Comparison of a Three and Four Phase Interleaved Bidirectional DC/DC-Converter for the Operation in an Energy Storage System in Wind Turbines

Home , Kiel

15th International Power Electronics and Motion Control Conference, EPE-PEMC 2012 ECCE Europe, Novi Sad, Serbia Comparison of a Three and Four Phase Interleaved Bidirectional DC/DC-Converter for the Operation in an Energy Storage System in Wind Turbines

Sonke¨ Grunau, Manuel Fox∗, Friedrich W. Fuchs∗∗ Institute for Power Electronics and Electrical Drives / University of Kiel, Kiel, Germany, email: [email protected], ∗[email protected], ∗∗[email protected]

Abstract—The increasing penetration of wind power to the of an unsymmetrical voltage fault at the point of common electrical grid leads to the need of a stronger contribution coupling (PCC) in case of adapted power consumption of wind turbines to grid control, concerning quality and ability of the ESS. If this PCC voltage collapses, the whole stability. For the purpose of controlling the active power amount of generated power cannot be delivered to the injection, depending on the needs of the grid and not mains. The ESS could buffer this power. depending on the power generation, energy storage systems The foreseen need for storage technologies in WT leads to can be used. In this paper this kind of system with a bidirectional interleaved DC/DC-converter and electric double- many investigations in this area. Storage technologies for layer capacitors as storage technology is investigated. The wind power applications are investigated in [4], [5] and dimensioning of the converter’s main elements and its perfor- [6]. Electric double-layer capacitors (EDLC) are qualified mance for three- and four-phase interleaved configurations but still too expensive as a storage solution for short term is studied and compared. Simulations are carried out in energy storage to smooth output power and to disburden MATLAB/Simulink/Plecs. the converter system in grid fault situations. But there is Keywords—Wind Energy, Energy Storage System (ESS), a trend of decreasing prices. In [1], [7] and [8] control Interleaved Buck Converter, Bidirectional. strategies for constant output power for different generator types are investigated with ESS and EDLC and batteries. The authors of [9] and [10] use ESS to disburden WT I.I NTRODUCTION systems during grid faults. In [11] and [12] DC/DC- The amount of power injection from regenerative energy converter topologies are investigated for this purpose. sources like wind turbines (WT) to the mains is increasing The authors conclude that interleaved buck converters are steadily. The proportion of power attained from wind qualified topologies for ESS. In [13] a real time simulation energy was growing with approximately 30% worldwide a of a WT with ESS is presented. Interleaved DC/DC- year in the last couple of years [1]. It can be foreseen that converters for electronic vehicle drives have been gained in some parts of the electrical grid, especially in coastal at the institute in former research work [14]. regions, the main part of energy feed in will come from In this paper an ESS system is presented and investigated. WT in the future. These decentralized feeders normally It consists of a bidirectional n phase interleaved buck supply their energy by means of frequency converters to converter (BC) and is connected in parallel to the DC the grid. As a negative consequence, unwanted effects like link of the full back-to-back converter system at its high high harmonics occur, also favorable ones, as for example voltage side, as shown in Fig. 1. On its low voltage side the the possibility to control the power feed in. Especially the storage unit which consists of EDLC is connected. This reactive power can be controlled to stabilize the grid volt- paper aims to dimension and to control such systems with age and by using energy storage systems (ESS) the active three and four phase interleaved configurations (n = 3 and power can be fed in independently from the fluctuations of n = 4 respectively) of a bidirectional DC/DC-converter the wind, as far as the ESS and the converter ratings allow. and to compare these configurations with respect to the The grid codes (GC), like the German GC [2], contain the systems performance and the effort of components used. requirements for grid connection of WT. Because of the In section II the investigated topology of the interleaved growth of wind energy it can be predicted that the GC will bidirectional BC is presented and the main components be expanded and sharpened in the future [3]. By means of the ESS are dimensioned. Also the control strategy is of ESS it is possible to smooth the active output power described. BC configurations with n = 3 and n = 4 are in order to reduce the unwanted grid effects, especially in compared in all chapters. A case study and simulations are weak grid regions. Sharpened GC requirements could lead presented in section III. This paper ends with a conclusion to the necessity for WT to contribute to primary frequency in section IV. control in the case of a low grid frequency. In this case the amount of active output power has to be increased. This amount could be supplied by an ESS. The converter II.E NERGY STORAGE SYSTEM system of a WT could be supported by an ESS in the case There are two general concepts of combining wind of a low voltage ride through (LVRT) at grid faults. The power generation and ESS. First, an aggregated storage DC link voltage could be controlled constant in the case system can be installed at a specific point in the wind Fig. 1. Wind energy conversion system with energy storage system at the DC link with generator-side-converter (GSC) and line-side-converter (LSC) power plant or in the grid. These ESSs typically have a large storage capacity and can be constructed for example as pumped hydro, compressed air or battery plants. These concepts tend to supply energy during peak loads and save energy during times of low energy consumption. In this paper, a second concept of distributed ESS is considered where each WT contains its own ESS. It is connected to the DC link of the full power converter system and consists of a bidirectional DC/DC-converter and EDLC as storage. Such a system can be used for example to smooth the active output power, to let the WT contribute in frequency control or to enhance the Fig. 2. Proposed topology of bidirectional 3-phase (black) and 4-phase behaviour during grid faults. To reach these targets the (grey) interleaved buck converter storage capacity has to be dimensioned. The state of charge of an EDLC depends on its actual EDLCs have a low energy density on the one hand, on the voltage. The maximum voltage U results from EDLC,max other hand they provide a large power density [15]. So they the number of the cells connected serially. Because the are able to buffer huge amounts of energy in short time ESS should always be able to work at rated power, periods and can be assembled for fast dynamic processes. the minimal voltage U of the EDLC correlates EDLC,min In [13] und [16] EDLCs are also used as storage units in with the maximum current load capacity I of the ESS,max ESS for wind power applications. For the simulations an DC/DC-converter and the EDLC. By (1) it can be seen, equivalent circuit from [1] with series resistances is used. that the useable energy of the ESS depends on the voltage The dimensioning of these EDLCs and the inductances is range. described in the next subchapters. Z tmax EESS = PESS(t) dt B. Dimensioning of EDL Capacitance tmin In this paper the ESS will be used and dimensioned 1 2 2 = · CEDLC · UEDLC,max − UEDLC,min (1) to reach three main targets: to smooth the output power 2 of a WT at fluctuating wind conditions, to contribute in IESS,max can be increased by paralleling IGBTs or using primary frequency control and to enhance the behaviour at IGBTs with a higher maximum current on the one hand, low voltage ride through (LVRT). For these single targets on the other hand by paralleling single phase BC. The the capacities of the EDLCs have to be calculated. These first way leads to less switching and measurement efforts. calculations base on the required energy which has to be Parallel BC can be operated interleaved and so the total stored. To make universally valid statements the results J current ripple can be reduced. are given as a factor e with W = s as unit. The number e The comparison in the following subsections refer to describes how many seconds the rated power PN can be the dimensioning of the DC/DC-converter’s inductances stored or released. for different interleaved configurations and to the EDLC capacitance. 1) Active Power Smoothing In this operation, the ESS serves as a filter of the A. Proposed Topology active power. The active power output depends on the The ESS consists of n bidirectional interleaved BC fluctuating wind conditions. If primary WTs inject their connected in parallel. The control signals are phase shifted power to the grid, as governments plan to increase the 2·π by n of a switching period (Ts = 1/fs). The system’s regenerative power injection steadily in the future, huge rated power is PN . IGTBs are used as switching devices power oscillations could result. By means of the ESS a with a maxmimum current of Ii,max. In this paper ESS smoothing of the injected power could be necessary to systems with n = 3 and n = 4 will be investigated and avoid negative effects to the grid. It is obvious that the compared. In Fig. 2 the proposed system is shown. The quality of this smoothing depends strongly on the size of BC’s high voltage side is connected to the DC link of the storage unit. The bigger the useable storage capacity, the back-to-back converter of the WT with UDC , its low the higher oscillations in longer time periods can be voltage side is connected to the energy storage unit, using smoothed [17] and power ramp rates can be implemented. EDLCs. Its voltage depends on its state of charge. Such The smoothing effect therefore also depends on the power management. For this purpose a fuzzy-controller could be pumped storage power plants and other fast active power used [4]. stations could bring in their power within tsec = x min In the point of balanced operation (BO), the ESS must as secondary power reserve and WT contribute with 2 % be able to save and to release power. In [16] it has been of PN , the time gap in between could be buffered by WT J proposed that 20% of the WT’s rated power PN should be with ESS with a capacity of e = 1.2 · x W . buffered for approximately 1 minute for this purpose, this For the construction of the energy storage system consid- J corresponds to e = 12 W . With such a big capacity good ering tsec = 2.5 min, the stored energy could be calculated smoothing results could be reached. These investigations with (4). J could not show how the smoothing capability changes with ef-control = 3 W (4) different storage sizes, i.e. with different e. In this paper, it is approximated that the ESS should be able to save and 3) LVRT Enhancement release 5% of PN for 30 seconds each, which corresponds In the case of a voltage breakdown only a part of the to (2). But further investigations in this field are necessary. generated power can still be injected into the grid due to limited power rating of the LSC. If WTs disconnected J eP-smooth = 3 W (2) in such a case from the grid, the loss of power injection could even enforce the grid fault. So WTs have to provide 2) Frequency Control Contribution a LVRT capability [20]. To strengthen the grid a reactive Actually, the European network of transmission is using current injection is also required [2] in the fault case. an amount of 3000 MW for frequency primary control The amount of injected reactive current Ir depends on purposes. This is about 1% of the peak load of 300 GW the voltage sag depth ∆U (p.u.) and a factor k which is [18]. For a frequency deviation of ±200 mHz this reserve set up by the grid provider and differs within 0 and 10. must be activated for at last 15 min. Generally WTs are In [2] it is defined as (5), taking a voltage dead band of operated by a maximum power point tracking (MPPT) ±10% around the nominal voltage without the need of algorithm to harvest the maximum power from the wind. injection into account. In this operation mode they can not provide primary Ir/IN control reserve without an ESS. If the whole primary k = (5) control was provided by wind energy power plants, the ∆U necessary storage capacity would be 750 MWh (3). By means of the ESS, the generated power Pgen which can not be fed into the grid in case of an LVRT, should E = 3000 MW · 900 s (3) be stored and released after fault clearance. To calculate = 750 MWh the storage size for this purpose the worst case should be considered, if there was a three phase breakdown to 0 for Today the primary control reserve is provided by big 150 ms and a fault clearance described by the worst case power stations. [18] defindes that 2 % of the nominal line in the LVRT characteristic [2] and shown in Fig. 3 power has to be reserved for primary control by each (black line). Calculations about this were done in [4] and power station with a nominal power above 100 MW. [10], but the need of reactive current injection was not Fewer resources, reducing the global CO2 emission and considered, which is done in this paper. In Fig. 3, three nuclear phase-out, will certainly result in a new situation. Smaller power stations and suppliers of renewable energies will take part in the control market. In Germany, for example, there is no regulation defining the minimal size of power stations for frequency control contribution. But all contributing power stations need to reserve a minimum power of 1 MW for primary control purposes [19]. A primary control with WTs or rather wind parks will cause a permanent reserve by reducing the general active power output by a few percent or a storage of the additionally required power by ESS. In this case the necessarily stored energy for primary control contribution would be 1 MW · 900 s = 250 kWh. By using a wind park with

10 WT the energy stored by one WT would have to be Fig. 3. LVRT characteristic referring to [2] with PESS (blue line) and 25 kWh. EESS (blue area) for a worst case consideration If the main part of the generated energy is produced by renewable fluctuating sources, the grid codes need to be areas are defined: Area I reaches from t0 to t1. In this adapted to the future development of the energy market. time span the voltage sag is 100 % and no energy can be For an economic optimization, a change of requirements injected into the grid. The whole amount of Pgen has to for primary control reserve could be one option. The be stored in the ESS. Between t1 and t2 area II can be additional reserves of the secondary control need be put definded. The grid voltage rises linearly. Depending on k into effect faster. In [18] it is defined that secondary full reactive current Ir has to be injected until t2. During control will only be activated if there is an (ACE) ”Area this time the LSC has no capacity to feed in active power Control Error”. In this case, secondary control needs to be as well, so Pgen still needs to be stored in the ESS. t2 can adapted after 15 min since fault occurrence. be calculated by (6). The regulation for the secondary control demand requires 0.9 − 1/k t = · (t − t ) + t (6) a full availability of secondary reserve after 5 min. If 2 0.9 3 1 1 From t2 on, in area III, up to t3, the reactive current is current fluctuation range ∆IL,max was selected to be the linearly decreased with the voltage regeneration and the same for all interleaved configurations. So the inductances total injectable power Sgrid of the LSC increases again. of each interleaved level configuration can be dimensioned In all areas, PESS can be calculated by (7). In regions I differently. and IISgrid is equal to Qgrid, the injected reactive power. Each branch of the interleaved DC/DC-converter behaves Ton q as a one phase BC. Depending on the duty cycle a = T 2 2 s PESS(t) = Pgen(t) − Sgrid(t) − Qgrid(t) (7) with the switch-on time Ton, at the individual branch inductances Ln varying voltages UL can be found. These k 10 As a worst case consideration, is assumend as and voltages differ between UL = UDC − UESS and UL = P = P E e gen N . By means of these equations ESS and −UESS. The current fluctuation range ∆IL,i of one can be calculated as shown in (9). branch can be calculated by (11). Index i indicates the Z t3 phase current, without this index the total output current EESS = PESS(t) dt (8) of the DC/DC-converter is meant. t0 EESS J UDC · Ts 2 eLVRT = = 1.399 W (9) ∆IL,i = a − a (11) PN Ln

4) Discussion of Results For DC/DC-converters with n phases, n different areas In the three previous subsections the required relative with different UL combinations in the phases can be storage capacity e for the different ESS applications has defined. They can be numbered from l = 0 to l = n − 1. been calculated. These applications require to feed in The variables a and l are related by (12). active power (f-control), to store active power (LVRT) or both (P-smoothing). So a point of balanced operation l l + 1 < a < (12) BO with a state of charge of the EDLC of EBO can be n n calculated. It can be designated as mean value of state of charge where all operation modes are available. A power For each area the current fluctuation range can be calcu- management should always return to this point during lated. By means of l, (13) can be derived to calculate ∆IL normal operation. In Fig. 4 the three storage requirements depending on a. are visualized. It can be seen that emin needs to be at J least 4.5 W in total. To ensure full operability during UDC · Ts l J ∆IL = (1 + l − n · a) a − (13) all states of charge, a larger emin = 7.4 W is required Ln n which corresponds to the sum of the single e-requirements. Because the ESS should deliver PN in all states of charge To determine the maximum current fluctuation range ∆IL,max, (13) has to be derived by a. A maximum can 1+2l be found for adImax = 2n . By means of (14) ∆IL,max can be calculated for different n.

UDC · Ts ∆IL,max = (14) Ln · 4n

To receive an equal current ﬂuctuation range for all n at the output of the DC/DC-converter, a necessary inductance Ln can be calculated by (15), using (14).

Fig. 4. Minimum dimensioning of the ESS, here with UEDLC L L = (15) n n and the current load capacity IL,max of the DC/DC- converter and the EDLCs is limited, the minimal voltage In [14] the inductor volume for a three-phase interleaved of the EDLCs can not be 0 in general. The minimum DC/DC-converter was calculated. This volume can be voltage U signiﬁcantly depends on I . It EDLC,min L,max predicted by the energy of the n inductors EL,n. For its can be increased by paralleling IGBTs, by the utilization of calculation the square of the maximum current amplitude IGBTs with higher current load capacity or by paralleling ˆ2 of each phase IL,i,max has to be used. The branch current single BC. The maximum voltage UEDLC,max, on the 1 ﬂuctuation range is maximum for a = 2 . For a given other hand, depends on how many EDLCs are connected UESS 1 maximum duty ratio amax = smaller than , in series. It does not depend on the interleaved level. In UDC 2 (10) the necessary capacity of the EDLC can be calculated. ∆IL,i,max0 can be calculated by means of (11). In (16) the dependence of ∆IL,i,max0 and the maximum possible 2 · emin · PN 1 branch current ripple (for a = 2 ) is shown, which is n- CESS = 2 2 (10) UEDLC,max − UEDLC,min times the value of ∆IL,max.

C. Design of Inductances 2 ∆IL,i,max0 = n · 4 · (amax − amax) · ∆IL,max (16) The total mean current IL,max is divided equally to all BC-branches which then have to carry IL,max/n. To By deﬁning an output current ripple percentage x in (17), make the BCs with different n comparable, the maximum the maximum phase current amplitude in (18) can be calculated by the currents in (19) and (20). is compared to a sawtooth signal and the switching times for the IGBTs result. At interleaved operation there is one ∆IL,max = x · IL,max (17) PI-controller for each BC-branch and the sawtooth signals ∆IL,i,max0 2·π Iˆ = I + (18) are shifted by n of one switching period to one another. L,i,max L,i,max 2 The current is equally divided to all branches resulting I in an equal stress for all devices. This control structure I = L,max (19) L,i,max n requires a measurement of all branch currents. There are 2 also advanced strategies, for example current sharing, in ∆IL,i,max0 = n · x · 4 · (amax − a ) · IL,max (20) max which fewer measurement sensors are necessary [22]. The Hence EL,n can be calculated by (21). inner loop controller is limited between [amin...amax], depending on the minimal switch-on times of the IGBTs. 1 ˆ2 EL,n = n · · Ln · IL,i,max (21) In high dynamic processes the controller operates at these 2 boundaries. The relation of EL,3/EL,4 can be plotted for different The controller of the outer loop generates the reference ∗ amax in Fig. 5. To receive a small output current ripple, value for the controller of the inner loop, IL, with the target to maintain the DC link voltage UDC constant. For controller design purposes the behaviour of the closed inner loop can be described as a PT1 delay-element. Its settling time is estimated by the current dynamics. Here

Fig. 8. DC link voltage control structure Fig. 5. EL,3/EL,4 depending on x and amax (red: 0.1, green: 0.2, blue: 0.5) a PI-controller is also implemented, as shown in Fig. 8. The current reference is limited up to n · IL,i,max. the volume of the inductors can be reduced using a higher Because the actuating variables of the controllers are interleaved level, especially if UEDLC,max is smaller than 1 limited, an anti-windup is used for each PI-controller. half of the DC link voltage (amax < 2 ). This is necessary but challenging for the design process, because the plant loses its linear characteristics in high D. Control dynamic processes. The IGBTs are switched by push-pull operation and receive the switching signals by a pulse width modulation III.C ASE STUDYAND SIMULATION (PWM). So there are only two switching states 1|0 and A. Model Despriction 0|1, allowed for each branch, as shown in Fig. 6. Because of the antiparallel diodes the current can ﬂow in both A Simulation model was built and components were directions. So no discontinuous operation can take place. A dimensioned in a laboratory setup scale with the properties shown in table I. This parameter setup is used for simulations, which are done with MATLAB/Simulink/Plecs. For the dimensioning of the storage, EDLC-modules, each

TABLE I MODEL PROPERTIES

PN 30 kW CESS 16.67 F Fig. 6. Two switching states, left: 1|0, right: 0|1 fs 2.5 kHz Ri,ESS 36 mΩ UDC 700 V m 3 cascaded control with an inner current-control loop and an CDC 12 mF Ln=3 0.519 mH outer voltage-control loop is implemented. Similar control IDC,max 42.8 A Ln=4 0.389 mH structures are presented in [10] and [21]. For control design purposes the bidirectional interleaved BC can be with a capacity of CEDLC = 50 F, capacitor voltage of treated as a conventional buck converter with one branch. UC = 56 V, internal series resistance of Ri = 12 mΩ PI-controllers, as shown in Fig. 7, are used to control and a maximum steady state current of IC = 650 A, are considered [23]. In order to maintain always full power PN transfer capability and considering Ri and m EDLC modules serially connected, the minimum useable EDLC voltage is given by (22), the maximum useable EDLC voltage can be calculated by (23).

PN UEDLC,min = − IL,max · m · Ri (22) IL,max PN Fig. 7. current control structure, here for n = 3 UEDLC,max = m · UC − Ri · (23) UC the currents in the inductances by generating the duty By means of (10) the dependence of e on IL,max can be cycle a as actuating variable for the PWM. This signal plotted in Fig. 9. It can be seen that e increases with a operations it can temporarily leave this marked area.

B. Simulation Results Due to complexity the wind turbine converter system is not simulated. The generator of the wind turbine system produces an unsteady current flow due to fluctuating wind conditions but the LSC should deliver a constant power to the grid. So the ESS has to maintain the DC link voltage Fig. 9. Useable storage capacities e depending on IL,max for the same constant by absorbing the resulting current difference EDLC stack configuration (3 EDLC modules, in total with CESS and I = I − I with I as the DC current from the Ri,ESS , connected in series) and a constant maximum power of PN diff gen grid gen GSC and Igrid as the DC current flowing into the LSC, as shown in Fig. 12. higher current load capability of the DC/DC-converter for the same EDLC configuration. To fulfil the requirement of emin = 4.5 s, at least a maximum current IL,max = 275 A is necessary. Here, it is important to mention that e refers to PN and not to the maximum power Sconv,max of the DC/DC-converter. PN remains constant, Sconv,max increases with an increasing IL,max. If e was related to Sconv,max, the curve in Fig. 9 would fall for high IL,max. The inductances can be calculated by (14) to Ln=3 = 0.519 mH and Ln=4 = 0.389 mH. Fig. 10 shows the Fig. 12. Simulation model (black) of the ESS, connected to the DC link of the WT

1) General behaviour To simulate different conditions of WT and failures, the simulation is taken out for three different Idiff set points, as shown in Fig. 13. The EDLC module has a state of charge

Fig. 10. Current fluctuation range ∆IL depending on a with operation range (red) and point of balanced operation (vertical red lines) for n = 3, Ln=3 = 0.519 mH (left) and n = 4, Ln=4 = 0.389 mH (right) current fluctuation range ∆IL for three and four times interleaved operation with the calculated inductances depending on the duty cycle. The area of operation is marked in red. The point of balanced operation is marked by the vertical red line. It is important that ∆IL is small for small a, because in such operation points the highest currents IL can flow and the total effective current should not exceed the IGBT’s current limitation. In Fig. 11 the regions of

Fig. 13. Simulation with n = 3, green: actual value, red: reference value

UESS = 123 V. Firstly a positive difference current of Fig. 11. Regions of steady state operation with IL,max = 450 A Idiff = 20 A (≈ 0.5 · IDC,nom) arises, simulating a very strong rise of the wind speed or a grid fault, starting at steady state operation are shown. For the simulation model t = 0.01 s. The DC link controller reacts fast due to its a maximum converter current of 450 A is chosen which small time constant. The DC link voltage maintains almost may not be exceeded. Because of the power invariance and constant, it has an overshoot of less than 0.15 % and is the fact that UDC always remains constant, the current at controlled to its nominal value within a few milliseconds. the DC link side IDC also remains constant for all states At t = 0.02 s the difference current changes abruptly to of charge of the EDLC when PESS = PN . The duty Idiff = −20 A. Nearly the same behaviour of the controller cycle a is proportional to UESS in steady state, UEDLC could be achieved again. Due to different dynamics this can be calculated considering Ri,ESS. During dynamic process needs more time and a higher overshoot, as well as a longer transient state occurs. In section II it was strategy of this DC/DC-converter is shown. In simulations shown that voltage at the BC inductances UL changes the functionality of the use in WT to reach the proposed with each switching state. Because of the big difference targets could be shown by using test signals representing of primary voltage UDC and secondary voltage UESS the grid fault conditions and fluctuating wind conditions. absolute value of UL and so the inverter dynamics for current falling situations is slower than for current rising REFERENCES situations. [1] S. M. Muyeen, R. Takahashi, T. Murata, and J. Tamura, “Integration 20 ms later, Idiff changes to an alternating current with an of an energy capacitor system with a variable-speed wind gener- amplitude of 10 A and an offset of 25 A at a frequency of ator,” IEEE Transactions on Energy Conversion, vol. 24, no. 3, pp. 740–749, 2009. 200 Hz. In the case of unsymmetrical grid faults at the DC [2] B. der Justiz BMJ, “Verordnung zu systemdienstleistungen link voltage of a converter system an oszillation of 100 Hz durch windenergieanlagen (systemdienstleistungsverordnung - sdl- typically arises [24]. As shown in the simulation the ESS windv).” http://www.gesetze-im-internet.de/sdlwindv/, July 2009. [3] M. Altin, O. Goeksu, R. Teodorescu, P. Rodriguez, B.-B. Jensen, is capable to stabilize the DC link voltage at even higher and L. Helle, “Overview of recent grid codes for wind power frequencies. Also, oscillations due to wind fluctuations can integration,” in Proc. 12th Int Optimization of Electrical and be controlled. Electronic Equipment (OPTIM) Conf, pp. 1152–1160, 2010. [4] C. Abbey and G. Joos, “Supercapacitor energy storage for wind energy applications,” IEEE Transactions on Industry Applications, 2) LVRT vol. 43, no. 3, pp. 769–776, 2007. [5] M. Swierczynski, R. Teodorescu, C. N. Rasmussen, P. Rodriguez, In Fig. 14 the simulation shows the behaviour of a and H. Vikelgaard, “Overview of the energy storage systems for condition similar to a grid fault. From t = 0.01 s, the wind power integration enhancement,” in Proc. IEEE Int Industrial total rated current of the WT can not be supplied to the Electronics (ISIE) Symp, pp. 3749–3756, 2010. [6] M. Swierczynski, R. Teodorescu, P. Rodriguez, C. Rasmussen, grid and flows into the storage unit. The DC link voltage and H. Vikelgaard, “Storage possibilities for enabling higher wind can be kept nearly constant the whole time. energy penetration,” in Proceedings of the EPE Wind Energy Chapter Symposium, 2010. [7] Q. Li, S. S. Choi, Y. Yuan, and D. L. Yao, “On the determination of battery energy storage capacity and short-term power dispatch of a wind farm,” IEEE Transactions on Sustainable Energy, vol. 2, no. 2, pp. 148–158, 2011. [8] B. C. Ummels, E. Pelgrum, and W. L. Kling, “Integration of large- scale wind power and use of energy storage in the netherlands’ electricity supply,” IET Renewable Power Generation, vol. 2, no. 1, pp. 34–46, 2008. [9] W. Qiao and R. G. Harley, “Grid connection requirements and solutions for dfig wind turbines,” in Proc. IEEE Energy 2030 Conf. ENERGY 2008, pp. 1–8, 2008. [10] T. Nguyen and D.-C. Lee, “Ride-through technique for pmsg wind turbines using energy storage systems,” Journal of Power Electronics, vol. 10, pp. 733–728, 2010. [11] W. Li, G. Joos, and C. Abbey, “A parallel bidirectional dc/dc converter topology for energy storage systems in wind applications,” in Proc. 42nd IAS Annual Meeting Industry Applications Conf. Conf. Record of the 2007 IEEE, pp. 179–185, 2007. [12] W. Li and G. Joos, “Comparison of energy storage system technologies and configurations in a wind farm,” in Proc. IEEE Power Electronics Specialists Conf. PESC 2007, pp. 1280–1285, 2007. [13] W. Li, G. Joos, and J. Belanger, “Real-time simulation of a wind turbine generator coupled with a battery supercapacitor energy storage system,” IEEE Transactions on Industrial Electronics, vol. 57, no. 4, pp. 1137–1145, 2010. [14] J. Schroeder, B. Wittig, and F. Fuchs, “High efficient battery backup system for lift trucks using interleaved-converter and increased edlc Fig. 14. LVRT-Simulation with n = 3, green: actual value, red: voltage range,” in IECON 2010 - 36th Annual Conference on IEEE reference value Industrial Electronics Society, pp. 2334 –2339, nov. 2010. [15] M. A. Guerrero, E. Romero, F. Barrero, M. I. Milanes, and E. Gonzalez, “Overview of medium scale energy storage systems,” IV.C ONCLUSION in Proc. CPE ’09. Compatibility and Power Electronics, pp. 93– 100, 2009. In this paper, a multi interleaved buck converter for [16] L. Qu and W. Qiao, “Constant power control of dfig wind turbines the use in an ESS connected to the DC link of a WT with supercapacitor energy storage,” IEEE Transactions on Industry Applications, vol. 47, no. 1, pp. 359–367, 2011. converter system, is investigated. The main components of [17] G. Suvire and P. Mercado, “Active power control of a flywheel this system are dimensioned for the purpose of stabilizing energy storage system for wind energy applications,” Renewable the wind turbine system during grid fault, to achieve a Power Generation, IET, vol. 6, pp. 9 –16, january 2012. [18] ENTSO-E, “Ucte operation handbook.” Webseite, Juni 2009. constant output power at fluctuating wind conditions and https://www.entsoe.eu/resources/publications/system-operations/ to contribute to frequency control of the grid. Current operation-handbook/. ripple and total inductor volume can be reduced by in- [19] M. Otte et al., “Beschluss az: Bk6-10-097.” Internet, 2011. http://www.bundesnetzagentur.de/DE/DieBundesnetzagentur/ creasing the interleaved level n, on the other hand the Beschlusskammern/1BK-Geschaeftszeichen-Datenbank/BK6/ measurement and switching effort rise. Within a case study 2010/BK6-10-000bis100/BK6-10-097bis-099/BK6-10-097 the components could be determined for an ESS with Beschluss 2011 04 12.pdf? blob=publicationFile abgerufen am 13.06.2012. rated power PN = 30 kW. A comparison of this setup [20] bdew, “Technische richtlinie erzeugungsanlagen am mittelspan- for different maximum currents IL,max shows that avail- nungsnetz.” Website, June 2008. http://www.tyskret.com/dansk/ able storage capacity increases with increasing I . vindenergi/mittelspannungsrichtlinie.pdf. L,max [21] N. P. W. Strachan and D. Jovcic, “Improving wind power quality Especially in the case of high investment costs for storage using an integrated wind energy conversion and storage system components this should be taken into account. A control (wecss),” in Proc. IEEE Power and Energy Society General Meeting - Conversion and Delivery of Electrical Energy in the 21st Century, pp. 1–8, 2008. [22] J. Abu Qahouq, L. Huang, and D. Huard, “Sensorless current sharing analysis and scheme for multiphase converters,” Power Electronics, IEEE Transactions on, vol. 23, pp. 2237 –2247, sept. 2008. [23] WIMA, “Datasheet.” Website, June 2012. http://www.wima.de/DE/ WIMA SuperCap MC Powerblock.pdf. [24] H. Miranda, R. Teodorescu, P. Rodriguez, and L. Helle, “Dc-link voltage oscillations reduction during unbalanced grid faults for high power wind turbines,” in Proc. 2011-14th European Conf. Power Electronics and Applications (EPE 2011), pp. 1–11, 2011.