Available online at www.sciencedirect.com Electric Power Systems Research 78 (2008) 399–408 A systematic approach to design and operation of a doubly fed induction generator Chad Abbey ∗ , Géza Joós 1 Department of Electrical and Computer Engineering at McGill University, Canada Received 7 November 2006; received in revised form 21 March 2007; accepted 23 March 2007 Available online 7 May 2007 Abstract The doubly fed induction generator (DFIG) is currently one of the most common topologies employed for wind turbine generators (WTGs). The system has the benefit of a back-to-back voltage sourced converter (VSC) of reduced rating, due to its connection to the rotor windings. This paper considers the impact of mechanical and electrical parameters on the kVA requirements of the two VSCs, which together with the dc link capacitor serve as the rotor winding’s power supply. This topology is contrasted with alternatives utilizing a diode rectifier-voltage sourced inverter pair and a set of design curves are generated. In addition to steady-state analysis, an operating strategy for reactive power allocation management is proposed. The theoretical considerations are validated with results obtained from representation of the system in an electromagnetic transient program. © 2007 Elsevier B.V. All rights reserved. Keywords: Power electronics; Reactive power; Wind energy; Doubly fed induction generator 1. Introduction The growth of wind energy has continued during the past decade and renewed focus on alternative sources of energy has contributed to its further development worldwide. In some cases, the amount of wind generation has approached that of conventional technologies and the impact of wind on system stability, operation and power quality can no longer be neglected. New recommendations and utility standards require that wind parks aid in voltage support and that turbines remain connected during system disturbances [1,2]. This implies the need for advanced control schemes requiring coordination of the wind park output power and reactive power for support of the system voltage. Amongst the types of large capacity WTGs, wound rotor induction machines represent an important percentage. Here, the back-to-back voltage sourced converters (VSCs) are connected between the supply and the rotor windings. This facilitates inde∗ Corresponding author at: Department of Electrical and Computer Engineering, 3480 University Street, Rm. 633 McConnell Engineering Bldg., Montreal, Quebec, Canada H3A 2A7. Tel.: +1 514 398 4667; fax: +1 514 398 4470. E-mail addresses: chad.abbey@mail.mcgill.ca (C. Abbey), geza.joos@mcgill.ca (G. Joós). 1 Tel.: +1 514 398 4667; fax: +1 514 398 4470. 0378-7796/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.epsr.2007.03.013 pendent control of the direct and quadrature components of the stator current through the rotor-side, while realizing a reduction in the ratings of the IGBT converters, as only a fraction of the power produced by the machine circulates through this path. Many researchers have demonstrated that this topology can be effectively used to separately control the generator’s speed and the delivered reactive power [3–9]. This paper considers the operating range of a doubly fed induction generator (DFIG), applied as a WTG, with emphasis on the impact of operating strategy and design parameters on the converter ratings, considered in part in Ref. [10]. The wind characteristic, speed range, gearbox ratio and allocation of reactive current between the two converters are considered, first from the steady-state point of view. We discuss the operating procedures for normal and transient modes and the implication of grid code requirements on the converter rating. Validation of the theory is performed using the representation of the system in electromagnetic transient programming software. 2. Wound rotor induction generator The wound rotor induction generator, through its slip rings, allows access to the rotor windings, which can be connected to external resistors or may be fed from a variety of converter 400 C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 its slip. In a truly optimized system, maximum rotor power will in fact occur for two operating points: (1) maximum speed and rated power and (2) lower output power but a larger magnitude of slip, as will be demonstrated. The requirement for delivery of reactive power adds an additional dimension to the problem. The issues of speed range and reactive power allocation are each treated in turn in what follows. 2.1. Reactive power considerations Through proper division of the reactive power between the two converters the overall rating of the two converters can be minimized. In order to facilitate the discussion, the parameter, K, will be defined here as the rotor-side converter compensation constant, representing the proportion of compensation supplied from the rotor-side converter and will be given by: K= Qr Qr + Qgrid (1) where Qr and Qgrid are defined as the equivalent1 reactive power injected into the machine from the rotor-side, and the reactive power supplied by the grid-side converter, respectively, Fig. 2: Fig. 1. Converter topologies for a wound-rotor induction generator and associated speed ranges. options, Fig. 1. Depending on the converter topology implemented this may restrict the generator speed range, due to the direction of flow of power from the rotor terminal. Diode rectifier front-ends (Fig. 1, top) and slip energy recovery drives (Fig. 1, bottom) will be limited to sub- and supersynchronous speeds, respectively. However, the bidirectional converter, using two IGBT voltage sourced converters (VSCs), has the flexibility to operate either above or below synchronous speed, at the cost of a second forced-commutated converter. For the latter arrangement, the reactive power compensation can be accomplished from the grid-side, the rotor-side converter or a combination of the two, Fig. 2. The choice of how to divide the compensation has an effect on the rms current, and consequently the kVA of the two converters. Equally important in determination of the kVA requirements of the converters is the speed range of the machine; the power carried across the bridge is a function of the generator power and Qr = Qr s (2) In the majority of the literature, the rotor-side converter supplies all of the reactive power compensation since the reactive power injected is amplified by a factor of s−1 . In the present work, the possibility of supplying some reactive power from the grid-side converter is considered. In steady-state operation this could have the benefit of reducing the overall kVA requirements of the converters. Transient effects related to realization of the low voltage ride through (LVRT) requirement are also important, and will be discussed in subsequent sections. In order to translate these features into the optimal allocation of reactive power, one needs to consider how the performance of the system varies with the compensation constant. The compensation constant will be chosen in order to minimize the complex power of the rotor-side and grid-side converters, at the peak rotor power points. Transient behavior and analysis of the other operating points will then be considered in the development of an operating strategy that aims to minimize the apparent power of the two converters and improve transient response following disturbances. 2.2. Equivalent circuit model The wound-rotor induction machine steady-state equivalent circuit model, Fig. 3, can be used, together with the steadystate models of the other components of the system to obtain information regarding the kVA ratings of both the grid-side and rotor-side converters. The equivalent circuit model can be used 1 Fig. 2. Reactive power sources in the doubly fed induction generator. Equivalent refers to the fact that the injected rotor converter reactive power, as seen from the machine, is multiplied by the inverse of the slip, as given by Eq. (2). C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 401 Fig. 3. Steady-state equivalent circuit model of wound-rotor induction machine. to derive equations that describe the relationship between the stator and rotor currents. Thus, for a desired stator current (which is dictated by the complex power and the stator voltage) the corresponding rotor-side current can be calculated. The matrix equation below is used to describe this relationship: ⎡ ⎤⎡ ⎤ ⎡ ⎤ 1 Rs + jXs Vm Vs ∠0◦ 0 ⎢ ⎥⎢ ⎥ ⎢ ⎥ 0 (Rr /s) + jXr ⎦ ⎣ Is ⎦ = ⎣ (Vr /s)∠δr ⎦ (3) ⎣1 1 jXm jXm Ir 0 In the case of a wind generator, the real power, which is a function of the mechanical input torque, follows from the wind speed. However, reactive power can be easily controlled and is related to the imaginary current of the magnetizing branch and the contribution from the rotor-side converter: isq = imq − irq (4) Therefore, it can be noted that the stator reactive power can be reduced by supplying a greater proportion of the magnetizing current from the rotor-side. The source of isq must also be considered, noting that it can be supplied from either the grid or from the grid-side converter in the form of igrid,q . The point to extract from the discussion is that the reactive power compensation can be applied from two sources: (i) rotor-side using irq or (ii) grid-side using igrid,q. How this is supplied will affect the converter currents, which are also related to the real power flows, which is considered in the following section. 2.3. WTG relationships When considering the effect of the WTG generator power on the power that must be carried across the dc link one must take into account the actual WTG power characteristic, as well as the influence of the gearbox ratio on the generator speed range, and consequently its slip. The output power of a wind generator is given by: Pm = CP (λ, β) ρA 3 ν 2 wind (5) where Pm is the mechanical output power of turbine; Cp the performance coefficient of the turbine; ρ the air density (kg/m3 ); A the turbine sweep area (m2 ); vwind the wind speed (m/s); λ the tip-speed ratio (rotor blade tip speed divided by the wind speed); and β is the blade pitch angle. A typical WTG output power versus wind speed characteristic is shown in Fig. 4. For WTGs there are three modes of operation: (i) maximum power point tracking (MPPT), where Fig. 4. Generator power vs. wind speed, corresponding turbine speed and the effect of the gearbox ratio on the operating range of the generator. Operation at maximum power point tracking, pitch regulated power beyond vrated . the speed is set to the optimum tip-speed ratio, λopt ; (ii) constant speed region, usually above rated nominal wind speed and (iii) WTG shut down, which occurs for wind speeds below cut-in wind speed, vcut-in , or above the upper wind speed limit (not shown in the figure, typically 20–25 m/s). Fig. 4 also shows how changing the gearbox ratio, nGB , affects the location of synchronous speed on the characteristic. For all three topologies in Fig. 1 it is essential to consider this parameter. With regards to the bidirectional topology, the choice of nGB will affect the amount of power that flows to or from the rotor. The following sections propose a methodology for selection of this parameter. 3. Converter rating minimization Conventionally, the reactive power compensation in a DFIG is supplied from only one of the converters, namely the rotorside converter. This is done for reasons of simplicity and since the reactive power supplied by the rotor converter is effectively amplified by a factor of s−1 . However, this may not be the optimal method since the kVA needs of the rotor-side converter can be greater than the grid-side converter. More specifically, when the speed of the generator deviates greatly from synchronous speed or when the output power approaches or reaches the rating of the machine, there is likely justification for deviating from this operating approach. Under these circumstances it may be desirable to supply some of the reactive power from the grid-side converter, relaxing the demand on the rotor-side device. 402 C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 This section addresses the minimization of the kVA requirements of the converters by considering the division of reactive power and the installed gearbox ratio and their relationship to the overall rotor apparent power requirements. The problem is formally developed and optimality conditions are determined. A design example is used to aid in illustration of the concepts. 3.1. Problem formulation min Srating (K, nGB ) (6) K,nGB Subject to: 0≤K≤1 (7) nGB,min ≤ nGB ≤ nGB,max (8) ωtur,min ≤ ωtur ≤ ωtur,max (9) where the turbine speed is related to the generator speed by: (10) Srating can be expressed by: Srating = || S r S grid ||∞ (11) Sr and Sgrid are the vectors of the apparent power magnitudes for the rotor and grid-side converters, respectively, evaluated at all wind speeds from vcut-in to vrated , and over the entire range of reactive power values. These quantities are evaluated using the equivalent circuit model (3) together with the wind speed characteristic (5). 3.2. Problem reduction As Eq. (11) implies, the search space constitutes all compensation constants and possible gearbox ratios, across the entire operating range of the WTG. Although the problem is still comparatively small in terms of dimensionality, both reduction of the search space and an operating strategy for K result if the converters are assumed to be lossless. Neglecting converter losses implies that the kVAs will be equal for a given operating point if the reactive powers of both converters are also equal. This taken together with Eqs. (1) and (2) gives the following relationship for K: K= Qr /s Qr /s + Qr Radius, R (m) vwind,nom (m/s) vcut-in (m/s) λopt 37.5 12 5 6.3 Table 2 Generator data The greater of the two converter ratings was chosen as the minimizing function, which intuitively suggests that this should result in the same rating for each converter following optimization. The reasoning for this choice is that generally in practice, the kVA ratings of back-to-back converters should be matched in order minimize cost. Matched converters are typically standard and thus, a reduction in cost is possible if they are chosen to be equal. Therefore, the premise is to attempt to limit the installed kVA of the two converters, using the free variables K and nGB . Therefore, the objective becomes: ωgen = nGB ωtur Table 1 Wind turbine data (12) Sbase (MVA) Vbase (V) Poles Xs (p.u) Rs (p.u) Xr (p.u) Rr (p.u) Xm (p.u) 2 690 4 0.117 0.00621 0.1136 0.00627 3.28 which reduces to: K= 1 1 + |s| (13) The constrained optimization problem can then be simplified using this condition by calculating the allocation constant for any given operating point. Furthermore, Eq. (13) forms the optimal operating strategy for selection of K. 3.3. Design example The outlined problem was then solved using data for a typical WTG utilizing the DFIG topology. Tables 1–3 provide the characteristics of the wind turbine, generator data and constraints. The characteristics curve, Cp , for the wind turbine used was that given in Ref. [11]. Here no upper or lower bounds were set for the generator slip, but the methodology could easily incorporate these values in order to respect mechanical limitations or other constraints. The emphasis was to develop an analytical model for optimally choosing the gearbox ratio. This could then be used to quickly determine the effect of different parameters that influence the optimal design parameter as well as the rated value of the two converters. In the following section, a graphical approach is taken; essentially mapping out the solution process performed by the optimization tool. This validates the solution but also provides valuable insight into the variation of the apparent power flow through the converters for changes in operating modes. Table 4 shows the results of the optimization process performed using the numerical method and using the graphical method in the following section. Both provide approximately Table 3 Electrical and mechanical operating ranges Qdfig lead/lag (p.u.) ωtur,min (rpm) ωtur,max (rpm) nGB,min nGB,max 0.5 9 21 94 224 C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 403 Table 4 Optimal gearbox ratio and converter ratings Solution method nGB Srating Analytical Graphical 118 116.5 0.2878 0.28–0.29 the same result, the advantage of the analytical method being the speed and precision once the solution method has been outlined, yet lacking the intuitive insight that the graphical procedure offers. 4. Parametric analysis Using the induction machine equivalent circuit a number of curves were obtained in order to gain an understanding of the dependence of the apparent power demands of the two converters on the gearbox ratio, as well as on the compensation constant. While the optimal values were determined in the previous section, it is informative to consider the variation of these parameters, particularly for the alternate converter topologies presented in Fig. 1. 4.1. Gearbox selection Here the effect of the gearbox ratio was investigated by calculating the infinite norm of the two converter ratings for different values of nGB . The full reactive power range was considered and easily enough one can arrive at the somewhat obvious conclusion that the limiting case is for full leading reactive power (as the internal requirements must be provided in addition to that supplied to the grid). Reactive power was allocated using the operational strategy defined by Eq. (13). Fig. 5 shows the results of the calculation, performed for different pole numbers. We note that as the number of poles is increased, the optimum gearbox ratio decreases, as expected. For the 4 pole machine the optimum gearbox ratio was found Fig. 5. Variation of required kVA rating vs. gearbox ratio, given for different poles, operation at Qdfig = 0.5 p.u., Kopt . Fig. 6. Calculated converter kVA vs. wind speed for (a) optimal nGB (116.5), Kopt and for two alternate WRIM topologies; and (b) for optimal nGB and different values of K. to be about 116.5, which balances the rotor apparent power carried at rated power with the rotor power needs at low speed operation—near zero real power production but high rotor power due to large values of slip. Next, the kVA requirements were calculated for the nGB,opt , over the wind speed range to demonstrate how the kVA changes for different operating points, Fig. 6. As can be noted, the kVA requirements starts from a first peak at 5 m/s (low power but high slip), decreases to zero at synchronous speed and then reaches the maximum value again at rated wind speed. This confirms the result from Fig. 5 and the previous section, showing that the gearbox value is optimum for the given constraints. The kVA requirements for the alternate converter topologies are also given, assuming MPPT operation, along with the effect of using a constant value for the compensation constant (1 and 0.5). Compared with the optimal case the maximum value of each of the alternatives is consistently higher. However, it is interesting to note that the variable speed drive topology requires a rating which is only twice that of the full converter topology with optimal operation. The higher apparent power and increased harmonic injection would be somewhat offset by the reduced converter cost, whereas the slip recovery drives does not appear to be competitive. Nonetheless, the operating strategy of 404 C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 Fig. 7. Calculated kVA for rotor-side and grid-side converters as a function of K. Calculated for operation at wind speed of 5 m/s and for different values of Qdfig . Optimal operating K is given by the dotted vertical line. these topologies will certainly differ and lack other beneficial characteristics that the full converter is able to provide. Although the main focus has been the impact of these design parameters on the converter rating, for completion the relationship between the gearbox ratio and the size of the machine is now considered. We can consider the relation for the mechanical torque of the machine: τm ∝ |φr ||φs | sin δ (14) which is related to the mechanical power by the speed of the machine: Pm = τm ωr (15) Therefore, as the gearbox ratio is increased, the electrical speed of the generator will also increase, and consequently the size of the machine required to meet the same power rating decreases. So, a tradeoff needs to be established between the machine design and that of the converter; however this discussion is deferred to another work. 4.2. Compensation constant Using the gearbox ratio of 116.5, the converter kVA requirements were obtained for different values of WTG reactive power, at different values of K. The results were obtained for the two peaks given in Fig. 6, corresponding to wind speeds of 5 and 12 m/s, Figs. 7 and 8, respectively. This provides a detailed understanding of the influence of the different parameters and validates the expression in Eq. (13). Here the kVA needs of the grid-side and rotor-side converters are given for operation at 0.5 and 0.25 leading VArs and unity power factor operation. The omission of lagging VArs is justified by the fact that smaller reactive power is required in these cases, due to the internal requirements of the generator. As can be noted the optimal K, as given by the intersection of the curves for the rotor and grid-side converters agrees with that predicted by Eq. (13), which is shown by the vertical line. Fig. 8. Calculated kVA for rotor-side and grid-side converters as a function of K. Calculated for operation at wind speed of 12 m/s and for different values of Qdfig . Optimal operating K is given by the dotted vertical line. It can be noted that the majority of the compensation is supplied from the rotor-side converter (70–85%) however, a small percentage should come from the grid-side converter in order to minimize the maximum rating of the two VSC’s, which is given by the point of intersection. The optimum point is a function of the operating slip only and corresponds with the value predicted by Eq. (13). Note that the actual reactive power supplied by each converter is equal; however, the reflected rotor reactive power represents the majority of the contribution. In summary, either a rigorous optimization method or a set of design curves can be used to determine the optimal parameters for minimization of the converter ratings. While the latter provides greater insight into the sensitivity with respect to the different parameters involved, the former will be favored where the intent is design focused. It is interesting to note that Kopt is independent of the machine and turbine characteristic, whereas the gearbox ratio is sensitive to the wind turbine characteristic curve. 4.3. Machine turns ratio Some mention should be made to the actual rotor current magnitude. In wound-rotor induction machines, the turns ratio between the stator side and rotor-side is typically not equal to 1. Here, by plotting the current in the rotor, considering the power curve and generator speeds given in Fig. 4, for decreasing values of turns ratio, one can determine the ratio for which the current in the grid-side and rotor-side converters becomes equal. Assuming again our objective is to equate the two ratings – both in terms of power and current – the optimum turns ratio falls close to 0.3, Fig. 9. Note here that the power transformation is not affected by this parameter, like for a transformer, and therefore, previous developments still hold. 4.4. Speed and power factor range The dependence of the converter rating on parameters related to the design of the WTG is also of interest, namely the upper C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 405 ing. This results in a nearly linear increase in converter rating, doubling for a change in pF from 0.99 to 0.85. 5. Operating considerations The converter ratings are defined for normal operation but certain concessions may need to be made for transient conditions, and as well, the compensation may need to be adjusted for different operating conditions. These issues are treated here in brief. 5.1. Steady-state Fig. 9. Magnitude of rotor and grid-side (in bold) converter currents over speed range for different machine turns ratios. Operation at rated leading reactive power (Qdfig = 0.5 p.u.). Table 5 Dependence of converter rating on maximum machine speed ωe,max kVA (p.u.) nGB 1.05 1.15 1.25 0.328 0.310 0.292 98 107 116 speed limit and the reactive power capability of the generator. Utilizing the developed methodology, the influence of these two parameters was determined, Table 5 and Fig. 10. With regards to the upper speed limit, we can note that as long as the generator’s electrical speed is allowed to vary up to 25% greater than synchronous speed, the optimal gearbox ratio and minimum converter rating can be achieved. As we further restrict the speed range, we ultimately reduce to the specific case of the variable speed drive topology, only that the flexibility to delivery reactive power from the grid-side converter is retained. The reactive power range of the WTG will be dictated by the grid code requirement in most cases. Fig. 10 shows the impact of imposing greater reactive power capability (expressed as the leading power factor at rated real power) on the converter rat- Fig. 10. Effect of power factor range (leading/lagging power factor capability at rated power, as dictated by interconnection standards) on converter kVA rating. As was alluded to in previous sections the same equation used to determine the compensation constant at the limiting operating points can form the basis for an operating strategy, Fig. 11. In this way, the kVA requirements of both converters can be equalized during normal operating. It should be noted that this is complicated by certain practical logistics. Most notably, it requires calculation of the rotor reactive power, and the internal magnetizing current will change slightly for different operating points. In the calculations performed this issue was addressed in two ways. One can assume a magnetizing current by taking the internal voltage to be 1 per unit. Stator and converter reactive powers then follow from this assumption and all other quantities are calculated. The compensation constant is then recalculated at the end. Alternatively, an iterative method can be used to determine the stator and supply converter contributions for a given K and wind speed. Either method works fine for generation of the curves; however, on-line operation requires further consideration. An open-loop approach (resembling the former case) can be used, relinquishing the benefit of exactly controlling the compensation constant to its optimum value. Opting for a look-up table is likely the simplest and most practical approach and should yield compensation constants that are close to the optimum value if a significant number of points are charted offline. Fig. 11. Operational strategy for selection of K as a function of slip. 406 C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 Fig. 12. Active crowbar protection for low voltage ride through in the DFIG. 5.2. Transient control Low voltage events on the grid, most often caused by faults on the transmission network or adjacent feeders, require special consideration for the control of the WTG but also in the hardware components required. Most utilities now demand that a WTG meet a low voltage ride through (LVRT) requirement, meaning that they must remain connected for all normally cleared contingencies [1,2]. While the LVRT characteristic itself is somewhat subjective, the implication is that the WTG must be able to operate during low voltage events; in extreme cases operation for zero voltage events is required. The most commonly cited LVRT solution for the DFIG is the active crowbar, whereby the rotor windings are short-circuited [12]. The consequence is that the rotor converter is not able to provide reactive current during and immediately following the fault (Fig. 12). As a result, the proposal here is that, regardless of the prefault compensation constant, it should be immediately set to zero upon activation of the crowbar, as is discussed in Ref. [13], shifting compensation entirely to the grid-side converter. Following recovery of the voltage, the normal operating conditions can then once again dictate the necessary compensation constant. This is easily implemented in the control algorithm using a logic signal that automatically sets the compensation constant to 0 should the rotor winding be short-circuited. Fig. 13. Variation of the kVA ratings of the rotor and grid-side converters as a function of K for operation at 5 m/s, slip = 0.45, Qdfig = 0.5 (validation of Fig. 7). optimum compensation constants and the minimum kVA match well with those values determined in the previous sections. 6.2. Allocation implementation The most straightforward method for implementation of the allocation strategy is offline calculation of the internal magnetizing requirements of the machine, which could then be incorporated into the control using a look-up table. This would avoid the use of major modifications to the control structure, circumventing the need for additional compensation blocks. Alternatively, the two reactive current loops could potentially be combined; however, it is not clear how equal contribution from each converter could be guaranteed, or how the control loop might respond during transients. Potential merits of this approach might include a simpler control structure; however further investigation is required. 6. EMTP validation In order to support the theory developed in the previous section and to investigate the behavior of the system in response to various disturbances, the WTG was represented in a commercial electromagnetic transient programming package and various simulations were performed. The objective was to reproduce the theoretical predictions and therefore, demonstrate the validity of the installed rating. This serves not only as mutual validation for the different representations of the system but also to address how the allocation strategy could be executed in the control scheme. 6.1. Steady-state performance The kVA requirements were obtained for the same operating conditions as for the curves generated using the equivalent circuit model. Again operation for wind speeds of 5 and 12 m/s was considered, Figs. 13 and 14. The compensation constant was slowly varied from 1 to 0 and the results were plotted. The Fig. 14. Variation of the kVA ratings of the rotor and grid-side converters as a function of K for operation at 12 m/s, slip = −0.25, Qdfig = 0.5 p.u. (validation of Fig. 8). C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 7. Conclusions This paper has discussed the parameters affecting the converter kVA requirements and reactive power management in a doubly fed induction machine. Steady-state analysis of the system for determining the impact of gearbox ratio and choice of operating point on the reactive power flow on the converter rating was performed. A methodology was proposed and demonstrated for minimization of the converter ratings and allocation of reactive power, considering these parameters. Based on this analysis, operating strategies were proposed for overall minimization of kVA flow on the rotor-side. Implementing optimal allocation of reactive power results in a 15% reduction in converter rating compared with full delivery from the rotor-side, whereas an almost 50% reduction is realized when compared with the less expensive diode-front-end converter topology. Appendix A. List of symbols A Cp imq irq isq Igrid,q Ir Is K Kopt nGB nGB,max nGB,min nGB,opt Pm Qdfig Qgrid Qr Qr Rr Rs s Sgrid Sr Srating vcut,in vrated vwind Vm Vr Vs Xm Xr Xs turbine sweep area performance coefficient of the turbine quadrature component of the magnetizing current quadrature component of the rotor current quadrature component of the stator current quadrature component of the grid side converter’s current rotor current stator current compensation constant optimum compensation constant induction machine turns ratio maximum induction machine turns ratio minimum induction machine turns ratio optimum induction machine turns ratio mechanical output power of turbine reactive power delivered from the DFIG generator system reactive power delivered from grid-side converter reactive power delivered from rotor-side converter reactive power delivered from rotor-side expressed on stator side rotor resistance expressed on stator side stator resistance generator slip grid-side converter kVA rating rotor-side converter kVA rating power electronic converter kVA rating wind turbine generator cut-in wind speed wind turbine generator rated wind speed wind speed voltage across magnetizing branch rotor voltage stator voltage magnitizing reactance rotor reactance expressed on stator side stator reactance 407 Greek letters β blade pitch angle δ angle between the stator and rotor flux vectors angle of rotor-side voltage expressed relative to stator δr voltage λ tip-speed ratio of the rotor blade tip speed to the wind speed ρ air density ωgen generator angular speed ωtur wind turbine angular speed ωtur,max maximum wind turbine angular speed ωtur,min minimum wind turbine angular speed References [1] S.M. Bolik, Grid requirements challenges for wind turbines, in: Proceedings of the Fourth International Workshop on Large Scale Integration of Wind Power and Transmission Networks for Offshore Wind Farms, 2003. [2] G. Joos, Grid code review, Integration of Renewable Energy Sources and Distributed Energy Resources Conference, Brussels, December 1–3, 2004. [3] R. Pena, J.C. Clare, G.M. Asher, Doubly fed induction generator using back-to-back PWM converters and its application to variable-speed windenergy generation, J. Proc.-Electr. Power Appl. 43 (3) (1996). [4] R. Datta, V.T. Ranganathan, Direct power control of grid-connected wound rotor induction machine without rotor position sensors, J. IEEE Trans. Power Electr. 16 (3) (2001). [5] J.B. Ekanayake, L. Holdsworth, X. Wu, N. Jenkins, Dynamic modeling of doubly fed induction wind turbines, J. IEEE Trans. Power Syst. 8 (2) (2003). [6] H. Akagi, H. Sato, Control and performance of a doubly-fed induction machine intended for a flywheel energy storage system, J. IEEE Trans. Power Electr. 17 (1.) (2002). [7] R. Datta, V.T. Ranganathan, A simple position-sensorless algorithm for rotor-side field-oriented control of wound-rotor induction machine, J. IEEE Trans. Ind. Electr. 48 (4) (2001). [8] R. Datta, V.T. Ranganathan, A method of tracking the peak power points for a variable speed wind energy conversion system, J. IEEE Trans. Power Electr. 16 (3) (2001). [9] L. Morel, H. Godfroid, A. Mirzaian, J.M. Kauffmann, Double-fed induction machine: converter optimization and field-oriented control without position sensor, J. IEE Proc.-Electr. Power Appl. 145 (4) (1998). [10] B. Rabelo, W. Hofmann, Optimal active and reactive power control with the doubly-fed induction generator in the MW-class wind-turbines, Proc. IEEE Conf. Power Electr. Drive Syst. 1 (2001) 53–58. [11] L. Ran, J.R. Bumby, P.J. Tavner, Use of turbine inertia for power smoothing of wind turbines with a DFIG, in: Proceedings of the 11th International Conference on Harmonics and Quality of Power, 2004, pp. 106–111. [12] J. Niiranen, Voltage ride through of a doubly-fed generator equipped with an active crowbar, in: Proceedings of the Nordic Wind Power Conference, Chalmers University of Technology, 2004. [13] V. Akhmatov, Variable-speed wind turbines with doubly-fed induction generators. Part IV. Uninterrupted operation features at grid faults with converter control coordination, Wind Eng. 27 (6) (2003) 519–529. Chad Abbey received his degree in electrical engineering from the University of Alberta in 2002. In 2004, he graduated with an M.Eng. degree from McGill University, in Montréal where he is currently pursuing his Ph.D. He is presently working with CANMET Energy Technology Centre, in Varennes, Québec where he is a Research Engineer and coordinates a research program on the modeling and integration of distributed generation. His current research interests include wind energy, distributed generation and their integration to the grid. Géza Joós graduated from McGill University, Montreal, Canada, with an M.Eng. and Ph.D. He is a Professor with McGill University, Montreal, Canada, since 408 C. Abbey, G. Joós / Electric Power Systems Research 78 (2008) 399–408 2001. He is involved in fundamental and applied research related to the application of high-power electronics to power conversion, including distributed generation, and power systems. He has published extensively and presented numerous papers and tutorials on these topics. His employment experience also includes ABB, the University of Quebec and Concordia University. He has been involved in consulting activities in Power Electronics and Power Systems, and with CEA Technologies as Technology Coordinator of the Power Systems Planning and Operations Interest Group. He is active in a number of IEEE Industry Applications Society committees and in IEEE Power Engineering Society and CIGRE activities and working groups dealing with Power Electronics and applications to Distributed Resources. He is a Fellow of the Canadian Academy of Engineering.