You are on page 1of 9



A Dynamic Wind Generation Model for Power Systems Studies

Ana I. Estanqueiro, Member, IEEE
AbstractIn this paper, a wind park dynamic model is presented together with a base methodology for its application to power system studies. This detailed wind generation model addresses the wind turbine components and phenomena more relevant to characterize the power quality of a grid connected wind park, as well as the wind park response to the grid fast perturbations, e.g., low voltage ride through fault. The developed model was applied to the operating conditions of the selected sets of wind turbine experimental benchmark data from Azores and Alsvik wind parks, both for steady and transient operation of the grid. The results show a fairly good agreement in the relevant range of frequencies and indicate the model may be used as a tool for power system studies. Index TermsDynamic models, power quality, power system studies, wind energy, wind turbines.

I. INTRODUCTION NE may state that the current greater challenge for the wind energy research area was caused by this sectors own extreme success. The high capacity installed in the latest years, mostly in European countries, introduced a new set of technological issues among grid planners [1][4] and transmission system operators (TSOs). These recent concerns are a TSO challenge: will the systems be capable to cope with the wind power generation in large quantities (aka high penetration) without requiring new models and system operation tools, increased performance of the wind turbines, or even a change in the transmission system conventional mode of planning and operation strategy? The concern is quite legitimate as it is the TSO responsibility to manage the system within safety boundaries and respond to ofcial regulatory bodies for the occurrence of serious events or even blackouts. The main TSO planning problem is the sudden disconnection from the grid of most of the installed wind capacity as a response to a fast grid perturbation originating a voltage dip. Most actual large wind parks are already capable of remaining connected to the grid, under such events, i.e., they have ride through fault (RTF) ability. This functionality allowed wind
Manuscript received September 6, 2006; revised February 19, 2007. The experimental campaigns carried out in Azores were sponsored by NATO SfS program under the project PO-Mistral. European Union R&D programs partly nanced several projects that contributed over the years to the nal results here presented, namely, projects POWERCONTROL (JOR-3-CT95-0067) and WIRING (JOR3-CT98-0245). The Portuguese Ministry of Economy, through INETI, nanced the participation in IEA Annex 21 that enabled to complete the modeling and validation process. Paper no. TPWRS-00608-2006. The author is with INETI-National Institute for Engineering, Technology, and Innovation, Lisbon 1649-038, Portugal (e-mail: Digital Object Identier 10.1109/TPWRS.2007.901654

parks to be seen as conventional power plants that behave (almost) as any other unit in the system. That is a sign of this technologys maturity, but it also brought a few obligations to this renewable sector related to the technology adult age, namely wind park models have to exist and be validated [5]; for high wind penetration, models of the wind generation need to be interfaced with the power system transient analysis tools; to identify the need for adaptation, when proved to be essential, of the existing wind capacity to have RTF capability for pre-identied perturbations; the disclosure of wind turbine constructive data by the manufacturers to enable the correct operation of the TSOs wind generation models. The concerns of the power system planners constitute clear signs of this sectors strong need for reliable models that may contribute to the characterization of these recent, and increasingly large, generating units. The dynamic models need to describe the time-dependent dynamic normal operation of the wind generation and its response to voltage dips. To achieve that, they also need to include the wind turbine main dynamics, as some authors recently started to refer [6]. In this paper, a dynamic model is presented and validated against experimental data obtained at two different wind parks/turbine manufacturers, for steady and transient operation of the power system and, therefore, may be integrated in common power system numerical platforms to be used as a planning tool. II. STUDYING POWER SYSTEMS WITH EMBEDDED WIND GENERATION The development of the International Electrotechnical Commission IEC 61400-21 standard during the later 1990s and its publication in 2001 [7], as well as the outcomes of several research projects [8][10], enabled to identify the characteristics of the wind turbines with higher inuence on the electrical power deliveredas well as the parameters more adapted to their quanticationto act as normalized quality indicators. The more relevant characteristics identied are presented in Table I. That important step enabled to assess and characterize the power quality of a grid connected wind turbine based on a experimental test and extrapolate the data to a wind park. It also allows, in the feasibility phase of a wind park, to use the knowledge about the wind turbine power quality and thus optimize the parks design characteristics to avoid the degradation of the existing network quality of service.

0885-8950/$25.00 2007 IEEE




the ride through fault capability. Since this standard was developed for single wind turbines only, the wind park design and its topology as well as the common characteristics of the existing transformers and other type of electrical equipment are not included and have to be addressed using adequate wind park dynamic models. This possibility is anticipated in the revision of the standard, currently underway. B. Local Constraints Assessment: Grid and Wind Flow The grid parameters to assess the system power quality with embedded wind power generation are similar to the ones used for conventional power system analysis, even for steady-state load ow simulation, thus available from every electric utility company. If assessed, the wind uctuations may enable to compute the local wind ow power spectral density. By applying one of the several methods available, e.g., the one based on the Shinozuka method included in the wind model of INPark [11] or any other similar approach, it is possible to generate a broad set of synthetic wind time series characteristics of the wind park future site that can be used as an input of the models. If the experimental characterization of the wind power spectral density is not achievable, another possibility is to use one of the well-known spectral functions from the literature (e.g., Kaimal [12] or Davenport [13]) as an input. Nevertheless, the non-experimented user should be aware of the large errors that may be introduced by this process and should use it with care only as a rough estimate of the local turbulence. Moreover, it should be noted that the described procedure to obtain the wind uctuations is only relevant for local power quality studies. For very fast transient assessment, as in the characterization of the RTF response of a wind turbine or park, the time dependency of the wind input is usually not relevant and may be neglected. C. Wind Park Design and Control The wind turbine micrositing (to avoid wake ow operation of some wind turbines) and the optimum use of available terrain determine the layout of the park power collecting systems (internal grid). It is important to highlight the fact that, with the increasing capacity of the individual wind turbines, actual and future wind parks have a quite lower number of machines then they used to. As a consequence, the smoothing effect of the uncorrelated uctuations of the active power [14] is expected to be reduced. However, modern wind technologies have much smoother power outputs than last decades typical ones, so this issue is not expected to have its overall relevance increased. The voltage and power conditioning control, both individually by wind turbine and globally by the wind generation unit, turn these spatially dispersed wind units into what is nowadays already seen as a new generation of power units. Even if that is not broadly realized within the sector, wind power is driving the development of new grid access codes for most countries, both for transmission and distribution networks. In relation to wind energy development, innovative TSOs strategies, as forecast techniques, renewable energy systems storage, and different orders of merit in the dispatch centers exist nowadays. New PS operation principles also appear: wind parks are clustering together with their monitoring and central control, even at the

A broader and alternative approach, proposed in this paper, consists in the application of wind generation dynamic models as planning tools in the wind generation grid integration study. This study may be done when the application of the wind park for grid connection is analyzed or during the design phase of a wind park (e.g., if the connection of a large wind park to a weak area of the grid is intended). These models may also be applied in a more relevant way, i.e., to be integrated in the TSO models and become part of the power system overall behavior assessment. The model presented in this paper, and the methodology developed to its application, relies on the completion of steps A to C from Table I to assess the main wind park dynamic characteristics. Therefore, the connection of large wind parks to the grid and the risk assessment of the power system operation with large quantities of embedded wind generation may be characterized during the last step (D), i.e., by performing the numerical simulation of the wind park with dynamic models under steady and transient operation of the grid. A. Wind Turbine Type and Parameters Identication The base technology of the wind turbines used in a wind park determines the type of power quality assessment to be performed. If variable-speed wind turbines are to be installed in a wind park, these will be connected to the grid through inverters. Therefore, it is more relevant to estimate the harmonic distortion in the local grid than the dynamic voltage uctuations, since these are possibly negligible. The wind turbine parameters needed to characterize the power quality of a wind turbine were published in IEC 61400-21 in 2001. Some of the parameters are under revision in order to include, among others,



distribution grid level, and the use of distribution system operators (DSOs) is referred to as of great relevance. These later concepts still correspond to incipient technologies, most in research and/or demonstration phases, and their inclusion in the actual wind park dynamic models requires a broader knowledge of the techniques to be implemented. D. Numerical Simulation of the Wind Parks Having fullled the three previous steps of the method, it is already possible to simulate and assess the behavior of a wind park, if adequate numerical models are available. In the recent past, there was a general tendency to use very simple mechanical and aerodynamic models to describe wind generation, e.g., [15]. This tendency seems to lie on the traditional separation between electrical and mechanical research areas. The separation of the wind turbines aerodynamic behavior from their power conversion electrical side has, as main consequence, lost the dependency of the mechanical part from the grid status and occurrences. Although that may be possible for certain specic circumstances, most of the simple aerodynamic models fail to describe the relevant power oscillations of the wind turbine output, even under normal steady operation of these equipments when compared to experimental data. Recently and possibly driven by the necessity to characterize with detail the dynamic response of the modern extremely large wind turbines, slightly more detailed aerodynamic and structural sub-models started to be used (e.g., [16]). Since the development of dynamic wind generating models is being strongly encouraged among the scientic community [5], several dynamic models are currently being developed and are expected to be available in the near future to implement the suggested methodology. Nevertheless, these models will have to be validated against experimental data to guarantee they produce replicable results for different sites and wind turbine types. The wind turbine models under development will be linked to the power system numerical platforms, resulting in a whole power system model. The situation may be simplied in the near future if wind parks models are provided to the TSOs by the manufacturers in a generalized format, all addressing and containing the same basic components of the wind energy conversion system in a standardized modular approach, as the ones existing for many years for the conventional power stations. The numerical simulation of a wind park composed of generalized wind turbines also offers the possibility to test different wind turbine technologies, internal and external grid layouts, and different grid conditions (or types of transient perturbations) that may affect the system power quality as well as the wind park transient stability. The recent need for the wind generation units to comply with the power system requirements in what concerns the response to very fast transients should also be a factor of acceptance (or not) of a wind park project. Finally, the type of models here presented can be applied (embedded in the overall power system models) in the feasibility phase of large wind park projects to assess the overall grid integration impact. When the results of the model show a poor performance of the wind power plant and a possible negative impact on the power system operation and/or the detection of

Fig. 1. Angular deection (ap) in the axial direction.

nonconformity to existing codes, that should be a motive for non-acceptance or re-scaling of a wind park project, or even the substitution of the wind turbine technology. III. INPARK DYNAMIC MODEL The INPark model is a dynamic detailed wind generation model that aims to characterize the electric power delivered by a wind turbine as well as the voltage uctuations induced at its terminals by the variable and time-dependent energy resource. This model started to be developed in the mid-1990s, mainly focused on the power quality assessment of wind turbines, and its characteristics are described in detail in [8][10]. Recently within the activities of International Energy Agency (IEA) Annex 21 research tasks, the model was extended in order to address a wider range of phenomena, and sub-models have been added to describe the wind generation behavior under transient operation of the power system. A. Rotor Sub-Model The INPark aerodynamic behavior of the rotor is based on the known Glauerts momentum/strip theory [17], and its characteristics are described with detail in [11]. For the current purpose, it included only the rst mode of the blade-hub joint movement (in both directions) in the torque, rotational speed, and mechanical power equations of the shaft. Within this assumption, the blades are modeled as simple rigid bodies, having the entire deformation place in the connection of the blade root to the hub. The connection in the axial direction is modeled through an articulated joint by a spring according to Fig. 1, initially used to describe the Dutch exible wind turbines [18]. The equilibrium equation for the momenta in the axial direction, neglecting the tower displacement, is (1)

The axial momentum due to the aerodynamic forces is obtained from the balance of the lift and drag forces in this direction using the widely used blade element/strip theory. Within this model, the aerodynamic forces are calculated under a quasi-



electrical quantities, namely, the active and reactive power (or (point of the correspondent currents) and the voltages at the common coupling) and other busbars in the vicinity of the wind turbine or wind park under study. B. BladeTower Interference Sub-Model Usually the models that try to describe the tower shadow effect fail to support themselves in a theoretical or even empirical basis, most being just a mere reproduction of experimentally detected oscillations. Therefore, an innovative approach based on the experimental results for circular section cylinders from Zdravkovich [17] was developed and implemented in the INPark model. Zdravkovich investigated the variation in the drag and lift coefcients of a cylinder due to the existence of others in its vicinity and experimentally determined the effects of their interference. There is a similar relative position of the cylinder bodiesstudied by Zdraykovichand the blades and the supporting tower of a wind turbine, if these are assumed as cylindrical. The application of this method consists on the construction of an interference coefcient matrix function of the blade radius and the relative angle between this one and the tower. It should be referred that, in the aerodynamic forces computation, the lift and drag coefcients are weighted by the ratio between the drag coefcient of the interfering wind prole that constitutes that sector of the blade (as a function of the attack angle) and the coefcients of an isolated cylinder (5a,b) Therefore, the matrix of interference coefcients for the wind rotor is constructed according to the following expressions: (6a) (6b) and being, respectively, the interference cowith efcient of lift and drag for cylindrical bodies, and the drag coefcient of an isolated cylinder. In (6), the subscript `` '' refers to the bidimensional (2-D) usual aerodynamic prole characteristics. This interference effect is therefore implemented blade by blade and analyzed independently; thus, the computation of the actuating forces and blades momenta also take this effect into account. The equations of the momentum/strip-theory classical method for the thrust and the torque, actuating on a blade element, are therefore replaced by (7) (8) (5) This dynamic approach of the wind rotor involves several structural and aerodynamic simplications justied by the objectives of the model, i.e., the characterization of the relevant C. Transmission System Sub-Model The mechanical transmission is modeled according to the scheme of Fig. 3 assuming that all the torsional movement oc-

Fig. 2. Lead-lag movement in the tangential direction.

stationary regime approach. After their computation, one obtains, for the blade element

(2) where and represent the projection of the lift and drag coefcients of the blade element along this axis, respectively, with and without tower interference. For the tangential direction, the forces and momentum in the articulated connection joint may be represented by a string of stiffness and a damping system , as illustrated by Fig. 2. Thus, the equation governing the lead-lag movement of one blade relatively to its standstill position may be written as (3) The moment associated with the aerodynamic forces along the tangential direction due to the wind ow is obtained, as before for the axial direction, with the aerodynamic lift and drag forces. For a blade element , one may write (4) In a wind turbine, the effective velocity of the ow in the rotor (or in a blade element) depends on the radius, being thus variable along it. Once the blades are modeled as rigid bodiesassuming that all the angular deection occurs at the hub connectionthe calculation of the axial deection angle and the is based on the determination (for each time lead-lag angle step of the quasi-stationary approach) of an equivalent value for the ow effective velocity, representative of the whole blade behavior. An identical process is followed to determine the lift and drag coefcients and the axial and tangential interference factors ( and ). Therefore, the mechanical torque produced , is obtained as the sum of each individual by the rotor, blade contributions (subject, or not, to the tower interference, depending on its azimuth angle, ) as follows:



bewhere the total rotor dispersion referred to the stator and mutual induccomes subdivided into self-induction tion coefcients (14)
Fig. 3. Power transmission sub-model.

E. Grid Sub-Models The local grid is modeled using the common power systems approach: pi-equivalents of internal and external lines, ideal step-up transformers, and the Thevenin equivalent of grid at the , normally the interconnection substation to the wind parks existing network. IV. MODEL VALIDATION PROCESS The validation process of wind turbine models is not an easy task to accomplish, not only due to the complexity of some models but also associated with the stochastic nature of the wind. Therefore, in order to easy and clarify the validation process, the limits of validity of the models should be predened, as well as the ltering of the experimental samples selected to be described by the numerical models. As an example, experimental data where climatic unstable conditions or very high turbulence were identied should be rejected for validation purposes. A. Grid Steady-State Operation 1) IEA Azores Case Study: The wind park dynamic model INPark, described in general terms in Section III, was applied to the IEA Annex 21 Azores case study. The study addresses a situation where a 0.55-MW small wind park operates in the Azorean Island of S. Jorge, being constituted by ve stall Nordtank wind turbines . These turbines are equipped only with a startup control system and maximum wind and power protections. There are no RTF or any other devices that equip modern wind turbines. The wind park is connected through a step-up transformer to a medium voltage (MV) 15-kV line that is part of the island weak and isolated power system. Detailed data about the S. Jorge power system and INPark input data for this case study may be found in [10]. A synthesis of the Azores parameters is presented in Appendix B. Within this experimental campaign, the wind velocity 3-D components were measured using a sonic anemometer installed at the rotor hub height at 1.5-D (rotor diameters) for the prevailing wind direction. This level of wind measurement detail and the sampling frequency of 40 Hz assured a good representation of the time-dependent resource for dynamic analysis. Samples with turbulence above 12% were rejected. Fig. 4 shows the comparison between the experimental and INPark simulated dataset (sample BB17) for the active power, in this case applied to the wind turbine #2 of the Azores S. Jorge wind park. Fig. 5 shows the comparison between experimental and INPark simulated values for the voltage at the wind turbine #2 terminals for the same dataset sample (BB17). In Fig. 6, the power spectra of the simulated and experimental values for a single wind turbine are presented (WT#2 as an

curs in the primary shaft and being mathematically represented by (9a) (9b) (10) is the angle of the primary shaft; the angular posiwhere the tion of gearbox with respect to a reference; and angle of the high-speed shaft (positive in the wind turbine rotarepresent, respectively, the torques tional speed sense). the electroof the wind rotor and the shaft transmission and magnetic torque of the generator. The transmission torsion angle . Rewriting the equations, one may be represented by obtains (11a) (11b)

in (9)(11a) is obtained using The rotor drive torque as input the geometric and aerodynamic characteristics of the blades and the rotor and by solving, in each quasi-stationary time-step, the system constituted by (1)(5), (7), and (8). D. Electric Generator Sub-Model for this particular standard The electromagnetic torque induction generator is obtained using the saturation model rst developed by Ferreira de Jesus [20] and implemented by Castro [21] for small hydro power plants. The direct and quadrature of the stator voltage may be written as components


(12b). Some well-known algebraic manipulations enable to write the coordinates) in the form rotor winding equations ( (13a) (13b)



Fig. 4. Azores WT #2 experimental and INPark simulated active power. Fig. 7. Azores case study: spectra of the wind park simulated and experimental active power (MV side).

Fig. 5. Experimental and INPark simulated voltage at Azores WT #2 terminals.

Fig. 6. Azores case study: spectra of simulated and experimental active power for wind turbine #2.

example). The highlighted frequencies correspond to the usual peaks for the blades and tower interference, rotor sampling (3p), yaw angle and/or misalignment (1p), and the higher frequency (here with the value of 8.42 Hz) is usually detected in the standard wind turbine active power output and is normally attributed to the induction generator. In Fig. 7, similar data are presented, i.e., the simulated and experimental spectra, here for the whole wind park active power,

measured and simulated at the interconnection substation (MV side). The results clearly depict the contribution of the two types of turbines (NTK 99 and NTK 150), being the p frequencies correspondent to the larger wind turbine. Both the experimental values and simulation results are consistent in what concerns the Hz), equal for both energy content of the higher peak ( wind turbines since the induction generators that equipped them were of the same type and size. 2) IEA Alsvik Benchmark Case Study: The benchmark test selected for the dynamic models validation within IEA Annex 21 was the Alsvik wind park case study mainly due to the high amount of constructive and structural data available for the Danwind 180-kW stall wind turbine. These turbines are equipped only with a startup control system and maximum wind and power protections. There are no RTF or any other devices that equip modern wind turbines. The park is constituted by four wind turbines. A synthesis of the detailed data for the Alsvik wind turbines used for the IEA Annex 21 benchmark test may be found in [22]. Added structural data are available from Ganander [23]. A synthesis of the Alsvik parameters is presented in Appendix C. Fig. 8 depicts the comparison of the INPark simulated and experimental values for the active power of the IEA benchmark case study of Alsvik. The same time series (sample d6 379) and comparison is displayed in the frequency domain in Fig. 9. Once again, all of the relevant characteristic and natural frequencies were retained by the model. Limitations on the experimental sampling frequency and the duration of these experimental datasets as well as on the full representation of the wind mechanical turbulencecommon cup anemometers were used in the experiment to measure the time-dependent horizontal wind speedconstitute barriers to the complete analysis of these data, in dynamic terms. B. Transient Operation: Response to a Grid Voltage Dip The available experimental dataset for a transient operation of the power system in IEA databank for the (non-doubly fed)



Fig. 8. Alsvik wind turbine #2 measured and simulated active power.

Fig. 10. Alsvik voltage dip dataset: measured and simulated voltage at the wind turbine terminals.

Fig. 9. Alsvik case study: spectra of simulated and measured active power for wind turbine #2.

Fig. 11. Experimental and simulated active power for the Alsvik voltage dip dataset.

standard induction wind turbines addressed in this section correspond to the recorded characteristic quantities of the Alsvik wind park response to a grid voltage dip. INPark model input data are the same presented in this section for the grid steady-state operation. The time-dependency of the wind was neglected for this simulation, being the mean wind speed set at 4.5 m/s. The results depicted in Figs. 10 and 11 show that the model retains the frequency and general pattern of the oscillation in the transient response, although their amplitude is more damped, possibly due to the poor characterization of some input dynamic parameters. C. Analysis of the Results The results of the INPark model agree fairly well with the experimental data for the IEA Azores case study, both in time and in frequency domain, with this latest considered even more indicative of the model adequacy. Characteristic frequencies corresponding to the turbine rotational speed and blade pasare clearly identied by the model as well as the sage turbine response to the wind high frequency energy content. Although a fair agreement was also obtained for the IEA Alsvik benchmark case study, in this situation, the simulation results

show a larger deviation from the experimental values than for Azores. A set of reasons exists for the higher deviations in the Alsvik simulation, among them the fact that not all the data for the INPark model were available (e.g., standard saturation characteristics were used for the generator due to non-availability of data) and the wind transducers used for this setup being common cup anemometers, rather than the sonic 3-D fast anemometry used in Azores. The model transient response simulated for the voltage dip measured in Alsvik agrees in an acceptable way with the experimentally obtained data. The model retains the transient oscillations main frequency and general pattern but shows a higher damping of their amplitudes. V. CONCLUSION Nowadays, the factors with major impact on the wind turbines behavior and overall power quality are identied. This fact, supported by the actual intense development of adequate dynamic wind generation models, allows the denition of methodologies to assess how wind power generation may affect the power system quality of service and its overall response to grid transient occurrences, including the response of these new power plants to severe voltage dips.



In this paper, a dynamic wind generation model for this purpose was presented and compared to experimental results. The validation process showed an overall good description of the measurements, especially taking into consideration the difculties to conduct these experiments, and the reduced quality of some of the available dynamic input parameters, as well as poor characterization of the wind turbulence for the benchmark selected case studies. The major limitation for these models spread and wider use seems to be the high level of detail needed for the wind turbine constructive data, most of it classied by the manufacturers. In the future, the development of wind turbine standard generalized dynamic models to embed in the power system usual numerical tools is a possible solution for this problem since these may be prepared to accept normalized input data from manufacturers in a black box format, therefore protecting the requested constructive condentiality.

Direct and quadrature components of stator voltage. Effective wind velocity. Electric generator inertia (angular velocity ). Flow angle. Gearbox ratio. Lift and drag coefcients. Mass of one blade element. Projection of along the axis, with and without tower interference. Projection of along the axis, with and without tower interference. Rotor angular velocity. Rotor inertia (blades, hub and primary ). shaft of angular velocity Shaft damping (referred to the primary shaft). Shaft stiffness (referred to the primary shaft). Stator angular velocity. Tangential velocity. Tower position, velocity, and acceleration. The moments acting on a blade element dr of mass dm, in the axial and rotational directions (other than with aerodynamic forces origin), in this model are as follows. Moment due to centrifugal force. Moment of the ctitious force due to the tower movement. Moment associated to the blade-hub spring connecting in the axial direction, in the connection with stiffness value and rigid for the remaining radius. Moment due to the blade inertia. Moment associated to Coriolis force due to movement in axial direction. Moment due to the mechanical damping (rotational). Moment due to the spring stiffness Moment due to the blade inertia. .

APPENDIX A NOMENCLATURE AND SYMBOLS Air density. Angular velocity of the wind turbine (rotor). Attack and twist angles. Axial and tangential interference factors. Axial velocity. Blade angular deection (ap) in the ow direction (axial). Blade angular deection (lead-lag) in the rotational direction. Blade angular velocity and acceleration (tangential direction). Blade angular velocity and acceleration in the axial direction. Blade azimuth angle. Blade inertia. Blade length. Blade number and chord. Blade radius. Blade stiffness, rotation direction. Blade-hub stiffness (axial direction). Damping coefcient, tangential direction. Direct and quadrature components of stator magnetic ux. Direct and quadrature components of rotor magnetic leakage ux referring to the stator. Direct and quadrature components of stator current.

APPENDIX B AZORES CASE STUDY DATA Wind rotor, drive train, and structural data (NTK150) kW; m; m;



Kgm2; Kgm2; rpm. Wind rotor, drive train, and structural data (NTK99) kW; m; m; Kgm2; Kgm2; rpm. Induction generator V; A; ; H; H;

APPENDIX C ALSVIK CASE STUDY DATA Wind rotor, drive train, and structural data (DanWind180); kW; m; m, Kgm2; Kgm2; rpm. Induction generator: V; A; H; H; mH; . ACKNOWLEDGMENT The author would like to thank NATO and the European Union programs that supported the initial phase of this research and the Portuguese Government that, through INETI, enable to complete this work by participating in the IEA Wind Task 21. The author would also like to thank Mr. J. Lameira and Mr. P. Costa for their cooperation on running INPark for some of the specied datasets and on the preparation of the graphic les. REFERENCES [1] DENA, Planning of the Grid Integration of Wind Energy in Germany Onshore and Offshore Up to the Year 2020, 2005, DENA Grid study. Deutsche Energie-Agentur. [2] A. I. Estanqueiro, A. R. Castro, J. Ricardo, M. Pinto, R. Rodrigues, and J. P. Lopes, How to prepare a power system for 12% wind energy penetration: The Portuguese case study, presented at the 2006 Nordic Wind Power Conf., Espoo, Finland. [3] R. Doherty, H. Outhred, and M. OMalley, Establishing the role that wind generation may have in future generation portfolios, IEEE Trans. Power Syst., vol. 21, no. 3, pp. 14151422, Aug. 2006. [4] E. D. Castronuovo and J. A. P. Lopes, On the optimization of the daily operation of a wind-hydro power plant, IEEE Trans. Power Syst., vol. 19, no. 3, pp. 15991606, Aug. 2004. [5] J. O. Tande, E. Muljadi, O. Carlson, J. Pierik, A. I. Estanqueiro, P. Srensen, M. OMalley, A. Mullane, O. Anaya-Lara, and B. Lemstrom, Dynamic models of wind farms for power system studiesStatus by IEA wind R&D Annex 21, in Proc. EWEC04, London, U.K., 2004. [6] T. Petru and T. Thiringer, Modeling of wind turbines for power system studies, IEEE Trans. Power Syst., vol. 17, no. 4, pp. 11321139, Nov. 2002.

[7] Wind Turbine Generator SystemsPart 21: Measurement and Assessment of Power Quality Characteristics of Grid Connected Wind Turbines, IEC Std. 61400-21:2001, 2001. [8] A. I. Estanqueiro, INPark Model Development: Theoretical Approach and Numerical Implementation, NATO PO- Mistral SfS Final Report, INETI, Lisbon, Portugal, 2001. URL last accessed Nov. 30, 2006. [Online]. Available: [9] A. I. Estanqueiro, WIRING Project Final Report (JOR3-CT98-0245) Part A: Theoretical Basis of INETIs INPark Wind Park and Local Grid Models, INETI, Lisbon, Potugal, 2002. URL last accessed Nov. 30, 2006. [Online]. Available: [10] A. I. Estanqueiro, WIRING Project Final Report (JOR3-CT98-0245), Part B: Application of the Models, INETI, Lisbon, Potugal, 2002. URL last accessed Nov. 30, 2006. [Online]. Available: download.aspx?id=944086CBC6FAA057B9F55D61BD933995. [11] A. I. Estanqueiro, Modelao dinmica de parques elicos, Ph.D. dissertation, TULisbon, (in Port), Lisbon, Portugal, 1997. [12] J. C. Kaimal, J. C. Wingaard, Y. Izumi, and O. R. Cot, Spectral characteristics of surface layer turbulence, Quart. J. R. Meteorol. Soc., no. 98, pp. 563589, 1972. [13] A. G. Davenport, The spectrum of horizontal gustiness near the ground in high winds, Quart. J. R. Meteorol. Soc., no. 87, pp. 194211, 1961. [14] N. H. Lipman, E. A. Bossanyi, P. D. Dunn, P. J. Musgrove, G. E. Whittle, and C. Maclean, Fluctuations in the output from wind turbine clusters, Wind Eng., vol. 4, no. 1, pp. 17, 1980. [15] J. Cidrs and A. E. Feijo, A linear dynamic model for asynchronous wind turbines with mechanical uctuations, IEEE Trans. Power Syst., vol. 17, no. 3, pp. 681687, Aug. 2002. [16] A. C. Hansen, P. Sorensen, F. Blaabjergand, and J. Bech, Dynamic modelling of wind farm grid integration, Wind Eng., vol. 26, no. 4, pp. 191210, 2002. [17] H. Glauert, Airplane propellers, in Aerodynamic Theory, W. F. Durand, Ed. Gloucester, MA.: Division L. Peter Smith, 1935 (reprint 1976), vol. IV, pp. 170360. [18] P. M. Bongers, Modelling and identication of exible wind turbines and a factorizational approach to robust control design, Ph.D. dissertation, TUDelft, Delft, The Netherlands, 1994. [19] M. M. Zdravkovich, The effects of interference between circular cylinders in cross ow, J. Fluids Struct., no. 1, pp. 239261, 1987. [20] J. M. Ferreira de Jesus, A model for saturation in induction machines, IEEE Trans. Energy Convers., vol. 3, no. 3, pp. 682688, Sep. 1988. [21] R. M. Castro, Gerador de induo: utilizao em pequenas centrais hidroelctricas, M.Sc. thesis, TULisbon, (in Port), Lisbon, Portugal, 1988. [22] J. T. G. Pierik, CSS Wind Farm Dynamic Model Validation. Alsvik Measurements and Simulations, Rep. ECN-C-05-000, 2005. URL last accessed Nov. 30, 2006. [Online]. Available: eRoomReq/Files/energy/12x276IEAAnnexXXIDynamicwindfarmmodels/0_116d4/Alsvik_evaluation_ver5.pdf. [23] H. Ganander, The use of a code-generating system for the derivation of the equations for wind turbine dynamic, Wind Energy, vol. 6, no. 4, pp. 333345, 2003. Ana Estanqueiro (M07) was born in Coimbra, Portugal, in 1963. She received the Electrical Engineer degree, M.Sc degree, and Ph.D. degree in mechanical engineering from the Technical University of Lisbon (TUL), Lisbon, Portugal, in 1986, 1991, and 1997, respectively. She has worked as a Research Scientist at National Institute for Engineering, Technology (INETI), Lisbon, Portugal, since 1987, being currently the Director of the Wind and Ocean Energy Research Unit as well as an Associate Professor at Universidade Lusiada, Lusiada, Portugal. Her research interests are broad within wind energy with a focus on dynamic models of wind turbine beneting from her electrical and mechanical background. Prof. Estanqueiro is currently Chair of the International Energy Agency (IEA) Wind Agreement and President of the PT IEP/IEC CTE 88Wind Turbines.