Overvoltage Characteristics in Symmetrical Monopole HB MMC ... · modules and analyses cable stresses during various station internal as well as dc side faults. In order to examine

Overvoltage Characteristics in SymmetricalMonopole HB MMC-HVDC Configuration

comprising Long Cable SystemsM. Goertz, S. Wenig, C. Hirsching, K. M. Schafer, S. Beckler, J. Reisbeck, M. Kahl, M. Suriyah, T. Leibfried

Abstract—This contribution focuses on high voltage directcurrent (HVDC) transmission systems comprising modular multi-level converters (MMC) equipped with half-bridge (HB) sub-modules and analyses cable stresses during various stationinternal as well as dc side faults. In order to examine relevantovervoltage characteristics affecting HVDC cable systems, asystematic approach to evaluate overvoltage stresses is presentedand an extensive set of time-domain simulations has been carriedout. Obtained results are relevant for considerations on insulationco-ordination of HVDC cable systems.

Keywords—Extruded dc cable system, half-bridge, HVDC,insulation co-ordination, OHL-cable mixed system.

I. INTRODUCTION

THE number of installed HVDC projects in symmetricalmonopole (SMP) configuration based on HB MMC tech-

nology is emerging at a fast pace [1]. In line with the gatheredproject and operational experiences, several articles have cov-ered the aspects of overvoltage stresses in SMP configurationcaused by station internal or dc side faults. The general systembehaviour under dc side faults is discussed in [2]. Studiesconcerning insulation co-ordination of converter stations orfocusing on cable overvoltages are conducted in [3] and [4]–[6], respectively. With regard to offshore projects [7] givesan insight into the gained experience and presents measuringresults of a fault recorder during a cable fault. Moreover,recent research focuses on the laboratory imitation of theoccurring overvoltage shape during dc side faults in SMPconfiguration [8], [9]. However, since standardized test levelsfor HVDC cable systems with extruded insulation are notyet available [10], [11], preliminary insulation co-ordinationstudies are still of essential importance to provide a reliablecable system design. This contribution analyses overvoltagesaffecting the cable system in case of very long land cablesystems and gives an indication of the impact of systemscomprising mixed overhead lines (OHL) and cable sections.The scope of this research is motivated by the designatedembedded interconnectors in Germany.

M. Goertz, C. Hirsching, K. M. Schafer, M. Suriyah and T. Leibfried arewith the Karlsruhe Institute of Technology (KIT), Institute of Electric EnergySystems and High Voltage Technology (IEH), Karlsruhe, BW 76131, Germany(e-mail of corresponding author: [email protected]).S. Beckler, S. Wenig, J. Reisbeck and M. Kahl are with the TransnetBWGmbH, Stuttgart, BW 70173, Germany (email: [email protected]).

Paper submitted to the International Conference on Power SystemsTransients (IPST2019) in Perpignan, France June 17-20, 2019.

II. SYSTEM DESCRIPTION AND MODELLING

A schematic of the considered SMP configuration is de-picted in Fig. 1 and underlying equipment design is statedin Tab. I. Converter equipment as well as cable terminationsare protected by surge arresters (SA). The non-linear voltage-current characteristic of the SAs is approximated by piece-wise linear resistances. Converter arm inductances are locatedon the dc side. Investigated transmission system configurationsare shown in Fig. 2. In order to evaluate the impact of verylong land cable systems, two different transmission systemconfigurations comprising either 200 km or 700 km onshorecable with extruded insulation are investigated. Then, resultsare extended to an OHL-cable mixed system with a totallength of 700 km. Along the cable system, a solid cableshield grounding is assumed every 5 km taking into accounta grounding resistance of RSJ = 5Ω and RS = 0.1Ω at jointsand cable terminations, respectively. Cable and OHL sectionsare modelled by frequency-dependent line models in accor-dance with the theory given in [12]. The submodule stacks ofthe MMCs are represented through a Type 4 detailed equivalentcircuit model according to [13]. Time-domain simulations arecarried out using PSCAD/EMTDC. An adequate solution timestep of 5µs is considered.

A. Control and Protection

Station 1 operates in an active/reactive power control mode,station 2 acts as a dc voltage/reactive power controlled sta-tion. The protection system of each converter station consists

TABLE ISELECTED PARAMETERS OF SYMMETRICAL MONOPOLE CONFIGURATION

Parameter Valuerated power Pr 1 GWnominal dc voltage (pole-to-ground) U0 ± 320 kVnominal ac voltage (valve-/ grid side) 330 kV / 400 kVline frequency f 50 Hzshort circuit level ac grid 45 GVAX/R ratio ac grid 10transformer configuration wye-deltanumber of submodules per arm N 256average arm sum voltage 640 kVaverage submodule voltage 2.5 kVsubmodule capacitor CS 8.5 mFarm inductance Larm 50 mHclearing time ac circuit breakers TC 80 msswitching impulse protective level of arresters 1.7 p.u. @ 3 kA

F4 F5

F1

RSRSJshield grounding

distance: 5 km

AC-CBsAC-CBs

Station

MMC 1

Station

MMC 2

F2

Transmission line configuration, see Fig. 2

SM

stack

LarmF3

HB submodules

Fig. 1. Schematic of symmetrical monopole configuration.

Cable length: 700 km

Cable length: 200 km

Cable length: 300 km Cable length: 300 kmOHL length: 100 km

F5 F6

Fig. 2. Single pole diagram of investigated transmission system configura-tions: 700 km cable, 200 km cable and 700 km OHL-cable mixed system.

of valve overcurrent, submodule under-/overvoltage loops aswell as a dc pole-to-ground voltage imbalance criterion. Theprotection loops comprise artificial delays in order to addressdelays due to data acquisition and processing. After protectiontripping, all IGBTs of the affected converter are blocked andan opening order is sent to the ac circuit breakers (AC-CBs).Protection thresholds and controller parameters are held con-stant for all investigated transmission system configurations.

B. Parametric Study Framework

With regard to voltage stresses affecting the cable system,a broad range of station internal as well as dc line faults areconsidered, as stated in Tab. II. In order to ensure that over-voltages are derived in consideration of worst-case conditions,different pre-fault converter operation modes and various faultinstants are taken into account.

III. SYSTEMATIC APPROACH TO EVALUATEOVERVOLTAGES

In order to assess occurring voltage stresses affecting thecable system, a systematic approach for calculating over-voltage parameter is developed. First, a parametric study isperformed by means of EMT-software. Then, obtained dataare evaluated during post-processing by numerical computingsoftware. Along the cable system, voltage measuring pointsare placed in equidistant sections with a length of 5 percentof total transmission length. For each run of the parametricstudy all voltage measuring points along the cable are takeninto consideration for post-processing. This measure allows to

TABLE IIPARAMETRIC STUDY FRAMEWORK

Description Configuration

1. power set pointat station 1

a) +Pr (ac in-feed), +Qr (cap.)b) −Pr (ac export), +Qr (cap.)c) 0GW (zero load), +Qr (cap.)

2. fault type

F1: positive dc pole-to-ground fault atcable terminationF2: phase a-to-ground fault at transformervalve-sideF3: positive arm p1-to-ground faultF4: positive dc cable core-to-screen-to-ground fault at 1km distance from station 1F5: positive dc cable core-to-screen-to-ground fault at 50% of transmission lengthF6: positive dc pole-to-ground fault atOHL-cable transition station (mixed line)

3. fault resistance 0.1Ω, 10Ω

4. faultsynchronisation

a) zero crossing of phase a-to-groundvoltage at transformer valve-sideb) zero crossing of ac current in phase a attransformer valve-side

5. fault instanta) ω · t = 0, b) ω · t = 45

c) ω · t = 90, d) ω · t = 225

e) ω · t = 270 after zero crossing

TP

TP

Fig. 3. Determination of time-to-peak value.

characterise the voltage shape at each point along the cable.Besides the absolute maximum overvoltage levels at eachmeasuring point, time-to-peak values and maximum voltagegradients are calculated. Here, the time-to-peak value TP is

(a)

-1.7p.u. -1.7p.u.-1.72p.u. -1.7p.u. -1.76p.u.

F1 F2 F3 F4 F5Fault Location

-1.8-1.7-1.6-1.5-1.4-1.3-1.2-1.1

-1-0.9-0.8-0.7-0.6-0.5-0.4-0.3-0.2-0.1

0

u dc,n

in p

.u.

200km cable, +Pr,

200km cable, -Pr

200km cable, 0 GW700km cable, +P

r,

700km cable, -Pr

700km cable, 0 GW

(b)

-0.61p.u.

1.7p.u.

-0.74p.u.-0.95p.u.

-0.68p.u.

F1 F2 F3 F4 F5Fault Location

-1-0.85

-0.7-0.55

-0.4-0.25

-0.10.050.2

0.350.5

0.650.8

0.951.1

1.251.4

1.551.7

1.85

u dc,p

in p

.u.

200km cable, +Pr,

200km cable, -Pr

200km cable, 0 GW700km cable, +P

r,

700km cable, -Pr

700km cable, 0 GW

Fig. 4. Absolute maximum overvoltages or polarity reversals along the cablesystem as a function of fault type F1 - F5, power set point and cable length:(a) voltage stresses along negative dc pole, (b) voltage stresses along positivedc pole.

defined as the time interval between ±5% of rated dc voltageU0 and the point in time of absolute maximum overvoltage. Itis important to mention that this definition is not in line withthe time to crest defined in [14] for impulse voltage test. Thefront of the overvoltage might consist of superimposed voltageoscillations or travelling wave phenomena, as exemplified inFig. 3. This issue has to be kept in mind when evaluating time-to-peak values. During the time-to-peak, absolute maximumvoltage gradients are determined. In order to avoid that numer-ical oscillations might distort the obtained gradient, an averagevoltage gradient along five solution time steps is calculated.

IV. IMPACT OF VERY LONG CABLE SYSTEMS

Within this section, results of the 200 km as well as 700 kmcable system are presented and the impact of long cablesystems on voltage stresses affecting the cable system isdiscussed. Obtained results are classified in overvoltages withsame polarity as dc operating voltage and overvoltages withopposite polarity to dc operating voltage. An overvoltage withopposite polarity to dc operating voltage might occur at thefaulted dc pole during the cable discharge process.

A. Absolute Maximum Overvoltage Levels

For each run of the parametric study absolute maximumvoltage levels of all voltage measuring points along thecable are determined during post-processing. Then, absolutemaximum voltage levels of all runs of the parametric studyrelated to the same fault type and the same power set pointare identified. Results are depicted in Fig. 4 (a)-(b) in per

unit (p. u.) of rated dc voltage U0. All investigated dc sidefaults (F1, F3, F4, F5) are located along the positive dcpole. Hence, a polarity reversal and an overvoltage with samepolarity as operating voltage can be observed at the positiveand negative dc pole, respectively. For both cable lengths,absolute maximum overvoltage levels occur during a cablefault at 50% cable length (F5) at zero load operation (0 GW).Absolute maximum overvoltage levels are 1.76 p. u. and1.72 p. u. for 200 km cable length and 700 km cable length,respectively. Occurring overvoltage levels reach higher valuesin the 200 km system than in the 700 km system for allinvestigated fault types.

(a)

absolute maximum overvoltage,

cable fault at 50% cable length (F5)

F2

F3

F4F1

F2

F3

F1, F4

(b)

absolute maximum

polarity reversal,

cable fault close to

station (F4)

Fig. 5. Worst case voltage profiles along the cable system as a function offault type F1 - F5 and cable length: (a) negative dc pole, (b) positive dc pole.

B. Voltage Profiles along the Cable SystemA worst-case voltage profile along the cable for each fault

type is shown in Fig. 5 (a)-(b). Within this study, a worst-case is defined as absolute maximum overvoltage or polarityreversal at each voltage measuring point along the cable forall simulation runs related to the same fault type. Hence,a voltage profile of the same fault type might consist ofdifferent simulation runs. As can be seen, highest overvoltagelevels occur in the middle of the cable of the healthy dc poleduring F5. This finding is also achieved in [5]. At the faultedcable, the absolute maximum polarity reversal can be observedsubsequent to a cable fault in the vicinity of the converterstation (F4). The absolute maximum polarity reversal occurs atthe cable termination adjacent to the faulted cable section. Theoccurring polarity reversal is below 1 p. u. and is independentof total cable system length. As shown in [4], [15], the voltagereversal at the faulted cable is caused by the intrinsic dischargeprocess of the cable through the fault impedance. As converters

have only limited impact on the cable discharge and thus onoccurring polarity reversal, the following sections focus on theovervoltage along the cable of the healthy dc pole.

(a)

first superposition

of uf1 and uf2

blocking instant of

MMC1blocking instant of

MMC2

uf1,uf2

(b)

first superposition

of uf1 and uf2 blocking instant at

MMC1 and MMC2

uf1,uf2

Fig. 6. Overvoltage build-up along the dc cable of the negative pole during acable fault at the positive pole (F5): (a) cable length 200 km, (b) cable length700 km.

C. Overvoltage Build-up during Cable Fault F5

This section describes the voltage build-up along the cableduring the worst-case fault F5. Figure 6 presents the cablecore-to-ground voltage udc,n at different voltage measuringpoints along the cable for both considered system lengthsduring the worst-case run of F5. As can be seen in the zoomedpart of Fig. 6 (b), both converters block their IGBTs at thesame time instant (green curve). For the 200 km system,blocking instants of both converters are slightly different. Atthe instant of IGBT blocking tby , the voltage udc,n at therespective converter station y ∈ 1, 2 consists of an impulsevoltage ufy superimposed on dc operating voltage, as depictedin Fig. 6. The impulse voltages uf1 and uf2 represent forwardtravelling waves that propagate from both converter stationsinto the cable. In case both converters block at the same timeinstant, a first superposition of uf1 and uf2 occurs in the middleof the cable. Otherwise, the location of the first superpositionmight deviate from the middle of the cable. Thus, the voltageat the point of superposition x0 can be written as the sum ofboth travelling waves superimposed on dc operating voltage:

u(x0, t) = U0 + uf1 · e−α·l1 + uf2 · e−α·l2 . (1)

The exponential parts in (1) describe the attenuationof the travelling wave along the cable distance ly .For simplicity, dispersion effects are neglected. Theattenuation constant α is frequency-dependent and is

α(10Hz . . . 1kHz) ≈ 0.36×10−3 1km . . . 1.6×10−3 1

km for theconsidered cable design. Applying (1), the superpositionof both travelling waves can be estimated based on theassumption that α is constant. In the 200 km system, theabsolute maximum overvoltage along the cable is caused bythe first superposition of uf1 and uf2. In the 700 km system,the superposition of uf1 and uf2 leads to a peak during thefront of the overvoltage, but not to the absolute peak voltage,see Fig. 6 (b). This is due to following reasons: i) for verylong cable systems the cable self attenuation effect mitigatesthe impulse voltages propagating along the cable, ii) theinitial impulse voltage at the converter stations ufy as wellas the voltage gradient prior to blocking decreases for verylong cable sections. However, it should be kept in mindthat system behaviour is affected by considered protectionsthresholds and blocking delays as well as converter controlprior to blocking.

(a)

770µs

8480µs

2460µs

3360µs2810µs

4215µs

795µs

8485µs

955µs

4325µs

F1 F2 F3 F4 F5Fault Location

0

2000

4000

6000

8000

10000

Tim

e-to

-pea

k va

lue

in µ

s

200km cable700km cable

(b)

0.59kV/µs

0.46kV/µs

0.28kV/µs

0.59kV/µs0.52kV/µs

0.55kV/µs

0.43kV/µs

0.3kV/µs

0.45kV/µs

0.52kV/µs

F1 F2 F3 F4 F5Fault Location

0

0.1

0.2

0.3

0.4

0.5

0.6

Vol

tage

gra

dien

t in

kV/µ

s

200km cable700km cable

Fig. 7. Impact of cable length on overvoltage characteristics: (a) fastest time-to-peak values, (b) absolute maximum voltage gradients.

D. Overvoltage Characteristics

Fastest time-to-peak values and absolute maximum voltagegradients of all voltage measuring points along the cable ofthe healthy pole under consideration of all runs are depictedin Figure 7 (a)-(b). In the 200 km system, fastest time-to-peakvalues TP are in the range of 700µs and occur during a faultat the cable termination (F1). In the 700 km system, fastesttime-to-peak values are in the range of several milliseconds.Moreover, voltage gradients during the front of the overvoltagetake on smaller values with increasing cable length. Figure 8shows a distribution of all calculated time-to-peak values. Forlong cable systems, a wider range of time-to-peak valuesexists. It should be noted that a phase-to-ground fault at

(a) (b)

Fig. 8. Histogram of all occurring time-to-peak values: (a) cable length200 km, (b) cable length 700 km.

transformer valve-side (F2) might evoke an overvoltage on thedc side where the peak value is reached only after several half-cycles of ac voltage. Therefore, certain TP values are between20 ms to 30 ms for both system lengths.

(a)

cable-OHL

transition

OHL-cable

transition

(b)

cable-OHL

transition

Fig. 9. Worst case voltage profiles along the cable system as a function offault type F1 - F6 and transmission line configuration. Considered systemlength 700 km: (a) negative dc pole, (b) positive dc pole.

V. IMPACT OF OHL-CABLE MIXED SYSTEMS

This section focuses on a transmission length of 700 km andextends results to mixed transmission systems. As introducedin Fig. 2, the mixed system comprises a 100 km OHLsection embedded between two cable sections, each with alength of 300 km. The mixed system is compared to the700 km system comprising solely cable sections. The worst-case voltage profiles under consideration of all investigatedfault types are depicted in Fig. 9. As can be seen in Fig. 9 (a),

absolute maximum overvoltages along the cable of the healthypole reach similar levels in both systems. However, the voltageprofiles in the mixed system show a strong location depen-dency along the cable. This is due to the fact that the embeddedOHL represents a discontinuity in surge impedance at thecable transition stations. Hence, the OHL-cable transitions leadto additional travelling wave reflections. Fastest time-to-peakvalues and absolute maximum voltage gradients of the cableovervoltage at the healthy pole are shown in Fig. 10. Themixed system shows faster time-to-peak values and significantsteeper voltage gradients than the system comprising solelycable sections. Especially, faults at the OHL-cable transitionstation (F6) lead to shortest front times and steepest voltagegradients of all investigated fault types.

(a)

4025µs

8480µs

4090µs3360µs3565µs

4215µs 4020µs

8485µs

6840µs

4325µs

3055µs

F1 F2 F3 F4 F5 F6Fault Location

0

2000

4000

6000

8000

10000

Tim

e-to

-pea

k va

lue

in µ

s

700km mixed700km cable

(b)

0.39kV/µs

0.46kV/µs

0.27kV/µs0.3kV/µs

0.41kV/µs

0.52kV/µs0.44kV/µs

0.52kV/µs

0.4kV/µs0.43kV/µs

1.14kV/µs

F1 F2 F3 F4 F5 F6Fault Location

0

0.2

0.4

0.6

0.8

1

1.2

Vol

tage

gra

dien

t in

kV/µ

s

200km mixed700km cable

Fig. 10. Impact of transmission line configuration on overvoltage charac-teristics. Considered system length 700 km: (a) fastest time-to-peak values,(b) absolute maximum voltage gradients.

VI. CHARACTERISTIC VOLTAGE SHAPES

The voltage curves at the location of absolute maximumovervoltage during a fault at 50% system length (F5) and aphase-to-ground fault at transformer valve-side (F2) are shownin Fig. 11. For F5, differences exist between the consideredtransmission system configurations during the overvoltagefront. The voltage shape during F5 consists of a voltageincrease up to a peak value followed by a temporary over-voltage (TOV) at a decreased level. The TOV level dependson considered SA characteristic and is approximately 1.5 p. u.for all investigated systems. The TOV persists until the cableis discharged through intrinsic shunt or stray impedances toground or auxiliary earthing breakers are applied, see [4] forfurther explanation. During F2, voltage oscillations can beobserved at the cable system until AC-CBs have cleared thefault. The amplitude of the voltage oscillation increases withdecreasing cable length as well as for mixed systems.

(a)

2.06 2.07 2.08 2.09 2.1 2.11 2.12 2.13time in s

-1.8

-1.7

-1.6

-1.5

-1.4

-1.3

-1.2

-1.1

-1

-0.9

u dc,n

in p

.u.

200km cable, F5, RF=0.1 , t=225°, 0 GW, meas. loc. 55%

700km cable, F5, RF=0.1 , t=90°, 0 GW, meas. loc. 50%

700km mixed, F5, RF=0.1 , t=90°, +P

r, meas. loc. 15%

(b)

2.08 2.1 2.12 2.14 2.16 2.18 2.2time in s

-1.75

-1.65

-1.55

-1.45

-1.35

-1.25

-1.15

-1.05

-0.95

-0.85

-0.75

-0.65

u dc,n

in p

.u.

200km cable, F2, RF=0.1 , t=225°, -P

r, meas. loc. 35%

700km cable, F2, RF=0.1 , t=270°, -P

r, meas. loc. MMC2

700km mixed, F2, RF=0.1 , t=270°, -P

r, meas. loc. 75%

Fig. 11. Worst case voltage curves at the location of absolute maximumovervoltage: (a) fault at 50% transmission length (F5), (b) phase-to-groundfault at transformer valve-side (F2).

VII. CONCLUSION

In order to assess relevant overvoltage characteristicsaffecting HVDC cable systems in SMP configuration basedon HB MMC technology, a detailed parametric study has beencarried out and a systematic approach to evaluate overvoltagesis presented within this paper. The following list provides anoverview of the main findings:

• A decrease of absolute maximum overvoltage level canbe observed for very long cable system lengths. Thisstatement is proven only for the considered cable systemlengths of 200 km and 700 km.

• An increasing cable system length leads to slower time-to-peak values and lower voltage gradients.

• Time-to-peak values have to be evaluated carefully as theovervoltage front might consist of superimposed voltageoscillations.

• In the considered systems, fastest time-to-peak valuesand absolute maximum overvoltage levels occur not atthe same measuring location. Therefore, a combinationof fastest time-to-peak value and highest overvoltagelevel for testing purposes might result in unrealistic cablestresses.

• In mixed OHL-cable systems, faster time-to-peak valuesand significant steeper voltages gradients occur comparedto systems comprising solely cable sections. For theconsidered systems absolute maximum overvoltage levelsare similar.

• The authors expect that short cable sections embeddedbetween OHL segments might lead to multiple super-positions of travelling waves for certain fault locations

and result in severe overvoltages. Therefore, overvoltagestresses affecting the cable system in OHL-cable mixedsystems are difficult to predict as system behaviour de-pends on projects specific parameters.

For the sake of completeness, it should be mentioned thatfurther project specific parameters such as converter stationdesign or the transformer vector group might impact onovervoltages [3], [7]. Future research is required in orderto examine if the analysed parameters such as overvoltagelevel, front time and voltage gradient might be critical stressesfor the cable insulation. It is worth mentioning that besidesthe investigated overvoltage characteristics, other parameterssuch as duration of TOV appear relevant for cable systemdesign. However, results provided within this paper representa profound starting point for further work on insulation-coordination of HVDC cable systems. Therefore, obtainedresults are of importance with regard to upcoming projectsin symmetrical monopole HB MMC-HVDC configuration.

REFERENCES

[1] M. Saltzer et. al., “Surge and extended overvoltages testing of HVDCcable systems,” Int. Conf. on Insulated Power Cables (Jicable’17),Dunkerque, France, Nov. 2017.

[2] F. B. Ajaei and R. Iravani, “Cable Surge Arrester Operation Due toTransient Overvoltages Under DC-Side Faults in the MMC-HVDCLink,” IEEE Trans. Power Del., vol. 31, no. 3, pp. 1213–1222, June2016.

[3] H. Saad, P. Rault, S. Dennetiere, “Study on Transient overvoltages in theConverter Station of HVDC-MMC links,” Int. Conf. on Power SystemTransients (IPST’17), no. IPST17-185, Seoul, Republic of Korea, June2017.

[4] S. Dennetiere, H. Saad, A. Naud, P. Honda, “Transients on DC cablesconnected to VSC converters,” Int. Conf. on Insulated Power Cables(Jicable’15), Versailles, France, June 2015.

[5] S. Mukherjee, M. Saltzer, Y.- J. Hafner, S. Nyberg, “Cable Overvoltagesfor MMC based VSC HVDC Systems: Interactions with Converters,”Int. Colloquium on H.V. Insulated Cables, CIGRE Study Committee B1,New Delhi, India, Oct. 2017.

[6] M. Goertz et. al., “Analysis of Overvoltage Levels in the Rigid BipolarMMC-HVDC Configuration,” 15th IET Int. Conf. on AC and DC PowerTrans. (ACDC 2018), Coventry, UK, Feb. 2019.

[7] M. Greve, M. Koochack Zadeh, T. Rendel, A. Menze, “Behaviour of theHVDC links with MMC technology during DC cable faults,” CIGREWinnipeg 2017 Colloquium, Study Committees A3, B4 & D1, Winnipeg,Canada, Sept. 2017.

[8] C. Freye, S. Wenig, M. Goertz, T. Leibfried, F. Jenau, “Transient VoltageStresses in MMC-HVDC links - Impulse Analysis and Novel Proposalsfor Synthetic Laboratory Generation,” IET High Voltage, vol. 3, no. 2,pp. 115–125, June 2018.

[9] T. Karmokar, M. Saltzer, S. Nyberg, S. Mukherjee, P. Lundberg, “Eval-uation of 320 kV extruded DC cable system for temporary overvoltagesby testing with very long impulse waveforms,” CIGRE General Meeting,Paris, France, Aug. 2018.

[10] “Recommendations for testing DC extruded cable systems for powertransmission at a rated voltage up to 500 kV,” CIGRE Tech. Rep. 496(WG B1.31), 2012.

[11] “High voltage direct current (HVDC) power transmission - Cables withextruded insulation and their accessories for rated voltages up to 320 kVfor land applications - Test methods and requirements,” IEC 62895:2017.

[12] A. Morched, B. Gustavsen and M. Tartibi, “A universal model foraccurate calculation of electromagnetic transients on overhead lines andunderground cables,” IEEE Trans. Power Del., vol. 14, no. 3, pp. 1032–1038, July 1999.

[13] “Guide for the development of models for HVDC converters in a HVDCgrid,” CIGRE Tech. Rep. 604 (WG B4.57), 2014.

[14] “High-voltage test techniques - Part 1: General definitions and testrequirements,” IEC 60060-1:2010.

[15] M. Goertz, S. Wenig, S. Suriyah, and T. Leibfried, “Determinationof transient overvoltages in a bipolar MMC-HVDC link with metallicreturn,” Proc. Power Sys. Computation Conf., Dublin, Ireland, June 2018.