling can be found in most EMT simulation software. In multi-conductor case, there are
several assumptions (e.g. using constant transformation matrix) in transforming phase quantities in to modal
quantities and hence the model approximately represents the fundamental frequency.
These models as outlined in references [65], [67], [A6] are based on modal theory in which the coupled
transmission line equations are decoupled using transformation matrices. As an example, consider the following
system of equations:
Y=GX (A1.3)
Using suitable transformation matrices, the above can be decoupled as shown below. Each decoupled mode can
be treated as a single phase circuit in the solution.
YM =GM XM (A1.4)
In general the transformation matrix from direct phase domain to modal domain is frequency dependent and
complex. However for most overhead transmission lines (e.g. horizontally symmetrical overhead lines) or simple
cable systems this transformation matrix is almost constant and also the imaginary part of the transformation matrix
is negligible compared with the real part. For such cases, the transformation matrices can be considered to be real
and constant with little error in the time domain solution. The assumption of the real constant transformation
matrices leads to efficient time domain modelling, since less number of convolution are required to solve for a given
multi-conductor transmission line.
Page 94
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
where all parameters are a function of frequency (ω), Isend/Vsend is the sending end current/voltage, Irec/Vrec is
the receiving end current/voltage,
(A1.6)
References
[A1] Watson, E.F. & Arrillaga, J, 2003. Power systems electromagnetic transient simulation. London: The
Institution of Electrical Engineers. ISBN: 0 85296 106 5.
[A2] Tleis, N.D, 2008. Power Systems Modelling and Fault Analysis Theory and Practice. Oxford: Newnes. ISBN:
978-0-7506-8074-5.
[A3] R. Meredith, “EMTP Modeling of Electromagnetic Transients in Multi-Mode Coaxial Cables by Finite
Sections”, IEEE Transactions on Power Delivery, Vol. 12, No. 1, January 1997.
[A4] Dommel, H.W: Digital Computer Solution of Electromagnetic Transients in Single- and Multiphase Networks,
IEEE Transactions on Power Apparaturs and systems, Vol. PAS-88, No. 4, April 1969, pp. 388-399.
[A5] Nguyen, T.T, "Earth-return path impedances of underground cables. I. Numerical integration of infinite
integrals", IEEE Proceedings on Generation, Transmission and Distribution, Vol. 145, Issue: 6, pp 621 – 626,
November 1998.
[A6] A. Ametani, “A highly efficient method for calculating transmission line transients”, IEEE transactions on PAS,
PAS-95, No 05, Sept/Oct 1976.
[A7] T.-C. Yu and J. R. Martí, “A robust phase-coordinates frequency-dependent underground cable model
(zcable) for the emtp,” IEEE Transactions on Power Delivery, vol. 18, no. 1, pp. 189–194, January 2003.
[A8] B. G. A. Morched and M. Tartibi, “A universal model for accurate calculation of electromagnetic transients on
overhead lines and underground cables,” IEEE Transactions on Power Delivery, vol. 14, no. 3, pp. 1032–
1038, July 1999.
Page 95
Annex 2. MEASUREMENT OF CABLE ELECTRICAL PARAMETERS
This Annex summarises how the cable vendors determine the electrical parameters given on data sheets. It is
important to understand this process when verifying cable model parameters against the data sheets. In some
cases, it’s helpful to duplicate the vendor test method in an EMT program when verifying consistent results.
Both the DC resistance and cable capacitance can be measured directly from the cable either at the factory or
once installed in the field. Each manufacturer may have a different procedure and it is important to understand
their method when reviewing the data as no standard measurement method exists. For a DC resistive or capacitor
bridge measurement, test sets are used to measure the cable parameters directly.
The conductor resistances of circuit A-B, B-C and C-A are measured, then the resistance must be corrected to
20°C. By averaging the measurements, the reported resistance per phase per unit distance is calculated .
Positive, negative, and zero sequence parameters for the above test report are calculated as follows:
For positive sequence values, balanced three phase voltage is applied to all three phase cables with the far end
bonded. For every phase conductor, voltage (phase to ground), current (I), active (P) and reactive power (Q),
power factor and frequency are measured.
Im Z pos Q I 2
(A2.1)
Re Z pos P I 2
The imaginary component of Zpos is modified to the desired frequency. The impedance Zpos is then adjusted for the
maximum operating temperature 90 °C utilizing known formulas [55]. The per-phase impedance Zpos (Ω/m) is then
calculated by averaging the values for all three phases.
The calculation of the zero sequence parameters on the above test report are based on the test setup shown in
figure A2-2. The voltage between phase and earth (V), current (IAB), active power (P), power factor, reactive
power (Q), and frequency are all measured.
Page 96
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Im Z pos 3Q I AB
2
(A2.2)
Re Z pos 3P I AB
2
Using the above formulas, the imaginary part of Zzero is modified to the desired frequency. The impedance Zzero is
adjusted for the maximum operating temperature 90 °C. Then, the per-phase impedance Zzero(Ω/m) is then
calculated as follows.
Re Z zero Im Z zero
2 2
Z zero (A2.3)
The theoretical calculations used by the manufacturers are based on the assumption that all current will return
through metallic screens. However, this is not true for installed cables and must be accounted for.
Page 97
Annex 3. MEASUREMENTS ON THE 100 km LONG 150 kV AC CABLE
CONNECTION OF THE HORNS REV 2 OFFSHORE WIND FARM
A new 100 km long 150 kV AC cable line was installed in Denmark in 2009. This cable line is a radial connection
from an offshore windfarm to an onshore substation in the Danish transmission network. There was a unique
opportunity to perform detailed impulse measurements during the installation process, for model verification
purposes, and to perform full scale switching measurements. Both measurements were performed in relation to on-
going PhD projects. The purpose, performance, outcome and use of the measurements are described in detail in
[A9] and [A10].
The measurement method described here, can be adapted for model verification of other systems. Furthermore,
the results obtained can be used for validation of new modelling methods.
A schematic diagram of the cable connection for the Horns Rev 2 wind farm is shown in Figure A3-1. The main
components installed directly at the onshore substation called Endrup are also illustrated.
Figure A3-1: Diagram of the Cable Connection between the Onshore Substation in the
Danish Transmission Network and the Horns Rev 2 Wind Farm
The cable line is divided into three sections; a long land cable section, a short land cable section and a submarine
cable. The land cable sections are of the same cable type. The submarine and the land cable are connected with a
direct buried joint, such that the whole cable line is operated as a single 100 km long cable line.
Land cable
The land cable line consists of three 150 kV single core 1200 mm2 XLPE cables. The cables are laid in a tight
trefoil with the bottom cables at 1.3 m depth, see Figure A3-2.
Page 98
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure A3-2: Cross Sectional Layout of the 150 kV Land Cable System
A picture of the cable is shown in Figure A3-3. This picture shows all cable layers: core, core screen, insulation,
insulation screen, metal screen strands, metalic screen laminate and outer covering.
Page 99
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Submarine cable
The submarine cable is a three core armoured cable. The cross section of the submarine cable is shown in Figure
A3-4.
When modelling these specific cables, corrections of core resistivity and insulation permittivity must be included,
according to Section 3.4. When using the obtained results for validation of new modelling methods, the below
parameter corrections should be included.
Core resistivity
The core conductor is compact stranded. It is therefore necessary to correct the conductor resistivity, as explained
in Section 3.4.4. For the Horns Rev 2 land cable, the conductor resistivity is 3.19 x 10-8 Ω.m (A3.1).
∙ 20.75
2.8264 10 3.19 10 (A3.1)
1200
Page 100
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Insulation permittivity
The semiconductive layers are considered as being part of the insulation and the permittivity is corrected as
explained in Section 3.4.1.
The corrected relative permittivity is given in (A3.2). The insulation thickness is increased from 17 mm to 20.1 mm.
40.85
2.5 20.75 2.95 (A3.2)
39.50
22.25
Screen
The cable screen has a first layer of copper wires with a 95 mm2 cross-section and a second 0.8 mm thick
aluminium foil used as a water barrier. The diameter of the copper wires was measured with a micro meter
measuring tool and it was equal to 1.11 mm.
The wires are spaced, see Figure A3-3. Consequently, the cable screen cannot be modelled as a solid layer.
Applying the same method as used to correct conductor resistivity. The correction of the copper wires' resistivity is
as done in (A3.3).
, ,
The simulation models used in A3.1-Measurements for Model Verification have both screen layers modelled
separately in order to improve the results of the intersheath mode.
Such high level of accuracy is not necessary in A3.2-Full Scale Measurements and the two layers are simplified
into one by using (A3.4).
(A3.4)
8.98 10 Ωm
Submarine Cable
The correction of the submarine cable layers is made using the same approach used for the land cable. The
corrected conductor resistivity is given by (A3.5), and the corrected insulation permittivity by (A3.6).
Page 101
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
15.25 (A3.5)
1.724 10
630
1.999 10 Ωm
37.75
2.5 15.25 3.105 (A3.6)
34.75
16.75
For model verification purposes, the field measurements were performed in 3 steps; measurements on a single
minor section, measurements on a single major section, measurements on multiple major sections (the entire land
part of the cable line).
Figure A3-5: Simulated Modal Velocity for the Single Minor Section of the Measured 150
kV Cable System
Page 102
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Because of the low velocity of the ground mode, only the coaxial and intersheath modes are tested in the field.
For the coaxial mode, the current flows in the core conductor and fully returns in the screen of the same phase and
for one of the intersheath modes, the current flows in one screen and fully returns in the screen of an adjacent
cable.
Therefore, for high frequencies, the impedance matrix for coaxial and intersheath modes can be simplified as
shown in (A3.7) and (A3.8) respectively.
(A3.7)
(A3.8)
where denotes an impedance of cable 1, Zgmij denotes the mutual earth impedance between cables i and j. VCSi
and VSGi are cable i conductor-screen and screen-ground loop voltages respectively and ICS1, ISG2 and ISG3 are the
conductor-screen loop current of cable 1 and screen-ground loop currents of cables 2 and 3 respectively. It should
be noted, that ISG3 = −ISG2.
For the intersheath mode, the earth return impedance therefore has an impact on the results. And by exciting only
the coaxial and intersheath modes of propagation, it is possible to evaluate the accuracy for separate impedance
parts of the full cable impedance. The cables are identical and in a tight trefoil, and therefore the different cable
impedances for each phase are assumed identical.
The purpose of the measurements is to obtain a base for validation of simulation models. The measurements
should validate the models for fast transient behaviour. According to IEC 60071-2 an impulse of 1.2x50 μs can be
used to simulate lightning overvoltages. The field measurements therefore use an impulse test to validate the cable
performance and model calculations for the cable terminal conditions. Using such an impulse as an input to the
cable, and measuring voltage/current in both ends of the cable gives results, where each reflection can clearly be
seen. This way, it is possible to validate modelled traveling time and the simulated cable parameters.
When performing simulations, one should use correct time step. Furthermore, during measurements, the sample
time should be sufficient. The single section of the underground 150 kV cable for the offshore HR2 wind farm is
1780 m long. The travelling time, assuming the speed of light, for 1780 m is approximately 5.94 μs. The front time
of the impulse, 1.2 μs, is therefore the dominating shortest time, and the time step and sample time should be no
larger than 0.12 μs. There should at least be 10 samples during the front time, in order to have an acceptable
resolution. The sample time should therefore be no less than 120 ns, which implies sampling frequency no less
than 8.33 MHz. The sampling frequency used for the measurements was 10 MHz.
The power source used for the impulse test is a 1.2/50 µs surge tester with a maximum output voltage of 7 kV.
For excitation of the coaxial mode, a 1.4 kV impulse is used and for excitation of the intersheath mode a 2 kV
impulse between the two sheaths is used. These impulses require a sampling frequency range of approx. 8 MHz
for measurements. For this short cable length, with an open end (and no cross-bounding point) the sending end
voltage will approximately double at the receiving end. In order not to risk any damage of the screen or measuring
equipment, the sending end voltage is limited to only 1,4 kV and 2 kV (between the sheaths). Normally the screen
is protected with a 5 kV surge voltage limiter.
The impulse generator was connected to the HV cable by use of a 5m long 300 mm2 Cu wire. The inductance of
such a wire is estimated as 1 μH per meter. The wire is therefore represented by a 5 μH inductance connected in
series with the surge generator. The two measurement setups are shown in Figure A3-6 and Figure A3-7. In Figure
Page 103
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
A3-6, the sending end voltage and current as well as the receiving end voltage, is measured for the excited phase.
In Figure A3-7, the sending end voltage and current for the excited sheaths, all core voltages as well as the
receiving end voltage for the excited sheaths are measured.
Figure A3-6: Excitation of the Coaxial Mode for a Single Minor Section
Figure A3-7: Excitation of the Inter-sheath Mode for a Single Minor Section
The results from measurements on a single minor section are shown in Figure A3-8 – Figure A3-10.
Page 104
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure A3-8: Results for Single Minor Section Measurements, Excitation of the Coaxial
Mode
Page 105
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
(A) SENDING END VOLTAGE, SCREEN AND CORE CONDUCTORS (B) SENDING END VOLTAGE, SCREEN AND CORE ONDUCTORS
CABLE 2, VSH2 AND VC2 RESPECTIVELY CABLE 2, VSH3 AND VC3 RESPECTIVELY
(C) SENDING END VOLTAGE, INDUCED VOLTAGE ON (D) SENDING END CURRENS, SCREEN CONDUCTORS CABLE 2
CORE CONDUCTORS AND 3
Figure A3-9: Results for Single Minor Section Measurements, Excitation of the Inter-sheath
Mode, at Sending End of the Cables
Page 106
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure A3-10: Results for Single Minor Section Measurements, Excitation of the Inter-
sheath Mode
As shown in equations (A3.7) – (A3.8), these results can be used to validate the impedances in the cable
impedance matrix. The measurements can therefore be used in order to verify a specific cable model or cable
parameter calculations. This has already been done in [A9]. The results are shown in Figure A3-11 and Figure A3-
12.
The coaxial mode showed no difference between measurements and simulations, and therefore calculations of
core impedance parameters were verified. There was however difference for the intersheath setup, and therefore
the proximity effect between wires of the screen, and between cables was included, as well as correct modelling of
the two screen layers (Al laminate and Cu wires) in a new cable model. The validation of this model improvement
can be seen in Figure A3-12, whereas detailed description of the model improvements and simulations can be
found in [A9].
Page 107
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure A3-11: Comparison of the Sending End Current and Receiving End Voltage on
Energized Screen Conductor for the Coaxial Mode
Figure A3-12: Comparison of the Sending End Current and Receiving End Voltage on
Energized Screen Conductor for the Intersheath Mode
Page 108
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure A3-13: Field Test Setup for the Single Major Section Measurements
Page 109
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
As the cross-bonding points are included for this measurement setup, and since the exact value of grounding
resistance was not known, the results cannot be used for validation of impedance and parameter calculations. The
results can however, be used to verify modelling of a cable line with cross-bonding points. This has been done in
[A9], and the results can be seen in Figure A3-15.
Page 110
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
There are several cross-bonding points in the setup for measurements on multiple major sections. The cable
segments lengths are given below.
Page 111
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
The generator used is the same 1.2/50 µs surge tester as before. For this test, one must ensure that the impulse is
sufficiently large so it will not be damped too much along the long cable line. The voltage should still be below the
protection level of the surge voltage limiter of 5 kV. If this is not taken into account, the SVL’s can influence the
results. A 4.3 kV impulse is used for excitation. The impulse generator was, as before, connected to the HV cable
by use of a 5 m long 300 mm2 Cu wire. The wire is therefore represented by a 5 μH inductance connected in series
with the surge generator. The field measurement setup is as shown in Figure A3-16.
Figure A3-16: Field Test Setup for the Multiple Major Section Measurements
The results from measurements on multiple major sections are shown in Figure A3-17.
Page 112
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
As the cross-bonding points are included for this measurement setup, and since the exact value of grounding
resistance was not known, the results cannot be used for validation of impedance and parameter calculations.
There are 22 cross-bonding points in this long line. These results should therefore not be used for validation of
modelling cross-bonding points. The results can however, be used to verify modelling of an entire long cable line,
without shunt compensation. This has been done in [A9], and the results can be seen in Figure A3-18.
Page 113
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Figure A3-18: Validation of a Cable Model for a Long Line, without Shunt Compensation
The set of measurements shown above can be used for various type of model validation. The results are split into
verifying the cable model itself, model of a cable with cross-bonding points and a model of a long cable line.
Measurements of only the coaxial and intersheath modes are acceptable and can be used for cable parameter and
modelling software verification. As cross-bonding points cause reflections and influence the measurement results,
measurements on a major cable section cannot be used for validation of cable parameter calculations. Field
measurements on a major section and a longer cable line can however be used for verifying the modelling as a
whole.
The cables are excited with a fast impulse, and the measurements thereby include a very high frequency spectrum.
In order to verify the model used for system studies, a full scale field test has also been performed. The results
from this test are given in Section A3.2.
Page 114
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
C A
B
The energization, steady-state and de-energization waveforms were recorded three times in all three measuring
points. The closing of the circuit breaker was made by synchronised switching at voltage zero crossing. The
voltage and current were transformed to values readable by the measurement equipment by inductive voltage
transformers and inductive current transformers.
The recorded waveforms were then compared with simulations performed using a commercial software in order to
evaluate the model accuracy.
SIMULATION SETUP
The cable was modelled as being connected to an equivalent network modelled by an ideal voltage source and a
2372ej87.91° MVA short-circuit power.
For the cross-bondings a ground resistance of 3 Ω and a wire inductance of 1 μH were considered. All the minor
sections of the cross-bondings were modelled. The shunt reactors parameters were obtained directly from the
respective test report.
The simulated and measured waveforms are compared in the next pages.
Page 115
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
Energization Transient
Current
The energization of Horns Rev 2 cable is made using synchronised switching. As the three phases are connected
at voltage zero crossing, the initial DC component in the shunt reactor installed in the middle of the cable is virtually
maximal.
Figure A3-20 shows the currents in all the three-phases of the shunt reactor in the moments after its energization.
The value of the initial DC component is of approximately 400 A, a value virtually equal to the magnitude of the AC
component. Figure A3-21 shows that this DC component can also be observed in the currents registered at the
sending end. During the first 50 ms there is a high-frequency component that is superimposed on the 50 Hz
component, increasing the current amplitude.
There is good agreement between the simulated and measured voltage and current at the sending end during the
first moments of the transient. Approximately 10 ms after the energization, the simulation accuracy starts to reduce,
as observed by comparing the current waveforms.
The divergence between simulations and measurements after this instant is most likely because the model does
not include the lines adjacent to the energized node. It is a common practice in the transient overvoltage studies as
it causes more severe results. As a result, however, the reflections in the adjacent lines and nodes are not
reproduced in the simulation (see Section 2.7), which can result in the divergence between simulations and
measurements.
500 500
Current [A]
Current [A]
0 0
-500 -500
-1000 -1000
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
time [s] time [s ]
Figure A3-20: Current in the three-phases of the shunt reactor in first 0.5s after the
connection; a) Measured b) Simulated
0 0
Current [A]
Current [A]
-500 -500
-1000 -1000
-1500 -1500
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
time [s] time [s]
Figure A3-21: Three-phases currents in the sending end in the first 0.2s after the
connection; a) Measured; b) Simulated
Voltage
Due to the synchronised switching the voltage transients are very small, which can be seen in Figure A3-22.
When comparing voltage and current it can be seen that current is much more distorted due to the synchronised
switching, at voltage zero crossing.
Page 116
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
100 100
Voltage [kV]
Voltage [kV]
0 0
-100 -100
-200 -200
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
time [s] time [s]
Figure A3-22: Voltage in the cable's sending end during its energization; a) Measured; b)
Simulated
Steady-state
Table A3-1 compares the peak voltage and current in the sending end, receiving end and shunt reactor. The
comparison is made between the measured results and the simulations.
Table A3-1: Comparison between the measured results and the simulations: peak voltage
and current in the sending end, receiving end and shunt reactor.
Voltage in the Voltage in the Current in the Current in the
sending end [kV] receiving end [kV] sending end [A] shunt reactor [A]
Measured 133 142.6 510.6 372.8
Simulated 132.2 134.4 523.4 369.3
The simulation results are in good agreement with the measurements with the exception of the voltage in the
receiving end, which is higher in the measurements.
The current into the shunt reactor is practically equal in the simulations and measurements. The shunt reactor was
modelled according to the information provided in the test report, and it was validated against the measurement
results shown in the same test report. Therefore, as the current into the shunt reactor is the same in simulations
and measurements, the voltage in that point, at a distance of 55.4 km from the sending end, is also the same.
Consequently, it is concluded that, for steady-state condition, the land cable model is accurate and the most likely
reason for the difference in the receiving end voltage is the submarine cable model.
The inaccuracy in the submarine cable model may be caused by several reasons: modelling of the return path,
inaccurate permeability and/or resistivity values of the seabed/water, electric connectivity between the screens
[A10], [A11]. All these factors are difficult to model as the necessary data is normally unavailable.
Page 117
POWER SYSTEM TECHNICAL PERFORMANCE ISSUES RELATED TO THE APPLICATION OF LONG HVAC CABLES
A3.3 Conclusion
The field measurements performed on the cable line for the offshore wind farm Horns Rev 2 show how cable
parameter calculations, and an EMT type cable model can be validated, as well as the modelling of the connection
as a whole, for system studies.
The field test results have already been used for model improvements and network model validation. The methods
of measurements can therefore be used for validation on other cable lines, as well as the results shown here can
be used for validation of new cable models.
For the purpose of parameter calculations and model validation, it is recommended to use an impulse to excite the
cable, while for verifying the equivalent network modelling for system studies, a full scale network voltage should
be used during energization transients.
The full scale measurement allows the validation of not only a cable model, but also the models of other equipment
installed, such as shunt reactors and transformers.
References
[A9] U. S. Gudmundsdottir, Modelling of long High Voltage AC cables in Transmission Systems. Fredericia,
Denmark: PhD. Thesis, Aalborg University, 2010, ISBN: 978-87-90707-73-6.
[A10] F. M. F. da Silva, Analysis and simulation of electromagnetic transients in HVAC cable transmission grids.
Fredericia, Denmark: PhD. Thesis, Aalborg University, 2011, ISBN: 978-87-89179-99-5.
[A11] K. Yamabuki, K. Kubori, “A transient analysis of a scaled model for a submarine cable connected with an
offshore wind farm”, Electric Power Systems Research 85 (2012) 59-63.
[A12] H. W. Dommel, EMTP Theory Book. 2nd ed. Microtran Power System Analysis Corporation, Vancouver,
British Columbia, Canada, 1996.
Page 118
ISBN: 978-2-85873-251-7