Design Study of Doubly-Fed Induction Generators for a 2MW Wind Turbine

Transcription

1 WIND ENGINEERING VOLUME 33, NO. 5, 29 PP Deign Study of Doubly-Fed Induction Generator for a 2MW Wind Turbine Rebecca Todd, Mike Barne and Alexander C. Smith School of Electrical & Electronic Engineering,The Univerity of Mancheter, PO Box 88, Sackville Street, Mancheter, UK, M6 QD, {rebecca.todd, mike.barne, ABSTRACT A deign tudy for a 2 MW commercial wind turbine i preented to illutrate two connection method for a tandard doubly-fed induction machine which can extend the low peed range down to 8% lip without an increae in the rating of the power electronic converter. Thi far exceed the normal 3% lower limit. The low peed connection i known a induction generator mode and the machine i operated with a hort circuited tator winding with all power flow being through the rotor circuit. A two loop cacaded PI control cheme ha been deigned and tuned for each mode. The purpoe of thi paper i to preent imulation reult which illutrate the dynamic performance of the controller for both doubly-fed induction generator connection method for a 2 MW wind turbine. A imple analyi of the rotor voltage for the doubly-fed connection method i included a thi demontrate the dominant component that need to be conidered when deigning uch advanced control trategie. Keyword: Doubly-fed, Induction generator, Wind turbine. LIST OF IMPORTANT SYMBOLS v rdq i rdq λ dq P Q pf T e p L m R r L r σ ω f Stator referred Rotor referred * erence value Direct and quadrature rotor voltage Direct and quadrature rotor current Direct and quadrature tator flux linkage Stator real power Stator reactive power Stator power factor Torque Differential operator Magnetiing reactance Rotor reitance Rotor reactance Total leakage inductance Slip frequency

2 498 DESIGN STUDY OF DOUBLY-FED INDUCTION GENERATORS FOR A 2MW WIND TURBINE. INTRODUCTION There i continuing interet in wind turbine, epecially thoe with a rated power of manymegawatt. Thi popularity i largely driven by both environmental concern and alo the availability of foil fuel. Legilation to encourage the reduction of the o called carbon footprint i currently in place and o interet in renewable i currently high. Wind turbine are till viewed a a well etablihed technology that ha developed from fixed peed wind turbine to the now popular variable peed technology baed on doubly-fed induction generator (DFIG). A DFIG wind turbine i variable peed with the rotor converter being controlled o that the rotor voltage phae and magnitude i adjuted to maintain the optimum torque and the neceary tator power factor [, 2, 3]. DFIG technology i currently well developed and i commonly ued in wind turbine. The tator of a DFIG i directly connected to the grid with a power electronic rotor converter utilied between the rotor winding and the grid. The variable peed range i proportional to the rating of the rotor converter and o by limiting the peed range to ±3% [4, 5, 6, 7] the rotor converter need only be rated for 3% of the total DFIG power whilt enabling full control over the full generator output power. Thi can reult in ignificant cot aving for the rotor converter [4]. The lip ring connection to the rotor winding however mut be maintained for reliable performance. The power generator peed characteritic hown in figure i for a commercial 2 MW wind turbine. The generator peed varie with wind peed however thi relation i et for a pecific location. A wind peed, and therefore machine peed, fall the power output of the generator reduce until the wind turbine i witched off when the power extracted from the wind i le than the loe of the generator and converter. An operating mode ha been propoed by a wind turbine manufacturer that i claimed to extend the peed range o that at lower peed the power extracted from the wind i greater than the loe in the ytem and o the ytem can remain connected. Thi propoed that the tandard doubly-fed (DF) connection i ued over the normal DF peed range and the o-called induction generator (IG) mode i ued to extend the low peed operation. Previou work ha illutrated that IG mode enable the DFIG to operate down to 8% lip [8]. Thi change in operation i achieved by diconnecting the tator from the grid in DF mode and then hort circuiting the tator to enable IG operation. All of the generator power flow through the rotor converter in IG mode. The IG curve i identical to the DF curve for ±3% lip. The etimated IG power extracted from the wind at low peed i obtained by extrapolating the curve for the DF mode. The reference torque required by both controller (DF and IG mode) can eaily be derived from thi curve. The torque peed data can then be tored in a look-up table o the reference torque i automatically varied with peed. The capability of modern DF wind turbine to vary the reactive power aborbed or generated [6, 9, ] allow a wind turbine to participate in the reactive power balance of the grid. The reactive power at the grid connection conidered in thi work i decribed, for the UK, by the Connection Condition Section CC [] available from the National Grid. The reactive power requirement for a wind farm i defined by figure 2. Point A - MVAr equivalent for.95 leading power factor at rated MW Point B - MVAr equivalent for.95 lagging power factor at rated MW Point C - MVAr -5 % of rated MW Point D - MVAr 5 % of rated MW Point E - MVAr -2 % of rated MW The objective of thi paper i to invetigate the controller performance of DF and IG mode for a 2MW, 69V, 4-pole DFIG uing machine parameter provided by the manufacturer. Thi i further reearch building on a previou paper which demontrated the teady-tate performance of the two mode of operation, DF and IG mode [8]. In [8] the author dicued the

3 WIND ENGINEERING VOLUME 33, NO. 5, DF IG 2 Generated power, MW Machine peed, rpm Figure : Power peed characteritic..2.8 % Rated power, MW A E C D B Figure 2: Grid connection requirement []. teady-tate efficiency for both connection. The teady-tate performance work illutrated that there were benefit to operating the machine in one connection method a oppoed to the other. Thi paper examine the controllability (i.e. tranient performance) of the 2 MW wind turbine. Reult of the full dynamic controller (current regulation, decoupling equation and vector control) in both DF mode and IG mode are hown. A detailed analyi of the

4 5 DESIGN STUDY OF DOUBLY-FED INDUCTION GENERATORS FOR A 2MW WIND TURBINE DFIG Grid P m P P g P rm Rotor ide converter DC link Grid ide converter P rg Figure 3: Doubly-Fed (DF) connection. component that form the rotor voltage over the full operating range in DFIG mode i preented a thi enable the dominant control component to be identified. Thi i particularly important when deigning advanced control cheme a an overview over the full operating range can be identified. Simulation model, which have been validated againt a 7.5kW laboratory rig [2], are applied to a realitic 2 MW wind turbine to enable concluion to be made regarding the propoed ue of IG mode in a real wind turbine. 2. CONNECTION METHODS Doubly-fed induction machine are commonly connected a hown in figure 3. The grid ide inverter (GSI) i controlled to maintain a fixed dc link voltage with a given power factor at the grid (in our cae unity). The rotor ide inverter (RSI) i controlled o the maximum energy i extracted from the kinetic energy of the wind whilt enabling the tator power factor to be controlled within the limit of the grid requirement though unity power factor i often deirable. An alternative connection method for a doubly-fed machine i hown in figure 4, here called the induction generator (IG) connection. The tator i diconnected from the grid and i hort-circuited. The rotor circuit i unchanged from figure 3. The GSI i controlled a in DF mode. The objective of the RSI i to control the tator flux linkage while extracting the maximum power from the kinetic wind energy. 3. CONTROLLER PERFORMANCE A cloed loop controller for both DF mode and IG mode ha been dicued in prior work [2] but only for a 7.5 kw laboratory tet rig. The dynamic of a 2 MW ytem are omewhat different and are invetigated in thi paper. The performance of the dynamic controller for both DF and IG mode are hown in thi ection for a 2 MW wind turbine. 3.. DFIG Mode (T and Q Control) The reference value for the controller in DF mode are torque (ee figure ) and tator reactive power to enable the grid code requirement [] to be achieved, figure 2. Two peed are invetigated in thi ection to enable the performance of the controller to be hown both above and below the 2% of rated power limit from the grid code requirement. A nominal generated power of 32 kw i achieved at 5 rpm (le than 2% of rated power) and

5 WIND ENGINEERING VOLUME 33, NO. 5, 29 5 Grid DFIG S/C P m P g P rm Rotor ide converter DC link Grid ide converter P rg Figure 4: Induction Generator (IG) connection. Torque, Nm Torque, Nm Stator reactive power, MVAr 5rpm Stator reactive power, MVAr 55rpm Figure 5: DF vector control reference and actual value. a nominal power of.25 MW i achieved at 55 rpm (greater than 2% of the rated power). The reference and actual torque, T e, and tator reactive power, Q, are hown for both peed in figure 5. The value of reference torque, T * e, for both peed i the pecific nominal torque for a given peed calculated from figure ; 2672 Nm for 5 rpm and 77 Nm for 55 rpm. A tep of 2 Nm i applied at both peed to illutrate the dynamic repone to a tep change in torque. The value of reference tator reactive power, Q *, at 5 rpm i varied between the limit pecified by the grid code requirement; initially 5% of the generated power with a tep at t=3.5 to +5% of the generated power. At 55 rpm the tator power factor, pf *, i initially.95 leading with a tep change at t=3 to unity pf and a final tep at t=4 to a.95 lagging pf. The vector control loop are tuned for a time contant of. and.9 for the T e and the Q loop repectively. The vector control i deigned to have a lower bandwidth than the current regulation.

6 52 DESIGN STUDY OF DOUBLY-FED INDUCTION GENERATORS FOR A 2MW WIND TURBINE Rotor current (d), ka Rotor current (q), ka 5rpm Rotor current (d), ka Rotor current (q), ka rpm Figure 6: DF current regulation reference and actual value. Stator flux linkage (d), Wb Stator flux linkage (d), Wb Torque, Nm rpm rpm Torque, Nm Figure 7: IG vector control reference and actual value. The actual rotor current direct, i rd, and quadrature, i rq, component correponding to figure 5 are hown in figure 6. The effect of the tep change in T * e i apparent on the i rq (the upercript indicate that the variable i referred to the tator) a expected. The i * rq component at 55 rpm contain mall tranient repone at t=3 and t=4 that are due to the tep change in the Q value. The tep change in Q *, hown in figure 5, caue a fat change in i * rd, figure 6, a there i initially an error between the reference and actual Q a the control take a hort while to repond. The current regulation i tuned to enure that the bandwidth prevent the controller reponding to uch tranient while till achieving a uitable peed of repone. The equation baed tuning ued to deign the controller give imilar value of proportional and integral gain for the current regulation direct and quadrature loop to thoe ued by Holdworth et al [] IG Mode (T and Flux Control) The reference value for the controller in IG mode are tator flux linkage and torque. Two condition are invetigated for the 2 MW generator in IG mode, tart-up and torque tep repone, at 4 rpm (minimum IG mode peed [2]) and 42 rpm (generated power at thi peed correpond to the upper power rating of rotor converter, 6 kw). The reference and actual torque, T e, and tator flux linkage, λ r (the upercript r indicate that the variable i referred to the rotor), for both peed are hown in figure 7.

7 WIND ENGINEERING VOLUME 33, NO. 5, Rotor current (d), A Rotor current (d), A Rotor current (q), A rpm rpm Rotor current (q), A Figure 8: IG current regulation reference and actual value. The teady-tate T e i the nominal value for the peed of operation, 32 Nm for 4 rpm and 48 Nm for 42 rpm derived from figure. A tart-up equence i required to etablih the rated λ r in the machine, for a given peed, by mean of a ramp, figure 7, before the machine can generate power. Once the controller reference λ r ha been etablihed in the machine, the T * e i increaed by mean of a controlled ramp to the nominal value for a given peed and then a tep repone of 5 Nm tep at 4 rpm and 2 Nm at 42 rpm i applied. The controller regulate the machine to track T * e a expected, ee figure 7. The vector control loop determine the reference rotor current value that are hown in figure 8. The i rd component initially increae rapidly to etablih the λ r and i approximately 3 time the nominal teady-tate value for a given load point. The current i within the rated limit at all time. The initial i rd can be ignificantly reduced if a lower repone of λ r i implemented. The i rq component i regulated by the torque loop to enable the deired power to be generated. Initially there i a light error due to the high i rd which affect the quadrature loop by the cro coupling term. Once nominal λ r i etablihed in the machine the direct and quadrature loop are decoupled. Again a T e tep caue a tranient pike in i * rq though the control i tuned to be lower than thi change in reference value. 4. CONTRIBUTION OF ROTOR VOLTAGE COMPONENTS The performance of both DF and IG mode ha been illutrated in the previou ection. Both controller are baed on an inner current loop and an outer control loop for torque and tator reactive power in the DF cae and torque and tator flux linkage in the IG cae. Decoupling equation were then added to the PI controller output to reduce the effect of cro coupling between the loop. The final part of thi work tudie the contribution of the teady tate component of rotor voltage, given in full in eqn ( and 2), for a 2 MW machine to ae the importance of decoupling equation at variou peed. The rotor voltage, v r, rotor current, i r, and the non-differential component of v r given by eqn ( and 2) are invetigated for the full DF peed range ( to 95 rpm) with the nominal torque determined from figure, and a tator power factor, pf, range of.9 lagging to.9 leading. Only the pf i conidered a the GSI i aumed to maintain unity pf at the rotor converter connection to the grid independent of the RSI. V R i p L L m m rd = r rd + λ d + pσird ωf λq ωfσi rq L L ()

8 54 DESIGN STUDY OF DOUBLY-FED INDUCTION GENERATORS FOR A 2MW WIND TURBINE 4 v rd, V v rq, V Figure 9: Rotor voltage variation with tator power factor. V R i p L L m rq = r rq + λ q + pσirq + ωf λd L L m + ω σi f rd (2) Figure 9 how the variation of v rdq for the peed and tator reactive power range invetigated. The v rd component i dominated in the teady-tate by the ω f σi rq term a the voltage drop acro R r i negligible and the λ q component i zero due to the choice of reference frame. Thi can be confirmed by comparing figure 9 with figure. The v rq in a 2 MW machine i dominated by the ω f (L m /L )λ d term a the low total leakage inductance, σ, reduce the effect of the i rd cro coupling term and the λ orientation frame et the λ q component to zero. The variation in v rq at contant peed (and therefore torque) i due to the cro coupling from the i rd which i regulating the tator reactive power, Q, and therefore pf. The v r magnitude i dominated by the v rq component and i ymmetrical 5rpm; the ynchronou peed for a 4-pole machine. Thi i confirmed by Park et al [3]. The teady-tate variation in the direct, i rd, and quadrature, i rq, rotor current component with repect to peed and Q i hown in figure. The i rd component regulate the tator power factor, pf, by controlling Q and the i rd component regulate T e. The value of i rd determine the proportion of the generator reactive power upplied by the tator and rotor circuit. An increaingly poitive i rd increae the proportion of Q from the rotor circuit while decreaing the Q from the tator until Q i exported by the tator. An increaingly negative i rd increae the Q from the tator circuit, reducing the Q from the rotor ide until Q i exported by the rotor. Q increae with T e to maintain the deired pf and o the i rd component will be higher for contant pf at higher peed. The i rq component i approximately contant at contant peed due to the contant torque and i poitive for generated power due to the orientation frame and the direct and quadrature axi alignment. The i r magnitude i within the rated value for all condition conidered in figure. The remainder of thi ection illutrate the rotor voltage, v rdq, teady-tate component from eqn ( and 2). The R r i rd term in v rd and the R r i rq term in v rq are imply i rdq, figure, caled by R r and o are not hown.

9 WIND ENGINEERING VOLUME 33, NO. 5, i rd, ka i rq, ka Figure : Rotor current variation with tator power factor. 2 ω f i rd σ, V ω f i rq σ, V Figure : Rotor voltage jσω f i rdq term. The jσω f i rdq cro coupling term of v rdq are hown in figure. The σω f i rq term contribute to v rd and σω f i rd form part of v rq. The σω f i rd component varie with both peed and tator reactive power a tator reactive power i proportional to torque for a given tator power factor. The σω f i rd component increae with peed a the load torque increae, figure. The σω f i rq component i the dominant term in the v rd component, eqn (), at non-ynchronou peed; the polarity i a reult of ω f and the magnitude i defined by the torque. The magnitude i i rdq caled by lip frequency, ω f, and the total leakage inductance, σ. Figure 2 how the j(l m /L )ω f λ dq component of v rdq. The (L m /L )ω f λ q term contribute to v rd ; the term i approximately zero due to the orientation frame. The (L m /L )ω f λ d term dominate the v rq component. The hape of the (L m /L )ω f λ d component i clearly influenced by ω f.

10 56 DESIGN STUDY OF DOUBLY-FED INDUCTION GENERATORS FOR A 2MW WIND TURBINE 2 ω f λ q L m /L, V ω f λ d L m /L, V Figure 2: Rotor voltage j(l m /L )ω f λ dq term. 5. DISCUSSION Thi analyi enable the v rd and v rq component to be characteried by the dominant term. The λ orientation frame reult in the λ q feed forward term in v rd being negligible and o the teady tate v rd component i a reult of R r i rd σω f i rq. Three ditinct region can then be identified, ub-ynchronou peed (low i rq due to low load o v rd i approximately R r i rd ), about ynchronou peed (ω f i around o v rd i approximately R r i rd ) and uperynchronou peed (i rd and i rq are comparable due to higher load torque and high tator power factor o v rd i approximately R r i rd σω f i rq ). The tranient repone of v rd for a tep in i * rd i dominated by the pσi rd. The p(l m /L )λ d term ha a negligible effect a the λ d term i contant auming a tiff grid. An i * rd tep affect both the teady tate value of v rq and the teady tate term in v rd. The teady tate v rq component i dominated by the λ d term, confirmed by Hopfenperger et al [9] (with the exception of ynchronou peed when the teady tate v rq i dependent on the R r i rq term). The tranient repone of v rq to an i * rq tep i dominated by the pσi rq term a the differential of the tep change in i rq i initially high. The p(l m /L )λ q term ha a negligible effect a λ q i approximately zero. The v rd term and the teady-tate term in v rq all experience a change in value due to the i * rq tep. 6. CONCLUSIONS Thi paper ha invetigated the controller repone for the DF and IG mode connection for a 2 MW DFIG wind turbine. The machine parameter for the 2 MW machine were provided, for a commercially available WRIM ued in wind turbine, by the manufacturer. The 2 MW machine parameter ued in thi work are not imply a linear caling of prior work on a 7.5 kw machine and o the characteritic are not identical between the two machine. Two area of analyi have been invetigated with repect to the 2 MW DFIG. Exiting imulation model have been ued to evaluate the controllability and teady-tate and tranient behaviour of a 2 MW DFIG in DF and IG mode. The outcome how that IG mode i a controllable mode of operation which will extend the low peed operation a rotor voltage decreae (a peed reduce) and o the voltage limit of the IGBT will be repected a will the

11 WIND ENGINEERING VOLUME 33, NO. 5, current and power limit of the machine and converter. The compoition of the rotor voltage wa invetigated in DF mode for the 2 MW DFIG. Thi howed how the importance of the decoupling equation on the performance of the DFIG varied with peed. ACKNOWLEDGEMENTS The author are grateful to FKI Indutrial Drive and the EPSRC for their upport. REFERENCES. Pena R, Clare J and Aher GM. Doubly Fed Induction Generator uing Back-to-Back PWM Converter and it Application to Variable-Speed Wind-Energy Generation. IEE Proceeding - Electric Power Application May 996; 43; 3; Kelber C and Schumacher W. Control of Doubly-Fed Induction Machine a an Adjutable Speed Motor/Generator, VSSHy 2 - European Conference Variable Speed in Small Hydro. 3. Ran L, Bumby JR and Tavner PJ. Ue of Turbine Inertia for Power Smoothing of Wind Turbine with a DFIG. th International Conference on Harmonic and Quality of Power 24; Müller S, Deicke M and De Doncker RW. Doubly fed induction generator ytem for wind turbine. IEEE Indutry Application Magazine 22; May/June; Hanen AD, Iov F, Blaaberg F and Hanen LH. Review of Contemporary Wind Turbine Concept and their Market Penetration. Wind Engineering 24; 28; 3; Chengwu L and Fengxiang W and Yong T. Deign and Implementation of A Doubly- Fed VSCF Wind Power Control Sytem. International Conference on Power Sytem Technology: PowerCon 22; 4; Hofmann W. Optimal Reactive Power Splitting in Wind Power Plant Controlled by Double-Fed Induction Generator. IEEE AFRICON September 999; 2; Smith S, Todd R, Barne M and Tavner PJ. Improved Energy Converion for Doubly-Fed Wind Generator. IEEE Tranaction on Indutry Application 26; 42; Hopfenperger B, Atkinon DJ and Lakin RA. Stator-Flux-Oriented Control of a Doubly- Fed Induction Machine With and Without Poition Encoder. IEE Proceeding - Electric Power Application July 2; 47; 4; Holdworth L, Wu XG, Ekanayake JB and Jenkin N. Comparion of Fixed Speed and Doubly-Fed Induction Wind Turbine During Power Sytem Diturbance. IEE Proceeding - Generation, Tranmiion and Ditribution May 23; 5; 3; National Grid, Connection Condition September 25; Rev 2; Iue Todd R, High Power Wind Energy Converion Sytem, Eng.D thei, School of Electrical and Electronic Eng., Univ. of Mancheter, UK, Park JW, Lee KW and Lee HJ. Control of Active Power in a Doubly-Fed Induction Generator Taking into Account the Rotor Side Apparent Power. 35th Annual IEEE Power Electronic Specialit Conference 24;

12