energies

Article Analysis of Aging through Operation Data Calibrated by LiDAR Measurement

Hyun-Goo Kim * and Jin-Young Kim

Korea Institute of Energy Research, Daejeon 34129, Korea; [email protected] * Correspondence: [email protected]; Tel.: +82-42-860-3376

Abstract: This study analyzed the performance decline of wind turbine with age using the SCADA (Supervisory Control And Data Acquisition) data and the short-term in situ LiDAR (Light Detection and Ranging) measurements taken at the Shinan located on the coast of Bigeumdo Island in the southwestern sea of South Korea. Existing methods have generally attempted to estimate performance aging through long-term trend analysis of a normalized in which wind speed variability is calibrated. However, this study proposes a new method using SCADA data for wind farms whose total operation period is short (less than a decade). That is, the trend of power output deficit between predicted and actual power generation was analyzed in order to estimate performance aging, wherein a theoretically predicted level of power generation was calculated by substituting a free stream wind speed projecting to a wind turbine into its power curve. To calibrate a distorted wind speed measurement in a nacelle anemometer caused by the wake effect resulting from the rotation of wind-turbine blades and the shape of the nacelle, the free stream wind speed was measured using LiDAR remote sensing as the reference data; and the nacelle transfer function,   which converts nacelle wind speed into free stream wind speed, was derived. A four-year analysis of the Shinan wind farm showed that the rate of performance aging of the wind turbines was estimated Citation: Kim, H.-G.; Kim, J.-Y. to be −0.52%p/year. Analysis of Wind Turbine Aging through Operation Data Calibrated Keywords: performance aging; normalized capacity factor; nacelle transfer function (NTF); LiDAR; by LiDAR Measurement. Energies 2021, 14, 2319. https://doi.org/ Shinan wind farm 10.3390/en14082319

Academic Editors: Davide Astolfi and Francesco Castellani 1. Introduction According to IEC 61400-1, which explains the design condition of wind turbines, the Received: 12 March 2021 design life-span of wind turbines should be more than twenty years [1]. Since a wind Accepted: 17 April 2021 turbine is a machine that operates continuously under repetitive fatigue load conditions, Published: 20 April 2021 its performance inevitably deteriorates as a result of aging during its twenty-year life cycle (blade erosion, efficiency reduction of gearbox, bearing, generator, etc.) [2]. Publisher’s Note: MDPI stays neutral The results of a comprehensive analysis of 5600 wind turbines in the UK and 7600 wind with regard to jurisdictional claims in turbines in Denmark [3] verified that performance due to aging deteriorated to a significant published maps and institutional affil- extent. Nonetheless, performance aging has not been taken into consideration during iations. evaluations of the economic feasibility of wind farm projects due to a lack of reliable estimation data. According to the results of another study on thirty years of operational records of 3200 wind turbines in Denmark [4], the rate of reduction of their capacity was calculated Copyright: © 2021 by the authors. to be −0.32%p (percent point) per year, but the report did not specify whether this rate Licensee MDPI, Basel, Switzerland. was due purely to a deterioration of the wind turbines or to variations in the annual wind This article is an open access article resources. The results of an analysis of wind farms located in Ontario, Canada, for five distributed under the terms and years [5] showed that if the annual wind speed change is calibrated every year, the rate conditions of the Creative Commons of reduction of their capacity was −1.0%p per year, which is a much higher rate than Attribution (CC BY) license (https:// that reported in Denmark. The estimation by Hughes (2012) [3] and that by Staffell and creativecommons.org/licenses/by/ Green (2014) [6], who calibrated wind resource variability using MERRA (Modern-Era 4.0/).

Energies 2021, 14, 2319. https://doi.org/10.3390/en14082319 https://www.mdpi.com/journal/energies Energies 2021, 14, x FOR PEER REVIEW 2 of 12

Energies 2021, 14, 2319 2 of 12

(2014) [6], who calibrated wind resource variability using MERRA (Modern-Era Retro- Retrospectivespective analysis analysis for Research for Research and Applications) and Applications) reanalysis reanalysis data, showed data, showed that the that normal- the normalizedized load factor load (or factor capacity (or capacity factor) factor)in onshore in onshore wind farms wind in farms the UK in thehad UK fallen had steadily fallen steadilyby −0.48%p by − per0.48%p year for per a year decade. for aHere, decade. the normalized Here, the normalized load factor load is the factor total ispower the total out- powerput over output a certain over period a certain of periodtime divided of time by divided the maximum by the maximumpotential power potential output power ad- outputjusted for adjusted monthly for variations monthly variations of wind speed. of wind speed. ByrneByrne etet al. (2020) (2020) [7] [7 ]and and Astolfi Astolfi et etal. al.(2020) (2020) [8] demonstrated [8] demonstrated the wind the wind turbine turbine aging agingof of Vestas V52 using V52 usingthe support the support vector vector machin machinee regression regression of operation of operation curves, curves, and found and foundthat the that performance the performance decline decline over overten years ten years was about was about 5%. Hamilton 5%. Hamilton et al. et (2020) al. (2020) [9] ana- [9] analyzedlyzed fleet-wide fleet-wide performance performance with with age age with with 917 917 wind wind farms farms installed installed before 20082008 inin thethe USUS andand showedshowed thethe performanceperformance decline decline of of− −0.54%p per year. InIn SouthSouth Korea, Korea, the the rate rate of of deterioration deterioration in in the the capacity capacity of of Sungsan Sungsan wind wind farm farm in in Jeju Jeju waswas reportedreported toto be be− −0.12%p0.12%p perper yearyear forfor fivefive yearsyears [[10],10], butbut thethe reliabilityreliability ofof thethe resultresult is is suspicioussuspicious becausebecause itit waswas obtainedobtained byby aa trendtrend analysisanalysis thatthat usedused onlyonly fivefive values values for for the the annualannual averageaverage capacitycapacity factor.factor. MostMost previousprevious studies studies have have estimated estimated the the deterioration deterioration of wind of wind farm farm capacity capacity by con- by ductingconducting a trend a trend analysis analysis of the of normalized the normaliz capacityed capacity factor, factor, which waswhich calibrated was calibrated according ac- tocording monthly to windmonthly speed wind variability. speed va However,riability. However, according accordin to one analysisg to one of analysis wind resource of wind variabilityresource variability off the western off the coast western of South coast Korea of South [11 ],Korea the uncertainty [11], the uncertainty of the yearly of the capacity yearly factorcapacity decreased factor decreased to 0.7% when to 0.7% the when analysis the an periodalysis was period set was to ten set years to ten or years longer, or longer, while thewhile monthly the monthly capacity capacity factor fell factor to 2.7% fell whento 2.7% the when analysis the period analysis was period set to 36was months set to or36 longer.months This or longer. implies This that, implies since the that, uncertainty since the of uncertainty the capacity of factor the capacity due to wind factor resource due to variabilitywind resource is comparable variability or is higher comparable than that or higher of the windthan that turbine of the deterioration wind turbine rate, deteriora- a long- termtion trendrate, a analysis long-term of moretrend than analysis ten years of more would than be ten needed years to would obtain be a reliableneeded qualitativeto obtain a estimatereliable qualitative of the rate ofestimate deterioration of the rate by a of monthly deterioration trend analysis.by a monthly trend analysis. ForFor windwind farmsfarms whosewhose totaltotal operatingoperating periodperiod isis short, short, any any estimate estimate of of performance performance agingaging would be be unreliable unreliable since since the the uncertainty uncertainty of wind of wind resource resource variability variability is more is more dom- dominantinant than than the wind the wind farm farmdeterioration deterioration rate. The rate. cumulative The cumulative installation installation capacity capacity of MW- ofclass MW-class wind turbines wind turbines in South in Korea South is Korea1512 MW is 1512 (as of MW the (asend of of the2019), end and of 2019),the cumulative and the cumulativenumber of installations number of installationsis over 600 turbines, is over 600around turbines, half of around which halfhave of a whichtotal operating have a totalperiod operating of less than period five of years less (Figure than five 1) years[12]. Thus, (Figure even1)[ if12 the]. Thus,monthly even wind if the speed monthly varia- windtion is speed calibrated variation in the is case calibrated of wind in turbines the case in of South wind Korea, turbines few in of South the available Korea, few cumu- of thelative available operational cumulative data cover operational a long enough data cover peri aod long to enable enough a reliable period toestimate enable of a reliable the rate estimateof wind ofturbine the rate deterioration of wind turbine using deterioration only the normalized using only capacity the normalized factor. capacity factor.

FigureFigure 1.1.Total Total wind wind turbine turbine installation installation capacity capacity in in South South Korea. Korea.

ThisThis studystudy aimsaims toto proposepropose aa newnew methodmethod ofof calculating calculating purelypurely mechanicalmechanical windwind turbineturbine performance performance aging aging rate rate using using SCADA SCADA data data whichwhich isis thethe rawestrawest windwind turbineturbine data.data. InIn other other words, words, a theoreticala theoretical prediction prediction of power of power generation generation is calculated is calculated through through the wind the turbine power curve based on the free stream wind speed, which is incident to a wind

Energies 2021, 14, x FOR PEER REVIEW 3 of 12

Energies 2021, 14, 2319 3 of 12

wind turbine power curve based on the free stream wind speed, which is incident to a wind turbine, and is then compared with the actual power generation data contained in theturbine, SCADA and record. is then compared with the actual power generation data contained in the SCADAThe record.deficit in power output between the predicted power generation and the actual The deficit in power output between the predicted power generation and the actual power generation is attributable to turbulence effects due to wake, site conditions such as power generation is attributable to turbulence effects due to wake, site conditions such wind shear caused by climate or terrain, operating effects such as the inflow angle and as wind shear caused by climate or terrain, operating effects such as the inflow angle and yaw misalignment, and the mechanical aging of the wind turbines among other factors yaw misalignment, and the mechanical aging of the wind turbines among other factors [6]. [6]. Although most of these are caused by stochastic effects, the dominant long-term cu- Although most of these are caused by stochastic effects, the dominant long-term cumulative mulative factor is wind turbine aging. Thus, the rate of mechanical deterioration of wind factor is wind turbine aging. Thus, the rate of mechanical deterioration of wind turbines turbines can be estimated by conducting a trend analysis of the power output deficit. can be estimated by conducting a trend analysis of the power output deficit. In addition, a LiDAR, a ground-based remote sensing device, was deployed and a In addition, a LiDAR, a ground-based remote sensing device, was deployed and a short-term campaign was conducted to accurately measure the free stream wind speed short-term campaign was conducted to accurately measure the free stream wind speed toward the wind turbines. The accuracy of the predicted power output calculation can be toward the wind turbines. The accuracy of the predicted power output calculation can be improved by calibrating the nacelle wind speed to the free stream wind speed. To that improved by calibrating the nacelle wind speed to the free stream wind speed. To that end,end, a a correlation correlation equation equation was was derived derived between between the the free free stream stream wind wind speed speed measured measured at at thethe hub hub height height of of a a wind wind turbine turbine by by LiDAR, LiDAR, and and the the nacelle nacelle wind wind speed speed measured measured by by the the nacellenacelle anemometer mountedmounted in in the the wind wind turbine. turbine. The The results results from from applying applying the proposedthe pro- posedmethod method to the Shinanto the Shinan wind farm, wind which farm, has which been has in commercial been in commercial operation sinceoperation 2009, since were 2009,compared were withcompared those with obtained those using obtained the conventional using the conventional method. method.

2.2. Analysis Analysis Data Data 2.1.2.1. Shinan Shinan Wind Wind Farm Farm TheThe Shinan Shinan wind wind farm farm is a is small-scale a small-scale wind wind farm farm (3 MW) (3 MW) that was that built was on built the onsandy the northernsandy northern shore shoreof Bigeumdo of Bigeumdo Island, Island, locate locatedd in the in thesouthwestern southwestern sea sea of of South South Korea Korea ◦ 0 00 ◦ 0 00 (34°46(34 46′33.833.8″ NN 125°56 125 56′15.315.3″ E).E). In In the the Shinan Shinan wind wind farm, farm, three three IEC IEC Class Class IIA IIA 1 1 MW MW wind wind turbinesturbines (Mitsubishi (Mitsubishi MWT-1000A MWT-1000A of of Japan), Japan), each each with with a cut-in a cut-in wind wind speed speed of 3 of m/s, 3 m/s, a hub a heighthub height of 69 ofm, 69and m, a androtor a rotordiameter diameter of 61.4 of m, 61.4 are m, arranged are arranged in line inin linethe east-west in the east-west direc- tiondirection (Figure (Figure 2). This2). study This study analyzed analyzed four years four years of SCADA of SCADA data datataken taken from fromFebruary February 2009 (i.e.,2009 the (i.e., farm’s the farm’s start of start commercial of commercial power powergeneration) generation) to April to 2013 April such 2013 as such wind as speed, wind windspeed, direction, wind direction, power poweroutput, output, yaw misalignme yaw misalignment,nt, ambient ambient temperature, temperature, etc. During etc. During the operationthe operation period, period, there there was was no nopower power curtailm curtailment,ent, and and the the Weibull Weibull distribution distribution of of wind wind speedspeed measured measured at at the nacelle-mounted anemomet anemometerer was the scale factorfactor ofof cc == 6.776.77 m/sm/s andand the the shape shape factor factor of of kk == 1.66. 1.66.

FigureFigure 2. 2. LocationLocation ( (leftleft)) and and layout layout ( (rightright)) of of the the Shinan Shinan wind wind farm farm (the (the wind wind turbines turbines and and the the LiDAR). LiDAR).

2.2.2.2. LiDAR LiDAR Measurement Measurement Campaign Campaign TheThe remote remote sensing sensing campaign campaign was was conduct conducteded from from November November 2009 2009 to toMarch March 2010 2010 (4 (4 months) using the Leosphere WindCube LiDAR which uses a pulsed erbium-doped fiber months) using the Leosphere WindCube LiDAR which uses a pulsed erbium-doped fiber laser at 1.54 µm wavelength and heterodyne detection. The vertical measurement heights laser at 1.54 μm wavelength and heterodyne detection. The vertical measurement heights were designed to include the wind turbine rotor area and the hub height (40 m, 70 m, 100 m, 130 m, ... above ground level). The sampling rate of SCADA and LiDAR is 1 Hz

Energies 2021, 14, x FOR PEER REVIEW 4 of 12

Energies 2021, 14, x FOR PEER REVIEW 4 of 12 Energies 2021, 14, 2319 4 of 12

were designed to include the wind turbine rotor area and the hub height (40 m, 70 m, 100 werem, 130 designed m, ... above to include ground the level). wind Theturbine sampling rotor area rate andof SCADA the hub and height LiDAR (40 m, is 701 Hz m, and100 m,andthe 130 data the m, datawere ... above were converted converted ground into level). intoa 10 amin The 10 minaverage. sampling average. The rate Theaccuracy of accuracySCADA of the ofand WindCube the LiDAR WindCube is LiDAR 1 Hz LiDAR and has thehasbeen data been verified were verified convertedthrough through ainto comparative a comparativea 10 min average. verifi verificationcation The accuracywith with SODAR SODAR of the WindCube(SOnic (SOnic Detection LiDAR And has beenRanging) verified underunder through variousvarious a terrains terrainscomparative such such as asverifi plain, plain,cation hilly hilly terrain,with terrain, SODAR and and urban urban(SOnic [13]. [13]. ADetection measurement A measure- And Ranging)algorithmment algorithm under of the variousof LiDAR the LiDAR andterrains a correctionand such a correction as methodplain, hillymethod in complexterrain, in complex and terrain urban terrain using [13]. computationalusing A measure- compu- mentfluidtational dynamicsalgorithm fluid dynamics (CFD)of the areLiDAR (CFD) well andare described wella correction described in Kim method and in Kim Meissner in and complex Meissner (2017) terrain [14 (2017)]. using [14]. compu- tationalThe fluid LiDAR LiDAR dynamics was was deployed deployed (CFD) are between between well described wind wind turbines turbinesin Kim #2 and and #2 Meissner and #3 at #3 the at (2017)Shinan the Shinan [14]. wind windfarm farmin orderThe in order LiDARto measure to was measure deployedthe sea the breeze, sea between breeze, which wind which is turbines a ismain a main #2wind, and wind, without#3 at without the interruption Shinan interruption wind of farm the of intheterrain order terrain features to featuresmeasure (Figure (Figurethe 2). sea Due2 ).breeze, Due to the to which conica the conical isl scanninga main scanning wind, setup setup withoutof WindCube of WindCube interruption [14], [the14 of ],scan- the terrainscanningning plane features plane at the at(Figure thehub hub height 2). height Due of to the of the the wind conica wind turbine turbinel scanning (the (the yellowsetup yellow of circle circleWindCube in in Figure Figure [14], 22)) mightthemight scan- bebe interfered partiallypartially byby the the blade blade rotation rotation (the (the white white circle circle of #2).of #2). To preventTo prevent this, this, the LiDARthe Li- ning plane at◦ the hub height of the wind turbine (the yellow circle in Figure 2) might be interferedisDAR rotated is rotated 35partiallyclockwise 35° clockwiseby the so thatblade so the that rotation scanning the scanni (the points whiteng points (four circle yellow(four of #2). yellow dots) To dopreventdots) not do overlap this, not overlapthe in theLi- DARrotorin the is plane. rotor rotated plane. 35° clockwise so that the scanning points (four yellow dots) do not overlap in theThe rotor comparison plane. of wind speed at the hubhub heightheight betweenbetween thethe SCADASCADA andand LiDARLiDAR data is depicted in Figure3. The frequency distribution of wind direction measured by data Theis depicted comparison in Figure of wind 3. The speed frequency at the hu distributionb height between of wind the direction SCADA measured and LiDAR by LiDAR during the campaign period revealed that the excellent north-northwest sea breeze dataLiDAR is depictedduring the in campaign Figure 3. periodThe frequency revealed distribution that the excellent of wind north- directionnorthwest measured sea breeze by with the mean wind speed of 10 m/s is the main wind as shown in Figure4a. Figure4b LiDARwith the during mean the wind campaign speed ofperiod 10 m/s revealed is the mainthat the wind excellent as shown north- innorthwest Figure 4a. sea Figure breeze 4b shows the wind rose of the third-generation reanalysis data, MERRA-2 at the same period withshows the the mean wind wind rose ofspeed the third-generationof 10 m/s is the mainreanalysis wind data, as shown MERRA-2 in Figure at the 4a. same Figure period 4b of the campaign. They share very similar characteristics with the LiDAR measurements. showsof the thecampaign. wind rose They of theshare third-generation very similar ch reanalysisaracteristics data, with MERRA-2 the LiDAR at the measurements. same period Figure4c shows the wind rose of MERRA-2 for the last ten years, in which southeasterly ofFigure the campaign. 4c shows theThey wind share rose very of MERRA-2similar ch aracteristicsfor the last tenwith years, the LiDARin which measurements. southeasterly winds blew in summer but north-northwesterly winds were dominant throughout the year, Figurewinds 4cblew shows in summer the wind but rose north-northweste of MERRA-2 forrly the winds last ten were years, dominant in which throughout southeasterly the and the frequencies of easterly and westerly winds where the wake effect occurred were windsyear, and blew the in frequencies summer but of north-northwesteeasterly and westerlyrly winds winds were where dominant the wake throughout effect occurred the very few. year,were andvery the few. frequencies of easterly and westerly winds where the wake effect occurred were very few.

Figure 3. Comparison of the daily-mean wind speed betweenbetween SCADA andand LiDARLiDAR datadata atat thethe hubhub height.height. Figure 3. Comparison of the daily-mean wind speed between SCADA and LiDAR data at the hub height.

(a) (b) (c) Figure 4. Wind frequency(a) roses at the Shinan wind farm from(b) 30 November 2009 to 17 March 2010.( c()a ) LiDAR measure- ment (dashed line: mean wind speed at 70 m above ground); (b) MERRA-2; (c) MERRA-2 over a period of 10 years. Figure 4. Wind frequency rosesroses atat the the Shinan Shinan wind wind farm farm from from 30 30 November November 2009 2009 to 17to March17 March 2010. 2010. (a) LiDAR(a) LiDAR measurement measure- ment (dashed line: mean wind speed at 70 m above ground); (b) MERRA-2; (c) MERRA-2 over a period of 10 years. (dashed line: mean wind speed at 70 m above ground); (b) MERRA-2; (c) MERRA-2 over a period of 10 years.

Energies 2021, 14, x FOR PEER REVIEW 5 of 12

For reference, MERRA-2 refers to third-generation reanalysis data (1 h average) that extensively assimilates satellite-based remote sensing data, showing a highly accurate es- timate for the southwest offshore of South Korea [15]. The MERRA-2 and LiDAR meas- urements showed a high correlation of the coefficient of determination, R2 = 0.78 and 0.95 for wind speed and wind direction respectively.

3. Analysis Methods

Energies 2021, 14, 2319 The method of calculating the rate of deterioration of the capacity factor from5 of 12the SCADA data of the wind farm and sort-term LiDAR measurements is presented in Figure 5 in the form of a schematic diagram. First, the free stream wind speed was measured through the short-term LiDAR meas- urementFor reference, campaign, MERRA-2 and the refersnacelle to transfer third-generation function (NTF; reanalysis the correlation data (1 h average)between thatwind extensivelyspeeds measured assimilates at the satellite-based upstream of remotewind tu sensingrbine using data, a showingremote sensing a highly device accurate and a estimatenacelle foranemometer the southwest on a offshorewind turbine) of South was Korea derived [15]. by The analyzing MERRA-2 the and correlation LiDAR mea- with 2 surementsregard to showed the nacelle a high wind correlation speed of ofthe the SCADA coefficient data. of determination, R = 0.78 and 0.95 for windSecond, speed the and nacelle wind directionwind speed respectively. of the SCADA was calibrated to the free stream wind speed using the derived NTF, and the calibrated speed was substituted into the wind tur- 3. Analysis Methods bine power curve to calculate the predicted power output. TheThird, method the difference of calculating between the ratethe actual of deterioration power output of the and capacity the predicted factor power from the out- SCADAput was data calculated, of the wind and farm the andrate sort-termof deterioration LiDAR of measurements the performance is presented of the wind in Figure turbine5 inwas the formestimated of a schematicby a monthly diagram. trend analysis.

FigureFigure 5. 5.Schematic Schematic diagram diagram of of analysis analysis of of rate rate of of deterioration deterioration of of performance performance of of a winda wind turbine. turbine.

3.1.First, Derivation the free of the stream Nacelle wind Transfer speed Function was measured through the short-term LiDAR mea- surementSince campaign, a nacelle andanemometer the nacelle is transferinstalled function in the rear (NTF; side the of correlationthe blade roots between on the wind wind speedsturbine’s measured nacelle, at itthe measures upstream wind of windspeedturbine that is deformed using a remote by the sensing nacelle devicegeometry and and a nacellethe wake anemometer caused byon the a blades’ wind turbine) rotation; was ther derivedefore, the by higher analyzing the wind the correlationspeed, the greater with regardthe deficit. to the In nacelle this regard, wind speed IEC 61400-12-2 of the SCADA [16], data.which recommends the guideline for wind Second, the nacelle wind speed of the SCADA was calibrated to the free stream wind turbine power performance, provides a rule on the creation of the NTF to calibrate the speed using the derived NTF, and the calibrated speed was substituted into the wind nacelle wind speed to the free stream wind speed. Kim et al. (2015) [17] verified various turbine power curve to calculate the predicted power output. types of NTFs according to the type of wind turbine, blades, nacelle geometry, and sensor Third, the difference between the actual power output and the predicted power output position via an analysis of the literature (Figure 2 of Kim et al., 2015 [17]). They defined was calculated, and the rate of deterioration of the performance of the wind turbine was estimated by a monthly trend analysis.

3.1. Derivation of the Nacelle Transfer Function Since a nacelle anemometer is installed in the rear side of the blade roots on the wind turbine’s nacelle, it measures wind speed that is deformed by the nacelle geometry and the wake caused by the blades’ rotation; therefore, the higher the wind speed, the greater the deficit. In this regard, IEC 61400-12-2 [16], which recommends the guideline for wind turbine power performance, provides a rule on the creation of the NTF to calibrate the nacelle wind speed to the free stream wind speed. Kim et al. (2015) [17] verified various types of NTFs according to the type of wind turbine, blades, nacelle geometry, and sensor position via an analysis of the literature (Figure2 of Kim et al., 2015 [ 17]). They defined the Energies 2021, 14, 2319 6 of 12 Energies 2021, 14, x FOR PEER REVIEW 6 of 12

NTF as a function of wind speed as well as turbulence intensity, thereby improving the functionthe NTF fitnessas a function [18]. of wind speed as well as turbulence intensity, thereby improving the functionAssuming fitness that the[18]. variations of wind flow passing through the wind turbine can be expressedAssuming with the that blades’ the variations aerodynamic of wind characteristics, flow passing that through is, thepower the wind coefficient turbine ( CcanP) orbe thrustexpressed coefficient with the (CT blades’). This studyaerodynamic conducted characteristics, a regression that analysis is, the of power the NTF coefficient using the (CCPP) -equationor thrust (Equationcoefficient (1)), (CT).C TThis-equation study conducted (Equation (2)), a regression and non-linear analysis function of the (NTFEquation using (3) the). NoteCP -equation that Equations (Equation (1) (1)), and C (2)T -equation are derived (Equation empirically (2)), [and19,20 non-linear]. function (Equation (3)). Note that Equations (1) and (2) are derived√ empirically [19,20].  3  a1 − CP Vcorrected,1 = Vnacelle + a3 (1) 𝑎a 𝐶 𝑉 2 𝑉 𝑎 (1) , 𝑎 q Vcorrected,2 = b1 (1 − CT) Vnacelle + b2 (2) 𝑉, 𝑏1𝐶 𝑉 𝑏 (2) V = c V3 + c V2 + c V + c (3) corrected𝑉,,3 𝑐1 nacelle𝑉 𝑐2 nacelle𝑉 𝑐3 nacelle𝑉 𝑐4 (3) where a , a , a , b , b , c , c , c , c are the fitting coefficients determined by the maximum where 1𝑎,𝑎2 ,𝑎3,𝑏1,𝑏2,𝑐1,𝑐2,𝑐,𝑐3 4 are the fitting coefficients determined by the maximum likelihood estimation. (a = 4.60, a2 = 3.84, a3 = 0.35, b = 1.06, b2 = 3.65, c = 0.0008, likelihood estimation. (𝑎1 = 4.60, 𝑎 = 3.84, 𝑎 = 0.35, 𝑏1 = 1.06, 𝑏 = 3.65, 𝑐1 = 0.0008, c2 = 0.0538, c3 = 0.3025, c4 = 2.8041). 𝑐 = 0.0538, 𝑐 = 0.3025, 𝑐 = 2.8041). TheThe powerpower coefficientcoefficient isis defineddefined asas thethe ratioratio betweenbetween rotorrotor powerpower andand freefree streamstream windwind power,power, which which refers refers to to the the fraction fraction of of wind power that thatis is absorbedabsorbed byby thethe rotor.rotor. TheThe thrustthrust coefficientcoefficient isis defineddefined as the the ratio ratio betw betweeneen the the thrust thrust force force and and the the dynamic dynamic force force of of free stream wind, which refers to the fraction of wind energy that is absorbed by the free stream wind, which refers to the fraction of wind energy that is absorbed by the rotor rotor [21]. The power and thrust coefficients of the Mitsubishi MWT-1000A wind turbine [21]. The power and thrust coefficients of the Mitsubishi MWT-1000A wind turbine are are shown in Figure6. shown in Figure 6.

FigureFigure 6.6. PowerPower curve,curve, powerpower andand thrustthrust coefficientcoefficientof ofthe theMitsubishi Mitsubishi MWT-1000A. MWT-1000A.

3.2.3.2. TrendTrend AnalysisAnalysis ofof Wind Wind Turbine Turbine Deterioration Deterioration TheThe differencedifference betweenbetween actualactual powerpower generationgenerationand and predictedpredicted power powergeneration generation can can bebe calculatedcalculated usingusing thethe followingfollowing sequence.sequence.

(1)(1) The nacelle wind wind speed speed (V (Vnacellenacelle) is) calibrated is calibrated to free to free stream stream wind wind speed speed (Vcorrected (Vcorrected) using) usingthe NTF the derived NTF derived in Section in Section 3.1. 3.1. (2)(2) The effectiveeffective windwind speedspeed ((VVeffeff) is calculatedcalculated toto compensatecompensate thethe effecteffect ofof airair densitydensity whenwhen thethe performanceperformance curve is applied as follows. In the equation below, below, ρ𝜌o refersrefers to the standard standard air air density density when when the the wind wind turbine turbine performance performance curve curve is derived. is derived. The ◦ Theseasonal seasonal temperature temperature change change in Korea in Korea is about is about26 °C, 26so accurateC, so accurate air density air densitycalcula- calculationtion is essential is essential for wind for resource . assessment. In this Instudy, this study,air density air density was calcu- was lated using SCADA data and the nearby weather station data (located 2.5 km behind) such as air temperature, air pressure, and humidity.

Energies 2021, 14, 2319 7 of 12

calculated using SCADA data and the nearby weather station data (located 2.5 km Energies 2021, 14, x FOR PEER REVIEW behind) such as air temperature, air pressure, and humidity. 7 of 12

1/3 Ve f f = Vcorrected(ρ/ρo) (4)

/ (3) The theoretically predicted power𝑉 𝑉 generation𝜌/𝜌 (Ppredicted ) is calculated by substituting(4) the effective wind speed (Veff) into the performance curve of the Mitsubishi MWT- (3) 1000AThe theoretically wind turbine predicted (Figure 6power). ‘Power generation output deficit’(Ppredicted) ( isdP calculated) refers to theby substituting difference the effective wind speed (Veff) into the performance curve of the Mitsubishi MWT- between predicted and actual (PSCADA) power generation in a 10 min average. 1000A wind turbine (Figure 6). ‘Power output deficit’ (dP) refers to the difference between predicted and actualdP (P=SCADAPpredicted) power− generationPSCADA in a 10 min average. (5) 𝑑𝑃 𝑃 𝑃 (5) (4) Statistical outliers that are not included within 95% of the statistical distribution (4) ofStatistical the 10 min outliers average that power are not output included deficit within were 95% eliminated. of the statistical Note that distribution the special of carethe 10 is min needed average when power treating output outliers deficit if were there eliminated. were power Note curtailments. that the special In this care study,is needed there when was notreating power outliers curtailment if there and were non-operating power curtailments. period (P InSCADA this study,≤ 0) were there ◦ ◦ excluded.was no power In addition, curtailment data other and non-operating than the 285–0 periodand 0–35 (PSCADAsectors ≤ 0) werewere alsoexcluded. excluded. FigureIn addition,7 shows data surface other than roughness the 285–0° length and distribution 0–35° sectors around were thealso LiDARexcluded. location Figure where the disturbed wind directions due to wind turbine tower shading, wake zone, 7 shows surface roughness length distribution around the LiDAR location where the dis- and topographic effect are identified [22]. The reason behind the selection of northerly turbed wind directions due to wind turbine tower shading, wake zone, and topographic winds (285–0◦ and 0–35◦) only was to single out only sea breeze cases that can exclude effect are identified [22]. The reason behind the selection of northerly winds (285–0° and influence factors due to wake and terrain features. The surface roughness length (z ) 0–35°) only was to single out only sea breeze cases that can exclude influence factors dueo was calculated by a linear least square method to fit a logarithmic wind speed profile to wake and terrain features. (Equation (6)) to measured wind speed data by LiDAR at 40 m, 70 m, and 100 m The surface roughness length (zo) was calculated by a linear least square method to heights above ground. In the equation below, u∗ and κ are the friction velocity and fit a logarithmic wind speed profile (Equation (6)) to measured wind speed data by LiDAR von Kármán constant, respectively. at 40 m, 70 m, and 100 m heights above ground. In the equation below, 𝑢∗ and κ are the   friction velocity and von Kármán constant, respectively.u∗ z V(z) = ln (6) κ𝑢∗ z𝑧o 𝑉𝑧 𝑙𝑛 (6) κ 𝑧 (5) Finally, the wind turbine deterioration rate was determined through the slope in the (5) analysisFinally, ofthe the wind monthly turbine average deterioration of 10 min rate averaged was determined power output through deficit the trend.slope in the analysis of the monthly average of 10 min averaged power output deficit trend.

FigureFigure 7. 7.Rose Rose diagram diagram of of the the surface surface roughness roughness length length around around the the LiDAR LiDAR location. location.

4.4. Results Results and and Discussion Discussion 4.1. Derivation of the Nacelle Transfer Function 4.1. Derivation of the Nacelle Transfer Function The curve fitting of the free stream wind speed measured a 10 min average at the hub The curve fitting of the free stream wind speed measured a 10 min average at the hub height of 69 m during the LiDAR campaign and the 10 min averaged nacelle wind speed washeight conducted of 69 m usingduring three the LiDAR NTFs, i.e.,campaign Equations and (1)–(3),the 10 min thereby averaged determining nacelle wind the fitting speed was conducted using three NTFs, i.e., Equations (1)–(3), thereby determining the fitting coefficients. The curve fitting results showed that the fittings (R2) of CP-NTF (Equation (1)), CT-NTF (Equation (2)), and non-linear NTF (Equation (3)) were 0.97, 0.99, and 0.97, respec- tively, and that CT-NTF had the best fitting result.

Energies 2021, 14, 2319 8 of 12

Energies 2021, 14, x FOR PEER REVIEW 2 8 of 12 coefficients. The curve fitting results showed that the fittings (R ) of CP-NTF (Equation (1)), CT-NTF (Equation (2)), and non-linear NTF (Equation (3)) were 0.97, 0.99, and 0.97, respec- tively, and that CT-NTF had the best fitting result. TheThe probabilityprobability distributiondistribution ofof thethe 1010 minmin averagedaveraged windwind speedspeed deficitdeficit accordingaccording toto thethe threethree NTFsNTFs isis shownshown inin FigureFigure8 ,8, where where the the distribution distribution of of the the wind wind speed speed deficit deficit in in CCTT-NTF-NTF waswas the the closest closest to to the the normal normal distribution, distribution, as wellas well as showingas showing the smallestthe smallest variance. vari- 2 C Consequently,ance. Consequently, R of RT2 -NTFof CT-NTF was the was smallest. the smallest. The windThe wind speed speed deficit deficit is the is differencethe differ- betweenence between the predicted the predicted free stream free stream wind speedwind speed calculated calculated by substituting by substituting the nacelle the nacelle wind speedwind intospeed the into NTF the and NTF the and actual the freeactual stream free windstream speed wind measured speed measured by LiDAR. by Finally,LiDAR. C -NTF was found to be the best fit function from the evaluation of the curve fit regression Finally,T CT-NTF was found to be the best fit function from the evaluation of the curve fit andregression the probability and the probability distribution distri of thebution deficit of (Figure the deficit8). (Figure 8).

FigureFigure 8.8. StatisticalStatistical distributionsdistributions ofof 1010 minmin averagedaveraged windwind speedspeed deficitsdeficits by by NTFs. NTFs.

4.2.4.2. AnalysisAnalysis ofof thethe WindWind TurbineTurbine Deterioration Deterioration Trend Trend

TheThe monthlymonthly average average of of normalized normalized capacity capacity factor factor (CF (𝐶𝐹m) is) calculatedis calculated by by the the following follow- equationing equation [5]. [5]. CFm CFm = 𝐶𝐹 (7) 𝐶𝐹 Pm/Pm (7) 𝑃⁄𝑃 where m stands for month (1, 2, ... , 12), P and P are the monthly average of wind power where m stands for month (1, 2, …, 12),m 𝑃 andm 𝑃 are the monthly average of wind for a long-term (over 10 years) period and a short-term (wind farm operation) period, power for a long-term (over 10 years) period and a short-term (wind farm operation) pe- respectively. In this study, wind power is calculated by hourly wind speed at the hub riod, respectively. In this study, wind1 power3 is calculated by hourly wind speed at the height from MERRA-2 data, i.e., P = ρV A (A: rotor area). hub height from MERRA-2 data, i.e., 2𝑃 𝜌𝑉𝐴 (A: rotor area). The rate of performance aging is +0.47%p per year when wind resource variability is notThe considered rate of performance and this positive aging rate is +0.47%p does not per make year sense. when However,wind resource the rate variability becomes is −not0.01%p considered per year and when this positive wind resource rate does variability not make is correctedsense. However, with 10 the year rate reanalysis becomes data.−0.01%p Moreover, per year thewhen rate wind becomes resource−0.11%p variability per is year corrected when with 35 year 10 year MERRA-2 reanalysis data data. is usedMoreover, as shown the rate in Figure becomes9. Consequently, −0.11%p per ityear is conjectured when 35 year that MERRA-2 the short-term data is capacity used as factorshown data in Figure would 9. not Consequently, be appropriate it is when conjectu estimatingred that the the performance short-term capacity aging rate factor of wind data turbines,would not even be ifappropriate the wind resource when estimating variability th weree performance to be corrected aging with rate the of highly wind accurateturbines, reanalysiseven if the data. wind The resource sinusoidal variability pattern were that to appears be corrected every with 12 months the highly in Figure accurate9 is duereanal- to theysis seasonal data. The variation sinusoidal of synoptic pattern that wind appear speedss every in Korea, 12 months i.e., the in strongest Figure 9 in is winter due to and the theseasonal weakest variation in summer. of synoptic Along with wind wind speeds speed in Korea, changes, i.e., diurnal the strongest variations in ofwinter atmospheric and the stabilityweakest byin seasonsummer. makes Along the with difference wind speed of power changes, output diurnal even larger. variations of atmospheric stability by season makes the difference of power output even larger. Figure 10 is a graph representing the monthly trend analysis of the power output deficit. The power output deficit shows one-year cycle of seasonal variability but a steady increasing trend is obvious. In other words, as performance deteriorated, the difference between theoretical and actual power generation tended to increase linearly. The slope of the trend analysis line in the monthly average of power output deficit for four years is 0.43 kW/month, which can be converted to an output reduction rate of −0.52%p per year (i.e.,

Energies 2021, 14, x FOR PEER REVIEW 9 of 12

the rate of deterioration of the capacity factor). One of the reason of seasonal variation would be different wind shear patterns by season in Monsoon climatology. For reference, the average slope of trend analyses of 1-, 2-, 3-, and 4-year periods was Energies 2021, 14, 2319 0.46 ± 0.15 kW/month which is a 6.5% difference (1-year period means each of the9 offour 12 years such as 2009, 2010, 2011, and 2012; 2-year period refers to consecutive two years such as 2009–2010, 2010–2011, and 2011–2012). Energies 2021, 14, x FOR PEER REVIEW 9 of 12

the rate of deterioration of the capacity factor). One of the reason of seasonal variation would be different wind shear patterns by season in Monsoon climatology. For reference, the average slope of trend analyses of 1-, 2-, 3-, and 4-year periods was 0.46 ± 0.15 kW/month which is a 6.5% difference (1-year period means each of the four years such as 2009, 2010, 2011, and 2012; 2-year period refers to consecutive two years such as 2009–2010, 2010–2011, and 2011–2012).

FigureFigure 9.9. TrendTrend analysisanalysis ofof windwind turbineturbine deteriorationdeterioration raterate by normalized capacity factor with the the correctioncorrection ofof windwind resourceresource variability.variability. ((•●:: monthlymonthly capacitycapacity factor,factor, blackblack andand blueblue lines:lines: harmonicharmonic regressionregression with with 95% 95% confidence confidence limits, limits, dashed dashed line: line: linear linear regression). regression).

Figure 10 is a graph representing the monthly trend analysis of the power output deficit. The power output deficit shows one-year cycle of seasonal variability but a steady increasing trend is obvious. In other words, as performance deteriorated, the difference between theoretical and actual power generation tended to increase linearly. The slope of the trend analysis line in the monthly average of power output deficit for four years is 0.43 kW/month, which can be converted to an output reduction rate of −0.52%p per year Figure 9. Trend analysis of wind turbine deterioration rate by normalized capacity factor with the (i.e.,correction the rate of wind of deterioration resource variability. of the capacity (●: monthly factor). capacity One factor, of the black reason and of blue seasonal lines: variationharmonic wouldregression be differentwith 95% confidence wind shear limits, patterns dashed by line: season linear in Monsoonregression). climatology.

Figure 10. Trend analysis of wind turbine deterioration by monthly average of power output defi- cit with the CT-NTF correction. (●: power output difference, ■ : standard deviation, black and blue lines: harmonic regression with 95% confidence limits, dashed line: linear regression).

The monthly trend analysis without applying the NTF showed the same slope of 0.43 kW/month (R2 = 0.77). However, the power output deficit decreased from 85.5 ± 5.4 kW to 41.3 ± 2.9 kW after applying the NTF, resulting in a reduction in both of the mean and the variance for four years by half. In spite of improving the accuracy of power generation prediction by employing NTF calibration, the effect of NTF calibration was not observed Figure 10.10. Trend analysisanalysis ofof windwind turbineturbine deteriorationdeterioration byby monthly monthly average average of of power power output output deficit defi- cit with the CT-NTF correction. (●: power output difference, ■ : standard deviation, black and blue with the CT-NTF correction. (•: power output difference, : standard deviation, black and blue lines: harmoniclines: harmonic regression regression with 95% with confidence 95% confiden limits,ce limits, dashed dashed line: linear line: linear regression). regression).

ForThe reference,monthly trend the average analysis slope without of trend applying analyses the ofNTF 1-, showed 2-, 3-, and the 4-year same periodsslope of was0.43 0.46kW/month± 0.15 (R kW/month2 = 0.77). However, which is the a 6.5% power difference output (1-yeardeficit decreased period means fromeach 85.5 ± of 5.4 the kW four to 41.3 ± 2.9 kW after applying the NTF, resulting in a reduction in both of the mean and the variance for four years by half. In spite of improving the accuracy of power generation prediction by employing NTF calibration, the effect of NTF calibration was not observed

Energies 2021, 14, 2319 10 of 12

years such as 2009, 2010, 2011, and 2012; 2-year period refers to consecutive two years such as 2009–2010, 2010–2011, and 2011–2012). The monthly trend analysis without applying the NTF showed the same slope of 0.43 kW/month (R2 = 0.77). However, the power output deficit decreased from 85.5 ± 5.4 kW to 41.3 ± 2.9 kW after applying the NTF, resulting in a reduction in both of the mean and the variance for four years by half. In spite of improving the accuracy of power generation prediction by employing NTF calibration, the effect of NTF calibration was not observed obviously in the trend analysis. This implies that the wind speed deficit and the power output deficit are attributable to independent causes. The wind speed deficit is mainly due to wake effect and nacelle geometry but the causes of power output deficit are not directly relevant to wake effect. However, it should be noted that a NTF would also be aging due to blade erosion but it was not considered in this study. The rate of performance aging (−0.52%p per year) of the wind turbines at the Shinan wind farm is within the range of −0.48%p per year and −1.0%p per year by month, which are similar to the performance aging rates of onshore wind farms in the UK, US and Canada. Note that this value is limited to the Mitsubishi MWT-1000A wind turbines installed at the Shinan wind farm, whereas the values for the UK, US, Denmark, and Canada are the results of a comprehensive analysis of wind turbines made in Europe with various machine types, production years, and operational periods. However, it is of great significance to quantitatively confirm the performance aging of Asian wind turbines installed in the Asian region. It also supports that the figures are similar to those around the world. The aging of the wind turbines was presumed to be the result of a gradual deterioration in performance due to mechanical aging, and not the result of sudden aging after a given number of years. The overall trend of performance aging progressed in a linear fashion which was observed in previous studies in the UK, Denmark, and Canada. According to South Korea’s carbon-neutral plan, the wind energy should supply 70 TWh of electricity by 2050, and the wind turbine capacity of 24.6 GW is expected to be installed. However, when considering the performance aging of −0.5%p per year, 1.8 GW should be added to the total capacity (+7.3%). This means that accurate prediction and anal- ysis of the performance aging of wind turbines is not only necessary for the development of control strategy, but is also a significant research field to accelerate carbon neutrality.

5. Conclusions The reliability of the method of estimating the rate of performance aging in existing wind farms was somewhat low due to the uncertainty of wind resource variability, when applied to wind farms with a short total operating period, i.e., less than 5 years. To overcome this problem, the difference between predicted and actual power generation using SCADA data, which is a monthly average of power output deficit, was calculated, and the rate of performance aging was estimated using a trend analysis. The main conclusions of the study are summarized as follows: (1) The trend analysis of power output deficit at the Shinan wind farm over 4 years showed that the rate of performance aging was –0.52%p per year, and that it pro- gressed linearly. This value was similar to the figure of –0.48%p per year obtained for onshore wind farms in the UK, and −0.54%p per year in the US. This result implies that generally about 10% of performance aging occurs over the 20 years of wind farm operation, and this figure should be reflected when evaluating the cost of wind energy and setting national target. (2) It was presumed that the nacelle wind speed is affected nonlinearly depending on the wind turbine’s control mode and related to the performance aging. Despite that the prediction accuracy of wind power output was significantly improved when the nacelle wind speed was corrected using the NTF based on CT curve, it was found that the performance aging is not directly affected by this correction. Since the wind turbine aging occurs in the mechanical drive, the nacelle wind speed is not affected by Energies 2021, 14, 2319 11 of 12

the aging of blade control mechanism. Therefore, it is conjectured that the correction using NTF is not mandatory when analyzing the performance aging of wind turbines.

Author Contributions: Data curation, J.-Y.K.; Investigation and writing, H.-G.K. All authors have read and agreed to the published version of the manuscript. Funding: This work was conducted under the framework of the research and development program of the Korea Institute of Energy Research (C1-2410). Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Not applicable. Acknowledgments: Thank you to DONGKUK S&C for providing the Shinan wind farm data. Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations

CFD Computational Fluid Dynamics IEC International Electrotechnical Commission LiDAR Light Detection and Ranging NTF Nacelle Transfer Function MERRA Modern-Era Retrospective analysis for Research and Applications SCADA Supervisory Control And Data Acquisition SODAR SOnic Detection And Ranging

References 1. International Electrotechnical Commission. Wind Turbine, Part. 1: Design Requirement, IEC 61400-1, 3rd ed.; IEC: Geneva, Switzerland, 2005. 2. Le, B.; Andrews, J. Modelling wind turbine degradation and maintenance. Wind Energy 2016, 19, 571–591. [CrossRef] 3. Hughes, G. The Performance of Wind Farms in the United Kingdom and Denmark; Foundation: London, UK, 2012; 48p. 4. Bach, P.F. Capacity Factor Degradation for Danish Wind Turbines; Paul-Frederik Bach Consulting: Switzerland, 2012; 4p. 5. Harrison, J. Economic Analysis of the Algonquin Power Company Amherst Island Wind Energy Generation System; Association to Protect Amherst Island: Amherst Island, ON, Canada, 2013. 6. Staffell, I.; Green, R. How does wind farm performance decline with age? Renew. Energy 2014, 66, 775–786. [CrossRef] 7. Byrne, R.; Astolfi, D.; Castellani, F.; Hewitt, N.J. A study of wind turbine performance decline with age through operation data analysis. Energies 2020, 13, 2086. [CrossRef] 8. Astolfi, D.; Byrne, R.; Castellani, F. Analysis of wind turbine aging through operation curves. Energies 2020, 13, 5623. [CrossRef] 9. Hamilton, S.D.; Millstein, D.; Bolinger, M.; Wiser, R.; Jeong, S.G. How Does Wind Project Performance Change with Age in the United States? Joule 2020, 4, 1–16. [CrossRef] 10. Shin, D.H.; Ko, K.N.; Huh, J.C. A study on method for identifying capacity factor declination of wind turbines. Int. J. Environ. Ecol. Eng. 2015, 9, 55–59. 11. Kim, H.G.; Jang, M.S.; Ko, S.H. Long-term wind resource mapping of Korean west-south offshore for the 2.5 GW project. J. Environ. Sci. Int. 2013, 22, 1305–1316. [CrossRef] 12. Ministry of Trade, Industry and Energy. New & Renewable Energy Statistics 2019; Korea Energy Agency: Seoul, Korea, 2020. 13. Kim, H.G.; Jeon, W.H.; Kim, D.H. Wind resource assessment for high-rise BIWT using RS-NWP-CFD. Remote Sens. 2016, 8, 1019. [CrossRef] 14. Kim, H.G.; Meissner, C. Correction of LiDAR measurement error in complex terrain by CFD—Case study of the Yangyang pumped storage plant. Wind Eng. 2017, 41, 226–234. [CrossRef] 15. Kim, H.G.; Kim, J.Y.; Kang, Y.H. Comparative evaluation of the third-generation reanalysis data for wind resource assessment of the southwestern offshore in South Korea. Atmosphere 2018, 9, 73. 16. International Electrotechnical Commission. Wind Turbines—Part 12-2: Power Performance of Electricity Producing wind Turbines Based on Nacelle Anemometry; IEC 61400-12-2; IEC: Geneva, Switzerland, 2008. 17. Kim, H.G.; Kang, Y.H.; Yun, C.Y. Derivation of nacelle transfer function using LiDAR measurement. Trans. Korean Soc. Mech. Eng. A 2015, 39, 929–936. [CrossRef] 18. Kim, H.G. Method of calibration of nacelle anemometer using LiDAR measurements. In Korea Patent, 10-1383792; Korea Institute of Energy Research: Daejeon, Korea, 2014. Energies 2021, 14, 2319 12 of 12

19. Ormel, F.T.; Donth, A.; Siebers, T.; Faucherre, C.; Luetze, H. Correction method for wind speed measurement at wind turbine nacelle. In European Patent, 1793123B1; GE International Inc.: Munich, Germany, 2016. 20. Martin, C.M.; Lundquist, J.K.; Clifton, A.; Poulos, G.S.; Schreck, S.J. Atmospheric turbulence affects wind turbine nacelle transfer functions. Wind Energy Sci. 2017, 2, 295–306. [CrossRef] 21. Burton, T.; Jenkins, N.; Sharpe, D.; Bossanyi, E. Wind Energy Handbook, 2nd ed.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2011; ISBN 978-0-470-69975-1. 22. Castellani, F.; Astolfi, D.; Mana, M.; Piccioni, E.; Becchetti, M.; Terzi, L. Investigation of terrain and wake effects on the performance of wind farms in complex terrain using numerical and experimental data. Wind Energy 2017, 20, 1277–1289.