VSC-HVDC System Protection A Review of Current Methods

1 VSC-HVDC System Protection: A Review of Current Methods Jared Candelaria, Student Member, IEEE, and Jae-Do Park, Member, IEEE Abstract—Currently classical thyristor-based high voltage di- rect current (HVDC) systems hold the market in bulk power transmission. However, recent advances in semiconductor tech- nology have led to voltage source converter based HVDC (VSC- HVDC) systems becoming a viable competitor. Not only is VSC- HVDC a competitor for transmission but it can also be used in multi-terminal systems, which have become an attractive option for renewable energy applications or for distribution in large cities. As more and more VSC-HVDC systems are installed, the protection of these systems must be taken into account. This paper explores different options and ideas for VSC system protection. Index Terms—HVDC, VSC, Protection I. INTRODUCTION Recently voltage source converter based high voltage di- rect current (VSC-HVDC) systems are becoming more of a competitor of classical thyristor-based HVDC systems [1]. As the converter power rating increases it may one day replace thyristor-based HVDC. VSC-HVDC is attractive because, un- like classical-HVDC, no reactive power support is needed to operate the system. In fact VSC’s can produce reactive power, and control active and reactive power independently [2]. This controllability allows VSC-HVDC converters to operate in systems with little or no AC support, something that classical HVDC cannot achieve without expensive support [3]–[6]. VSC’s are also advantageous in multi-terminal systems. Multi- terminal systems consist of three or more converters to create a HVDC network. Applications of multi-terminal systems include distribution into large cities, renewable energy inter- connects, and even ship power systems [7]–[10]. VSC’s are better suited for multi-terminal systems as the power flow can be changed by changing the direction of the current. Classical HVDC requires the DC voltage polarity to be changed, which can be difficult [3], [11], [12]. Currently, all of the installed VSC-HVDC systems are either back-to-back converters or are connected through underground cable. No overhead DC lines have been installed as of yet. It is said that the absence of overhead DC greatly reduces the risk of DC faults. However, in the case of a cable-connected system, a ground fault is almost always permanent, either by the failed cable insulation or the cable damage by an outside source. As more of these systems are installed into the bulk power system, protection of the VSC-HVDC systems must be a priority. VSC-HVDC systems are, by design, vulnerable to faults on the DC systems. Classical current-sourced-converter- based (CSC) HVDC naturally are able to withstand short circuit currents due the DC inductors limiting the current during fault conditions [8], [13]. When a fault occurs on the DC side of a VSC-HVDC system the IGBT’s lose control and Fig. 1. VSC Operation (a) Normal. (b) Positive Line-to-Ground Fault. the freewheeling diodes act a bridge rectifier and feed the fault [7], [9], [10], [13]–[17], as shown in Fig. 1. The types of faults possible on a HVDC system are as follows. • Positive line to ground fault • Negative line to ground fault • Positive line to negative line fault • Overcurrent • Overvoltage A challenge associated with the protection of VSC-HVDC systems is that the fault current must be detected and extin- guished very quickly as the converters fault withstand rating is generally only twice the converter full load rating [10]. Fault detection is also important, especially on multi terminal systems, in order to isolate the fault and restore the system to working order. This paper attempts to review the issues and the current methods of VSC-HVDC protection. II. POSSIBLE FAULTS A. Line-to-Ground A line-to-ground fault (ground fault) occurs when the pos- itive or negative line is shorted to ground. In overhead lines faults may occur when lightning strikes the line. This may cause the line to break, fall to the ground and create fault. In this situation the fault is always permanent and the line must be isolated for repair. Ground faults may also occur by objects falling onto the line, such as trees, providing a path to ground. In some cases when an object causes the ground fault it may fall away from the line and the system can be restored. If the fault persists the line would have to be taken out of service until the fault path can be cleared. 2 Underground cable is almost completely immune to line- to-line faults, as insulation, conduit and the earth separate the cables. However, they can still occur. The insulation of the cable can fail due to improper installation, excessive voltage/current, exposure to the environment(water, soil, etc) or cable aging [13]. When this occurs, the broken insulation will allow a path for current to flow to ground. As the fault persists the integrity of the insulation is reduced causing the fault to worsen. A ground fault may also occur when a person inadvertently cuts through one of the lines. This can happen during construction projects. In either case the fault will always be permanent and will require a complete shutdown of the line as well as a costly repair. When a line-to-ground fault occurs, the faulted pole rapidly discharges capacitor to ground. This causes an imbalance of the DC link voltage between the positive and negative poles. As the voltage of the faulted line begins to fall, high currents flow from the capacitor as well as the AC grid. These high currents may damage the capacitors and the converter [16]. B. Line-to-Line As stated before, a line-to-line fault on a cable-connected system is less likely to occur on the cable. In an overhead system, line-to-line faults can be caused by an object falling across the positive and negative line, they may also occur in the event of the failure of a switching device causing the lines to short. A switching fault, which is independent of how the converter stations are connected together, causes the positive bus to short to the negative bus inside the converter. A line- to-line fault may be either temporary or permanent. C. Overcurrent While overcurrent protection is important during line-to- line and line-to-ground faults, it must also operate when the system is being overloaded. Overload conditions may occur in two-terminal systems when the load increases past the rating of the converter or as a result of a fault on another part of the system. For example, if three VSC’s are feeding a common load and one VSC is dropped due to a permanent fault, the remaining two must supply the load. This will result in elevated currents that may overload the converters. In this situation the overcurrent protection would need to operate. Another option to avoid a wide spread blackout would be to shed non-critical loads. D. Overvoltage Overvoltages may occur in overload or in fault conditions. Overvoltages are only a concern during line-to-ground faults. When the fault occurs, the capacitor discharges rapidly to ground. The current flows through the ground then back to the unfaulted line and finally back to the source, which causes the voltage on the healthy pole to increase to 2 p.u. [16]. Overvoltage is also a concern during the loss of a converter. The loss of an inverter can cause voltage spikes due to excessive power, quickly charging the DC link capacitors. A rectifier loss is only of concern when the loss is temporary. When the rectifier returns suddenly it can cause overvoltages similar to that of an inverter loss [18]. Unbalanced AC network phases, mainly caused by faults, can lead to adverse effects on the operation of the VSC [19]. The configuration of the transformer linking the VSC to the AC grid can affect the voltage level on the DC side under AC faults. It can be seen that when the VSC is connected in a Y/∆,that a ground fault on the AC can result in an overvoltage of 1.5 p.u. [20]. III. DC PROTECTION WITH AC DEVICES Protection of DC systems can be done with conventional AC devices such as circuit breakers and fuses. The advantages of using AC devices include: • Less expensive than DC counterparts • Shorter lead time • Mature science • More familiar devices Below the methods of protection are shown. A. AC Circuit Breakers Placing AC circuit breakers on the AC side of the VSC is the most economical way to protect the DC system. They are commonly available and can be replaced in a shorter amount of time. However, AC circuit breakers result in the longest interruption time as a result of their mechanical restrictions [7]. Currently, the best interrupting time for an AC circuit breaker is two cycles [21]. When using an AC circuit breaker, the voltage of the DC capacitors will be monitored as well as the current in each DC line at each converter. These values will be fed back to a standard relay, which will monitor over/undervoltage, as well as overcurrent. When a DC fault occurs, the capacitors will discharge rapidly causing the voltage to decrease. The current on the faulted line will increase over the rated value. Once the relay senses one or more of these conditions, it will trip the breaker. In an attempt to restore the system, the relay will enter a re-closing cycle in which the relay will close back in and sense the voltage and current of the DC system. If the fault is cleared the system will return to normal, but if a permanent fault is detected the relay will lock out the breaker. The relay identifies a permanent fault by the re-closing sequence. A typical industry standard for re-closing on AC systems is that two attempts will be made; this can be applied to the VSC systems as well. After two attempts without success, the relay determines that the fault is permanent and will not allow the breaker to close. In back-to-back or two-terminal transmission systems, dif- ferential protection may be used to protect each converter, as shown in Fig. 2. The differential relay (Note: 87 is the ANSI standard number for a differential relay) will measure the current entering the converter as well as the current leaving the converter. If the current entering does not match the current leaving the differential relay, it will trip the AC breaker [22], [23]. In back-to-back systems one relay could also monitor the AC current at the sending VSC as well as the receiving VSC. If the current at one end does not match the current at the 3 Fig. 2. Back to Back Differential Protection. Fig. 3. Two-Terminal Differential Protection other end, the relay would know a fault has occurred and the VSC’s would be tripped offline. In a two terminal transmission system, two relays would be required and the current readings would have to be sent to the other relay via communication as can be seen in Fig. 3. This type of differential protection is common in AC systems [24] as could be applied to VSC protection. Also, compensation for losses would have to be taken into account. While AC circuit breakers are inexpensive, easy to use, and widely used, they also shut down the entire converter. This is problematic in the case of ground faults where the faulted line could be isolated and the system could run mono-polar using ground as a return path [16]. AC circuit breakers are also inconvenient in multi-terminal systems, which will be discussed later. B. Fuses Fuses on the AC side are generally not a good solution for protection of the VSC [25]. This is because a fuse is a thermal device that is only allowed one operation. Fuses do not have the ability to distinguish whether a fault is temporary or permanent. To a fuse every fault is permanent and therefore the system would not be able to be restored until the fuse was physically replaced. The only place that a fuse may be an acceptable alternative is for a non-critical load, or in areas where space is limited, such as a ship. Fuses are used for protection [25], but are mainly for AC protection and other DC devices protect the DC line. The DC devices coordinate with the fuse such that they will trip before the fuse. The fuse will only operate in the event that the DC protection fails. Fig. 4. IGBT-CB Fault Blocking Capabilities. Overall, fuses should be used only when necessary and all other options are not possible. IV. DC PROTECTION WITH DC DEVICES While AC devices are an economical way to protect the DC system, DC devices are a better option whenever possible. DC protective devices can act faster than their AC counterparts, as well as sectionalize lines. This allows the operation of unfaulted lines to continue. The methods of protection using DC devices are shown below. A. IGBT Circuit Breakers An IGBT circuit breaker (IGBT-CB) utilizes the blocking capability of the solid-state device. Like the other IGBTs in the converters, the IGBT-CBs are configured with an anti-parallel diode. The only drawback to the IGBT-CB is that it is a uni- directional device. This is illustrated in Fig. 4. When a fault occurs on the DC line, the IGBT is able to block the fault current (represented by the dashed line in Fig. 4). If the fault occurs on the converter side, the anti-parallel diodes conduct and allow current to flow (represented by the solid line in Fig. 4). In this scenario the IGBT-CB must rely on the blocking of the IGBTs in the converter [7]. For two-terminal systems, IGBT-CBs can be placed at each converter station, one on the positive line and one on the negative line, as can be seen in Fig. 4. Fast acting DC switches are used in conjunction with the IGBT-CB, which is used to isolate the line once the fault current has been cleared. It should be noted that the switch cannot break current and may only be opened once the fault has been extinguished. As with AC, the DC current of each line and the DC voltage of each capacitor will be sensed. Once the control system senses a fault on the line, an appropriate IGBT-CB will receive a gate signal to block the current. Once the fault current has been extinguished the fast acting DC switches will open, isolating the line. To determine if the fault is temporary or permanent, the DC switches and the IGBT-CB will close. If the fault has cleared the system will return to normal operation. If the fault is still present, the line will be isolated again and a permanent fault will be determined. The advantage of using an IGBT-CB is that the entire converter is not shutdown in the case of a ground fault. This 4 Fig. 5. ETO Based VSC-HVDC Converter. allows the faulted line to be isolated and have the system continue to run mono-polar. The IGBT-CB also opens faster than its AC counterpart. The disadvantage to the IGBT-CB is that it cannot protect against DC rail faults in the rectifier [25]. B. Converter Embedded Devices Converter embedded devices are active, protective compo- nents that are installed inside a VSC to detect and isolate DC faults. This method eliminates the use of additional devices, reducing the footprint of the converter station and also possibly cutting cost. However, a redesign of the converter is required. In [10], the converter uses two Emitter Turn Off (ETO) devices in an anti-parallel configuration to achieve both switching and protection. The ETO has a higher voltage and current rating than IGBT. Fig. 5 illustrates the converter configuration. In normal operation the ETO’s X act as the switching devices, while ETO’s Y act as the anti-parallel diode, and are constantly fired on. In the event of a fault the ETO’s X are blocked while ETO’s Y continue to feed the fault. Once the fault has been identified as permanent the Y ETO’s will be gated off. Another protection method has replaced the typical IGBT and anti-parallel diode is a combination switching device with the submodule shown in Fig. 6 [6], [17]. The converter submodule provides two different levels of protection. The first level protects the converter from being shutdown during a switching device failure. In the event of a switching device failure, the submodule will close switch K1, shorting out the defective submodule. This allows the converter to continue to operate by using redundant modules for un-altered system performance. The second level of protection reacts under fault conditions. As stated earlier, the switching device blocks and the anti-parallel diode conducts to feed the fault under fault conditions. The freewheeling diodes in VSC’s are not able to withstand large surge currents, and may be damaged before the fault is cleared. The solution in [17] is to bypass the IGBT and the anti-parallel diode with a press-pack thyristor K2. The proposed press-pack thyristor is able to withstand high surge currents, protecting the anti-parallel diode until the fault can be cleared. This protection method allows the converter additional con- trol and increases the current rating of the switching devices. It also cuts down on the number of components required for protection because the protection is embedded in the converter. Fig. 6. VSC-HVDC Converter Submodule. The disadvantage is that the entire converter must be shut down in the event of a permanent fault. This works well in two- terminal systems but may cause problems in multi-terminal systems. V. MULTI-TERMINAL SYSTEM PROTECTION VSC-HVDC systems are very appealing in multi-terminal systems, as power flow can be changed not by voltage polarity but the direction of the current. The possible applications for multi-terminal VSC-HVDC (MT-VSC-HVDC) systems are used in renewable energy applications and in distribution of power in mega cities. The protection strategies for MT-VSC- HVDC utilize both AC and DC protection. A. AC Protection As stated previously, DC protection can be achieved by using AC circuit breakers on the AC systems. This strategy can be applied to MT-VSC-HVDC as well. A ”hand shaking” method is proposed in [9]. This method, in addition to using AC circuit breakers, implements fast acting DC switches. The switches are only used to isolate lines and cannot break load or fault current. Each VSC will receive current measurements from their respective DC switches. When a fault occurs all of the AC circuit breakers associated with the MT-VSC- HVDC system will trip. Next, each VSC must determine which one of its respective switches to open. This is done by measuring the magnitude and direction of the current through each switch. The switch that will be selected is the one with the largest positive fault current. The hand shaking method defines positive as out of the node and negative into the node. Fig. 7 illustrates the example system given in [9]. When a fault occurs on Line 1, VSC1 receives current measurements from SW11 and SW31. VSC1 senses that the current through SW11 is positive and the current through SW31 is negative. Through the hand shaking method VSC1 opens SW11. VSC2 receives current measurements from SW12 and SW22. Once again the current through SW12 is positive and the current through SW22 is negative and switch SW12 is selected. VSC3 receives current measurement from SW33 and SW23. The current direction for both switches is measured as positive. The switch with the highest magnitude of current is selected. At this point Line 1, the faulted line, is isolated, and Line 3 is open at one end. At this point the system must enter a re-closing mode. First, all of the AC 5 Fig. 7. (a) Current Flow During a Fault. (b) Fault Isolation with the Hand Shaking Method. breakers will close back in, re-energizing the VSC’s. Next, the fast DC switches of the non-faulted DC lines must be closed. The VSC’s only re-close switches when the voltage of its respective line is near the voltage of the VSC terminals. Fig. 8 shows the re-closing method presented in [9], where it can be seen that only SW33 will be able to re-close. During the fault VSC1 chose to open SW11, leaving SW31 closed. Once the AC breakers re-close, and the VSC1 is