Control Scheme for a Cascaded Multilevel Converter Used in Low-Voltage-Ride-Through Tests of Grid-Connected Wind Turbines Fernanda Carnielutti∗ , Benhur Tessele† , Jean de Paris‡ , Jorge Massing‡ and Humberto Pinheiro‡ ∗ Power Electronics and Control Research Group - GEPOC Federal University of Santa Maria - UFSM, Campus Cachoeira do Sul, Brazil Email: firstname.lastname@example.org † Grupo de Pesquisa em Automação e Sistemas - GPAS Federal University of Santa Catarina - UFSC, Florianópolis, Brazil Email: email@example.com ‡ Power Electronics and Control Research Group - GEPOC Federal University of Santa Maria - UFSM, Campus Santa Maria, Brazil Emails: firstname.lastname@example.org, email@example.com, firstname.lastname@example.org Abstract— This paper describes the implementation of a control scheme for a symmetrical cascaded multilevel converter used as a fault generator for low-voltage-ride-through (LVRT) tests of grid-connected wind energy conversion systems (WECSs). The converter is able to emulate the main faults that can happen in real power systems, with no signiﬁcant impact on the grid during the LVRT tests. The proposed control scheme is divided in two levels: a central unit and local controllers in each converter power cell. The cells are composed of a back-to-back singlephase full-bridge converter and a local controller, a Digital Signal Processor (DSP), that controls the DC bus voltage through the grid-side converter and performs the modulation of the outputside converter. The central controller is also a DSP, that generates the voltage references and the synchronization signals for the power cells. Experimental results of LVRT tests are shown to demonstrate the good performance of the converter and the proposed control scheme. I. I NTRODUCTION The amount of installed wind power in constantly increasing in many countries around the world. According to the European Wind Energy Association (EWEA), over 13GW from Wind Energy Conversion Systems (WECS) were installed across Europe during 2015, with Germany leading the market . As a result, Power System Operators (PSOs) are beginning to require the active participation of WECS in the control of the power system to which they are connected, in a way such as their dynamic behavior more closely resembles that of conventional synchronous generators. This task can be accomplished by modern WECSs, as the majority of commercially available wind turbines are now connected to the grid via static power converters, allowing variable speed operation and active participation in the control of the power system to which they are connected . The control actions that WECS must perform in order to actively take part in the control of a power system are comprised of a list of operational requirements presented in the grid codes of the PSOs –. Among the operational l-))) requirements ﬁrst developed for conventional power plants (hydro, thermal, etc.), and now being updated for wind power, are reactive power compensation, primary frequency control and low-voltage-ride-through (LVRT) capability, this last one being one of the most important grid code requirements. During certain types of faults in the grid, the WECS must remain connected during a speciﬁc time interval, and, in some cases, provide reactive support to help the reestablishment of the system. For example, Fig. 1 shows the LVRT curve presented in the German Grid Code . The horizontal and vertical axis represent, respectively, the time and the amplitude of the peak line-to-line grid voltage. In Fig. 1, the fault occurs at time t = 0. As long as the values of the line-to-line grid voltages remain inside the blue area, the wind turbine must stay connected to the grid. Inside the lightgray area, the turbine must disconnect, if there are instabilities in the system. Note that, during a 150ms interval, even with zero voltage, the turbine must stay connected. Other Grid Codes present similar operational requirements, with LVRT curves adapted to the speciﬁc characteristics of their respective power systems. As a result, wind turbine manufacturers must test and certify the capacity of their WECS to withstand these operational requirements. Concerning LVRT, there are four possible ways to emulate faults in grid for tests of grid-connected wind turbines . Fault generators using a bidirectional two-level rectiﬁer/inverter is one of the possible ways to emulate the faults, and have already been presented in literature , – . However, as for conventional two- or three-level converters the number of levels in the output line-to-line voltages is limited, large LC ﬁlters are needed at their output terminals, and undesired interactions among the wind turbine controllers and these ﬁlters may arise. In this context, this paper describes the implementation of a cascaded multilevel converter used to emulate faults in the grid g %Vll 100% Must stay connected Power Cell A1 Limit 1 B1 15% Must disconnect Grid Limit 2 t(ms) 0 150 700 1500 C1 Input 3Φ transformer 45% di M sc ay on ne ct 70% 3000 A2 B2 Wind turbine under test C2 A3 Figure. 1. LVRT curve presented in the German Grid Code (VDEW, 2007). for LVRT tests of grid-connected wind turbines. The converter is able to emulate the main faults that can occur in real power systems. As multilevel converters reduce the distortions in the output line-to-line voltages, the output LC ﬁlter can be reduced or even totally suppressed, mitigating its interactions with the WECS under test. The symmetrical cascaded multilevel converter presented in this paper is comprised of three single-phase cells per phase. Each cell is composed of a grid-side full-bridge rectiﬁer, a DC bus and a WECS-side full-bridge inverter, as shown in Fig. 2 (back-to-back conﬁguration). Usually, symmetrical cascaded multilevel converters have in their front-end non-controlled rectiﬁers –, as this conﬁguration is easier to control. The back-to-back topology was chosen in this work as, for the LVRT application, power must ﬂow from the grid to the WECS and vice-versa. The proposed converter is a 100kVA prototype, and is a ﬁrst step towards the implementation of a new testing system rated for higher powers The control of the symmetrical cascaded multilevel converter is divided in two levels: a central unit and local controllers in each cell, as seen in Fig. 3. The cell is connected to the secondary of the transformer by a ﬁlter inductor Lf . Both the central and local controllers are implemented with a TMS320F28335 Digital Signal Processor (DSP), from Texas Instruments. This is a simpler architecture when compared to , for example, where, besides a central controller, the control of each power cell is achieved by means of an association of a DSP and an FPGA. The proposed converter is a 100kVA prototype, and is a ﬁrst step towards the implementation of a new testing system rated for higher powers, so that larger WECS can be tested. The proposed control scheme will be described in Section II of this paper, and Section III brings experimental results to validate its performance. Finally, the conclusions are drawn in Section IV. II. D ESCRIPTION OF THE P ROPOSED S YSTEM In this Section, the control levels presented in Fig. 3 are described in detail. A. Central Controller The central controller is implemented with a TMS320F28335 ﬂoating point DSP from Texas Instruments, B3 C3 vag vbg Output sensors vcg vcgf vbgf vagf Figure. 2. Grid, symmetrical cascaded multilevel converter and wind turbine under test. sr1 iL Secondary of the transformer sr 3 s3 s1 Lf Bypass sr 2 sr 4 iL vg Power Cell s2 vcc s1 , s2 , s3 , s4 sr1 , sr 2 , sr 3 , sr 4 Local DSP SCIrx SCItx - Central DSP Data Package - Cell Data Package SCItx Figure. 3. s4 Central DSP SCIrx Central Controller Control system and communication protocols. and is responsible for the overall control of the symmetrical cascaded multilevel converter. It communicates with the converter power cells through a bidirectional ﬁber optic channel. The central DSP has three UART (Universal Asynchronous Receiver/Transmiter) serial interfaces, and each one of them is used to communicate with the three cells of a given phase: phase a with UARTa, phase b with UARTb, and phase c with UARTc. All UARTs from the central and the local DSPs are set to a Baud Rate of 2.34MBd, the faster rate allowed. The bytes in a package are transmitted and received using the UARTs FIFO (First In, First Out) modules, as can TABLE I PARAMETERS FOR A LVRT TEST. Central DSP UARTc UARTb UARTa FIFORx FIFORx FIFOTx FIFORx FIFOTx Before/after Fault Parameters Description Input value Positive Sequence (SP) 0-100% of the grid voltage peak Negative Sequence (SN) 0-100% of the grid voltage peak Zero Sequence (ZN) 0-100% of the grid voltage peak Grid Frequency 30-120 Hz Angle 0-240o Test total time 0-59min, 0-59s, 0-100 tens of ms Fault Parameters During the Test Description Input value Type of fault Symmetric, asymmetric or phase jump Positive Sequence (SP) 0-100% of the grid voltage peak Negative Sequence (SN) 0-100% of the grid voltage peak Zero Sequence (ZN) 0-100% of the grid voltage peak Phase Jump Angle 0-240o Fault duration 0-10000ms FIFOTx Cell A1 Cell B1 Cell C1 FIFOTx Local DSP FIFORx FIFOTx Local DSP FIFORx FIFOTx Local DSP FIFORx Cell A2 FIFOTx Local DSP FIFORx Cell B2 FIFOTx Local DSP FIFORx Cell C2 FIFOTx Local DSP FIFORx Cell A3 FIFOTx Local DSP FIFORx Cell B3 FIFOTx Local DSP FIFORx FIFOTx Local DSP FIFORx Cell C3 Cell Data Package: Byte 1 Byte 2 Address DC Bus Status Central DSP Data Package: Byte 2 Byte 1 Address COMP11 Byte 3 COMP12 Byte 4 COMP21 Byte 5 COMP22 Byte 6 CRC Figure. 4. Central DSP, Local controllers in each power cell and data packages ﬂow and conﬁguration. be seen in Fig.4. The packages that the central DSP sends to the cells contains six bytes, also shown in Fig.4: the cell address, the comparators for the upper IBGTs s1 and s3 of the inverter (each comparator is broken in two bytes - COMP11 and COMP12 for s1 and COMP21 and COMP22 for s3 ) and a CRC (Cyclic Redundancy Check) byte, used to check if the package was corrupted during the transfer process. If the CRC calculated by the cells upon receiving a package is equal to the CRC that was sent, the cell proceeds to update the control of the DC bus voltage and the modulation of the inverter. If the CRCs are different, the local DSP awaits the arrival of a new package. The central DSP operates with a 12kHz frequency, while the cells have two operating frequencies: 2kHz for the inverters, and 12kHz for the control of the DC bus voltage by the rectiﬁers. As a result, the data packages are not sent to all the cells of a given phase at the same time, but delayed, representing the phase angle for the Phase-Shift (PS) modulation , that is, 60o in this case. The instant in which a cell receives its associated package (FIFO receiver interruption) is used to synchronize the cell to the central DSP, resulting in the correct phase angles for the PS modulation. This can be seen in Fig. 5, where the carrier of the central DSP is shown, as well as the carriers for the cells of phase a. Here, the arrows represent the FIFO interruptions for the cells. On the other hand, the central DSP receives the packages of the power cells, containing its address and the status of the DC bus voltage, in an interruption associated to that speciﬁc cell. For the performance of a new LVRT test, the before/after fault and during fault parameters, listed in Table I, are set in the central DSP. This enters in Mode 1 of operation and sends to all power cells a command for their rectiﬁers to pre-charge their DC bus capacitors, while the inverters remain inhibited. After a deﬁned pre-charge period, the cells send back to the central controller the status of their DC buses. If all the DC voltages are correct, the central DSP changes the mode of operation from pre-charge (Mode 1) to execution of the LVRT test (Mode 2). During Mode 2, the central DSP sends a command for the cells to enable their inverters, while the rectiﬁers keep regulating the DC bus voltages. The central DSP calculates the voltage references for the inverter legs of the power cells before, after and during the emulation of the fault in the grid. The voltage references are obtained by means of the carrier-based geometrical modulation described in , that, together with the PS modulation, is able to cancel harmonic components on the synthesized output line-to-line voltages. When the LVRT test is completed, the central DSP enters in Mode 3 (Idle Mode), and waits for another set of parameters to perform a new LVRT test. It also inhibits both the rectiﬁers and inverters of the cells, conﬁguring them to stay idle in Mode 3 and wait for a new test to be performed. B. Local Controllers The local controllers are also TMS320F28335 DSPs, and are responsible for the control of the DC bus voltage by the rectiﬁer and for the PWM modulation of the inverter of each cell. As previously stated, at the beginning of a new LVRT test, the central DSP conﬁgures the cells through the UART/ﬁber optic channel to enter in Mode 1 and perform the pre-charge of their DC buses. When the receiver FIFO in the local DSP receives an entire package, comprised of six bytes, it generates an interrupt; the cell checks the CRC byte and the address. If both are correct, the carriers of the rectiﬁer (12kHz) and of the inverter (2kHz) are synchronized to the central DSP, and the local control actions are performed (regulation of the DC bus voltage and update of the comparators for the inverter IGBTs). After the DC bus voltage pre-charge period, the cells check the value of the DC bus voltage and send to the central DSP their addresses and the status of the DC bus; if all voltages are correct, the central DSP commands the cells to start performing the LVRT test (Mode 2), sending the data Central DSP and DSPa1 - 2kHz rectiﬁers - 12kHz vgf1 Low-pass filter (.) 2 vgf2 -1 (.) 2 + 0º + vgf ¸ vdc2 60º vdc2 ref Notch filter 2 vdcf evdc2 + - C 2 uPI PI 120º Modulator u ff ´ iref vgf ig - DSPa2 - 2kHz Low-pass filter + DSPa3 - 2kHz vg u KP -+ eig Comp sr1 Comp sr2 Figure. 6. Figure. 5. Carrier of the central DSP and of the rectiﬁers of the cells in 12kHz and carriers for the inverters of the three cells of phase a, with phase-shift of 60o in 2kHz. packages described in the previous Section. As already stated, the inverter switches at 2kHz, and the carriers of the same phase are phase-shifted of 60o , in order to cancel harmonic components in the total output line-to-line voltages. Finally, when the LVRT test is completed, the central DSP enters in Mode 3, commanding the cells to inhibit both their rectiﬁers and inverters. The cells and the central DSP enter in an idle mode, waiting for a new LVRT test to be performed. C. DC Bus Voltage Control The control of the DC bus voltage, vdc , is performed by the local DSPs of the cells, with a sampling frequency of 12kHz. The block diagram of the control is shown in Fig. 6, where C is the capacitance of the DC bus capacitor. A signal proportional to the square of vdc passes through a discretetime notch ﬁlter, designed such as to reject frequencies around 120Hz. The outer voltage loop is composed by a PI controller with an anti-windup, resulting in the control action upi . The grid voltage at the secondary winding of the transformer, vg , is measured and the synchronization signal is obtained by two discrete-time second-order low-pass ﬁlters. The output of the ﬁlters is then normalized, resulting in vgf . The current reference iref is obtained as the multiplication of upi and vgf . The inner current loop was implemented by means of a proportional controller, that transforms the current error eig in the control action u. To compensate for possible disturbances caused by vg , the normalized grid voltage vgf is added to u in a feedforward action, resulting in uf f . This last one passes through a modulator, where it is modiﬁed such as to assume values between 0 and TPER (the maximum value of the DSP carrier), resulting in the comparators Compsr1 and Compsr2 for the rectiﬁer IGBTs sr1 and sr3 (sr2 and sr4 complementary to, respectively sr1 and sr3 ). Block diagram of the DC bus voltage control algorithm. In the next Section, experimental results are given to demonstrate the good performance of the proposed system. III. E XPERIMENTAL R ESULTS This Section brings experimental results to demonstrate the good performance of the system during LVRT tests. Fig. 7 shows the prototype of the symmetrical cascaded multilevel converter. An equipment under test was connected to the output of the converter through an LC ﬁlter with L = 0.9mH and C = 15μF, in order to emulate a WECS. The DC bus voltages of the cells were regulated to 100V. The modulation index m for the pre and post fault intervals is equal to 0.9, and the amplitudes of the phase voltages during the fault are calculated according to . The pre-charge period for the DC bus voltages of the cells is deﬁned as 5s. After this interval, if all the DC buses are regulated to the correct voltage value, the central DSP enables the inverters of the cells, and the LVRT test is performed. The total time was deﬁned as 3s, and, at half this period, a fault in the grid was emulated, lasting 1.5s. Fig. 8 shows the three-phase currents at the primary windings of the grid-side transformer, as well as the DC bus voltage of one of the cells of phase a, during a complete LVRT test for a B-Type fault (phase-ground). It can be seen that the converter has a very low impact on the grid, as, during the transient events of load connection and disconnection (when the inverter is enabled after the pre-charge period) and during the beginning and end of the fault (during the performance of the LVRT test), the grid-side currents have limited transients. The good performance of the DC bus voltage controller can also be seen, acting fast during the transients to regulate the DC bus voltage in the desired 100V value. Regarding the voltages, Figs.9 and 10 show, respectively, the phase voltages at the output of the multilevel converter (vag , vbg and vcg in Fig.2) and on the equipment under test (WECS), after the LC ﬁlter (vagf , vbgf and vcgf in Fig.2), B1 A1 B2 B3 Power cell and inductor Central DSP and RPI A2 A3 C1 Figure. 9. Phase voltages at the output of the multilevel converter, showing the transients at the beginning and end of a LVRT test for a B-Type fault (phase-ground). C3 C2 Figure. 7. converter. Implemented prototype of the symmetrical cascaded multilevel Fault emulation Pre-charge of the DC bus voltages Test Figure. 8. Three-phase currents measured at the primary windings of the grid-side transformer, and the DC bus voltage of one of the cells of phase a, during a complete LVRT test for a B-Type fault (phase-ground). Figure. 10. Filtered phase voltages on the equipment under test, showing the transients at the beginning and end of a LVRT test for a B-Type fault (phase-ground). showing the transients at the beginning and end of the LVRT test. Figs. 11, 12 and 13 show the same waveforms for a CType fault (phase-phase). In Figs. 10 and 13, it can be seen that the voltages on the equipment under test are almost sinusoidal. Fault emulation IV. C ONCLUSIONS This paper presented the implementation of a symmetrical cascaded multilevel converter used for LVRT tests of gridconnect WECS. The converter emulates the main faults that can occur in real power systems, and is comprised of three single-phase cells per phase. Each one is made of a grid-side full-bridge rectiﬁer, a DC bus and a full-bridge inverter at the output (WECS) side. The control is divided in two levels: a central control unit and local controllers in each power cell. Each local Digital Signal Processor (DSP) controls the DC bus voltage through the rectiﬁer and performs the modulation of Pre-charge of the DC bus voltages Test Figure. 11. Three-phase currents measured at the primary windings of the grid-side transformer, and the DC bus voltage of one of the cells of phase a, during during a complete LVRT test for a C-Type fault (phase-phase). the inverter. The central controller is also a DSP, that generates the voltage references for the inverters and is responsible ACKNOWLEDGMENTS The authors would like to thank the Power Electronics and Control Research Group - GEPOC, of the Federal University of Santa Maria - UFSM, and the National Counsel of Technological and Scientiﬁc Development - CNPq. R EFERENCES Figure. 12. Phase voltages at the output of the multilevel converter, showing the transients at the beginning and end of a LVRT test for a C-Type fault (phase-phase) Figure. 13. Filtered phase voltages on the equipment under test, showing the transients at the beginning and end of a LVRT test for a C-Type fault (phase-phase). for the synchronization of the converter power cells to the central timer. Communication protocols were designed and implemented to interconnect the control subsystems. Experimental results were presented for the emulation of B- and C- type faults. It was shown that the converter has a very low impact on the grid, as, during the transient events of load connection and disconnection (when the inverter is activated after the pre-charge period) and during the beginning and end of the fault (during the execution of the LVRT test), the grid-side currents have limited transients. The phase voltages on the equipment under test are almost sinusoidal, due to the multilevel characteristic of the output voltages of the symmetrical cascaded multilevel converter implemented. The proposed converter is a 100kVA prototype, a ﬁrst step towards the implementation of a testing system rated for higher powers.  The European Wind Energy Association, “Wind in power: 2015 european statistics,” in www.ewea.org, February 2016.  I. J. Gabe, Análise e Controle de Aerogeradores com Conversor Pleno Durante Afundamentos de Tensão. Doctoral Thesis, Universidade Federal de Santa Maria, 2012.  R. T. P. R. B. B. J. L. H. M. Altin, Gksu, “Overview of recent grid codes for wind power integration,” in 12th International Conference on Optimization of Electrical and Electronic Equipment (OPTIM), May 2010, pp. 1152–1160.  National Grid U.K., “The grid code 2014,” in www.nationalgrid.com, August 2014.  Networks and System Rules of the German Transmission System operators, VDN-e.v. beim VDEW, “Transmission code 2007,” in www.vdnberlin.de, August 2007.  Operador Nacional do Sistema ONS, “Procedimentos de rede 2009,” in www.ons.org.br, May 2009.  Energinet Denmark, “Technical regulation 3.2.5 for wind power plants with a power output greater than 11 kw,” in www.energinet.dk, September 2010.  C. Wessels, R. Lohde, and F. W. Fuchs, “Transformer based voltage sag generator to perform LVRT and HVRT tests in the laboratory,” in 14th International Power Electronics and Motion Control Conference, EPE-PEMC 2010, September 2010.  J. S. D. L.-A. O. A. Miguel Garca-Gracia, M. Paz Comech, “Voltage dip generator for wind energy systems up to 5mw,” Applied Energy, vol. 86, no. 4, pp. 565–574, April 2009.  J. Dirksen, “Lvrt: Low voltage ride-through,” DEWI Magazine, no. 40, pp. 56–60, August 2013.  R. Lohde and F. W. Fuchs, “Laboratory type pwm grid emulator for generating disturbed voltages for testing grid connected devices,” in 13th European Conference on Power Electronics and Applications, 2009. EPE 09, September 2009.  C. P. D. D. H.-R. W. D. S. A. Richter, J. von Bloh, “Control of a medium-voltage test generator,” in IEEE Power Electronics Specialists Conference, 2008. PESC 2008, June 2008.  C. Saniter and J. Janning, “Test bench for grid code simulations for multi-MW wind turbines, design and control,” IEEE Transactions on Power Electronics, vol. 23, no. 4, pp. 1707–1715, July 2008.  D. G. H. B. P. McGrath and W. Y. Kong, “A decentralized controller architecture for a cascaded h-bridge multilevel converter,” IEEE Transactions on Industrial Electronics, vol. 61, no. 3, pp. 1169–1178, March 2014.  Siemens, Robicon Perfect Harmony Medium-Voltage Liquid-Cooled Drives - Catalog D 15.1. Siemens, 2012.  G. Electric, MV6 Series - The ﬂexible, general purpose medium voltage drive. General Electric, 2014.  T. Atalik, M. Deniz, E. Ko, C. . Gerek, B. Gltekin, M. Ermis, and I. adrc, “Multi-DSP and -FPGA-based fully digital control system for cascaded multilevel converters used in facts applications,” IEEE Transactions on Industrial Informatics, vol. 8, no. 3, pp. 511–527, August 2012.  D. G. Holmes and T. A. Lipo, Pulse Width Modulation for Power Converters: Principles and Practices, 1st ed. Wiley Inter-Sciences, 2003.  F. Carnielutti, H. Pinheiro, and C. Rech, “Generalized carrier-based modulation strategy for cascaded multilevel converters operating under fault conditions,” IEEE Transactions on Industrial Electronics, vol. 59, no. 2, pp. 679–689, February 2012.  R. F. de Camargo, Método de Sincronização Aplicado a Conversores PWM Trifásicos. Doctoral Thesis, Universidade Federal de Santa Maria, 2006.