HVDC SuperGrids with Modular Multilevel Converters - the Power ...

2012 – 9th International Multi-Conference on Systems, Signals and Devices

HVDC SuperGrids with Modular Multilevel Converters - the Power Transmission Backbone of the Future Noman Ahmed1, Staffan Norrga1, Hans-Peter Nee1 1)

Electrical Energy Conversion (E2C) KTH Royal Institute of Technology Stockholm, Sweden

Abstract—In order to transmit massive amounts of power generated by remotely located power plants, especially offshore wind farms, and to balance the intermittent nature of renewable energy sources, the need for a stronger high voltage transmission grid is anticipated. Due to limitations in ac power transmission the most likable choice for such a grid is a high-voltage dc (HVDC) grid. However, the concept of the HVDC grid is still under active development as different technical challenges exist, and it is not yet possible to construct such a dc grid. This paper deals with prospects and technical challenges for future HVDC SuperGrids. Different topologies for a SuperGrid and the possibility to use modular multilevel converters (M2Cs) are presented. A comprehensive overview of different submodule implementations of M2C is given as well as a discussion on the choice between cables or overhead lines, the protection system for the dc grid and dc-side resonance issues. Keywords- HVDC grid, Multi-terminal, Multilevel Converters, Power Transmission, Voltage Source Converter (VSC)

I.

INTRODUCTION

In Europe the total electricity consumption increased by 32.8% from 1990 to 2007. The average annual growth rate of final electricity consumption across the EU-27 countries was 1.7% in this period [1]. This rate is not particularly high in comparison to countries in rapid economic expansion such as China, but it is causing concern, since the existing high-voltage ac (HVAC) grid in Europe is operated close to its limits. In combination with the trading mechanisms of free electricity markets, increased congestion problems are expected. Additionally, renewable energy sources such as wind power may generate power at other instants and other locations than those that are optimal from a power system perspective. From an energy perspective, hydro power could be used to balance the variable energy output of the renewable energy sources. However, such balancing actions would create even more congestion as the different energy sources are geographically spread. As most weather systems have a geographical distribution of approximately 1000 km, the effects of aggregation can only be accounted for if wind farms can be interconnected by a sufficiently strong grid covering several thousands of kilometers.

978-1-4673-1591-3/12/$31.00 ©2012 IEEE

Arif Haider2, Dirk Van Hertem3, Lidong Zhang4, Lennart Harnefors4,1 2) 3)

Etteplan Industry AB, Västerås, Sweden ESAT/ELECTA, K.U.Leuven, Belgium 4) ABB, Ludvika, Sweden

Given the current transmission system, massive congestion problems are foreseen. This can jeopardize the development of the internal European energy market, which aims to have a competitive and sustainable energy supply with sufficient security of supply. In order to make full use of all energy sources, an electric transmission system of substantially higher capacity than today is needed. A solution based on upgrading the existing HVAC grid is difficult especially in densely populated areas, because of public opposition and limited availability of right of way (ROW). In most of central Europe, and in many other locations, the only alternative for a fundamental upgrade of the power system is a high-voltage dc (HVDC) grid. Such a grid would allow a relative straightforward underground implementation of a grid. Until recently, no obvious converter candidate for such an HVDC SuperGrid was available. Depending on the power semiconductors used in the converter topology, the HVDC transmission can be classified as thyristor based linecommutated converters (LCCs) and insulated gate bipolar transistor (IGBT) based voltage source converters (VSCs). LCCs, having high efficiency, do not perform well in meshed grid connections as the voltage must be reversed if the direction of power is reversed. Additionally, LCCs suffer from being a burden to the interconnected HVAC grid because of the need for reactive power. Another problem related to the same phenomenon is the risk for commutation failures, normally during inverter operation [2]-[4]. VSCs, however, can change power flow through the reversal of dc current direction rather than voltage polarity. As a result, the VSC technology allows converters to connect to a common dc voltage, and cross-linked polyethylene (XLPE) cables can be used. VSCs can provide reactive power and are not prone to commutation failures. They also do not have as serious requirements for the short-circuit ratio of the connected ac system. They have smaller foot-print and allow black start. Since the late nineties, therefore, HVDC transmission based on two-level or three-level VSCs has been promoted and also installed in various parts of the world [2]-[4]. An alternative VSC topology is the modular multilevel converter (M2C) [5], which has the potential to have efficiencies higher than 99% and is also promising with respect

to reliability and fault handling. The M2C is highly scalable with respect to the number of levels. The number of submodules can be increased or decreased as the number of levels increases or decreases to get the desired output voltage. The aim of this paper is to describe the prospects and challenges of a potential future HVDC grid for Europe. The paper has been organized as follows. In Section II, the concept of SuperGrids is presented with different possible topologies. Section III deals with the possibility of replacing existing HVAC lines by HVDC lines. In Section IV, different HVDC converter topologies are described. Protection issues for SuperGrids are discussed in Section V. In Section VI, cable systems for the SuperGrid are discussed. Section VII deals with the dc-resonance problem in a meshed grid. Section VIII concludes this paper. II.

SUPERGRIDS

The SuperGrid is seen as a solution to allow massive integration of renewable energy sources into the energy system. As such it acts as a transmission system that securely spreads electric power across national borders or even larger [6], [7]. It serves as an additional transmission structure besides the existing HVAC grids, which would not be able to handle the amount of power if the projected amount of renewable resources should be connected. An upgrade is needed as they are already operated close to their limits. The up-grading of HVAC grids with overhead transmission lines in Europe is very difficult because of legal, political, and environmental concerns. The ac expansion options, both overhead and underground, are often limited by voltage or transient instability problems, increased short-circuit current levels and impacts of unaccepted network loop flows. The alternative solution to the problem of expanding power transfer capability in the future is an HVDC grid. Different topologies for possible SuperGrids are shown in Fig. 1 [7]. A. Grid Topologies A first potential topology (Fig. 1a) is a simple multiterminal system, which can be described as a dc bus with several tappings. While being the simplest form of a multiterminal HVDC system, without dc meshes and no redundancy in the dc system itself, it can hardly be considered a grid in itself as it offers no redundancy. Nevertheless, such a topology can be very useful as a component within a hybrid ac-dc grid to reinforce the ac system. The proposed SouthWest Link in Sweden is an example of such a setup. A second possible configuration (Fig. 1b) is to have a grid of independent dc lines, where all the nodes remain ac nodes. The transmission between those nodes is done using dc lines, which means that each circuit forms an independent dc line with two converters. In this topology, all dc lines are fully controllable. The grid can consist of a mixture of LCC and VSC HVDC lines, and different lines can operate at different voltages. This setup makes it quite straightforward to incorporate existing HVDC lines in the new dc grid. The set of HVDC lines between continental Europe and Scandinavia could be considered as (the beginning of) such a grid. The third topology (Fig. 1c) implies a meshed dc system,

Figure 1. Different possible topologies for SuperGrids

with a number of connections between the ac and dc system. The dc system consists of dc nodes connected to each other without converters in between. The dc system is generally meshed, and multiple paths between nodes are possible. This has clearly an influence on the reliability, but also in the way in which this system must be protected. The fourth topology (Fig. 1d) is a meshed dc system with additional converters (shown in red). These converters have their dc sides connected in series with the dc line and the ac side is connected to the ac grid via transformer T2 on either side of the main transformer T1 for the M2C, preferably the side with the lower voltage. The additional converter maybe either VSCs or LCCs. In the case of VSCs they could even be small M2Cs. From a cost and robustness point-of-view, however, a double thyristor converter may be preferable [8]. Fig. 2 shows an example using a double thyristor converter.

bipolar dc line, by using the same conductors, towers and foundations, making not only this conversion cheaper but also requires less time than any other methods of conversion used for upgrading [9], [10]. In the future it is expected that during the development of SuperGrids, conversion of ac lines into dc will get more interest especially where construction of new lines is not possible. IV.

HVDC CONVERTERS FOR SUPERGRIDS

Until recently, no obvious converter candidate for such HVDC SuperGrids was available but the following advantages of the VSC over LCC makes it more favorable.

Figure 2. Connection of thyristor converter with DC line

The voltage across capacitor C can be controlled by means of thyristor converter. By changing voltage across the capacitor the power-flow in the dc lines can be controlled, thus providing additional possibilities for power-flow control using a converter that has a rating which is only a small fraction of the M2C. The drawback of the second topology for a deployment at larger scale is that more converters are needed for each branch, unlike topologies 1 and 3, which need 1 per node. Also, with the increase in number of branches and with frequent changes in load flow, topology 2 is not advisable. Since converters are expensive and cause losses in the dc grid, topology 3 is advisable as it provides redundancy rather than only forming an alternative path to the ac grid. Topology 1 can be seen as first step in building a SuperGrid. Topology 4 provides additional possibilities for power-flow control, also in the dc grid, but with a lower cost than for topology 2. III.

REPLACEMENT OF EXISTING HVAC LINES WITH HVDC LINES

The SuperGrid is not something that will be built in one single stage. Similar to the construction of an ac grid, a possible scenario is to first build point-to-point HVDC connections which later may be connected and integrated into a dc SuperGrid. During this development process, some existing ac lines could be replaced by dc lines to make them an integral part of a SuperGrid. Another possibility is that after the accomplishment of the SuperGrid, some parts of the existing ac grid must be upgraded or replaced by dc lines to enhance their power transfer capability and system stability. In recent years, interest in converting existing ac lines to dc has grown rapidly. In Europe, where the construction of new overhead transmission lines is very difficult because of public opposition and less available space, converting an existing ac line into a dc line is an appealing solution to increase its power transfer capability. This can be done by two methods, either by replacing only ac insulators without any tower modification or by replacing conductors with a possible tower modification. A power upgrading is possible by converting a single circuit ac line into a monopolar dc line or a double circuit ac line into a

•

The direction of flow of power can be changed by changing the direction of current.

•

Independent control of active and reactive power.

•

Less severe requirements of short-circuit ratio for the connected ac system.

•

Forced commutation allows black start.

•

Smaller converter station foot-print.

•

Light weight and economical extruded polymer dc cables can be used.

Existing two-level and three-level VSC-HVDC converters have some technical complexities which increases with the increase in levels and voltage magnitude. For high voltage and high power applications these converters with high number of series connected IGBT devices also produce valve stresses and electromagnetic interference (EMI) problems if controlled improperly. Taking into account the requirements of SuperGrids, these converters show some drawbacks, which limit their performance and application in SuperGrids. •

High switching frequencies cause high switching losses and reduced efficiency.

•

The valve currents have very high slope which produces undesired EMI problems.

•

If short circuits at the dc bus are not disconnected immediately, the stored energy of the dc capacitor produces extremely high currents which cause damage to the semiconductor devices of the VSC.

A. Multilevel Converters In comparison to two-level converters, the switching frequency per switch of multilevel converters is substantially lower. Hence, multilevel converters may have significantly lower losses. The existing multilevel topologies are categorized as •

Diode clamped or neutral point clamped (NPC)

•

Flying capacitors or capacitor clamped (FC)

•

Cascaded H-bridge with separate dc sources

•

Modular multilevel cascaded bridge converter (M2C) (without separate dc sources)

For higher number of levels the NPC converter requires high numbers of clamping diodes, which makes its structure quite complex. Also voltage balancing is a challenge with a high number of levels. Similarly, for higher number of levels the capacitor clamped multilevel converter topology requires a high number of capacitors with very high overall energy storage capability. The capacitors hence become bulky and expensive. For active power transmission, the inverter control becomes complicated. Furthermore, the construction becomes complicated and more complex as the number of levels increases. For cascaded H-bridge converters finally, separate dc sources are needed which makes its application limited [11], [12]. B. Modular Multilevel Converters The M2C that was proposed by Marquardt and Lesnicar [5], is now getting popularity due to its advantages over the conventional topologies. As shown in Fig. 3(a), the M2C is highly scalable with respect to the number of levels. The basic component of an M2C is called a submodule. The number of submodules can be increased or decreased as the number of levels increases or decreases to get the desired output voltage. Three submodule topologies that have been proposed for M2C are the half bridge submodule (HBSM), full bridge submodule (FBSM) and clamp double submodule (CDSM) [13].

(a)

S1

1) HBSM: It is mainly composed of two IGBT switches, two anti-parallel diodes and a dc storage capacitor C0 as shown in Fig. 3(b). The terminal voltage of each submodule can either be switched to zero or a voltage V. 2) FBSM: It consists of four IGBTs with anti-parallel diodes and a capacitor as shown in Fig. 3(c). This topology allows positive, negative as well as zero voltages. 3) CDSM: In normal operation CDSM shown in Fig. 3(d) represents two equivalent HBSMs. The half bridges are connected positive terminal to negative terminal so the insert and bypass switch positions are interchanged. In an M2C these submodules are connected in cascaded style which gives the freedom to connect as many modules required without increasing the complexity of the circuit. The converter arm represents a controllable voltage source controlled with modulation techniques to get the desired output. The main advantages of M2C (regardless of submodule implementation) are [14] •

The internal arm currents are continuous.

•

The inductors inserted in the arms limit the ac-side current in case of a short circuit at the dc side.

•

No separate energy sources are required for submodule capacitors.

•

No common dc link capacitor is required.

One manufacturer makes use of stacked press-pack IGBTs. In this case M2C is called cascaded two-level (CTL) converter [15]. For two-level or three-level converters with series connected switches operating at higher voltages, voltage

C0 S2

(b)

(c)

S3

S1 C0

C0 S5

S4

S2

(d) Figure 3. (a) Modular multilevel converter (b) Half bridge submodule (c) Full bridge submodule (d) Clamp double submodule

balancing is a major challenge as it requires extremely precise control of hundreds of simultaneously switching devices within a valve. For M2C the equivalent task is to balance the voltage across the individual submodule but at a slower time scale. The advantages of M2C over other existing converter topologies

probably makes it most suitable candidate for HVDC grids. The losses in two-level VSC stations are around 1.6% of the rated transmission capacity and around 70% of these are dissipated in the IGBT valves [16]. With M2Cs the valve losses are expected to be much lower. The choice of submodule implementation has a significant effect on valve losses. In [17], it is shown that the conventional HBSM with IGBTs has around 0.5% valve losses of rated power, while CDSM and FBSM implementation approximately give 35% and 70% more losses respectively. Hence, with reference to losses, the HBSM implementation for M2Cs would be preferred in future SuperGrids. V.

PROTECTION ISSUES

The protection system may be one of the major challenges for the future SuperGrid. When a short-circuit occurs, immediately the entire dc grid discharges into the fault. The VSCs lose control as the IGBTs are blocked and the fault current continues to flow through the anti-parallel diodes across the IGBTs. The stored energy at the dc bus results in extremely high currents and subsequent damage if short circuits at the dc bus cannot be disconnected immediately [13], [14]. Even if the line inductance has an influence in the early stages of the fault, the only limit to the final fault current is the resistance of the dc lines, which should be as low as possible as this influences the losses and the rating of the line. The availability of the SuperGrid depends on the capability to withstand dc faults. Fast interruption of fault currents is essential for grid reliability. In a meshed system it is important to disconnect only the faulted part of the grid, but keep the remaining system operational [18]. Due to the rapid rise in current, the fault clearing (including detection) should happen within less than 5 ms. Within this time frame, a suitable protection system should provide: fault detection, correct identification of the fault location and selectivity in opening the correct dc breakers. At the same time, redundancy is needed. The short time interval does not allow for communication between measurement devices. Traditional protection devices for ac systems such as impedance relays cannot be used and the protection system technology for VSC-HVDC which is currently available is able to protect point-to-point connections, and can be extended to protect simple multi-terminal schemes. However, for the protection of meshed HVDC grids, there is no suitable protection method available and a fundamentally new protection methodology is needed. The HBSM is the commonly used submodule implementation for M2C, but for a meshed grid this topology requires dc circuit breakers as the conventional method used to handle dc-side faults by ac breakers requires long clearing times. The dc breaker must be significantly faster than ac breakers used today so that the prospective fault current does not cause any damage to the components involved. The only known breakers to meet these requirements employ semiconductor switches to interrupt the current. This may imply significant losses in normal operation due to the on-state voltage drop of the semiconductor switches. Recently however, a hybrid semiconductor HVDC breaker has been proposed

Fast Disconnector

Auxiliary DC Breaker

Current Limiting Reactor Residual DC Current Breaker

Main DC Breaker

Figure 4. Modular hybrid dc breaker

[19]. The main benefit of this concept is that the current does not pass through the main semiconductor switch in normal operation, whereby the losses can be kept very low as shown in Fig. 4. In the future, it is anticipated that fast and reliable dc breakers will be commercially available for high voltages and currents. Considering this, the HBSM will probably be the topology used in M2C for SuperGrids as it has the lowest losses and the lowest installation costs. When using the FBSM submodule topology, the shortcircuit problem is less severe. The FBSM has the ability to cut off arm currents of any direction by impressing the appropriate polarity of terminal voltages in the arm. Compared to the HBSM, the additional switching states are not very useful in normal operation, because a reverse voltage polarity is not required at the dc-bus for HVDC applications. A severe drawback of FBSM is its losses, which are almost doubled as compared to HBSM due to doubling of the number of semiconductors used. Another drawback is higher initial cost due to a high number of semiconductor components. This, however, has to be put into relation to the cost of a potential dc circuit breaker, which would not be necessary with a FBSM implementation of the M2Cs. It has also to be kept in mind that an FBSM implementation is probably more expensive than an HBSM converter plus a dc circuit breaker. The CDSM is the most recently proposed submodule topology for the M2C. In [13] it is shown that the CDSM can decrease the fault current to zero within 1 ms after detection, without tripping the ac-side switches, which makes all the converters fully operable for immediate restarting. This new topology seems favorable for converters in a meshed dc grid as it does not require as many IGBTs as required in the FBSM implementation. This topology has 35% more losses than a conventional HBSM but this has to be put against possibly reduced requirement for arm inductance, reducing the cost of the converter station. VI.

CABLES AND OVERHEAD LINES

In the future HVDC SuperGrid it is not obvious that only cables or only overhead lines should be used. However, in general the possibility of using overhead lines is very limited considering environmental concerns, public opposition and limited ROW. Also, with overhead lines, there is a higher risk of flash over due to lightning. The permitting process of laying underground cables is much easier than building overhead

Figure 6. Passive damping circuit

Figure 5. Transmission capacity (available or announced) for HVDC systems

field strength of the cable. In the future voltage ratings up to 800 kV are anticipated. VII. DC-SIDE RESONANCE

lines. Two major types of cables namely mass impregnated (MI) and Extruded cables are available for HVDC transmission. MI cables are insulated with special paper and impregnated with a high viscosity compound. Extruded cables are insulated with extruded-polyethylene-based compounds. Paper insulated cables impregnated with low viscosity oil also exist, but are limited to land cables or short submarine crossings, as they need hydraulic systems for oil feeding. The choice of cable technology for offshore HVDC transmission depends on whether the polarity should be changed (as for LCC-HVDC) or not. If the polarity of the voltage should be changed MI cables should be used. However, if a considerable part of the cable should be on-shore, MI cables are usually avoided due to the moist-shielding led layer. Apart from these requirements, the choice between extruded cables and MI is free. The latter is more expensive due to a more complex design and difficult handling. Fig. 5 shows a comparison of power transfer capacity of LCC-HVDC and VSC-HVDC using overhead lines and cables. The maximum voltage ratings for the two types differ substantially. For extruded cables the current maximum voltage rating is 320 kV, whereas the corresponding value is 500 kV for MI cables. For bulk power transmission the MI cable has proven to be more reliable but requires long jointing times. For extruded cables the jointing time is much less due to prefabricated joints [20]. For the anticipated VSC-HVDC SuperGrids, extruded cables are preferred due to their low weight and easy installation for both land and submarine applications. SuperGrids demand transmission of several GW of power. With the current status of cable technology that demand requires parallel circuits. Use of copper conductors is possible, but this increases the cost. Additionally, with copper conductors the weight is approximately three times higher compared to a cable with aluminium conductors having the same transmission capacity. If the use of aluminium conductors is foreseen, the transmission capacity can be increased either by increasing the conductor area or by increasing the voltage. The conductor area cannot be increased beyond a certain limit for practical reasons, probably 3000 mm2 [21]. Thus, the only possibility is to increase the transmission voltage. This can be done by increasing the insulation thickness or by increasing the electric

The problem of dc-side resonance has been discovered in point-to-point VSC-HVDC transmissions, where the inductance of the dc transmission line and the dc capacitance at the VSC terminals create a dc resonance circuit. The dc resonance is only of concern if the resonant frequency is lower than the fundamental frequency. The dc-side resonance excites oscillations in the dc circuit which impose a bandwidth limitation on the dc voltage controller. A common solution to the dc resonance problem for a point-to-point VSC-HVDC link is to apply a notch filter in the direct-voltage control loop to reduce the resonance peak [22]. In a highly meshed HVDC grid many resonances can be excited. This is foreseen to cause problems whenever abrupt steps in power are imposed on the grid. Especially, faults may generate severe oscillations. Because of the limited line resistance, these oscillations are poorly damped. In order to handle the oscillations two basic groups of measures can be taken. •

Passive damping circuit, see Fig. 6

•

Converter control

Passive damping circuits are lossy and costly and must, therefore be minimized. Active control of the dc-side voltage by means of the converters might be very challenging, but should be investigated. Another reasonable alternative is to investigate the passive behaviour of the converter with respect to dc-side oscillations. VIII. CONCLUSIONS The concept of SuperGrids allows massive integration of remote sources of renewable energy. The unavailability of dc breakers that enable isolation of faulted segments in SuperGrids is a major challenge to overcome. A protection system (including dc breaker) with reasonable operating time (< 5 ms) is required for the development of SuperGrids. Limitation of cable systems for higher ratings is another challenge. Building a meshed HVDC grid is very complicated. Technically VSC-HVDC with M2Cs makes it more realistic. It is the most probable technology to be used especially if a cable-based grid is considered. With expected higher number of branches and frequent changes in load flow, topology 3

presented in the paper is advisable for SuperGrids. Topology 4 can also be used as it provides additional power-flow possibilities. From a loss point-of-view the HBSM implementation for the M2C performs well, but short circuit issues might call for another submodule implementation. For longer transmission distances dc cables are the only solution for going underground and sea. It is expected that dc cables having desired voltage rating for SuperGrid will be available in the future. This will certainly be a great benefit when designing SuperGrids. Development of SuperGrids needs new power corridors. Hence, in the future replacement of ac lines with dc lines will receive more interest. For building SuperGrids dcside resonance problems should also be investigated. It is likely that SuperGrids will be used for interconnection of offshore generation but they can also be used for strengthening the onshore network. The SuperGrid is not something for the immediate future. Similar to the construction of an ac grid, a possible scenario is to first build point-to-point HVDC connections which later may be connected and integrated into a SuperGrid.

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

REFERENCES [1] [2]

[3]

[4]

[5]

[6] [7]

ENER18 Electricity consumption, European Environment Agency. M. P. Bahrman and B. K. Johnson, “The ABCs of HVDC transmission technologies,”, Power and Energy Magazine, IEEE Vol. 5 , Issue. 2, 2007 , pp. 32 – 44 N. Flourentzou, V. G. Agelidis and G. D. Demetriades, “VSC-based HVDC power transmission systems: An overview,” IEEE Transactions on Power Electronics, Vol. 24, No.3, March 2009. J. Pan, R. Nuqui, K. Srivastava, T. Jonsson, P. Homberg and Y. J. Hafner, “Ac grid with embedded VSC HVDC for secure and efficient power delivery,” IEEE Energy 2030, Atlanta, GA USA, 17-18 November, 2008. A. Lesnicar and R. Marquardt, “An innovative modular multilevel converter topology suitable for a wide power range,” IEEE PowerTech Conference, Bologna, Italy, June 23-26, 2003. S. Gordon, “Supergrid to the rescue,” Power Engineer 2006;20(5), pp. 30-33 D. Van Hertem and M. Ghandhari, “Multi-terminal VSC HVDC for the European supergrid: obstacles,” Renewable & Sustainable Energy Reviews , Vol. 14, Issue 9, December 2010, pp. 3156-3163

[16] [17]

[18]

[19]

[20]

[21]

[22]

H. Xie, L. Ängquist and H.-P. Nee, “Design study of a converter interface interconnecting energy storage with the dc-link of a StatCom,” IEEE Transactions on Power Delivery, Vol. 26, No. 4, October 2011. M. L. Matele, A. Clerici and G. Valtorta, “Power upgrading by converting ac lines to dc,” 3rd AFRICON Conference, 1992 , pp. 474– 478 A. Clerici, L. Paris and P. Danfors, “HVDC conversion of HVAC lines to provide substantial power upgrading,” IEEE Transactions on Power Delivery, Vol. 6, 1991, pp. 324–333 J.-S. Lai and F.Z. Peng, “Multilevel converters a new breed of power converters,” Industry Applications Conference, 1995. Thirtieth IAS Annual Meeting, IAS '95., IEEE 1995, Vol. 3, pp. 2348–2356 J. Rodríguez., J.-S. Lai and F. Z. Peng, “Multilevel inverters: A survey of topologies, controls, and applications,” IEEE Transactions on Industrial Electronics, Vol. 49, No. 4, August 2002. R. Marquardt, “Modular multilevel converter: An universal concept for HVDC-networks and extended dc-bus-applications,” IEEE International Power Electronics Conference, June 2010, Japan. S. Allebrod, R. Hamerski and R. Marquardt, “New transformerless, scalable modular multilevel converters for HVDC transmission,” IEEE Power Electronics Specialists Conference (PESC), Rhodes, Greece, June 15-19, 2008. B. Jacobson, P. Karlsson, G. Asplund, L. Harnefors and T. Jonsson, “VSC-HVDC transmission with cascaded two-level converters,” Cigré Session, Paris, France, 2010. HVDC Light transmission losses, http:www.abb.com, 2010. T. Modeer, H.-P. Nee and S. Norrga, “Loss comparison of different submodule implementations for modular multilevel converters in HVDC,” EPE 2011, Birmingham, UK. D. Van Hertem, M. Ghandhari, J. B. Curis, O. Despouys and A. Marzin, “Protection requirements for a multi-terminal meshed dc grid,” Cigré Symposium The electric power system of the future, Bologna, 2011. J. Häfner and B. Jacobson, “Proactive Hybrid HVDC breakers – A key innovation for reliable HVDC grids,” Cigré symposium, 13-15 September 2011, Bologna, Italy M. Jeroense, “HVDC, the next generation of transmission highlights with focus on extruded cable systems,” Proceedings of 2008 International Symposium on Electrical Insulating Materials, Sep. 7-11, 2008, Japan M. Jeroense, A. Gustafsson, P. Sunnergårdh, L. Carlsson, “The status and future of HVDC Light cable systems,” Cigre SC B4 2009 Bergen Colloquium. L. Zhang, L. Harnefors, H.-P. Nee, “Interconnection of two very weak ac systems by VSC-HVDC links using power-synchronization control,” IEEE Transactions on Power Systems, vol. 26, no.1, pp. 344-355, Feb. 2011.