‘Contents lists available at ScienceDirect = ena ny Renewable Energy ELSEVIER journal homepage: www. com/locate/renene A parametric model for wind turbine power curves incorporating ® environmental conditions oa Yves-Marie Saint-Drenan *”, Romain Besseau *, Malte Jansen °, lain Staffell °, Alberto Troccoli ‘-*, Laurent Dubus ‘*‘, Johannes Schmidt ‘, Katharina Gruber Sofia G. Simoes ®, Siegfried Heier * 4 MINES norsk, PSL Research Universi. OLE Centre Observation, macs Frey, O09, Sophia Antipas, Pace Cente for Emronmental oy, perl Cllege London, London, SW? INE UK ‘RIE PESDRDIOS, Ps. France {School fiona Sconces, Ukr of East Ala, Norwich NRA 7 OK “Worl Energy ond Meteorol Counc (WEMC), Norwich NRA 7, OK "insite for usta anomie Developer, Unies of Nua Resources and fe Slee, 190, Vena, Ase ‘CENSE Cente or Enronmencl and Sustalnaity Research, NOVA Sco o Scence and Technol) NOVA University Usbon, 2829-516 Caparica, Pore! niet of Kase Kase. Germany ARTICLE INFO ABSTRACT ‘ree Nang Received 19 August 2019 Received in revised orm Ari 2020 ‘Accepted 23 Apel 2020, ‘lable onine 13 Nay 2020 ‘Awind turbines power curve relates its power production tothe wind speed it experiences, The typical shape of a power curve is well known and has been studied extensively. However, power curves of in vidual turbine models can vary widely from one another. This is due to both the technical features of the turbine (power density, cu-in and eut-out speeds, limits on rotational speed and aerodynamic ef ficiency), and environmental factors (turbulence intensity, ai density, wind shear and wind veer. Data ‘on individual power curves are often proprietary and only avalable through commercial databases. We therefore develop an open-source model for pitch regulated horizontal axis wind turbine which can a ‘generate the power curve of any turbine, adapted to the specific conditions of any site. This can employ rane fone of sx parametric models advanced inthe literature and accounts for the eleven variables mentioned above. The model is described, the impact of each technical and environmental feature is examined, and Ics then validated against the manufacturer power curves of 91 turbine models, Versions ofthe model Parametric model (Open-source model Model vabéaton ate made available in MATLAB, R and Python code for the community. Crown Copyright © 2020 Published by Elsevier Led. All rights reserved. 1. Introduction, evolve over time due to wear of turbine components, ditt accu- ‘mulation on the wind turbine blades, and many other effets [6]. Evaluation of the power curve during the lifetime of a wind farm is “The power curve of a wind turbine relates the speed of the wind flow intercepted by the wind turbine rotor to its electrical output. A ower curve is needed at different stages of the lifetime of a wind farm, Prior to its market introduction, the power curve of a newly designed turbine must be assessed to validate its performance. Project developers use power curves together with wind infrma- tion to evaluate the economic viability of developing a wind farm. ‘When operating, the aerodynamic efficiency of a turbine may * Coresponding author. Email edie ywesmariessin-renan@ mines paseehit (M,.Saiat- Drenan) ep: dora 0 106jsenene 202004125 ‘0580-1481/cown Copyright © 2020 Pie ty Elsevier dA ight eer therefore useful to monitor the state of health of the turbines [5] and degradation due to ageing [11.35/41], Power curves are also used to estimate the aggregated power production of wind farms, and their integration into national power systems and electricity ‘markets [16,42] “There has been extensive research on methods for assessing power curves over the last decades [|4,18.23,33.43.47.49]. Indeed, the quality of power curves i a critical issue since the risk involved inthe building and operation of wind farms depends directly on the accuracy of this information. [deally, power curves should be ‘measured in a wind tunnel under controlled conditions. Due tothe large dimensions of modern wind turbines, power curves can onlyAL Suin-Drenan tal / Renewable Energy 157 (2020) 754-765 15 bbe evaluated in real outdoor conditions, making robust assessment difficult due to the spatial and temporal variations of the wind speed. In addition, measurement procedures recommended in the IEC standard 61400-12-1 [20] are continuously improved [30}- All these activities jointly contribute to the increased accuracy of assessed power curve and in a reduction of the acquisition time needed to evaluate them, Power curves are assessed and made available by turbine ‘manufacturers after the correction of different issues, such as tur- bbulence intensity, wind shear, wind veer, up-flow angle; following, the procedures defined by the IEC standard 61400-12-1 [20] These power curves can be found in the product sheets of wind turbines fr in databases which collate numerous power curves, such as ‘hhewindpower.net [44] or WindPRO [22], which are used in this study. While convenient, these databases are not freely available. Unfortunately, power curves of many wind turbines remain diffi- cult to find and, when available, information such as the reference turbulence intensity or the air density is frequently missing. This Jack of information leads to a non-negligible uncertainty in the owercalculation ofa given turbine at best, and tothe impossibility ‘of performing such calculation at worst. Iris particularly an issue in prospective analyses of the energy mix where power curves are required for each individual turbine installed across a wide area [1642]. This paper addresses these issues mentioned above: the availability of power curves and the consideration of enviconmental parameters, by proposing a parameterised power curve model ‘where the impact of turbulence intensity, wind shear, wind veer and air density are explicitly considered, There are different possibilities for estimating the power pro- duction from wind speed data when the power curve is unknown, One approach consists of using a statistical model whose parame- ters are trained on joint measurements of power output and ‘meteorological inputs fora historical period. An impressive number ‘of analytical and statistical tools have been identified, including, polynomial models, linearised segmented models, neural networks and fuzzy methods (26, 40]. If statistical models fook similar to a power curve at first glance and can be exchanged in some applications, they differ on some ‘important points. Firstly, statistical models capture the relationship ‘between wind speed and net (rather than gross) power production, including potential wake effects, the impact ofthe local orographyy, ‘wind turbine availability or even systematic errors in the wind speed data, These factors should be disentangled as the gross tur- bine production is of interest. Secondly. supervised statistical ‘models require training data and therefore they cannot be used to ‘model a planned wind farm or to simulate a fleet of wind farms ‘where available measurements do not yet exist In the latter case, the use of power curve is unavoidable and the lack of information is ‘often addressed by choosing equivalent power curves based on the similarity between the desired turbine and those for which a power curve is available [4]. However, even this approach lacks a widely recognised! and validated rationale. {AS described in numerous studies on the dynamics of wind turbines {19,46}, the behaviour of the power production of a tur- bine can be estimated asa function ofthe wind speed using general characteristics of the turbine and a power coefficient model. To the ‘best of our knowledge, the use of such models for generating power ‘curves has so far not been systematically studied and validated. The physicaly-based approach suggested here uses the rated power and the rotor dimension as the main input parameters, and allows other operating characteristics, which are also standard informa- tion, to also be specified (such as cut-in and cut-out wind speed, or minimal or maximal rotational speed). This work relies on existing analytic power coefficient functions describing the aerodynamic efficiency of the blade published in the literature such as eg. Ref [10,19]. AS a consequence, the parameterised model is valid as Jong.as the analytic power coefficient functions are. Those models ‘were developed for horizontal-axis wind turbine but other input data stemming from blade measurements or numerical calculation ‘an be used instead. Finally, the model proposed here offers the possibility to explicitly account for environmental factors such as, turbulence intensity, air density. wind shear and veer. The effect of aerodynamic obstacles surrounding a wind turbine on its power is, ‘not considered inthis study because it is an information specific to cach turbine. This paper is structured in six main parts. A comprehensive description is given in 2of the different steps necessary to evaluate ‘a power curve from general characteristics of a wind farm, such as the rotor area or the nominal power. This section also includes a discussion on the consideration and influence of external env ronmental parameters. Owing tothe numerous input parameters of the model, a sensitivity analysis and statistical analysis of these parameters are described in Section 3 and section 4, respectively. The results of our validation are summarised in section 5, where the ‘model output has been compared to power curve from ‘tewindpower.net [44] database. Some insights on the limitations and possible improvements to the proposed model are discussed in section 6, along with its possible domains of application. imple- mentations of the proposed model in Python, R and MATLAB are also provided as supplementary material to this paper: 2. Methodology 24. Operating regions of a wind turbine Power curves are traditionally divided into four operation re- sions, as shown in Fig. 1 and detailed below. At very low wind speeds, the torque exerted by the wind on the blades is insufficient to bring the turbine to rotate, The wind speed at which the turbine starts {0 generate electricity is called cut-in wind speed and is typically between 3 and 4 m/s. Region I corresponds to wind speeds below this cut-in wind speed. Power can be consumed in this re- gion from turbine electronics, communications and heating/de- ‘cing of blades, although these ancillary loads are not included in ower curves. Above the cut-in wind speed, there is sufficient torque for rotation, and power production increases with the cube of wind speed before reaching a threshold corresponding to the rated po- wer of the turbine (oF nominal power) that is designed to not exceed. The lowest wind speed at which the nominal power is reached is called the rated (or nominal) wind speed and is typically between 12 and 17 m/s. Region I is delimited by the cut-in and the tated wind speed, and corresponds fo an interval where the wind ‘turbine operates at maximal efficiency. There are, however, some ‘exceptions to the optimal operation of the wind turbine in this region. Firstly, while an optimal operation requires the rotational speed to be proportional tothe wind speed, the speed of rotation is ‘bounded by lower and upper limits. Secondly, at high wind speeds, the turbine can sometimes be deliberately operated at lower power to reduce rotor torque and noise levels [25] For wind speeds above the rated wind speed, the wind turbine is designed to keep output power at the rated power, which cannot be exceeded. This can be achieved by means of a stall regulation or pitch control. The latter solution consists in adjusting the pitch angle of the blades to keep the power at the constant level and is ‘overwhelmingly used in modern large turbines. Region Il corre- sponds to wind speed values where the turbine operates at its rated power, and is bounded by the nominal wind speed and the cut-off ‘wind speed, which is introduced below. The forces acting on the turbine structure increase with wind16 YM Saint rena et a Renwable Energy 157 (2020) 754-768 2500 { i 1 = ez c tt 1 “) ol Smeets Se eae re lode pitch angle ty oars Wind speed fs) Fig. 1 Opting repos of pia pitch eeulted wind tube and evolution of ich angle ation speed and Upsped ao TSR) wh wind speed speed, and at some point the structural condition of the turbine can be endangered. To prevent damage, a braking system is employed to bring the rotor to a standstill [50], The cut-off wind speed cor- responds to the maximum wind speed a wind turbine can safely support while generating power and is usually about 25 ms. Re- ‘gion V includes all wind speeds larger than the cut-off wind spe Some manufacturers have introduced storm control in larger~ bladed turbine models, where the power is gradually reduced (eg. from 21 m/s up to25 m/s) to prevent such drastic loss of power at the cut-out speed. Fig. 1 shows the different operating regions described above as well asthe evolution of the main operating parameters of a wind turbine: pitch angle, rotor speed and tip-speed ratio (TSR). This visual representation is based on previous works [2,7.10,24) 22. Parametric wind turbine power curve ‘The wind power calculation for regions I I and IV is trivial with the information typically available on a wind turbine.’ However, the description of the power curve in region II is complex and ‘methodologies to improve itare still being researched, Between the cut-in and the rated wind speeds, the wind power production Fyr can be calculated by Eq, (1) asa function of the wind speed Viys alt density p, ot0r area Arwor and power coefficient Cp(2, 8), with J being the tip-speed ratio and the blade angle [19] The cit-in toad ate wind pes a6 wel asthe ated power are ypc ‘sven a wind fine product sets. raat avalube, tng partes can tl beestimatd, as explaed ater. Parr oA 8) “ Rotor area is straightforward to obtain and data on air density, are readily available, although it should be noted that density varies ‘ver space and time (for example between 1.1 kg/m? and 13 kg/m? between summer and winter in Germany [42]. That said, the parameter with the largest uncertainty in Eq, (|) is the power co- efficient C,(2,0)¢ which ultimately depends on the wind speed. 221. Parametric model ofthe power coefficient C5(2.8) ‘The power coefficient C2, 8) expresses the recoverable fraction ofthe powerin the wind flow. This quantity is generally assumed to bea function of both tip-speed ratio 2 and blade pitch angle @ [19] ‘The power coefficient can either be evaluated experimentally or calculated numerically using blade clement momentum (BEM), computational fluid dynamics (CFD) or generalised dynamic wake (GDW) models [10.12,37} A less accurate but convenient altema- tive consists in using numerical approximations. A few empirical relations can be found inthe literature (see eg. Ref. |19]}) with the _general form: {3 8) @ The above equation reflects the general relationship of the po- wer coefficient Gy with and g. Different values can be found for the ‘model coefficients ¢, In our study, six different parameter sets from different authors have been considered [10,13,29,38,45]. These different parameterisations ae listed and illustrated in A. We limit the extent ofthe analysis to six parameterisations but our approach can be easily extended to any other parametric models or numer- Jeal data, (2/2, — €38 — C428 — €58% — coe + cg 41 = + c8)!~c10(6? 41) | 222, Determination ofthe blade pitch angle, 8, as a function of. wind speed IF we assume that the wind turbine is designed to achieve its ‘maximum efficiency in region Il, the blade pitch angle can be set to zero between the cut-in and the nominal wind speed. Indeed, i is ‘usually assumed that the blade pitch is only used fo limit the power production to the nominal power in region Ill [2.10.25] and our ‘modelling assumption seems therefore reasonable. That said, pitch angle can be also used in regulation strategies that aim to limit rnoise emissions or mechanical effects on the turbine structure within region Ml [25]. Such strategies are not considered in the present work and their integration in our modelling approach may be the subject of future developments. 223, Determination ofthe tip-speed ratio 2.as a function of the wind speed ‘The tip-speed of the blade is equal to the product ofthe rota- tional speed of the rotor « and the rotor radius, Djxor/2. We can therefore express the tip-speed ratio as a function of the rotor rotational speed and radius as well as of the wind speed Vis as follows: (Drawer /2) va @ As explined above, the aerodynamic efciency ofthe wind turbine depends on the tip-speed ratio. The maximum power Yield in regions therefore obtained fr Hye namely the Vale that maximises G fora given wind sped:AL Suin-Drenan tal / Renewable Energy 157 (2020) 754-765 7 Dog = arg max C2.) @ riod Considering Eq. (2), if the wind turbine is operating at constant tip-speed ratio, the rotational speed of the rotor « should vary proportionally to the wind speed Vjys. Tis is only possible in the ‘operating range ofthe turbine, which is bounded by tpn td “ma. ‘This constraint should be taken into account in the estimation of according to Eq. (3). Based on previous works (eg Refs. [2.24)), we use a simple approach, which consists in estimating the value of & using the rotational speed « as follows: tin (cma maxon p23 - Vas) ® As illustrated in Fig, 1, the rotational speed « given by Eq. (5) corresponds (0 gy but is bounded between tn and ty. It ean bbe observed that the maximum value of Cp with constrained rota- tional speed w is obtained with Eq, (5) due to the monotonic ‘behaviour ofthe function Cp(A) for 4< Jaye and 2> ep respectively. It should be stressed that the validity of the relationship proposed inthis section is imited to the case of pitch regulated plant and itis simplified: it doesnt take into account regulation mechanisms such a noise reduction or load mitigation, 23. Considering the effect of external parameters on the power, ‘As summarised later in section 2.4, the relationships and ‘modelling assumptions described in section 2.2 are sufficient for estimating the power curve of a wind turbine. However, this power curve corresponds to ideal operating conditions and external fac- tors should be taken into consideration to better simulate the ‘behaviour of awind turbine in real conditions. These factors are the turbulence intensity, air density, wind shear and wind veer, inflow angle and wake effets ‘Wake effects are strongly dependent on the specific layout of a ‘wind farm, particularly the number of turbines and their spacing. ‘Wake losses amount to approximately 11-13% for turbines spaced 7 {09 turbine diameters apart [5.17] As the losses are time-varying, ‘due (© wind speed and its prevailing direction, they can not be considered further in the present work. The inflow angle results {rom the effect of the orography on the wind but, since it depends fon the site and less on the wind turbine itself, it is not considered here. The effects of the remaining parameters on the power curve are evaluated next 234. The effect of turbulence intensity on the power curve The power curve derived in the previous section corresponds to the ideal case of a laminar and stationary wind conditions, which rarely occurs in practice. Since the relationship between wind po- ‘wer and wind speed is non-linear, the effect of high frequency variations in the wind speed on the power must be taken into consideration [28]. This is usually realised by considering the tur- bbulence intensity (TI) defined as: atu) naa 6) Inthe above equation, yu) represents the mean wind speed and «a(u) the standard deviation of the wind speed measured ata fre- {quency of 1Hz or higher ina time period of 10 min [20] Typical values forthe average turbulence intensity range from 5 t0 15% When no time series ofthe turbulence intensity is available, iis usual to assume a constant value ofthe turbulence intensity for a particular ste Numerous works have been produced to evaluate and model the effect ofthe turbulence intensity on the power production of wind ‘turbines [3.5] In this work, the impact of the turbulence intensity ‘on the power curve is pragmatically calculated by assuming that short-term variations of the wind speed” follow a Gaussian distr- bution with mean U = yu) and standard deviation UTI (see eg, Albers []). With this assumption the effect ofthe turbulence in- tensity on the power curve fora wind speed U can be considered by ‘making a convolution between the original power curve and a Gaussian Kernel of mean U and standard deviation U-Tl and taking, the resulting power for the wind speed U. This calculation is illus- ‘wated in Fig. 2 In Fig. 2, the upper, middle and lower plots represent respec- tively the original power curve, the different Kernels and the final power curve. The modified power value is the weighted average of power values at wind speeds between 9 and 13 where the weights are the blue kernel of the middle plot. The vertical light grey lines represent the weighted average. The same procedure is iterated for ‘each wind speed with the different kernels represented in the ‘middle plot. The effect of the turbulence intensity on a power curve is, illustrated in Fig, 3 for different values of the turbulence intensity between 0 and 15%, using the power curve of a 2-MW wind turbine ‘with a rotor diameter of 80 m. This example shows clearly that the effect of the turbulence intensity can be significant, especially around the nominal wind speed. This parameter is therefore of paramount importance for the estimation of the power curve in real condition. It will be taken into consideration in the comparison ‘of the model output with manufacturers power curves in section 5 ‘As can be seen in Fig. 3, the turbulence intensity has no effect on the sudden power decrease as the wind speed exceeds its cut-off value. It was indeed decided not to apply the smoothing effect of the turbulence intensity in this region since the cut-off is not activated based on high frequency wind speed but based on a longer time average. In addition, an hysteresis implemented for the restart of the wind turbine as the wind speed decreases below the ‘cut-off value hinders using the kernel convolution approach for the calculation of the Tl effect on the power production 23.2. The effects of the air density on the power curve ‘With the approach proposed in this work, the consideration of air density on the power curve is explicit and straightforward, as is illustrated for values varying between 1.15 and 1.3 kgy/m? in Fig. 4 ‘The reference value forthe air density is set to 1.225 kg/m?, which lies in the middle of the variation interval It can be observed in Fig. 4 that the impact of varying air density on the power curve is ‘much lower than the effect of the turbulence intensity. Yet, i im- pacts the power curve across the whole range of Region Il, where the frequency of occurrence is generally high and a careful consideration of this external factor should therefore be made. 23.3, The effects ofthe wind shear and the wind veer on the power ‘Wine speed is not uniform across the wind turbine’s rotor plane, asi increases with height through the atmospheric boundary layer. ‘The vertical wind profile can be described in several ways [34], such as the logarithmic profile which depends on the roughness length, {riction velocity and stability parameter, In many applications, the simpler power law model is used which relates the ratio of the * The pial maximum of the spectral density of dhe wind speed has is maximum inthe feqency range of st 1/100 Ho, white even large ind lot response higher than 1/10 Ha) [i "18 YM Saint rena et a Renwable Energy 157 (2020) 754-768 3 3 8 8 2) Power curve corresponding to a turbulence intensity of 0% ° 3 10 18 20 Ey Wind speed [m/s] Power production [kW] g b) Wind speed and TI dependant Gaussian smoothing kemels ( 3 10 15 20 2 Wind speed [m/s] a3 g Power production [kW] é 00 ) Power curve corresponding to a turbulence intensity of 10% °9 5 10 15 20 Ey Wind speed [m/s] Fg.2. strato of he meth used tale te eet of urbulencs ntensiy on the power cure the upper mile and lwer pls represent respectively the orpinal power ‘uve the dileent Keats and the sl power cave. An cumple of elation ora wind speed of Il preted z Te oe . Sen - g ine a Ay 25) & nf Sam : § sao ts i 2 oye me ~ jl : B eo ee 28 i jo bm 2 a ee 5 wine pei ee ‘Fig. L Ilustration of the effect of the turbulence intensity an a power curve. ‘Wind speed [mm/s] le. ustration ofthe effect of he deny on power curveYeAL Sint Dena ta Renee Energy 157 (2020) 754-768 0 Height above ground {m] Wind speed ratio V(zyV(Hub) (+ Height above ground fm] [Absolute diference in wind direction Fig 5. lustation of wing shear and wind weer ars the wind turbine rotor The lt pane ilstrates2urbin oto died int erzotal bands, corresponding thse in (91 The mde and rig panes illustrate the variation n wind spec ad dection wth hist ‘wind speeds at two heights with the power ofthe ratio of the two heights: @)=wan)-(Z)" @ ‘Where uz) is the wind speed at hub height, yu». and u(z) is the ‘wind speed at height z. In this equation, « isthe Hellman or shear coefficient, which quantifies the wind shear and varies typically between 0 and 0.4. The effect of the wind shear on the vertical profile of the wind speed in the region ofa rotor area i illustrated in the middle panel of Fig 5. In this example, a hub height of 60 m. and a rotor diameter of 80 m have been assumed. Wind veer is defined as the change in wind direction as a function of height. It has been shown that wind veer does exist in {ypical wind situations [21] To consider wind veer, we assume that the change in wind direction is zero at hub height and varies in- early with height according to (2) = v-(2— Zina) (8) In the above equation, the parameter v quantifies the evolution of the difference in wind direction A(z) as a function of the height difference (2 ~ zy) Based on the statistical analysis of Ivanell etal. [21], we assume that this parameter can vary between O and 0.75") ‘m. The wind veer is illustrated in the right plot of Fg. 5 To evaluate the impact of wind shear and veer on the power curve, we follow the approach that is recommended ina revision of the IEC standard [3], which consists in replacing the wind speed at ‘hub height by a rotor equivalent wind speed Uc, which is defined Usy= (EG) weosanir® @ ‘The coefficients 4; correspond to the area of elementary hor zontal bands ofthe rotor area a illustrated inthe let panel of Fig. LU, and Ag, corresponds respectively to the wind speed and varia- tion ofthe wind direction with respect to that at hub height inthe "horizontal band. The influences of the wind shear and veer on the power curve of 2 2-MW wind turbine with a rotor diameter of 80 m and a hub height of 60 mare represented in Fig. 6. The variations in the power curve illustrated in these two plots were obtained by replacing the ‘wind speed at hub height by the equivalent wind speed calculated with Eq. (9) In the upper plot of Fis. 6, the impact ofthe wind shear is barely, visible, which is in agreement with the work of Wagner etal. [48] The limited effect ofthe wind shearon the power production can be explained by two factors. Firstly, larger values of the cubic wind speed above hub height are balanced by lower values below hub height. I should be however noted that this balancing effect be- ‘comes limited as the vertical distance to the hub height increases Secondly, the impact of wind speed values far from the hub height are limited by the area of the horizontal rotor band, which de- ‘creases with increasing distance to the hub height In the lower plot of Fig. 6 it can be observed that the impact of the wind veer is small yet greater than that of the wind shear. Indeed, the effect of wind veer is larger that of the wind shear because the effective wind speed decteases above and under the nub so that there is no balancing effect. et, the weighting resulting, {rom the horizontal bands of rotor area limits the effect of the vertical change in wind direction on REWS. ‘We can also observe in Fig. 6 that the wind shear and veer are ‘not impacting the power curve in the cut-off region; we consider indeed that the cut-off is activated based on the measurements of ‘wind speed at hub height. ‘The wind shear and veer are both integrated in the codes accompanying this paper. It necessitates information on the hub height as addtional parameter forthe generation of power curve. 24, Summary of the modelling approach The main computational steps described in the previous sec- tions are summarised in Fg. 7. In the first step, the rotor speed is ‘evaluated as a function ofthe wind speed. This is achieved using the ‘optimal tip-speed ratio evaluated with Ea (4) and the relationship ‘between wind speed and tip-speed ratio given by Eq. (2), respecting the operational range of the rotor speed lone In the Second step, the evolution of the power coefficient with wind speed is, evaluated using the rotor speed » and a power funetion C(2,2)-The analytical expression given in Eq (2) is used in this work, but other expressions can be implemented instead. In order to differentiate the form ofthe Cy(i,8) to its maximal value, the function Cy(,8) is scaled so that its maximal value is the newly introduced parameter Cpomaxe The power output of the wind turbine is evaluated in the third step using Eq. 1). This curve isscaled by the nominal power of the turbine, and then the cutin and cut-off wind speeds are applied. The fourth and final step introduces the effect of the70 YM Saint rena et a Renwable Energy 157 (2020) 754-768 . el eo a im lame E a sam 8 § 1000 7 2 é é oom 5 E 500] Bed : : pow OR eewacnwn | accencemaiu at ‘Wind speed [rvs] ‘Wind speed [m/s] (b) Saal 7 Eo E on g 2 1800] jos E a 3 a He ~ nt & g se 2 sw z° : : os Wind speed [mis] Wind speed fms] Fg. Influence ofthe wind shear (2) and the wind ver) onthe ower curve ofa yi 2-MW wind turbine (80. rotor ameter a ub- ight of 60 CpAB) Cymax — (min @max) rotor (VeuVeo) Pnom (TI,Shear,Veer) =. speed ofthe rotor 2) Caeulation ofthe power 4 ° Coefient aa function ofthe |3) Calculation af trblence. wind speed free power curve = : 1) Consideration ofthe effet ak te is the wind T,shearandveeron the power cue j “Sh | i erolona(oeg el ga” * ba = arog) tes et GO) ase) : Pan =f (rey T Shea. Yer) Vws) F oveer=0) Fig. 7 Row chart epeseating the main computation sep forthe estimation of power cave fo characters a wind ubeAL Suin-Drenan tal / Renewable Energy 157 (2020) 754-765 ze turbulence intensity on the power curve. This is the only external effect that is considered explicitly in our model, since other effects can be applied by evaluating a rotor equivalent power curve [3] ‘The input parameters of the model are listed in the upper line of. Fig. 7. These input parameters can be gathered in three groups of Alifferent natures. Firstly, wo parameters Cy(2.8) and Cyrmn are related to the aerodynamic efficiency of the blades. The second group of parameters (71 and pj.) ae related to external conditions. Finally, the lat group of parameter includes 6 design characteristics ‘of wind turbines that can in most cases be found in manufacturer's product sheets. ‘As argued in Ref, [31], energy models should be made open to improve the quality of science and aid the productivity of other researchers. We therefore implement the model outlined in Fig. 7in three widely-used programming languages, Python, Rand MATLAB, ‘The power curve for an arbitrary wind turbine can be generated by simply specifying the rotor diameter and nominal power output. Sensible defaults are given for all other parameters, or they can be Customised as desired. The model code is available from Github _ntps:{/github.com/YvesMSaintDrenan/WT.PowerCurveModel 3. Analysis of the sensitivity of the power curve to the model parameters A sensitivity analysis was performed to assess the relative importance of the different parameters of the model. A reference set of parameters was determined and the sensitivity of the power curve to each parameter was evaluated by varying each parameter individually across a typical range. The set of reference parameters and their variation interval are given in Table I. The present sensitivity analysis is limited to a univariate anal- ysis: one parameter is varied at the time. The sensitivities of the parameters are not quantified as in the Mortis screening method [27] or generalised sensitivity analysis [39] but only qualitatively assessed For this purpose, the sensitivities are visually represented by lines of different colours for the different values of the varied parameters in the plots of Fg. 8 ‘The sensitivity af the reference power curve ta variations ofthe rotor area and the nominal power are displayed in the first row of Fig. 8 (plots (a) and (b)). These two parameters yield the largest sensitivity to the output power and should thus be treated with the greatest caution. However, their level of uncertainty is negligible as they both are design parameters and most manufacturers mention them directly in the name of the turbine (e.g. Vestas VB0-2000, Enercon E82 £2/2.0 MW, GE Haliade 150-6 MW, Gamesa G114- 2.0 MW, Bonus B82/2300 ...). Tes interesting to note that for very small rotors, the power. ‘nntet Reference values and variation intra ofthe diferent parameters considered a the sensitivity anal Parameters Teerencevalve_Valton inten Rotor diameter wm 40-120 Nominal power 2000 kw 150-250 Kw Cain wind spec spins oS mis {cat ot wind speed 25mis 20-30 mis Minimal rotation speed 10rpm 0-15 9m Maximum Gy value aot 03-058 ( parameterision Dateralfio} Stats tal Thongam etal [5 De Kooning et 19) ‘iene ea 29} Daveral (10) curves move away froma cubic increase and even decrease at high ‘wind speed. This is due to the rotational speed that increases with decreasing rotor area to reach the optimal TSR. As soon as the rotational speed is bounded by the maximal rational speed in- crease ofthe wind speed brings about decrease of the power co- efficient which can ultimately results ina decrease of the power (blue curves in 8-3}. “The effects of variations ofthe cut-in an cut-off wind speeds on the reference wind turbine ae illustrated in the second row of 8 (plots (6) and (4). It can be observed thatthe cu-off wind speed has the largest impact on the power curve while the sensitivity of the power curve fo the cut-in wind speed is moderate. It should hhowever be noted thatthe frequency of acurrence of wind speed in the neighbourhood ofthe cut-in wind speed can be high while wind speed values close to the cut-off wind speed are much less frequent. The two wind speeds are therefore both important pa- sameters for the estimation ofthe anaval energy production of a ‘wind turbine but they both are of lesser importance compared to other parameters such as the nominal power or the rotor area, ‘The sensitivities of the model output to the minimum and ‘maximum rotation speeds are displayed in the third row of Fis 8. can be observed that the effect of the minimum rotational speed is limited to a wind speed interval of 3-9 mis while that of the maximum rotational speed can be observed for wind speed values close to the nominal wind speed. This is due tothe fact thatthe rotation speed is unconstrained between these two intervals. The effect of the maximum rotor speed is hardly visible. This is due to two reasons: firstly, the wind speed corresponding to the maximum rotation speed is very close tothe nominal speed and, secondly, the decrease of the Cp value with the wind speed at ‘maximal roation speed is relatively smal. The sensitivity ofthe ‘model to the minimum rotation speed is more pronounced and should accordingly be carefully chosen since the frequency of foccurtence of the wind speed in the interval 3-9 mys high In the last row of pts the sensitivity ofthe model output tothe scaled Cp mode and the maximal value ofthe power coefficient are displayed. We rescaled the shape ofthe Cp model tothe magnitude of Cp values given by the model and proposed to scale the output using the new parameters Gm: which corresponds to the ‘maximal power coefficient given by the model. This is very instructive since as can be observed in the plots) and (f) of 8, the power curve is not sensitive to the choice of the scaled parame- terisation but the effect of variations ofthe parameter Cy ya 09 the power curve is significant. This shows that the choice Of the Cp model snot critical but the choice ofan accurate value for Gyna i decisive for an accurate calculation ofthe wind power production. 4, Statistical analysis of the most sensitive model input parameters ‘While parameters such as the nominal power and the rotor area are readily available for each turbine, this is not the case for other pparameters such as the maximum power coefficient or the rotor ‘minimal and maximal speeds. In order to address such situations, a statistical analysis of the model parameters was performed, whici ‘ean be used as guidance to the choice of unknown parameters. For this, we used the database of wind turbines and power curves provided by thewindpower.net [44], which includes extended data ‘on the main turbine characteristics. As of May 2019, this commer- cial database contains about 780 turbines models 41. Maximum value of the power coefficient Gp mae ‘The value of Cymax has been evaluated for 600 wind turbines using the power curve and characteristics of the turbine bym2 YM Saint rena et a Renewable Energy 157 (2020) 754-768 Wing power production [kW a i L Z é Nominal power kW] Wind speed imi} o a 28 Cutt wind speed avs] z : 8 5 = |a) mere oaeea|| fi, so ‘co E 5 ao nS t = E i > B z™ leo 7 Mo. ‘Wd epee) F"l0 man 3 so g i 2 o lo. 8 Mind speed iva ze) womens) IE ace 7 8 5 :* i 2 bg ee ee = 9) ban $1500 E ea ~ See z* el ‘ind speed ins} : ‘Wi spe rl g é 1 =a) IE z. : i 5 03 SF ey = h) [= name ComCoarTD] fom in i bo : a 0. los wee Fig & luce ofthe erent parameters on the power cure In ech panel ne input parame is aie ces typi rang as ven in the panes gen inverting Eq. (1) and selecting the maximum value. This process is illustrated in the two plots of Fig. 9. The power coefficient is plotted as a function of wind speed forall turbines inthe left plot, and a histogram of the maximum values for each turbine is shown in the right plot. It can be noticed that there are some potentially cor- rupted values of Cy these will be discussed in the next section, The ‘most frequent value is 0.44 and 80% of the values are between 0.4 and 0.5. A dependency of this parameter on further characteristics such as eg. the size of the turbine can be expected but no clear dependencies could be identified with the available data. Based on this short analysis, we therefore recommend using a value of 0.44 When this information is not available. 42, Cut-in and cut-off wind speeds The distributions of the cut-in and cut-off wind speeds of the wind turbine information contained in thewindpowernet [44] dataset are displayed in Fig. 10. It can be observed that the cut-in wind speed are between 1 and 5 m/s. Most values are distributed around! 3 m/s with 90% of the values between 2 and 4 mjs. The cut-AL Suin-Drenan tal / Renewable Energy 157 (2020) 754-765 7a | i z i ; : LJ I Fe i i Bos Pea 2 vanaepsosinss Neer Fig 8. Lt: Power coef as a anton ofthe wind speed fr the power curves sll heinpace et [4] data Rig esto othe maximal power ‘oeticentevated fora ava power eres, 00 gm 1 § mo 2 6 0 1 0 2 9% 9 Wind speed [mvs] Fig. 10. Dstibution of the cata wind speed fom the European dataset of ‘ewindpoweret 4 ‘off wind speed are between 15 and 30 ms. The most frequent values are 20 and 25 mis, with a share of all wind turbines of respectively 12% and 70%, When information on the cut-in and/or cut off wind speeds is unavailable, values of respectively 3 and 25 ms can be recommended based on the present analysis. 43, Minimal and maximal rotational speed ‘As can be observed in Fig. 11, the minimum and maximum, rotational speeds exhibit a strong dependency on the rotor diam- eter. The same statistical analysis as those presented in the two previous subsections could therefore not be carried out. Instead, we adopted a similar approach as that described in Ref. [15] and fitted this dependency using an exponential function, which gives the ‘wo following expressions for the minimal and| maximal rotation speed as a function of the rotor diameter: a 1046 558 fi Ding wit {4 eaeait ao) 7 = 705.406 n= win {§ = 705,408 ay The minimum and maximum rotation speed can be estimated ‘with Eq, (10) and Eq, (11) when information on these characteristics is missing. tor diameter Fig Minit and maximal rotational speed 35a fenton ofthe rexr iameter (aca saree thewndponernet Fea. 44, Validation ofthe parametric power curve mode A validation of the model introduced in 2 has been conducted using the power curve of 91 wind turbines with a nominal power greater than 1 MW as provided by their manufacturers and avail- abje in the thewindpower:net (44) dataset. For this validation, the power curve model has been run with specific information on the ‘nominal power and rotor area, while other model inputs are set to the reference values described in 4 and the air density is set to 1.225 kgm’. As the level of turbulence intensity corresponding to ‘each power curve is unknown, they are compared to model outputs obtained with values of the turbulence intensity ranging between and 10%, Asa consequence, a quantitative validation could not be ‘conducted, as this unknown parameter would have Co be optimised in the model (which would instead be calibration). Rather, a qual- itative validation is performed through a visual comparison of the ‘model output to the database of power curves, whose main out- comes are described in this section. In Fig. 12, the result of the validation is given for three wind turbines which show a close correspondence between model output and power curve from the database. While this does not completely exclude the possibility of 4 systematic modelling error, the degree of similarity across the plots suggests the model can synthesise realistic power curves fora variety of wind turbines. Tecan be observed that the best matches between model output and turbine data are obtained for different values of the turbulence intensity: 25,5 and 7.5% for the wind turbine 258, 263 and 270. As mentioned above, information on the turbulence intensity is not available in the database, which hinders a proper validation of the ‘model and represents a non-negligible source of uncertainty for power estimation made with these power curves taken from ‘manufacturers and other sources. This highlights an advantage of the model we present, because it allows the value ofthe turbulence intensity to be controlled, or varied, This also applies othe effect of air density, allowing exploration of how the power curve evolves ‘with the altitude, temperature and season, Examples of wind turbines found to have the highest difference between model and turbine power curves in the validation are given in Fig 13. The two plots on the right side of Fig. 13 show that this difference stems from a mismatch between the maximal value ‘of the power coefficient assumed in our model and the actual value ‘ofa wind turbine. In one case, for turbine 408 this s possibly due to an error in the measured power curve, which actually exceeds the Betz limit. ‘Such values are known to result from power curves measured with shaded anemometer [36]. This observation highlights the need for cateful screening for data quality when using poweret YM Saint rena et a Renewable Energy 157 (2020) 754-768 A a : z : fe i Bae ca é ee Power costiciet Cp (+ re " | a Ba f . ey a “| Wine pees in) ‘Wine speed is) Fig 2. comparion of the mode output wih the pave ures the wind turbines which ae modelled wih igh uty The thee owt how the wind tubes 258,252 and of ewindpowene 3)AL Suin-Drenan tal / Renewable Energy 157 (2020) 754-765 8 oo a sett cae : erie (samme =, Esme, z : i) a3 i i ; 7 4a, yo --— Manutctinr poner cure 7 ie se vise Santee Feet ay Eremcecetee Erisaeaass oem ee eo asain / aren Power kW] Wind speed rvs} Fe 13, compan of the model utp wih the owe curves of wa wind turbines which are model wth ower qty. The two rows show the wind turbines £29 and408 of ‘ewindpoweret 4 ‘curves from manufacturers and databases. Tn the examples presented so far the shape of the power coef- ficient function matches with that from the wind turbine database However, turbine 404 in Fig. 14 exhibits a different shape which does not conform to the majority of other wind turbines. In this ‘ase, i is difficult co identify the reason for the observed difference (modelling error, correction of the effect of turbulence intensity, shaded wind measurements...) and further validation work would bbe needed to get a deeper insight in the performance of the model and possible sources uncertainty on the power curves. ‘The different examples presented above summarise most situ- ations encountered in the validation work. In order to give an ‘overview on the match between the model output and the data- base of power curves, all power curves have been represented by blue lines in Fig. 15. Since the error ofthe model principally oceurs in region Ml, the power curves have been scaled by normalising, them by the rotor area so that all power curves are similar in that area, Indeed, it can be observed that many of them overlap for wind speed values between 3 and 10 m/s. The model outputs obtained with the standard parameters but with power to rotor area ratio of (0.25, 0.375 and 055 and a turbulence intensity of 5% have also been tepresented in this plot. This final analysis shows that even in cases with the largest errors, such as those presented in Figs. 13 and 14, the model is able to reliably reproduce the behaviour ofthe majority of power curves seen across the wind industry, 5. Conclusion ‘We propose here an approach to estimate the power curve of a ‘wind turbine from its main characteristics. The present work has aggregated existing knowledge on wind turbine operation to address 2 current need for energy modelling applications. The model, with 12 parameters, offers the possibilty to adapt ‘the turbulence intensity and air density tothe actual conditions of a specific site. A sensitivity analysis has been conducted and estab- lished that nominal power, the rotor area and the maximal Cp value as the most influencing parameters. Choosing the nominal power ‘or the rotor area i straightforward since these two parameters are ‘often used to characterise a turbine and are even frequently con- tained in the turbine model name. Then, a statistical analysis of the remaining parameters was conducted to suggest default values,76 YM Saint rena et a Renewable Energy 157 (2020) 754-768 Power i) Wd speeding) Power coeticient Ga : * windspecainis) i 4. compari ofthe del cup with the pes curve of wind urine 4D of thewinpowernet 44, wich hat an ukawn denen. Powered ye er ‘ea pees) Fig. Compan ofl power carves conan in the datas ve ues) wth the power aut ela fos poe orator area ef 25,0375 an 013 KW? re curves (out thewindpccet [1 Fr strpetation ofthe eleences to ‘olor i ts gure legend, the reader i elered tthe Web version ofthis atile) relying on an extensive database of turbine characteristics. {qualitative validation has been conducted where the model output hhas been compared to power curves taken from a wind turbine dataset. This validation revealed that the model yields realistic ower curves when compared to those from the database for most wind turbine models. However, large differences are observed for a limited number of wind turbines that deserve further analysis, since it is unclear whether they result from modelling issues or data quality issues within the database. This highlights the need for caution when using power curves found online. Another conclusion ofthis validation is that the diferent power curves contained in the database obviously correspond to different level of turbulence i tensity. This information is generally not given and leads to an uncertainty that is avoided by the use of our model. ‘The proposed model is not aimed at replacing power curve ‘measurements campaigns as described in the IEC 61400—12 [20], which are essential for the characterisation of wind turbines and the monitoring of the energy production of a wind farm. Instead, it can represent helpful additional information to cross check results or to provide a robust best estimate of a turbine’s performance (either for existing or hypothetical future turbines). ts added value is clear for the estimation of the total wind power production in regions where only a limited subset of turbine characteristics are available. ‘The present approach being based on the assumption that a turbine is always operated to yield the maximum possible output, potential improvement of the proposed model may consist in integrating control strategies that result in sub-optimal yield pro- duction such as eg. noise emission limitation or smooth discon- nection at cut-of. In addition, further validation work would be needed to get a better insight in the performance and weaknesses of the proposed method. A generalisation of our approach to encompass stall regulated turbines and vertical axis turbines may be also considered in the future. This model offers alot of flexibility and can therefore be used in, simulation ofthe wind power production in energy mix analysis, I ‘maybe now be interesting to evaluate the impact ofthe different sizing parameter on the anual energy yield of a wind turbine but also to evaluate the expected yield of future wind turbines. Finally, to assist with the use of the parameterised model, we have developed implementations of the model able to generate power curves in MATLAB, R and Python, which can be found as supplementary material of this paper (Hips//eithub.com) ‘YwesMSaintDrenan/WT.PowerCurveModel) Declaration of competing interest ‘The authors declare that they have no known competing financial interests or personal relationships that could have Appeared to influence the work reported in this papet. ‘CRediT authorship contribution statement Yes-Marie Saint-Drenan: Writing - original draft, Sofware, Methodology, Conceptualization. Romain Besseau: Writing - original draft, Software, Methodology, Conceptualization. Malte Jansen: Conceptualization, Writing - review & editing, Software Iain Staffell: Conceptualization, Writing - review & editing, Soft- ware. Alberto Troccoli: Conceptualization, Writing - review & editing. Laurent Dubus: Conceptualization, Writing ~ review & editing. Johannes Schmidt: Software. Katharina Gruber: Software.{YeAL Sint Dena ta Renee Energy 157 (2020) 74-768 1 Sofa G. Simaes: Project administration, We Siegfried Hei ing - review & editing Methodology, Conceptualization. Acknowledgements The author would like to acknowledge the thewindpowernet [44] team for the compilation and regularly update of their wind turbine and power curve database. This work has been partly conducted in the framework of the Copernicus C3S energy and ERANET CLIMZPOWER projects. Copernicus Climate Change Service (C38) is a programme being implemented by the European Centre or Medium-Range Weather Forecasts (ECMWF) on behalf of the European Commission (contract _number: 2018/C35~ 426 — Lot ~ WEMC). The project CLIM2POWER is part of ERAACS, an ERA-NET initiated by JPI Climate, and funded by FORMAS (SE), [BMBF (DE), BMWEW (AT), FCT (PT), EPA (IE), ANR (FR) with co- funding by the European Union (Grant 690462). Malte Jansen and Technology through the strategie project UIDJAMB/04085/2013. Appendix A. Parametric power coefficient models C3(7.) The power coefficient Cp expresses what fraction of the power in, the wind the wind turbine extracts. This quantity is generally assumed to be a function of both tip-speed ratio and blade pitci langle. The power coefficient can whether be evaluated experi ‘mentally oF calculated numerically using BEM, CFD or GDW ‘models. A convenient alternative consists in using: numerical ap- pproximations, and a few empirical relations can be found in the literature. These expressions can be formulated with the general form below: Cpl,B) = €1(€2/ C58 — Cai — C58" —eG)e 7 +C5h Ip! = (2+ €98)"" e106? +1), : Jain Staffel! were funded by the Engineering and Physical Sciences, (aly Research Council through the IDLES programme (EP/RO45518)1). CCENSE is funded by the Portuguese Foundation for Science and ‘able a2 Conicents ofthe iret parameteration of found nthe Meraare (Sones ct al, 2001 2005) (Heer 2018) __(Thongamea, 2009) (De Kooning eta, 2015) (Ochi etal, 2014) (Dale, 2014, aan 05 05176 om om fom ous toss o Sas ous Blade pitch angle: 0° ° Power coefficient C, Blade pitch angle: 1° Power coofficent C, Tip speed ratios Blade pitch angle: 3° Tip speed ratio A Blade pitch angle: 5° ame Power coefficient ©, Tip speed ratio s Tip speed ratio A Fg 16. Comparison othe diferent Cy models ound in the erature for Bade pch angle values af 0,1 3a 578 YM Saint rena et a Renwable Energy 157 (2020) 754-768 References 11] A Albers Turbulence and shear normalisation of wind turbine power curve, {European Wind Energy Conference and Exhibition 2010, EWEC 2010, vl 62010, pp. 4116-4123 [2] ANt Aves, Demartino, F. Ricard, Asessment of the peak response of S-EMW HAW under combaned wind and ssamic Induced loa Open Gonstrce Build Tecpol 111) 2017) 44h 457, hp bentamnopenceny) Fuutexrjroscrpiati [5] Lt Beda LA Sean, nuence of turbulence intensity on wind urine ower curves in: Energy Procedia, vol 137, 2017, p. 553-58 [4] Becker D Tran, Compleon of wind trite dts co wind integration "ties applying random forests and kenearest neighbors, App Ene 208 zany PSE aon jnovocencedccoracrc a [5] J. Bosch Stel, AD. Hawkes, Temporally explicit and spatially resolved ‘lal ofshore wind energy penta ergy 165 (2018) 75-781. Np) fron stenedetcom scence ale pls 9505442 18318 [6] Crown fst Veriton of Wind Turbine Powter Carve" Summary of Projet Ress, Masters hes, Techn University of Denmark, Denar, Doi. 17] F campagnolo, V. Petrovic. Wind tunnel testing of poster maximiaton onitlstateges applied tomlin loting wi power platform frceedings af the 2h lnternadonal Gcean aod Polar Engineting Cone nce, 2016 op, 309-316, {Gi ice 6140-121, 602, Wind Turbines — Part 12-1: Power Peformance Measurements of Electricity Producing Wind Turbines Sanda eration letfotceeal Commision, 2015. fo] Aiton, Wagner. counting forthe eft of turbulence on wind urine ower cuves, in: Joural of Phys: Conference Sere, vol 524 2014 WO] Dat. LiL Wen. Long, Research on power coeicient of wind turbines ‘sed on SCADA data, Rene. Energy 86 (2016) 206-215 [nn] J:Bal W. Yang) Ca, DL. tong. Ageing assessment of wind tubing ‘ver ume by erecting wad arn Sada aia. Rene Eetay 118 (2018), 195-26 atime montis rans and esi cont fa id ‘ystems, hit jwarwcencelrectcom}cencelatiet ‘Sto a8n7a0289b [12] JCDay, Hu, DS. Lis, We. Modeling and analysis of deceive p= ‘aneat magnet symdbronous generator wind tune based wind toe eral network made, Proc: DME} Power Energy 226 (1) 2012) 62-72. 113) [DM De Kooning Gevaert} Van De Yyver, TL Vandoorn, - Vacevele, ‘Online estimation ofthe power coeficient versus tip-sped ratio curve of ‘Wind arbines im IECON Proceedings (Industral Hectic Conference) 2013, pp. 1782— 1797 [ns] D-tiot, |. Caogan tects of win shear and urate on ind tusbine Power curves, Wind Energy 1 (1090) 10-14 tassels) wien cont-10E [ns] [PSD cards, Wind Turbine Database: Modeling and Analysis with Focus ‘on Upsling, Masters these, Chalmers university of tchnoogy eteborg, 1061 L Gonzalez Aparicio, A. Zucker. F, Caren. F. Monfrt . Hu. Badger. Exes dataset Pat wind power generation European Meeorloge evved Hig resolidon RES generation time seis for present 3nd UE Scenarios pee suropaet, 2016 [ua] F Gonzalez Longa, P. Wal Tea, Wake effect tn wind fam porto rane steady-state an dynamie havior, Renew. Eneay 291) (2042) 29-338, Taiplwaresciencetret eomiscenee ariel) ‘hoo 48111005155. [n8) | Goaschall J eink, How to improve te estimation of power curves for ‘wind erie, Enron. Res et 3 (1) (2008 Lig) S."Heer, "Cid tegration of wind energy. ps:idlory/10.1002) S7aitts70s274, "2014 haps search ctosefongPa-Helert2Cs SE 2C 2014 Gr iteration sof wind ener. (20) Iec, ec 61400-1212 Power Performance Messurements of Elcrcy Po- cing Wind Turbines 2005. [at] Svanll Mielsen, | Sorensen, K Hansen D,Hensiagsn, The impuc of ‘wind direction in atmosphere bon interacting wakes athens ev wid fats, In: Torque 2010 2010, pp. 407-426. [22] Jens iladse, Jon. Kebberup, Ret Madsen. Thomas Jacobsen, Morten Lager, Ads. Sorensen Lasse Svenningsen, Mauro Mot, Karas Breda, Rabin Fak, Sean Chin eter Site, The Windpre Ml Editon 27, Ea International A's 2010 iw re feet) manus printMANUAL 27 pa. 123) Kc taser, W.Langreder, H.Hohlen, J, Hestrup, Turbulence eoreton For ower carves, Wind Energy 2007; pp. 139-162, Fray MUI tem, EE Bachyns, Aaa, Ets of hyeodynamie modeling in ts Fully coupled simlatons of a semi-submersible wind turbine in: Enemy Proved ol 24,201, pp 1-362. [a5] N to, Pujol Pacem Gonzaer, | Bramoa, A. Masaguer, Development Sf soull'scale wind energy stems aaplable fo eimateconditons using ‘catering torque contol” Pitch contol and CAES strategy. in ner Tonal Conference on Renewable Energies and Power Quality (CREFO?) Nalaga (Spain, 2017, p. 494-09. 25) Mc ty, SS. Kumar Al Sehakimar, GEP. Kumat A comprehensive ceview (wid turbine poner curve modeling techasques, Renew Sasa. Eney few. 30 (2014) 452-460. pw scieneedectcomjalence cep ‘4038113007296, 127) MD. Mors, Factor sampling plans for prchiminary computational exper tents, Technometis 33 2) (1991) 161-174 aay Pit Novgau,H Holtinen, A Mul-Trbine Power Curve Approach, 200, [23] Po. Oehiens AW. Manjonge, AD. Oar, Mathematical analy a ip speed ‘ato of wind turbine andi eet on power cofcen Int. Math Sat Compu4(1) 2014) et. hep imsecomndex ppnow 20. [30] Ere. Fea Power curve Working Croup. 2015, [BI] 5. eninge, Decaras, L Heh, Quoi, L Stal, The importance of ‘pen data and software: energy research aging behind? Energy Pl. 101 Gorn) “aitezis hapa iw seteneeireceamscenetartepi [32] 5 Pennnger 1 Stal, Long term pattems of european py ouput using 30 Yeas of validated houilyveabalyst and stlite data, Energy 114 (2016) 13511265 Ttpwwrscienedinet coms aricepi) ‘0s442i6311744, 133] E Rateshide, A Tindal C Johnson. A. Graves E Simpson, Bless . Haris D. Schoborg: fects of complex wind regnes on turbine etfrmance. i ‘Anerkan Wind Eneray Association WINDFOWER Conerence No. May, 200, pret ts. [bal L SthallenberrRodriguce, A methodological review to estimate techno. ‘conor wind energy production. Renew. Susan, Eeruy Rev. 21 (2013) Sees bibiimocenceeonacec icp 135) D.Shin Ka Comparative anal of degradation rates for inland and seaside wind trbines in compinice with the international lectrtecical commision standard Energy 118" (2017) 1180-1186. "pv Sehencerez comscencarclepiS03605442163 15809. 136) Dshin. ic Ka, Application of the nacelle taster function by 2 acl ‘ounte ight detection snd ranging system to wind turbine power pelo mance measurement, regis 12 (0) (2019) 137) | Sooeweg Ponder, W: Kling Dynamic modelling of wind turbine with ‘oubly ed inton generator in: 3001 Power EngincenngSocety Summer Meesng conference Proceedings (Cat NOOICHS/262) lI. 201 pp 644-640, hp iccupure see agldocament 970114 130) JE Sootwes, SW de Haan 1 Paine, WIL King, General mode for representing varable spect! wind rirbine in power system dynamics sim Tats Power Stems IEEE Transactions on 18 (1) (203) 144-151 139) Lat Sobol Senet estimates fr nonnea msthemtia models, athe Iti! Modeling snl Computatanal Expenment 1 (4 (1085) 407-44 1401 V'Sohont, Scopes, RK. Nema Areal review on wind trie power curve modeling techniques and ther apptiations In wind based enerey systems Energy 2018) 1-18, 2016, [at) stall Green, How does wind frm performance decline with age? Renew. Boegy 66 (20¢4) 775-786 iw scence comsciene] arepivs09601481 13005727. 12] Stal 5 reninges. Using bias corrected reanalysis to simulate current Sd future wind power output Enery 114 (2016) 1224-1239 143) Sumner hss, influence of stospherc stably on wind turblae po ‘wer performance curves, j Sl Energy En 128 (4) (200) 331 |) thewindpowernt, tewindpowetstet" Daubase. 2018, pew. thewindboweriet 145] 15 Thongam, P. Bouchard H. zzad, M. Ouhrouche, Wind speed sensorless Inaximun power oun uaeking conta of variable speed wand eeray eon ‘erin systems in lc [Mach (Des) (Co) 2008 {EMDC) 8. (EE) Ut) 209, pp 1832-1837 16) Tan 0 ou, C Si M, Stan, 2. Chen labor, Wind turbine power uve design for opal power generation in wid arms considering wake ‘tec, Energies 103) (2017) 1-19. a7) Re Wagner. Antoniou, SM. Pedersen, MS. Courtney, HE jargensen, Profile i tne peoance messes Wind Enay 1 (28) 148) R Wagner M. Courtney, TJ Larsen, US. Paulsen, Simulation of Sear and Turbines Impact on Wind Turbine Perfomance 2010. Rise R172, a9) S Wharton, [ke Lundguist Atmospheric sally flees wind turbine power ‘lection, Environ ett 7 (1) (2012 150] Ay. Wood, BF Wolenborg. Power Cencation Operation and Cont, joa ‘ad Sons, USA. 1996, Chap 5