EP2609663A1

EP2609663A1 - Device and method for filtering in electrical power networks

Info

Publication number: EP2609663A1
Application number: EP10856339.6A
Authority: EP
Inventors: Jyri ÖÖRNI; Aki Leinonen
Original assignee: Merus Power Dynamics Oy
Current assignee: Merus Power Dynamics Oy
Priority date: 2010-08-24
Filing date: 2010-08-24
Publication date: 2013-07-03
Also published as: WO2012025660A1; EP2609663A4

Abstract

A device (600) for compensating a distorted current waveform of a three- phase network (900) comprises: - a control unit (400) configured to determine a compensating current (I10) based on a measured load current (I_1L), - a capacitor unit (300) for storing energy, and - a combination of a switch bridge (200) and inductors (L1, L2, L3) arranged to transfer energy from the three-phase network (900) to the capacitor unit (300) and to return energy from the capacitor unit (300) to the three-phase network (900) so as to generate the compensating current (I10), wherein the switch bridge (200) is arranged to alternately increase and decrease an inductor current (I_1E) of a first inductor (L1 ) of said inductors (L1, L2, L3) by connecting the first inductor (L1) to a node (RAIL2) of the capacitor unit (300) at a switching frequency (fMOD) such that an averaged value of the inductor current (I1 E) is substantially equal to the compensating current (I_1D), wherein the inductors (L1, L2, L3) comprise amorphous ferromagnetic material, and wherein the switching frequency (fMoo) is in the range from 5 kHz to 50 kHz, preferably in the range from 7 kHz to 18 kHz.

Description

DEVICE AND METHOD FOR FILTERING IN ELECTRICAL POWER NETWORKS

FIELD OF THE INVENTION

The invention relates to compensating distortion of current waveforms in electrical power networks.

BACKGROUND

Poor power quality may cause various problems in electrical power networks. The power flow should preferably have a pure sinusoidal waveform and it should remain within specified voltage and frequency tolerances. In today's electrical networks, deviations from these ideal conditions are frequent due to increasing non-linear and other loads disturbing the power network.

Waveform distortion may be caused e.g. due to thyristor power control, which utilizes phase fired power regulation. Deviations from the ideal sinusoidal waveform may also be caused e.g. by rapidly fluctuating loads such as arc furnaces, and welding devices.

A distorted current waveform may be difficult or impossible to correct by using only passive electrical filters.

A distorted waveform may be decomposed into spectral components. In particular, the waveform may be represented as a sum of harmonic components. The waveform distortion may be at least partially compensated by suppressing harmonic components.

SUMMARY An object of the invention is to provide a device arranged for filtering out harmonic current components. An object of the invention is also to provide a method for filtering out harmonic current components. According to a first aspect of the invention, there is provided a filter device (600) for compensating a distorted current waveform of a three-phase network (900), the device (600) comprising :

- a control unit (400) configured to determine a compensating current (I-|_D) based on a measured load current ( _L),

- a capacitor unit (300) for storing energy,

and

- a combination of a switch bridge (200) and inductors (L1 , L2, L3) arranged to transfer energy from the three-phase network (900) to the capacitor unit (300) and to return energy from the capacitor unit (300) to the three-phase network (900) so as to generate the compensating current (I-|_D),

wherein the switch bridge (200) is arranged to alternately increase and decrease an inductor current (I-|_E) of a first inductor (L1 ) of said inductors (L1 , L2, L3) by repetitively connecting the first inductor (L1 ) to a node (RAIL2) of the capacitor unit (300) at a switching frequency (fMOD) such that an averaged value of the inductor current (I-|_E) is substantially equal to the compensating current ( _D), wherein the inductors (L1 , L2, L3) comprise amorphous ferromagnetic material, and wherein the switching frequency (fMOD) is in the range from 5 kHz to 50 kHz, preferably in the range from 7 kHz to 18 kHz. According to a second aspect of the invention, there is provided a method for compensating a distorted current waveform of a three-phase network (900), the method comprising:

- measuring a load current ( _L),

- determining a compensating current (I-|_D) based on the measured load current ( _L),

- generating the compensating current ( _D) by using a combination of a switch bridge (200) and inductors (L1 , L2, L3) to transfer energy from the three-phase network (900) to the capacitor unit (300) and to return energy from the capacitor unit (300) to the three-phase network (900),

wherein the switch bridge (200) is arranged to alternately increase and decrease an inductor current (I-|_E) of a first inductor (L1 ) of said inductors (L1 , L2, L3) by repetitively connecting the first inductor (L1 ) to a node (RAIL2) of the capacitor unit (300) at a switching frequency (fMOD) such that an averaged value of the inductor current (I-|_E) is substantially equal to the compensating current (ho), wherein the inductors (L1 , L2, L3) comprise amorphous ferromagnetic material, and wherein the switching frequency (fMOD) is in the range from 5 kHz to 50 kHz, preferably in the range from 7 kHz to 18 kHz.

Thanks to the invention, the size of the filter device may be reduced, while still maintaining the capability to provide high compensating currents at high efficiency.

In an embodiment, the combined electrical losses for transferring energy from the network to the capacitor unit and back to the network may be e.g. smaller than or equal to 3%.

Thanks to the relatively high switching rate, acoustic noise generated by the device may be reduced and/or shifted to a frequency range which is less annoying to human ears. In an embodiment, the filter device and the method according to the present invention may be used for compensating reactive power.

The embodiments of the invention and their benefits will become more apparent to a person skilled in the art through the description and examples given herein below, and also through the appended claims.

BRI EF DESCRI PTION OF THE DRAWINGS In the following examples, the embodiments of the invention will be described in more detail with reference to the appended drawings, in which

Fig. 1 shows a filtering device arranged to compensate non-ideal waveforms of currents drawn by a load,

Fig. 2 shows functional units of the filtering device, shows an inductor unit, a switch bridge, and a capacitor unit, shows, by way of example, a timing chart for generating a compensating current, shows, by way of example, adjusting the duty cycle of a control signal in order to generate a compensating current waveform, shows six sectors defined by six switching vectors in the complex plane, shows, by way of example, correction of waveform distortion and phase imbalance, shows, by way of example, a distorted load current, and a compensating current generated by the filtering device, shows, in a three dimensional view, an inductor core comprising amorphous ferromagnetic material, shows, in a three dimensional view, an inductor comprising cable winding, and shows, in a three dimensional view, an inductor comprising foil winding.

DETAILED DESCRI PTION

Referring to Fig. 1 , an active filter device 600 may comprise an inductor unit 100, a switch bridge 200, a capacitor unit 300, a control unit 400, and a passive filter 500. The inductor unit 1 00 and the switch bridge 200 are arranged to together operate as a voltage step-up device, which transfers electrical energy from the network 900 to the capacitor unit 300. In particular, the inductor unit 1 00 and the bridge 200 may pump electrical energy from a lower network voltage to a higher voltage of the capacitor unit 300. At a later stage, the inductor unit 1 00 and the switch bridge 200 may also deliver electrical energy from the capacitor unit 300 back to the network, i.e. back from a higher capacitor voltage to a lower network voltage.

A load 800 may be connected to nodes T1 , T2, T3, TN of a three-phase electric power network 900. The load 800 draws load currents _L, i, I NL from the nodes T1 , T2, T3, TN. Based on the measured load currents I-|_L, I₃L, INL, the filter device 600 generates compensating currents I-|_D, I2D, o, I ND- As a result, the nodes T1 , T2, T3, TN draw total currents l-u, I2T, i, I NT from the network 900. For example, the total current In of the first phase PH1 is equal to the sum of the load current _L and the compensating current _D.

Compensating currents I-|_D, I2D, o may be generated by controlling average values of alternately increasing and decreasing currents _E, I2E, UE coupled through the inductors L1 , L2, L3. The average values of the inductor currents may, in turn, be controlled by controlling the average values of voltages at the nodes 62, 72, 82 of the switch bridge 200 by pulse width modulation (PWM).

The frequency f_MoD of PWM modulation may be, for example, in the range of 5 to 50 kHz. In particular, the frequency of PWM modulation may be e.g. in the order of 1 0 kHz, and the compensating currents _D, I2D, o may be arranged to compensate harmonic current components e.g. up to the 50th harmonic of the mains frequency. For example, in case of the mains frequency 50 Hz, harmonic current components up to 2.5 kHz may be compensated (= 50th harmonic of the mains frequency 50 Hz).

The operation of the bridge 200 may cause additional rapid current fluctuations which are advantageously filtered out before the compensating currents are coupled to the nodes T1 , T2, T3, TN. The additional fluctuations may be filtered out e.g. by using the passive filter 500. The operation of the filter device 600 may be controlled by the control unit 400.

The filter device 600 may comprise a voltage sensor unit 440 for monitoring voltages V_CA, V_CB of energy storage capacitors of the capacitor unit 300.

The filter device 600 may further comprise a user interface 450 e.g. for selecting an operating mode of the filter device 600. The device 600 may have different operating modes e.g. for suppressing predetermined harmonic noise components, for suppressing all harmonic noise components, and/or for compensating reactive power. The interface 450 may comprise e.g. a display and a keypad (not shown).

The filter device 600 may comprise current sensors M1 for monitoring load currents l _L, I2L, i drawn from network nodes T1 , T2, T3. The filter device 600 may comprise a voltage sensor unit 41 0 for measuring voltages of network terminals T1 , T2, T3 The filter device 600 may comprise a current sensor unit 420 for measuring compensating currents _D, I2D, o- The filter device 600 may comprise a current sensor unit 430 for measuring inductor currents I -| _E, I2E, UE-

The voltages V _T, V₂T, V_3T of the nodes T1 , T2, T3 may be monitored by voltage sensors of a voltage sensor unit 41 0. The filter device 600 may comprise the voltage sensor unit 41 0. The voltage sensor unit 410 and/or the current meters M1 may also be external components connectable to the filter device 600.

The total current Ι₂τ of the second phase PH2 is equal to the sum of the load current l₂i_ and the compensating current I₂D- The total current Ι₃τ of the third phase PH3 is equal to the sum of the load current l₃i_ and the compensating current l_3D. The total current Ι _Ντ of the neutral phase PHN is equal to the sum of the load current l_Ni_ and the compensating current I _ND-

I -I L denotes a load current drawn from (or provided to) the node T1 of the first phase PH1 by the load 800. I₂i_ denotes a load current drawn (or provided) from/to the node T2 of the second phase PH2. I_3L denotes a load current drawn (or provided) from/to the node T3 of the third phase PH3. l_Ni_ denotes a load current drawn (or provided) from/to the neutral node TN. PHN denotes the neutral phase. The voltage V _T may denote the voltage of the node T1 with respect to the neutral node TN (star configuration).

Voltage values V _T, V₂T, V_3T measured by the voltage sensor unit 410 may be sent as signals S_V-IT, S_V2T, S_V3T to the control unit 400. Current values I -| _L, l_3L measured by the current sensors M1 may be sent as signals Sn_L, S|₂L, S_!3L to the control unit 400. Current values I -| _D, I2D, o measured by the current sensor unit 420 may be sent as signals S-| _D, S₂D, S₃D to the control unit 400. Current values _E, I2E, UE measured by the current sensor unit 430 may be sent as signals S-| _E, S₂E, S₃E to the control unit 400. Voltage values V_CA, V_CB measured by the voltage sensor unit 440 may be sent as signals S_VCA, S_VCB to the control unit 400.

The switch bridge 200 may be controlled by sending control signals S-|_A, S_2A, S_3A, from the control unit 400 to the switch bridge 200. Auxiliary control signals S-|_B, S_2B, S_3B may be sent from the control unit 400 or they may be generated in the bridge 200 based on the control signals Si_A, S_2A, S_3A.

The filter device 600 may have connection terminals TND, T1 D, T2D, T3D, which can be connected to the nodes TN, T1 , T2, T3 when the device is installed.

The switch bridge 200 may also be called as a converter bridge. The filter device 600 may also be called as a voltage source converter (VSC). Among other things, the filter device 600 can generate sinusoidal voltages for the three phases with desired amplitude, frequency and phase angle.

The device 600 can be used for reactive power compensation and/or for compensation of harmonic currents. With fast vector control, the device provides ability to control active and reactive power independently. During operation, the voltages at the nodes 62, 72, 82 (with respect to the node 303) alternate between the negative voltage -V_CB and the positive voltage V_CA- The desired current (or voltage) waveform may be generated by controlling the voltage and current output of the switch bridge 200 by Pulse Width Modulation (PWM). By using the pulse width modulation, the magnitude and phase of the current can be controlled freely and almost instantaneously within certain limits.

Referring to Fig. 2, the inductor unit 100 may comprise inductors L1 , L2, L3. The inductors L1 , L2, L3 store energy during voltage step-up and step-down operations. The first inductor L1 has nodes (or terminals) 61 , 62, the second inductor L2 has nodes (or terminals) 71 , 72, and the third inductor L3 has nodes (or terminals) 81 , 82. The inductors L1 , L2, L3 store energy "reactively", and they may also be called as "reactors".

The switch bridge 200 may be arranged to connect the inductors L1 , L2, L3 alternately to the positive node RAIL1 and to the negative node RAIL2 of the capacitor unit 300 at a switching frequency f_MoD such that averaged values of the inductor currents _E, I2E, UE are substantially equal to the corresponding compensating currents I-|_D, I2D, ho-

In particular, the switch bridge 200 may be arranged to connect the inductor L1 alternately to a positive node RAIL1 and to a negative node RAIL2 of the capacitor unit 300 at the switching frequency f_MoD such that an averaged value of the inductor current _E is substantially equal to the compensating current I-|_D. In other words, the switch bridge 200 may be arranged to alternately increase and decrease the current _E of the inductor L1 by repetitively connecting and disconnecting the first inductor L1 to/from the negative node RAIL2 of the capacitor unit (300) at the switching frequency fiviOD-

The magnitudes of the compensating currents I-|_D, I2D, o may be controlled by controlling timing of operation of the switches of the switch bridge 200. Additional electrical noise caused by opening and closing the switches may be reduced or eliminated by (passive) low-pass filtering. The cores of the inductors L1 , L2, L3 may comprise ferromagnetic material, which is in an amorphous state. Thanks to the amorphous state, losses in the material can be very low. Consequently, the currents coupled through the inductors L1 , L2, L3 may fluctuate at the high frequency f_MoD without causing excessive energy loss.

The compensating current I-|_D for the first phase PH1 may be provided by low-pass filtering an inductor current _E coupled through the first inductor L1 . The compensating current \_2D for the second phase PH2 may be provided by low-pass filtering an inductor current I₂E coupled through the second inductor L2. The compensating current l_3D for the third phase PH3 may be provided by low-pass filtering an inductor current I₂E coupled through the third inductor L3. The passive filter unit 500 may comprise inductors L1 1 , L12, L13, and capacitors C1 , C2, C3 which are arranged to form an LCL low pass filter (the acronym "LCL" refers to "inductor-capacitor-inductor") together with the inductors L1 , L2, L3 of the inductor unit 100. The passive filter unit 500 may comprise resistors R1 , R2, R3 to provide damping near the resonance frequency of the LCL filter. The common node 503 of the capacitors C1 , C2, C3 may be connected to the neutral phase (i.e. to terminal TN) in order to reduce electrical noise caused by opening and closing the switches of the bridge 200. The order of the resistors R1 , R2, R3 and the capacitors C1 , C2, C3 may also be interchanged. In other words, the resistors may R1 , R2, R3 may have a common node, which is connected to the neutral phase.

In order to further suppress switching noise, the common node 303 of the capacitors CA CB may be connected to the neutral node TN of the network 900 via optional inductors L4 and/or L5.

The switch bridge 200 may comprise switches SW1 A, SW1 B, SW2A, SW2B, SW3A, SW3B and diodes D1 A, D1 B, D2A, D2B, D3A, D3B. The switch bridge 200 may comprise three pairs of switches. A first pair SW1 A, SW1 B is connected between the first positive rail RAIL1 and the second negative rail RAIL2. The common node 62 of the first pair is connected to a first inductor L1 . A second pair SW2A, SW2B is also connected between the rails RAIL1 , RAIL2. The common node 72 of the second pair is connected to a second inductor L2. A third pair SW3A, SW3B is connected between the rails RAIL1 , RAIL2. A common node 82 of the third pair is connected to a third inductor L3.

Diodes D1 A, D1 B, D2A, D2B, D3A, D3B are connected anti-parallel with the switches. During normal operation, the absolute values of the voltages V_CA, VCB of the capacitors CA, CB are higher than peak mains voltage present in the nodes T1 , T2, T3, and the polarity of the diodes is arranged such that there is no uncontrolled discharging of the capacitors CA, CB via the diodes.

The switches may be controlled by the control signals S-|_A, S-m, S₂A, S₂B, S_3A, S_3B. The signal S-|_A may control operation of the switch SW1 A, the signal S-|_B may control operation of the switch SW1 B, the signal S₂A may control operation of the switch SW2A, the signal S₂B may control operation of the switch SW2B, the signal S₃A may control operation of the switch SW3A, the signal S₃B may control operation of the switch SW3B.

The control signals S-|_A, S-m, S_2A, S_2B, S_3A, S_3B may be e.g. logical signals. A value of 1 may refer to a situation where the corresponding switch is in a conducting state (closed switch). A value of 0 may refer to a situation where the corresponding switch is in a non-conducting state (open switch).

In an embodiment, the control signals S-m, S_2B, S_3B may be generated by the switch bridge 200 itself based on the signals S-|_A, S_2A, S_3A. In other words, it may be sufficient to send only the control signals S-|_A, S_2A, S_3A, from the control unit 400.

Normally, both switches (e.g. SW1 A and SW1 B) of a pair should not be in the conducting state simultaneously, because this would mean short-circuiting charged capacitors CA, CB of the capacitor unit 300. In an embodiment, the second switch (SW1 B) of a pair (leg) may be arranged to be in the non-conducting state when the first switch (SW1 A) is in the conducting state, and the second switch (SW1 B) may be arranged to be in the conducting state when the first switch (SW1 A) is in the non-conducting state. The switch bridge 200 comprises three legs, and each leg has two operating states. Thus, the whole bridge 200 has eight (=2³) possible different states. The (desired) state of the switch bridge 200 may be defined by three-dimensional switching vectors SW. The eight switching vectors are listed in Table 1 .

Table 1 . Switching vectors SW for a 3-leg switch bridge.

The first element (S-|_A) of a switching vector SW may define the state of the switch SW1 A, the second element (S₂A) of the switching vector SW may define the state of the switch SW2A, and the third element (S₃A) of the switching vector SW may define the state of the switch SW3A. A control signal sent from the control unit 400 may comprise a value of a switching vector SW. For example, the signal S-|_A, may define the first element of the switching vector SW the signal S₂A may define the second element, and the signal S₃A may define the third element.

The switch bridge 200 may be e.g. an IGBT bridge (comprising Insulated Gate Bipolar Transistors) or a MOSFET bridge (comprising Metal Oxide Semiconductor Field Effect Transistors). The switching frequency f_MoD of each individual switch of the switch bridge 200 may be e.g. in the range from 5 kHz to 50 kHz, preferably in the range from 7 kHz to 18 kHz. The capacitor unit 300 may comprise a first capacitor CA and a second capacitor CB connected in series between the positive and negative rails RAIL1 , RAIL2. The common node 303 of the capacitors CA, CB may be connected to the neutral phase (i.e. to the node TN). The capacitance value of the capacitor CA may be substantially equal to the capacitance value of the capacitor CB (however, they may also be different).

The capacitor unit 300 may comprise voltage balancing resistors (not shown), which may be arranged to reduce a voltage unbalance between the capacitors CA and CB due to different leakage currents. The voltage balancing resistors may also act as discharging resistors arranged to reduce the voltage of the capacitors CA, CB to a safe level when the device 600 is not operating. The device 600 may further comprise additional switches and resistors (not shown) for pre-charging the capacitors prior to stable operation. The current sensor unit 420 may comprise current sensors M2 for monitoring the compensating currents I-|_D, I2D, o- The current sensor unit 430 may comprise current sensors M3 for monitoring the inductor currents _E, I2E, UE- The voltage sensor unit 440 may comprise voltage sensors M4 for monitoring the capacitor voltages V_CA, V_CB- The current sensors M1 , M2, M3 may comprise e.g. current transformers and/or Hall effect current sensors. The voltage sensors may be e.g. Hall effect voltage sensors.

The operation of the device is now discussed by referring to Figs 3 and 4.

During normal operation, the voltage V_CA over the first capacitor CA, i.e. the voltage of the first rail RAIL1 with respect to the node 303 is higher than the peak line to neutral voltage of the nodes T1 , T2, T3. This prevents uncontrolled charging of the capacitor CA via the diodes D1 A, D2A, D3A. Also the voltage difference V_CB over the second capacitor DB is higher than the peak voltage difference between each node T1 , T2, T3 and the node 303. This prevents uncontrolled charging of the capacitor CB via the diodes D1 B, D2B, D3B.

The switch bridge 200 should be operated such that the absolute values of the voltages V_CA, V_CB are kept below a predetermined limit, in order to avoid damaging the switches SW1 A, SW1 B, SW2A, SW2B, SW3A, SW3B and the capacitors CA, CB due to electric breakdown.

A nominal RMS line-to-neutral voltage of the nodes T1 , T2, T3 may be e.g. from 230 V to 400 V. The positive peak voltage of the nodes T1 , T2, T3 corresponding to the RMS voltage 230 V is +325 V (=the RMS voltage multiplied by V2). The negative peak voltage of the nodes T1 , T2, T3 corresponding to the RMS voltage 230 V is - 325 V. In this case, the voltage of the positive rail RAIL1 may be kept e.g. in the range of +350 V to +450 V, and the voltage of the negative rail RAIL2 may be kept e.g. in the range of - 350 V to -450 V with respect to the common node 303. RMS refers to the root mean square.

IcA denotes a current to the first capacitor CA, and I_CB denotes a current to the second capacitor CB. V_Li , V_L2, and V_L3 denote voltages over the respective inductors L1 , L2, L3. l_Si A denotes a current conducted by the switch SW1 A or by the diode D1 A. I_{S B} denotes a current conducted by the switch SW1 B or by the diode D1 B. The timing chart of Fig. 4 illustrates operation of the device 600 in a simplified situation, where the contribution of the inductors L2, L3 and the switches SW2A, SW2B, SW3A, SW3B is not taken into account.

TON denotes the length of an interval during which the switch SW1 A is set to the conducting state, and TOFF denotes the length of an interval during which the switch SW1 A is set to the non-conducting state. T_MOD denotes the combined length of said two intervals, i.e. T_MOD = TON + TOFF- The switching frequency (switching rate) f_MoD of the switch SW1 A is equal to 1 /T_MOD- The duty cycle of the switch SW1 A is equal to T₀N/TMOD- Referring to Fig. 5a, a desired current waveform of a compensating current o may be provided by adjusting the duty cycle TON/TMOD-

Referring back to Fig. 4, the time derivative of the compensating current _D may depends on the duty cycle ΤΌΝ/TMOD- For example, the duty cycle TON TMOD may be selected such that the generated compensating current _D slowly decreases.

At the time t_a, the switch SW1 B is set to the conducting state (S1 B =1 ), and the inductor L1 is connected between the network node T1 and the negative rail RAIL2. The inductor current _E increases. Consequently, energy may be transferred from the network node T1 to the inductor L1 .

At the time t_b, the switch SW1 B is set to the non-conducting state (S1 B =0), and the inductor L1 is disconnected from the negative rail RAIL2. The inductor current _E starts to decrease, and the polarity of the voltage V_Li over the inductor L1 is reversed. The inductor current l _E may now flow via the diode D1 A to the positive rail RAIL1 Consequently, energy may be transferred from the inductor L1 to the capacitor CA.

At the time t_c, the switch SW1 B is set to the conducting state (S1 B =1 ), and the above-mentioned cycle may be repeated.

In this simplified example, the first capacitor CA was charged and the second capacitor CB was discharged by the inductor L1 . The charging state of the capacitors CA, CB may be balanced by timed operation of all switches of the switch bridge 200. In particular, when the network node T1 has a positive voltage, the capacitor CB may be charged by transferring energy from those network nodes T2, T3 which have a negative voltage.

Energy stored in the capacitor CA may be transferred back to the network node T1 , and/or to the other network nodes T2, T3 by adjusting timing of the control signals S-|_A, S₂A, S₃A, i.e. by adjusting timing of the states of the legs of the switch bridge 200.

In particular, a compensating current ( _D) may be generated by: - transferring energy from a network node (T1 ) to an inductor (L1 ),

- charging a capacitor (CA) by transferring energy from the inductor (L1 ) to the capacitor (CA), and

- discharging the capacitor (CA) by repetitively connecting the inductor (L1 ) to the capacitor (CA) by a switch (SW1 A) in order to generate the compensating current ( _D).

The method for modifying a current waveform may comprise:

- measuring a load current ( _L),

- transferring energy from a network node (T1 ) to an inductor (L1 ),

- discharging the capacitor (CA) by repetitively connecting the inductor (L1 ) to a the capacitor (CA) by a switch (SW1 A) in order to generate the compensating current ( _D),

wherein the voltage (V_CA) of said capacitor (CA) is higher than the voltage of the network node (T1 ), the inductor (L1 ) comprises amorphous ferromagnetic material, and a switching frequency (fMOD) of said switch (SW1 A) is in the range from 5 kHz to 50 kHz.

Energy transfer from the network 900 to the capacitors CA, CB comprises storing energy in the inductors L1 , L2, L3. During normal operation, the voltage of the positive rail RAIL1 is always higher than the voltage of the network node T1 . Therefore, the compensating current I-|_D provided by the network node T1 can charge the capacitor CA via the diode D1 A only when the inductor L1 provides an additional voltage difference V_Li , which increases the voltage of the node 62 to the voltage of the positive rail RAIL1 (e.g. during the period from t_b to t_c in fig. 4) or slightly above the voltage of the positive rail RAIL1 . The voltage difference V_Li for charging the capacitor CA is generated by consuming energy previously stored in the inductor L1 . Energy may be stored in the inductor L1 by connecting it to the negative rail RAIL2 by the switch SW1 B (e.g. during the period from t_a to t_b in Fig.4). Transferring energy from the capacitor CA to the network node T1 also comprises storing energy in the inductor L1 . When the switch SW1 A is closed, the absolute value | _E| of current I-|_E from the capacitor CA increases, wherein energy is transferred from the capacitor CA to the node T1 . As the absolute value | _E| of the current increases, a portion of the energy removed from the capacitor CA is stored in the inductor L1 . In this case the energy stored in the capacitor CA cannot be directly transferred to the network node T1 , but it is utilized for charging the second capacitor CB via the diode D1 B. Energy stored in the second capacitor CB may be fed to the network by closing the switch SW1 B.

The operation of the filter device 600 may comprise e.g. the following steps in the following order:

- transferring energy from a network node T1 to an inductor L1 ,

- transferring energy from the inductor L1 to the first capacitor CA,

- transferring energy from the first capacitor CA to the inductor L1 ,

- transferring energy from the inductor L1 to a second capacitor CB, and

- transferring energy from the second capacitor CB to the network node T1 . The filter device 600 may also be understood to operate such that the inductor currents _E, \∑E, UE fluctuate around the desired compensating current values l _D, I2D, ho-

Connecting the inductor L1 to the positive rail RAIL1 by the switch SW1 A (or by the diode D1 A) decreases the inductor current I-|_E (see e.g. the interval from t_b to t_c in Fig. 4). Connecting the inductor L1 to the negative rail RAIL2 by the switch SW1 B (or by the diode D1 B) increases the inductor current l _E (see e.g. the interval from t_b to t_c in Fig. 4). The averaged inductor current I-|_E, when averaged over a single modulation period T_MOD, may be substantially equal to the compensating current I-|_D. The control unit 400 may be arranged to control the bridge 200 so that the average values of the fluctuating inductor currents _E, l_2E, UE are substantially equal to the compensating current values I-|_D, I2D, ho- An optimum switching sequence for the switch bridge 200 may be determined by using space vector transformations, in particular by using the so-called synchronous reference frame theory, which includes the concept of the zero sequence voltage and the concept of the zero sequence current.

The switches of the bridge 200 may be operated such that the device 600 generates the required compensating currents I 1 D; I2D; UD- The control unit 400 may determine the switching vectors and the durations for each applied switching vectors such that the compensating currents _D, I2D, o may be generated.

As an intermediate step, also a voltage reference vector VECT_V and a zero- sequence voltage reference u_z* may be determined. The complex-valued voltage reference vector VECT_V represents voltages of the nodes 62, 72, 82 averaged over a modulation period (T_MOD), and the real-valued zero- sequence voltage reference u_z* represents a phase imbalance at the nodes 62, 72, 82 averaged over a modulation period (T_MOD)- The voltage reference vector VECTv and the zero-sequence voltage reference u_z* define a desired voltage output of the bridge 200. The complex-valued voltage reference vector VECTv and the real-valued zero-sequence voltage reference u_z* may be calculated e.g. by the alpha-beta transformation from desired average voltages of the nodes 62, 72, 82.

The desired average voltages of the nodes 62, 72, 82 may be determined based on the instantaneous voltages V _T, V₂T, V_3T of the network nodes T1 , T2, T3 and based on the desired compensating currents I -| _D, I2D, o such that the device 600 generates the compensating currents I 1 D; I2D; UD when the output of the bridge 200 is coupled to the network terminals T1 , T2, T3 via the inductors L1 , L2, L3.

After the voltage reference vector VECT_V and the zero-sequence voltage reference u_z* needed for generating the desired compensating currents _D, I₂D, UD have been determined, the corresponding switching sequence of the bridge 200 may be determined by using the vector diagram of Fig. 5b. The 3-leg bridge 200 can be controlled by using the eight different switching vectors listed in Table 1 . The switching vectors (0,0,0) and (1 , 1 ,1 ) are so- called zero switching vectors, which may be used to control the zero sequence voltage. The remaining six switching vectors may be associated with boundaries of sectors in the complex plane, as shown in Fig. 5b. The vectors (1 ,0,0) and (1 , 1 ,0) define a first sector (I). The vectors (1 , 1 ,0) and (0, 1 ,0) define a second sector (I I). The vectors (0, 1 ,0) and (0, 1 , 1 ) define a third sector (I I I), the vectors (0, 1 ,1 ) and (0,0,1 ) define a fourth sector (IV), the vectors (0,0, 1 ) and (1 ,0, 1 ) define a fifth sector (V), and the vectors (1 ,0, 1 ) and (1 ,0,0) define a sixth sector (VI). The sectors are indicated by the Roman numerals I - VI. The angular width of each sector is 60 degrees. Re refers to the real axis and Im refers to the imaginary axis of the complex plane.

The desired voltages at the nodes 62, 72, 82 may be generated by selecting and using the two switching vectors, which are closest to the determined voltage reference vector VECT_V in the complex plane. For example, when the voltage reference vector VECT_V resides in the 1 st sector, the switching vectors (1 ,0,0) and (1 ,1 ,0) may be used for implementing the desired voltage vector VECTv.

In addition to the selected ones, the zero switching vectors (0,0,0) and (1 , 1 , 1 ) may be used in the beginning and in the end of each modulation half period in order to generate the desired zero-sequence voltage. The determined voltage reference vector VECT_V resides in one of the six sectors shown in Fig. 5b. When the mains frequency of the three phase network is 50 Hz, the corresponding cycle period is 20 ms. During the 20 ms cycle, the voltage reference vector VECT_V draws a complete circle in the complex plane. This means that during the 20 ms cycle, six different modulation schemes should be applied.

Table 2 indicates a modulation sequence (i.e. a switching sequence) based on the two switching vectors selected by using the sector diagram of Fig. 5b. Each modulation period may be started by using the zero switching vector (0,0,0), followed by the first selected switching vector, the second selected switching vector, and finally by the other zero switching vector (1 , 1 , 1 ). The order is selected such that the state of only one control parameter (Si_A, S₂A, S₃A) needs to be changed at a time, i.e. so that only one element of the switching vector is changed at a time. Yet in other words, the first active switching vector may be selected such that the states of switches in only one leg of the bridge 200 needs to be changed at a time.

We assume here that only one switch of each leg of the switch bridge is in the conducting state at a time. After the midpoint of the modulation period, the switching vectors are applied in reverse order. T_Zi denotes duration for using the zero switching vector (0,0,0,), T-i denotes duration for using the first selected switching vector, T₂ denotes duration for using the second selected switching vector, and T_Z2 denotes duration for using the zero switching vector (1 , 1 , 1 ). The duration of the 1 st modulation half period may be equal to the duration of the 2nd modulation half period. The period TON shown in Fig. 4 may be equal to the sum T +T₂+Tz2+Tz2+ T₂+ T-i .

Table 2. Modulation sequence when the voltage vector VECT_V resides in sector I.

The duration (T_Zi , T-i , T₂, T_Z2) of each switching state of the modulation sequence may be calculated from the values of the length of the voltage reference vector VECT_V, from the phase angle of the VECT_V, and from the magnitude of the zero sequence reference voltage u_z*. In particular, the durations of the periods T _; T₂, T_Zi , T_Z2 may be calculated by using the following equations:

M _ _{{Ti + T2 )} (3)

r_d=^ (5)

T_z2 =(X - k)T_z (6) where T_MOD denotes the duration of the modulation period (T_MOD = 2·(Τζι+Τι , 2+TZ2)), IVECTvl denotes the length of the voltage vector VECT_V, V_CA and VCB denote the voltages of capacitors, Θ^* denotes the angle of the voltage vector VECTv, θι and θ₂ denote the angles of the active (selected) switching vectors forming the sector where the voltage vector VECT_V resides, T_z = zi+Tz2, Uz* denotes the zero-sequence voltage reference calculated by using the synchronous reference frame theory, and k denotes a weighting coefficient for providing the desired zero-sequence voltage reference. All angles are defined in relation to the real axis of the stationary reference frame.

The control unit (400) of the filter device (600) may be configured to: - determine a voltage vector (VECT_V) based on measured load currents ( _L, I₂L, I3L.) and based on measured voltage values (V _T, V₂T, V_3T) of the three- phase network (900),

- select switching vectors (SW) based on the determined voltage vector (VECTv),

- determine durations (T _; T₂) for each selected switching vector (SW) based on the determined voltage vector (VECT_V), and

- drive the switch bridge (200) by using the selected switching vectors (SW) and the determined durations (T _; T₂).

The durations T_Zi , T _; T₂, T_Z2 of the applied switching vectors may be determined such that the following conditions are fulfilled:

- 1 st priority: The capacitor voltages V_CA and V_CB should be kept below a maximum limit in order to avoid damaging the capacitors, the switches, and the diodes.

- 2nd priority: The capacitor voltages V_CA and V_CB should be kept above a minimum limit in order to prevent uncontrolled currents via the diodes of the bridge.

- 3rd priority: The compensating currents _D, 1₂D, 3D should be generated. -4th priority: Capacitor voltages V_CA and V_CB should be kept in the vicinity of a predetermined optimum voltage in order to ensure that the device 600 is capable of absorbing and providing energy in rapidly varying load conditions.

Fig. 6 shows compensation of both waveform distortion and phase imbalance. A thyristor controlled load 800 may cause waveform distortion of the type shown in Fig. 6. During the interval INTER1 , the load current l _L of the first phase may be smaller than an extrapolated reference current I -I L,REF, which approximates pure sine wave. Thus, energy could be transferred from the first phase to the capacitor unit 300 during the period INTER1 in order to compensate waveform distortion. During another interval INTER2, the load current l_2L of the second phase may be higher than an extrapolated reference current I₂L,REF, which approximates pure sine wave. In this case, energy could be transferred from the capacitor unit 300 to the second phase during the interval INTER2 in order to compensate waveform distortion. Energy stored during the first interval INTER1 can be used during the second interval INTER2. Compensating currents I-|_D, I2D, I3D may be determined, generated, and coupled to the nodes T1 , T2, T3 in order to draw pure sine wave currents l-u, I₂T UT from the network.

When filtering currents of a three-phase network, the other legs of the switch bridge 200 and the other inductors L2, L3 may also operate in a similar fashion, and also the other network nodes T2, T3 may participate in the energy transfer steps. Referring to Fig. 6, energy may be extracted e.g. from the first phase (node T1 ), and returned to e.g. the second phase (node T2).

It is known that reactive power may be compensated by using inductors and/or capacitors. However, compensation of reactive power by using only passive components is problematic when the reactive power varies rapidly in time. The device and the method according to the present invention may also be used for compensating rapidly varying reactive power.

Referring to Fig. 7, waveforms of the compensating currents I-|_D, I2D, o may be determined based on the measured load currents I-|_L, I2L, i even when the waveforms of the load currents are arbitrary.

A compensating current _D needed for compensating the distortion of a load current _L may be determined e.g. by subtracting the instantaneous value of the load current _L(t) from the instantaneous values 11 L_,REF(t) of a reference current waveform. The compensating currents for the other phases PH2, PH3 may be determined in the same manner.

Fig. 7 shows, by way of example, a distorted load current _L, and a compensating current I-|_D determined by subtracting the load current _L from a sinusoidal reference current I -I L.REF- The lowermost curve of Fig. 7 shows a substantially sinusoidal total current In, which is drawn from the network 900 when the load current I-|_L of the first phase PH1 and the compensating current I-|_D are summed at the node T1 . The reference current I I L.REF is a target current or a "desired" current, which should be drawn from the network in an optimum situation. The reference current I H_,REF may be e.g. a substantially sinusoidal current. The reference current I -I L.REF may be selected such that phase imbalance is compensated. The reference current I H_,REF rnay also be selected such that reactive load will be compensated.

Reference currents I₂L,REF, I2L,REF for the other phases PH2, PH3 may be determined and used, respectively. The "optimum situation" depends on the application. In different operating modes, the device 600 may be set to correct of phase imbalance, suppress predetermined harmonic noise components, suppress all harmonic (overtone) components, and/or compensate reactive power. An attempt to simultaneously correct all these non-idealities would often lead to an over- rated system, where the inductors, capacitors and transistors would be very large. In certain applications, it is not necessary to correct all these non- idealities. A user may set the desired desired correction mode e.g. via the user interface 450 of the active filter device 600 (Fig. 1 ). The term "harmonic current component" refers to a spectral component of a current waveform, whose frequency f_q is equal to an integer multiple of the fundamental mains frequency f₀ of the network, i.e. f_q = q · fo, where q is an integer. The fundamental mains frequency f₀ is typically equal to 50 Hz or 60 Hz.

The device 600 may be arranged to suppress one or more current components selected from a group consisting of current components whose frequency f_q is in the range of 2-f₀ to 25-f₀. The attenuation of each component may be individually selectable. The attenuation of a current component may be in the range of 0 to 1 00%.

In addition to the current components individually selected from the range of 2-f₀ to 25-fo, the device 600 may be arranged to suppress all current components whose frequency f_q is in the range of 26 f₀ to 50-f₀. The maximum frequency component of a compensating current l _D, which can be generated by the device 600 depends on the switching frequency f_MoD and on the properties of the low pass filter 500.

The device 600 may be arranged to suppress one or more current components selected from a group consisting of current components whose frequency f_q is in the range of 2-f₀ to 50 f₀.

The device 600 is typically arranged to operate such that the component at the fundamental mains frequency f₀ (1 st harmonic) is not significantly suppressed.

Referring to Fig. 8a, the cores 50 of the inductors L1 , L2, L3 may comprise amorphous ferromagnetic material 51 so as to enable a high switching rate for the switches of the bridge 200.

In particular, the core 50 may comprise a plurality of stacked layers of amorphous material 51 . Insulating material 52 may be disposed between the amorphous layers in order to minimize eddy currents. The insulating material 52 may be e.g. paper, plastic or lacquer.

The cores of the inductors L1 , L2, L3 may comprise ferromagnetic material, which is in an amorphous state. Thanks to the amorphous state, magnetic hysteresis loss in the material can be low. Also the magnitude of eddy currents may be low. Consequently, the direction of the current coupled through the inductors L1 , L2, L3 may contain high frequency components without causing excessive energy loss.

Amorphous ferromagnetic alloys can be fabricated by e.g. by rapid cooling and solidification a liquid alloy (e.g. Fe, Co, or Ni with B, C, Si, P, or Al). For example, the core of the inductors may comprise an amorphous alloy Fe₈oB₂o- During manufacturing, the cooling rate during solidification may be e.g. in the order of 10⁶ °C /s.

In particular, the material of the core may be an amorphous iron-based alloy, which comprises 85-95% iron, 6-10% silicon, and 1 -5 % boron (by weight). The core loss is advantageously less than 3 W/kg at 1 0 KHz sinusoidal modulation frequency (when magnetic flux density = 0.1 Tesla).

A long sheet of amorphous material may be rolled together with the insulating material 52 so as to form the core 50. The core 50 may be subsequently cut into two parts so as to form an air gap GAP1 . The cutting of the core into two parts also facilitates installation of the core through the windings 40a, 40b of the inductors L1 , L2, L3.

SX, SY and SZ denote orthogonal directions.

Referring to Fig. 8b, the windings may be made of round or flat wire 60. For example, insulated aluminum or copper wire may be used. The inductor may comprise two coils 40a, 40b connected in series or in parallel. Alternatively, the inductor may have only one coil 40a. The use of two coils 40a, 40b may provide a more compact construction and improved cooling.

The coils 40a, 40b may comprise so called Litz wire (not shown). The Litz wire comprises a bundle of individually insulated wires connected in parallel such that the skin effect and the proximity effect losses may be reduced.

Referring to Fig. 8c, the windings may be made from metal sheet so as to provide a compact and stable mechanical construction, to provide silent operation, and/or to minimize eddy current losses at high operating frequencies. For example, aluminum or copper sheet may be used to implement the winding(s) 40a, 40b of the inductors L1 , L2, L3. The different turns of the coils 40a, 40b may be isolated from each other e.g. by lacquer, plastic foil or paper. The filter device 600 may be used for compensating a current waveform distortion when the distortion is caused by the load 800. Thus, the device 600 may prevent electrical disturbances generated by the load 800 from propagating via the network 900 to other devices. The currents l-u, I2T, i drawn from the network 900 may be substantially sinusoidal. In particular, the corrected waveforms of the currents In, I2T, i may comply with power quality specifications adopted by an electricity distribution network operator. The load 800 causing waveform distortion may comprise e.g. a variable speed drive, an electric motor during rapid change of rotation speed, a transformer operating near saturation, a gas discharge lamp, a solid state rectifier, a welding device, or an arc furnace.

The drawings are schematic. For the person skilled in the art, it will be clear that modifications and variations of the devices and methods according to the present invention are perceivable. The particular embodiments described above with reference to the accompanying drawings are illustrative only and not meant to limit the scope of the invention, which is defined by the appended claims.

Claims

1 . A device (600) for compensating a distorted current waveform of a three- phase network (900), the device (600) comprising :

- a capacitor unit (300) for storing energy,

and

- a combination of a switch bridge (200) and inductors (L1 , L2, L3) arranged to transfer energy from the three-phase network (900) to the capacitor unit

(300) and to return energy from the capacitor unit (300) to the three-phase network (900) so as to generate the compensating current (I-|_D),

wherein the switch bridge (200) is arranged to alternately increase and decrease an inductor current (I-|_E) of a first inductor (L1 ) of said inductors (L1 , L2, L3) by connecting the first inductor (L1 ) to a node (RAIL2) of the capacitor unit (300) at a switching frequency (fMOD) such that an averaged value of the inductor current (I-|_E) is substantially equal to the compensating current ( _D), wherein the inductors (L1 , L2, L3) comprise amorphous ferromagnetic material, and wherein the switching frequency (fMOD) is in the range from 5 kHz to 50 kHz, preferably in the range from 7 kHz to 18 kHz.

2. The device (600) of claim 1 , wherein the combination of the switch bridge (200) and the inductor (L1 ) is arranged to transfer energy from a node (T1 ) of the three-phase network (900) to the capacitor unit (300) and to return energy from the capacitor unit (300) to said node (T1 ).

3. The device (600) of claim 1 or 2 wherein the capacitor unit (300) comprises a first capacitor (CA) and a second capacitor (CB) connected in series, and the common node (303) of the capacitors (CA, CB) is connected to the neutral phase (TN) of the network (900).

4. The device (600) according to any of the claims 1 to 3 wherein the voltage of a positive node (RAIL1 ) of the capacitor unit (300) is higher than than a peak voltage (V _T) of the three-phase network (900).

5. The device (600) according to any of the claims 1 to 4, wherein the control unit (400) is configured to:

- determine a voltage vector (VECT_V) based on measured load currents ( _L, I₂L, I3L.) and based on measured voltage values (V _T, V₂T, V_3T) of the three- phase network (900),

- select switching vectors (SW) based on the determined voltage vector (VECTv),

6. The device (600) according to any of the claims 1 to 5, wherein the switch bridge (200) comprises insulated gate bipolar transistors.

7. A method for compensating a distorted current waveform of a three-phase network (900), the method comprising :

- measuring a load current ( _L),

- determining a compensating current (I -| _D) based on the measured load current ( _L),

- generating the compensating current ( _D) by using a combination of a switch bridge (200) and inductors (L1 , L2, L3) to transfer energy from the three-phase network (900) to the capacitor unit (300) and to return energy from the capacitor unit (300) to the three-phase network (900) ,

wherein the switch bridge (200) is arranged to alternately increase and decrease an inductor current (I -| _E) of a first inductor (L1 ) of said inductors (L1 , L2, L3) by connecting the first inductor (L1 ) to a node ( RAI L2) of the capacitor unit (300) at a switching frequency (fMOD) such that an averaged value of the inductor current (I -| _E) is substantially equal to the compensating current ( _D), wherein the inductors (L1 , L2, L3) comprise amorphous ferromagnetic material, and wherein the switching frequency (fMOD) is in the range from 5 kHz to 50 kHz, preferably in the range from 7 kHz to 18 kHz.

8. The method of claim 7 comprising transferring energy from a node (T1 ) of the three-phase network (900) to the capacitor unit (300) and transferring energy from the capacitor unit (300) back to said node (T1 ).

9. The method claim 7 or 8 wherein the capacitor unit (300) comprises a first capacitor (CA) and a second capacitor (CB) connected in series, and the common node (303) of the capacitors (CA, CB) is connected to the neutral phase (TN) of the network (900).

10. The method according to any of the claims 7 to 9 wherein the voltage of a node (RAIL1 ) of the capacitor unit (300) is higher than than a peak voltage (V-ιτ) of the three-phase network (900).

1 1 . The method according to any of the claims 7 to 10 comprising:

- determining a voltage vector (VECT_V) based on measured load currents (l-i L, I2L, I3L.) and based on measured voltage values (V _T, V₂T, V_3T) of the three-phase network (900),

- selecting switching vectors (SW) based on the determined voltage vector (VECTv),

- determining durations (T _; T₂) for each selected switching vector (SW) based on the determined voltage vector (VECT_V), and

- driving the switch bridge (200) by using the selected switching vectors (SW) and the determined durations (T _; T₂).

12. The method according to any of the claims 7 to 1 1 , wherein the switch bridge (200) comprises insulated gate bipolar transistors.