Digital Object Identifier: /TPWRD URL:

Transcription

1 Leterme, W., Tielens, P., De Boeck, S., Van Hertem, D. (214). Overview of rounding and Configuration Options for Meshed HVDC grids. IEEE Transactions on Power Delivery. Digital Object Identifier: 1.119/TPWRD URL: This is the author s version of an article that has been accepted for publication in IEEE Transactions on power delivery. Changes are made to this version by the publisher prior to publication. c 214 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.

2 1 Overview of rounding and Configuration Options for Meshed HVDC rids Willem Leterme, raduate Student Member, IEEE, Pieter Tielens, raduate Student Member, IEEE, Steven De Boeck, raduate Student Member, IEEE and Dirk Van Hertem, Senior Member, IEEE Abstract This paper provides an overview and comparison of the possible grounding and configuration options for meshed HVDC grids. HVDC grids are expected to play a key role in the development of the future power system. Nevertheless, the type of grounding and the base configuration for the grid have not yet been determined. Various studies related to multiterminal HVDC or meshed HVDC grids often assume one specific configuration and grounding scheme and take it for granted. However, as there exist a large number of options, an overview is needed to balance pros and cons. In this paper, the influence of the different grounding options on fault behavior is investigated for point-to-point connections. Furthermore, the impact of the grounding type on the system fault behavior is investigated with electromagnetic transient simulations. Next, the suitability of a configuration to serve as base configuration in a meshed DC grid is investigated and compared in terms of extensibility and flexibility. In this evaluation, the grounding type, the number and location of grounding points in a grid are considered as well. Finally, an overview of the most important conclusions is given in a summarizing table. Index Terms DC fault, grounding, HVDC grid configuration, transient analysis, VSC HVDC I. INTRODUCTION MESHED HVDC grids are expected to play a key role in the development of the future power system since they are an interesting option to transport large amounts of renewable energy from the remote sources to the load centers and to fundamentally upgrade the existing AC network. In Europe, the construction of 126 km of (mainly submarine) HVDC links has already been planned in the next ten years [1]. For connecting remote offshore wind farms to the mainland, HVDC is the preferred option [2]. If these links are interconnected at a later stage, a meshed DC grid, also often referred to as supergrid, on top of the underlying AC network can be created [3]. To complete this supergrid, different choices regarding technology have to be made. Although only multi-terminal CSC (Current Source Converter) HVDC systems exist today, VSC (Voltage Source Converter) HVDC is considered to be more appropriate for the future meshed DC grid [4]. Some advantages of VSC HVDC are increased controllability, cheaper cable technology and the possibility to connect to weak AC grids [], [6]. Furthermore, new technologies for VSC converters have resulted in an increased converter efficiency close to the one of CSC [7]. W. Leterme, P. Tielens, S. De Boeck and D. Van Hertem are with KU Leuven, Belgium (EnergyVille/Electa research group, Electrical Engineering Department ESAT, Kasteelpark Arenberg 1 (PB244), 31 Heverlee). The work of Willem Leterme was supported by a research grant from the Research Foundation-Flanders (FWO). Contact: willem.leterme@esat.kuleuven.be, pieter.tielens@esat.kuleuven.be, steven.deboeck@esat.kuleuven.be, dirk.vanhertem@esat.kuleuven.be For VSC HVDC, several options exist regarding configuration and grounding. These include a low impedance grounded asymmetric monopolar, a high impedance grounded symmetric monopolar or a low or high impedance grounded bipolar configuration [8]. For a meshed HVDC grid, it is currently not clear which configuration and grounding type must be chosen. Yet these choices have a large impact on system cost, protection system design and extensibility of the grid. Moreover, the number of options increases significantly when considering all grounding options such as solid grounding, grounding through an impedance or leaving the system ungrounded [9]. In the literature, various configurations and grounding types have been used to study multiterminal HVDC or HVDC grids. In DC fault studies, this leads to different conclusions. In [1] and [11], a solid grounded bipolar configuration is used to evaluate fault currents, respectively to investigate fault clearing possibilities and the effect of DC grid topologies. The fault behavior of a two-level converter and a modular multilevel converter (MMC) for a symmetric configuration is compared in [12]. In [13], fault behavior for a symmetric configuration is examined for different types of grounding at both AC and DC side. As for protection studies, the same differences can be noted. In [14], a solid grounded bipole multiterminal system is considered for fault detection using wavelets. On the contrary, [1] and [16] use a high impedance grounded symmetric configuration for the design of fault location methods. In [17], topologies for a multiterminal HVDC network are evaluated on criteria such as flexibility and redundancy. This study focused on network topology rather than on HVDC configuration. The main contribution of this paper is to provide an overview and comparison of grounding and configuration options for meshed HVDC grids. Furthermore, a transient analysis for pole-to-ground faults in point-to-point connections is performed for different types of grounding impedances. As HVDC grids are still in a study phase, the transient analysis is based purely on transient simulations using PSCAD/EMTDC [18], which need to be validated by an experimental test setup. The feasibility of using each configuration and corresponding grounding option as a base configuration in a meshed HVDC grid is evaluated using criteria such as extensibility and flexibility for post-fault operation. Moreover, the influence of the number and location of grounding points on the normal operation and fault behavior of the system is investigated. The extensions on the work done in [9] are twofold. First, the transient study shows the fault behavior for different grounding types. Second, the general remarks stated in [9] are more concretely associated to each configuration and grounding option for a clearer overview. The paper is organized as follows. In section II, the base Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

3 2 +Un +Un grounding options is discussed. The impact of the grounding type on fault transients is further elaborated in the next section. +Un/2 Un/2 +Un Un A A A A (a) Asymmetric monopolar configuration (b) Symmetric monopolar configuration (c) Bipolar configuration (d) rounding options B +Un/2 B Un/2 Fig. 1. Base configurations and grounding options for point-to-point HVDC connections configurations and grounding options are discussed. Section III provides the models and conclusions of the transient simulations for each grounding option. Next, the suitability for each configuration and grounding option in a grid is discussed in section IV. In this section, a case study is performed to show the flexibility of the configurations for post-fault operation. Finally, in section V, the conclusions are given and summarized in a table. II. ROUNDIN AND CONFIURATION OF POINT-TO-POINT CONNECTIONS This section discusses the impact of base configuration and grounding type on system design. The two base configurations for HVDC, asymmetric and symmetric monopole, are shown in Figs. 1(a) and 1(b). An asymmetric configuration can be extended to a bipolar configuration, shown in Fig. 1(c). The symbol in Fig. 1 refers to the possible locations for grounding the HVDC system and can represent any of the basic grounding options, shown in Fig. 1(d). For the asymmetric configuration only low impedance grounding can be used 1. On the other hand, for symmetric and bipolar configurations each grounding option shown in Fig. 1(d) can be used. In this section, the steady-state operation after a pole-to-ground fault for the different configurations in combination with the 1 The distinction between low and high impedance grounding is in this paper based on the impedance at low frequencies. +Un B B Un A. Asymmetric monopolar configuration In normal operation, the metallic return conductor of an asymmetric monopole is operated at near-zero voltage and the positive pole voltage is equal to the nominal voltage U n of the converter. In order to limit the voltage rise on the metallic return, the asymmetric monopole needs to be low impedance grounded, with solid grounding as an ideal case. Normally only one converter is grounded, since earth currents can occur in normal operation when both converters are solidly grounded [19]. In case of a pole-to-ground fault on the positive pole in a solidly grounded system, the steady-state post-fault voltage does not exceed the nominal voltage. The voltage on the metallic return remains near-zero. On the other hand, the fault current shows a steep increase and has a high steady-state value. In most VSC HVDC converter topologies, the fault current is uncontrolled and continuously supplied after the fault [2]. Hence, fast fault detection and clearance is required. Basically, alternatives for solid grounding are grounding through a resistor or an inductor (Fig. 1(d)). More advanced grounding schemes, using a mixed configuration or active components at selected grounding locations are possible, however the analysis of these schemes falls out of the scope of this paper. Increasing the grounding resistance decreases steady-state fault currents whereas the voltage on the metallic return increases. In practice, there is always a resistance present in the grounding path [21]. Furthermore, if both converters are grounded, resistance grounding limits the steady-state current flowing through the earth in normal operation. An inductor in the ground path does not affect steady-state operation or steady-state fault currents. By contrast, it affects fault transients by limiting the rate-of-rise of the fault current. B. Symmetric monopolar configuration For the symmetric monopolar configuration, the steadystate voltage in normal operation on each pole is half the nominal converter voltage U n. In case of pole-to-ground faults, the steady-state fault voltage of the healthy pole can reach values up to the converter voltage U n. A neutral point can be made available by means of large impedances. In Fig. 1(b), this has been done by using the DC capacitors. Low impedance grounding (solid, resistance or inductance) as well as high impedance grounding (capacitance, ungrounded) of the neutral point is possible. If the grounding is of the low impedance type, the steady-state post-fault voltage on the midpoint of the capacitors is clamped to zero. The steadystate post-fault voltage on the capacitor of the healthy pole is U n. For high impedance grounding, the post-fault steadystate voltage on the midpoint of the capacitors is nonzero. The nominal converter voltage U n is shared between the capacitors. The steady-state fault current is for every grounding type zero, unless the converter transformer is Yg-connected at the converter side or omitted. Fault clearance time constraints can be less stringent than for the asymmetric configuration (e.g. Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

4 3 V ac L R k + u Converter 1 P1 ac ac Yg V dc - N1 2 km 2 km P2 N2 Converter 2 u k Yg R ac L ac V ac Fig. 2. Simulation model for the asymmetric monopolar configuration existing links with this type of configuration rely on fault clearing on the AC side, which is relatively slow). C. Bipolar configuration For the bipolar configuration, both low and high impedance grounding are possible. In case of solid grounding, the steadystate post-fault voltage on the healthy pole is the nominal converter voltage U n for a pole-to-ground fault. Analogous to an asymmetric monopole, the metallic return conductor is operated and rated at near-zero voltage. The fault current for a low impedance grounded bipolar configuration has the same properties as the fault current for a low impedance asymmetric configuration. rounding the bipole through a resistor or an inductor is possible as well. Similar conclusions for fault behavior can be drawn as for the asymmetric configuration. In case of a high impedance grounded bipolar configuration, the advantage of a low rated metallic return is lost. The steady-state post-fault voltage on the metallic return can reach the nominal converter voltage U n for an ungrounded system. Furthermore, the steady-state post-fault voltage on the healthy pole can reach up to 2 U n. The steady-state fault current for a single pole-to-ground fault in case of an ungrounded system is zero. III. POLE-TO-ROUND FAULT TRANSIENTS IN POINT-TO-POINT CONNECTIONS The transient behavior of a pole-to-ground fault has an impact on the protection system design and fault recovery. In this section, transient simulations are performed with PSCAD/EMTDC [18] for different grounding options as described in section II. The parameters for the different grounding types have been chosen for demonstration purposes. To enable a clear extraction of the effects of grounding and configuration on a fault transient in a HVDC system, simulations are performed for a point-to-point connection. The insights obtained from these simulations can further be used for an assessment of grounding and configuration of meshed HVDC grids, as given in the next section. The fault transient in meshed grids however is also influenced by many other factors such as fault location within the grid or grid topology [11]. A. DC system A point-to-point connection consisting of two converters connected by a 4 km cable is considered (e.g. Fig. 2 for an asymmetric configuration). Converter 1 controls the DC voltage, whereas converter 2 controls the active power. System grounding is provided only at converter 2 to avoid earth currents. A similar model is used for the symmetric configuration, where both converters are grounded in the same way as this does not lead to steady-state earth currents. The cable and converter parameters are enlisted in Tables I and II in the Appendix. The AC grid is modeled by a voltage source and an equivalent grid impedance. The converter transformer has a Yg- winding configuration and a leakage reactance u k. The HVDC converter topology is the half-bridge MMC topology [22]. It is modeled using a Thévenin equivalent model with inclusion of IBT (Insulated ate Bipolar Transistor) blocking [23], [24]. Converter control is implemented as described in [2]. Capacitors at the DC side are used for the symmetric configuration to provide the neutral point at the DC side of the converters. As DC capacitors are not strictly required for the MMC and only serve for filtering purposes [26], a smaller value can be chosen, in this case 1 µf. The neutral point can alternatively be provided by other large impedances. For the asymmetric configuration, both the cases with and without DC capacitors have been considered. The grid and converter parameters are obtained from [14] and [27]. The DC cables are modeled using the frequency dependent model of PSCAD [28]. Positive and negative pole cables are modeled identically. The cable consists of a copper conductor, XLPE insulation with semi-conducting screens (resp. 2 and 1 mm for inner and outer screen), a copper screen and outer insulation. The cable screen is directly grounded at both sides. Protective inductors in series with the cables as presented in [29] are not considered. In the pre-fault situation, a current of 1 ka flows from converter 2 to converter 1. A pole-to-ground fault on the positive pole is simulated by connecting the positive cable conductor to the cable screen and earth. A solid pole-to-ground fault is applied at the middle of the positive pole, i.e. at 2 km from the converter stations. Converter IBTs are blocked when the current through a converter arm exceeds ka. B. Results The results of the transient simulations are consecutively shown for the asymmetric and the symmetric configuration. 1) Asymmetric configuration: The grounding options considered for this configuration are solid grounding (resistor of 1 mω), resistance grounding (resistor of 1 Ω) and inductance grounding (inductor of mh). Figs. 3 and 4 show the currents and voltages measured at the positive pole and the metallic return at both cable ends when a fault is applied at 1 ms. Fig. 3 includes the case with DC capacitors, whereas for Fig. 4 the DC capacitors are omitted. The currents shown are the fault Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

5 (a) Positive pole current at grounded side (d) Positive pole current at ungrounded side (b) Positive pole voltage at grounded side (e) Positive pole voltage at ungrounded side (c) Voltage on metallic return at grounded side (f) Voltage on metallic return at ungrounded side Fig. 3. Voltages and currents after a positive pole-to-ground fault for different types of grounding for an asymmetric configuration (Solid line: Solid grounding, Dashed line: Inductance grounding, Dotted line: Resistance grounding). DC capacitors of 1 µf at each converter terminal (a) Positive pole current at grounded side (d) Positive pole current at ungrounded side (b) Positive pole voltage at grounded side (e) Positive pole voltage at ungrounded side (c) Voltage on metallic return at grounded side (f) Voltage on metallic return at ungrounded side Fig. 4. Voltages and currents after a positive pole-to-ground fault for different types of grounding for an asymmetric configuration (Solid line: Solid grounding, Dashed line: Inductance grounding, Dotted line: Resistance grounding). No DC capacitors at the converter terminals. currents, obtained by substracting the pre-fault current from the post-fault current. The current at the grounded converter side (Fig. 3(a)) increases steeply after the travelling wave caused by the fault reaches the terminal. In the first phase, the cable and the DC capacitor discharge into the fault. Simultaneously, the submodule capacitors start to discharge. By blocking the IBTs, the submodule capacitors are prevented from being fully discharged and the fault is fed through the antiparallel diodes of the IBTs. At the ungrounded side (Fig. 3(d)), the first current peak is half of that of the grounded side because of difference in line termination. Moreover, the fault current behavior at the ungrounded side is the same for each grounding type for the first milliseconds. This is a consequence of the travel time of a wave over the negative pole cable. The positive pole voltages at both grounded and ungrounded side reach a steady-state value close to zero (Figs. 3(b) and 3(e)). There is however a large difference between the negative voltages on the metallic return at the grounded and ungrounded side (Figs. 3(c) and 3(f)). The negative voltage at the ungrounded side reaches a value of -9 kv because of the voltage drop caused by the fault current flowing over the metallic return cable. The type of grounding influences the transient fault behavior as shown in Fig. 3. Resistance grounding attenuates the peaks in the transient fault current and limits the steady-state fault current. However, the steady-state post-fault voltage at the ungrounded side reaches -16 kv. Even at the grounded side, the negative pole voltage reaches -1 kv because of the total fault current flowing through the grounding resistor. With inductance grounding, the rate-of-rise and first peak of the fault current is reduced. The ideal inductor has no effect on steady-state currents and voltages. For the MMC, the DC capacitors can be further downsized or even omitted [7]. The results of the transient simulations for the case without DC capacitors is depicted in Fig. 4. In Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

6 1 Solid Inductor Resistor 1 Ungrounded Capacitor No Capacitors (a) Positive pole currents after a pole-to-ground fault Solid Inductor Resistor (c) Positive and negative pole voltages after a pole-to-ground fault (b) Positive pole currents after a pole-to-ground fault Ungrounded Capacitor No Capacitors (d) Positive and negative pole voltages after a pole-to-ground fault Fig.. Voltages and currents after a pole-to-ground fault for different types of grounding for a symmetric configuration. this case, the DC current is initially supplied by the converter submodule capacitors. The rate of rise of the current is lower due to the converter arm reactors. Because of the absence of the large DC capacitors, the DC voltage collapses faster. Blocking of the IBTs stops the discharge of the submodule capacitors and hence reduces the initial peak current. In the steady-state post-fault phase, the six-pulse behavior of the uncontrolled rectifier is more noticeable than for the case with large capacitors. Although that the transient behavior differs to a certain extent, the conclusions regarding grounding for the case without DC capacitors and with DC capacitors are the same. 2) Symmetric configuration: This section discusses the transients for a pole-to-ground fault for the different grounding options for a symmetric configuration (Fig. 1(b)). The system has a pole-to-pole voltage of 32 kv and a pole-to-ground voltage of 16 kv. The converters are both grounded at the midpoint of the DC capacitors. The options considered for low impedance grounding are solid grounding, grounding through a resistor of 1 Ω or an inductor of mh (see Figs. (a) and (c)). For high impedance grounding, a capacitor of 1 µf, high resistance grounding and leaving the system ungrounded are considered (Figs. (b) and (d)). Because of high degree of similarity between the high resistance grounded and ungrounded case, only the latter is shown in the figure. As both converters are for each simulation identically grounded, the voltages and currents at only one side are shown. In Fig. (a), the current measured at the positive pole is shown for solid, resistance and inductance grounding. In case of solid grounding, the current of the positive pole increases steeply after the incidence of the travelling wave caused by the fault because of the discharge of the DC capacitors. On a longer timescale, the current shows a damped oscillation, eventually decaying to zero. During the transient, the current within the converter remains limited. In the early transient, a discontinuity can be noticed each time a travelling wave reaches the terminal. Because of the DC capacitors, the voltage decay is rather smooth in comparison to the current (Fig. (c)). The steady-state post-fault voltage is zero on the positive pole and -32 kv (nominal converter voltage) on the negative pole. The transient behavior shows no significant excursions from these voltages. The DC capacitor of the faulted pole is fully discharged, while the capacitor of the healthy pole carries the nominal converter voltage. The transient behavior is influenced by type of grounding. The capacitors can be low impedance grounded through a small resistance or an inductance (Figs. (a) and (c)). Similar to the asymmetric case, the peak currents are attenuated by resistance grounding whereas the rate-of-rise is limited when inductance grounded (Fig. (a)). On a longer timescale, the resistance grounding forms an additional damping element in the damped oscillation. With a grounding inductor, the first positive oscillation has a lower maximum and the frequency of the oscillation has changed. The voltages (shown in Fig. (c)) for both resistance and inductance grounding show a less smooth behavior in comparison to solid grounding. Alternatively, high impedance grounding can be applied (Figs. (b) and (d)). If the capacitors are ungrounded, the main contribution to the fault current seen in Fig. (b) is the discharge of the negative pole cable. Hence, the fault current is much lower than when the DC capacitors are solidly grounded. The transient voltage excursions are somewhat higher than for the solidly grounded system (Fig. (d)). In steady-state, the voltage on the healthy pole is the nominal converter voltage, more or less equally distributed over the DC capacitors. High resistance grounding of the capacitors gives very similar results for the transient behavior as the ungrounded case. However, in this case, the DC capacitor of the healthy pole carries the nominal voltage in the post-fault steady-state. When grounding through a capacitor, the effective capacitance in case of ground faults is decreased because this capacitor is in series with the capacitor in case of pole- Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

7 6 B1 D1 B D B D A1 C1 A C A C B2 D2 A2 C2 (a) Asymmetric monopolar grid with earth return (b) Symmetric monopolar grid (c) Bipolar grid with metallic return B1 D B1 B1 D1 A1 C1 A1 C1 D A1 C1 B2 B2 B2 D2 A2 C2 A2 C2 A2 C2 (d) Bipolar grid with metallic return and asymmetric monopolar tappings Fig. 6. Possible VSC HVDC grid configurations (e) Bipolar grid with metallic return and symmetric monopolar tappings (f) Bipolar grid with metallic return and bipolar tappings with earth return to-ground faults. This slightly diminishes the first peak of the current in the positive pole and accelerates the decay in comparison to solid grounding. In steady-state, the grounding capacitor clamps a voltage on the midpoint, depending on its size. The nominal voltage is in this case shared between the DC capacitors as well. In case the DC capacitors are omitted, the DC system is ungrounded. The fault voltage and current are plotted in Figs. (d) and (b). The major contribution to the fault current for pole-to-ground faults is the discharge of the cables. The fault behavior is similar to the ungrounded case with DC capacitors. 3) Conclusion of simulations: The transient fault current in low impedance grounded systems is largely influenced by the grounding configuration. Due to the high rate of rise, protection schemes for low impedance grounded grids must act on a very short timescale. As grounding of the converter is a determining factor for the first transient, it is important to adapt settings for the protection scheme whether the converter is grounded or not. For a symmetric configuration, the transient fault current is mainly delivered by discharge of DC capacitors and cables. The grounding impedance has an impact on the maximum fault current. A main factor affecting the maximum voltages on each component is whether the neutral point in the symmetrical system is low or high impedance grounded. IV. ROUNDIN AND LAY-OUT OF HVDC RIDS This section evaluates the different options for the layout of a HVDC grid. The type of base configuration and grounding as well as the location of grounding in the grid are discussed. The base configurations are compared in terms of extensibility and flexibility for post-fault operation. Furthermore, also the availability of the different configurations is described. Beside the grounding type, the number and location of grounding points in the DC grid is an important factor. This has an influence on the presence of earth currents and the dimension of the earth electrodes [3]. In an European overlay grid, earth currents might need to be avoided as they can endanger human safety, enter the AC grid [31] or have environmental impacts [32]. A. Asymmetric monopolar grid In an asymmetric monopolar grid (Fig. 6(a)), the metallic return conductor can have a lower voltage rating because of low impedance grounding. For this type of grids, a fast acting fault clearance is needed because of high fault currents. In case of faults, the whole faulted link is lost. In case of multiple grounding points in the system, earth currents will flow in normal operation. These currents can be restricted by using resistance grounding. Alternatively, the grid can be low impedance grounded at only one single point and high impedance grounded elsewhere (e.g. as proposed in [33] and [34]). Several backup grounding points with an active device are required in case of an outage of the effectively grounded point. Moreover, the grounding points need to be dimensioned to sustain the total fault current supplied by all converters. A drawback is different fault behavior in the system depending on the location of the low impedance grounding. Extensive fault studies might be needed for every fault situation and grounding point to determine the correct settings for protective devices. Additionally, the location of the low impedance grounding influences the voltage rating of cables and converters. As both conductors carry the nominal current, large voltage drops can occur in the system. Converters and cables remote from a grounding point have to be rated for these voltages. The asymmetric monopolar grid is extensible with asymmetric monopoles or can be extended to a bipolar configuration. B. Symmetric monopolar grid In Fig. 6(b), a symmetric monopolar grid is shown. The grid can be grounded at the midpoint of the DC capacitors at each converter. Alternatively, the grounding point can be made available using other large impedances. For any of the grounding options, the system is effectively high impedance Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

8 7 P 1 pu P 1 pu V = +1 pu V = pu V = 1 pu A1 A2 Fig. 7. Three terminal DC system C1 C2 P =. pu P =. pu B1 B2 P =. pu P =. pu grounded. Time constraints for the protection of the symmetric monopolar grid can be less stringent as there is only a transient fault current. Analogous to the asymmetric monopolar grid, the whole link is lost in case of faults. No steady-state earth currents flow in normal operation when the system is grounded at multiple points. For standardisation of e.g. converter insulation requirements or protection settings, it is conceivable that each converter is grounded in the same manner. A symmetric monopolar grid is, besides other symmetric monopoles, extensible with high impedance grounded bipoles. Every extension must have cables and converters rated to carry the nominal converter voltage. This cost can be high in comparison to the power rating of the extension (e.g. a small wind farm). C. Bipolar grid Fig. 6(c) shows a bipolar grid configuration. In this scheme, both positive and negative pole are operated at nominal voltage, while a metallic return conductor is operated at nearzero voltage. The bipolar grid can be low or high impedance grounded. In contrast to a monopolar grid, still half of the faulted link is available in case of pole-to-ground faults. By unbalanced operation of the bipolar grid, more flexibility is available for post-fault operation compared to the monopolar grids, if the faulted section of the bipolar grid was initially partially loaded. A low impedance grounded bipolar grid is similar to an asymmetric monopolar grid regarding voltage rating of the poles and the need for fast acting fault clearance. Compared to the asymmetric monopolar grid, the power that can be transported over a link is doubled at the cost of only one extra cable. The bipolar grid can be low impedance grounded at multiple locations. In balanced steady-state operation, no (or only small) earth currents flow. Analogous to the asymmetric monopolar grid, earth currents in unbalanced operation can be limited by resistance grounding or low impedance grounding at only one point. A low impedance grounded bipolar grid is extensible with other low impedance grounded bipoles or asymmetric tappings between the metallic return conductor and one pole (Fig. 6(d)). The latter option provides a possibility to extend the grid by smaller rated systems at a reduced cost with respect to a symmetric grid. However, grid control can Current (pu) Losses (pu) Pole cables Metallic return Negative Converter Power Share (%) (a) Maximum current Negative Converter Power Share (%) (b) Network losses Fig. 8. Currents for different converter load sharing settings be more difficult as the system will be continuously operated in unbalanced conditions. The high impedance grounded bipolar grid is similar to the symmetric monopolar grid when considering cable voltage ratings, protection requirements and grounding locations. The metallic return must be rated at the nominal converter voltage. Moreover, if the cable voltage ratings are the same as used in the monopolar grids, the nominal converter voltage must be halved compared to monopolar grids. The high impedance grounded bipolar grid is extensible with other high impedance grounded bipoles, symmetric monopoles (Fig. 6(e)) and asymmetric tappings (Fig. 6(d)). However, insulation requirements of these tappings will be higher compared to asymmetric tappings on a low impedance grounded bipolar grid. Finally, a bipolar grid with metallic return cables can be extended by bipolar tappings with earth return (Fig. 6(f)). In case of unbalanced operation, earth currents are possible because of the low impedance grounded tapping. D. Case study Steady-state post-fault operation for the different grid types is illustrated using the simple three-terminal network shown in Fig. 7. The base configuration for the grid in Fig. 7 is a bipole, since the results of this configuration for balanced operation can be extended to monopolar configurations. The converters are modeled as simple voltage sources and the lines are represented by a resistance of. pu (U b = 32 kv, P b = 1 MW). The pre-fault operation settings are shown in Fig. 7. Converter A1 and A2 act as voltage regulators and set their voltages to 1 and -1 pu. At terminals B and C, 1 pu power is extracted, equally divided between both converters of the bipole. In normal operation, the metallic conductors carry no current. Hence, this solution could also be achieved using a double symmetric or asymmetric grid configuration. After a line outage between converters A1 and B1, different solutions for post-fault steady-state operation are possible. A first possibility is to continue operation with the same settings as before the fault occurred. The negative pole between Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

9 8 terminals B and C carries no current. Power is transported from terminal A to terminal B via the path A1-C1-B1-B2-A2. As the positive pole between terminals A and C is shared for power transport to both converters, this pole is most loaded. The metallic return conductors carry no current in this situation. So in terms of flexibility for post-fault operation, a symmetric operated bipolar configuration offers no advantage with respect to monopolar configurations. Another possibility is to change the power settings at converters B and C. The power setting of the negative pole converter can be increased while the positive pole converter s power setting is accordingly decreased. In this unbalanced operation, the metallic return conductors start carrying current while the current through the lines is decreased. To find an optimal setting, different objectives can be used, e.g. minimize maximal current in the network or minimize network losses. In Fig. 8, maximum current, network losses and currents through the metallic return are plotted for different degrees of converter power sharing. At A, pre-fault power settings are applied. It is clear that both objectives are in favor of unequal converter sharing, but optima are reached at different levels. The choice of operation interacts with the type of grounding, as described earlier. In the first case, multiple converters can be solid or low impedance grounded without significant steadystate earth currents. In the second case, earth currents can flow when multiple converters are low impedance grounded. Either the system is operated with only one solid grounded converter or an active device is used to switch from multiple grounded converters in pre-fault situation to only one solid grounded converter post-fault, to avoid large earth currents. V. CONCLUSION In this paper, an overview of possible grounding options and grid configurations for the future meshed HVDC grid is presented. A summary is given in Table III, indicating that grounding type and grid configuration are closely connected and have a direct impact on the HVDC grid design and protection. An important factor is whether the system will be low or high impedance grounded. Besides the grounding type, the number and location of grounding points in the system must be chosen. These choices affect cable insulation requirements, fault behavior after pole-to-ground faults and earth currents in normal operation. By performing transient simulations, the effect of different grounding impedances is demonstrated. As transient behavior is influenced by the grounding impedance it influences system protection requirements. The grid configuration is linked with the type of grounding. Asymmetric grid configurations can only be low impedance grounded, whereas symmetric grid configurations are effectively high impedance grounded, either through capacitors or other large impedances. The bipolar grid configuration can be either low or high impedance grounded. The grid configuration has an impact on extensibility of the grid, as not all configurations are compatible for interconnection. It also influences flexibility for post-fault operation. As the bipolar configuration has inherent redundancy, it offers higher flexibility for postfault operation than the monopolar configurations. APPENDIX TABLE I CABLE PARAMETERS Outer radius [mm] ρ [Ω m] ǫ r [-] µ r [-] Copper Core XLPE Insulation Copper Screen PE Insulation TABLE II RID AND CONVERTER PARAMETERS AC Primary Voltage V ac 4 kv AC Inductance L ac.1367 H AC Resistance R ac 3.78 Ω Transformer Ratio 4/18 kv Transformer Leakage Reactance u k.1 p.u. DC Voltage V dc 32 kv Converter Reactor 1 mh Converter Resistance.1 Ω Number of Submodules 1 Submodule Capacitance 1 mf REFERENCES [1] ENTSO-E, 1-Year Network Development Plan 212, ENTSO-E, Tech. Rep., 212. [2] J. De Decker and P. Kreutzkamp, Offshore Electricity rid Infrastructure in Europe, Final Report, Offshorerid, Tech. Rep., 211. [3] D. Van Hertem and M. handhari, Multi-terminal VSC HVDC for the European supergrid: Obstacles, Renew. Sust. Energ. Rev., vol. 14, no. 9, pp , 21. [4] E. Koldby and M. Hyttinen, Challenges on the road to an offshore HVDC grid, in Proc. Nordic Wind Power Conf., Bornholm, Denmark, 1-11 Sep. 29, 8 pages. [] L. Zhang, L. Harnefors, and H.-P. Nee, Modeling and Control of VSC- HVDC Links Connected to Island Systems, IEEE Trans. Power Syst., vol. 26, pp , may 211. [6] J. Arrillaga, Y. Liu, and N. Watson, Flexible Power Transmission: The HVDC Options. Hoboken, NJ, USA: John Wiley & Sons, 27. [7] S. Allebrod, R. Hamerski, and R. Marquardt, New transformerless, scalable Modular Multilevel Converters for HVDC-transmission, in Proc. IEEE PESC 8, Rhodes, reece, 1-19 Jun. 28, pp [8] W B4-2, HVDC rid Feasibility Study, Cigré, Tech. Rep., 212. [9] S. De Boeck, P. Tielens, W. Leterme, and D. Van Hertem, Configurations and Earthing of HVDC rids, in Proc. IEEE PES M 213, Vancouver, Canada, 21-2 Jul. 213, pages. [1] M. K. Bucher, M. M. Walter, M. Pfeiffer, and C. Franck, Options for ground fault clearance in HVDC offshore networks, in Proc. IEEE ECCE, Raleigh, NC, USA, 1-2 Sep. 212, pp [11] M. Bucher, R. Wiget,. Andersson, and C. Franck, Multiterminal HVDC Networks: What is the Preferred Topology? IEEE Trans. Power Del., vol. 29, no. 1, pp , Feb 214. [12] X. Chen, C. Zhao, and C. Cao, Research on the fault characteristics of HVDC based on modular multilevel converter, in Proc. IEEE EPEC, Winnipeg, Canada, 3- Oct. 211, pp [13] J. Rafferty, L. Xu, and D. Morrow, DC fault analysis of VSC based multi-terminal HVDC systems, in Proc. IET ACDC 212, Birmingham, UK, 4- Dec. 212, 6 pages. [14] K. De Kerf, K. Srivastava, M. Reza, D. Bekaert, S. Cole, D. Van Hertem, and R. Belmans, Wavelet-based protection strategy for DC faults in multi-terminal VSC HVDC systems, IET ener. Transm. Distrib., vol., no. 4, pp , 211. [1] O. Nanayakkara, A. Rajapakse, and R. Wachal, Traveling-Wave-Based Line Fault Location in Star-Connected Multiterminal HVDC Systems, IEEE Trans. Power Del., vol. 27, pp , Oct [16] J. Yang, J. Fletcher, and J. O Reilly, Short-Circuit and round Fault Analyses and Location in VSC-Based DC Network Cables, IEEE Trans. Ind. Electron., vol. 9, pp , Oct Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must

10 9 TABLE III OVERVIEW OF HVDC RID CONFIURATIONS Asymmetric monopolar grid Symmetric monopolar grid Bipolar grid Operating voltages a, U n U n/2, U n/2 U n,, U n U n/2,, U n/2 Cables/P b 2 2 3/2 3 rounding type Low impedance High impedance c Low Impedance High Impedance rounding points Single grounding point Multiple grounding points Single grounding point Multiple grounding points Multiple grounding points d Multiple grounding points d Cable Voltage Rating, U n U n, U n U n,, U n U n, ±U n/2, U n Steady-state fault current Large Zero Large Zero Protection requirements Fast acting protection Less stringent time constraints Fast acting protection Less stringent constraints Extensibility Asymmetric monopoles Symmetric monopoles Low impedance grounded bipoles High impedance grounded bipoles Post-fault flexibility after loss of cable Upgrade to bipolar grid High impedance grounded bipoles Asymmetric monopoles between metallic return and pole Low Low High High Asymmetric monopoles between metallic return and pole a Pole-to-ground voltage (U ptg) b Converter power P U ptg c rounded through DC capacitors or resistors d Earth currents possible [17] O. omis-bellmunt, J. Liang, J. Ekanayake, R. King, and N. Jenkins, Topologies of multiterminal HVDC-VSC transmission for large offshore wind farms, Electr. Power Syst. Res., vol. 81, no. 2, pp , 211. [18] PSCAD, PSCAD User s manual, Manitoba HVDC Research Centre, 21. [Online]. Available: reference-material [19] ABB, HVDC Light: It s time to connect, ABB, Tech. Rep., 212. [2] J. Candelaria and J.-D. Park, VSC-HVDC system protection: A review of current methods, in Proc. IEEE PSCE 211, Phoenix, AZ, USA, 2-23 Mar. 211, 7 pages. [21] T. önen, Electrical Power Transmission System Engineering Analysis and Design, 2nd Edition. Boca Raton, FL, USA: CRC Press, 29. [22] A. Lesnicar and R. Marquardt, An innovative modular multilevel converter topology suitable for a wide power range, in Proc. IEEE PowerTech 23, Bologna, Italy, Jun. 23, 6 pages. [23] U. nanarathna, A. ole, and R. Jayasinghe, Efficient Modeling of Modular Multilevel HVDC Converters (MMC) on Electromagnetic Transient Simulation Programs, IEEE Trans. Power Del., vol. 26, pp , Oct [24] H. Saad, J. Peralta, S. Dennetiere, J. Mahseredjian, J. Jatskevich, J. Martinez, A. Davoudi, M. Saeedifard, V. Sood, X. Wang, J. Cano, and A. Mehrizi-Sani, Dynamic Averaged and Simplified Models for MMC-Based HVDC Transmission Systems, IEEE Trans. Power Del., vol. 28, pp , Apr [2] U. nanarathna, S. Chaudhary, A. ole, and R. Teodorescu, Modular multi-level converter based HVDC system for grid connection of offshore wind power plant, in Proc. IET ACDC 21, London, UK, Oct. 21, pages. [26]. Adam,. Kalcon, S. Finney, D. Holliday, O. Anaya-Lara, and B. Williams, HVDC Network: dc fault ride-through improvement, in CIRÉ Canada Conf. on Power Sys., Halifax, Canada, 6-8 Sept. 211, 8 pages. [27] J. Peralta, H. Saad, S. Dennetiere, J. Mahseredjian, and S. Nguefeu, Detailed and averaged models for a 41-Levels MMC -HVDC system, IEEE Trans. Power Del., vol. 27, pp , Apr [28] PSCAD, PSCAD EMTDC User s uide, Manitoba HVDC Research Centre, 21. [Online]. Available: reference-material [29] F. Deng and Z. Chen, Design of Protective Inductors for HVDC Transmission Line Within DC rid Offshore Wind Farms, IEEE Trans. Power Del., vol. 28, pp. 7 83, Jan [3] E. Kimbark, Direct current transmission: Volume I. Hoboken, NJ, USA: John Wiley & Sons, [31] R. Zeng, Z. Yu, J. He, B. Zhang, and B. Niu, Study on Restraining DC Neutral Current of Transformer During HVDC Monopolar Operation, IEEE Trans. Power Del., vol. 26, pp , Aug [32] T.. Magg, H. D. Mutschler, S. Nyberg, J. Wasborg, H. Thunehed, and B. Sandberg, Caprivi Link HVDC Interconnector: Site selection, geophysical investigations, interference impacts and design of the earth electrodes, Cigré, Tech. Rep., 21. [33] N. ibo, K. Takenaka, S. Verma, S. Sugimoto, and S. Ogawa, Protection scheme of voltage sourced converters based HVDC system under DC fault, in Proc. IEEE Power Eng. Soc. Asia Pacific Transmission and Distribution Conf. Exhibit., 6-1 Oct. 22, pp [34] M. Takasaki, N. ibo, K. Takenaka, T. Hayashi, H. Konishi, S. Tanaka, and H. Ito, Control and protection scheme of HVDC system with selfcommutated converter in system fault conditions, Elect. Eng. in Japan, vol. 132, no. 2, pp. 6 18, 2. Willem Leterme (S 12) was born in 1989 in Izegem, Belgium. He received the M.Sc. degree in electrical energy engineering from KU Leuven, Leuven, Belgium in 212. Currently, he is working at KU Leuven in the research group Electa, dept. of Electrical Engineering, to obtain a Ph.D. degree. His research interests include transient fault studies and design of protection algorithms for meshed VSC HVDC grids. Pieter Tielens (S 11) was born in Belgium in He received the M.Sc. degree in electrical energy engineering from the University of Leuven (KU Leuven), Leuven, Belgium, in 211, where he is currently working towards a Ph.D. degree. He is a research assistant in the research group Electa, department of Electrical Engineering of the KU Leuven. His fields of interest are the integration of renewable energy resources into the electricity system and inertialess grids. Steven De Boeck (S 11) was born in 1988 in Diest, Belgium. He received the M.Sc. degree in electrical energy engineering from the University of Leuven (KU Leuven), Leuven, Belgium, in 211. Currently, he is working towards a Ph.D. degree at KU Leuven in the research group Electa, dept. of Electrical Engineering. His fields of interest are new methodologies to maintain the security of the power system and the use of PFCs, such as PST and HVDC, in emergency plans. Dirk Van Hertem (S 2-SM 9) was born in 1979, in Neerpelt, Belgium. He received the M.Eng. in 21 from the KHK, eel, Belgium and the M.Sc. in electrical engineering from the KU Leuven, Belgium in 23. In 29, he received the Ph.D., also from the KU Leuven. In 21, he was a member of EPS group at the Royal Institute of Technology, in Stockholm, Sweden, where he was the program manager for controllable power systems for the EKC 2 competence center. Since spring 211 he is back at the University of Leuven where he is an Assistant Professor in the Electa group. His special fields of interest are power system operation and control in systems with FACTS and HVDC and building the transmission system of the future, including offshore grids and the supergrid concept. He is an active member of IEEE PES and IAS and Cigré. Copyright c 214 IEEE. Personal use is permitted. For any other purposes, permission must