Method for controlling a grid-forming converter, computer program and grid-forming converter.

The invention relates to a method for controlling a grid-forming converter, the con- verter comprising semiconductor switches and a control unit which controls the semi- conductor switches. The invention further relates to a computer program for computer controlled execution of such a method and a grid-forming converter which is ar- ranged for executing such a method.

Generally speaking, the invention relates to the use of the technology of converters in the form of grid-forming converters that shape the voltage waveform in the electrical grid. An example of such grid-forming converter are the Smart Transformers (ST), which are also called Solid State Transformers, Power Electronics Transformers or Intelligent Universal Transformers. Such converters make use of electronic control, in particular software or other kind of logic control, for controlling a plurality of semicon- ductor switches which are switched in a manner such that a desired energy transfer from a primary side of the converter to a secondary side of the converter or in the op- posite direction can be achieved.

Faults in such converters or converter-fed grids are difficult to handle, due to the low overload capability of the ST. On one side, the low voltage (secondary) side con- verier of ST can protect itself by blocking rapidly the semiconductor switches when its current is higher than a threshold value, avoiding the semiconductors destruction, but, on the other side, it loses the vertical selectivity of the fault protection relays (grid-breakers) in the downstream grid. Indeed, in case of a fault in the ST-fed grid, the ST may be the first one to intervene, due to the low ampacity of its converter, with respect to the breaker intervention characteristics, that have higher current interven- tion thresholds.

It is an object of the invention to improve this situation.

The object of the invention is achieved by a method for controlling a grid-forming con- verter, the converter comprising semiconductor switches and a control unit which controls the semiconductor switches, the method comprising at least the following steps:

a) detecting a fault in the converter or in the grid connected to the converter, which fault may lead to a malfunction of the converter,

b) changing an actual operating mode of the control unit into a fault operating mode for controlling the semiconductor switches, which protects the converter from the malfunction and at the same time maintains the selectivity grid-breakers in the grid.

The fault detection can be achieved measuring the current at the converter point of connection with the electrical grid or in any point of the electrical grid, by means of lo- cal measurements or by means of a communication infrastructure or communication- based protection system. Regarding this last possibility, the breakers installed in the converter-fed grid measure an increase in the current, and recognized the fault pres- ence, communicate it to the grid-forming converter. Received the fault signal from the breakers, the grid-forming converter applies what described in this invention. In ab- sence of communication infrastructure or communication-based protection system, the converter relies on local measurements, such as current, voltage, frequency, phase measurements.

Such grid-forming converters control the voltage in the fed grid, e.g. in the grid con- nected to the primary side or the secondary side of the converter. In case of grid fault the converter can protect itself by blocking the semiconductor-based switching ele- ments. However, the response of the converter may result faster than the reaction of any breaker relay in the grid, resulting in the loss of grid-breaker selectivity in the dis- tribution grid. The converter may interrupt the short circuit current before the breaker in charge of breaking the current, thus interrupting the power supply to the whole dis- tribution network. The proposed protection method allows to protect grid forming con- verter and at the same time to respect the grid breakers selectivity.

Therefore, since the selectivity of the grid-breakers is maintained by the invention, in other words, the proposed method allows to respect the grid-breakers selectivity.

This allows to integrate a grid-forming converter without any changes in the protec- tion systems of existing electrical distribution grids. For maintaining the selectivity grid-breakers in the grid, for example, the converter may by further operated, while the fault is present, until a grid-breaker in a grid connected to the converter triggers.

The malfunction of the converter may be a destruction of one or more of its semicon- ductor switches due to high short circuit currents. The proposed approach uses the fast controllability of the converter to provide fast fault clearing and safety to the con- verter switching circuit.

According to an advantageous embodiment, in fault operating mode the converter blocks as first reaction the semiconductor switches of the phase or phases involved in the fault. This protects the elements of the converter. The blocking of the semicon- ductor switches can be done, for example, by switching off the semiconductor switches.

According to an advantageous embodiment, in fault operating mode the converter waits for a short transient which allows to dissipate the fault-caused transient energy from the semiconductor switches and then gradually increases the output voltage provided by the semiconductor switches in the fed grid. The short transient can be, for example, in the area of a few microseconds or it may reach to several cycles of the grid voltage. The gradual increase of the output voltage can be applied continu- ously or stepwise, or such increase functions could be mixed.

According to an advantageous embodiment, a voltage gradient is applied for gradu- ally increasing the output voltage. The voltage gradient can be a non-linear gradient, for example an exponential function. According to an advantageous embodiment, the voltage gradient is a linear ramp. This allows for soft and smooth recovery of the output voltage.

According to an advantageous embodiment, the gradually increasing of the output voltage is done from a starting voltage, e.g. zero Volt, to a target voltage, e.g. the nominal voltage of the grid.

According to an advantageous embodiment, upon at least one trigger condition the controller starts to switch the semiconductors in order to create a squared waveform of the output voltage of the converter. Using such square-wave mode instead of usual sinusoidal voltage waveform, the converter is able to provide the full short cir cuit current (at nominal voltage) using a square-based voltage waveform. It mini- mizes the switching losses, maintaining the semiconductor temperature below the maximum one. In the square-wave mode the switching frequency of the semiconduc- tor switches of the converter can be significantly lower than in sinusoidal mode. For example, in square-wave mode the switching frequency may be the nominal grid fre- quency of the grid fed by the converter (e.g. 50 Hz), or lower. If deemed necessary to reduce the switching losses, DC current can be supplied to the fault, leaving the switching elements turned ON during the fault.

According to an advantageous embodiment, the controller evaluates, by means of an automatic fault detection method, whether the fault is close to the converter and/or the closest breaker has an intervention current higher than the converter ampacity, if both of these conditions are not fulfilled then the controller starts to operate the power converter with constant maximum overcurrent. In this operation mode, the grid voltage in the faulty phase will be less than or equal to the nominal value, but still can provide partial voltage supply. The value of constant maximum overcurrent can be determined by the detail parameters of the used power semiconductor devices, and switching losses reduction-based strategies. The percent of overcurrent operation of the power semiconductor devices is dependent on the amount of power losses that can be reduced via the switching losses reduction-based strategies.

According to an advantageous embodiment, the controller evaluates whether the fault is close to the converter and/or the closest breaker has an intervention current higher than the converter ampacity, if one or both of these conditions is fulfilled then the controller starts to switch the semiconductors in order to create a squared wave- form of the output voltage of the converter. Therefore, if the fault is close to the con- verter, a trigger condition for starting the square-wave mode is given. If the closest breaker has an intervention current higher than the converter ampacity, another trig ger condition for starting the square-wave mode is given. The controller can get infor- mation about the position of the closest breaker and/or of the converter ampacity, for example, by storing such parameters in a memory of the control unit. The squared waveform of the output voltage helps that the converter can withstand higher over- load, due to the reduced switching losses of the semiconductors. Additional switching components (e.g., thyristors) can be added in order to support the converter to pro- vide further short-circuit capability in the fed grid. These switches can work in coordi- nation with the proposed invention. The term“ampacity” means the nominal maxi- mum current of the grid-forming converter or a specific phase of the converter.

According to an advantageous embodiment, the method comprises using a commu- nication-based protection system.

According to an advantageous embodiment, the method comprises detecting the fault in the converter or in the grid connected to the converter and/or changing the actual operating mode of the control unit into the fault operating mode using a com- munication-based protection system.

The object of the invention is further achieved by a computer program with instruc- tions executable on a computer for computer controlled execution of a method of the aforementioned kind. By such computer program the same advantages can be achieved as mentioned above.

The object is further achieved by a grid-forming converter having a primary side to be connected to a primary grid and a secondary side to be connected to a secondary grid, the converter comprising semiconductor switches and a control unit which con- trols the semiconductor switches for converting electrical energy from the primary grid to the secondary grid and/or from the secondary grid to the primary grid, wherein the control unit is arranged for executing a method of the aforementioned kind. By such converter the same advantages can be achieved as mentioned above.

According to an advantageous embodiment, no ferroresonant circuit is connected in parallel to the converter. Such ferroresonant circuit may be used in other applica- tions, for example uninterruptable power supplies, for providing an external short-cir- cuit path to the converter, which allows for injecting higher currents than the con- verter would be able to inject. However, as a result of the avoidance of a ferroreso- nant circuit, in the present invention electrical parts can be saved and the complete grid can be realized with less efforts.

According to an advantageous embodiment, the control unit comprises a computer and/or logic circuitry for executing the method of the aforementioned kind. The com- puter maybe a microprocessor, microcontroller or similar unit which is able to execute a software program. The logic circuit maybe a FPGA.

According to an advantageous embodiment,

a) the primary grid is a high voltage grid and the secondary grid is a medium voltage (MV) grid or

b) the primary grid is a medium voltage grid and the secondary grid is a low voltage (LV) grid.

Advantageous applications of the invention are any LV and MV distribution grid, where a ST is integrated, as well as islanded grids and microgrids.

In a nutshell, the technology proposed in this patent allows to protect the converter from overloading during short circuits, and at the same time to respect the vertical se- lectivity of breakers in the grid. If a fault occurs, the converter blocks as first the semi- conductors of the phases involved in the fault. Then, it may wait for a short transient (variable from few microseconds to several fundamental cycles) in order to let dissi- pate the fault transient energy and following applies a voltage ramp up to the nominal voltage, that can take several hundreds milliseconds. The fault can be considered as an impedance to ground or to neutral conductor or to another phase, thus, when the voltage increases, the current increases together. The breakers in the grid are set in order to create a vertical selectivity. Only the breaker near the fault is called to inter- vene, and if it fails, the upper level breaker intervenes. However, the upper level breaker will intervene for higher current than the lower level one, in order to avoid false trips and because its nominal current is higher. Increasing the voltage linearly, the current increase linearly too, forcing to open before the breaker with the lowest current rating in the fault path. If the fault is near the converter, and the responsible breaker has an intervention current higher than the converter converter ampacity, the converter semiconductors may begin to impose a voltage squared waveform, that en- ables to withstand higher overload, due to the reduced switching losses.

Advantages in bullet points:

• Minimization of the fault transient, with the clearance of it in few hundreds of milliseconds.

• Perfect integration of grid-forming converters (e.g., STs) in the current distribu tion grid, respecting the existing vertical selectivity of the protection system.

• The converter can work on single-phase base, opening only the faulty-phase, leaving in operation the other two phases.

• With the proposed method, the impact of the short circuit current from rotating machines present in the grid is minimized, due to the reduced voltage needed for letting the breakers trip.

• The grid-forming converters (e.g., ST) is able to provide the full short circuit current in case of high breaker intervention current rating.

The invention is now described by examples, using the following figures:

Fig. 1 - voltage and current profile over time of a first embodiment and

Fig. 2 - voltage and current profile over time of a second embodiment and

Fig. 3 - square-wave modulation of output voltage and

Fig. 4 - a converter connected to a medium voltage and a low voltage grid.

A first embodiment of the approach proposed in this patent is shown in Fig.1.

Point 1 : A fault occurs in any of the grid phases. This approach can work in single-phase, two-phase, or three-phase mode. In reaction to the fault, the current rises.

Point 2:

The current reaches a maximum threshold, e.g. defined by the manufacturer of the converter to protect the converter hardware. The converter semiconductor switches involved in the grid faults trip, isolating each phase under fault. The voltage is dropped instantaneously to zero, due to the opening of the circuit (switches open).

Point 3:

The converter keeps the semiconductor switches open for a certain period that can vary from one sampling time (e.g., few microseconds) to several fundamental cycles (e.g., 40-60 ms), in order to dissipate the fault energy initially accumulated. At point 3, the converter begins to apply a voltage ramp, that can reach the steady state in the order of hundreds of milliseconds (e.g., 200-400ms), depending on the connec- tion rule of loads and generators. The current begins to increase.

Point 4:

The increase of the current leads all breakers involved in the faulty path to measure an increase in the current. The breaker that first opens is the one nearest the fault, due to the smaller intervention current. Applying the voltage ramp or any other smooth transition allows to respect the vertical selectivity, because the higher level breakers will trigger later than the one nearest the fault. When the breaker opens at point 4, the current drops due to the clearing of the fault, and it continues to increase together with the voltage increase.

Point 5:

At point 5 the normal conditions are restored, with nominal voltage. The current de- pends solely by the load demand, because the fault is cleared

The aforementioned converter protection works best if the intervention current of the breakers is below the converter rating (or converter ampacity). However, Fig. 2 shows a second embodiment of the approach proposed in this patent which is opti mized if a higher current is requested for letting the breaker trip. The embodiment of figure 2 is the same as the embodiment of figure 1 from the be- ginning until point 3 is reached. At point 3 the converter begins to apply the voltage increase function. In this operating phase, the converter may output a squared wave- form of the output voltage, as further described hereinafter. The converter may in- stantly, at point 3, switch to square-wave mode, or it may begin with normal sinusoi- dal voltage waveform and then, when a certain voltage or current is reached, change to a square-wave mode.

If a higher current is requested for letting the breaker trip, the converter has to pro- vide that current in order to avoid the permanence of the fault or long disconnection time, due to the intervention of the thermal curve of the breaker. To provide the full short circuit current, the converter applies the nominal voltage, and, as soon as the maximum allowable current is reached, the converter begins to work in square-wave mode. It imposes a squared voltage waveform at nominal frequency (or at a reduced frequency with respect to the nominal switching one), as shown in Fig. 3, that allows to minimize the switching losses (and thus the semiconductor temperature), and thus to inject the full short circuit current, although for a short amount of time. Eventually, a DC current can be supplied to the fault point, leaving the switching elements in an ON state. This allows to reduce the thermal stress in the semiconductors, and thus to increase the maximum current injection at the fault point. In this way, the breaker can sense a current higher than its intervention current limit, and open the faulty phase. As soon as the fault is cleared, the current goes back below the converter ampacity and the converter may restore the sinusoidal voltage waveform, or may continue in square-wave mode.

In presence of a communication infrastructure or communication-based protection system, the converter can upgrade its dynamic performance coordinating with the grid breakers. An automatic fault detection method allows this coordination by means of the aforementioned communication infrastructure. As operation example, the con- verter blocks its switching elements to prevent damage to its hardware and signal the fault status to all the breakers. Then the converter restores gradually the voltage to the nominal value, increasing gradually the short-circuit current. The breakers, being already pre-alerted from the grid-forming converter of the fault presence, triggers and open the circuit as soon as their intervention current thresholds are met.

Example of operations with vertical selectivity:

Let us consider the network in Fig. 4. It is composed of a converter ST, a ST-fed grid (e.g. a Low Voltage (LV) grid), and, in the first branch, of several breakers with their protection intervention currents. If a fault occurs at the end of the line (Point A), the Breaker A must intervene, because it is the nearest to the fault. The ST begins to in- crease the voltage after blocking the conduction immediately after the fault. The cur- rent increases and when it reaches the intervention current of Breaker A, its relay triggers the Breaker A to open. Breaker B and Breaker C do not measure a current high enough to be forced to intervene. In this case, the machine Q, present in a par- allel branch, does contribute to the fault current only in limited way, due to the de- creased voltage level in the grid.

In case the fault is located in point B, the Breaker A does not see the short circuit cur- rent and Breaker B must intervene. With respect to the previous case, the voltage in- creases more, because Breaker B has higher current rating. As soon as the Breaker B intervention current is met, its relay forces the opening of the breaker. Breaker C sees the short circuit current, but it does not intervene due to the current level that is lower the intervention one.

In case the fault is located in the point C, the Breaker C needs a short circuit current higher than the ST ampacity. Thus, the ST decides to finish the voltage ramp (or in- creasing the voltage enough to have the desired short circuit current) and to switch in square-wave mode. In this mode, the ST is able to inject higher current than in the si- nusoidal case, because it decreases its switching losses, switching to low frequency (50 Hz) instead of several kHz.

In all the previous cases, as soon as the fault is cleared, the current is immediately reduced, following the voltage ramp.