Fault-tolerant DC voltage network

Abstract

The invention relates to a DC current network comprising a first and a second line, each of which is designed to take on the load of the respective other line at least temporarily according to the situation, a first power switch device for selectively connecting one end of the first line to ground, a second power switch device for selectively connecting the other end of the first line to ground, a third power switch device for selectively connecting one end of the second line to ground, a fourth power switch device for selectively connecting the other end of the second line to ground, a first converter which is connected to one end of the first line via a fifth power switch device and to one end of the second line via a sixth power switch device, a second converter which is connected to the other end of the first line via a seventh power switch device and to the other end of the second line via an eighth power switch device, a controller which is configured such that in the event of a disturbance, the disturbed active line is grounded via the respective two power connection devices, the first and second converter limit the DC voltage to a specified maximum direct current, the disturbed line is separated from the converters by the respective two switch devices connected to the converters, and the respective other line is switched through to the converters via the power switch devices which are connected to the converters if the respective other line has not already been switched through.

Description

[0001] Fault-tolerant DC voltage network

[0002] The present invention relates to a fault-tolerant DC voltage network and a method for a DC power network for switching off a line in the event of a fault.

[0003] The necessary replacement of fossil fuels with electrical energy places new demands on the functioning of electrical grids. In the area of ​​long-distance transmission of large amounts of power – including nationwide and Europe-wide – electronically controllable direct current grids (“DC grids”) are particularly necessary to supplement the existing alternating current grids (“AC grids”). These new types of grids are connected to the AC grid at specific, selectable locations by power electronic converters. The new types of DC grids must meet the highest demands in terms of reliability and availability, as they must contribute significantly to the secure long-distance transmission of high amounts of power and to the stabilization of the AC grids.This is made possible by the ability of power electronic converters to provide controllable reactive power, which has become available since thyristor converters were replaced by innovative topologies and semiconductors. The Modular Multilevel Converter ("MMC"), with state-of-the-art silicon IGBTs as semiconductors, has emerged as the preferred topology worldwide. The introduction of silicon carbide semiconductors is currently under development.

[0004] Historically, converters have long been considered a critical component of DC grid reliability and availability. However, the modular, fault-tolerant structure of the MMC, advances in semiconductor technology, and many years of global operational experience have led to a new assessment: insulation faults and mechanical damage in large-scale power grids will be considered the dominant source of failure in the future.

[0005] The state of the art for AC grids has remained essentially unchanged for many decades. The following discussion therefore refers to electronically controllable DC grids and the additional aspects that arise from coupling these grids with AC grids.

[0006] A fundamental and well-known problem in a DC grid arises when the grid has closed meshes to ensure redundant lines: During normal operation, certain target values ​​for the active power are generally specified for the converter stations, which are then set at the interfaces to the AC grid and transmitted via the DC grid. In a DC grid with closed meshes, however, there are multiple lines that can participate in the power transmission. Since the ohmic resistance of the lines determines the distribution of the currents in a DC grid, there is generally a desire to be able to control the current distribution in a targeted manner so that it meets certain objectives. These objectives include, for example, adhering to maximum current values ​​in order to avoid thermal stress on individual lines.Possible actuators for this function are power converters, which inject relatively small, additional voltages into the lines of the DC grid. These so-called power flow controllers (PFCs) are preferably inserted into the line nodes of the DC grid. This means that they do not have to supply active power and therefore do not require a (high-voltage isolated) energy feed. Furthermore, it is known that in order to function during normal operation, they only need to generate a relatively low voltage compared to the nominal voltage of the DC grid. Both of these boundary conditions make it possible to keep the technical complexity and costs of the PFC relatively low. However, in order to limit or interrupt the current in a line in the event of a typical fault in a DC grid, a voltage greater than the DC grid voltage is required.

[0007] DC circuit breakers (DCBs), which can fulfil this task, represent an essential component of known solutions for fault control in DC networks. However, the technical complexity of all known designs is very high because the requirements and boundary conditions for a DC circuit breaker are considerably more demanding than for an AC circuit breaker. These more difficult boundary conditions are: a) The fault current in the event of an insulation fault or short circuit in a DC network can rise to several times the nominal current within a few milliseconds, since the increase is essentially only limited by the line inductances. b) The possible final value of the fault current is fundamentally only limited by the ohmic line resistances and can therefore be up to approx.two orders of magnitude above the rated current c) The fault current has no natural zero crossing, so that conventional mechanical circuit breakers cannot interrupt the current d) A suitable, functioning DC circuit breaker must generate a defined counter voltage (typically 1.2 to 1.6 times the rated DC voltage) and absorb a very high level of energy to reduce the current in the line in question. Electronic DC circuit breakers which contain a series connection of semiconductor switches (e.g. IGBT) can at least undercut the permissible delay time until a sufficiently high counter voltage builds up. However, the outlay for such an electronic high-voltage switch is very high. In addition, semiconductor switches are not suitable for absorbing high levels of energy, so additional elements are required.A major disadvantage in this regard is that the energy absorption capacity of available varistors is also exceeded by orders of magnitude. The DC circuit breaker, which is always switched on during normal, undisturbed operation of the DC grid, also exhibits high power dissipation (on-state power loss), which leads to unacceptable energy losses during continuous operation.

[0008] Known concepts and designs for DC circuit breakers (DCBs) therefore generally also use one or more mechanical switches, which bridge the semiconductor switches during undisturbed continuous operation. In this way, the high energy losses of the semiconductor switches can be avoided. However, if an unexpected fault occurs in the DC network, the delay time typical of mechanical switches from the switching command to the mechanical opening of the contacts is disruptive. This principle-related disadvantage applies to all DCBs that use mechanical switches to avoid on-state power losses. However, with improved electromechanical drive methods, it is technically possible to reduce these delay times to values ​​below a few milliseconds. Direct electrodynamic drives based on the eddy current principle (Thompson) have proven to be particularly suitable, with delay times of up to approx.1msec achieved.

[0009] Although fast-switching mechanical switches are available, this concept is still disadvantageous because both semiconductor switches and high-quality mechanical switches are used, which negatively affects the manufacturing costs.

[0010] It is therefore the object of the present invention to provide a DC power grid, in particular a HVDC power grid, that does not have the aforementioned disadvantages and enables simple shutdown of a line in the event of a fault. In particular, it should be possible to operate only with mechanical switches if necessary. This object is achieved by the subject matter of the independent patent claims. Preferred embodiments are the subject matter of the dependent claims.

[0011] The present invention is based on the discovery that other additional relevant delay times in the system until a fault shutdown is successfully completed can be significantly reduced through suitable measures. This, in turn, enables the use of mechanical circuit breakers that do not have to meet stringent requirements. In particular, this eliminates the need for additional semiconductor switches and / or very fast mechanical switches if necessary.

[0012] Furthermore, the invention is based on the finding that it is advantageous to connect the faulted line to earth via circuit breakers so that the considerable energy present in the faulted system can be dissipated, which facilitates the switching operations. As soon as the faulty line has been connected to earth by circuit breakers at both ends, a standby line, i.e. the second line (if only two lines are present), can be connected to the converters via circuit breakers. This connection is relatively straightforward because the converter can have the property of limiting the current, at least during the switching interval. As soon as the standby line is connected at both ends, the load that was previously carried by the faulty line can now be carried by the standby line.

[0013] Connecting the standby line is not necessary if it was already live. In this case, the standby line virtually seamlessly takes over the load previously carried by the faulty line. Typically, such line systems are designed or capable of handling the load from the faulty neighboring line, in addition to its permanent power, at least for a certain period of time.

[0014] By initially grounding the faulty line, a low-inductance circuit is created, each formed by a converter output on the DC side, the grounding switch (i.e., the circuit breaker that enables the ground fault), and the disconnector (i.e., the circuit breaker that enables the connection or disconnection of the line from the converter). This local circuit, which is no longer loaded by the severed long line, allows for a rapid current change (via the converter). The invention proposes a DC power grid, preferably an HVDC power grid, including corresponding components and switches, with a first and a second line, each designed to take over the load of the other line at least temporarily, depending on the situation, a first power switching device for selectively connecting one end of the first line to ground,a second power switching device for selectively connecting the other end of the first line to earth, a third power switching device for selectively connecting one end of the second line to earth, a fourth power switching device for selectively connecting the other end of the second line to earth, a first converter, wherein the first converter is connected to one end of the first line via a fifth power switching device and is connected to one end of the second line via a sixth power switching device, a second converter, wherein the second converter is connected to the other end of the first line via a seventh power switching device and is connected to the other end of the second line via an eighth power switching device, a control device which is configuredthat in the event of a fault, the faulty active line is earthed via the respective two power switching devices, the first and second converters limit the DC voltage to a specified maximum DC current, the faulty line is disconnected from the converters by the respective two switching devices connected to the converters, and the other line is connected to the converters via the power switching devices connected to the converters, if not already connected.

[0015] Furthermore, the invention proposes a method for a DC power grid for switching off a line in the event of a fault and distributing the load to a second line, comprising the following steps: connecting at least one of the lines on both sides to the DC side of a converter, wherein each of the converters is connected to an AC network with its other side and each converter is capable of regulating the DC side independently of the AC side, detecting a fault on the at least one line, connecting the faulty line on both sides to earth by means of a circuit breaker, limiting the maximum DC current of the two converters to a predetermined maximum current, disconnecting the faulty line on both sides from the converters by means of a circuit breaker, connecting the undisturbed line to the converters on both sides by means of a circuit breaker in the eventthat the undisturbed line is not already connected to the converters via these circuit breakers. Although the present invention works with conventional converters and is therefore not limited to the use of modular multilevel converters (MMCs), these are particularly well suited for the inventive solution, as explained in more detail below. In particular, conventional converters with a downstream bidirectional DC / DC converter can be used to regulate the output current. This also makes it possible to regulate the DC side independently of the AC side, at least for a short time, as required for shutdown. It is this property that also makes MMCs particularly suitable.

[0016] At the time of shutdown, the converter should initially limit the current to the current present at the time and then reduce it to a relatively low value, such as 10 A or zero. This then allows the mechanical switches to open without problems.

[0017] DC power grids are preferably HVDC (high-voltage DC). More than two lines can be present, so the load can be distributed across several undisturbed lines in the event of a fault.

[0018] Preferably, the system includes sensors for detecting faulty lines. This sensor function can be integrated into the converters, preferably MMCs.

[0019] In addition to the sensors that detect the fault, control units that supply the switching pulses to the circuit breakers are preferably provided to trigger the switching process. This control function can also be integrated into the converters.

[0020] According to one embodiment, the invention is implemented using MMC in the converter stations and utilizes some of its specific properties, as explained in more detail below.

[0021] The lines can be, in particular, overhead lines, underground cables, or submarine cables. In particular, the invention enables the continued, trouble-free operation of the AC and DC networks with the desired target values ​​for active and reactive power in the event of faults in the power grid.

[0022] Preferred embodiments are explained in more detail below with reference to the accompanying figures. Fig. 1 shows a schematic representation of an embodiment according to the invention,

[0023] Fig. 2 : an embodiment of a circuit breaker,

[0024] Fig. 3 : an alternative embodiment for a circuit breaker, and

[0025] Fig. 4: an embodiment of a converter with bipolar DC / DC converter

[0026] Fig. 1 shows a schematic view of one embodiment. This embodiment is intended to be a basic configuration; of course, instead of two lines, it could also be a branched line system, which could then also be completely or partially deactivated accordingly.

[0027] The DC power grid 1 has converters 2, 3 on either side of lines 4, 5. Lines 4, 5 are shown in the usual schematic manner, including parasitic capacitances 9 and inductances 8. Lines 4, 5 can both carry current during normal operation, or one of the two lines can be provided as a redundant line that takes over the load of the faulty line in the event of a fault. If both lines are under load, one of the two lines can also take over the load of the faulty line in the event of a fault, at least for a certain period of time.

[0028] Each of the lines (more than two lines are also possible) can be connected to the two converters 1, 2 via a pair of circuit breakers 10a, 10b, 11a, 11b at its two ends. Each of the two lines is connected to ground via another pair of circuit breakers 12a, 12b, 13a, 13b, as shown in Fig. 1. Optionally, additional inductors 6, 7 can be provided for DC throttling or smoothing the DC current.

[0029] The converters 2, 3 can be connected on the AC side to typical AC grids or to power generation systems, such as power plants or generators, in which case a transformer is often interposed. Furthermore, the converters 2, 3 (or all converters involved in a power system) can be connected to each other via a ground wire 14.

[0030] The earth wire can be laid as an additional wire together with wires 4, 5 and, if necessary, other wires, or can be formed by the cable shield around these wires.

[0031] The following describes the procedure for disconnecting a line in the event of a fault. For this purpose, assume that initially only line 4 is live, i.e., line 4 is connected to converters 2 and 3 via circuit breakers 10a and 10b at both ends. Circuit breakers 12a and 12b, however, are open.

[0032] Power line 5 is separated from the converters 2, 3 by means of the circuit breakers 11a, 11b and is preferably earthed via the circuit breakers 13a, 13b.

[0033] If a fault is detected on line 4 by sensors (not shown), the faulty line is grounded by switches 12a, 12b so that the energy present in the faulty system can dissipate. Before, at the same time, or shortly afterward, switches 10a, 10b are opened to isolate the faulty line from converters 2, 3. This opening is easily possible due to the grounding on the one hand and the current-limiting and current-controlling properties of the converters, preferably MMCs, on the other hand, in contrast to the situation without grounding and / or without the presence of converters with these properties. The faulty line 4 is now completely shut down.

[0034] At the same time, shortly before, or shortly after, redundant line 5 is connected by closing switches 11a and 11b. This connects redundant line 5 to the converters. If the line was previously grounded, switches 13a and 13b must be opened at approximately the same time to prevent current from flowing to ground.

[0035] Line 5 now carries the load that previously flowed through line 4. To enable the process as described, a centralized or distributed control device is required to switch all relevant switch pairs in a timely manner. This control device, like the sensors, can be housed in the converters.

[0036] However, it is also possible that line 5 was previously connected to converters 2 and 3 via switches 11a and 11b, thus also carrying load alongside line 4. In this case, line 5 then assumes all power alone, at least for a period of time until line 4 is operational again.

[0037] Thanks to the converter's current-limiting and current-controlling properties, the current can be adjusted as desired. If the total current decreases accordingly after line 4 is switched off (i.e., only line 5 continues to carry its current as before), the converter can increase the current in line 5, so that ultimately the previous total current is now carried solely through line 5.

[0038] For the specified method according to this embodiment, this means that in this case, switches 11a and 11b were already initially switched on and therefore do not need to be switched on again. Grounding via switches 13a and 13b was then not present.

[0039] After a line has been switched off, it can be reconnected to the system after a short time or after the fault has been rectified by first isolating it from earth via the two earth switches and then or simultaneously establishing a connection to the converters via the two isolating switches.

[0040] The detection of the line fault is particularly easy at the upper points K in Fig.1, because there the voltage deviates transiently from the nominal voltage due to a fault on the line.

[0041] Due to the fact that the energy is diverted from the faulty line by earthing on the one hand and the current-limiting and current-controlling effect of the connected converters on the other hand, the converters can be easily separated from the faulty line using the switches.

[0042] The converters can not only limit the current to the current level, thus preventing any increase, but can also reduce the current to very low values, such as 10 A or zero. This function can be achieved very conveniently using MMCs. This now also allows the use of very simple and therefore inexpensive mechanical switches instead of complex switches and / or semiconductor switches. In large systems, this represents a decisive cost advantage. One possible switch for this purpose, which is very fast yet inexpensive, is the vacuum switch.

[0043] In the manner according to the invention, faulty lines can be switched off easily, quickly, and cost-effectively. Earthing removes disruptive energy from the system. Switching off is therefore quicker, simpler, and more cost-effective. Switching off can therefore be used specifically to search for faulty lines and / or their causes by systematically switching many lines in a network off and on in a short space of time. Switching off lines solely for troubleshooting is possible with the prior art, but is complex and cost-intensive. The invention makes such shutdowns easy to handle, thus creating new possibilities for troubleshooting. Since the shutdowns can occur quickly, the connected AC systems are practically unaffected, as the inverters store enough energy to briefly compensate for the shutdown of individual lines.

[0044] Switching off a faulty line or several lines, i.e. a faulty system, is therefore easy because connected systems, DC and AC systems, are not disturbed and practically do not notice it.

[0045] Figures 2 and 3 show possible designs for the aforementioned circuit breakers. The circuit breakers can all be identical or different from one another. Preferably, inexpensive, mechanical switches with low switching capacity (capable of switching neither high current nor high voltage) are used.

[0046] Fig. 2 shows a possible structure. The internal circuit of a first embodiment for a switching unit / circuit breaker 20 comprises an electromechanical circuit breaker 21, which preferably has a drive with a shortened delay time of only a few milliseconds compared to other commercially available designs. A high-blocking semiconductor diode 22, which may be of commercially available design, is connected in parallel with the switch. Also connected in parallel is a series circuit comprising a snubber capacitor 23 and a snubber resistor 24, which serve to reduce the voltage gradients across the switch 21 and the semiconductor diode 22. Also connected in parallel is a high-ohmic resistor 25, which optionally serves for static balancing of the switch voltages in the case of series circuits. The power switching unit 20 thus essentially consists of known, commercially available components.

[0047] However, none of the additional components mentioned are absolutely necessary; the switch alone can also be used. The use of a diode connected in parallel is particularly advantageous, as this limits the voltage seen by the switch, namely to the forward voltage of the diode. A diode should therefore be present next to the switch if the switch can only withstand low voltages. This allows for flexible adaptation to the requirements of various mechanical high-voltage switches, so that the most economically or technically suitable switch can be selected. It is well known that switching off high currents and the requirement for a high arc voltage of the switch in particular lead to very high wear and tear and therefore considerable expenditure.If switches with high arcing speeds are available, the invention makes it possible to dispense with the wiring using parallel diodes and / or RC attenuators. However, very low-complexity, cost-effective switches can generally only generate very small arc voltages of typically a few tens of volts due to their design. This applies, for example, to vacuum interrupters. In this application, the invention enables the mechanical switches to only generate a very small arc voltage, which is greater than the forward voltage of the parallel diodes. For such switches, the current to be switched off by the mechanical switch can be selected not only to be approximately zero, but also to be very small, but slightly above zero in the current direction of the active switching off of the mechanical switches - as best corresponds to the switching capacity of these switches.

[0048] Another embodiment of the switching unit 30 is shown in Fig. 3. This variant can be used when the permissible voltage of the semiconductor diodes 32 is significantly lower than that of the available power switches 31. Fig. 3 shows an example of a series connection of three semiconductor diodes. However, any number is possible. The optional capacitors 33 and resistors 34 and 35 are also shown, similar to Fig. 2.

[0049] Both variants of switching units 20 and 30 can also be connected in series as required to obtain switching units for any desired high voltage. A key advantage of these switching units is that the technically unavoidable tolerances of the circuit breakers' delay times do not impair their reliable voltage distribution and switching capability. As explained below, the preferred operating mode of the switching units ensures that the circuit breakers are only subjected to the forward voltage of the semiconductor diodes connected directly in parallel during their switching operation. This makes it possible to implement the circuit breakers using a series connection of several low-voltage switches. This makes it possible to achieve particularly short delay times in combination with extremely high voltages.

[0050] Fig. 4 shows an embodiment of a converter that can be used, but is not an MMC. A conventional inverter 40 is connected to a bipolar current controller 41 on its DC side, preferably with a capacitor 42 interposed. The resulting converter has current-limiting and current-controlling functions and allows the DC side to be controlled independently of the AC side. It can be connected to the circuit shown in Fig. 1 via terminals 43 and 44.

[0051] In the following, a possible switching scenario for switching off a faulty line is explained in detail using Fig. 1.

[0052] This sequence begins with the detection of a fault at time (ti) on line 4 in line system 1 of Fig. 1. Although only one line is referred to below as being faulty, it is clear to the person skilled in the art that this can also affect a line system and thus an entire line system can be switched off.

[0053] After the fault has been detected, the defective line 4 should be disconnected from the “rest” of the DC network at all its terminals as quickly as possible.

[0054] Detection can be carried out in a variety of known ways. It is known that fast and spatially precise fault detection and localization is difficult and prone to its own errors. The invention enables the transient deviation of the DC voltage from the nominal voltage at the terminals of the line systems to be used as a simple, reliable fault detection. The detectors for the fault should preferably measure here. Alternatively, the discharge current in the circuit elements 23, 24 of Fig. 2 of the switching units 20 can also be detected for this purpose. The first converter station in which the fault is detected is generally a power-feeding station that is spatially closest to the fault. It is advantageous and technically feasible to transmit the detection of the fault to all other converter stations using modern communication technology.This can then also communicate target values ​​for the power to be transmitted between all stations.

[0055] When using MMCs for converter stations, the following features of these devices are utilized: a) The MMC's direct currents can be quickly adjusted using the control system and maintained within narrow tolerance ranges. b) The MMC's AC and DC variables can be independently adjusted using the control system. In particular, large power differences between the AC and DC power can be compensated for by the MMC's internal energy storage (capacitors in the submodules) (for periods of typically 5...10 ms).

[0056] As a result of these boundary conditions, it can be ensured that the DC currents of the MMC (even after a fault occurs) initially remain at the previous operational value, or within a narrow, specified tolerance band around this value. (This value is not a maximum overcurrent limit, but rather the (significantly lower) value required for the operationally specified power setpoint.)

[0057] The first step is to switch on the electromechanical switches 12a, 12b, which are connected to the ground wire 14. After a certain communication runtime, this is done at all participating stations if an entire line system needs to be shut down.

[0058] The delay time of the switches 12a, 12b and that of the communication is typically in the range of a few milliseconds.

[0059] By "grounding" the defective line or defective line system, any potential arc at the fault location is extinguished. The defective line system is discharged, and the fault current in it decays due to the natural losses in this line system in typically 100 ms to 500 ms. It is now de-energized.

[0060] After actuating switches 12a and 12b as described above, the MMCs can control their direct currents to zero very quickly (typically 0.5 ms). This is made possible by the fact that a circuit with lower inductance (by more than two orders of magnitude) now exists via the circuit breakers, the DC choke 6, 7, and the DC current of the MMC. The desired switches 10a and 10b, which are required to isolate the defective line or line system, can now be opened. A variety of known electromechanical switches can be used because they now switch under approximately "zero volts / zero amperes" conditions.

[0061] As a final step, the DC voltage is ramped back up to operating values. (This time period can also be implemented as < 1 ms to several ms and is adjustable via control technology.) The entire fault clearance process therefore typically takes 5 ms if the delay time of the electromechanical switches (100) is specified at approximately 2...3 ms. Within this time period, the MMCs, as already explained, can use their internal energy storage to ensure that the AC networks continue to achieve the required transmission power without disruption. The invention therefore makes it possible to continue transmitting the full, undisturbed power via the remaining line systems. This time period is limited only by the thermal time constant of these lines, which is at least several minutes up to approximately 10 minutes.

[0062] During this time, “attempts” to partially or completely restart the defective line system can now be made, as this can be done without disrupting the power transmission.

Claims

Patent claims 1. A DC power grid with a first and a second line, each of which is designed to take over the load of the other line at least temporarily depending on the situation, a first power switching device for selectively connecting one end of the first line to earth, a second power switching device for selectively connecting the other end of the first line to earth, a third power switching device for selectively connecting one end of the second line to earth, a fourth power switching device for selectively connecting the other end of the second line to earth, a first converter, wherein the first converter is connected to one end of the first line via a fifth power switching device and is connected to one end of the second line via a sixth power switching device, a second converter,wherein the second converter is connected to the other end of the first line via a seventh power switching device and is connected to the other end of the second line via an eighth power switching device, a control device which is configured such that in the event of a fault, the faulty active line is grounded via the respective two power switching devices, the first and second converters limit the DC voltage to a predetermined maximum DC current, the faulty line is disconnected from the converters by the respective two switching devices connected to the converters, and the respective other line is connected to the converters via the power switching devices connected to the converters if not already connected.

2. The device according to claim 1, wherein the power grid is a HVDC power grid.

3. Device according to one of the preceding claims, wherein further lines are provided which are connected in the same way and can thus be switched off.

4. Device according to one of the preceding claims, wherein the converter has a current-limiting and current-controlling function.

5. Device according to one of the preceding claims, wherein the converter is capable of regulating the DC side independently of the AC side.

6. Device according to one of the preceding claims, wherein the converter consists of an inverter with a downstream bidirectional DC / DC converter for regulating the output DC current.

7. Device according to one of the preceding claims, wherein the converter is a modular multilevel converter, MMC.

8. Device according to one of the preceding claims, with a sensor device for detecting a faulty line.

9. Device according to one of the preceding claims with a control device for sending control pulses to the circuit breakers.

10. Apparatus according to any one of the preceding claims, wherein at least some of the circuit breakers are mechanical switches.

11. Device according to claim 10, wherein a diode is connected in parallel with the mechanical switch to limit the switching voltage.

12. Device according to one of the preceding claims, wherein at least some of the circuit breakers are vacuum switches.

13. A circuit breaker for a device according to any one of claims 1-12, wherein a diode is connected in parallel to a mechanical switch.

14. A method for a DC power system to switch off a line in the event of a fault and distribute the load to a second line, comprising the following steps: connecting at least one of the lines on both sides to the DC side of a converter, each of the converters being connected to an AC network with its other side and each converter being able to control the DC side independently of the AC side, Detecting a fault on at least one line, Connecting the faulty line to earth on both sides using a circuit breaker, limiting the maximum DC current of the two converters to a specified maximum current, disconnecting the faulty line from the converters on both sides using a circuit breaker, Connecting the undisturbed line to the converters on both sides using a circuit breaker in case the undisturbed line is not already connected to the converters via these circuit breakers.

15. The method of claim 14, wherein the converter instructs the switching of the power switches.

16. Method according to claim 14 or 15, wherein the converters belonging to the line to be switched off operate synchronously with one another.

17. The method according to claim 14, 15 or 16, wherein, in the event of a fault being detected, the converter reduces its DC current from a current current to a predetermined maximum current.

Images